You are on page 1of 101

V O L U ME 3 , ISS U E 10 OC T O BE R 2 0 1 9

MASS
M ONTHLY A PPL ICATIO N S IN
STRE N G TH SPO R T

E R IC H E LMS | G R E G N UCK O LS | MIC HAEL ZO URDO S | ERIC T REXL E R


The Reviewers
Eric Helms
Eric Helms is a coach, athlete, author, and educator. He is a coach for drug-free strength and
physique competitors at all levels as a part of team 3D Muscle Journey. Eric regularly publishes
peer-reviewed articles in exercise science and nutrition journals on physique and strength sport, in
addition to writing for commercial fitness publications. He’s taught undergraduate- and graduate-
level nutrition and exercise science and speaks internationally at academic and commercial
conferences. He has a B.S. in fitness and wellness, an M.S. in exercise science, a second Master’s
in sports nutrition, a Ph.D. in strength and conditioning, and is a research fellow for the Sports
Performance Research Institute New Zealand at Auckland University of Technology. Eric earned pro status as a natural
bodybuilder with the PNBA in 2011 and competes in the IPF at international-level events as an unequipped powerlifter.

Greg Nuckols
Greg Nuckols has over a decade of experience under the bar and a B.S. in exercise and sports
science. Greg earned his M.A. in exercise and sport science from the University of North Carolina
at Chapel Hill. He’s held three all-time world records in powerlifting in the 220lb and 242lb classes.
He’s trained hundreds of athletes and regular folks, both online and in-person. He’s written for
many of the major magazines and websites in the fitness industry, including Men’s Health, Men’s
Fitness, Muscle & Fitness, Bodybuilding.com, T-Nation, and Schwarzenegger.com. Furthermore,
he’s had the opportunity to work with and learn from numerous record holders, champion athletes,
and collegiate and professional strength and conditioning coaches through his previous job as Chief Content Director for
Juggernaut Training Systems and current full-time work on StrongerByScience.com.

Michael C. Zourdos
Michael (Mike) C. Zourdos, Ph.D., CSCS, has specializations in strength and conditioning and
skeletal muscle physiology.  He earned his Ph.D. in exercise physiology from The Florida State
University (FSU) in 2012 under the guidance of Dr. Jeong-Su Kim. Prior to attending FSU, Mike
received his B.S. in exercise science from Marietta College and M.S. in applied health physiology
from Salisbury University. Mike served as the head powerlifting coach of FSU’s 2011 and 2012
state championship teams. He also competes as a powerlifter in the USAPL, and among his best
competition lifts is a 230kg (507lbs) raw squat at a body weight of 76kg. Mike owns the company
Training Revolution, LLC., where he has coached more than 100 lifters, including a USAPL open
division national champion.

Eric Trexler
Eric Trexler is a pro natural bodybuilder and a sports nutrition researcher. Eric has a PhD in Human
Movement Science from UNC Chapel Hill, and has published dozens of peer-reviewed research
papers on various exercise and nutrition strategies for getting bigger, stronger, and leaner. In
addition, Eric has several years of University-level teaching experience, and has been involved in
coaching since 2009. Eric is the Director of Education at Stronger By Science.

2
Letter from
the Reviewers

W
elcome to the October 2019 issue of MASS.
The issue kicks off with a discussion of cluster sets and “intraset rest” strategies.
Next, Mike breaks down the first study comparing velocity and RPE for esti-
mating distance from failure. Eric Helms analyzes a study investigating whether diet soda
helps people consume fewer calories, and Eric Trexler digs into the fat-free mass index
research.
In the back half of the issue, Greg digs into the first study analyzing whether anthro-
pometric characteristics can predict if you’re best-suited for the sumo or conventional
deadlift, and Mike reviews a systematic review on the effects of intra-session cooling. Eric
Trexler examines some research suggesting that males and females may have different
responses to nitrate supplementation, and we finish the written content with a systematic
review on the placebo effect in exercise research.
For the videos this month, Mike discusses concurrent training for marathon runners,
and Eric Helms begins a series on energy intake for optimizing lean mass. Most people
don’t know this, but Mike has actually ran a few marathons, and I know quite a few MASS
readers are also serious about trying to maintain their size and strength while improving
endurance performance, so we think you’ll find that video enjoyable and informative.
Enjoy!

The MASS Team


Eric Helms, Greg Nuckols, Mike Zourdos, and Eric Trexler

3
Table of Contents

6
BY G R EG NUCKOL S

Twice the Sets, Half the Reps? The Devil is in the Details
In this article, I reviewed another study looking at the effects of training to failure versus stopping
shy of failure. That study gave me an excuse to dig into the cluster set and intra-set rest literature.
If you’re interested in cluster sets and intra-set rest training, you’ll enjoy this article.

18
BY M I CHAEL C. ZOUR DOS

Intraset RPE and Velocity Provide Similar Estimates of Reps Performed


It is generally accepted that velocity gives you a better gauge of intraset proximity to failure than
RPE, but do we really know this? Although the reviewed study leaves a lot to be desired, it is
actually the first to explore this topic.

30
BY E RI C HEL MS

Diet Sodas and Adherence: Helpful or Counterproductive?


Do “diet” sodas actually help you diet, or do they make it harder? Does the calorie-free flavor
curb your sweet tooth, or set you up for failure by causing cravings? As you’ve probably learned
by now from reading MASS, the answer depends on the context – both the situation, and the
individual.

40
BY E R I C T R EXL ER

Using Fat-Free Mass Index to Forecast Long-Term Gains for Males and
Females
At some point in their lifting career, just about every lifter wonders, “Exactly how big can I get?”
While fat-free mass index (FFMI) isn’t going to tell you the exact limit to your potential, or perfectly
weed out steroid-users from drug-free lifters, it’s still a helpful metric for approximating muscularity
and understanding how big most people can get. Read this article to figure out how big male
and female lifters tend to get, and what this means for your natural potential for muscle growth.

53
BY G R EG NUCKOL S

Can We Predict if You’re Better off Deadlifting Conventional vs. Sumo?


There are a lot of recommendations floating around concerning whether you should deadlift sumo
or conventional based on how you’re built. In spite of the numerous recommendations, there was
no actual research on the topic. That changed with the study reviewed in this article.

4
62
BY M ICHAEL C. ZOUR DOS

Be Cool … Literally
Sitting in a vest with crushed ice and drinking a below-0 degree Celsius beverage has been
investigated as a method to improve exercise performance. Does it actually work? A new meta-
analysis provides us with answers.

73
BY E R I C T R EXL ER

Do the Performance Effects of Beetroot Juice and Dietary Nitrate Differ


Between Males and Females?
Supplements that promote nitric oxide are in just about every pre-workout supplement on the
market. Nitrate, a primary component of beetroot juice, is a notable nitric oxide promoter that
is also found in a variety of fruits and vegetables. Several studies have reported performance
benefits of dietary nitrate for males, but there’s a considerable lack of research in females. Read
this article to find out if nitrate actually improves performance, and if this effect is similar for both
males and females.

85
BY G R EG NUCKOL S

The Placebo Effect Impacts Performance More Than You Might Expect
The placebo effect is a well-known psychological phenomenon, but we often forget about it
in exercise research. We focus on how much a supplement, device, or treatment improves
performance relative to a placebo, but a lot of the “real-world” improvements in performance
may actually be attributable to placebo effects.

95
BY M I CHAEL C. ZOUR DOS

VIDEO: Concurrent Training for a Marathon


You may know that running long distance isn’t the ideal form of cross-training to maximize your
strength. However, maybe you don’t care, and you want to run a marathon anyway. I think this
is cool, as once upon a time I didn’t care and ran some, despite knowing the negative effects on
strength and size. Therefore, this video shows you how to accomplish completing a marathon
while affecting strength as little as possible.

97
BY E RI C HEL MS

VIDEO: Energy Intake to Optimize Lean Mass, Part 1


“You must be in a surplus to build muscle”– An oft-repeated adage in the evidence-based
community, but perhaps an overly simplified statement. In part 1 of this series, we draw on
multiple reviews published in the last 20 years, and mechanistic primary research to explore
how energy intake influences hypertrophy, and when and why a surplus might or might not be
necessary to grow muscle.

5
Study Reviewed: Impact of Two High-Volume Set Configuration Workouts on
Resistance Training Outcomes in Recreationally Trained Men. Karsten et al. (2019)

Twice the Sets, Half the Reps?


The Devil is in the Details
BY G RE G NUC KO LS

In this article, I reviewed another study looking at the effects of training


to failure versus stopping shy of failure. That study gave me an excuse to
dig into the cluster set and intra-set rest literature. If you’re interested in
cluster sets and intra-set rest training, you’ll enjoy this article.

6
KEY POINTS
1. In a recent study, two groups of trained young men either performed 4 sets of
10 to failure for all exercises, or 8 sets of 5 for all exercises, aiming to only get
halfway to failure on all sets.
2. Both groups had similar increases in squat and bench press strength, and
experienced similar biceps and anterior deltoid hypertrophy.
3. The group training to failure experienced greater quad growth.
4. There were some methodological issues with this particular study, but it serves
as a good launching off point to discuss cluster sets and intra-set rest training.

B
uilding off of the study by Car- ing you to do the same total amount of
roll et al (2) that I reviewed last reps with the same load, while staying
month, I dug into another study further from failure and maintaining a
investigating the effects of training to higher average rep velocity and pow-
failure versus stopping each set shy of er output. With cluster sets, this is ac-
failure. In the presently reviewed study complished by taking short rests within
(1), one group of trained men trained each set, while with intra-set rest, this
for six weeks, performing all exercises is accomplished by doubling the num-
for 4 sets of 10 reps to failure, while an- ber of sets, halving the reps per set, and
other group performed all exercises for shortening the rest intervals between
8 sets of 5 with loads they should be able sets. Although cluster sets are popular,
to lift for 10 reps. Strength gains in the the longitudinal evidence in their favor
squat and bench press were similar be- is lacking. However, when training with
tween groups, as were biceps and anteri- lower loads (below ~80% of 1RM), in-
or deltoid hypertrophy. However, vastus tra-set rest training seems to be a prom-
medialis (inner quads) hypertrophy was ising option that may improve strength
significantly greater in the group train- gains.
ing to failure.
The main study I’m reviewing had a
couple major issues that are discussed Purpose and
in the interpretation section. However, Hypotheses
this study gave me an excuse to dig into
the cluster set and intra-set rest litera- Purpose
ture. Cluster sets and intra-set rest ac-
The purpose of this study was to com-
complish the same basic purpose: allow-

7
Table 1 Subject characteristics

Failure group (n=9) Nonfailure group (n=9)

Body mass (kg) 78.4 ± 24.3 76.0 ± 13.8

Fat mass (%) 23.4 ± 11.7 15.1 ± 8.4

Fat mass (kg) 18.6 ± 15.4 11.8 ± 8.0

Fat-free mass (%) 76.6 ± 11.7 84.9 ± 8.4

Fat-free mass (kg) 58.8 ± 13.6 64.2 ± 11.5

1RM bench press (kg) 71.0 ± 26.8 90.0 ± 29.1

1RM squat (kg) 87.2 ± 25.3 102.2 ± 28.5

pare the effects of two training programs


differing only in proximity to failure for Subjects and Methods
each set: One group trained to failure
on each set, while the other group did Subjects
half as many reps per set at the same in- 18 young males participated in this
tensity, but did twice as many sets. Out- study. They all had at least 2 years and
comes of interest were body composi- at most 5 years of prior training experi-
tion, squat and bench press strength, ence, and none of the subjects competed
upper- and lower-body power, and var- in any strength or physique sports. Un-
ious measures of hypertrophy. fortunately, the groups may have wound
up differing in meaningful ways at base-
Hypotheses line. While body mass was similar be-
The authors hypothesized that gains in tween groups, the subjects in the group
upper and lower body power would be not training to failure tended to be lean-
greater when not training to failure, hy- er, have more fat-free mass, and have
pertrophy would be greater when train- greater bench press and squat strength
ing to failure, and both groups would (though these differences weren’t statis-
improve squat and bench press strength tically significant, likely due to high het-
to similar degrees. erogeneity within both groups).

8
Table 2 Exercises performed in the 2 training routines

Training program 1 Training program 2


(back, arm extensors, and lower body)

Bench press Lateral pull-down

Chest press Barbell pull-over

Barbell curl Barbell lying arm extension

Seated dumbbell curl Barbell close grip press on bench

Dumbbell deltoid raise Parallel squat

Barbell shoulder press Deadlift

Barbell shoulder front raise Machine leg curl

Experimental Design ing ultrasound. Strength performance


To ensure that the results of the study was assessed via 1RM bench press and
would not be unduly influenced by squat, and power performance was as-
learning effects, the subjects all started sessed via vertical jump height and
with one week of familiarization with bench press power at 50% of 1RM. All
the squat and bench press. After this of these tests were repeated after the
familiarization week, the subjects un- end of the training intervention.
derwent body composition testing us- Training took place two days per
ing a Bod Pod, the circumferences of week for six weeks. Both groups per-
their arms and thighs were measured, formed the same exercises, seen in Ta-
and muscle thicknesses of their vas- ble 2. Training frequency for each mus-
tus medialis (a quad muscle), triceps, cle group was just once per week. The
and anterior deltoids were assessed us- groups were also matched for rep vol-

9
Figure 1 Changes in muscle thickness between groups

9 *
7
muscle thickness
Changes (mm) in

5
failure
3

1 nonfailure

9
Vastus medialis Elbow flexor Anterior delt
thickness thickness thickness

* = a significant difference between groups (p<0.05)

ume and (theoretically) intensity, but The other group performed 8 sets
differed in terms of the number of sets of 5 reps not to failure for all exer-
performed and proximity to failure. cises, with 1 minute of rest between
One group performed 4 sets of 10 sets (in order to match the routines
reps to failure for all exercises, with 2 for work-to-rest ratios). They start-
minutes of rest between sets. Failure ed with their self-estimated 10RM,
was defined as the subjects self-deter- and adjusted loads based on their ef-
mined rep max (in other words, they fort-based (not reps-in-reserve-based)
didn’t fail the last rep, but they pushed RPE. They were recommended to start
each set to RPE 10, when they didn’t each set with an effort-based RPE of
think they could complete another 5, and to not allow their effort-based
rep). If the description of the methods RPE at the end of a set to exceed 7. If
is correct, the subject did not decrease their final effort-based RPE for a set
their training weights if they were un- exceeded 7, the load was decreased by
able to complete all 10 reps in a set, 2.5kg for the next set. This isn’t spelled
but rather they were allowed an extra out in the methods, but I believe their
30 seconds of rest between sets. If they loads were increased by 2.5kg if their
felt like they could do more than 10 effort-based RPE at the start of a set
reps in a set, the load was increased by was 4 or below.
2.5kg.

10
Figure 2 Changes in 1RM strength between groups in bench press and squat

14
12
Changes (kg) in

10
1RM strength

6
4 failure
2
nonfailure
0
Bench press 1RM Squat 1RM

Findings Interpretation
Of all of the variables measured, the I discussed the effects of training at
only change that was significantly dif- different proximities to failure in the last
ferent between groups was for vastus issue of MASS, and I recently wrote a
medialis thickness (p = 0.026). That was huge article about it for my own web-
actually a pretty large difference, with site, so I actually don’t want to discuss
the group training to failure seeing an this article in the context of failure ver-
increase of ~6%, while the group training sus non-failure training. Rather, I want
further from failure only increased vas- to discuss them in the context of cluster
tus medialis thickness by 0.6% (which sets, which is a subject we haven’t touched
wasn’t even a significant within-group on in MASS recently (Mike covered
change). In the discussion of the paper, an acute study [3] several months ago,
the authors also tried to spin the chang- but we haven’t yet reviewed a longitudi-
es in body composition as favoring the nal training study). This wasn’t exactly a
group training to failure, and the change cluster set study, but it was pretty similar.
in bench press power as favoring the First, however, I’d like to express a bit
group staying further from failure, but of skepticism about the way the authors
none of those between-group compar- of this study standardized intensity. The
isons were particularly close to signifi- authors don’t report relative volume loads
cant (p > 0.16 for all), and none of the (sets x reps x %1RM) or mean training
differences were practically meaningful intensities, even though those factors
either (d < 0.10). were supposed to be equated. I’m con-

11
cerned that intensity may not have been more like 7-8+ reps in the tank, especial-
equated. Specifically, I think the group ly for squats.
training further from failure was train- Thus, if anything, I’m surprised by
ing with lower intensities. Loads were how well the non-failure group did in
lowered when their effort-based RPE this study. They were likely training at
increased above 7 by the end of a set. lower intensities than the failure group,
In research on the squat (4) and bench and considerably further from failure
press (5) using this same effort-based than virtually anyone would recommend
RPE scale, subjects reached an RPE of for strength or hypertrophy. However,
about 7 (7.6 for squat and 6.8 for bench) strength gains were still similar between
with just a 10% decrease in velocity groups, and upper body hypertrophy (bi-
when training with 70-80% 1RM loads. ceps and front delt thicknesses) was sim-
In those same studies, velocity dropped ilar as well, in spite of the fact that the
off by more than 50% in the squat (0.53 non-failure group was exclusively doing
m/s to 0.26 m/s) and nearly 70% in the training that many people would think
bench press (0.49 m/s to 0.15 m/s) be- of as “junk volume.” And if I’m correct
tween the fastest rep and the rep before that the non-failure group was training
failure. Since velocity decreases basically super far from failure, the lack of quad
linearly as one approaches failure, if the growth with just eight sets of squats per
subjects in the non-failure group were week is pretty unsurprising.
actually going halfway to failure (as was
One thing that was weird about this
the authors’ intention), they would have
study is how drastically volumes differed
needed a ~25% velocity drop for each set
for the exercises and muscles tested.
of squats, and a ~35% velocity drop for
Squats were trained just once per week,
each set of bench press. However, if ef-
and there were no other quad-dominant
fort-based RPE is already around 7 with
exercises (the only other lower body lifts
just a 10% velocity loss, it likely exceeds
were deadlift and leg curls). Bench, on
7 well before someone is actually halfway
the other hand, effectively had a fre-
to failure when training with ~10RM
quency of twice per week, and was pretty
loads. Thus, especially for squats (which
directly trained with three different ex-
were already given an effort-based RPE
ercises (bench press and chest press on
of 7.6 with just a 10% velocity loss), I
day 1, and close grip bench on day 2).
strongly suspect the subjects ended up
The front delts would have been stim-
training with loads well below 75% of
ulated directly by 5 different exercis-
1RM, and did not actually go halfway to
es across two days (bench press, chest
failure on each set. Instead of leaving 5
press, shoulder press, front delt raises,
reps in the tank on all sets, I wouldn’t be
and close grip bench), the biceps by
surprised if the subjects had something

12
Figure 3 Examples of traditional, cluster set, and intra-set rest protocols

Traditional training

6 reps 2 minutes rest 6 reps 2 minutes rest 6 reps

Cluster sets

30s 30s 30s 30s 30s 30s


rest rest rest rest rest rest

2 reps 2 reps 2 reps 2 minutes rest 2 reps 2 reps 2 reps 2 minutes rest 2 reps 2 reps 2 reps

Intra-set rest

3 reps 1 minute 3 reps 1 minute 3 reps 1 minute 3 reps 1 minute 3 reps 1 minute 3 reps
rest rest rest rest rest

four different exercises across two days Like I said, though, I want to primarily
(barbell curl, dumbbell curl, reverse grip focus on cluster sets in this interpretation
bent over rows, and pull-downs), and section. Traditionally, cluster sets are de-
the quads by just one exercise on one day fined as sets of an exercise interspersed
(squats; I guess you could also make an with short rest intervals, with each mini-
argument for deadlifts, but I wouldn’t be set not taken to failure (thus distinguish-
very receptive to it). With that in mind, ing it from rest-pause training). So, for
this study suggests that going close to example, instead of doing sets of 6 reps
failure doesn’t matter much for strength with 2 minutes between sets, you might
(within the context of this study, with do 3 mini-sets of 2 reps, with 20 seconds
short-term non-periodized training), between each mini-set, and 2 minutes of
and it only matters much for hypertro- rest after 3 mini-sets (thus accomplish-
phy when volume is low. There was a big ing 6 reps during the cluster). The point
difference in quad hypertrophy between of cluster sets is to better maintain rep
4 sets to failure and 8 sets far from fail- speed and quality throughout a set, since
ure, but not much of a difference in bi- the brief rest intervals decrease metabo-
ceps and front delt hypertrophy between lite buildup and keep you further from
16-20 sets to failure and 32-40 sets far failure. It is thought that doing so will
from failure. enhance gains in strength or power, de-
pending on how the cluster training is

13
implemented. The strategy used in the press strength (9.7% vs. 4.9%). Finally,
current study is reminiscent enough to a study by Oliver et al (9) is a bit of an
cluster sets in both purpose and execu- oddball. It compared traditional training
tion that it’s functionally similar - doing (4x10 with 2 minutes between sets) to
twice as many sets, with fewer reps per intra-set rest training (8x5 with 1 min-
set, but shorter rest periods between sets. ute of rest between sets), and found that
Though the terminology wasn’t used in intra-set rest training led to larger gains
the present study, this strategy has been in both squat and bench press strength
previously termed “intra-set rest.” (63.8 vs. 48.5kg for squat, and 15.1kg vs.
There are a few longitudinal studies 9.1kg for bench press).
that are comparable to the present study When we add the present study (1)
that use either cluster sets or intra-set into the mix, we see two studies (Karsten
rest. For the most part, the theory be- and Nicholson) where traditional train-
hind cluster sets hasn’t really worked out ing and either cluster training or in-
in practice. Hansen et al (6) and Nich- tra-set rest training led to similar gains
olson et al (7) both compared cluster in strength, two studies (Hansen and
loading to traditional high-load training Lawton) where traditional training
(80-95% 1RM). In Hansen et al, tradi- proved superior, and one (Oliver) where
tional heavy training led to significantly intraset rest training led to larger gains
(p < 0.05) larger strength gains in the in strength.
squat than cluster training (18.3% vs. If there’s anything resembling a trend
14.6%). In Nicholson et al, traditional in these three studies, it’s that cluster
heavy training (4 sets of 6 at 85%) and or intra-set rest strategies may produce
two different cluster set arrangements (4 similar or greater strength gains when
cluster sets of 6 reps with 1 rep in each training with lower loads (both Karsten
mini-set with either 85% or 90% 1RM) and Oliver based training on 10RM
led to similar increases in squat strength loads, with most or all of the training
(15.28kg for traditional, 15.83kg for performed with loads below 10RM
clusters at 85%, and 17.22kg for clusters loads, or ~75% 1RM), but may be det-
at 90%). Moving on to studies investi- rimental with higher loads (both studies
gating intra-set rest, Lawton et al (8) where traditional training led to larger
compared traditional training (4x6 with strength gains – Hansen and Lawton
about 4 minutes between sets) against – mostly used loads in excess of 80%
intra-set rest training (8x3 with about of 1RM). There may be a logical expla-
2 minutes between sets) with relative nation for that. With higher loads, any
loads equated, and found that traditional training you do will be heavy enough to
training led to larger increases in bench strongly stimulate the non-hypertrophic

14
APPLICATION AND TAKEAWAYS
For the time being, the research is still murky, but you may make better strength gains
by doing intra-set rest training during the offseason when you’re training with loads
below 80% 1RM. Cluster sets, especially with heavier loads, don’t seem to boost
strength gains. It’s still unclear how cluster sets and intraset rest training influence
hypertrophy.

mechanisms of strength gains, and in- on the new 1RMs established during
troducing additional rest just serves to week 4, and week 8 was a deload and
reduce the overall stress and difficulty testing week. Week 9-12 followed the
of the training. With lower loads, how- same pattern. The weakness of this de-
ever, going closer to failure leads to fa- sign was that loads weren’t autoregulat-
tigue, drops force output, and makes the ed for each subject as they were in the
training slightly less specific for strength present study, but a strength is that you
gains, while introducing more rest peri- knew all subjects were actually training
ods helps keep force output higher, en- with the same intensity, as opposed to
hancing specificity. the approach used in the present study
The prior study that’s the most com- where the non-failure group was likely
parable to the present study is the one training at considerably lower intensities
by Oliver et al, where intraset rest train- than the failure group. The Oliver study
ing led to larger strength gains. I think also had better ecological validity, as the
it’s worth examining the differences accessory lifts were the same for both
between the two studies to understand groups and resembled how people tend
why the intraset rest protocol in Oli- to train accessory lifts (3 sets of 10, likely
ver et al enhanced strength gains, while to failure), as opposed to the presently
the intraset rest protocol in the present reviewed study where the accessory lifts
study failed to enhance strength gains. were trained the same way as the prima-
I think the biggest factor is that inten- ry lifts (who would ever do 8 sets of 5 in-
sity was more tightly controlled in the credibly submaximal front delt raises?).
Oliver study. For both groups, training Ultimately, I think cluster sets are nice
was organized into three four-week in theory, but probably don’t pan out very
blocks. During weeks 1-3, intensity in- well in practice (at least for the power
creased from 65% to 75% 1RM, while lifts; I could see its utility for weight-
the fourth week was a deload and test- lifters, track and field athletes, and team
ing week. During weeks 5-7, loads again sport athletes). Once weights get above
increased by 65% to 75% 1RM, based 80% of 1RM, you probably don’t need to

15
do anything too fancy for your training
to have a robust effect on strength. How- Next Steps
ever, during periods when you’re training I’d honestly just love to see a direct rep-
with lower loads, it may make sense to lication of the Oliver study. I’d also like
use an intra-set rest approach. You still to see an intraset rest study where ter-
get the benefits of training with lower mination of each exercise was based on
loads (namely, giving your body a break reps in reserve. For example, one group
for incessant heavy loading), you may could do sets of 10 at 70% 1RM until
be able to keep gaining strength (if you they reach 2 RIR, and one group could
generally just maintain strength during do sets of 5 at 70% 1RM with shorter
the “offseason”) or increase strength at rest intervals until they reach 2 RIR. I’d
a faster rate, and as long as you’re still be interested in acute fatigue and recov-
doing “normal” accessories fairly close ery after both training designs, as well as
to failure, I doubt you’re missing out on longitudinal strength gains and hyper-
much (if any) hypertrophy, especially if trophy.
you terminate an exercise at the same
overall proximity to failure. For exam-
ple, instead of doing sets of 10 with 2
minutes between sets until you have 2
reps left in the tank, you could do sets
of 5 with the same load with 60 seconds
between sets until you have 2 reps left
in the tank; since you’re terminating the
session after reaching the same overall
level of straining, I think recovery time
would be similar between these two set-
ups as well. I think intra-set rest training
might be good for adherence for some
lifters as well. Powerlifters, as a species,
generally hate doing blocks of 8-12 reps,
but if you can still get a lot of the same
benefits while just doing more sets of
4-6 reps instead, I think a lot of lifters
would find that more palatable.

16
References

1. Karsten B, Fu YL, Larumbe-Zabala E, Seijo M, Naclerio F. Impact of Two High-Volume


Set Configuration Workouts on Resistance Training Outcomes in Recreationally Trained
Men. J Strength Cond Res. 2019 Jul 29.
2. Carroll KM, Bazyler CD, Bernards JR, Taber CB, Stuart CA, DeWeese BH, Sato K, Stone
MH. Skeletal Muscle Fiber Adaptations Following Resistance Training Using Repetition
Maximums or Relative Intensity. Sports (Basel). 2019 Jul 11;7(7).
3. Tufano JJ, Halaj M, Kampmiller T, Novosad A, Buzgo G. Cluster sets vs. traditional sets:
Levelling out the playing field using a power-based threshold. PLoS One. 2018 Nov
26;13(11):e0208035.
4. Chapman M, Larumbe-Zabala E, Gosss-Sampson M, Colpus M, Triplett NT, Naclerio F.
Perceptual, Mechanical, and Electromyographic Responses to Different Relative Loads in
the Parallel Squat. J Strength Cond Res. 2019 Jan;33(1):8-16.
5. Chapman M, Larumbe-Zabala E, Gosss-Sampson M, Triplett NT, Naclerio F. Using Per-
ceptual and Neuromuscular Responses to Estimate Mechanical Changes During Continu-
ous Sets in the Bench Press. J Strength Cond Res. 2018 Feb 22.
6. Hansen KT, Cronin JB, Pickering SL, Newton MJ. Does cluster loading enhance lower body
power development in preseason preparation of elite rugby union players? J Strength Cond
Res. 2011 Aug;25(8):2118-26.
7. Nicholson G, Ispoglou T, Bissas A. The impact of repetition mechanics on the adaptations re-
sulting from strength-, hypertrophy- and cluster-type resistance training. Eur J Appl Physiol.
2016 Oct;116(10):1875-88.
8. Lawton T, Cronin J, Drinkwater E, Lindsell R, Pyne D. The effect of continuous repetition
training and intra-set rest training on bench press strength and power. J Sports Med Phys
Fitness. 2004 Dec;44(4):361-7.
9. Oliver JM, Jagim AR, Sanchez AC, Mardock MA, Kelly KA, Meredith HJ, Smith GL,
Greenwood M, Parker JL, Riechman SE, Fluckey JD, Crouse SF, Kreider RB. Greater gains
in strength and power with intraset rest intervals in hypertrophic training. J Strength Cond
Res. 2013 Nov;27(11):3116-31.

17
Study Reviewed: Rating of Perceived Exertion and Velocity Loss as Variables for
Controlling the Level of Effort in the Bench Press Exercise. Varela-Olalla et al. (2019)

Intraset RPE and Velocity


Provide Similar Estimates of
Reps Performed
BY MIC HAE L C . ZO URD O S

It is generally accepted that velocity gives you a better gauge of intraset proximity
to failure than RPE, but do we really know this? Although the reviewed study leaves
a lot to be desired, it is actually the first to explore this topic.

18
KEY POINTS
1. This study examined the relationship between intraset Borg rating of perceived
exertion (RPE; not repetitions in reserve-based RPE) and velocity loss during sets
to failure at low and moderate loads with reps performed at various points in a set
of the Smith machine bench press.
2. Both RPE and velocity loss were highly correlated with reps performed, with
perhaps a slight edge for velocity loss.
3. Importantly, the traditional Borg RPE scale was used to assess intraset RPE instead
of reps in reserve-based RPE, which severely limits the application. However, this
article does discuss the usefulness of both velocity loss and absolute velocities to
assess proximity to failure.

O
n a mood state scale of “I’m out on the 1.0 m/s set) was also calculated.
of caffeine” to “I just hit a dead- The authors then performed regression
lift PR,” I trend toward the caf- equations with velocity and RPE indi-
feine-less state when scientific literature vidually and together to examine how
states that using velocity to predict repe- strongly these methods correlated with
titions in reserve (RIR) is more accurate the amount of reps performed in a set
than using rating of perceived exertion up to that point. The correlations be-
(RPE). That contention may be true, but tween both RPE and velocity loss with
my annoyance stems from the fact that percentage of reps completed were very
there is no experimental evidence actu- strong (r2 > 0.90), and unsurprisingly,
ally showing this. We shouldn’t perpet- incorporating both RPE and velocity
uate opinions as fact if they are unsup- loss improved the ability of a regression
ported in the scientific literature. This equation to predict reps performed. Al-
study (1) examined the ability of RPE though intraset RPE and velocity loss
and velocity loss during sets to failure both had high correlations with reps
on the Smith machine bench press to performed, this study is a bit frustrat-
predict reps performed. Over three ing, as it used Borg RPE instead of reps
different days, seven men performed in reserve (RIR) based RPE. Further,
Smith machine bench press sets to fail- this study didn’t actually show a predic-
ure at loads corresponding to 1.0 m/s, tion of RIR from RPE or velocity loss;
0.70 m/s, and 0.50 m/s, and were asked rather, it correlated each metric, taken
to provide a Borg RPE after every 5, 3, at different points during a set, with the
and 2 reps, respectively. The velocity loss percentage of reps completed at that
between each interval (i.e. every 5 reps point. Despite these frustrations, there

19
Table 1 Subject characteristics

Smith machine
Number Age (years) Body mass (kg) Height (cm) Training experience
bench 1RM (kg)

7 men 22 ± 1.9 77.8 ± 11.7 176 ± 68 83.6 ± 16.4 > 1 year

Data are mean ± SD


Subject characteristics are from Valera-Olalla et al. 2019 (1)

is some benefit from examining these


data, so I’ll do my best to provide some Subjects and Methods
actionable takeaways in this article.
Subjects
Seven men with a training experience
Purpose and of at least one year participated. The par-
Hypotheses ticipants, however, were not very well-
trained, as their Smith machine bench
press one-repetition maximum (1RM)
Purpose was only slightly greater than their body
The purpose of this study was to exam- mass, on average. Specific subject details
ine the strength of the relationships be- are in Table 1.
tween the percentage of reps performed
during sets to failure and both intraset Study Design
RPE and velocity loss, and to investi- The methods in the paper were not de-
gate if including both RPE and veloc- scribed particularly clearly, so I’ve done
ity loss together in the same regression my best to interpret and explain them
equation improved the strength of the as clearly as possible. This study con-
correlation. sisted of two familiarization sessions
and three experimental sessions. During
Hypotheses the first familiarization session, subjects
The authors predicted that both RPE had anthropometrics (height and body
and velocity loss would be similarly cor- mass) assessed, then performed a Smith
related with the percentage of reps per- machine bench press 1RM. During the
formed during sets to failure, and that second familiarization session, subjects
including RPE and velocity loss in the performed one set to failure at a load
same regression equation would im- they could initially press at 0.70 m/s to
prove the strength of the relationship. familiarize the subjects with what a true
10RPE set felt like.
During each of the three experimen-

20
tal visits, subjects performed a modified models – examined the agreement be-
1RM test before performing one set to tween actual reps performed and RPE
failure with a load that corresponded to and velocity loss predictions of reps per-
an initial average concentric velocity of formed.
1.0, 0.70, or 0.50 m/s. The loads used for
the set to failure were determined during
the modified 1RM test. Subjects with a Findings
1RM greater than 80kg performed two
reps at 60kg after performing a single at Reps Performed, Velocity Load, Load
80kg, rather than working up to a true Used
1RM. I’m not sure why this modifica- First, let’s report simple observations.
tion was implemented, but this is the Table 1 shows how many reps were per-
reason for calling it a “modified” 1RM formed at each load lifted, the best rep
test. (often the first rep) and last rep velocity
at each load, and the percentage veloci-
Assessing RPE and Velocity Loss ty loss from the best rep to the last rep.
During the sets to failure with the 1.0 The 1.0, 0.70, and 0.50 m/s loads cor-
m/s load, the subjects were asked for an responded to 43.5%, 63.7%, and 77.3%
RPE on the 1-10 Borg scale after ev- of 1RM, respectively. As you would ex-
ery 5 reps, and velocity loss was calcu- pect, more reps were performed with the
lated between every 5 reps as well. This 1.0 m/s load compared to 0.70 m/s load,
scale can be seen in Table 2 here. Im- which in turn yielded more reps than the
portantly, the Borg RPE scale (2) simply 0.50 m/s load. Velocity loss was similar
gauges effort and not RIR, which limits between the 1.0 (~80%) and 0.70 m/s
the applicability of this study (more on (~80%) loads, but both of these loads
that later). During the set to failure with had greater velocity loss from the fastest
0.70 m/s, RPE and velocity loss were to last rep compared to 0.50 m/s load.
assessed during every 3-rep increment,
and the same was done during every RPE and Velocity Loss Prediction of Reps
2-rep increment during the set to fail- Performed
ure with the 0.50 m/s load. Regression Both RPE (r2 = 0.89) and velocity loss
equations were then used with RPE and (r2 = 0.91) were highly correlated with
velocity both individually and together reps performed. These correlations sim-
to examine their relationships with the ply show that RPEs were rated higher
percentage of reps performed in the set. and the magnitude of velocity loss was
Bland-Altman plots – graphical repre- greater as more reps were performed,
sentations of agreement between two which is obviously to be expected. Fur-

21
Table 2 Descriptive velocity data, load used, and reps performed

Load 1 ( 1.00 m/s) Load 2 ( 0.70 m/s) Load ( 0.50 m/s)

Variables Mean ± SD (range) Mean ± SD (range) Mean ± SD (range)

AV Best (m/s) 1.00 ± 0.05 (0.91 - 1.06)# 0.71 ± 0.03 (0.68 - 0.76)* 0.51 ± 0.03 (0.47 - 0.55)*^

AV Last (m/s) 0.20 ± 0.08 (0.11 - 0.31) 0.15 ± 0.03 (0.10 - 0.19) 0.15 ± 0.03 (0.11 - 0.18)

VL (%) 80.20 ± 9.0 (67.1 - 88.9) 79.0 ± 4.1 (75.1 - 85.4) 69.3 ± 6.5 (61.7 - 78.3)*^

Load (kg) 36.4 ± 5.8 (30.0 - 45.0)# 53.2 ± 8.6 (40.0 - 65.0)* 64.6 ± 9.9 (52.5 - 77.5)*

Repetitions 41.4 ± 15.8 (24.0 - 74.0)# 19.3 ± 4.1 (13.0 - 25.0)* 9.6 ± 1.4 (8-12)*^

Data are mean ± SD


From Valera-Olalla et al. 2019 (1)
AVbest = Average velocity of the best rep in the set; AVlast = Average velocity of the last rep in the set; VL = Velocity loss
* = Significantly different than the 1.00 m/s load; # = Significantly different than the 0.70 m/s load
^ = Significantly different than both the 1.00 and 0.70 m/s loads

ther, Figures 1A and 1B show that both that they necessarily agree (3); thus, a
RPE and velocity loss predicted the Bland-Altman plot is necessary to de-
amount of reps left with similar accuracy. termine agreement. The main takeaway
Specifically, the percentage of reps actu- from the Bland-Altman plots (Figure
ally completed is on the x-axis in Figure 2AB) is that they suggest some degree
2, and on the y-axis is the percentage of inaccuracy of both RPE and velocity
of reps which either RPE (Figure 1A) loss to predict reps performed, which is
or velocity (Figure 1B) predicted had not surprising as it is more difficult to
been completed at that point in the set. “agree” than to “correlate.” Further, the
RPE had a standard error of estimate of plots do not definitively show one meth-
9.24%, while velocity had a standard er- od (RPE or velocity loss) to be better
ror of estimate of 8.52%. than the other. The plots presented by
the authors, however, appear to have a
Agreement of RPE and Velocity Loss with few issues. First, the dashed lines on the
Reps Performed top and bottom represent the limits of
The authors also used Bland-Altman agreement and should reflect the range
plots, which visually display the level of of the confidence intervals listed on the
agreement between the two methods. top of the graph; however, the range of
Although two methods may highly cor- the lines is slightly different from the
relate (as seen above), that does not mean confidence intervals. Additionally, even

22
Figure 1 RPE and velocity loss predictions of reps performed
though a Bland-Altman is a graphical
representation of agreement, some au-
thors have recommended against using A RPE
100
them for repeated measures data (which y = 0.89x + 6.62
r = 0.94 (0.92 - 0.96)
90
is what these data are) (4). In short, SEE: 9.24%
identity line
80
while there are issues with the plots, I 70
think we can be confident that neither

%REP predicted
60
intraset RPE or velocity loss were truly 50
excellent predictors of the percentage of 40

reps performed in a set. 30

20

Interpretation
10

0
0 10 20 30 40 50 60 70 80 90 100

These data are both interesting and %REP actual

frustrating. They are interesting, because


they show that RPE and velocity loss B
VL
provide pretty similar estimates of how 100
y = 0.91x + 5.44
many reps had been performed at var- 90
r = 0.95 (0.93 - 0.97)
SEE: 8.52%

ious points throughout a set. However, 80


identity line

the data are frustrating for various rea- 70


%REP predicted

sons: 1) Borg RPE was used despite the 60

authors being aware of the RIR-based 50

40
RPE scale, 2) they included RPE and
30
velocity predictions in the regression
20
equations when really far from failure
10
(~36 reps), which is not used in a prac- 0
tical setting, 3) a Smith machine bench 0 10 20 30 40 50 60 70 80 90 100

%REP actual
press was used instead of free weights,
and 4) relatively novice subjects were From Valera-Olalla et al. (1)
RPE = Rating of Perceived Exertion; VL = Velocity Loss.
used. So, before providing actionable These figures show the percentage of reps actually completed plotted against the percentage
of reps which RPE (Figure 1A) and velocity loss (Figure 1B) predicted were completed when

takeaways from this study, we must first an RPE was called or velocity loss was assessed during the repetition intervals. The dashed
line is the “identity” line. The identity line is exactly at a 45-degree angle, which indicated a 1:1
relationship, or a perfect prediction. The regression line for both RPE (1A) and velocity loss (1B)
offer some critiques and limitations of is tightly conformed to the identity line.

this study. To do this, I’m going to kick


us off with some whining. RIR-based RPE scale in the introduc-
tion of the paper, yet still used the Borg
Critiques and Limitations RPE scale. My first thought is “why”?
The authors cited the utility of the Hackett and colleagues showed in sepa-

23
Figure 2 Bland-Altman plots
rate studies in 2012 (5) on the squat and
bench press and in 2016 (6) on the chest
press, leg extension, and leg press that RPE
r2 = 0.031
gauging intraset RIR was more accurate A 95% Cl: -19.19 - 19.13
Systematic bias ± random error: -0.003 ± 9.78%

for predicting proximity to failure and 40

levels of effort than using Borg RPE. In

Difference actual-predicted %REP


30

+1.96 SD
fact, in the Hackett studies, it was com- 20

monplace for subjects to report moder- 10

ate RPEs on the Borg scale even when 0


bias

reaching failure. This prompted further -10

studies to use the RIR-based RPE scale -20


-1.96 SD

to assess intraset proximity to failure (7). -30

The authors did not refer to any of these -40

data, so it is possible they aren’t aware of 0 10 20 30 40 50

mean actual-predicted %REP


60 70 80 90 100

it; however, I doubt that is the case. In


the presently reviewed study, the authors
“anchored” RPE by having subjects per- B
VL
r2 = 0.023
form a set to failure during the famil- 95% Cl: -17.49 - 17.39
Systematic bias ± random error: -0.053 ± 8.89%

iarization to practice what a 10RPE felt


40
Difference actual-predicted %REP

like. However, it is likely that this an-


30

+1.96 SD
choring didn’t completely negate any is- 20

sues using Borg RPE, as the effort-level


10

bias

descriptors at submaximal RPEs were 0

still ambiguous. The authors stated


-10

-1.96 SD
that one of the purposes of the study
-20

was to examine if the “goodness of fit”


-30

(i.e., the r2) to predict the percentage of


-40
0 10 20 30 40 50 60 70 80 90 100

reps performed during a set would be mean actual-predicted %REP

improved if RPE was included in the


same multiple regression with veloci-
From Valera-Olalla et al. (1)
RPE = Rating of Perceived Exertion
These figures examine the agreement between the percentage of reps predicted by RPE (A) and velocity loss (B)

ty loss compared to a linear regression


with the actual percentage reps performed

with velocity loss as the sole predictor.


In that sense, sure, Borg RPE would be done in this study), adding another vari-
novel, but we already know it’s inferior able – no matter what it is (you could
for determining intraset proximity to include subject’s shoe size) – will virtu-
failure to the RIR-based scale, and the ally always increase the r2 value to some
stated purpose itself is not very sound. degree. Also, it’s not helpful to include
When you run a multiple regression (as two metrics in the same regression that

24
people aren’t doing 40-rep sets. From a
design perspective, using the RIR-based
RPE scale during more normal rep sets
RPE AND VELOCITY LOSS (i.e. <20) would have been far preferable
PROVIDE PRETTY SIMILAR for actually comparing the accuracy of
intraset RPE and velocity loss to predict
ESTIMATES OF HOW MANY RIR. In other words, this study doesn’t
provide us with the comparison that we
REPS HAD BEEN PERFORMED really need.

AT VARIOUS POINTS Furthermore, as we know from the pre-


viously referenced Hackett studies (5, 6),
THROUGHOUT A SET. Borg RPE is an especially poor estimate
of effort at close proximities to failure,
which is where RIR-based RPE shines
(7). As stated above, the anchoring in
this study may have helped the estimates
are so highly correlated with each other.
in this study closer to failure, but it cer-
In fact, this is called “multicollinearity,”
tainly didn’t fully negate the inaccuracy.
and it’s something you’re taught to avoid
Unfortunately, the authors only ran the
on day 1 of learning about multiple re-
regressions with all data points included
gression. Rather, you should see if those
and not an additional regression to ex-
two methods agree with each other with
amine the accuracy of RPE and veloci-
agreement statistics such as Bland-Alt-
ty loss when closer to failure. This could
man and Mountain Plots (Folded Cu-
have been done by conducting a sepa-
mulative Empirical Distribution Plots).
rate analysis on only RPEs and velocity
I can see a rationale for Borg RPE to loss after the set was close to completion
generally gauge fatigue during extreme- (i.e. 75% of reps had been completed).
ly high-rep sets. In this study, about However, we can infer some results for
~41 reps were performed at the 1.0 m/s how RPE and velocity loss fared later in
load (43.5% of 1RM), which meant that the set in Figure 1. Although the r-val-
when an RPE was called for after 5 reps, ues are similar between RPE (0.94) and
there was, on average, an RIR of 36. Ob- velocity loss (0.95) in Figure 1AB, by
viously, it’s difficult to accurately predict simply looking at the scatter plots, we
RIR when about 10 reps from failure, so see that on the bottom left hand corner
accurately predicting 36 RIR is nearly of the plots (i.e. toward the beginning
impossible. Therefore, a general gauge of of the set and far from failure), the in-
effort (i.e. Borg RPE) is OK early in such dividual data points for RPE (panel A)
a high-rep set; however, in practice, most

25
are pretty tightly conformed to the iden-
tity line (dashed line), but these points
get farther away from the identity line
at the opposite end, which suggests the
OFTENTIMES, INDIVIDUALS
relationship between RPE and reps per- CAN MAINTAIN THE SAME
formed is worse when closer to failure.
The opposite is true for velocity loss, in VELOCITY FOR A FEW REPS
that more data points are tightly con-
formed to the identity line closer to fail- DURING MODERATE- TO
ure. This shouldn’t be surprising, as the
aforementioned Hackett studies (5, 6)
HIGH-REPETITION SETS;
showed lifters to underrate Borg RPE
on various exercises (squat, bench, chest
THUS, VELOCITY LOSS WILL
press, leg extension, and leg press) when BE VERY MINIMAL ON THE
close to failure (within three reps), and
even recorded submaximal Borg RPE FIRST FEW REPS OF A SET.
ratings when reaching failure during
a set. Velocity, on the other hand, does
not decline linearly during a high-rep
thus, applying the degree of velocity loss
set to failure. Oftentimes, individuals
universally to predict the percentage of
can maintain the same velocity for a few
reps that have been completed is not a
reps during moderate- to high-repeti-
great metric. Besides, it’s more import-
tion sets; thus, velocity loss will be very
ant to predict RIR (how far someone is
minimal on the first few reps of a set.
from failure), rather than how many reps
Another issue is the use of velocity loss have been completed. To do this, it seems
rather than absolute velocities, which we more prudent for individuals to find out
have discussed previously. It’s difficult to how many RIR they have at specific ab-
give normative velocity loss percentages solute velocities. For example, a 1RM
because one person may have a starting squat is commonly reported around
velocity of 0.70 at 70% of 1RM, while 0.25 m/s in the literature (8, 9, 10),
another lifter has an average velocity of which would put someone at about an 8
0.60 m/s on their first rep. Therefore, at RPE (2 RIR) when they are at 0.35 m/s.
a velocity loss of 40%, the first individual Therefore, I would recommend finding
would be at a velocity of 0.48 m/s with what absolute velocities correspond to
the other individual at a velocity of 0.36 what RPE/RIR. This is actually easy to
m/s. This would theoretically leave the do in practice. Simply perform a set to
latter individual much closer to failure; failure at various intensities (70, 75, 80,

26
APPLICATION AND TAKEAWAYS
1. Borg RPE and velocity loss are both decent predictors of the percentage of reps
completed at various points throughout a set with velocity loss being better than
Borg RPE when closer to failure.
2. This study has glaring issues that limit its application. However, from other data,
we know that RIR-based RPE is effective at accurately gauging proximity to failure
when close to failure during low- to moderate-rep sets.
3. This study does, however, highlight some of the inconsistencies with velocity loss.
When using velocity, it may be better to use absolute velocity values as opposed
to velocity loss and cross-reference those values with RPE/RIR. As discussed, this
can be accomplished by performing a set to failure and recording velocity on each
rep.

and 85% are probably the most relevant) the ability to grind on the last few sets.
and track velocity throughout the set. The 0.15 m/s last rep is also faster than
Afterward, you can match up each re- Helms et al (9), who reported a 10RPE
corded average concentric velocity with to be 0.10 ± 0.04 m/s on the bench press.
the RPE/RIR at that point in the set. Be However, this difference in velocity at
aware, however, that when you do this, failure between studies can be explained
you will probably find very little change by the current study using a Smith ma-
in velocity over the first few reps at 70% chine, a two-second pause on the chest,
of 1RM, so at lower to moderate inten- and a cohort of fairly novice lifters ver-
sities, I might only match up RPEs with sus free-weight bench press and well-
velocities once velocities start to consis- trained lifters in Helms’ study. Therefore,
tently decline on every single rep. if using absolute velocities to correspond
A final interesting note about this study with RPE/RIR, not only should the ve-
is that it reported the last rep on the 0.70 locities be individualized, but the veloci-
and 0.50 m/s sets to be 0.15 m/s, while ties should be exercise-specific.
the last rep on the 1.0 m/s set was 0.20
m/s. Theoretically, these should all be
the same; however, it seems likely that Next Steps
metabolic fatigue played a much greater The next steps are simple. We need a
role in fatigue on the 1.0 m/s set, as ~41 study that assesses intraset velocity and
reps were performed. Practically, this RIR-based RPE on the squat, bench
probably translated to some shaking of press, and deadlift and a statistical anal-
the arms and instability, which reduced ysis that directly compares the accuracy

27
of RPE and velocity for predicting RIR.
These potential data can reveal if the
objectivity of velocity is indeed superi-
or to RPE for gauging RIR or if RPE
gets the job done by itself. No matter the
findings, certain personality types will
still make RPE a poor choice for some,
but the practicality in these data would
be wide-ranging.

28
References

1. Varela-Olalla D, del Campo-Vecino J, Leyton-Román M, Pérez-Castilla A, Balsalo-


bre-Fernández C. Rating of perceived exertion and velocity loss as variables for controlling
the level of effort in the bench press exercise. Sports biomechanics. 2019 Jul 29:1-5.
2. Borg GA. Psychophysical bases of perceived exertion. Med sci sports exerc. 1982 Jan
1;14(5):377-81.
3. Altman DG, Bland JM. Measurement in medicine: the analysis of method comparison stud-
ies. Journal of the Royal Statistical Society: Series D (The Statistician). 1983 Sep;32(3):307-
17.
4. Myles PS, Cui J. I. Using the Bland–Altman method to measure agreement with repeated
measures.
5. Hackett DA, Johnson NA, Halaki M, Chow CM. A novel scale to assess resistance-exercise
effort. Journal of sports sciences. 2012 Sep 1;30(13):1405-13.
6. Hackett DA, Cobley SP, Halaki M. Estimation of Repetitions to Failure for Monitoring
Resistance Exercise Intensity: Building a Case for Application. The Journal of Strength &
Conditioning Research. 2018 May 1;32(5):1352-9.
7. Zourdos MC, Goldsmith JA, Helms ER, Trepeck C, Halle JL, Mendez KM, Cooke DM,
Haischer MH, Sousa CA, Klemp A, Byrnes RK. Proximity to Failure and Total Repetitions
Performed in a Set Influences Accuracy of Intraset Repetitions in Reserve-Based Rating of
Perceived Exertion. Journal of strength and conditioning research. 2019 Feb.
8. Zourdos MC, Klemp A, Dolan C, Quiles JM, Schau KA, Jo E, Helms E, Esgro B, Duncan
S, Merino SG, Blanco R. Novel resistance training–specific rating of perceived exertion scale
measuring repetitions in reserve. The Journal of Strength & Conditioning Research. 2016
Jan 1;30(1):267-75.
9. Helms ER, Storey A, Cross MR, Brown SR, Lenetsky S, Ramsay H, Dillen C, Zourdos MC.
RPE and velocity relationships for the back squat, bench press, and deadlift in powerlifters.
The Journal of Strength & Conditioning Research. 2017 Feb 1;31(2):292-7.
10. Izquierdo M, González-Badillo JJ, Häkkinen K, Ibanez J, Kraemer WJ, Altadill A, Eslava J,
Gorostiaga E. Effect of loading on unintentional lifting velocity declines during single sets of
repetitions to failure during upper and lower extremity muscle actions. International journal
of sports medicine. 2006 Sep;27(09):718-24.

29
Study Reviewed: Do Low-Calorie Sweetened Beverages Help to Control Food
Cravings? Two Experimental Studies. Maloney et al. (2019)

Diet Sodas and Adherence:


Helpful or Counterproductive?
BY E RI C HE LMS

Do “diet” sodas actually help you diet, or do they make it harder? Does
the calorie-free flavor curb your sweet tooth, or set you up for failure
by causing cravings? As you’ve probably learned by now from reading
MASS, the answer depends on the context – both the situation, and the
individual.

30
KEY POINTS
1. This research examined ad libitum (as desired) food intake after exposure to
a food craving stimulus or control in two different experiments: 1) in diet soda
non-consumers versus frequent consumers, and 2) among frequent diet soda
consumers when diet soda was or wasn’t available during ad libitum eating.
2. In experiment 1, diet soda non-consumers ate more after exposure to a craving
stimulus than a control; however, frequent consumers did not. In contrast, in
experiment 2, frequent consumers ate more following a craving stimulus than a
control; however, they ate less when diet sodas were available versus when they
were not.
3. Frequent consumers of diet soda being “protected” against craving-induced
overeating was only observed in the first, but not second study. However, more-
or-less supporting the theme that diet sodas may aid weight loss efforts, less total
food was consumed when diet soda was available to the participants versus when
unavailable in the second study.

T
he present study (1) examined associations reported between BMI and
an aspect of the hotly debated diet soda consumption, this is almost
topic of whether diet sodas are certainly reverse causation; individuals
helpful or harmful for weight loss. In the with a higher BMI are more likely to
lay public, the crux of the argument is be on a diet and subsequently consume
normally on faulty grounds, with most artificial sweeteners (3), as randomized
anti-diet soda proponents claiming diet controlled trials broadly report replacing
sodas result in weight gain despite hav- regular soda (and sometimes even water)
ing zero calories, due to “chemicals,” “in- with diet soda results in modest weight
sulin,” “hormones,” “toxins,” or by harm- loss (4). However, like I discussed last
ing “gut health.” To be charitable, there month in my review of processed foods’
is very preliminary data that at high (but effect on ad libitum energy intake (5),
still within assumed-to-be-safe) dosage what is factually correct is not always the
ranges, a small minority of the available same as what is useful. Indeed, we need
artificial sweeteners could plausibly have more data on how artificial sweeteners
a minor negative impact on the gut mi- impact craving-related ad libitum food
crobiome; however, the only causal hu- consumption in applied conditions rep-
man data are from a controversial study resentative of the “real world.” This study
on the rarely used sweetener saccharin set out to do just that and examined how
(2). While there are sometimes positive consumers and non-consumers of diet

31
soda behave in ad libitum eating scenar- would be protected from overeating fol-
ios after exposure to food cravings, and lowing a craving stimulus if diet soda
how diet soda availability moderates this was available, and would overeat when
behavior. In the following article, I’ll re- diet soda was not available.
view the findings of this study and ex-
plain their relevance.
Subjects and Methods
Purpose and Research Experiment 1
Questions 120 male and female university staff
and students, age 31.44 ± 8.54 years,
Purpose participated in this study and were di-
vided into two 60-person groups based
The broad purpose of this research,
on whether they were habitual consum-
conducted across two experiments, was
ers of diet soda or non-consumers. At
to determine the mechanisms (cog-
baseline, all participants were given a
nitive/behavioral) by which diet soda
series of questionnaires related to eating
might affect eating behavior. Specifically,
behavior, attitudes and beliefs about diet
by priming hedonic eating with a choc-
soda, and enjoyment of diet soda. They
olate craving stimulus, this study aimed
were also asked about their eating atti-
to observe ad libitum caloric intake in
tudes toward chocolate. Table 1 shows
frequent diet soda consumers compared
comparative characteristics of each
to non-consumers (experiment one) and
group, with an emphasis on their eating
in frequent consumers when diet soda
attitudes. Following group assignment,
was and was not available (experiment
each group was exposed to a craving and
two).
a control stimulus in a random order in
Hypotheses separate lab visits. In the craving stim-
ulus condition, they were given a choc-
Experiment 1: Energy intake would olate bar which they had to hold, open,
be greater after a hedonic craving stim- and smell for a given amount of time. In
ulus relative to the control stimulus in the control condition, they were instead
non-consumers of diet soda, and fre- given a wooden block of the same shape
quent diet soda consumers would be and size. Then, following the control or
protected from overeating during the craving stimulus, the participants were
craving stimulus by seeking out diet soda provided both sweet and savory snack
to drink instead. foods, their regular soda of choice (or
Experiment 2: Frequent consumers full-sugar version of their preferred diet

32
soda in the diet soda consumer group), Table 1
water, and their diet soda of choice (or Characteristics
Frequent consumers of Non-consumers of diet

diet version of their preferred full-sug-


diet soda (n=60) soda (n=60)

ar soda in the diet soda non-consumer


Age (y) 30.45 (9.17) 32.43 (7.81)

BMI (kg/m2)
group), and were free to eat and drink
26.29 (4.26) 22.80 (3.48)*

TFEQ
whatever they wanted. The energy con-
Disinhibition 8.35 (2.62) 6.33 (2.77)*
sumed was measured and compared be-
DEBQ
tween groups and conditions. As a side
Restraint 3.24 (1.07) 2.63 (1.10)**
note, there was also a visual test where Emotional 2.96 (0.90) 3.01 (0.89)
images of regular and diet sodas and wa- External 3.25 (0.47) 3.30 (0.53)
ter were shown to the participants fol- Attitudes and beliefs

lowing the craving or control stimulus Appetite and weight

before the ad libitum eating occurred to


5.61 (6.07) 2.58 (1.16)*
management

see if the frequent consumers paid more Palatability and


enjoyment
5.10 (1.25) 3.31 (1.43)*

attention to images of diet sodas after ex- ACQ

periencing cravings. This served a useful Trait functional 34.66 (14.60) 35.53 (15.50)

purpose in that the researchers told the Trait guilt 44.56 (16.66) 31.91 (19.04)*

participants they were only investigating Trait craving 51.62 (18.27) 47.07 (22.46)

attentional focus after a craving stim- Values are means with standard deviations in parentheses

ulus, and not ad libitum energy intake.


TFEQ = three factor eating behavior; DEBQ = Dutch eating behavior questionnaire;
ACQ = attitudes to chocolate questionnaire
* = p < .001, ** = p < 0.05 frequent consumers vs. non-consumers
They were told that the food was provi-
sioned simply as thanks for volunteering
and because they had to fast for at least ipants exposed to the same craving or
three hours prior to the study. The par- control stimulus, followed by provision
ticipants were unaware their food intake of food and the allowance of ad libi-
was monitored and measured and that tum eating and drinking. In this case,
energy intake was the principal variable though, all 172 participants were fre-
assessed. However, while this served the quent diet soda consumers who were
useful purpose of providing additional equally divided into four groups: 1) the
control, the attentional data didn’t add control stimulus followed by food pro-
much to the principal findings, so I’ll vision with diet soda available, 2) the
skip over them for the purpose of this control stimulus followed by food pro-
review. vision without diet soda available, 3) the
craving stimulus followed by food pro-
Experiment 2 vision with diet soda available, and 4)
the craving stimulus followed by food
The second experiment was conduct-
provision without diet soda available.
ed in a similar manner, with the partic-
The measured variables I’ll review were

33
Table 2 Descriptive statistics for diet soda (available vs. unavailable) and craving (vs. control) groups

Craving diet soda Control diet soda Craving diet soda Control diet soda
Characteristics
available available unavailable unavailable

N 43 43 43 43

Age (y) 29.05 (12.94) 27.00 (9.48) 26.86 (11.95) 28.16 (12.66)

BMI (kg/m2) 26.94 (4.67) 25.90 (4.19) 27.36 (3.97) 27.54 (3.71)

DEBQ

Restraint 3.25 (0.78) 3.15 (0.87) 3.16 (0.84) 3.25 (0.75)

Emotional 2.91 (0.91) 2.79 (0.84) 3.05 (0.81) 2.75 (0.76)

External 2.92 (0.85) 2.93 (0.73) 2.86 (0.91) 2.83 (0.75)

Values are means with standard deviations in parentheses


DEBQ = Dutch eating behavior questionnaire;

the same, and the group breakdowns the diet soda drinkers in either condi-
are shown in Table 2. tion were not statistically significant).
In experiment two, even when diet soda
was available, the participants (who were
Findings all diet soda drinkers) consumed more
The principal finding in experiment during the craving stimulus condition.
one was that frequent consumers of diet With that said, the total calories con-
soda did not consume more energy fol- sumed was lower when diet soda was
lowing the craving condition compared available compared to when it was not,
to the control condition, while the diet even if diet soda availability didn’t pre-
soda non-consumers did consume more vent an increased consumption of ener-
energy following the craving stimulus gy following the craving stimulus (Fig-
condition (Figure 1). ure 2).
However, this finding was not repli-
cated in experiment two. Importantly,
in experiment one, there was a with-
Interpretation
in-group difference between conditions, There are a couple of ways to look at the
not a between-group difference (i.e. diet findings from a big picture perspective.
soda non-consumers ate more food First, you could scratch your head at the
during the craving condition than the conflicting finding of diet soda drinkers
control condition, but the differenc- being protected against craving-induced
es in how much they ate compared to overeating in experiment one, but not in

34
Figure 1 Mean energy intake following craving and control
conditions in frequent and non-consumers

900
* Control
Calories (kcal)

Craving
700

500

300

100

Frequent consumers Non-consumers

*p = 0.037
Error bars represent standard error of the mean

experiment two, and question the utility question I posed in the teaser of this
of diet sodas for dieters. But, another per- article (“Does the calorie-free flavor curb
spective is that when diet soda was avail- your sweet tooth, or set you up for failure by
able in experiment two, during either causing cravings?”), I’d say the evidence
the control or craving condition, overall leans toward diet sodas being helpful.
food consumption was lower compared If you’re the curious type, you might
to when diet soda was not available. That wonder why there was a disparity at all.
isn’t the same as being protected from Even for those who just want the black
craving-induced overeating as was sug- or white answer, it’s worth considering
gested in experiment one, but it at least the “why” of the findings. Reason being,
suggests a reduced magnitude of over- I think the utility of diet soda might dif-
eating occurs when diet soda is available. fer based on the individual situation.
If you’re application focused and sim- To help you understand why I be-
ply want a black or white answer to the lieve that’s the case, we have to think

35
Figure 2 Mean energy intake in the craving and control conditions and in the
diet soda available and unavailable groups

*
900 *
800 * Control
700 Craving
Calories (kcal)

600
500
400
300
200
100
0
Diet sodas available Diet sodas unavailable

*p < 0.05
Error bars represent standard error of the mean

deeper about who is really compared in they didn’t assess beliefs about the ef-
this study. Ask yourself, why do people fectiveness of diet soda for weight and
choose to drink diet soda? Notice fre- appetite control. Since everyone was a
quent consumers had significantly high- frequent consumer, questions were not
er BMI, restraint and disinhibition (im- asked specifically about diet soda be-
pulsivity) toward eating, guilt associated liefs; however, not everyone consumes
with eating chocolate, and higher be- diet soda for the same reason. Certainly,
liefs that diet soda was effective in con- if you recruit a large group of diet soda
trolling appetite and weight compared consumers, you’ll get many who display
to non-consumers (Table 1). Essentially, attitudes and traits associated with di-
this was also a comparison of a group eting, which will affect the group mean.
more focused on weight loss compared But, you’ll also capture a fair number of
to a group less focused on weight loss. people who are not dieting, but simply
In experiment two, which consisted looking for ways to prevent weight gain
solely of frequent diet soda consumers, by controlling “empty calorie” intake.

36
APPLICATION AND TAKEAWAYS
It’s not totally clear whether or not diet sodas will protect you fully against hedonic
food cravings. However, at the very least, we can conclude that drinking diet soda
will at least mitigate the total amount of craving-induced eating that occurs after
exposure to an environmental food cue like an advertisement, the smell of cookies
when you walk by Subway, or your significant other whipping out a chocolate bar
while you’re on a diet.

You also might capture people who ar- ~20% eyeballing the means displayed in
en’t dieting or trying to prevent weight the bar graph of Figure 1) in the craving
gain, but trying to limit sugar based condition than the control among diet
on their beliefs about sugar’s effect on soda consumers in experiment one. If
health. We know a fair bit about the at- you are dieting, and you’re put in a situa-
titudes of the 60 diet soda drinkers in tion where you are exposed to images of
experiment one – indeed, at the group diet soda and given a bunch of non-di-
level, they are much more characteristic et friendly food alongside diet soda, you
of “dieters” than non consumers – but may very well “batten down the hatches”
we don’t have the same knowledge about and stick to your diet with more vigor.
the four groups of 43 diet soda drinkers In fact, it’s quite possible that because the
in experiment two. They didn’t differ in two groups were specifically asked ques-
any of the measured metrics (Table 2), tions about their beliefs regarding diet
but it’s possible (especially when bro- soda as a weight loss tool in experiment
ken up into multiple smaller sample size one, while they were not in experiment
groups) that the groups’ reasons were two, that the protective effect occurred,
not homogeneous for consuming diet as the questions acted as goal primers.
soda and some groups had more or few- To tie it all together, while it’s specu-
er “dieters” than others. lative on my part, it’s possible that diet
Based on prior research (6), the au- soda only acts to protect fully against
thors theorized that diet sodas can be cravings when you are actively on a diet.
helpful, not just by satisfying cravings, The diet soda is a reminder of your choice
but rather because the act of seeing or to drink diet soda because you are on a
drinking a diet soda serves as a reminder diet, reminding you of your goals, and
that you are indeed on a diet, and that serving to bolster motivation. Also, as
exposure reinforces your goals. In fact, an aside, among dieters who have a very
while not statistically significant, ener- “set” consistent pattern of eating and
gy intake was actually lower (by about drinking behavior, diet soda can be used

37
as a direct swap for calorie-containing
beverages, which, when all else remains
the same, effectively reduces calories.
However, in non-dieting conditions, it
may have a more minor effect, not pro-
tecting against cravings completely, but
serving to curb cravings and blunt total
energy intake in an ad libitum setting
(e.g. you might eat fewer total calories at
a party if there is diet soda available and
you seek it out).

Next Steps
I would like to see similar research con-
ducted in heterogeneous groups of diet
soda consumers, differentiated by their
reasoning for drinking diet soda. We
could compare individuals who choose
to drink diet soda for weight loss or
weight loss maintenance reasons against
individuals who consume diet soda for
other reasons and see if there was a dif-
ference in ad libitum energy intake in
various conditions. That would really
tell us what the primary mechanism is
by which diet sodas might impact im-
proved adherence. If only the weight
loss or weight loss maintenance group is
positively impacted, we can safely con-
clude the effect of diet soda is due to
being reminded of your goal. However,
if an effect is observed in diet soda con-
sumers unconcerned with body weight,
we could conclude there might also be
an effect related to it satisfying cravings.

38
References
1. Maloney NG, Christiansen P, Harrold JA, Halford JC, Hardman CA. Do low-calorie sweetened
beverages help to control food cravings? Two experimental studies. Physiology & behavior. 2019 Sep
1;208:112500.
2. Ruiz-Ojeda FJ, Plaza-Díaz J, Sáez-Lara MJ, Gil A. Effects of sweeteners on the gut microbio-
ta: a review of experimental studies and clinical trials. Advances in Nutrition. 2019 Jan 1;10(sup-
pl_1):S31-48.
3. Miller PE, Perez V. Low-calorie sweeteners and body weight and composition: a meta-analysis of
randomized controlled trials and prospective cohort studies. The American journal of clinical nutri-
tion. 2014 Jun 18;100(3):765-77.
4. Rogers PJ, Hogenkamp PS, De Graaf C, Higgs S, Lluch A, Ness AR, Penfold C, Perry R, Putz P,
Yeomans MR, Mela DJ. Does low-energy sweetener consumption affect energy intake and body
weight? A systematic review, including meta-analyses, of the evidence from human and animal stud-
ies. International Journal of Obesity. 2016 Mar;40(3):381.
5. Hall KD, Ayuketah A, Brychta R, Cai H, Cassimatis T, Chen KY, Chung ST, Costa E, Courville A,
Darcey V, Fletcher LA. Ultra-processed diets cause excess calorie intake and weight gain: an inpa-
tient randomized controlled trial of ad libitum food intake. Cell metabolism. 2019 May 16.
6. Papies EK, Hamstra P. Goal priming and eating behavior: Enhancing self-regulation by environ-
mental cues. Health Psychology. 2010 Jul;29(4):384.

39
Study Reviewed: Upper and Lower Thresholds of Fat-Free Mass Index in a Large
Cohort of Female Collegiate Athletes. Harty et al (2019)

Using Fat-Free Mass Index to


Forecast Long-Term Gains for
Males and Females
BY E RI C T RE X LE R

At some point in their lifting career, just about every lifter wonders, “Exactly
how big can I get?” While fat-free mass index (FFMI) isn’t going to tell you
the exact limit to your potential, or perfectly weed out steroid-users from
drug-free lifters, it’s still a helpful metric for approximating muscularity and
understanding how big most people can get. Read this article to figure out
how big male and female lifters tend to get, and what this means for your
natural potential for muscle growth.

40
KEY POINTS
1. Fat-free mass index (FFMI) has been used as a proxy for muscularity and has
previously been used to estimate upper limits for lean mass accretion in drug-
free lifters.
2. The current study (1) measured FFMI in a large sample of 372 female athletes.
The 97.5th percentile value was 23.9, but multiple athletes had values over 25,
and one athlete had a value over 27.
3. Current evidence suggests that it isn’t rare for males to have an FFMI as high as
28, or for females to have an FFMI as high as 24. Lifters can use FFMI to help plan
out their weight gain phases, but shouldn’t use it to arbitrarily limit themselves or
make baseless steroid accusations about others.

A
large percentage of lifters, at above 25. Based on this information, it
some point in their career, have was implied that values above 25 might
an interest in getting really, really be considered a “red flag” with regard to
huge. Inevitably, these lifters will reach a steroid use. In the online fitness world,
point in their training career where they the idea of a “natural limit” of 25 became
begin to wonder exactly how big they law, despite the very, very notable lim-
could get if they absolutely maximized itations of the Kouri study (for example,
their natural potential. People have pro- aiming to determine the upper limits of
posed several ways to estimate a lifter’s human potential based on 74 commer-
genetic limit for lean mass gains, using cial gym-goers).
tools such as a variety of online calcula- Unfortunately, the FFMI literature has
tors, multiple joint and bone measure- been male-dominated to date, and there
ments, the ratio of muscle to bone, and hasn’t been as much public discourse re-
fat-free mass index (FFMI). The beauty garding a proposed upper limit for female
of FFMI is that it is remarkably easy to muscularity. The current study sought
calculate, using very common measure- to characterize FFMI values in a wide
ments that are hard to mess up. range of sports, and to provide a prelim-
In 1995, Kouri et al (2) published inary estimate of the FFMI upper lim-
a study assessing FFMI in resis- it for female athletes. Results indicated
tance-trained males, both with and that FFMI differed between sports and
without a history of steroid use. The au- was positively associated with indices of
thors noticed that their drug-free lift- bone health. Most surprisingly, at least
ers all had values below 25 kg/m2, while three participants had values above 25,
many of the steroid users had values well with a maximal recorded value of 27.2!

41
This article will discuss what these num-
bers tell us about FFMI upper limits for Table 1 Body composition variables of
subjects in the current study (1)
both males and females, and how to use
FFMI to guide your next bulking phase. Mean ± SD

Height (cm) 167.55 ± 7.50

Purpose and Hypotheses Weight (kg) 69.46 ± 13.04

BF % 24.18 ± 5.48

Purpose
FM (kg) 17.50 ± 7.18
The authors stated that they had three
central purposes; to report sport-spe- FFM (kg) (w/ BMC) 52.93 ± 7.38

cific norms for FFMI in female ath- BMC (g) 2543.79 ± 364.77
letes, to determine if these values dif-
fered between sports, and to estimate BMD (g/cm3) 1.204 ± 0.096

an “upper limit” for FFMI in female SD = standard deviation; cm = centimeters; kg = kilograms;


athletes. BF% = body fat percentage (using DEXA);
FM = fat mass (using DEXA);
FFM = fat-free mass (using DEXA);
BMC = bone mineral content; g = grams;
Hypotheses BMD = bone mineral density

The authors hypothesized that FFMI


values would significantly differ be-
Design
tween sports, with particularly low val-
ues expected in sports that they con- This study featured a very simple
sidered to be “weight-sensitive.” Such design: recruit a big group of ath-
sports included cross country, gymnas- letes, measure body composition us-
tics, dance, swimming and diving, syn- ing dual-energy x-ray absorptiometry
chronized swimming, wrestling, and (DEXA), then crunch some num-
weightlifting. bers. The researchers followed typical
pre-visit guidelines for DEXA mea-
surements, but weren’t able to stan-
Subjects and Methods dardize menstrual cycle phase or com-
petitive season (in-competition or
Subjects off-season). These aren’t particularly
critical details, and they are extremely
This study sampled 372 female colle-
understandable, as studying collegiate
giate athletes, representing a wide range
athletes is extremely difficult from a
of sports. Table 1 shows the general de-
logistical perspective.
scriptive characteristics of the sample.

42
Table 2 Fat-free mass index values for each sport (1)

Sport N FFMI Range

Cross country 11 16.56 ± 1.14c,d 14.71 - 18.58

Gymnastics 35 18.62 ±1.12d 16.16 - 20.84

Dance 2 17.86 ± 1.17 17.03 - 18.68

Swim & dive 31 18.16 ± 1.67d 15.86 - 22.77

Synchronized swimming 29 17.27 ± 1.47c,d 15.07 - 20.68

Wrestling 19 19.15 ± 2.47a,b 15.04 - 24.45

Olympic weightlifting 15 19.69 ± 1.98a,b 17.19 - 23.84

22 18.98 ± 2.50a 15.27 - 24.38

Basketball 20 18.64 ± 1.87 15.96 - 22.37

Ice hockey 16 17.96 ± 1.04d 16.22 - 19.76

Lacrosse 40 18.58 ± 1.84d 15.94 - 26.35

Rugby 99 20.09 ± 2.23a,b 15.73 - 27.20

Volleyball 20 18.04 ± 1.13d 15.35 - 20.72

Water polo 13 18.35 ± 1.92 14.85 - 22.85

Total† 372 18.82 ± 2.08 14.71 - 27.20

All FFMIRaw data is presented as mean ± SD


a = significantly different from cross country (p < 0.05)
b = significantly different from synchronized swimming (p < 0.05)
c = significantly different from olympic weightlifting (p < 0.05)
d = significantly different from rugby (p < 0.05)
† = significant difference between sports (p < 0.001)

After the scans were done, FFMI was and assessing correlations between
calculated using the following equa- FFMI and markers of bone health.
tion:
Total lean mass (kg) + Total bone mineral content (kg)
A Note On Height Adjustments
(Height [m])2 If you check out the original paper,
you’ll see that they applied a couple of
For statistics, they were interested in different height adjustments to their
reporting “typical” FFMI ranges for FFMI values. One adjustment equa-
each sport, comparing between sports, tion was derived from a previous study

43
Figure 1 Fat-free mass index values for each sport

30

25
FFMI

20

15

10
Cross country

Gymnastics

Dance

Swimming & dive

Synchronized
swimming

Wrestling

weightlifting

Track

Basketball

Ice hockey

Lacrosse

Rugby

Volleyball

Water polo

Total combined
Olympic

Sport

FFMI values for each sport. The top and bottom of each box represent the maximum and minimum, while the middle bar represents
the mean value observed for that sport.

by Kouri et al (2), and the other was tweaking the exponent in the denom-
derived from a study my colleagues inator, with the “ideal” value probably
and I published in 2017 (3). The gen- falling somewhere around the 2.4-2.6
eral premise for height correction is range, but the most common approach
that FFMI, if not adjusted, tends to is to use regression-based mathemati-
be a bit biased, with higher values ob- cal adjustments.
served in taller individuals. Gains in In this study, the authors found vir-
body mass don’t perfectly scale with tually no relationship between height
height squared, and if we were to cube and FFMI. This is probably related to
height instead of squaring it in the the fact that not every athlete in the
FFMI equation, that would lead to an sample was necessarily trying to get as
over-correction. You could probably big as possible, as certain sports tend to
roughly account for this height bias by

44
Table 3 FFMI
sports combined

All Sports Weight-sensitive sports Other sports

Percentile FFMI (kg/m2) FFMI (kg/m2) FFMI (kg/m2)

5 15.96 15.71 16.21

10 16.38 16.11 16.95

15 16.92 16.35 17.24

20 17.20 16.78 17.51

30 17.63 17.19 17.83

40 18.10 17.63 18.41

50 18.52 18.14 18.72

60 18.93 18.52 19.23

70 19.55 18.94 19.91

80 20.26 19.65 20.71

90 21.52 20.65 22.16

self-select and reinforce specific height is a little easier on the eyes. Finally, Ta-
and body composition characteristics. ble 3 presents some percentile ranges
So, for this study, the authors (wisely) for all sports combined, in addition to
ditched the height-adjusted values. separate values for weight-sensitive and
non-weight-sensitive sports. The au-
thors reported that the 97.5th percentile
Findings value was 23.9, which was operationally
Table 2 shows the exact breakdown defined as the “upper threshold,” or the
of FFMI values by sport. That’s a large upper limit that most female athletes can
amount of numbers, so the same data are realistically shoot for. The lowest ob-
presented graphically in Figure 1, which served value was 14.7, and the highest

45
was 27.2. In addition, FFMI was posi-
tively correlated with both bone mineral
content (r = 0.53, p < 0.001) and bone THESE RESULTS SUGGEST
mineral density (r = 0.46, p < 0.001).
THAT IT’S NOT PARTICULARLY
ATYPICAL FOR FEMALES TO
Interpretation
I want to focus the majority of this dis-
ACHIEVE FFMI VALUES WELL INTO
cussion on the concept of using FFMI THE MID-20S, AND IF IT ONLY
to establish upper limits for musculari-
ty, but first I want to address the other TOOK A FEW HUNDRED PEOPLE
results of this paper. The authors found, TO OBSERVE A VALUE OVER
unsurprisingly, that FFMI differed
among sports. This isn’t shocking, but it’s 27, YOU CAN BE CERTAIN THAT
still helpful to actually quantify patterns
that we know to be true. I used to do
THERE ARE PLENTY OF DRUG-
a bunch of research on college athletes FREE FEMALES OUT THERE
that involved muscle ultrasound scans.
One day, a fellow researcher walked by WITH VALUES COMFORTABLY
and whispered, jokingly, “Do you real- HIGHER THAN THAT.
ly need an ultrasound to know if you’re
looking at an athlete?” Point taken, but
there’s value in building a quantifiable
profile for athletes on a sport-by-sport articles by Mike, Eric Helms, and I)
(or even position-by-position) basis, can lead to both performance and injury
as it allows us to understand the body issues in athletes, and this seems to be
composition metrics that appear to be particularly prevalent in female athletes
associated with high-level performance (4).
and injury reduction. We also know that disordered eating
Along those lines, the authors found is a common concern in female colle-
that FFMI was correlated with bone giate athletes, and researchers and prac-
mineral density. This is pretty notable titioners use extremely careful language
for female athletes in weight-sensitive when discussing body composition with
sports, in which bone (and other) inju- this population. This study is admittedly
ries are quite prevalent. We’ve known a bit dated, but researchers in the 1980s
for a while that low energy availability surveyed 42 female collegiate gymnasts,
(as discussed in three previous MASS and found that 28 of them were told by

46
you were above 25, you weren’t natural.
In 2017, I published a paper showing
FFMI CAN BE USEFUL FOR that FFMI values exceeding 25 were
not only possible, but pretty common,
PLANNING PURPOSES. in high-level American football players.
In fact, 31.3% of the Division I players
HOWEVER, IT’S REALLY exceeded 25, and we observed several
IMPORTANT TO REMEMBER individuals above 28. Some people were
pretty bothered by the finding, and sug-
WHAT FFMI IS NOT GOOD FOR: gested that the study was no more than
IMPOSING LIMITS ON YOURSELF evidence of rampant steroid use. I dis-
agreed then, and I still do. But now, as
AND MAKING BASELESS we’ll discuss, there’s even more evidence
to support my contention.
STEROID ACCUSATIONS.
As a reminder, the conclusions of the
Kouri paper (2) rested on shaky ground
all along. They sampled only 74 males
a coach that they were too heavy, and 21 from commercial gyms, and their inclu-
of these 28 were using at least one un- sion criteria only required that subjects
healthy weight-control behavior (5). So, had been lifting for at least two years. If
while these initial FFMI findings might you train at a standard commercial gym,
seem obvious, they give researchers and take a look around. Excluding the peo-
coaches a body composition metric, ple who have ever used steroids, select
which likely relates to both performance 74 guys. Do you think you’ve isolated
and injury risk, that isn’t fat-focused. As the upper and lower boundaries of hu-
a result, they can more readily discuss man potential for any measurable char-
objective, performance-oriented body acteristic?
composition goals with their athletes The other aspect of the Kouri paper in-
without dwelling on fat mass, which is volved estimating FFMI of Mr. America
pretty useful. winners from 1939-1959. They selected
this time range because they felt it was
The “maximum limit” for fat-free mass still plausible to believe that the compet-
index itors had not yet started using anabolics.
As I mentioned in the introduction, a I don’t feel comfortable using this data to
large segment of the online fitness world support or challenge any hypothesis, be-
has long viewed 25 as the maximal up- cause body composition was essentially
per limit for a natural, male lifter. So, if determined by educated guesses. None-

47
theless, there were several Mr. America in a recent study (6). Our study featured
winners that were comfortably above 25, a smaller sample size (266 versus 372),
and those physiques were built using the and a different selection of sports, so
limited training- and nutrition-relat- one wouldn’t expect our values to line
ed knowledge and resources that were up perfectly. However, we also reported
available over 60 years ago. a female with a value over 25, and sev-
Overall, there’s just never been strong eral individuals above 20. These results
evidence to suggest that a male can- suggest that it’s not particularly atypical
not exceed 25, but there was still some for females to achieve FFMI values well
pushback from reporting values above into the mid-20s, and if it only took a
25 in high-level male athletes. The pres- few hundred people to observe a value
ent study has reported values above 25 over 27, you can be certain that there
in females, so I would imagine there are are plenty of drug-free females out there
at least a few people losing their minds with values comfortably higher than
over it. that.

Upper thresholds for fat-free mass index Upper thresholds for fat-free mass index
in females in males
The current study recruited a sample As I mentioned previously, our study
that was likely to contain some pretty in football players found a large number
muscular females. They got a big group of males with an FFMI over 25 (3), and
of people (n = 372), and they were sure the 97.5th percentile cut-off was 28.1.
to include sports that reward strength The authors of the current study (1)
and power, such as wrestling, Olympic recently published a very similar study
weightlifting, and rugby. In fact, they in 209 male collegiate athletes (7), and
found 99 female collegiate rugby play- their results line up quite nicely with
ers, which is a surprisingly huge number our reported findings. Just by a visual
based on the sample sizes for all other estimate from their figures, it looks like
sports. The biggest rugby player clocked they had three male athletes at or above
in with an FFMI of 27.2, but high values ~29, and one of them was comfortably
weren’t restricted to the rugby players. A above 30. For their entire sample, which
lacrosse player had a value of 26.35, and included several sports that place min-
there were also individual values above imal value on attaining maximal levels
24 for the wrestling and track & field of lean mass (such as cross country, golf,
teams. These values might seem high, and swimming), the 97.5th percentile
but my colleagues and I reported simi- was 28.3. Finally, a 2018 study (8) sam-
lar data from female collegiate athletes pled 95 large male athletes competing in

48
American football, powerlifting, Sumo, to know exactly how many people can
or shot put, for the purpose of deter- achieve the remarkable FFMIs that have
mining upper limits for muscularity. The been reported in recent literature, but
results indicated that the average FFMI you could theoretically be one of them,
value was slightly over 25, there were and so could the random jacked person
numerous athletes with FFMIs well that everyone accuses of being on ste-
above 25, and a handful of athletes had roids.
values above 30. So, based on the three What you can use FFMI for, is to help
most recent studies in resistance-trained plan out your weight gain phases. You
males, it seems like you don’t have to look can estimate the FFMI of people who
extremely hard to find drug-free males are excelling in whatever you want to
in the high 20s. Frankly, this shouldn’t excel in, whether that’s bodybuilding,
be shocking; a 1999 study on 36 Sumo powerlifting, or some other athletic
wrestlers included two subjects with endeavor. You can also determine your
FFMIs above 36, and the authors found “ideal” offseason body-fat percentage
that an FFMI of >30 tended to separate (BF%) or the highest body-fat percent-
the elite Sumo wrestlers from their sub- age you’d be comfortable with at the
elite counterparts (9). peak of your weight gain phase. You can
then plug them into the equation below,
Using fat-free mass index which gives you a weight to shoot for at
Fat-free mass index can be really use- the given body-fat percentage that you
ful, but there are a couple of things it chose.
should not be used for. If you read our
FFMI x Height(m)2
study from back in 2017, you’ll see that
we aimed to report the 97.5th percentile
BF%
100

observed, rather than identify a univer- Based on the data currently available,
sal “maximum limit.” The value was not it seems that values substantially over 28
intended to be the maximum value pos- are probably a stretch for a lot of males,
sible, but rather an upper threshold that and values above 24 are likely a stretch
most people could feasibly hope to aim for a lot of females. However, if you have
for, because most people are, by defini- good genetics and tend to respond well
tion, not outliers. Along these lines, you to training, you may be able to aim high-
should not use FFMI to impose restric- er.
tive limits on your lifting goals, and you
definitely shouldn’t use FFMI to make Limitations of FFMI studies
baseless steroid accusations about oth-
Whenever you read a paper that in-
ers. We simply don’t have enough data
tends to establish normative values or

49
APPLICATION AND TAKEAWAYS
1. It doesn’t seem particularly unusual for males to achieve FFMI values as high as
28, and for females to achieve FFMI values as high as 24. For college athletes,
the highest observed values for males and females have been ~31 and ~27.
2. The body of FFMI literature is still small, so it would be premature to say we
have a firm understanding of the natural limits of muscularity. We really need
some huge studies that include genetically gifted, exceptionally well-trained
individuals to enhance our understanding.
3. You shouldn’t use FFMI to place restrictive, arbitrary limits on yourself or to
make baseless steroid accusations about others, but you can use it as a tool to
help plan your weight gain phases.

upper limits for FFMI, there are a few tions for the use of FFMI. As we’ve dis-
key considerations to keep in mind. It’s cussed, it’s slightly biased toward higher
certainly important to consider whether values in taller people. It fails to directly
or not steroid users were effectively ex- account for differences in the density of
cluded from participation. Many studies fat-free mass, which varies on the basis of
measure athletes that are subject to var- age, sex, race, and several other factors. It
ious drug-testing procedures, but it’s al- fails to directly account for the fact that,
ways important to realistically consider at least to some degree, lean mass and
how many drug users could have theo- fat mass are gained and lost in unison;
retically slipped through the cracks. The this is why you can find plenty of Sumo
second (and most important) question wrestlers with values well into the mid-
is, “Did this sample include any freaks?” 30s (9), but you’re unlikely to find much
And I can assure you, I mean “freak” in of that among a sample of contest-ready
the best possible way. If we want to make natural bodybuilders. Finally, as noted
inferences about upper limits for muscu- previously, there are several methods for
larity, we have to make sure the sample trying to estimate an individual’s genet-
is large enough and contains people who ic ceiling for muscle gain, and they each
are genetically gifted enough and well- have pros and cons. Fat-free mass index
trained enough to at least be close to is not a perfect proxy for muscularity, but
the top end of human potential. That’s it seems to generally get the job done.
a lofty goal, so most samples fall short, Despite some limitations, FFMI can
with some falling way shorter than oth- be useful for planning purposes. How-
ers. ever, it’s really important to remember
There are also plenty of inherent limita- what FFMI is not good for: imposing

50
limits on yourself and making baseless
steroid accusations.

Next Steps
Fat-free mass index is pretty straight-
forward, so complex study designs ar-
en’t really needed at this time. Instead,
we need more data using large data sets,
with samples that include people who
have been hitting the weights hard for a
long time, and (ideally) some genetically
gifted lifters. As more studies like this
come out, we should continue to devel-
op a better understanding of exactly how
big the typical male or female lifter can
realistically hope to get over the course
of their lifting career.

51
References
1. Harty PS, Zabriskie HA, Stecker RA, Currier BS, Moon JM, Jagim AR, et al. Upper and lower
thresholds of fat-free mass index in a large cohort of female collegiate athletes. J Sports Sci. 2019
Oct;37(20):2381–8.
2. Kouri EM, Pope HG, Katz DL, Oliva P. Fat-free mass index in users and nonusers of anabolic-an-
drogenic steroids. Clin J Sport Med Off J Can Acad Sport Med. 1995 Oct;5(4):223–8.
3. Trexler ET, Smith-Ryan AE, Blue MNM, Schumacher RM, Mayhew JL, Mann JB, et al. Fat-Free
Mass Index in NCAA Division I and II Collegiate American Football Players. J Strength Cond Res.
2017;31(10):2719–27.
4. Mountjoy M, Sundgot-Borgen JK, Burke LM, Ackerman KE, Blauwet C, Constantini N, et al. IOC
consensus statement on relative energy deficiency in sport (RED-S): 2018 update. Br J Sports Med.
2018 Jun 1;52(11):687–97.
5. Rosen LW, Hough DO. Pathogenic Weight-Control Behaviors of Female College Gymnasts. Phys
Sportsmed. 1988 Sep;16(9):140–4.
6. Blue MNM, Hirsch KR, Pihoker AA, Trexler ET, Smith-Ryan AE. Normative fat-free mass index
values for a diverse sample of collegiate female athletes. J Sports Sci. 2019 Aug;37(15):1741–5.
7. Currier BS, Harty PS, Zabriskie HA, Stecker RA, Moon JM, Jagim AR, et al. Fat-Free Mass Index
in a Diverse Sample of Male Collegiate Athletes. J Strength Cond Res. 2019 Jun;33(6):1474–9.
8. Abe T, Buckner SL, Dankel SJ, Jessee MB, Mattocks KT, Mouser JG, et al. Skeletal muscle mass in
human athletes: What is the upper limit? Am J Hum Biol Off J Hum Biol Counc. 2018;30(3):e23102.
9. Hattori K, Kondo M, Abe T, Tanaka S, Fukunaga T. Hierarchical differences in body composition of
professional Sumo wrestlers. Ann Hum Biol. 1999 Apr;26(2):179–84.

52
Study Reviewed: Anthropometrical Determinants of Deadlift Variant
Performance. Cholewa et al. (2019)

Can We Predict if You’re Better off


Deadlifting Conventional vs. Sumo?
BY G RE G NUC KO LS

There are a lot of recommendations floating around concerning whether


you should deadlift sumo or conventional based on how you’re built. In
spite of the numerous recommendations, there was no actual research on
the topic. That changed with the study reviewed in this article.

53
KEY POINTS
1. When subjects with prior training experience, but not deadlifting experience,
maxed on the sumo and conventional deadlifts, their 1RMs were virtually identical,
on average.
2. Their reps-to-failure performance with 60% of 1RM was virtually identical for
both deadlift styles as well.
3. While people with longer torsos relative to their total height were a bit better at
sumo deadlifts (and vice versa), the strength of that relationship was weak. No
other anthropometric variable was associated with performing particularly better
with one style of deadlift over the other.

F
or as long as I can remember, I’ve standing height (people with longer tor-
seen popular posts and articles sos were more likely to perform better
claiming that you should choose in the sumo deadlift). However, even
your deadlifting style based on various that was a very weak predictor. Thus, it
anthropometric factors like arm length, seems that simple anthropometric mea-
leg length, torso length, or just overall surements aren’t equipped to predict the
height. I wouldn’t be shocked if I wrote style of deadlift that lifters will initially
such an article at some point. Howev- be best suited to.
er, these recommendations have always
been based on a conspicuous lack of data
and are often contradictory. There’s a rea- Purpose and
son for that: Until now, no one had ever
studied the effects of anthropometrics on
Hypotheses
relative performance in the conventional
Purpose
and sumo deadlifts.
The purpose of the study was to see if
The presently reviewed study (1) mea-
any combination of body segment lengths
sured a slew of body segment lengths
could predict whether people would per-
and 1RM conventional and sumo dead-
form better in the sumo or conventional
lift strength in a sample of young sub-
deadlift.
jects who had prior training experience,
but no prior deadlifting experience. The
Hypotheses
only statistically significant predictor
of performing better in one style or the The authors hypothesized that sub-
other was the ratio of sitting height to jects with longer relative arm length and

54
shorter relative femur lengths would per- Table 1 Participants characteristics
form better in the conventional deadlift
Sex (n)
than the sumo deadlift. Male (28); Female (19)

Height (m) 1.73 ± 0.09

Subjects and Methods


Weight (kg) 77.6 ± 21.0

Age (years) 21.7 ± 3.0

Resistance training
Subjects experience (months)
11.0 ± 14.3

The subjects were 47 young lifters (28


males and 19 females) with some de- would be significant predictors), fol-
gree of training experience, who did not lowed by loaded sumo and conventional
deadlift as part of their training routine. deadlifts. After the subjects warmed up,
This population was chosen because they performed their deadlifts for sets of
the authors wanted subjects who could 2 reps while the researchers added 5-10
quickly learn both the sumo and conven- kg per set. They reported their estimated
tional deadlifts, but whose performance reps in reserve after each set, and the re-
in both lifts would not be influenced by searchers cut them off once they report-
prior experience with either lift. More ed four reps in reserve. This process was
details can be seen in Table 1. repeated for both deadlifting styles and
was used to give the researchers an idea
Experimental Design of the sorts of 1RM numbers to expect
Testing took place over four days. On from the subjects.
the first day, the researchers took an- During the third and fourth days
thropometric measurements and taught (which were both preceded by at least
the subjects how to correctly perform 48 hours without exercise), the subjects
unloaded conventional and sumo dead- worked up to a 1RM deadlift with ei-
lifts. The anthropometric measurements ther the sumo or conventional deadlift
taken were as follows: standing height, technique. The variant they performed
body mass, upper arm length, forearm on the third day was determined by flip-
length, wrist girth, hand length, seated ping a coin. After the 1RM was achieved
height, torso height, upper leg length, (defined as a technical max; testing was
lower leg length, and ankle girth. cut off when the lifters could no longer
During the second day, the subjects maintain a neutral spine), the subjects
performed counter-movement jumps rested for 5 minutes and performed one
for maximum height and maximum set to failure at 60% of their day 3 1RM
crunches in one minute first (to see if (they used the same load for days three
explosiveness or anterior core strength and four, even if the subjects recorded

55
Table 2 Overview of study protocol

Day 3 (at least 48 hours after Day 4 (at least 48 hours after
Day 1 Day 2
Day 2) Day 3)

Anthropometrics Loaded deadlift training Lift style randomly assigned Deadlift 1RM test

Deadlift repetitions to fatigue


Unloaded deadlift training Deadlift 1RM test
with 60% of Day 3’s 1RM

Deadlift repetitions to fatigue


with 60% 1RM

different 1RMs on both days). They did to their total height were more like-
all of the reps with a 1-2 second con- ly to perform better in the sumo than
centric and a 1-2 second eccentric tem- the conventional deadlift. Nothing else
po. The set was terminated when the was a significant predictor, and as you
subjects could no longer maintain their can see in Figure 1, sitting to standing
technique, or when they could no longer height ratio wasn’t a very strong predic-
maintain the required tempo. For both tor either.
the 1RMs and the sets to failure at 60%
1RM, straps were allowed so that the
subjects would not be limited by their Interpretation
grip. As with any study, this present study
had both strengths and weaknesses. The
study population wasn’t perfect, but it
Findings was probably the best population that
The subjects had similar sumo and con- one could feasibly recruit. Since they
ventional deadlift 1RMs (116.4 ± 36.5 did have some prior training experience,
for sumo, and 115.6 ± 35.5 for conven- they were likely able to attain pretty de-
tional). Reps to failure with 60% of the cent deadlifting technique with two fa-
day 3 1RM were also similar with both miliarization sessions (since the deadlift
variations (20.8 ± 6.4 reps for sumo, and isn’t a horribly complex lift), but their
20.7 ± 8.9 for conventional). lack of prior deadlifting experience
There was a weak, albeit significant, would mean that they weren’t way bet-
positive correlation between the subjects’ ter at one technique or the other simply
sitting height to standing height ratio due to the specificity of previous dead-
and sumo to conventional deadlift ratio lift training. A perfect population for
(r = 0.297; p = 0.043). In other words, this sort of study would be a cohort of
subjects who had longer torsos relative lifters who were highly trained in both

56
the sumo and the conventional dead-
lift, with similar amounts of experience Table 3 Performance results

with both variations, but that would be Sumo deadlift (kg) 116.4 ± 35.6
an incredibly hard population of people
to find and recruit for a study. However, Conventional deadlift (kg) 115.6 ± 35.5

it’s possible (likely, even) that the con- Sumo deadlift repetitions to
20.8 ± 6.4
ventional:sumo deadlift strength ratio fatigue

that someone displays the first time they


Conventional deadlift
try the lifts isn’t perfectly representative repetitions to fatigue
20.7 ± 8.9

of what their conventional:sumo dead-


lift strength ratio would be if they spent
a few years training both lifts. Thus, I people choose their deadlift style based
think this study was designed to answer on arm length wasn’t actually based on
a question like, “the first time someone any data, as far as I can tell. Thus, this is
tries deadlifting, which deadlift style the first and only study that has actual-
should they try first, based on their an- ly investigated whether anthropometrics
thropometry?” rather than, “what dead- influence whether someone will be rela-
lift style is going to be the strongest for tively better at the conventional or sumo
someone in the long run, based on their deadlift.
anthropometry?” In terms of the actual results, I feel
However, even with that limitation, pretty confident saying that this study
this study is a valuable addition to the found no meaningful predictors of rel-
literature. Since 2010, the recommen- ative performance between the deadlift-
dation has been floating around in the ing styles. The sitting:standing height ra-
literature (2) that people with longer tio correlation was weak, and as you can
arms should do conventional deadlifts, see from the actual figure, any prediction
and people with shorter arms should one tried to derive would be incredibly
do sumo deadlifts. I’ve seen quite a few tenuous (and that was the strongest cor-
people cite that paper as scientific evi- relation); it explained less than 9% of the
dence for the proposal that deadlift style variance in the sumo:conventional 1RM
should be based on arm length, but it’s ratio.
important to note that the 2010 paper Now, that doesn’t mean that nothing
was simply providing an overview of the predicts the style of deadlift one will be
deadlift and giving recommendations best-suited to. If I had to hazard a guess,
(some based on actual data, and some I wouldn’t be surprised if greater hip ab-
based on the author’s logic or anecdotal duction range of motion was a positive
experience); the recommendation that predictor of greater relative performance

57
Table 4 Correlations (r) between anthropometric predictors and the
sumo to conventional deadlift ratio

Predictor Pearson correlation p-value

Arm to height ratio 0.037 0.807

Arm to sitting height ratio -0.088 0.556

Thigh to height ratio -0.093 0.533

Lower leg to thigh ratio -0.015 0.920

Lower leg to height ratio -0.182 0.220

Sitting height to height ratio 0.297 0.043

Abdominal crunches 0.156 0.296

Counter movement jump 0.025 0.871

in the sumo deadlift, for example. Fur- both styles (your strongest technique
thermore, body mass seem to be a pre- may or may not be your most comfort-
dictor, at least among elite powerlifters, able technique). You can’t just take some
with more lightweight lifters of both body segment measurements and come
sexes favoring sumo deadlifts. Howev- away with a good prediction.
er, I think your best bet for determining I’d like to make clear that while anthro-
the deadlift style you’re best-suited to is pometrics don’t predict which style of
simply training both sumo and conven- deadlift someone will be best-suited to,
tional deadlift, and seeing which one im- anthropometrics should be at least some-
proves in strength the fastest. Or, if you what predictive of deadlift performance
don’t aim to compete in powerlifting, overall. Namely, long arms relative to to-
you could instead choose your primary tal height should predispose someone to
deadlift style based on comfort once you being a good deadlifter, simply because
get plenty of reps under your belt with long arms decrease your range of motion

58
Figure 1 Correlation between sitting to standing height ratio and sumo to conventional deadlift ratio

1.25
r = 0.297 p = 0.043
1.2
Ratio of sumo deadlift 1RM to

1.15
conventional deadlift 1RM

1.1

1.05

0.95

0.9

0.85

0.8
0.45 0.47 0.49 0.51 0.53 0.55 0.57

Ratio of sitting height to standing height

at the bottom of the lift, and most peo- pick as good of parents are generally
ple are weakest near the floor. However, pretty weak at both lifts, with anthro-
I would bet that even relative arm length pometrics playing a much smaller inde-
is a fairly weak predictor of overall dead- pendent role.
lift strength. Just as relatively long arms Finally, I’d just like to make it clear
should help the deadlift, they should that while average 1RMs were similar
hinder performance in the bench press. for both styles of deadlift, the spread of
However, among competitive powerlift- individual performances in favor of one
ers, there’s a moderately strong correla- style or the other was quite large. Most
tion (r ~ 0.70) between bench press and subjects performed pretty similarly in
deadlift performance – in other words, both deadlift styles (within about 5% or
even for lifts that are helped and hin- so), but one subject pulled almost 20%
dered by completely different anthro- more sumo than conventional, and sev-
pometric factors, people who are high- eral subjects pulled more than 10% more
ly trained and have a propensity to be conventional than sumo. MASS readers
strong are generally pretty good at both, should be savvy enough to know that
and people who are less trained or didn’t we can’t always (or can’t generally) map

59
APPLICATION AND TAKEAWAYS
For the time being, I still don’t think there are any incredibly strong predictors of
which deadlift style you’re best suited to. Your best bet is to spend time training
both, and letting your performance and comfort tell you which style suits you best.

group averages onto individuals, but


since this study showed each individual
data point, it allows for a nice illustra-
tion of this crucial point.

Next Steps
The ideal study would be the one I pro-
posed at the start of the interpretation
section: Run the same study on a pop-
ulation of people who train both styles
of deadlift equally. In that study, more
measures could be added, such as hip
abduction range of motion. Single-joint
strength measures would also add to
the study, to see if relative joint-level
strength is predictive of performing bet-
ter with one deadlift style over the other
(i.e. if people with stronger quads rela-
tive to their hip extensors are relative-
ly better at sumo deadlifts, or if people
with stronger spinal erectors relative to
their hip extensors are relatively better at
conventional deadlifts).

60
References
1. Cholewa JM, Atalag O, Zinchenko A, Johnson K, Henselmans M. Anthropometrical Determinants
of Deadlift Variant Performance. J Sports Sci Med. 2019 Aug 1;18(3):448-453.
2. Hales M. Improving the Deadlift: Understanding Biomechanical Constraints and Physiological
Adaptations to Resistance Exercise. Strength Cond Jour. 2010 Aug;32(4):44-51.

61
Study Reviewed: Cooling During Exercise Enhances
Performances, but the Cooled Areas Matter. A Systematic
Review with Meta-Analyses. Douzi et al. (2019)

Be Cool … Literally
BY MIC HAE L C . ZO URD O S

Sitting in a vest with crushed ice and drinking a below-0 degree Celsius
beverage has been investigated as a method to improve exercise
performance. Does it actually work? A new meta-analysis provides us
with answers.

62
KEY POINTS
1. This meta-analysis examined if cooling the body during both aerobic and anaerobic
exercise can significantly increase performance.
2. The results showed a moderate effect for cooling to improve aerobic exercise
performance and a small effect for cooling to improve anaerobic performance;
however, none of the anaerobic studies were on resistance training.
3. Importantly, the results showed that for cooling to be effective for anaerobic
performance, the ambient temperature must already be hot. Further, while this
meta-analysis did not include resistance training studies, the resistance training
data that does exist is promising.

I
f you see someone wearing a cool- dicate a small but significant benefit for
ing vest or chugging ice-cold water per-cooling (effect size = 0.27) versus
before or during training, they may no cooling for anaerobic performance.
not be setting a new fashion trend or be This meta-analysis also examined which
crazy thirsty. Rather, this person may be cooling techniques were most effective,
purposefully cooling their body. Cool- what areas of the body should be cooled,
ing the body before exercise (pre-cool- sex differences, training status, and what
ing) or during exercise (per-cooling) has magnitude cooling should be used. This
been shown to improve performance, article brings about a new topic for
especially when ambient temperatures MASS; thus, we will cover each of the
are high (2). However, data have shown above points to provide a practical guide
that endurance exercise performance has for how to use per-cooling.
been improved with body-cooling even
when ambient temperatures were fairly
normal (25ºC) (3). This article exam- Purpose and Research
ines a meta-analysis (1) that evaluated
the effects of per-cooling on both aero-
Questions
bic and anaerobic exercise performance.
Purpose
The results indicated that per-cooling
had a significant (p<0.001) and moder- The purpose of this meta-analysis was
ate effect (effect size = 0.60) to improve to examine the effects of per-cooling on
performance versus no cooling for aero- both aerobic and anaerobic exercise per-
bic performance across 36 studies. Only formance. The meta-analysis also eval-
9 studies were included for anaerobic uated different methods of cooling and
performance; however, the data did in- what areas of the body should be cooled.

63
Table 1 Systematic review inclusion criteria

Study applied cooling during exercise and described the cooling technique and
1
application site of cooling

The cooling technique had to reduce body temperature (i.e. a cooling vest or cooling
2
drink) or cause a cooling sensation (i.e. water spray)

Study was a crossover design with one condition being per-cooling and the other being
4
no cooling

Meta-analysis inclusion criteria from Douzi et al. 2019 (1)

Hypotheses per-cooling studies.


No hypothesis was given, as is custom- In addition to just determining if
ary in a meta-analysis. per-cooling was effective for both aero-
bic and anaerobic performance, the au-
thors of the meta-analysis also examined
Subjects and Methods the following related to per-cooling: 1)
sex differences, 2) influence of training
Study Inclusion status, 3) aerobic versus anaerobic ex-
As I’m sure everyone knows, a me- ercise, 4) type of cooling methods, 5)
ta-analysis does not have its own sub- most effective site of cooling, 6) com-
jects; rather, it analyzes data from all paring cooling in both hot (>28ºC) and
of the independent studies that fit its temperate (22-28ºC) ambient condi-
criteria to determine the overarching tions.
findings on a topic. To be included in A total of 45 studies were included. 36
this meta-analysis, the individual stud- studies examined aerobic exercise, and
ies had to meet the criteria listed in Ta- 9 examined anaerobic exercise perfor-
ble 1. mance. Studies were classified as aero-
Studies that did not satisfy all crite- bic if the duration of the performance
ria in Table 1 were excluded. For exam- task was ≥ 76 seconds, and studies were
ple, if a study was a crossover design, anaerobic if the performance task was
but used pre-cooling, it was excluded <76 seconds.
since the meta-analysis only looked at

64
Table 2

3
to be the best methods of per-cooling.

4 Decreasing skin temperature about 2ºC is preferable.

2
already high.

4 Decreasing skin temperature about 2.5ºC is preferable.

Findings high (33.4 ± 3.4ºC, effect size = 0.57)


and temperate (23.4 ± 2.4ºC, effect size
A summary of the main findings is in = 0.86), although the finding for tem-
Table 2. perate conditions is only based on three
studies. Cooling parts of the body (i.e.
Aerobic Exercise head and neck) and fluid ingestion (i.e.
Overall, per-cooling significantly im- ice slurry) aided performance; however,
proved aerobic exercise performance cooling the entire body, surprisingly, did
(effect size = 0.60) with no difference not significantly augment performance
between the sexes or for training sta- versus no cooling. The average decreases
tus for per-cooling benefits (p > 0.05). in skin temperature and core tempera-
Importantly, per-cooling was effective ture were 2.1 ± 0.4ºC and 1.0 ± 0.7ºC,
when the ambient temperature was both respectively. The exercise duration of

65
the 36 aerobic studies ranged from 10
minutes to 90 minutes (average = 45.0 ± Interpretation
25.3 min). I was excited when seeing this title
since this is a new topic for MASS. The
Anaerobic Exercise quick rehash of the findings is that cool-
Per-cooling had a small but significant ing the body during exercise is helpful
benefit for anaerobic performance (effect for both aerobic and anaerobic perfor-
size = 0.27) without any differences be- mance. However, there is much more to
tween the sexes or for people of different discuss, including: the method and area
training status. Dissimilar from aerobic of cooling, how long and how much to
exercise, only whole-body cooling gar- cool the body, and – of course – what
ments showed a significant performance mechanism is driving performance ben-
improvement, and cooling only helped efits. Since this is a new topic for MASS,
anaerobic performance in hot condi- we will keep the interpretation simple
tions (>28ºC). The average decreases in and clarify each point raised above so
skin temperature and core temperature that you have an idea of how to actually
were 2.6 ± 1.1ºC and 1.3 ± 1.0ºC, re- implement this. Further, we will discuss
spectively. The exercise duration of the 9 the studies on lifting that exist but were
anaerobic studies ranged from 5 seconds not included in this meta-analysis.
to 69 seconds (average = 21.3 ± 21.6 sec-
onds). Importantly, no anaerobic studies What Methods are Used to Cool?
included were on resistance training; We can divide cooling methods into
rather, these studies involved mostly two main types. The first is external cool-
intermittent sprint exercise. However, ing, which is typically accomplished by
studies examining per-cooling’s effects wearing a cooling vest or garment, but
on resistance training do exist, and we could also occur using air ventilation or
will elaborate on that data in this arti- water spray. There’s also internal cooling,
cle’s interpretation. which is ingesting a very cold beverage
(-1ºC to 4ºC), usually an ice slurry or
Aerobic vs. Anaerobic ice-cold water. When a beverage is used,
As you might expect after reading the it is typically ingested ~6.5oz at a time
above, the authors reported a statistically intermittently throughout long duration
greater benefit of per-cooling for aerobic exercise. Indeed, Riera et al (4) found an
exercise compared to anaerobic exercise ice slurry to improve performance versus
(p<0.01). no cooling when ~6.5oz were ingested
every 5km during a 20km cycling time
trial. Although the ice slurry is used in-
termittently, the external cooling meth-

66
ods are usually applied for the totality of with per-cooling. However, anaerobic
aerobic exercise. exercise performance only benefited
For anaerobic exercise, no study to date from per-cooling in hot temperatures
has used cold beverage consumption. (>28ºC). I do think there is some ap-
For the external strategies, the cooling plication to lifting since it’s not too un-
vests are typically filled with crushed common to train in hot conditions. Any
ice, while cooling garments can be filled other non-air conditioned garage gym
with cold water and then sealed closed. lifters in hot climates out there? Not
Oftentimes, research will use garments to mention, not all gyms have air con-
that only cover one portion of the body. ditioning, and if you train at a CrossFit
For example, the current meta-analysis gym, then you likely train in the heat too
found that cooled neck collars were ef- and might consider per-cooling.
fective at improving aerobic exercise per-
formance. For anaerobic performance, The Lifting Data
although there was a significant benefit Even though no resistance training
for cooling, it is important to note that studies were included in this meta-anal-
only whole-body cooling garments im- ysis, lifting studies do exist. In fact, a
proved performance. The anaerobic re- recent systematic review from Latella
sults are only based on nine studies, so et al (7) included six per-cooling resis-
it is possible that other methods or sites tance training studies and noted that
of application are effective and we just four of those studies showed enhanced
don’t know yet. In most studies, the ex- resistance training performance of some
ternal cooling garments have been used kind when cooling the body between
both during recovery periods only (5) or sets. Specifically, Bacon et al (8) found
during the totality of exercise (6). Fur- that rock climbers performed more reps
ther, this meta-analysis (1) reported an across multiple sets of pullups when ice
average decrease in skin temperature of bags were put on the upper body between
2.6ºC and decrease in core temperature sets; however, in this study, the rest pe-
of 1.3ºC across all anaerobic studies, riods were 17 minutes, which is incon-
so as long as you follow the guidelines sistent with real-world training. Galo-
above, then you can likely expect about za et al (9) observed men with at least
those changes in skin temperature. one year of training experience perform
~21% more reps over 4 sets of preach-
Hot vs. Temperate Ambient Conditions er curls at 70% of 1RM when ice bags
According to the meta-analysis, aero- were put on the subjects’ biceps during
bic exercise performance was improved one-minute interset rest intervals. How-
in both hot and temperate conditions ever, the Galoza study was not a cross-

67
over design, as two groups (cooling and from Grahn et al (13), who also used
control) were used; this is why it couldn’t palm cooling. Grahn et al found acute
be included in the present meta-analy- bench press reps to failure to be im-
sis. Interestingly, multiple studies (10, proved with interset palm cooling versus
11, 12, 13) have shown cooling only the no cooling when core temperature was
palms to improve resistance exercise elevated to 39ºC (13). Further, in tem-
performance, which is inconsistent with perate conditions, Grahn et al conduct-
the meta-analysis conclusions that cool- ed multiple longitudinal protocols and
ing the whole body is the best way to go found that bench press volume capabil-
to improve anaerobic performance. Spe- ities were increased by 40% after three
cifically, Kwon et al 2010 (10) reported weeks and pull-up volume was increased
that trained men performed more reps by 144% over two weeks compared to
over four sets of bench press at 85% of no cooling when palm cooling was ap-
1RM when cooling the palms between plied during three-minute interset rest
sets versus no cooling in a crossover intervals during twice a week training
design. Five years later, Kwon and col- in both protocols. The Grahn study also
leagues completed a similar study in conducted a longitudinal trial in which
women and found palm cooling to lead 10 men benched 2 times per week for
to greater reps performed on the bench 10 weeks with no cooling between sets
press at 85% of 1RM when compared for the first 4 weeks and palm cooling
to heating; however, no control con- between sets for the final 6 weeks. Im-
dition was used in the study on wom- portantly, loads in this 10-week protocol
en (11). In the Kwan studies, the palm were increased based on successful com-
was cooled during three-minute rest pletion reps in the previous week. Grahn
intervals using a hand cooling device reported no difference in the load used
that was set to 10ºC, and the ambient between the first and last training ses-
conditions were temperate. Similar to sions of the 4 weeks with no cooling (97
the Kwon studies, Caruso et al (12) also ± 25 kg), but significantly greater load
found palm cooling (to 15ºC) to cause used in the last session of the 6 weeks of
greater maintenance of power output cooling (110 ± 27 kg) and a 22% increase
over four sets of leg press compared to in bench press 1RM after the 10 weeks.
no cooling in temperate conditions. The Therefore, the authors attributed most
data from Caruso represents the first (if not all) of the increase in bench press
study to show that cooling of the upper 1RM to cooling. Obviously, you can get
extremities can improve performance stronger without palm cooling, so the
on solely lower body exercise. The most longitudinal results from Grahn seem
impressive results from per-cooling on a bit of a stretch. When considering all
resistance training performance come of the data specifically related to lifting,

68
the results do seem pretty promising. increased central nervous system fatigue
To summarize, the data pretty consis- and resultant drops in motor output in
tently show cooling to improve volume the 40-degree condition. Interesting-
capabilities, cooling has worked in both ly, Siegel et al (16) found pre-training
hot and temperate conditions for lift- ice slurry ingestion to lead to improved
ing, palm cooling is the most common strength on a two-minute isometric bi-
method, ice bags on the exercising mus- ceps test versus no cooling. However,
cle have also been effective, and cooling the main research question in Siegel’s
during the interset period seems to be study was to examine body-cooling on
the way to go. running performance, thus a ~40-min-
ute run was done prior to the biceps
Mechanisms and Resistance Training test. These studies, along with the
Possibilities Grahn et al (13) study discussed above
There are three suggested mecha- suggest that when core temperature can
nisms for body-cooling: 1) thermoreg- be lowered, resistance training perfor-
ulatory, 2) central, and 3) psychophysi- mance may increase, which is consis-
ological (2). Thermoregulatory changes tent with the proposed central mecha-
with body-cooling manifest as reduc- nism of per-cooling.
tions in skin temperature, which have A less-supported mechanism behind
been theorized to reduce cardiovascular body cooling is the potential psycho-
strain (14). Therefore, the reduction in physiological component, which is
cardiovascular strain makes it logical to say that body cooling could affect
for much of the cooling data to be on the interaction between the mind and
aerobic exercise. body. While a specific psychophysi-
I think the central mechanism is most ological mechanism has not been di-
likely to explain the mostly positive re- rectly shown, it is common for studies
sults discussed in the section above for that see performance improvements
lifting. During lifting, core temperature to also have subjects report increased
can rise quickly, and when core tem- thermal sensation and satisfaction and
perature is too high (>40ºC), areas of decreased rating of perceived exertion
the brain that drive motor activation during and following cooling (2).
can be inhibited (15). Indeed, Nybo et Lastly, I understand this sounds im-
al (15) observed that during prolonged practical, but I really don’t think it is.
isometric knee extensions (>30 sec- You can purchase a cooling vest for as
onds), strength was greater with a core little as $30 and outfit it with ice inside
temperature of 38ºC versus 40ºC, and or just bring some ice packs to the gym.
the authors attributed the difference to Most of us spend $30 per month on

69
APPLICATION AND TAKEAWAYS
1. This meta-analysis shows that per-cooling can augment both aerobic and anaerobic
performance.
2. For anaerobic performance, the results of the meta-analysis show that the entire
body should be cooled; however, palm cooling is the most beneficial method of
application in the available resistance training studies.
3. Core temperature can rise pretty high during resistance training, which is likely
why the available per-cooling studies on resistance training show mostly positive
results. Practically, cooling is not too hard to do, so it might be worth a shot.

supplements, and this vest would be a


one-time purchase, and ice packs would Next Steps
be even cheaper. It’s hard to outright I would like to see if the enormous ben-
recommend, but I don’t think cooling efits for cooling seen in the Grahn study
is that hard to try, and if we are looking can be replicated. A pretty straightfor-
for any edge, then body-cooling could ward design with two groups training in
be worth a shot. Although, not related temperate conditions could help. One
to per-cooling specifically, cooling the group would use palm cooling between
body consistently post-exercise (twice sets during every session, and the other
a week for 12 weeks) with cold-water group would use no cooling. If this lon-
immersion has been shown to attenuate gitudinal study does show a benefit, then
anabolic signaling and potentially blunt a follow-up to that would be to compare
the satellite cell response to training different methods of cooling.
(17), which we have written about pre-
viously. While we don’t know for sure
if this happens with consistent pre- or
per-cooling, it is necessary to note that
this has occurred over the long-term
with post-exercise cooling. In short,
there is little to no risk with cooling the
body once or every so often, but if cool-
ing is used consistently, then you could
see diminished anabolic signaling.

70
References
1. Douzi W, Dugué B, Vinches L, Al Sayed C, Halle S, Bosquet L, Dupuy O. Cooling during exercise
enhances performances, but the cooled body areas matter: A systematic review with meta‐analyses.
Scandinavian journal of medicine & science in sports. 2019 Jul 24.
2. Stevens CJ, Taylor L, Dascombe BJ. Cooling during exercise: an overlooked strategy for enhancing
endurance performance in the heat. Sports Medicine. 2017 May 1;47(5):829-41.
3. Eijsvogels TM, Bongers CC, Veltmeijer MT, Moen MH, Hopman M. Cooling during exercise in
temperate conditions: impact on performance and thermoregulation. International journal of sports
medicine. 2014 Sep;35(10):840-6.
4. Riera F, Trong TT, Sinnapah S, Hue O. Physical and perceptual cooling with beverages to increase
cycle performance in a tropical climate. PLoS One. 2014 Aug 1;9(8):e103718.
5. Duffield R, Dawson B, Bishop D, Fitzsimons M, Lawrence S. Effect of wearing an ice cooling jacket
on repeat sprint performance in warm/humid conditions. British journal of sports medicine. 2003
Apr 1;37(2):164-9.
6. Tyler CJ, Sunderland C. Cooling the neck region during exercise in the heat. Journal of athletic
training. 2011 Jan;46(1):61-8.
7. Latella C, Grgic J, Van der Westhuizen D. Effect of Interset Strategies on Acute Resistance Training
Performance and Physiological Responses: A Systematic Review. The Journal of Strength & Condi-
tioning Research. 2019 Jul 1;33:S180-93.
8. Bacon NT, Wingo JE, Richardson MT, Ryan GA, Pangallo TC, Bishop PA. Effect of two recov-
ery methods on repeated closed-handed and open-handed weight-assisted pull-ups. The Journal of
Strength & Conditioning Research. 2012 May 1;26(5):1348-52.
9. Galoza P, Sampaio-Jorge F, Machado M, Fonseca R, Silva PA. Resistance exercise inter-set cooling
strategy: effect on performance and muscle damage. International journal of sports physiology and
performance. 2011 Dec 1;6(4):580-4.
10. Kwon YS, Robergs RA, Kravitz LR, Gurney BA, Mermier CM, Schneider SM. Palm cooling delays
fatigue during high-intensity bench press exercise. Medicine & Science in Sports & Exercise. 2010
Aug 1;42(8):1557-65.
11. Kwon YS, Robergs RA, Mermier CM, Schneider SM, Gurney AB. Palm cooling and heating delays
fatigue during resistance exercise in women. The Journal of Strength & Conditioning Research. 2015
Aug 1;29(8):2261-9.
12. Caruso JF, Barbosa A, Erickson L, Edwards R, Perry R, Learmonth L, Potter WT. Intermittent
palm cooling’s impact on resistive exercise performance. International journal of sports medicine.
2015 Oct;94(10):814-21.
13. Grahn DA, Cao VH, Nguyen CM, Liu MT, Heller HC. Work volume and strength training re-
sponses to resistive exercise improve with periodic heat extraction from the palm. The Journal of
Strength & Conditioning Research. 2012 Sep 1;26(9):2558-69.
14. Lee DT, Haymes EM. Exercise duration and thermoregulatory responses after whole body precool-
ing. Journal of applied physiology. 1995 Dec 1;79(6):1971-6.

71
15. Nybo L, Nielsen B. Hyperthermia and central fatigue during prolonged exercise in humans. Journal
of applied physiology. 2001 Sep 1;91(3):1055-60.
16. Siegel R, Maté J, Watson G, Nosaka K, Laursen PB. Pre-cooling with ice slurry ingestion leads to
similar run times to exhaustion in the heat as cold water immersion. Journal of sports sciences. 2012
Jan 1;30(2):155-65.
17. Roberts LA, Raastad T, Markworth JF, Figueiredo VC, Egner IM, Shield A, Cameron‐Smith D,
Coombes JS, Peake JM. Post‐exercise cold water immersion attenuates acute anabolic signalling
and long‐term adaptations in muscle to strength training. The Journal of physiology. 2015 Sep
15;593(18):4285-301.

72
Study Reviewed: No Longer Beeting Around the Bush: A Review of Potential Sex
Differences with Dietary Nitrate Supplementation. Wickham and Spriet. (2019)

Do the Performance Effects of


Beetroot Juice and Dietary Nitrate
Differ Between Males and Females?
BY E RI C T RE X LE R

Supplements that promote nitric oxide are in just about every pre-workout
supplement on the market. Nitrate, a primary component of beetroot juice,
is a notable nitric oxide promoter that is also found in a variety of fruits and
vegetables. Several studies have reported performance benefits of dietary
nitrate for males, but there’s a considerable lack of research in females. Read
this article to find out if nitrate actually improves performance, and if this effect
is similar for both males and females.

73
KEY POINTS
1. The body of research on dietary nitrate and beetroot juice supplementation is
growing, but there’s a glaring lack of research on female participants.
2. The current review article (1) evaluated potential sex differences and found that
nitrate generally has less pronounced effects on blood pressure and exercise
economy in females than males.
3. Nitrate has been shown to enhance strength and power, and its direct effects
on muscle contractile function are the most plausible mechanisms for these
effects. At this time, there is insufficient evidence to conclude that males enjoy
larger strength or power improvements than females. However, it is a plausible
possibility, and we need substantially more research in this area.

N
itric oxide has an interesting the nitrate/beetroot literature focuses
history. It was discovered in the on endurance exercise rather than re-
1700s, but its role in physiology sistance exercise, so its applications for
remained unknown for a couple hun- strength athletes are poorly understood.
dred years. Then, in the 1980s, a series The literature has another glaring flaw:
of experiments showed that nitric oxide very, very few studies with female par-
played a critically important role in car- ticipants. The current review paper (1)
diovascular physiology by inducing re- evaluated the existing literature to ex-
laxation of blood vessels (2). In the years plore sex-based differences related to
since, it was named “Molecule of the nitrate supplementation. The authors
Year” by the prestigious journal Science, suggest that, compared to men, women
earned a Nobel Prize for the scientists have higher baseline plasma nitrite and
who discovered its role in the body, and experience larger increases in plasma
now we find some kind of nitric oxide nitrate, but experience smaller reduc-
precursor in just about every pre-work- tions in blood pressure and less consis-
out supplement on the market. tent improvements in exercise economy
As the supplement research has following nitrate supplementation.
evolved over the last few decades, it has So, does this mean that males will en-
become clear that dietary nitrate, often joy performance improvements from
consumed as beetroot juice, is one of nitrate supplementation, and females
the more promising nitric oxide-pro- will miss out? Read on to find out.
moting ingredients out there. Unfor-
tunately, the overwhelming majority of

74
tion somewhat differently. Finally, the
authors noted some areas requiring fur-
PEOPLE OFTEN THINK OF NITRIC ther research in the future.

OXIDE PURELY AS A BLOOD FLOW


PROMOTER, AND THINK THAT Subjects and Methods
As a narrative review, there aren’t re-
ENHANCING “THE PUMP” IS ITS ally any methods to highlight or scruti-
nize. As opposed to a systematic review
ONLY POTENTIAL CONTRIBUTION or meta-analysis, authors for this type
TO EXERCISE APPLICATIONS. IN of paper are free to highlight the litera-
ture that they deem most relevant to the
REALITY, NITRIC OXIDE EXERTS topic. From my perspective, they did an
A WIDE RANGE OF EFFECTS excellent job covering the breadth of the
literature available and delivered the in-
THROUGHOUT THE BODY. formation in an unbiased manner, with
no evidence of cherry-picking.

Purpose and Hypotheses As a review, they didn’t recruit any


This paper is a narrative review, in subjects of their own, but they were in-
which authors simply review the studies terested in assessing the overall repre-
that are currently available on the topic. sentation of males and females within
As a result, the paper is more descriptive the entire body of research. They found
than hypothesis-driven, and hypotheses more than 100 studies using male-on-
aren’t always directly stated in this type ly samples, compared to only 7 studies
of work. The purpose was quite clear: with only females.
the authors set out to review the existing
literature documenting male and female
responses to beetroot juice and dietary Findings and
nitrate supplementation. The authors Interpretation
also highlighted physiological differenc-
Before we get into sex-based differenc-
es between males and females that could
es, let’s establish a firm understanding of
theoretically cause differential respons-
how nitrate is supposed to work. Nitric
es to nitrate, which leads one to believe
oxide is a gas and has an extremely short
that they likely suspected that males and
half-life, so directly supplementing with
females may respond to supplementa-
nitric oxide is off the table. Instead, we

75
Figure 1 The effects of nitric oxide on exercise-related outcomes (3)

Dietary L-Citrulline
NO3-
L-Arginine
Bacterial
anaerobes

NO2-
Acidosis
NO
hypoxia

Exercise Mitochondrial Ca2+ Vasodilation Glucose Muscle


efficiency respiration handling uptake fatigue

Dietary nitrate (NO3-) is converted to nitrite (NO2-), which is then converted to nitric oxide (NO). Nitric oxide then exerts
a wide range of effects throughout the body, many of which have the potential to influence exercise performance.

tend to supplement with precursors to the oral cavity. That’s an important (but
nitric oxide; as a result, when nitric oxide sometimes overlooked) detail; this bac-
production is naturally increased during terial conversion is 100% necessary for
exercise, we can be confident that pro- nitrate to work, so antibacterial mouth-
duction won’t be impaired by a lack of wash absolutely eliminates all potential
precursor availability. You can form ni- effects of dietary nitrate intake. Nitrite is
tric oxide from arginine (and, indirectly, then converted to nitric oxide in a pro-
from citrulline), but we’ll restrict our fo- cess that is stimulated by conditions of
cus to the nitric oxide pathway that in- acute hypoxia and acidosis, as we often
volves nitrate. observe during high-intensity exercise.
When you consume nitrate, it is con- People often think of nitric oxide purely
verted to nitrite by bacteria present in as a blood flow promoter, and think that
enhancing “the pump” is its only poten-

76
Table 1 Summary of key anthropometric, physiological, and metabolic differ-
ences between males and females (1)

Males Females

Larger in stature (Wells 2007) Smaller in stature (Wells 2007)

Greater total mass Lower total mass (Wells 2007)

Greater relative lean mass Greater relative fat mass


Anthropometrics (Wells 2007) (Wells 2007)

Greater proportion of lean


Greater proportion of lean
mass in the shoulders, chest,
mass in the thighs
lower gluteal region
(Abe et al. 2003)
(Abe et al. 2003)

Greater hemoglobin levels Lower hemoglobin levels

[Hemoglobin] and red blood Greater red blood cell mass Lower red blood cell mass
cell mass
Greater oxygen carrying Lower oxygen carrying
capacity (Murphy 2014) capacity (Murphy 2014)

Greater relative proportion of Greater relative proportion of


type II glycolytic skeletal type I oxidative skeletal

(Haizlip et al. 2015) (Haizlip et al. 2015)

Lower skeletal muscle Greater skeletal muscle


capillarization capillarization

Greater intramuscular
Greater glycolytic enzyme
Fuel stores and utilization triglyceride stores
activity (Green et al. 1984)
(Tarnopolsky et al. 2007)

Greater reliance on Greater capacity to mobilize


carbohydrate during intramuscular triglyceride
submaximal exercise stores (Tarnopolsky et al.
(Tarnopolsky 2000) 2007)

Greater reliance on fat during


submaximal exercise
(Tarnopolsky 2000)

tial contribution to exercise applications. Notable physiological differences between


In reality, nitric oxide exerts a wide range males and females
of effects throughout the body. Most Now, let’s briefly consider some rele-
notably, there is some evidence to sug- vant characteristics that tend to differ
gest that nitric oxide reduces the energy between males and females. As summa-
cost of exercise, enhances the contrac- rized in Table 1, males tend to be taller
tile function of muscle, increases cellular and heavier, with a relatively higher pro-
glucose uptake, and attenuates muscle portion of lean mass to fat mass. Males
fatigue, in addition to increasing blood also tend to have more red blood cells
flow. These effects are summarized in and hemoglobin, and lower capillary
Figure 1. density of skeletal muscle, in addition
to a greater proportion of type 2 mus-

77
cle fibers. As such, it’s relatively unsur- Figure 2 The most likely mechanisms by
prising that males tend to rely compara- which nitrate may increase strength and power (6)
tively more on carbohydrate metabolism
during exercise, whereas females tend
to utilize more fat during exercise. In
addition to these fairly well-known sex NO3-
differences, the authors of the current
review noted that females tend to have
higher baseline plasma nitrite levels, de-
spite having similar nitrate levels, which NO2-
may indicate that females convert nitrate
to nitrite more efficiently than males (1).
In summary, nitrate is thought to exert
effects related to some combination of NO
blood flow, oxygen delivery, mitochon-
drial function, and/or muscle function,
and there are notable sex-based differ-
ences that could potentially influence
each of these outcomes. Now, let’s take RyR sGC activity
nitrosylation
a look at what the researchers found as
they reviewed the evidence available. cGMP

Nitrate effects in males and females


PKG
This review paper concisely summa-
rized the current nitrate literature, with
a focus on three main outcomes: blood Ca2+ Ca2+
release sensitivity
pressure, exercise economy (that is, the
amount of energy expenditure required
for a given amount of exercise), and
performance. For relatively untrained
males, nitrate seems to induce a small
reduction in blood pressure. Some stud- Ftw df/dt Vmax Pmax
ies have also reported small blood pres-
sure reductions for relatively untrained Muscle contractile function
females, but the effect isn’t as large as in
untrained males. This difference may re- Dietary nitrate (NO3-) is converted to nitrite (NO2-), which is then converted to nitric oxide (NO).
Nitric oxide is thought to nitrosylated ryanodine receptors (RyR) and increase guanylyl cyclase (sGC)
activity, thereby increasing calcium (Ca2+) release and myofibrillar calcium sensitivity. This results
late to the fact that females have higher in enhanced twitch force (Ftw), rate of force development (dF/dt), estimated maximal shortening
velocity (Vmax), and maximal power (Pmax).

78
blood nitrite levels at baseline, or the fact itive conclusions. In addition, many of
that females typically have lower blood the performance studies utilized pret-
pressure than males. However, as sub- ty well-trained samples and investigat-
jects become more and more aerobical- ed acute (single-day) supplementation
ly fit, nitrate’s blood pressure-lowering protocols. This is notable, because the
effect seems to diminish. In theory, you male literature would indicate that more
could argue that this “lack of an effect” is robust effects are typically observed in
a good thing; if you have normal blood less-trained individuals, and when using
pressure, you wouldn’t want your dietary chronic (rather than acute) supplemen-
supplement to induce an acute episode tation protocols.
of hypotension. While the possibility of performance
In relatively untrained males, nitrate differences is not yet conclusively set-
supplementation has been shown to re- tled, the research seems to indicate that
duce the oxygen cost of submaximal ex- females have, at the very least, different
ercise, thereby improving exercise econ- baseline nitrite levels and differential
omy by around 3-5%. As a result, there blood pressure responses when com-
are quite a few studies reporting small pared to males. These apparent discrep-
performance improvements across a ancies are interesting and informative,
wide range of performance tests in this but the majority of MASS readers are
population. In fairly well-trained males, probably more interested in strength ef-
the research generally indicates that fects than blood pressure. So, let’s take a
these improvements in exercise economy closer look at strength and power-relat-
and endurance performance diminish. ed outcomes.
In the current review, only seven studies
evaluating performance in female-only A focus on lifting performance
samples were identified, which high- In terms of physiological effects, nitric
lights the glaring lack of female research oxide is most widely known as a vaso-
in this area. Effects on exercise econo- dilator that promotes blood flow. As a
my and performance were equivocal in result, most lifters tend to think that ni-
females, with the available studies re- tric oxide precursors, such as nitrate and
porting fairly mixed results. While this citrulline, inherently work via increased
sounds like the male literature is a lit- blood flow. In reality, it’s highly unlikely
tle bit more promising than the female that the delivery (or clearance) of blood
literature, it’s important to note a few is a critically limiting factor during resis-
caveats. The body of female literature is tance exercise performance, and there’s
far too limited, with only seven studies, limited justification for the idea that de-
to identify clear patterns or make defin- livering (or clearing) more blood would

79
Figure 3 Potential sex differences associated with dietary nitrate supplementation (1)

• Oral microbiome (+)


• Oral hygiene practices (?)
• NO3 · dose (?)

• Mitochondrial function (?) • Baseline plasma [NO2-] (+)


• Ca2+ handling (?) • ∆ plasma [NO3-] (+)
• Cross-bridge cycling (?) • ∆ plasma [NO2-] (+)
• Plasma [cGMP] (?)

• Skeletal muscle as a
storage reservoir for
• Baseline blood pressure (*)
NO3- (?)
• Post-NO3- blood pressure (-)

• NO3- excretion (?)

NO3- = dietary nitrate; + = response more robust in females; - = response less robust in females; * = lower in females; ? = unknown

be inherently beneficial for acute per- most likely related to direct effects on
formance effects. As discussed in the the contractile function of muscle.
current review (1), the evidence is also When a muscle contraction occurs,
a little shaky when it comes to nitrate there are a couple of very critical steps in
significantly influencing mitochondri- the process: Calcium is released from the
al efficiency. Larsen et al (4) did report sarcoplasmic reticulum, and that calci-
that nitrate supplementation enhanced um binds to a myofibrillar protein called
mitochondrial efficiency in a mixed troponin. Plenty of calcium release, plen-
sample of males and females, but oth- ty of binding to troponin, plenty of force
er studies have found no such effect (1, production. As reviewed by Coggan and
5). At this time, it seems that strength Peterson (6), nitrate increases nitric ox-
and power-related effects of nitrate are ide production, and evidence suggest

80
that nitric oxide can increase ryano-
dine receptor nitrosylation and guany-
lyl cyclase activity (Figure 2). This not NITRATE SUPPLEMENTATION
only increases calcium release, but also
makes the myofibril more sensitive to
DOES INDEED INFLUENCE THE
the force-promoting effects of calcium, CONTRACTILE FUNCTION OF
thereby enhancing force production and
shortening velocity.
MUSCLE IN FEMALES, AND
Aside from the sciencey, mechanistic IMPROVEMENTS IN VARIOUS
minutiae, there’s some (admittedly lim- STRENGTH AND POWER
ited) evidence that this carries over into
actual resistance exercise. Most notably, OUTCOMES HAVE BEEN OBSERVED
Mosher et al (7) found that nitrate sup- IN MIXED SAMPLES CONTAINING
plementation enhanced repetitions to
fatigue on three consecutive bench press BOTH MALES AND FEMALES. IT
sets taken to failure, using a load equiv-
alent to 60% of the subjects’ one-rep
SEEMS PREMATURE TO CONCLUDE
maxes. Unfortunately, Mosher’s study THAT NITRATE AFFECTS FEMALE
used male participants only, but it’s one
of the few nitrate studies looking at a
MUSCLE FUNCTION TO A LESSER
resistance exercise outcome with much DEGREE THAN MALE MUSCLE.
ecological validity at all. Coggan et al (8)
used a mixture of male and female sub-
jects, and found that nitrate supplemen- observe an increase in low-frequency
tation enhanced leg extension power, but torque production.
only at pretty high contraction veloci-
ties (360 degrees per second). Similarly, There’s a pretty serious lack of research
Hoon et al (9) found nitrate supplemen- pertaining to nitrate’s effects on resis-
tation to attenuate muscular fatigue in a tance exercise outcomes, but we can at
leg extension test using a mixed sample least make some observations based on
of male and female participants. Final- the evidence that’s currently available.
ly, a recent study by Wickham et al (10) It definitely seems as if nitrate has the
studied a female-only sample, and while capacity to influence muscle function,
they found no benefits for exercise econ- and preliminary results suggest that its
omy or cycling performance, and many effects are most notable during explo-
of their muscle function outcomes were sive, high-velocity muscle actions (8),
unaffected by supplementation, they did and its effects are more pronounced in
the fatigued state (11). Finally, we have

81
APPLICATION AND TAKEAWAYS
1. There’s a huge lack of research pertaining to nitrate supplementation in females,
and there are notable anthropometric, physiological, and metabolic differences
between sexes that could theoretically yield divergent responses to nitrate
supplementation.
2. Most of the research showing performance improvements with nitrate
supplementation has been done using untrained or recreationally active males
doing endurance exercise. When it comes to resistance exercise, it’s still unclear
if males experience a greater performance boost than females, or if performance
benefits will carry over to well-trained individuals of either sex.
3. There’s enough mechanistic and preliminary research to tentatively consider
nitrate to be a promising ergogenic aid, but that doesn’t mean you necessarily
need to supplement. Aiming to achieve daily nitrate intakes of 400-800mg, either
from food or supplementation, has potential benefits with minimal likelihood of
adverse effects.

observed that, at the very least, positive sistance exercise performance. But, for
effects on contractile function are ob- now, the preliminary results are promis-
served in samples that at least contain ing. So, let’s discuss a few practical con-
some females. However, can the results siderations for those interested in nitrate
of Mosher et al (7) be reliably replicat- supplementation.
ed? Are these effects larger in men than First, an effective nitrate dose tends to
women? Will women actually benefit be at least 400-800 mg (preferably to-
to a meaningful degree when utilizing a ward the higher end), which is equiv-
test protocol that approximately resem- alent to 6.4-12.9 mmol. A fairly con-
bles “real-world” resistance exercise? If sistent observation is that studies with
we replicate these studies in well-trained chronic (multiple-day) dosing protocols
athletes instead of recreationally active observe more reliable effects than acute
subjects, will we see similar results? For (single-dose) studies. A very recent
now, these questions remain unanswered. study (12) helps explain why that might
be, as its findings suggest that humans
Practical considerations for beetroot/ni- have the ability to store nitrate and ni-
trate supplementation trite within muscle, and that this stor-
We still have plenty to learn about ni- age reservoir becomes depleted follow-
trate supplementation and exactly how ing exercise. Notably, the sample in this
much potential it has to enhance re- study included both males and females,

82
so this does not appear to be a male-spe- in various strength and power outcomes
cific finding. Finally, the nitrate supple- have been observed in mixed samples
ments that are currently on the market containing both males and females. It
aren’t great. A recent study (13) evalu- seems premature to conclude that ni-
ated 24 different beetroot juice products trate affects female muscle function to
on the market; nitrate content varied a lesser degree than male muscle, but
pretty substantially between samples of we need more research to reach a more
the same exact product, and there was definitive conclusion. This isn’t the only
a 50-fold range in nitrate content be- question that remains; as summarized
tween the products with the highest and in Figure 3, we also need answers to a
lowest dose. Fortunately, it’s totally real- variety of lingering questions pertaining
istic to obtain ergogenic doses of nitrate to how sex might affect nitrate dosing,
from simple dietary modifications, and the conversion of nitrate to nitrite, mi-
at least one study has shown this to sig- tochondrial function, and intramuscular
nificantly enhance multiple performance storage of nitrate and nitrite.
outcomes (14). Foods with high nitrate
content include celery, beets, spinach,
rocket (arugula), and a variety of other Next Steps
fruits and vegetables. This review made it abundantly clear
In conclusion, the current review article that there are several key questions that
certainly identified some discrepancies need to be pursued. We need stud-
between the male and female nitrate lit- ies evaluating potential sex-based dif-
erature. Namely, women appear to have ferences pertaining to optimal nitrate
higher plasma nitrite levels, lower resting dosing strategies, intramuscular storage
blood pressure, a smaller blood pressure and depletion of nitrate and nitrite, the
reduction from nitrate, and less consis- effects of nitrate on exercise economy
tent improvements in exercise economy and mitochondrial efficiency, and intra-
and performance tests involving endur- muscular calcium kinetics and contrac-
ance exercise modalities (such as run- tile properties following nitrate intake.
ning, swimming, and cycling). However, I’d love to see a female study similar to
when it comes to resistance training, the Mosher et al (7) to evaluate the effects
most likely ergogenic mechanism relates of dietary nitrate in the context of a test
to direct effects on muscle function. At protocol that approximately reflects “re-
this time, we have some evidence to sug- al-world” resistance training. It’d also be
gest that nitrate supplementation does great to see studies that directly com-
indeed influence the contractile function pare male and female responses, rather
of muscle in females, and improvements than drawing conclusions by comparing

83
male-only and female-only studies with
key methodological differences. Finally,
we’ll need to determine if the effects of
dietary nitrate vary throughout the men-
strual cycle, and determine if habitually
increased nitrate intake actually enhanc-
es training adaptations over time.

84
References
1. Wickham KA, Spriet LL. No longer beeting around the bush: a review of potential sex differences
with dietary nitrate supplementation. Appl Physiol Nutr Metab. 2019 Sep;44(9):915–24.
2. Stuart-Smith K. Demystified … Nitric oxide. Mol Pathol. 2002 Dec;55(6):360–6.
3. Bailey SJ, Vanhatalo A, Winyard PG, Jones AM. The nitrate-nitrite-nitric oxide pathway: Its role in
human exercise physiology. Eur J Sport Sci. 2012 Jul 1;12(4):309–20.
4. Larsen FJ, Schiffer TA, Borniquel S, Sahlin K, Ekblom B, Lundberg JO, et al. Dietary inorganic
nitrate improves mitochondrial efficiency in humans. Cell Metab. 2011 Feb 2;13(2):149–59.
5. Whitfield J, Ludzki A, Heigenhauser GJF, Senden JMG, Verdijk LB, van Loon LJC, et al. Beetroot
juice supplementation reduces whole body oxygen consumption but does not improve indices of
mitochondrial efficiency in human skeletal muscle. J Physiol. 2016 Jan 15;594(2):421–35.
6. Coggan AR, Peterson LR. Dietary Nitrate Enhances the Contractile Properties of Human Skeletal
Muscle. Exerc Sport Sci Rev. 2018 Oct;46(4):254–61.
7. Mosher SL, Sparks SA, Williams EL, Bentley DJ, McNaughton LR. Ingestion of a Nitric Oxide
Enhancing Supplement Improves Resistance Exercise Performance. J Strength Cond Res. 2016
Dec;30(12):3520–4.
8. Coggan AR, Leibowitz JL, Kadkhodayan A, Thomas DT, Ramamurthy S, Spearie CA, et al. Effect
of acute dietary nitrate intake on maximal knee extensor speed and power in healthy men and wom-
en. Nitric Oxide. 2015 Aug 1;48:16–21.
9. Hoon MW, Fornusek C, Chapman PG, Johnson NA. The effect of nitrate supplementation on mus-
cle contraction in healthy adults. Eur J Sport Sci. 2015;15(8):712–9.
10. Wickham KA, McCarthy DG, Pereira JM, Cervone DT, Verdijk LB, van Loon LJC, et al. No effect
of beetroot juice supplementation on exercise economy and performance in recreationally active fe-
males despite increased torque production. Physiol Rep. 2019 Jan;7(2):e13982.
11. Tillin NA, Moudy S, Nourse KM, Tyler CJ. Nitrate Supplement Benefits Contractile Forces in Fa-
tigued but Not Unfatigued Muscle. Med Sci Sports Exerc. 2018 Oct;50(10):2122–31.
12. Wylie LJ, Park JW, Vanhatalo A, Kadach S, Black MI, Stoyanov Z, et al. Human skeletal muscle
nitrate store: influence of dietary nitrate supplementation and exercise. J Physiol. 2019 Jul 27; doi:
10.1113/JP278076.
13. Gallardo EJ, Coggan AR. What’s in Your Beet Juice? Nitrate and Nitrite Content of Beet Juice
Products Marketed to Athletes. Int J Sport Nutr Exerc Metab. 2019 01;29(4):345–349.
14. Porcelli S, Pugliese L, Rejc E, Pavei G, Bonato M, Montorsi M, et al. Effects of a Short-Term
High-Nitrate Diet on Exercise Performance. Nutrients. 2016 Aug 31;8(9).

85
Study Reviewed: The Placebo and Nocebo Effect on Sports Performance: A
Systematic Review. Hurst et al. (2019)

The Placebo Effect Impacts


Performance More Than You
Might Expect
BY G RE G NUC KO LS

The placebo effect is a well-known psychological phenomenon, but


we often forget about it in exercise research. We focus on how much
a supplement, device, or treatment improves performance relative to
a placebo, but a lot of the “real-world” improvements in performance
may actually be attributable to placebo effects.

86
KEY POINTS
1. In a meta-analysis, placebo and nocebo effects both have small but meaningful
impacts on performance.
2. The placebo effect is larger if someone thinks they’re ingesting a banned substance,
such as anabolic steroids or erythropoietin (EPO).
3. Interestingly, for caffeine, up to two-thirds of the “real-world” increase in performance
with caffeine usage may be attributable to the placebo effect (though caffeine itself
absolutely still has a real physiological effect).

A
ny time a new supplement effects (the placebo effect’s evil twin
study drops, savvy readers know – worsening performance when given
that one of the first things to a treatment you expect to harm perfor-
check is whether the study was place- mance) have a small but notable effect
bo-controlled. Placebo controls are im- on physical performance. Furthermore,
portant, because part of the effect you when people think they’re consuming a
get from any treatment is the effect you banned substance, or when they’re ma-
get from simply receiving a treatment, nipulated into believing the placebo has
even if that treatment doesn’t actually already provided them a performance
do anything. If you can do 10 reps with boost, the placebo effect is even larger.
a given weight without taking a pill, 12
reps with a sugar pill, and 13 reps with
a caffeine pill, then simply taking a pill Purpose and Hypotheses
gives you 2 reps, with caffeine only giv- Since this was a meta-analysis, the
ing you 1 extra rep, not 3. purpose was simply to statistically ana-
We mostly focus on how much a giv- lyze all of the research investigating the
en supplement, device, or treatment effects of placebos and nocebos on ex-
improves performance in excess of the ercise performance. No hypotheses are
boost provided by a placebo. We gener- stated for meta-analyses.
ally pay less attention to the boost pro-
vided by the placebo itself. However, a
recent meta-analysis (1) fills that gap, Methods
analyzing the research that has inves-
The authors ran a search for studies
tigated the magnitude of the placebo
investigating the effects of placebos
effect on exercise performance. Over-
and nocebos on physical performance.
all, it seems that placebo and nocebo
To be included in the meta-analysis,

87
Figure 1 Exclusion process of identified studies
Identification

Records identified through Records identified through


electronic databases other resources
n = 4026 n=8
Screening

Records screened Records excluded


n = 4034 n = 3992
Eligibility

Full text articles assessed Full text records excluded


for eligibility n = 10
n = 42 No control treatment (n=3)
Non-healthy participants
(n=2)
Inadequate placebo (n=2)
Non-sporting outcome
measure (n=2)
Included

Failure to deceive
Studies included in analysis
participants (n=1)
n = 32

a study needed to be published in an Once the studies were collected, the


English-language peer-reviewed jour- authors calculated effect sizes for each
nal, include healthy subjects, and re- study and “aggregated” them. The au-
port at least one objective measure of thors don’t actually explain how they
performance (i.e. not just pain, fatigue, performed the meta-analysis itself
or perceived exertion). Furthermore, (Fixed effects model? Random effects
to quantify the placebo effect, studies model? Simply taking a weighted av-
needed to include a no-treatment con- erage of the effect sizes?), which is …
trol condition or measurement. odd.

88
Figure 2 Effect sizes for various placebo interventions

1.5

Large

1
Effect size (d)

Medium

0.5

Small

Trivial
0
Over placebos

Nutritional placebos

Amino acid placebo

Overall placebo effect

Overall nocebo effect

Caffeine placebo

Mechanical placebos

EPO placebo

+ placebo

TENS placebo

Steroid placebo
Preconditioning

Findings placebos or nocebos, while 12 investi-


gated mechanical placebos. Most of the
More than 4,000 studies were studies investigated the placebo effect,
screened, with 32 eventually meet- while only 5 examined the nocebo ef-
ing the inclusion criteria, represent- fect. Of the studies using placebos, 4
ing 1,513 total participants. Of those used overt placebos (i.e. they told the
32 studies, 20 investigated nutritional subjects when they were giving them a

89
small effect (d = 0.21). Cold water im-
mersion, sodium bicarbonate, ischemic
preconditioning, carbohydrate, beta al-
ONE OF THE MOST LOGICAL anine, kinesio tape, and magnetic wrist-
AND INTERESTING FINDINGS bands were all found to have trivial or
null effects. Overt placebos also had no
OF THIS META-ANALYSIS significant effect.
WAS THE MORE “SERIOUS” The nocebo effect was also small (d =
0.37). Since there were only five stud-
PLACEBOS INDUCED LARGER ies investigating the nocebo effect, the
researchers didn’t investigate whether
PLACEBO EFFECTS. the type of nocebo used moderated the
outcomes.
placebo), while 5 augmented the place-
bos with preconditioning via augment-
ed feedback (which I’ll explain in the
Interpretation
next section). The first thing I’d like to reiterate is
that the authors don’t make it clear
Overall, nutritional and mechanical
how they actually performed their me-
ergogenic aids significantly improved
ta-analysis, so it’s hard to tell if they
performance. The effect sizes for both
mucked something up. I think they
were small (d = 0.35 for nutritional
just extracted the effect size from each
placebos, and d = 0.47 for mechanical
study and either averaged them or took
placebos). Unsurprisingly, the placebo
a weighted average. If that’s what they
effects generated by placebos claiming
did, their point estimates for each effect
to be banned substances were larger
size may be pretty alright (just skim-
(d = 1.44 for steroids, and d = 0.81 for
ming the reported effect sizes, it doesn’t
EPO). Interestingly, preconditioning
seem like there’s an extreme amount of
procedures also had a large effect on
variability within each sub-category),
performance (d = 0.82). Sham trans-
but they may have either more or less
cutaneous electrical nerve stimulation
variance around the mean estimate than
(TENS) was also reported to have a
they “should.” There are a few other sta-
large effect size (d = 0.86). Small ef-
tistical things I could complain about,
fect sizes were noted for placebo amino
but I feel like I probably do too much
acids and caffeine (d = 0.36 and 0.40,
of that anyways, and the things I want
respectively). A completely fictitious
to complain about probably wouldn’t
sport supplement was found to have a
meaningfully change my interpreta-

90
tion of this meta-analysis. However, we
probably need to be a little more cau-
tious with these findings than we would THE PLACEBO EFFECT OF
otherwise need to be if their statistical
approach was more transparent. CAFFEINE MAY BE JUST AS
One of the most logical and interest- LARGE AS THE “TRUE” EFFECT
ing findings of this meta-analysis was
the more “serious” placebos induced OF CAFFEINE, IF NOT LARGER.
larger placebo effects. Since the place-
bo effect is a psychological phenome-
non based on expectancy (you think and experience a disproportionate in-
what you’re taking will improve perfor- crease in strength. However, in place-
mance, and that belief is what actually bo-controlled research (i.e. when peo-
improves performance), it makes since ple who aren’t taking steroids are still
that placebos that you think will have given injections to make them think
a larger effect do actually have a larger they’re taking steroids), it seems that
effect. I think just about anyone would the boost steroids provide for strength
expect to improve their strength when gains is disproportionately smaller than
taking steroids, or improve their endur- the boost they provide for hypertro-
ance performance when taking EPO; phy (6), with steroids helping subjects
those substances are banned in most build 3.3 times more muscle (4.23% vs.
competition because they do cause 13.92% increases in combined triceps
large improvements in performance, and quadriceps cross-sectional area),
after all. And though just three of the while only helping them build 1.7 times
studies in this meta-analysis investigat- more strength (35 vs. 60 kg combined
ed banned substances (2 for steroids, 1 increase in squat and bench press 1RM).
for EPO; 2, 3, 4, 5[NOTE]), all three Since this meta-analysis found that the
of those studies reported large effects, placebo effect itself generates a large
and the mean effect for those studies improvement in performance when
(d = 1.23) was more than three times people think they’re taking banned
larger than the mean effect for all of the substances, it does make me think the
studies in this meta-analysis (d = 0.37). reports of large, fast strength increases
I’ve always wondered about the degree in strength reported when people start
to which the increase in strength people taking steroids may be based as much
report when they start taking steroids is on expectancy as the actual physiologi-
attributable to expectancy. People often cal effects of the drugs.
claim that they build a lot more muscle

91
to one-half of the effect attributable to
the actual physiological effects of caf-
feine.
AS A GENERAL RULE, IT PAYS
I was surprised that placebo TENS
TO BE OPTIMISTIC; IF YOU treatment was so effective. TENS units
consist of electrodes that are placed on
EXPECT TO PERFORM WELL the skin, with a current passed through
AND HAVE GOOD RESULTS the electrodes that is sufficient to stim-
ulate the underlying nerves, but gener-
FROM YOUR TRAINING, YOU ally low enough that it does not cause
muscular contraction. Generally you can
PROBABLY WILL, AND IF YOU feel a TENS unit working (it tingles at
EXPECT TO PERFORM POORLY low voltage and can be mildly uncom-
fortable at higher voltages), though you
AND HAVE LACKLUSTER can’t feel the current if the voltage is low
enough. For placebo TENS treatment,
RESULTS, THAT WILL you turn the unit on, and you may even
LIKELY BECOME A SELF- rig it up so that it will show that it’s
operating at a low voltage, but you don’t
FULFILLING PROPHECY. run a current through it. While I sup-
pose the show of placing the electrodes
and turning the machine on may repre-
sent a “bigger” placebo treatment than
Another interesting note is that the simply giving someone an unmarked
placebo effect of caffeine may be just as pill, I would have thought that people
large as the “true” effect of caffeine, if not would be highly skeptical that the ma-
larger. Meta-analyses find that caffeine chine was really doing anything if they
improves performance relative to place- didn’t feel anything. However, it does
bo, with small effect sizes in the 0.2-0.4 appear that sham TENS treatment
range (7). This meta-analysis found that is quite an effective placebo. I doubt
the placebo effect for caffeine was asso- MASS readers can really do much with
ciated with an effect size of about 0.4. that information, but I thought it was
Thus, in the “real world,” when people cool.
take caffeine before a workout, the total
The finding that preconditioning
effect may actually be quite large (i.e. in
caused a large improvement in perfor-
the d = 0.6-0.8 range), with about one-
mance is fascinating. For a positive pre-
half to two-thirds of the effect attribut-
conditioning study, you generally test
able to expectancy, and about one-third

92
APPLICATION AND TAKEAWAYS
The placebo effect has a small but notable effect on performance under most
circumstances. While it’s hard to placebo yourself, expectancy effects in general can
influence performance, so try to maintain positive self-talk about your training, and
try to avoid catastrophizing things that happen in your life, as doing so could harm
performance by triggering negative expectancy effects.

subjects at baseline, test them again preconditioning is combined with a


when giving them a placebo, while al- placebo treatment, the total effect is (on
tering the test to make the subjects feel average) more than twice as large as the
like they performed better, and then effect of the placebo alone.
test them another time with the placebo Predictably, overt placebos didn’t im-
but without the test alteration. For ex- pact performance. An overt placebo is
ample, you could test someone’s maxi- a placebo someone knows they’re get-
mal bench press reps with 100kg on day ting. If you give someone a pill and say,
1. 48 hours later, you could give them “this is a placebo pill that does noth-
a placebo pill, tell them it’s caffeine, ing,” that’s an overt placebo (some-
and test their maximal bench press reps times called an open label placebo). It
with 90kg, while telling them it’s 100kg may sound ludicrous that the effects
(with altered plates so the bar still looks of overt placebos even need to be re-
like it’s loaded to 100kg). Their perfor- searched, but interestingly, overt place-
mance should be better, which they’ll bos have been found to improve subjec-
attribute to the placebo pill. 48 hours tive outcomes (such as pain or nausea)
later, you give them the placebo pill compared to a no-treatment control
again, and test their maximal bench (8), which I find fascinating. However,
press reps with 100kg. Since they al- when you’re objectively measuring ex-
ready have the belief that the placebo ercise performance, people don’t get a
pill (which they think is caffeine) im- placebo effect when they know they’re
proves their performance, they’ll prob- taking a placebo.
ably perform much better on this test
Finally, I want to draw attention to
with 100kg than they would have if you
the fact that the mean nocebo effect was
hadn’t preconditioned them (i.e. session
just as large as the mean placebo effect.
1: 100kg with no placebo; session 2:
When people think something will im-
100kg with placebo). Preconditioning
prove performance, it generally does,
essentially works to amplify expectancy,
and when people think something will
and this meta-analysis found that when

93
harm performance it generally does. In
a previous MASS article, I reviewed a Next Steps
study showing that in some cases, be- I’m still stuck on the idea of the pla-
liefs about your genetics can affect your cebo effect of steroids. A deception
physiology and perceptions more than study would be really cool, involving
your actual genetics themselves do (10). four groups. All four groups are giv-
I’m not sure how relevant that finding en pills. Group 1 is given oral steroids
is in this particular meta-analysis (I and are told they’re being given oral
don’t think many people intentionally steroids. Group 2 is given oral steroids
take supplements or use interventions and are told they’re being given an in-
that they think will hinder their perfor- ert control. Group 3 is given an inert
mance), but negative expectancy effects control and are told they’re being given
are a pretty generalizable concept. It’s oral steroids. Group 4 is given an inert
something to keep in mind with your control and are told they’re being given
self-talk and the interactions you have an inert control. Since this meta-anal-
with your clients. For example, if you ysis found that overt placebos don’t
expect that your strength is going to improve exercise performance, group
drop when you go into a calorie defi- 2 would show us the “true” physiolog-
cit, it probably will. It may have done ical effect of steroids, group 3 would
so anyways, but the drop will likely be show us the placebo effect of thinking
larger if you expect it to take a hit. Or you’re taking steroids, group 1 would
if you expect a night of bad sleep to show us the “stacked” effect (placebo +
hinder performance, your performance actual physiological effect), and group
will probably be lower (even though re- 4 would be the control group. I would
search tends to find that one night of predict that groups 1 and 2 would easily
bad sleep doesn’t have much of an im- gain the most muscle, but I wouldn’t be
pact on acute performance; it tends to shocked if groups 2 and 3 gained simi-
negatively affect skill performance and lar amounts of strength.
tactical decision-making, but not things
like force output or endurance [9]). As
a general rule, it pays to be optimistic;
if you expect to perform well and have
good results from your training, you
probably will, and if you expect to per-
form poorly and have lackluster results,
that will likely become a self-fulfilling
prophecy.

94
References
1. Hurst P, Schipof-Godart L, Szabo A, Raglin J, Hettinga F, Roelands B, Lane A, Foad A, Coleman
D, Beedie C. The Placebo and Nocebo effect on sports performance: A systematic review. Eur J
Sport Sci. 2019 Aug 20:1-14.
2. Ariel G, Saville W. Effect of anabolic steroids on reflex components. J Appl Physiol. 1972
Jun;32(6):795-7.
3. Maganaris CN, Collins D, Sharp M. Expectancy Effects and Strength Training: Do Steroids Make
a Difference? The Sport Psychologist. 200;14(3):272-278.
4. Heuberger JAAC, Rotmans JI, Gal P, Stuurman FE, van ‘t Westende J, Post TE, Daniels JMA, Mo-
erland M, van Veldhoven PLJ, de Kam ML, Ram H, de Hon O, Posthuma JJ, Burggraaf J, Cohen
AF. Effects of erythropoietin on cycling performance of well trained cyclists: a double-blind, ran-
domised, placebo-controlled trial. Lancet Haematol. 2017 Aug;4(8):e374-e386.
5. I think the authors accidentally cited the wrong study for EPO. I think citation 4 above is the study
they meant to cite.
6. Bhasin S, Storer TW, Berman N, Callegari C, Clevenger B, Phillips J, Bunnell TJ, Tricker R, Shirazi
A, Casaburi R. The effects of supraphysiologic doses of testosterone on muscle size and strength in
normal men. N Engl J Med. 1996 Jul 4;335(1):1-7.
7. Grgic J, Grgic I, Pickering C, Schoenfeld BJ, Bishop DJ, Pedisic Z. Wake up and smell the coffee:
caffeine supplementation and exercise performance-an umbrella review of 21 published meta-anal-
yses. Br J Sports Med. 2019 Mar 29. pii: bjsports-2018-100278.
8. Charlesworth JEG, Petkovic G, Kelley JM, Hunter M, Onakpoya I, Roberts N, Miller FG, Howick
J. Effects of placebos without deception compared with no treatment: A systematic review and me-
ta-analysis. J Evid Based Med. 2017 May;10(2):97-107.
9. Kirschen GW, Jones JJ, Hale L. The Impact of Sleep Duration on Performance Among Competitive
Athletes: A Systematic Literature Review. Clin J Sport Med. 2018 Jun 14.
10. Turnwald BP, Goyer JP, Boles DZ, Silder A, Delp SL, Crum AJ. Learning one’s genetic risk changes
physiology independent of actual genetic risk. Nature Human Behavior. 2019; 3:48-56.

95
VIDEO: Concurrent Training
for a Marathon
BY MIC HAE L C . ZO URD O S

You may know that running long distance isn’t the ideal form of cross-
training to maximize your strength. However, maybe you don’t care, and
you want to run a marathon anyway. I think this is cool, as once upon a
time I didn’t care and ran some, despite knowing the negative effects
on strength and size. Therefore, this video shows you how to accomplish
completing a marathon while affecting strength as little as possible.
Click to watch Michael's presentation.

96
Relevant MASS Videos
1. Concurrent Training Attenuates Anaerobic Adaptations Part 1. (Volume 1 Issue 3)
2. Concurrent Training Attenuates Anaerobic Adaptations Part 2. (Volume 1 Issue 4)

References
3. Wilson JM, Marin PJ, Rhea MR, Wilson SM, Loenneke JP, Anderson JC. Concurrent training: a
meta-analysis examining interference of aerobic and resistance exercises. The Journal of Strength &
Conditioning Research. 2012 Aug 1;26(8):2293-307.
4. Murach KA, Bagley JR. Skeletal muscle hypertrophy with concurrent exercise training: contrary
evidence for an interference effect. Sports medicine. 2016 Aug 1;46(8):1029-39.

97
VIDEO: Energy Intake to
Optimize Lean Mass, Part 1
BY E RI C HE LMS

“You must be in a surplus to build muscle”– An oft-repeated adage in the evidence-


based community, but perhaps an overly simplified statement. In part 1 of this
series, we draw on multiple reviews published in the last 20 years, and mechanistic
primary research to explore how energy intake influences hypertrophy, and when
and why a surplus might or might not be necessary to grow muscle.
Click to watch Eric's presentation.

98
References
1. Houston ME. Gaining weight: the scientific basis of increasing skeletal muscle mass. Canadian
journal of applied physiology. 1999 Aug 1;24(4):305-16.
2. Phillips SM. A brief review of critical processes in exercise-induced muscular hypertrophy. Sports
Medicine. 2014 May 1;44(1):71-7.
3. Rossow LM, Fukuda DH, Fahs CA, Loenneke JP, Stout JR. Natural bodybuilding competition
preparation and recovery: a 12-month case study. International journal of sports physiology and
performance. 2013 Sep;8(5):582.
4. Hulmi JJ, Isola V, Suonpää M, Järvinen NJ, Kokkonen M, Wennerström A, Nyman K, Perola M,
Ahtiainen JP, Häkkinen K. The Effects of Intensive Weight Reduction on Body Composition and
Serum Hormones in Female Fitness Competitors. Frontiers in physiology. 2017 Jan 10;7:689.
5. Hector AJ, McGlory C, Damas F, Mazara N, Baker SK, Phillips SM. Pronounced energy restriction
with elevated protein intake results in no change in proteolysis and reductions in skeletal muscle
protein synthesis that are mitigated by resistance exercise. FASEB journal: official publication of the
Federation of American Societies for Experimental Biology. 2018 Jan;32(1):265.
6. Pasiakos SM, Vislocky LM, Carbone JW, Altieri N, Konopelski K, Freake HC, Anderson JM, Fer-
rando AA, Wolfe RR, Rodriguez NR. Acute energy deprivation affects skeletal muscle protein syn-
thesis and associated intracellular signaling proteins in physically active adults. The Journal of nutri-
tion. 2010 Apr;140(4):745.
7. Carbone JW, Pasiakos SM, Vislocky LM, Anderson JM, Rodriguez NR. Effects of short-term en-
ergy deficit on muscle protein breakdown and intramuscular proteolysis in normal-weight young
adults. Applied Physiology, Nutrition, and Metabolism. 2014 Jun 19;39(8):960-8.
8. Mountjoy M, Sundgot-Borgen JK, Burke LM, Ackerman KE, Blauwet C, Constantini N, Lebrun
C, Lundy B, Melin AK, Meyer NL, Sherman RT. IOC consensus statement on relative energy de-
ficiency in sport (RED-S): 2018 update. British journal of sports medicine. 2018 Jun;52(11):687.
9. Fagerberg P. Negative Consequences of Low Energy Availability in Natural Male Bodybuilding: A
Review. International journal of sport nutrition and exercise metabolism. 2018 Jul 1;28(4):385.

99
Just Missed the Cut
Every month, we consider hundreds of new papers, and they can’t all be included in MASS.
Therefore, we’re happy to share a few pieces of research that just missed the cut. It’s
our hope that with the knowledge gained from reading MASS, along with our interpreting
research guide, you’ll be able to tackle these on your own.

1. Ortiz et al. A Systematic Review on the Effectiveness of Active Recovery Interventions on


Athletic Performance of Professional-, Collegiate-, and Competitive-Level Adult Athletes
2. Trezise and Blazevich. Anatomical and Neuromuscular Determinants of Strength Change
in Previously Untrained Men Following Heavy Strength Training
3. Dinyer et al. Applying the Critical Power Model to a Full Body Resistance Training
Movement
4. Schoenfeld et al. Calculating Set-Volume for the Limb Muscles with the Performance of
Multi-Joint Exercises: Implications for Resistance Training Prescription
5. Wilk et al. Changes of Power Output and Velocity During Successive Sets of the Bench
Press with Different Duration of Eccentric Movement
6. Leite et al. Does exercise-induced muscle damage impair subsequent motor skill
learning?
7. Dias et al. Effect of Acute and Chronic Xenon Inhalation on Erythropoietin, Hematological
Parameters, and Athletic Performance
8. Ghaedi et al. Effects of a Paleolithic Diet on Cardiovascular Disease Risk Factors: A
Systematic Review and Meta-Analysis of Randomized Controlled Trials
9. Marzliger et al. Effects of Lengthening Velocity During Eccentric Training on Vastus
Lateralis Muscle Hypertrophy
10. Jäger et al. Eight Weeks of a High Dose of Curcumin Supplementation May Attenuate
Performance Decrements Following Muscle-Damaging Exercise
11. Bloedon et al. Impact of anthocyanin-rich whole fruit consumption on exercise-induced
oxidative stress and inflammation: a systematic review and meta-analysis
12. Haischer et al. Impact of Cognitive Measures and Sleep on Acute Squat Strength
Performance and Perceptual Responses Among Well-Trained Men and Women
13. Vieira et al. Resistance Training with Repetition to Failure or Not on Muscle Strength and
Perceptual Responses
14. Coratella et al. Specific prime movers’ excitation during free-weight bench press
variations and chest press machine in competitive bodybuilders
15. Carzoli et al. The effects of eccentric phase duration on concentric outcomes in the back
squat and bench press in well-trained males
16. Kamali et al. Transcranial direct current stimulation to enhance athletic performance
outcome in experienced bodybuilders

100
Thanks for
reading MASS.
The next issue will be released to
subscribers on November 1.

Graphics by Kat Whitfield, and layout design by Lyndsey Nuckols.

101

You might also like