The Mellanby Effect in Moderate

and Heavy Drinkers

Herbert Moskowitz,3b John Daily and Robert Henderson

INTRODUCTION

The M ellanby effect is nam ed for E. M ellanby, who, in 1919, reported that the behavioural
im pairm ent at a given blood alcohol level was greater when the blood alcohol level was
rising than w hen it was falling.2 W hile m any subsequent studies have replicated this example
o f acute tolerance, estimates o f its m agnitude have been com prom ised by experim ental prob­
lems. F or exam ple, m any studies have determ ined blood alcohol levels (BAC) using venous
samples whose alcohol level lags in time com pared to arterial and brain levels during the
rising BAC phase. O ther studies have failed to control for practice effects since, typically,
they have m easured subjects sequentially on a task — first during the rising and then on
the falling phases o f a single adm inistration. Clearly, during the falling period, subjects will
have been m ore practised on the task.
A nother problem has been that m any o f these studies have utilised extremely rapid rates
o f adm inistration o f alcohol with doses larger than 1 gm /kg bodyweight, given in five to twelve
minutes. Since rapid intake is associated with an initially greater degree o f im pairm ent, it
would be hard to estimate the additional deficits associated with the rising BAC phase in
norm al alcohol consum ption.
This current study attem pted to control the factors discussed above. It had one additional
feature. H alf the subjects were very heavy drinkers and h alf were m oderate drinkers, so as
to determ ine w hether there was an interaction between the developm ent o f acute tolerance
to alcohol and the existence o f chronic tolerance, typically found in the heavy drinker.
In the study, practice effects were balanced by requiring subjects to attend two experim en­
tal sessions, once for testing on a rising BAC curve and once for testing on a falling BAC
curve. H alf the subjects were first tested on a rising and then a falling BAC phase, and the
opposite was done for the other h alf o f the subjects. Accurate estimates o f brain alcohol
level were obtained by use o f a breath sam pling gas chrom atograph. Breath alcohol levels
are in equilibrium with arterial blood alcohol levels which are in turn in equilibrium with
brain alcohol levels. Thus, Unbiased estimates o f brain BAC were obtained for both rising
and falling blood alcohol phases. Alcohol adm inistration was given at rates more typical
o f social drinking situations, ranging betw een 0.32 and 0.35 g alcohol per kg bodyweight
p e r hour.
The «ttfdy utilised five behavioural perform ance measures: hand steadiness while stand­
ing and sitting; body ss?ay in tjbe lateral and anterior-posterior planes; and finally, a divided
attention task. D ue to technical difficulties, the divided attention task produced unreliable
data and results are not included.

8 California State University, Los Angeles and University o f California, Los Angeles.
bThis study was supported by the National Highway Traffic Safety Administration, U.S. Department
o f Transportation, and the National Institute o f Alcohol Abuse and Alcoholism, U.S. Department of
Health, Education, and Welfare. It was performed at the Systems Development Corporation. Opinions
expressed are those o f the authors and not necessarily the sponsors.

184
 The Mellanby Effect in Moderate and Heavy Drinkers 185

METHOD

Subjects
Forty male subjects were used: twenty m oderate drinkers and twenty heavy drinkers. These
categories were defined by alcohol quantity-frequency consum ption scales and personal inter­
views regarding recent drinking history.

Response Measures
H and steadiness was m easured by am ount o f time a metal stylus inserted in a hole made
contact with the metal plate. This m easure was taken both while standing and while sitting.
Body sway was m easured by attaching two strings at chest height to the subject’s side and
back and measuring the excursions o f the string in the subject’s anterior-posterior plane and
in the lateral plane.

Alcohol Treatments
Alcohol was adm inistered in mixed drinks containing 80-proof vodka. It was intended that
m oderate drinkers consume sufficient alcohol to achieve a BAC o f 0.10% and heavy drinkers
achieve 0.15%. The alcohol treatm ents adm inistered were designed to produce increases o f
roughly 0.02% BAC per hour in m oderate drinkers. In heavy drinkers the treatm ents were
also designed to produce increases o f 0.02% BAC per hour until they reached 0.10% BAC,
and then the alcohol rate was increased slightly so BACs were to increase 0.025% per hour.
Thus, m oderate drinkers reached their BAC levels in, roughly, five hours, arfd heavy drinkers
in, roughly, seven hours, during the day w hen testing was on the rising BAC. BAC samples
were taken at frequent intervals. Behavioural testing was tim ed by frequent breath sampling
so as to produce perform ance measures at 0.01% BAC.

RESULTS

BAC levels dropped 0.020% per hour for heavy drinkers and 0.017% per hour for m oderate
drinkers on the falling BAC test days. This difference is to be expected in terms o f the in­
creased metabolism rate for alcohol typically found in chronically tolerant heavy
drinkers.
It was desired that the rate o f change o f alcohol levels be similar during7the behavioural
testing on the rising and fiftlling phases. D uring the rising BAC test days, BAC levels rose
0.023% per hour S trt the m oderale drinkers aad 0.024% for the heavy drinkers. Thus, the
rates o f change o f alcohol level during the rising BAC test sessions were, roughly, 20 to 25
per cent jgreater fhan on the falling “BAS test lays.
Figure 1 shows perform ance scores for all subjects, both heavy and m oderate drinkers,
on the four behavioural response measures for the conditions o f rising and falling BAC at
increments o f 0.01% changes in BAC level plus an initial pretest level. It can be seen that
the data supports the existence o f the M ellenby effect in that perform ance decrem ent was
always greater at every BAC level during the rising BAC test days in contrast to the falling
BAC test days. The data was tested for statistical significance using a repeated measures
m ultivariant analysis o f variance based on a linear hypothesis model, and the difference
due to the rising and falling curve was found significant for all response measures. It should
be noted, however, that the difference in resistance to im pairm ent produced by testing on
the falling versus the rising BAC phase is quite small, equivalent in effect to between 0.01%
and 0.02% differences in BAC level.
186 H. Moskowitz, J. Daily and R. Henderson

A ------A Rising

O ------- O FALLING
30
o
oUJ
z UJ
OZ
>
_ 20
00 o
ZD .
z:
10
.--Cr
£ O-----O------- 0 - — 0 - — o-
L A T E R A L SWAY
—i— i- H------- 1-------1--------1

TEST
BLOOD ALCOHOL CONCENTRATION 1

Figure 1 Perform ance o f fo u r behavioural m easures as a fu n ction o f BA C fo r rising and

fa llin g BA C phases.

Figure 2 shows differences in perform ance decrem ents at various BAC levels as a function
o f w hether the subject is a heavy or m oderate drinker. It can be seen that the experienced
drinker is m ore resistant to the effects o f alcohol.
It m ight be asked w hether the existence o f chronic tolerance in the heavy drinker would
affect the developm ent o f acute tolerance, as expressed in the M ellanby phenom enon.
 The Mellanby Effect in Moderate and Heavy Drinkers 187

-A HEAVY
30

20-
Cd LU
UJ >
CD O

10 -
O
A

H------ 1------ 1------ h H------ 1------ 1------ H H------ h

30-

20 ----O- --- O'

a n t e r io r / p o s t e r io r sway
10 ■I--------- 1--------- 1--------- 1---------- 1----------1----------1----------1----------1----------1

5-
--O'

O
C
OD A
STANDING
HAND-STEADINESS
-I------ 1------ 1------ 1

hO 15-1
<

<
-J
Q_
L
C
U
O
10-
o
cCLn
A
Li_
O O
CO
Q
z
o HAND-STEADINESS
o
--- 1------ +— —I—
PRE- 0.0 .01 .02 .03 .04 .05 .06 .07 .08 .09 .10 .12 .13 .14
TEST
BLOOD ALCO HO L CONCENTRATION %

Figure 2 Performance o f heavy and moderate drinkers as a function o f BA C.

Figure 3 shows perform ance during the rising and falling BAC phases separately for
heavy and m oderate drinkers on one o f the measures, i.e. lateral sway. The results here are
typical o f all the perform ance measures. It can be seen that there is as m uch or greater develop­
m ent o f acute tolerance in the chronically tolerant heavy drinkers as in the non-tolerant
18S H. Moskowitz, J. LXaily and R. Henderson

m oderate drinkers. Clearly, there is evidence for the developm ent o f acute tolerance as ex­
pressed in the M ellenby phenom enon, and the acute tolerance occurs independently o f the
existence o f the chronic tolerance in the heavy drinker.

H EA V Y DRINKERS
A — ;---- A RISING

PRE- 0.0 .01 .02 .03 .04 .05 .06 .07 .08 .09 .10 .11 .12 .13 .14
BLOOD A LCO H O L CONCENTRATION %

M O D ERATE DRINKERS
40

30

!> 20

10 -

L A T E R A L SWAY

0 H------------1----------- 1----------- 1------------1----------- 1------------1------------1----------- 1----------- 1----------- 1------------1----------- 1----------- 1------------1----------- 1----------- 1
PRE- 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 .10 .11 12 13 14
TEST
BLOOD ALCO H O L CONCENTRATION %

Figure 3 Rising and falling BA C performance o f heavy and moderate drinkers on Lateral
Sway.

T he results o f this aspect o f the study are in conflict with the suggestion offered by
Jellinek (1960) that chronic heavy drinkers would be expected to show less acute tolerance
and specifically a sm aller M ellanby effect than m oderate drinkers. This suggestion was based
on the belief that the chronic tolerance induced by heavy drinking would have protected
the drinker from some o f the im pairing effects o f alcohol from the very start o f the drinking
session. The results herein obtained conform more closely with the expectations o f the theory
o f tolerance developed by LeBlanc, K alant, LeBlanc, and G ibbins.2 They propose that the
result o f the developm ent o f tolerance by heavy chronic drinking is a change in the rate
and degree o f final am ount o f acute tolerance exhibited at each drinking session, in com­
parison with that shown by a naive or m oderate drinker. Thus both m oderate and heavy
drinkers would begin to exhibit behavioral im pairm ent at approxim ately the same threshold
level in the rising BAC curve. However, the rate of increase in im pairm ent for the heavy
 The Mellanby Effect in Moderate and Heavy Drinkers 189

drinker would be slower and reach a lower level at a given BAC level than for a m oderate
drinker.
It might be wondered why the M ellanby effect is quite small in size in this study, in
comparison with larger differential effects for rising and falling BAC curves reported in other
studies. It was our hypothesis that this was due to the difference in the rate o f adm inistration
o f alcohol. To test this, Moskowitz and Burns3 perform ed an additional study in which per­
formance decrem ent under alcohol was exam ined as a function o f the rate o f drinking for
five different groups. The study determ ined that the more rapid the rate o f drinking the
greater the degree o f perform ance decrements at the same BAC level.

REFERENCES

1. Kalant, H., LeBlanc, A. E., and Gibbins, R. J. Tolerance to, and dependence on, some non-opiate
psychotropic drugs. Pharmacological Review 23, 135-191, 1971.
2. Mellanby, E., Alcohol: its absorption into and disappearance from blood under different con­
ditions. British Medical Research Committee, Special Report Series, No. 31. London: H.M.S.O.,
1919.
3. Moskowitz, H. and Burns, M., Effects of rate of drinking on human performance. Journal of
Studies on Alcohol 37, 598-605, 1976.