Health Psychology Copyright 2008 by the American Psychological Association

2008, Vol. 27, No. 1, 34 – 42 0278-6133/08/$12.00 DOI: 10.1037/0278-6133.27.1.34

The Relationship Between Cardiac Reactivity in the Laboratory

and in Real Life

Derek W. Johnston Martti T. Tuomisto

University of Aberdeen University of Tampere and Tampere University Hospital

Geoffrey R. Patching
University of Stockholm

Objective: An excessive cardiovascular response to acute stress is a probable risk factor for cardiovas-
cular (CV) disease. Such reactivity is usually assessed from the CV response to laboratory stressors.
However, if it is a risk factor, correlated responses must occur in real life. Design: In the present study,
we investigated the relationship between the heart rate (HR) response to five laboratory stressors and HR
reactivity in the field. Measures: HR variation, the response to a real life stressor (public speaking), and
the increase in HR with periods of self-reported tense arousal. Ambulatory HR, activity and posture were
measured continuously over a 7-hr period. Results: The HR increase to laboratory stressors did not relate
to HR variation consistently, but it did relate to the other two field measures. Conclusion: The results
suggested that a tendency to increased HR reactivity may be a risk factor for cardiovascular disease when
combined with exposure to stress.

Keywords: heart rate, cardiac reactivity, laboratory stressors, real life stressors, public speaking

An excessive cardiovascular response to acute stress, cardiovas-
cular hyper-reactivity, is a probable risk factor for cardiovascular
disease (Treiber et al., 2003) and may be one mechanism linking
various behavioral and psychological risk factors such as hostility
to cardiovascular disease (Smith & Ruiz, 2002). Hyperreactivity is
often assessed by measuring the cardiovascular (CV) response to
controlled laboratory stressors such as mental arithmetic and video
games (Johnston, Anastasiades, & Wood, 1990), public speaking
tasks (Kamarck, Schwartz, Janicki, Shiffman, & Raynor, 2003), or
a battery of such tasks (Tuomisto, 1997). If such reactivity is a
cause of disease, it cannot just be a laboratory phenomenon. The
responses seen in the laboratory must capture a characteristic of
the individual that occurs in real life either with sufficient fre-
quency to cause, for example, persistent elevations in blood pres-
sure or arterial damage or perhaps with sufficient force to trigger
a myocardial infarction or cerebrovascular accident in those with
compromised arteries (Johnston, 2002). For almost two decades
researchers have attempted to determine if CV reactivity seen in
the laboratory relates to CV reactivity in real life, i.e., extra-
Derek W. Johnston, School of Psychology, University of Aberdeen;
Martti T. Tuomisto, Department of Psychology, University of Tampere and
Department of Psychiatry, Tampere University Hospital; and Geoffrey R.
Patching, Department of Psychology, University of Stockholm.
The research reported in this article was supported by a grant from the
Medical Research Council to Derek Johnston.
We are grateful to Dr. Julia Hay for useful comments on an earlier
version of the manuscript.
Correspondence concerning this article should be addressed to Derek W.
Johnston, School of Psychology, College of Life Sciences and Medicine,
William Guild Building, University of Aberdeen, Aberdeen AB24 2UB,
Scotland. E-mail: d.johnston@abdn.ac.uk
laboratory settings; however, the relationships observed are often
weak or inconsistent (Turner et al., 1994). While not enabling a
definitive conclusion, this research has highlighted the various
issues which if systematically explored might clarify the relation-
ship between reactivity in the laboratory and in the field. The most
critical of these issues appears to be the nature and specificity of
the stressors studied in the laboratory and the field.
The Nature of the Laboratory Stressor
Few researchers are explicit about their choice of laboratory
stressors, but it is likely that most sample a range of disparate tasks
in the hope that they will include in the laboratory session tasks
that are analogous to activities that participants carry out in real
life or tap psychophysiological processes that occur frequently in
real life. There are at least four methods used to relate laboratory
and field reactivity. In early studies (reviewed by Turner et al.,
1994), investigators typically correlated the CV response to each
of a variety of laboratory stressors with their preferred measure of
CV reactivity in real life. In such studies no one task was assigned
priority over the others, and it was presumably the investigators’
expectation that the response to each task will correlate equally
well with the field measure.
A second approach has been taken by Kamarck, Debski, and
Manuck (2000) who have argued on essentially psychometric
grounds that a more reliable, and perhaps more representative,
measure of reactivity is obtained by aggregating responses over
tasks and task repetitions. They find such an aggregated measure
relates more reliably to field measures. Thirdly, Johnston et al.
(1990) have proposed that the response to particular stressors or
classes of stressors best predicts responsiveness in the field. They
showed that the response to what Obrist (1981) termed “active

34
 CARDIAC REACTIVITY IN THE LABORATORY AND IN REAL LIFE
coping tasks” predicted HR responsiveness in the field, but the
response to passive coping tasks did not, and they generalized from
this. However, there have been few systematic tests of their posi-
tion. Finally, the actual task may be unimportant as long as it
provokes the relevant physiological response. For example, van
Doornen and van Blokland (1992) maintain that the cold pressor
task is a strong predictor of real life responses because it provokes
a strong noradrenergic response, which they presume also under-
lies responsiveness in real life.
Measuring Responsiveness in Field Settings
In addition to these four methods of assessing relevant labora-
tory reactivity, there are at least three methods of measuring
responsiveness in the field that have been related to laboratory
reactivity. The most general method is from some measure of CV
variation (Floras, Hassan, Jones, & Sleight, 1987; Harshfield,
James, Schlussel, Yee, Blank, & Pickering, 1988; Johnston et al.,
1990; Kamarck et al., 2003). Such measures are based largely on
the assumption that while CV variation is multiply determined,
hyperresponsive individuals will respond more or more often to
everyday stressors and therefore show increased CV variation.
This measure has the strength of potentially capturing the effects
of stress over long periods of measurement. However, it is very
general, and the link with stress is assumed rather than directly
measured. In addition there are many other determinants of HR
variability, such as systematic variation in HR frequency that may
relate to sympathetic/parasympathetic balance (Task Force of the
European Society of Cardiology the North American Society of
Pacing Electrophysiology, 1996) and the related reduction in HR
variability predictive of recurrent coronary heart disease. We are
concerned with HR variation that is unsystematic in the frequency
and temporal domain and will use the neutral term “HR variation”
to describe it. Another approach is to examine the response to a
specific stressor in the field such as an examination (Davig,
Larkin, & Goodie, 2000; van Doornen & van Blokland, 1992) or
public speaking (Matthews, Manuck, & Saab, 1986). Such an
approach offers a much clearer specification of the field stressor
but only measures a very limited range of field situations that may
be quite similar to the laboratory tasks to which they are being
related. The final approach relies on self-reports of stress, emo-
tional arousal, or current demand. The CV responses during peri-
ods of high and low stress or demand are then related to laboratory
reactivity (Anastasiades & Johnston, 1991; Jain, Schmidt,
Johnston, Brabant, & von zur Muhlen, 1998; Kamarck et al.,
2003). This approach is less constrained than that of examining the
effects of a specific stressor, since participants are likely to en-
counter and be stressed by different events. The most obvious
deficiency of this approach is that it relies on the participant’s own
detection of stress or emotional arousal and hence the measure of
stress and the effects of stress may be confounded.
In this article, we examined the four approaches identified above
in the assessment of reactivity in the laboratory and the three
methods of characterizing reactivity in the field. In the laboratory,
we explored the predictive power of five different laboratory
stressors, including the cold pressor test, stressors that involve
active or passive coping, and a measure derived by aggregating
reactivity over all five stressors. In the field, heart rate and the
control variables of activity and posture were assessed continu-
35
ously over a 7-hour period while diary measures of tense arousal
were taken every 30 min. All participants spoke in public as part
of their usual academic training during the measurement period.
Reactivity during the day was assessed in three ways: by the
average HR variably over the complete period, through the HR
response to public speaking, and from the within subject covaria-
tion of self reported tense arousal and heart rate. In all cases,
activity and posture were controlled for. The most general predic-
tion was that participants who manifested the largest HR response
to the laboratory stressors would show the greatest HR variation
during the day, the largest HR response to public speaking, and
show a larger increase in HR with increases in self-reported tense
arousal. Based on our earlier work (Johnston et al., 1990), we
hypothesized that these effects would be strongest when related to
the HR response to active coping tasks. In contrast, Kamarck et al.
(2000) would anticipate that the average of the laboratory re-
sponses would predict ambulatory reactivity best, and if the find-
ings of van Doornen and van Blokland (1992) replicate, then one
might expect that the response to the cold pressor would be the
strongest predictor.
Method
Participants
Participants were 66 healthy male undergraduates from the
University of St. Andrews, Scotland, with a mean age 21.3 years
(range 17–32), weight 73 kg (range 57–94) and height 180 cm
(range 167–193) recruited from campus wide advertisements. Par-
ticipants were paid £12 for taking part in the study. The Ethics
Committee of the School of Psychology of St. Andrews University
approved the study, and participants gave informed consent.
Laboratory Sessions
Apparatus
The electrocardiogram (ECG), respiration, and skin conduc-
tance were recorded using a 6-channel Grass Model 7D polygraph,
digitized on a CED 1401 Plus interface (Cambridge Electronic
Design), stored and subsequently processed on an IBM compatible
486 PC. Continuous finger blood pressure was recorded using the
vascular unloading method with a Finapres recorder (Ohmeda
Medical) and digitized using the CED 1401 Plus. As the focus in
the present article is on the relationship between the HR response
to laboratory tasks and HR reactivity in the field, the report will be
confined to HR. The ECG was recorded in the modified lead two
position using standard disposable ECG electrodes (Blue sensor,
Medicotest A/C). The ECG was digitized at 1K and the occurrence
of the R-wave detected from a combined zero crossing and max-
ima detection method in Spike 2 software (Cambridge Instruments,
Ltd.) and was stored as interbeat interval (IBI) in msec and as HR
in bpm. HR is used throughout this article.
The Laboratory Tasks
The laboratory tasks were, in order of presentation: (1) tracking
task from a standard test battery (Debski et al., 1991; Kamarck et
al., 1992) lasting 5 min; (2) a computerized version of Ravens
progressive matrices, lasting 8 min (Steptoe, Fieldman & Evans,
36 JOHNSTON, TUOMISTO, AND PATCHING
1993); (3) a 6-min clip from the end of the film The Shining; (4)
a social problem-solving task in which participants were asked to
think about a minor problem in their domestic lives for 2 min and
then present possible solutions verbally for a further 2 min; and (5)
cold pressor, in which participants immersed their right hand in
heavily iced water for 1 min. A 2-min baseline preceded all tasks,
and there were additionally 5-min rest periods at the beginning and
end of the experimental period during which a relaxation tape was
played. After each task, there was a recovery period of at least 3
min during which no data were collected. The Raven’s is a typical
active coping task while the cold pressor and viewing the film are
passive coping tasks. We regard the social problem-solving task as
an active coping task while the tracing task, which has a frustrating
element, is more difficult to classify.
Procedure
The laboratory sessions took place between 9 and 11 in the
morning. The participants sat in a large over stuffed armchair in a
warm room next to the experimenter’s control room. Approxi-
mately 40 min were spent attaching electrodes and familiarizing
the participant with the experimental situation. Participants were
given verbal instructions for each task by the experimenter who
entered the room between tasks.
Ambulatory Recording
Apparatus
The ambulatory recording was carried out on a digital RM-10
recorder (Parametric Recorders, Ltd.) that consisted of a recorder
unit and event marker and plug-in modules measuring ECG, pulse
transit time, activity, and posture. The device weighs about 500 g
and was contained in a pouch attached to the participant’s belt.
ECG was recorded using the same techniques as described for the
laboratory session. Pulse transit time was derived from a photo-
electric transducer attached to the left earlobe. Pulse transit time
data are not reported in this article. The activity device consisted
of an accelerometer containing one piezoelectric strip (L/AC1).
The posture device consisted of a tube 2 mm thick and 125 cm
long filled with sterilized water. A small pressure transducer at one
end (MX900, Medex Inc.) continuously measured hydrostatic
pressure in the tube. The ECG was sampled at 500 Hz; the R-wave
is detected and HR calculated. The accelerometer was sampled at
100 Hz, rectified and averaged over 32 samples. The posture
device was sampled at 5 Hz. All the processed signals were stored
at 5 Hz. The RM-10 system and its use were described by
Tuomisto, Johnston, & Schmidt (1996).
Procedure
The participants attended the laboratory early on a day in which
they were due to present a paper at a tutorial as part of their course
of studies. Eighty percent of the ambulatory sessions occurred with
14 days of the laboratory testing (median gap 5.5 days); four were
more than 28 days. ECG electrodes were attached as described
above. The accelerometer was attached using surgical tape approx-
imately half way along the left thigh muscle. The plastic tube for
measuring posture was placed on the participant’s skin from calf to
chest and attached with surgical tape. The transducer was attached
just above the ankle with a Velcro band. The sensitivity of the
accelerometric gain was determined by simulating walking with
maximum force in the laboratory and adjusting the gain to avoid
ceiling effects but retain sensitivity. Before leaving the laboratory,
we calibrated the posture device by having the participants’ stand,
sit in a hard upright chair, and lie horizontally. The participants
were instructed in the use of a simple diary to be completed every
30 min during the day and immediately before and after they
presented at the tutorial. Diary entries were prompted auditorily by
a preprogrammed watch, which we supplied. Participants recorded
the exact time of their tutorial and their presentation and were
instructed to complete diary entries at these times. The diary
allowed the participants to record their activities, social situation,
the demands placed on them, the control they experienced and
current mood. They had to record how tired, hurried/rushed, in-
volved/interested, happy, irritable/angry, sad/depressed, and anx-
ious/tense they were on 7-point scales from “not at all” to “very
much.” Following Jain et al. (1998), we combined the scales of
hurried/rushed, irritable/angry and anxious/tense into a measure of
tense arousal. Cronbach alphas for this scale when assessed over
the first 6 hr of recording (i.e., separate alpha’s calculated every 30
min and involving at least 58 participants) averaged .67 and ranged
between .59 and .78.
Computer Analysis of Signals
The ambulatory recordings were replayed to file and edited with
the RMOS software (Parametric Recorders). Accelerometer read-
ings that were off scale were recorded as maximum values and
artifacts in the HR record removed using the RMOS editing
facilities, which detect improbably abrupt changes of HR between
beats. Missing data were replaced by linear interpolation. Artifacts
were infrequent. The data were then reduced to 60-sec means and
stored as ASCII files for subsequent analysis. The data reduction
methods are described more fully in Tuomisto et al. (1996).
Statistical Analysis
Laboratory data. All the laboratory data were examined for
artifacts (which were very rare) before the ECG derived HR was
reduced to 1-min means. The 1-min means for each stressor were
averaged to produce a mean value for each stressor and the 1-min
means for 2 min before each stressor were averaged and used as
the baseline. Preliminary analyses showed that baseline and reac-
tivity were strongly related, participants with lower HRs showing
larger HR increases to most of the stressors; we therefore used
residualized scores to free the measure of reactivity from the
baseline.
Ambulatory data. In deriving a measure of HR variation, we
used the autoregressive (AR) modeling methods we have de-
scribed in earlier papers (including Jain et al., 1998; Johnston et
al., 1990; Johnston et al., 1993, Johnston et al., 1994; Tuomisto et
al., 1996). We have found that HR series based on 1-minute means
can be adequately modeled as third order AR processes. We
carried out 4 modeling processes; univariate models incorporating
only lagged HR and multivariate models additionally incorporat-
ing concurrent and lagged activity, posture, and activity plus
posture. The resulting measures of variation are the standard
deviation of the 1-min HR means and the standard deviations after
 CARDIAC REACTIVITY IN THE LABORATORY AND IN REAL LIFE 37

allowing for the autocorrelation in the HR series and after allowing
for activity, posture and activity plus posture. The values, models
and fits were very similar to those reported in our earlier papers
(see above) and will not be presented. The periods when the
participants spoke during the tutorial were identified from the
times recorded in the participants’ diaries and the marker entries
onto the RM10 recorder. If a participant had not marked the exact
period when they spoke then the total tutorial was taken as the
stressful event. HR, activity and posture averaged over the com-
plete stressful episode was calculated. It was important to deter-
mine a control period during the day with which the stressful
period of speaking could be compared. This was done by searching
each participant’s record visually for periods that included a diary
entry and had similar activity, posture and duration as the tutorial
period. This search was done on the computer display of the
ambulatory records, excluding HR so the analyst could not be
biased towards selecting control periods of low HR (and hence
overestimating the effects of the tutorial). For the analysis of the
relationship between tense arousal and HR the average HR, activ-
ity and posture for the 5 minutes prior to each diary entry were
calculated.
Results
The means for heart rate during the stressors and immediately
preceding baselines are shown in Table 1 in the order that the tasks
were completed. All the stressors produced a change in HR, with
the exception of the initial 2 min of the social problem-solving
task, during which the participants considered possible solutions to
the problem that they had identified. As this had no effect on HR
and was not independent of the later section of the task, it will not
be considered further. It will be noted that watching the film was
associated with a decrease in HR. HR and HR variation during the
ambulatory period are shown in Table 2.
Following Kamarck et al. (2000), we averaged the response to
the battery of stressors. However, before using such a score, it is
advisable to confirm that the responses can be legitimately
grouped. The residualized HR response to the five tasks was
analyzed using principal components analysis. While such an
analysis should at best be regarded as suggestive on such a small
sample, it showed that one factor could be extracted on which all
the tasks loaded. The loadings were Ravens matrices .85, Tracing
.84, Problem Solving .71, Cold Pressor .62, and the Film .46. This
factor, of eigenvalue 2.52, explained 51% of the variance. The
Table 2
Heart Rate and Heart Rate Variation During a 7-Hr Weekday
Period
(SD)
Duration of measurement period (min) 436.1 (47.4)
Mean heart rate 85.1 (10.6)
Mean SD of HR 13.5 (3.1)
Mean residual allowing for autoregression (Res1) 7.0 (1.1)
Mean residual allowing for autoregression and 5.7 (0.9)
activity (Res2)
Mean residual allowing for autoregression and 6.1 (1.0)
posture (Res3)
Mean residual allowing for autoregression, activity, 5.2 (0.8)
and posture (Res4)
Cronbach’s Alpha (.75) for the simple average of the five re-
sponses was satisfactory.
The relationship between the response to the laboratory stressors
and HR and HR variation is shown in Table 3 for both the
individual tasks and the average response to the five tasks. Based
on earlier findings and associated theory, we had hypothesized that
the tasks involving active coping, (the Ravens Matrices and Prob-
lem Solving), would be associated with field measures of reactivity
while the tasks involving passive coping, the film and the cold
pressor, would not. The results clearly do not support these pre-
dictions, as the response to cold pressor is the only response to
predict field HR variation, which it does with some reliability
across all the measures of HR variation. The average response to
the laboratory stressors did not relate to field measures in a
convincing way.
In the field, HR was measured during a specific real life stressor,
a tutorial presentation, and during a control period. The mean
values for heart rate, activity, and posture are shown in Table 4. It
can be seen that the stress (tutorial) and control periods are well
matched on duration, activity, and posture. As expected, HR and
anxiety are markedly higher during the tutorial presentation. The
correlations between the HR response to the laboratory stressors
and the HR increase during the tutorial stressor, expressed as a
Table 3
Correlation Between Heart Rate Responsivity to Laboratory
Stressors and Heart Rate Variation in the Field

Field measures

Laboratory task HR SD HR Res1 Res2 Res3 Res4

Table 1
T Value Tests Difference Between Task and Baseline Level Tracing .03 .11 .01 .02 .04 .06
Ravens ⫺.04 .09 .04 ⫺.03 .10 .04
Task Baseline (SD) Task (SD) t Film ⫺.01 .25 .10 .10 .17 .14
Problem solving .16 .02 .13 .20 .14 .23
Tracing 70.21 (9.90) 72.71 (9.52) 3.32 Cold pressor .17 .35** .27 *
.33* .30* .34**
Ravens matrices 70.84 (9.86) 72.59 (9.48) 2.58 Average response .09 .24 .16 .19 .22 .24
Film 69.96 (9.92) 67.46 (9.05) ⫺4.63
Problem solving (thinking) 69.01 (9.34) 69.19 (9.27) 0.38 Note. HR ⫽ mean ambulatory heart rate; SD HR ⫽ SD of ambulatory
Problem solving (speaking) —* 75.27 (9.39) 10.68 heart rate; Res1 ⫽ residual heart rate after allowing for autoregression;
Cold pressor 68.17 (9.42) 72.94 (8.20) 5.43 Res2 ⫽ residual heart rate after allowing for autoregression and activity,
Average response 69.64 (9.47) 72.71 (8.41) 5.01 Res3 ⫽ residual heart rate after allowing for autoregression and posture;
Res4 ⫽ residual heart rate after allowing for autoregression, activity and
Note. Mean baseline and task heart rates (bpm) SD in parentheses. posture.
*
The same baseline was used for both aspects of the problem-solving task. *
p ⬍ .05. ** p ⬍ .01.
38 JOHNSTON, TUOMISTO, AND PATCHING

Table 4 Table 6
Heart Rate, Activity, Posture, and Anxiety During a Tutorial Regression of HR on Tense Arousal and Tense Arousal ⫻
Presentation and a Control Period Reactivity

Measure Control (SD) Tutorial (SD) t Arousal t(57) Arousal ⫻

Task coefficient coefficient Reactivity t(57)
Duration (min) 16.2 (12.7) 19.2 (14.6) 1.76 ns ( p ⬍ .08)
***
Activity 2.40 (3.70) 2.62 (3.97) 0.53 ns Tracing .82 4.56 .00 .13
Posture 122.00 (18.9) 124.7 (19.21) 1.69 ns ( p ⬍ .10) Ravens matrices .82 4.51*** .02 .55
Heart rate (bpm) 79.83 (14.15) 92.09 (16.43) 7.08 p ⬍ .001 Film .83 4.62*** .04 1.00
Anxiety (1–7 scale) 1.22 (1.29) 3.97 (1.70) 11.05 p ⬍ .001 Problem solving .82 4.79*** .08 2.67**
Cold pressor .83 4.95*** .08 3.27**
Average .84 4.86*** .10 2.44*
*
p ⬍ .05. **
p ⬍ .01. ***
p ⬍ .001.
residualized change score from the control period, are shown in
Table 5. All the correlations were positive, and the correlations
between the response to tracing and the cold pressor were signif-
icant. The largest correlation was with the average response to the The relationship between arousal and HR and the moderating
five laboratory stressors. The HR response to a number of labo- effect of reactivity on that relationship are shown in both the
ratory stressors predicts the HR response to a real-life public- simple case when only tense arousal and reactivity were consid-
speaking stressor. There was no support for the prediction that ered and when activity and posture were taken into account. Tense
stressors involving active coping would be the strongest predictor arousal related to HR, with increases in tense arousal being asso-
of the HR response to public speaking. ciated with higher HR. This effect was moderated by the HR
The final field test was the relationship between self-reported response to problem solving, the cold pressor, and the average
emotional arousal and HR. We predicted that participants who response to all the stressors. These effects are shown in Figures 1,
produced large responses to laboratory stressors would show 2, and 3 where it can be seen that the increase in HR associated
greater increases in HR in periods of increased emotional arousal. with an increase in tense arousal is greater in the more reactive
The relationship between reactivity and the regression of HR on participants. This effect is most pronounced in relation to the
emotional arousal was tested using hierarchical linear modeling response to the cold pressor. Activity and posture, as expected, are
(Raudenbush & Bryk, 2002) to test what Raudenbush and Bryk strongly related to HR. Tense arousal is still related to HR when
term a “slopes-as-outcomes” model. Using the procedure HLM activity and posture are allowed for by entering them in the model
(Hierarchical Linear Models) of the program HLM5 (V5.05; and while reactivity still moderates that relationship the effects are
Raudenbush, Bryk, Cheong, & Congdon, 2001), we fitted two slightly altered suggesting that some of the increase in HR that is
level models. Level 1 consisted of repeated measures of tense linked to reactivity may be related to metabolic processes involved
arousal, HR, activity, and posture. Level 2, within which Level 1 in physical activity.
was nested, was the residualized HR response to the laboratory
stressors. All variables were entered uncentered. The relationship Discussion
between reactivity and the slope of tense arousal on HR was tested
The cardiovascular response to laboratory stressors may index a
both in isolation and when concurrent activity and posture were
general tendency towards an enhanced CV response to stress that
controlled for by entering them into the model as covariates. As
places individuals at increased risk of CV disease or its clinical
before the predictive power of the responses to the individual tasks
consequences (Tuomisto, Majahalme, Kähönen, Fredrikson, &
and the average response was tested. Fifty-nine participants pro-
Turjanmaa, 2005). If this is indeed the case, the CV response to
vided 790 usable data points for this analysis, i.e. approximately 13
laboratory stressors should relate reliably to CV reactivity in the
data points per participant with a range of 4 to 17. The results of
field. In this study, we found that the HR response to most of
the multilevel modeling are shown in Tables 6 and 7.
laboratory stressors studied did not relate to field measures of HR
variation, a proxy for HR reactivity in the field. An exception was
the HR response to the cold pressor test. The average response to
Table 5 the laboratory stressors and to several of the individual stressors
Relationships Between Heart Rate Response to Laboratory did relate to the HR response to a real life stressor, presenting a
Stressors and a Real Life Stressor, Presenting a Paper in a paper orally in a tutorial. In addition, the increase in heart rate
Tutorial associated with increases in self-reported tense arousal was greater
in participants with larger heart rate responses to several laboratory
Stressor Correlation tasks and the average of these tasks.
Tracing .29* The present study enables one to examine a number of rather
Ravens matrices .21 different conceptions of the relationship between the CV responses
Film .19 to laboratory challenge and different measures of reactivity in real
Problem solving .21 life. These shall be considered separately for the three field mea-
Cold pressor .27*
Average response .33**
sures. The most general position is that the response to any
laboratory stressor should relate to all measures of CV reactivity in
*
p ⬍ .05. **
p ⬍ .01. the field. The data on HR variation clearly do not support that
 CARDIAC REACTIVITY IN THE LABORATORY AND IN REAL LIFE 39

Table 7
Regression HR on Tense Arousal, Tense Arousal ⫻ Reactivity, Activity, and Posture

Arousal ⫻
Arousal t(57) Reactivity t(57)
Task coefficient coefficient coefficient coefficient Activity t(58) Posture t(58)

Tracing .69 5.02*** .04 1.72 .42 8.42*** .20 8.23***

Ravens matrices .67 5.00*** .05 2.70** .42 8.39*** .20 8.30***
Film .68 5.00*** .04 1.22 .42 8.36*** .20 8.21***
Problem solving .68 5.01*** .06 2.32* .42 8.34*** .24 8.28***
Cold pressor .68 5.03*** .03 1.65 .42 8.41*** .20 8.05***
Average .69 5.22*** .09 2.82** .41 8.35*** .20 8.32***
*
p ⬍ .05. **
p ⬍ .01. ***
p ⬍ .001.

view. Only the HR response to the cold pressor relates to measures
of HR variation.
A much more probable view is that the response to a particular
class of laboratory tasks uniquely relates to reactivity in the field.
The most widely used analysis by task is the active versus passive
coping distinction proposed many years ago by Obrist (1981). We
(Johnston et al., 1990) hypothesized that the HR response in the
field related most strongly to the HR response to active coping
tasks in the laboratory. This was tested in this study by comparing
tasks with active coping features, a computerized version of
Ravens Progressive Matrices and a social problem-solving task
and two passive coping tasks, watching a harrowing film and the
cold pressor test. There was no support for the hypothesized link
between active coping tasks and HR variation, indeed the only
reliable predictor of HR variation in the field was the response to
the cold pressor, the gold standard passive coping task. Including
the present investigation, we have now examined the relationship
between active coping and field measures of HR variation in seven
separate data sets. In two studies of healthy volunteers, the HR
response to active coping tasks related reliably to HR variation in
the field (Johnston et al., 1990, 1994) but in two of them it did not
(Johnston et al., 1993). Support for the hypothesized link was also
obtained in a studies measuring rate pressure product (Jain et al.,
1998) and HR in patients with panic disorder (Anastasiades et al.,
1990) in which the stressors involved both active coping and
anxiogenic tasks. Measures of HR variation are affected by many
processes that are likely to be unrelated to the response to stress,
such as diurnal variations in HR and the effects of metabolic
activity and posture. We have made allowances for these using an
autoregressive model incorporating the simultaneous measurement
of activity and posture. In some of our studies, this has strength-
ened the relationship (Anastasiades et al., 1990; Jain et al., 1998;
Johnston et al., 1990, 1994), but it did not do so in the present
study.
In addition to the unique responses to individual stressors, the
CV response to laboratory stressors can be considered as an
undifferentiated whole, as Kamarck et al. (2003) essentially do
when arguing for averaging responses across many stressors. Fac-
tor analysis confirmed that it was appropriate to group the response
to the five tasks in the present study. However, averaged reactivity
was no more convincingly related to HR variation than the re-
sponse to the individual tasks and less so than the relationship to
the cold pressor task.
The findings in relation to a specific real life stressor, presenting
a paper in a tutorial, were different. The responses to all the
laboratory tasks were positively correlated with the HR increase
associated with the tutorial and the responses evoked by tracing
and the cold pressor were reliably so. However, the correlations
involving responses to the two active coping tasks were not sig-
nificant despite speaking in a tutorial having many of the features
of an active coping task. The average response across the five tasks
related reliably to the response to public speaking. This suggests

Figure 1. Relationship between Tense Arousal and Heart Rate for par- Figure 2. Relationship between Tense Arousal and Heart Rate for par-
ticipants in top (dotted line) and bottom (dashed line) quartile for HR ticipants in top (dotted line) and bottom (dashed line) quartile for HR
response to problem solving and the average participant (solid line). response to cold pressor and the average participant (solid line).
40 JOHNSTON, TUOMISTO, AND PATCHING
Figure 3. Relationship between Tense Arousal and Heart Rate for par-
ticipants in top (dotted line) and bottom (dashed line) quartile for average
reactivity and the average participant (solid line).
that the relationship between laboratory and real life responses is
stronger or more consistent when there is some evidence that the
HR response in real life is definitely stress related. The relationship
between laboratory and field was quite substantial. Reactivity was
treated as a continuous variable in the analyses reported. However,
if the participants are split at the median on average reactivity, then
low reactors show an elevation of 7 bpm during the tutorial, high
reactors 17 bpm. The relationship between HR responses in the
laboratory and a specific real-life stressor are variable across
studies. Positive relationships have been reported in some (Davig
et al., 2000; Kamarck et al., 2000; Matthews et al., 1986; Turner,
Carroll, Dean, & Harris, 1987; van Doornen & van Blokland,
1992) but not all studies (Turner, Girdler, Sherwood, & Light,
1990; Warwick-Evans, Walker, & Evans, 1988), and as Kamarck
et al. (2000) point out, the relationships are often specific to
particular tasks and cardiovascular measures. Davig et al. (2000)
did not find that aggregating across five laboratory tasks improved
the correlations with the CV response to proposing or defending a
thesis.
The final field measure examined in the present study was the
relationship between self-reported tense arousal and HR. Tense
arousal, the sum of ratings of feelings of hurry and rush, irritation
and anger, and anxiety and tension, related strongly to HR in the
field. However, the findings were quite variable when considering
the individual laboratory tasks with only the response to problem
solving and the cold pressor convincingly moderating the tense
arousal relationship. As well as disconfirming the view that all
tasks provoke a response that relates to field reactivity, the present
result provides no support for the primacy of active coping since
the significant results involve one active and one passive task.
There is support for Kamarck’s position, however, since average
HR reactivity moderates this measure of reactivity in the field.
While the main effect of tense arousal on HR was not reduced
when variations in activity and posture were controlled for, the
moderating effects of reactivity were inconsistently effected, but
still evident. In two earlier studies, we examined HR when partic-
ipants reported distress or psychological arousal in real life, in each
case high reactors produced greater increases in HR (Anastasiades
& Johnston, 1991) or a measure incorporating HR (Jain et al.,
1998). These earlier studies simply averaged the HR levels from
periods of high or low distress or arousal. This ignores a great deal
of information and involves some uncomfortable assumptions
about the data being averaged. In this article, multilevel modeling
was used to test the relationship between the aggregate score of
several self report scales that assessed aspects of tense arousal and
HR. Such modeling is a much more powerful and defensible
method of examining the relationship between simultaneously
assessed behavioral or subjective states and physiological re-
sponses that are thought to covary within individuals over time.
Tense arousal and HR related convincingly within participants and
this relationship was moderated by CV reactivity. Kamarck et al.
(2003) have used similar methods to examine the relationship
between laboratory reactivity HR and blood pressure response to
demanding events in daily life. They also found that high reactors
to laboratory challenge showed greater cardiovascular responses to
demanding events in their life. The relationship between tense
arousal and HR remained significant when activity and posture
were controlled for.
The final position to consider is that one particular stressor
relates strongly and uniquely to field reactivity. Van Doornen and
van Blokland (1992) proposed that the CV response to the cold
pressor is predictive of real life stress responses, although that is
contradicted by the findings of other studies in the present series
(Johnston et al., 1990, 1993). Nevertheless, in this study the
response to the cold pressor was the only laboratory measure to
relate to all field measures and it did so with reasonable power in
most cases. Van Doornen and van Blokland (1992) maintained that
the common feature of the cold pressor and reactivity in the field
was strong noradrenergic drive. However, while the cold pressor
does have substantial noradrenergic effects, it is not clear why they
consider noradrenergic drive the most likely common mechanism
since the field stressor they studied did not have a particularly
large effect on plasma noradrenaline. In addition the principal
components analysis carried out in the present study did not
suggest that the HR reactivity evoked by the cold pressor was
markedly different from that evoked by the other tasks studied.
The presentation of a paper to a tutorial group was a very
convenient and effective real life stressor. In many UK universities
such presentations are reasonably frequent, usually one or two for
each course per semester, and while not necessarily assessed are
sufficiently important to cause some anxiety without being so
critical that students might be reluctant to take part in ambulatory
studies during them. Finding an appropriate comparison period for
ambulatory stressors is often difficult. We focused on a period of
similar duration in which the subject had a similar level of activity
and the same posture. We were broadly successful in our match-
ing, and it was clear that anxiety and heart rate were reliably higher
when presenting at the tutorial. In real life it is, of course, not
possible to achieve that same level of control as in the laboratory.
Tutorials can take rather different forms with different tutors and
in different subjects and can vary from students standing to read
out a paper, or discussing a series of slides in an art history class,
to presenting a translation or having a fairly informal conversation
with the tutor and fellow students. Such presentations can vary
greatly in duration and in addition, we had to rely on the partici-
pants signaling when they were actively contributing to the tuto-
rial. The recorded durations lasted from four to 82 minutes. It is
very likely that some participants failed to record the duration
accurately and the long times are overestimates. If this period
 CARDIAC REACTIVITY IN THE LABORATORY AND IN REAL LIFE
included the time after the student’s active involvement had ceased
this might well mean that a period of low stress, and hence lower
HR, was included in the stress period. Despite this HR was
elevated over the tutorial period. While we were able to match the
control and tutorial periods for general activity and posture, we
were unable to control for the possibly differing nature of the tasks
that our participants were undertaking in control and tutorial
periods. It is certain that they were speaking during tutorial and
this may not have been the case in the control period. This could
contribute to the elevations in HR, but this could not provide an
explanation of the link to the laboratory response to stress because
speech was not involved in most of the laboratory stressors. Indeed
the HR response to only laboratory stressor involving speech,
social problem solving, did not relate to the HR response to the
tutorial. We consider that the oral presentation to a tutorial group
is a useful real life stressor that is common, somewhat stressful and
offers a reasonable measure of control.
Overall, the results of the present study suggest that the HR
response to some laboratory stressors does relate to stress related
HR reactivity in real life. There is some support for aggregating the
response to a variety of stressors to provide a more robust measure
of laboratory reactivity as advocated by Kamarck et al. (2000) but
no support was found for Johnston et al.’s (1990) position that
active coping tasks provoke a CV response that relates most
strongly to reactivity in the field. The apparent effectiveness of the
cold pressor in provoking a HR response that relates all the
measures of reactivity in the field, while anticipated by van
Doornen and van Blokland (1992), was rather surprising and may
repay further study. With the exception of the cold pressor, the
laboratory measures related most clearly to field measures that
were specifically associated with objective or self-reported stress.
This may suggest that most laboratory stressors will relate to the
occurrence of arterial disease as a function of both the hyperreac-
tivity that some individuals display to stress and their exposure to
stress in daily life. The cold pressor, the most physical of the
challenges used in this study, may relate to disease by tapping a
reactivity that is not so tied to stress and is active in the many
situations that increase HR in real life.
References
Anastasiades, P., Clark, D. M., Salkovskis, P. M., Middleton, H.,
Hackman, A., Gelder, M. G., et al. (1990). Psychophysiological re-
sponses in panic and stress. Journal of Psychophysiology, 4, 331–339.
Anastasiades, P., & Johnston, D. W. (1991). Models relating cardiovascu-
lar responses in the laboratory and in real life. In M. Johnston, M.
Herbert, & T. Marteau (Eds), Proceedings of the 4th Annual Conference
of the European Health Psychology Society (pp. 6 –7). Leicester, En-
gland: Health Psychology Section, British Psychological Society.
Davig, J. P., Larkin, K. T., & Goodie, J. L. (2000). Does cardiovascular
reactivity to stress measured in the laboratory generalize to thesis and
dissertation meetings among doctoral students. International Journal of
Behavioural Medicine, 7, 216 –235.
Debski, T. T., Kamarck, T. W., Jennings, J. R., Young, L. W., Eddy, M. J.,
& Zhang, Y. X. (1991). A computerized test battery for the assessment
of cardiovascular reactivity. International Journal of Biomedical Com-
puting, 27, 277– 89.
Floras, J. S., Hassan, M. O., Jones, J. V., & Sleight, P. (1987). Pressor
responses to laboratory stresses and daytime blood pressure variability.
Journal of Hypertension, 5, 715–719.
Harshfield, G. A., James, G. D., Schlussel, Y., Yee, L. S., Blank, S. G., &
41
Pickering, T. G. (1988). Do laboratory tests of blood pressure reactivity
predict blood pressure changes during everyday life? American Journal
Hypertension, 1, 168 –174.
Jain, A., Schmidt, T. F. H., Johnston, D. W., Brabant, G., & von zur
Muhlen, A. (1998). The relationship between heart rate and blood
pressure reactivity in the laboratory and in the field: Evidence using
continuous measures of blood pressure, heart rate and physical activity.
Journal of Psychophysiology, 12, 362–275.
Johnston, D. W. (2002). Acute and chronic psychological processes in
cardiovascular disease. In K. W. Schaie, H. Leventhal, & S. L. Willis
(Eds.), Effective health behavior in older adults (pp. 55– 64). New York,
NY: Springer.
Johnston, D. W., Anastasiades, P., & Wood, C. (1990). The relationship
between cardiovascular responses in the laboratory and in the field.
Psychophysiology, 27, 34 – 44.
Johnston, D. W., Schmidt, T., Vagt, S., McSorley, K., Albus, C.,
Klingmann, I., et al. (1994). The relationship between cardiovascular
reactivity in the laboratory and heart rate responsiveness in real life:
Active coping and ␤ blockade. Psychosomatic Medicine, 56, 369 –376.
Johnston, D. W., Vogele, C., Anastasiades, P., Kitson, C., McSorley, K., &
Steptoe, A. (1993). The relationship between heart rate responsiveness in
the laboratory and in the field: Two studies. Journal of Psychophysiol-
ogy, 7, 217–229.
Kamarck, T. W., Debski, T. T., & Manuck, S. B. (2000). Enhancing the
laboratory-to-life generalizability of cardiovascular reactivity using mul-
tiple occasions of measurement. Psychophysiology, 37, 533–542.
Kamarck, T. W., Jennings, J. R., Debski, T. T., Glickman-Weiss, E.,
Johnson, P. S., Eddy, M. J., et al. (1992). Reliable measures of behav-
iorally evoked cardiovascular reactivity from a PC-based test battery:
Results from student and community samples. Psychophysiology, 29,
17–28.
Kamarck, T. W., Schwartz, J. E., Janicki, D. L., Shiffman, S., & Raynor,
D. A. (2003). Correspondence between laboratory and ambulatory mea-
sures of cardiovascular reactivity: A multilevel modeling approach.
Psychophysiology, 40, 675– 683.
Matthews, K. A., Manuck, S. B., & Saab, P. G. (1986). Cardiovascular
responses of adolescents during a naturally occurring stressor and their
behavioural and psychophysiological predictors. Psychophysiology, 23,
198 –209.
Obrist, P. A. (1981). Cardiovascular psychophysiology. New York: Aca-
demic Press.
Raudenbush, S., & Bryk, A. (2002). Hierarchical linear models: Applica-
tions and data analysis methods. Thousand Oaks, CA: Sage.
Raudenbush, S., Bryk, A., Cheong, Y. F., & Congdon, R. (2001). HLM5
Hierarchical linear and nonlinear modeling. Lincolnwood, IL: Scien-
tific Software International.
Smith, T. W., & Ruiz, J. M. (2002). Psychosocial influences on the
development and course of coronary heart disease: Current status and
implications for research and practice. Journal of Consulting and Clin-
ical Psychology, 70, 548 –568.
Steptoe, A., Fieldman, G., & Evans, O. (1993). An experimental study of
the effects of control over work pace on cardiovascular responsivity.
Journal of Psychophysiology, 7, 290 –300.
Task Force of the European Society of Cardiology the North American
Society of Pacing Electrophysiology (1996). Heart rate variability: Stan-
dards of measurement, physiological interpretation and clinical use.
Circulation, 93, 1043–1065.
Treiber, F. A., Kamarck, T., Schneiderman, N., Sheffield, D., Kapuku, G.,
& Taylor, T. (2003). Cardiovascular reactivity and development of
preclinical and clinical disease states. Psychosomatic Medicine, 65,
46 – 62.
Tuomisto, M. T. (1997). Intra-arterial blood pressure and heart rate reac-
tivity to behavioral stress in normotensive, borderline, and mild hyper-
tensive men. Health Psychology, 16, 554 –565.
42 JOHNSTON, TUOMISTO, AND PATCHING

Tuomisto, M. T., Johnston, D. W., & Schmidt, T. F. H. (1996). The
ambulatory measurement of posture, thigh acceleration, and muscle
tension and their relationship to heart rate. Psychophysiology, 33, 409 –
415.
Tuomisto, M. T., Majahalme, S., Kahonen, M., Fredrikson, M., &
Turjanmaa, V. (2005). Psychological stress tasks in the prediction of
blood pressure level and need for antihypertensive medication: 9 –12
years of follow up. Health Psychology, 24, 77– 87.
Turner, J. R., Carroll, D., Dean, S., & Harris, M. G. (1987). Heart rate
reactions to standard laboratory challenges and a naturalistic stressor.
International Journal of Psychophysiology, 5, 151–152.
Turner, J. R., Girdler, S. S., Sherwood, A., & Light, K. C. (1990).
Cardiovascular responses to behavioral stressors: Laboratory-field gen-
eralization and inter-task consistency. Journal of Psychosomatic Re-
search, 34, 581–589.
Turner, J. R., Ward, M. M., Gellman, M. D., Johnston, D. W., Light, K. C.,
& van Doornen, L. J. P. (1994). The relationship between laboratory and
ambulatory cardiovascular activity: Current evidence and future direc-
tions. Annals of Behavioral Medicine, 16, 12–23.
van Doornen, L. J. P., & van Blokland, R. W. (1992). The relationship
between cardiovascular and catecholamine reactions to laboratory and
real-life stress. Psychophysiology, 29, 173–181.
Warwick-Evans, L., Walker, J., & Evans, J. (1988). A comparison of
psychologically induced cardiovascular reactivity in laboratory and
natural environments. Journal of Psychosomatic Research, 32, 493–
504.

Correction to Jacobsen et al (2007)

In the article, “Systematic Review and Meta-Analysis of Psychological and Activity-Based Inter-
ventions for Cancer-Related Fatigue” by Paul B. Jacobsen, Kristine A. Donovan, Susan T.
Vadaparampil, and Brent J. Small (Health Psychology, 2007, Vol. 26, No. 6, pp. 660-667), the text
directing readers to view supplementary materials online was omitted.
Supplementary materials to this article may be viewed at: http://dx.doi.org/10.1037/0278-
6133.26.6.660.supp

DOI: 10.1037/0278-6133.27.1.42