- Email: [email protected]
Baseline characteristics of participants in phase I of the trials of hypertension prevention
Baseline Characteristics of Participants in Phase I of the Trials of Hypertension Prevention P. K. Whelton, MD, P. R. Hebert, PhD, J. Cutler, MD, W. B. Applegate, MD, K. A. Eberlein, MPH, M. J. Klag, MD, M. E. Keough, MPH, S. Hamill, N. 0. Borhani, MD, J. Hollis, PhD, and A. Oberman, MD, for the Trials of Hypertension Prevention Collaborative Research Group Phase I of the Trials of Hypertension Prevention was designed to test the effectiveness and safety of three life-style (weight loss, sodium restriction, and stress management) and four nutrition supplement (calcium, magnesium, potassium, and jish oil) interventions in reducing diastolic blood pressure (DBP) in persons with a high-normal blood pressure. A total of2 182 persons with a DBP between 80 and 89 mm Hg met the eligibility criteria for participation in phase I and were randomized to one of the active intervention or control treatment groups. Most were white (82%), male (70%)) married (76%), nonsmoking (88%), college graduate (53%)) full-time employees (9 I%). The average blood pressure prior to entry into the trial was 124.9 mm Hg systolic and 83.8 mm Hg diastolic. A variety of baseline observations, including sociodemographic characteristics, personal and family medical history, health habits, diet, and biologic measurements, were documented before randomizaation and compared among the seven active intervention and control groups. As might be expected in a randomized trial of this sample sire, the distribution of measured baseline characteristics was virtually identical in the treated and control groups. Based on this finding and the knowledge that randomizaation procedures were implemented without deviation from the phase I protocol, it is probable that unknown potential confounders were also equally distributed at entry into the study. Given the achievement of high rates of follow-up, subsequent differences in blood pressure are unlikely to have been due to baseline differences between the active treatment and control groups, and can probably be attributed to effects of the active interventions. Ann Epidemiol 1992;2:295-310. KEY WORDS:
Blood pressure, hypertension,
prevention,
clinical trials, nutrition,
epidemiology.
INTRODUCTION Prevention of hypertension represents an attractive complement to our current strategy of treating those with established hypertension. The effectiveness and safety of nonpharmacologic therapy in patients with established hypertension have been well demonstrated but less information exists to document the role of such treatment in persons
From the Welch Center for Prevention, Epidemiology and Climcal Research, The Johns Hopkins Medical Institutions, Baltimore, MD (P.K.W., M.J.K.); The Ch anning Laboratory, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA (P.R.H., K.A.E.); Prevention and Demonstration Research Branch, Division of Epidemiology and Clinical Applications, National Heart, Lung, and Blood Institute, Bethesda, MD (J.C.); Department of Preventive Medicine, University of Tennessee, Memphis, TN (M.B.A.); The East Boston Neighborhood Health Center, Boston, MA (M.E.K.); Preventive Cardiology Program, Department of Medicine, New Jersey Medical School, Newark, NJ (S.H.); Department of Internal Medicine, University of California, Davis, CA (N.O.B.); Kaiser Permanente Center for Health Research, Portland, OR (J.H.); and Division of General and Preventive Medicine, Department of Medicine, University of Alabama at Birmingham, Birmingham, AL (A.O.). Address reprint requests to: P.K. Whelton, MD, Welch Center for Prevention, Epidemiology and Clinical Research, Carnegie 292, Johns Hopkins Hospital, 600 North Wolfe St., Baltimore, MD 21205. Received July 3, 1991; revised January 13, 1992. 0 1992Elsevier Science Publishmg Co., Inc.
1047-2797/921$05.00
296
Whelton
et al.
TOHP PHASE I: BASELINE CHARACTERISTICS
AEP Vol. 2. h’o. 3 May 19Y2:295-3/P
with a highnormal blood pressure ( 1, 2). With this in mind, the Trials of Hypertension Prevention (TOHP) Cooperative Research Group was formed in 1986 to conduct a two-phase study of the efficacy and safety of nonpharmacologic therapy in the prevention of hypertension, The primary purpose of this article is to report the baseline characteristics of participants in the TOHP phase I trial and to assess the extent to which measurements at entry differed between those assigned to the active and those assigned to the control treatment arms. A secondary purpose is to describe crosssectional relationships between blood pressure and other baseline variables such as sociodemographic characteristics, medical history, body mass index (BMI), alcohol intake, and various measures of nutritional status.
METHODS Phase I of the TOHP was a ten-center randomized, controlled clinical trial designed to measure the effectiveness and safety of nonpharmacologic interventions in reducing diastolic blood pressure (DBP) in persons with a high-normal blood pressure. Seven nonpharmacologic treatments were evaluated: three life-style interventions (weight loss, sodium restriction, and stress management) and four nutrition supplement interventions (magnesium, calcium, potassium, and fish oil). The interventions were tested individually in a randomized, controlled, parallel-group trial conducted under a common protocol with fixed periods of treatment and follow-up. Participants were randomly allocated to either active or control (usual care for life-style participants and placebo for nutrition supplement participants) treatment arms. Details of the allocation process and power estimates were presented elsewhere (3). Enrollment began in September 1987 and ended in October 1988. A volunteer’s suitability for participation in the trial was established over the course of three baseline screening visits (SV,_,), each 10 to 30 days apart. Treatment allocation was determined at a fourth baseline status review visit (SRV). The sequence and content of the four baseline data collection visits are outlined in Table 1 for each of the intervention and corresponding control groups. Trial eligibility was based primarily on age (30-54 years), general health status, and confirmation of a highnormal DBP (80-89 mm Hg). In general, the exclusion criteria were intended to eliminate individuals who were hypertensive (DBP S- 90 mm Hg or use of blood pressure medications within 2 months of the first evaluation), had cardiovascular disease, had a contraindication to any of the TOHP phase I interventions, or might have difficulty complying with the treatment or follow-up requirements of the trial. Another important eligibility criterion was a BMI less than approximately 36 kg/m2 (<0.0514 lb/in’). Participants were further classified according to BMI into highweight (men, 26.08-36.14 kg/m’; women, 24.26-36.14 kg/m2) and low-weight (men, <26.08 kglm2; women, <24.26 kg/m2) strata. The high-weight stratum corresponds roughly to 110 to 160% of ideal body weight according to the 1983 Metropolitan Life Insurance tables (4). Those in the high-weight stratum were candidates for any of the seven phase I interventions while their counterparts in the low-weight stratum were eligible for all but the weight loss intervention. Baseline observations were collected to (1) establish an individual’s eligibility for participation in the trial, (2) describe the sociodemographic and clinical characteristics of the study population, (3) provide a means to assess the influence of potential confounding factors, and (4) provide a standard against which to evaluate treatment effects. In order to reduce observer bias, data collectors were blinded to the treatment
TABLE 1 hypertension
Content and sequence of trial-wide baseline data collection prevention (TOHP), by treatment group Baseline
Data collected
297
Whelton et al. TOHP PHASE 1: BASELINE CHARACTERISTICS
AEP Vol. 2, No. 3 May 1992: 295-310
dunng phase I of the trials of visits0
First (SV, )
Second ( SV, )
Third (SV3 )
All
All All
Blood pressure
All
Weight
All
Height Sociodemographic profile MedIcal history Physical activity report Psychological well-being profile 24-Hour diet recall Food frequency report Nonfastmg blood specimen 24-Hour urine specimens
All All All
Fourth (SRV)
All All All All All
All
All
Hassles Scale Pill count Side effects Cardiovascular Anthropometric
Supplements reactivity measurements
Supplements Supplements Stress Weight
’ All = all groups; Stress = stress management and control groups; Supplements = nutntion supplements and placebo control Weight = wqht loss and control groups; SV = screenmg mit; SRL’ = status rewew visit.
groups;
assignment of the participants. As a means to ensure the use of standardized methods and a stringent level of quality control, all data collectors were trained centrally and were required to pass an initial series of certification examinations as well as periodic recertification evaluations. This included rigorous training in the use of a standardized protocol for measuring blood pressure and continuous quality control monitoring of the blood pressure technician’s performance. Blood pressure measurements were obtained using the Hawksley random-zero modification of the standard mercury sphygmomanometer (5). In addition to the previously described baseline measurements of blood pressure, medical history information, weight, and height, a 24-hour diet recall interview was conducted and a food frequency questionnaire was administered at SV, to assess baseline nutritional status (6, 7). The 24-hour diet recall data were collected by the TOHP staff trained and certified by registered dietitians from the trials’ nutrient data center, while the food frequency questionnaire was self-administered. Two baseline 24-hour urine specimens were collected at SV, and SRV. Values from each of these were averaged to estimate sodium, potassium, and creatinine excretion at entry into the trial. For all participants, blood samples were obtained to determine trial eligibility status. These analyses were performed by local clinical center laboratories; the results are not reported here. Physical activity was assessed by self-report of the frequency of physical activity resulting in perspiration (8). The Psychological General Well-Being Index was used to measure overall quality of life (9). The Psychological General Well-Being Index is a 22-item instrument that measures subjective well-being by assessing an individual’s level of anxiety, depression, positive well-being, self-control, general health, and vitality. The results from each of these subscales can be summed to provide an overall estimate of well-being, ranked from high to low. This score has successfully been employed in a number of blood pressure studies, including a recent trial comparing the imp&t of antihypertensive medications on quality of life (10). Additional anthropometric, physiologic, and psychosocial measurements were
298
Whelton et al. TOHP PHASE I: BASELINE CHARACTERISTICS
AEP Vol. 2, No. 3 1992: 295-310
May
obtained in selected groups of participants at entry into the trial (see Table 1). Skinfold thickness, waist and hip circumference measurements, and results of a submaximal bicycle ergometer test were obtained from those assigned to the weight loss intervention group, their counterparts in the weight loss usual care control group, and a small sample of randomly selected high-weight participants in the other life-style intervention arms (to maintain observer blinding). The bicycle ergometer test was primarily conducted as a safety measure. The Lazarus Hassles Scale, which measures both the perceived frequency and intensity of stressful situations, was administered to participants in the stress management group and their control subjects (11). In addition, blood pressure and heart rate response to two cardiovascular stress tests (Atari Breakout Game and Reverse Star Tracing Test) were measured in this group (12). Twenty-four-hour urinary magnesium and calcium excretion was measured for participants assigned to the stage I nutrition supplement intervention arms. Blood samples from participants in the nutrition supplement intervention arms were sent to a central laboratory at the University of Pittsburgh for measurement of serum magnesium levels prior to stage I and to a central laboratory at Harvard Medical School for measurement of lipid profiles and plasma fatty acid phospholipid levels prior to entry into stage II of the nutrition supplement intervention. Duplicate laboratory measurements were obtained in a 5% sample of all urine, plasma, and serum specimens. The results of these analyses provided reassuring evidence of our central laboratories’ quality control procedures. For example, the mean difference between replicate urinary sodium values was 4.94 mmol/L and more than 90% of replicate measurements were within 10% of each other. Group data are expressed as a mean value + standard deviation of the mean or as the proportion of study participants with an attribute. To assess the similarity of the treatment groups at baseline, means or proportions for each variable of interest were compared. Statistical procedures used to evaluate the probability that differences in means or proportions could be attributed to chance alone included a two-tailed, nonpaired Student’s t test (means) and a chi-square test of homogeneity (proportions). To identify cross-sectional associations between blood pressure and other baseline variables, univariate relationships were explored initially. Simple linear regression was used to assess the strength of these associations and to test for the presence of significant relationships. Subsequently, multiple linear regression models were used to determine whether particular variables were significantly related to blood pressure after adjusting for the effects of potential confounders. Probability results higher than 0.001 are expressed as exact values.
RESULTS Table 2 provides descriptive data that characterize the overall baseline sociodemographic and medical characteristics of the 2182 participants who met the eligibility criteria for phase I of the TOHP and were randomly assigned to one of the seven active interventions or their corresponding control groups. As can be seen, phase I participants had a mean age of 43 years and almost 50% were in the age range of 40 to 49 years. The majority were white, male, and married. However, 15% were black and almost one-third were women. The vast majority had received a high school education, and a little more than half were college graduates. Over 90% were employed full-time and half of the remainder worked part-time. Thus, the typical participant in TOHP phase I was a well-educated, middle-aged, white male with a full-time job.
Whelton et al. TOHP PHASE I: BASELINE CHARACTERISTICS
AEP Vol. 2, No. 3 May 1992: 295-310
TABLE Trials
2
Baseline
of Hypertension
sociodemographic Prevention
Demographic and social Age (mean, in y) % 30-39 % 40-49 % 50-54 % Male % White % Black % College graduatesb % Employed full-time % Married Health behawor and family history Cigarette smokmg (%) Never Former Current Alcohol intake (g/wk) Vigorous exerctse (ttmesiweek)’ Psychologwal general well-being index score % with history of premature CVD death in parentsd
and medical
characteristics
299
of the 2182 participants in phase 1 of the
(TOHP)” 43.0 ? 6.5 31.3 48.4 20.3 70.1 82.2 15.0 52.5 90.9 76.4
56.4 32.0 11.6 45.3 + 67.6 2.2 ? 2.4 80.6 t_ 11.8
Physical examination Systolic BP (mean, m mm Hg) %
124.9 ” 8.4 1.9 25.7 48.0 19.4 5.0 83.8 2 2.8 45.4 28.6 26.0 74.0 + 9.1 27.6 + 3.4 27.4 + 4.3
9.1
* Data presented as mean k standard dew&on, unless otherwise specified h Four-year degree program. ’ Exercise resulting in persplratlon. d Less than 60 y of age at death. CVD = cardiovascular dwase; BP = blood pressure.
A somewhat lower percentage of TOHP participants were current smokers than is the case for the general population (13). The data presented on alcohol intake and exercise frequency indicate that TOHP phase I participants tended, on average, to consume approximately four 12-02 cans of beer or 4-0.2 glasses of wine per week and participated in only modest amounts of physical activity. The overall baseline psychological well-being score suggests that the phase I cohort tended to exhibit a better than average psychological profile (9, 10). The objective during screening was to enroll persons with an average DBP in the high-normal range. The results in Table 2 demonstrate that this objective was accomplished. The majority had an entry blood pressure between 80 and 85 mm Hg diastolic (74%) and less than 130 mm Hg systolic (76%). The average BMI of 27.6 kg/ m2 for men and 27.4 kg/m2 for women contrasts with corresponding averages of approximately 26 and 25 kg/m2 for their age-matched counterparts in the general population (14). However, the BMI at entry into phase I varied substantially from 18.2 to 36.5 kg/m2 in men and 18.1 to 37.3 kg/m2 in women. The nutrient intake pattern reported at the baseline 24-hour recall (Table 3) is in keeping with what one might expect given the age, race, and sex distribution of the cohort (15). Recognizing that 24-hour recall estimates are generally thought to provide a somewhat more accurate estimate of nutrient intake (7), there was substantial consistency between corresponding results obtained by the 24.hour recall interviews and the food frequency questionnaire. For instance, estimated average total calorie intake (both sexes combined) was 2360 kcal/d by the 24-hour recall method and 2443 kcalld by the food frequency questionnaire (data not shown). The 24-hour urinary
300
Whelton et al. TOHP PHASE I: BASELINE CHARA~ERISTICS
TABLE 3 Baseline nutritional of Hypertension Prevention’
Variable
characteristics
of the 2182 participants
Estimate
24-Hour diet recall (n = 2157) Calories (kca1124 hr) Males Fern&s ‘16Kcal from fat Males
36.8 36.3
% Kcal from saturated fat Males Females % Kcal from monosaturated Males Females o/o Kcal from polyunsaturated Males Females Sodium (mmoli24 hrt Males Females Potassium (mmol124 hr) Males Females Sodium.potassium ratio Males Females Calcium (mgi24 hr) Males Females Magnesium (mgi24 hr) Males Females
12.9 12.6 fat 13.7 13.3 fat 7.1 7.2 180.8 t 122.5 r 87.0 2 61.3 z!z 2.3 t 2.2 t
in phase I of the Trials
Variable
2614.0 2 1029.0 1769.4 t 738.8
Females
AEP Vol. 2, No. 3 May 1992: 295-310
94.7 70.8 35.7 26.2 1.2 1.2
941.8 -c 684.1 c
613.9 469.6
284.7 t 201.3 2
157.9 119.4
Estimate
Food frequency report (n = 2154) Calories (kcal124 hr) Males Females o/ Kcal from fat
2591.0 2095.4
I 1244.1 ” 981.3
Males 39.1 Females 40.7 % Kcal from saturated fat Males 14.2 Females 14.7 Sodium (mmoV24 hr) Males 182.2 I 98.0 Females 145.4 I 84.9 Potassium (mm&24 hr) Males 94.2 ” 40.3 Females 78.4 t 33.8 Calcium (mg/24 hr) Males 960.1 -c 611.3 Females 870.9 + 583.6 Fiber (g/24 hr) Males 17.2 r 8.2 Females 14.2 2 6.9 24sHour urinary excretionb (n = 2045) Sodium (mmoli24 hr) Males 170.4 rt: 68.5 Females 130.6 z!z 53.8 Potassium (mm01124 hr) Males 67.1 4 24.5 Females 50.0 f 18.0 Sodium~potassium ratio Males 2.7 ir I.I Females 2.8 r 1.2
* Data presented as mean 2 standard deviation, unless otherwise specified. h Averageof collections received at screening visit 1 and status review visit
estimates of sodium and potassium intake provided values that were slightly lower than the corresponding estimates by 24-hour recall and food frequency assessments. Given the potential for excretion of sodium and potassium in stool and perspiration, as well as by other nonrenal mechansims, these lower urinary values are to be expected. Our values are similar to those obtained in two previous primary prevention of hypertension trials (16-18). Because of the variability in sodium intake, our baseline average of sodium excretion at SV, and SRV provides limited information regarding an individual’s urinary electrolyte excretion pattern (19). Recognizing this, only 3.7% of our participants had a baseline urinary sodium excretion value lower than our optimal sodium intervention goal of 60 mmol/24 h. Selected baseline characteristics of the three life-style and four nutrition supplement intervention cohorts and their corresponding usual-care or placebo-control groups are presented in Tables 4 and 5. Each of the 12 treatment groups had baseline characteristics that were similar to the overall pattern described in Table 2. More importantly, the seven active intervention groups groups were similar with respect to sociodemographic
and their
corresponding
profile, health
control
status, nutrient
301
Whelton et al. TOHP PHASE I: BASELINE CHARACTERISTICS
AEP Vol. 2, No. 3 May1992: 295-310
intake,
laboratory
Although
an
achieved
values,
comparison
a conventional
imbalance weight
at entry
respectively; Overall By design, Compliance
more
than
with pill-taking
period
that they “consumed”
values
the most
difference
notable
between
the run-in
for stage I was very high.
at least 66% of their more
than
assigned
90% of their
for the 528 stage II participants.
(96%)
the
(73% male versus 63% male,
consumed
was even better
cally, at least 88% of the 507 participants reported
of a gender
had to consume
80% of them
control
instances,
controls
during
into the trial.
corresponding
in several
and their corresponding
all 698 stage 1 participants
scores at entry
and
into the trial was the presence
P = 0.01). compliance with pill-taking
In practice,
well-being
of active
level of significance
loss participants
pills.
and psychological
unadjusted
for whom
information
at least 90% of their assigned
pills.
Specifi-
was available
pills during
the washout
prior to stage II. Prior to both stage I and stage II of the trial, compliance
the pill-taking
regimen
was similar
high rates of compliance mentioned
baseline
ing placebo
run-in
ments
can probably
inclusion phase,
in nonpharmacologic
in the active
in calendar
packs,
be attributed
and exclusion
the intrinsic
intervention
treatment
to a combination
criteria,
motivation
trials,
with
groups.
These
of the previously
the use of a compliance-monitor-
of those who volunteer
our decision
and the attentiveness
and control
to participate
to supply the nutrition
supple-
of the trial staff to the concerns
of the
study participants. .A side effects questionnaire administered
at SV, . A significant
pants
and control
intervention
group,
stools, which
and constipation, the number
which
complaints
and control
in side effect rates between
only for loose/frequent
group. Given
due to chance
a wide range of symptomatic supplement
difference
intervention
intervention
differences
nutrition
groups was noted
in the calcium potassium
covering
to all potential
active
was
partici-
intervention
were more common
was more common
of comparisons,
in the
these probably
reflect
alone.
The cross-sectional relationships of blood pressure with sociodemographic, dietary, and other baseline indicators are presented in Table 6. Among the sociodemographic variables, about
age was significantly
1 mm Hg higher
for women
to have a higher
unemployed,
participants
relationship
consumed
current race-,
cigarette
per week. smokers
and sex-adjusted an average
positive estimated
of borderline
by 24-hour
the 24-hour
dietary
recall,
significance
and SBP. None
recall were significantly
None
questionnaire, was positively 50-mmol significantly
of the
other
nutritional
were significantly associated with
increase related
in urinary
pressure
(P = 0.054), of the other
associated
standard
between nutritional
but
There
there
calorie
was a intake,
variables
from
SBP or DBP. Sodium
was significantly with a 0.2-mm
as measured
Age-,
with DBP,
50 beatsimin. BMI,
with either
questionnaire, was associated
variables,
eight
2 mm Hg, among
associated
and
was
and SBP,
who were nonsmokers,
and significantly
blood
intake, as estimated by the food frequency SBP such that a 50-mmol higher intake SBP.
by almost
trend
There
of alcohol
(approximately
higher,
SBP was
less-educated,
SBP and DBP.
1 mm Hg for every
between
(SBP).
significant
for black,
intake
to their counterparts
of approximately
relationship
relationship,
100 g of alcohol
SBP was significantly
compared
trend
self-reported
heart rate was positively
increase
was no significant
blood pressure was a borderline
to have a higher
between
a rise of 0.8 mm Hg for every
drinks)
with
positive
with systolic
SBP and a nonsignificant
and unmarried
a significant with
associated
for each 5 years of age. There
by the
food
related to Hg higher frequency
associated with either SBP or DBP. Urinary potassium SBP, the latter being almost 1 mm Hg higher for a
potassium
to DBP such
that
excretion.
Urinary
an increase
sodium-potassium
of 2 sodium-potassium
ratio was standard
Baseline characteristics
Demographic and social Age (Y) % Male % White % Black % College graduates % Employed fuii~time % Married Health status Systoltc BP (mm Hg) Diastolic BP (mm Hg) Heart rate (beats/min) Body mass index (kg/m’) Males Females % History of premature CVD death in parentsh Cigarette smoking % Never % Former % Current
TABLE 4
8.5 2.7 9.0 3.6 4.3
124.8 + 83.7 z? 73.2 2
27.2 -c 16.8 2 11.3
55.0 33.3 11.6
6.6
43.4 2 70.9 78.0 19.6 54.1 92.0 77.7
Active (n = 327)
58.9 30.8 10.3
27.2 t 27.0 rt 8.7
125.1 r 83.9 +: 73.8 r
42.6 rt 71.7 76.5 20.9 56.3 91.6 76.2
3.3 4.4
8.1 2.8 9.5
6.5
8.5 2.8 a.7 3.2 4.5
125.3 ? 83.9 +13.9 r 27.3 r 28.4 rt a.7
62.8 29.3 7.9
6.9
43.4 -r 71.1 83.9 13.2 56.6 91.7 77.7
Active (n = 242)
58.8 33. I 8.1
27.4 + 27.4 t 8.2
124.1 t 83.6 t 73.8 L
43.0 + 70.6 83.8 15.0 50.6 89.7 76.9
3.3 4.3
8.4 2.7 9.4
6.6
Control (n = 320)
Stress management
and control groups’
Control (n = 417)
Sodium resttiction
of life-style intervention
50.0 35.7 14.3
29.5 lr 29.6 ? 8.1
124.3 it 83.7 IL 73.7 ?I
43.1 r 72.7 81.8 16.6 50.0 92.5 76.3
Active (n = 308)
2.7 3.4
8.4 2.6 9.0
6.0
55.3 33.3 11.4
29.6 +29.4 III 7.1
124.6 A 84.0 t 74.8 -c
42.4 t 62.9’ 76.6 21.1 43.9 90.2 79.6
Control (n = 256)
Weight loss
2.5 3.3
a.1 3.0 9.2
6.2
AEP Vol. 2, No. 3 May 1992: 29.5-310
Whelton et al. TOHP PHASE I: BASELINE CHARACTERISTICS
303
304
Whelton et a1. TOHP PHASE I: BASELINE CHARACTERISTICS
AEI’Vol. 2, No. 3 295-310
May 1992:
from
fat
Fem& Sc?dium (mmollZ4 hr) Males Females Potassium (mmnl124 hr) Males Females Calcium (mg/24 hr) Males Females Magnesium (rug124 hr) Males Females 24-Hnur urirrary excretion Sodium (mm&24 hr) Potassium (mm&24 hr) Sodium-potassium ratio Calcium (mg/24 hr) Magnesium (mgi24 hr) Serum results Serum magnesium (mg/dL) Omega 3 fatty acidsd Eicosapentaenoic Docosahexaenoic Oleic acidd
Males
% Kcal
155.3 59.5 2.8 183.2 105.7
162.5 -f 77.‘)” 61.5 +‘ 25.0 2.8 I? 1.21 LV2.8 2 101.7 107.7 +- 41.2 0.2
278.4 _t 13&V 201.5 c 133.7
279.6 z 137.2 IVY.4 _t 113.7
1.9 2
858.0 -+ 492.5’ 638.1 ‘- 3V5.3
698.2 438.5’
1.9 2
0.2
i- 64.7 + 23.0 + 1.21 t 81.7 i- 40.6
61.0 t
102O.YI 742.8 r
33.5
28.0
85.1 t
37.0 22.3
87.1 _t 61.3 4
173.6 z!I 88.3 I3il.i + 75.0”
90.6 83.6
37.9 38.2
179.0 i 123.1 +
36.3 36.Y
928.0 676.0 273.9 Lt 129.2 189.0 + 74.9
262.7 t 124.8 201.2 r 90.6 149.9 t 74.6 59.5 Ifi 24.4 2.7 1. 1.2
304.1 ‘- 158.7 198.8 lir 139.1 63.1 16.7 1.0 91.4 42.4
1.Y f
146.4 t 62.7 -t 2.5 t lVO.7 +104.1 t 0.2
649.0 393.8
0.61 rt; 2.07 rt 9.55 I
144.8 t 56.4 ir 2.7 2
0.47 0.82 1.39
47.9 22.8 1.2
zt 550.4 rt 2Y6.1”
32.0 21.7
912.7 i 525.2 667.6 rt 338.1
1013.7 k 605.7 r.
rt 2
86.2 65.1
29.6 21.4
83.2 i 63.9 4
Vi.8 + 60.1 k
38.2 24.0
168.9 2 86.6 i30.8 -+ 56.2
99.8 52.1
166.1 rf: 72.0 121.6 t 6Y.4
183.1 f 103.6 i:
35.7 38.4
36.6 35.5
36.7 35.2
0.60 0.69 1.26
7I.1 23.4 1.3
156.9 i 59.2 _t 2.9 +
0.57 ‘1.95 + 9.51 2
392.5 92.1
302.6 t lY2.5 t
5 520.5 AT 290.0
32.9 22.0
85.3 t 60.Y ? 897.5 537.5
74.1 69.4
159.6 2 112.6 2
36.6 35.4
306
Whelton
et al.
AEP Vol. 2, No. 3 May 1992: 295-310
TOHP PHASE I: BASELINE CHARACTERISTICS
TABLE 6 Cross+ectional relationships of blood pressure with other baseline variables among the overall TOHP study group (n = 2182) Systolic blood pressure (mm Hg) Variable Demographic
Diastolic blood pressure (mm Hg)
p Coefficient
P value
0.218 0.687 0.587 -0.508 -0.797 -0.260
0.013 0.028 0.253 - 0.060 -0.316 -0.014
0.163 0.830 0.128 0.627 0.148 0.923
0.008
0.003
0.001
0.353
1.639 -0.174 0.116 0.002 - 0.030
0.004 0.663 0.121 0.904 0.526
- 0.307 - 0.052 - 0.003 0.021 - 0.004
0.109 0.698 0.904 0.001 0.821
0.364 0.005 0.013 0.012 0.005 0.003 0.007
0.054 0.142 0.148 0.154 0.704 0.113 0.214
-0.027 < -0.001 - 0.003 - 0.001 0.001 -0.001 -0.001
0.666 0.705 0.339 0.699 0.836 0.275 0.526
0.188 0.002 0.008 0.015 0.004
0.217 0.421 0.293 0.508 0.027
0.002
0.641
0.005 0.017 0.034
0.102 0.045 0.841
/? Coefficient
P
value
and social
Age (Y) Female sex Black race College education’ Employed full-time” Married” Health status Alcohol intake (g/wk)’ Smoking status’ Current smoker Past smoker Vigorous exercise (times/wkpb Heart Rate (beatsimin)” Body Mass Index (kg/m2)” 24-Hour diet recall Total calories (1000/d)” Fat (g/24 hr)” Saturated fat (g/24 hr)” Monosaturated fat (g/24 hr)’ Polyunsaturated fat (g/24 hr)” Sodium (mmol/24 hr)” Potassium (mmoU24 hr)”
Food frequency Total calories (1000/d)” Fat (g/24 hr)” Saturated fat (g/24 hr)” Fiber (g/d)” Sodium (mmoU24 hr)” Potassium (mm01124hr)” 24-Hour urinary electrolyte excretion Sodium (mmol/d)” Potassium (mmolid)” Sodium,potassium ratio”
0.069 0.001 0.003 0.006 0.001 < -0.001 0.001 - 0.002 0.150
0.179 0.281 0.245 0.429 0.106 0.811 0.136 0.392 0.009
’ Controlled for age, race, and sex
bExercise
resulting in perspiration.
deviation units (a ratio higher by 2.22) was associated with a DBP that was approximately 0.3 mm Hg higher. In a multivariate model that included age, race, sex, educational status, current smoking status, reported alcohol intake, total caloric intake, and 24-hour urinary sodium and potassium excretion, SBP remained positively and significantly associated with age, female gender, current smoking status, alcohol consumption, and 24-hour urinary potassium excretion (P < 0.001, P = 0.033, 2’ = 0.004, P = 0.006, and P = 0.054, respectively). In a corresponding multivariate analysis of the cross-sectional associations between baseline variables and DBP, only 24-hour urinary excretion of sodium achieved a conventional level of significance (P = 0.030).
AEP Vol. 2, No. 3 May 1992: 295-310
Whelton et al. TOHP PHASE I: BASELINECHARACTERISTICS
307
DISCUSSION Random allocation of participants to treatment groups and blinding of outcome assessment are among the most important design features of the modem clinical trial (20, 2 1). Given an adequate sample size and proper implementation of the process, random assignment should not only prevent bias in allocating study subjects to the various treatment arms but also provide groups that are quite similar at entry into the trial. This applies not only to similarity in variables that are known to influence the outcome of interest but also to unrecognized factors, which frequently account for a majority of variance in the dependent variable. Documentation of strict adherence to a formal randomization process and demonstration of similarity for variables measured at baseline provide the best assurance that the objectives of randomization have been achieved (20). In this respect, our results are very gratifying. Despite the complexity of the trial, there was no evidence of any important deviation from the protocol-approved randomization process at any of the ten clinical centers. Given this and our large sample size, we saw the expected degree of similarity for baseline observations in the active and control treatment groups. Specifically, only 14 of the more than 200 comparisons of baseline observations revealed a statistically significant (I’ < 0.05) difference. This is about the number that one would expect to occur by chance alone. The most noteworthy baseline discrepancy was a slight difference in gender distribution between participants in the weight loss active intervention and control groups. By design, enrollment in the trial was confined to compliant, middle-aged volunteers with a high-normal DBP. Since phase I of the TOHP and other similar trials were conducted in highly selected samples of the general population, their results should be generalized with caution (16-18). However, the pathophysiologic mechanisms of hypertension are probably quite similar for participants in these selected samples and their counterparts in more representative samples of the general population. Thus, the selective nature of our study sample is unlikely to have interfered with the primary purpose of TOHP phase I, which was to test the hypothesis that selected nonpharmacologic treatments lower the DBP in persons with a high-normal DBP at baseline. All of the cross-sectional associations noted between baseline blood pressure and sociodemographic and behavioral variables have been noted in previous reports (22). A striking finding was the fact that all but two (heart rate, and 24-hour urinary sodiumpotassium ratio) of the significant associations were noted for SBP but not for DBP. Presumably, this reflects the fact that the range of SBP, at entry into the trial, was relatively wide whereas the trial entry criteria ensured that the corresponding range of DBP was quite restricted. In addition to having been reported by other authors, the associations of higher blood pressures with older age, less education, increased alcohol consumption, a faster heart rate, and a higher urinary sodium-potassium excretion ratio are all plausible from a biologic standpoint. These associations also tend both to validate our measurement instruments and to reflect the precision of our phase I data collection methods. The finding of a positive relationship between current cigarette smoking and blood pressure was somewhat surprising as an inverse association has been noted in most cross-sectional studies. However, a positive association between cigarette smoking and blood pressure was also noted in the primary prevention of hypertension trial (17). The lack of an association between BMI and blood pressure was also unexpected but may merely reflect the fact that the study design resulted in a somewhat restricted distribution of both variables at entry into the trial. Finally, the positive
308
Whelton ec al. TOHP PHASE I: BASELINE CHARACTERISTICS
AEP Vol. 2. No. 3 May 1992: 295-310
relationship between urinary potassium excretion and SBP was unanticipated (23, 24). Given the large number of probability tests that were performed, this association may reflect a chance finding.
PARTICIPATING
INSTITUTIONS
Clinical
Centers-The
Health,
Baltimore,
AND
PRINCIPAL
Johns Hopkins
STAFF
University
School of Hygiene
and Public
MD: Paul K. Whelton,
MD (Principal Investigator), Lawrence Appel, MD, Jeanne Charleston, RN, Arlene Taylor Dalcin, RD, Craig Ewart, PhD, Linda Fried, MD, Delores Kaidy, Michael J. Klag, MD, Shiriki Kumanyika, PhD, Lyn Steffen, MPH, and W. Gordon Walker, MD; University of Alabama at Birmingham, Birmingham, AL: Albert Oberman, MD (Principal Investigator), Karen Copeland, RD, Heidi Hataway, MS, James Raczynski, PhD, Neil Rappaport, PhD, Mildred Sehn, and Roland Weinsier, MD; University of California at Davis, Davis, CA: Nemat 0. Borhani, MD (Principal Investigator), Edmund Bernauer, PhD, Patricia Borhani, Carlos de la Cruz, Andrew Ertl, Doug Heustis, Marshall Lee, MD, Wade Lovelace, Ellen O’Connor, Liz Peel, and Carolyn Sugars, RD; East Boston Neighborhood Health Center, East Boston, MA: James 0. Taylor, MD (Principal Investigator), Beth Walker Corkery, MPH, Denis A. Evans, MD, Mary Ellen Keough, MPH, Martha Clare Morris, MPH, Eleanor Pistorino, RN, and Frank Sacks, MD; University of Mississippi, Jackson, MS: Herbert G. Langford, MD (Principal Investigator), Mary Cameron, MS, Dianne Chantanop, RN, Sheila Corrigan, PhD, Stephanie Jennings, MS, John Kiley, MD, Judy Mahalak, and Nancy King Wright; University of Tennessee, Memphis, TN: William B. Applegate, MD (Principal Investigator), Amy Brewer, RD, Laretha Goodwin, RN, Stephen Miller, MD, Joe Murphy, PhD, Judy Randle, Jay Sullivan, MD, and Shirley Vossberg, RD; New Jersey Medical School, Newark, NJ: Norman L. Lasser, MD (Principal Investigator), David M. Batey, PhD, Lee Dolan, Sheila Hamill, Pat Kennedy, RD, and Vera 1. Lasser, MA; University of Pittsburgh, Pittsburgh, PA: Lewis H. Kuller, MD (Principal Investigator), Arlene W. Caggiula, PhD, N. Carole Milas, MS, and Monica E. Yamamoto, DrPH; Kaiser Permanente Center for Health Research, Portland, OR: Thomas M. Vogt, MD (Principal Investigator), Charles Coultrera, Denise Ernst, MA, John Givi, PhD, Merwyn R. Greenlick, PhD, Stephanie Hertert, Jack Hollis, PhD, Marlene McKenzie, RN, Margaret Raker, RD, Steve Smith, MA, Victor Stevens, PhD, and Betsy Wagner, MA; St. Louis University School of Medicine, St. Louis, MO: Jerome D. Cohen, MD (Principal Investigator), Mildred Mattfeldt-Beman, RD, Connie Brinkmann, RN, Katherine Roth, RD, and Lana Shepek, RD. Coordinating School,
Boston,
Center-Brigham MA:
and Women’s
Hospital
and Harvard
Medical
Charles H. Hennekens, MD (Principal Investigator), Julie Buring, ScD, Cristina Cann, Nancy Cook, ScD, Ellie Danielson, MIA, Kim Eberlein, MPH, David Gordon, MAT, Patricia Hebert, PhD, Janet Lang, PhD, Jean MacFadyen, Sherry Mayrent, PhD, Elizabeth Reilly, Bernard Rosner, PhD, Suzanne Satterfield, MD, Heather Tosteson, PhD, and Martin Van Denburgh. Project Office-National Heart, Lung, and Blood Institute: Jeffrey A. Cutler, MD (Scientific Project Officer), Erlca Brittain, PhD, Marilyn Farrand, RD, Peter Kaufmann, PhD, Ed Lakatos, PhD, and Eva Obarzanek, PhD. Central Laboratory-University of Minnesota, Minneapolis, MN: John Belcher, PhD (Project Director), Andrea Dommeyer, Ivan Mills, and Peggy Neibling.
AEP Vol. 2, No. 3 May 1992: 295-310
Whelton et al. TOHP PHASE I: BASELINE CHARACTERISTICS
309
University, Boston, MA: Margo Woods, ScD Nutrient Data Center-Tufts (Project Director), B. J. Kremen Goldman, RD, and Elaine Blethen, RD. Lipid Laboratory-Channing Laboratory, Brigham and Women’s Hospital, Bose ton, MA: Frank Sacks, MD (Director of Lipid Laboratory). Data and Safety Monitoring Committee-Jeremiah Stamler, MD (Chairperson), W. Stewart Agras, MD, Marianna Fordyce-Baum, PhD, C. Morton Hawkins, ScD, Theodore Kotchen, MD, Laurence McCullough, PhD, and Ronald Prineas, MB, PhD.
This work was supported by National Heart, Lung, and Blood Institute research grants H637852, HL367853, HL37854, HL37872, HL37884, HL37899, HL37904, HL37906, HL37907, and HL37924.
REEERENCES 1. Langford HG, Blaufox MD, Oberman A, et al. Dietary therapy slows the return of hypertension after stopping prolonged medication, JAMA. 1985;253:657-64. 2. Stamler R, Stamler J, Grimm R, et al. Nutritional therapy for high blood pressure: Final report of a four-year randomized controlled trial-the hypertension control program, JAMA. 1987;257:1484-91. 3. Satterfield S, Cutler ]A, Langford HG, et al, for the Trials of Hypertension Prevention Collaborative Research Group. Trials of Hypertension Prevention phase I design, Ann Epidemiol. 1991;1:455-71. 4. Metropolitan Life Insurance C!ompany. Metropolitan height and weight tables, Stat Bull. 1983;64:4. Lancet. 1970;1:337-8. 5. Wright BM, Dore LF. A random-zero sphygmomanometer, 6. Pennington JA. Associations between diet and health: The use of food consumption measurements, nutrient databases, and dietary guidelines, ] Am Diet Assoc. 1988;88:1221-4. 7. Anderson SA. Guidelines for use of dietary intake data, J Am Diet Assoc. 1988;88:1258-60. 8. Washburn RA, Montoye H]. The assessment of physical activity by questionnaire, Am J Epidemiol. 1986;123:563-76. 9. Dupuy HJ. The Psychological General Well-Being (PGWB) Index. In: Wenger NK, Mattson ME, Furberg CD, Elinson J, eds. Assessment of Quality of Life in Clinical Trials of Cardiovascular Therapies. New York: Le Jacq Publishing; 1984:170-83. therapy on the 10. Croog SH, Levine S, Testa MH. The effects of antihypertensive quality of life, N Engl J Med. 1986;314:1657-64. 11. Kanner AD, Coyne JC, Schaefer C, Lazarus RS. Comparison of two modes of stress measurement: Daily hassles and uplifts versus major life events, J Behav Med. 1981;4:1-39. 12. Rosner BA, Appel LA, Raczynski JM, et al, for The Trials of Hypertension Prevention Collaborative Research Group: A comparison of two automated monitors in the measurement of blood pressure reactivity, Ann Epidemiol. 1990;1:57-69. 13. The Health Benefits of Smoking Cessation. A report of the Surgeon General 1990. Rockville, MD: US Department of Health and Human Services, Public Health Service, Centers for Disease Control, Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health; 1990:269. DHSS publication no. (CDC) 90-8416. reference data and prevalence of over14. Najjar MF, Rowland M. Anthropometric weight, United States, 1976-1980, Vital Health Stat. [ll]. 1987; 238. DHHS publication no. (PHS) 87-1688. 15. Carrel MD, Abraham S, Dresser CM. Dietary intake source data: United States, 1976-1980, Vital Health Stat. [ll] 1983; 231. DHHS publication no. (PHS) 83-1681. 16, Stamler R, Stamler J, Gosch KC, et al. Primary prevention of hypertension by nutritional-hygienic means, JAMA. 1989;262:1801-7.
310
Whelton et al. TOHP PHASE I: BASELINE CHARACTERISTICS
AEP Vol. 2, No. 3 May 1992: 295-310
17. The Hypertension Prevention Trial Research Group. The Hypertension Prevention Trial (HPT): Three-year effects of dietary changes in blood pressure, Arch Intern Med. 1990;150:153-62. 18. Brown KM, Oberman A, VanNatta ML, Forster JL. Baseline characteristics in the Hypertension Prevention Trial, Controlled Clin Trials. 1989;1O(suppl):4OS-64s. 19. Liu K, Cooper R, McKeever J, et al. Assessment of the association between habitual salt intake and high blood pressure: Methodological problems, Am J Epidemiol. 1979;110:219-26. 20. Meinert CL. Randomization and the mechanics of treatment masking. In: Meinert CL, ed. Clinical Trials. Design, Conduct, and Analysis. New York: Oxford University Press; 1986:90-91. 21. Hulley SB, Feigal D, Martin M, Cummings SR. Designing a new study: IV. Experiments. In: Hulley SB, Cummings SR, eds. Designing Clinical Research. Baltimore: Williams & Wilkins; 1988: 110-27. 22. Committee on Diet and Health, Food and Nutrition Board, Commission on Life Sciences, and National Research Council. Diet and health. Implications for reducing chronic disease risk. In: Hypertension. Washington, DC: National Academy Press; 1989:549-62. 23. Whelton PK, Klag MJ. Potassium in the homeostasis and reduction of blood pressure, Clin. Nutr. 1987;6:76-85. 24. Walker WC, Whelton PK, Russell RP, Saito H, Hermann J. Relation between blood pressure, renin, renin substrate, angiotensin II, aldosterone, and urinary sodium and potassium in 574 ambulatory subjects, Hypertension. 1979;1:287-91.
Blood pressure, hypertension,
prevention,
clinical trials, nutrition,
epidemiology.
INTRODUCTION Prevention of hypertension represents an attractive complement to our current strategy of treating those with established hypertension. The effectiveness and safety of nonpharmacologic therapy in patients with established hypertension have been well demonstrated but less information exists to document the role of such treatment in persons
From the Welch Center for Prevention, Epidemiology and Climcal Research, The Johns Hopkins Medical Institutions, Baltimore, MD (P.K.W., M.J.K.); The Ch anning Laboratory, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA (P.R.H., K.A.E.); Prevention and Demonstration Research Branch, Division of Epidemiology and Clinical Applications, National Heart, Lung, and Blood Institute, Bethesda, MD (J.C.); Department of Preventive Medicine, University of Tennessee, Memphis, TN (M.B.A.); The East Boston Neighborhood Health Center, Boston, MA (M.E.K.); Preventive Cardiology Program, Department of Medicine, New Jersey Medical School, Newark, NJ (S.H.); Department of Internal Medicine, University of California, Davis, CA (N.O.B.); Kaiser Permanente Center for Health Research, Portland, OR (J.H.); and Division of General and Preventive Medicine, Department of Medicine, University of Alabama at Birmingham, Birmingham, AL (A.O.). Address reprint requests to: P.K. Whelton, MD, Welch Center for Prevention, Epidemiology and Clinical Research, Carnegie 292, Johns Hopkins Hospital, 600 North Wolfe St., Baltimore, MD 21205. Received July 3, 1991; revised January 13, 1992. 0 1992Elsevier Science Publishmg Co., Inc.
1047-2797/921$05.00
296
Whelton
et al.
TOHP PHASE I: BASELINE CHARACTERISTICS
AEP Vol. 2. h’o. 3 May 19Y2:295-3/P
with a highnormal blood pressure ( 1, 2). With this in mind, the Trials of Hypertension Prevention (TOHP) Cooperative Research Group was formed in 1986 to conduct a two-phase study of the efficacy and safety of nonpharmacologic therapy in the prevention of hypertension, The primary purpose of this article is to report the baseline characteristics of participants in the TOHP phase I trial and to assess the extent to which measurements at entry differed between those assigned to the active and those assigned to the control treatment arms. A secondary purpose is to describe crosssectional relationships between blood pressure and other baseline variables such as sociodemographic characteristics, medical history, body mass index (BMI), alcohol intake, and various measures of nutritional status.
METHODS Phase I of the TOHP was a ten-center randomized, controlled clinical trial designed to measure the effectiveness and safety of nonpharmacologic interventions in reducing diastolic blood pressure (DBP) in persons with a high-normal blood pressure. Seven nonpharmacologic treatments were evaluated: three life-style interventions (weight loss, sodium restriction, and stress management) and four nutrition supplement interventions (magnesium, calcium, potassium, and fish oil). The interventions were tested individually in a randomized, controlled, parallel-group trial conducted under a common protocol with fixed periods of treatment and follow-up. Participants were randomly allocated to either active or control (usual care for life-style participants and placebo for nutrition supplement participants) treatment arms. Details of the allocation process and power estimates were presented elsewhere (3). Enrollment began in September 1987 and ended in October 1988. A volunteer’s suitability for participation in the trial was established over the course of three baseline screening visits (SV,_,), each 10 to 30 days apart. Treatment allocation was determined at a fourth baseline status review visit (SRV). The sequence and content of the four baseline data collection visits are outlined in Table 1 for each of the intervention and corresponding control groups. Trial eligibility was based primarily on age (30-54 years), general health status, and confirmation of a highnormal DBP (80-89 mm Hg). In general, the exclusion criteria were intended to eliminate individuals who were hypertensive (DBP S- 90 mm Hg or use of blood pressure medications within 2 months of the first evaluation), had cardiovascular disease, had a contraindication to any of the TOHP phase I interventions, or might have difficulty complying with the treatment or follow-up requirements of the trial. Another important eligibility criterion was a BMI less than approximately 36 kg/m2 (<0.0514 lb/in’). Participants were further classified according to BMI into highweight (men, 26.08-36.14 kg/m’; women, 24.26-36.14 kg/m2) and low-weight (men, <26.08 kglm2; women, <24.26 kg/m2) strata. The high-weight stratum corresponds roughly to 110 to 160% of ideal body weight according to the 1983 Metropolitan Life Insurance tables (4). Those in the high-weight stratum were candidates for any of the seven phase I interventions while their counterparts in the low-weight stratum were eligible for all but the weight loss intervention. Baseline observations were collected to (1) establish an individual’s eligibility for participation in the trial, (2) describe the sociodemographic and clinical characteristics of the study population, (3) provide a means to assess the influence of potential confounding factors, and (4) provide a standard against which to evaluate treatment effects. In order to reduce observer bias, data collectors were blinded to the treatment
TABLE 1 hypertension
Content and sequence of trial-wide baseline data collection prevention (TOHP), by treatment group Baseline
Data collected
297
Whelton et al. TOHP PHASE 1: BASELINE CHARACTERISTICS
AEP Vol. 2, No. 3 May 1992: 295-310
dunng phase I of the trials of visits0
First (SV, )
Second ( SV, )
Third (SV3 )
All
All All
Blood pressure
All
Weight
All
Height Sociodemographic profile MedIcal history Physical activity report Psychological well-being profile 24-Hour diet recall Food frequency report Nonfastmg blood specimen 24-Hour urine specimens
All All All
Fourth (SRV)
All All All All All
All
All
Hassles Scale Pill count Side effects Cardiovascular Anthropometric
Supplements reactivity measurements
Supplements Supplements Stress Weight
’ All = all groups; Stress = stress management and control groups; Supplements = nutntion supplements and placebo control Weight = wqht loss and control groups; SV = screenmg mit; SRL’ = status rewew visit.
groups;
assignment of the participants. As a means to ensure the use of standardized methods and a stringent level of quality control, all data collectors were trained centrally and were required to pass an initial series of certification examinations as well as periodic recertification evaluations. This included rigorous training in the use of a standardized protocol for measuring blood pressure and continuous quality control monitoring of the blood pressure technician’s performance. Blood pressure measurements were obtained using the Hawksley random-zero modification of the standard mercury sphygmomanometer (5). In addition to the previously described baseline measurements of blood pressure, medical history information, weight, and height, a 24-hour diet recall interview was conducted and a food frequency questionnaire was administered at SV, to assess baseline nutritional status (6, 7). The 24-hour diet recall data were collected by the TOHP staff trained and certified by registered dietitians from the trials’ nutrient data center, while the food frequency questionnaire was self-administered. Two baseline 24-hour urine specimens were collected at SV, and SRV. Values from each of these were averaged to estimate sodium, potassium, and creatinine excretion at entry into the trial. For all participants, blood samples were obtained to determine trial eligibility status. These analyses were performed by local clinical center laboratories; the results are not reported here. Physical activity was assessed by self-report of the frequency of physical activity resulting in perspiration (8). The Psychological General Well-Being Index was used to measure overall quality of life (9). The Psychological General Well-Being Index is a 22-item instrument that measures subjective well-being by assessing an individual’s level of anxiety, depression, positive well-being, self-control, general health, and vitality. The results from each of these subscales can be summed to provide an overall estimate of well-being, ranked from high to low. This score has successfully been employed in a number of blood pressure studies, including a recent trial comparing the imp&t of antihypertensive medications on quality of life (10). Additional anthropometric, physiologic, and psychosocial measurements were
298
Whelton et al. TOHP PHASE I: BASELINE CHARACTERISTICS
AEP Vol. 2, No. 3 1992: 295-310
May
obtained in selected groups of participants at entry into the trial (see Table 1). Skinfold thickness, waist and hip circumference measurements, and results of a submaximal bicycle ergometer test were obtained from those assigned to the weight loss intervention group, their counterparts in the weight loss usual care control group, and a small sample of randomly selected high-weight participants in the other life-style intervention arms (to maintain observer blinding). The bicycle ergometer test was primarily conducted as a safety measure. The Lazarus Hassles Scale, which measures both the perceived frequency and intensity of stressful situations, was administered to participants in the stress management group and their control subjects (11). In addition, blood pressure and heart rate response to two cardiovascular stress tests (Atari Breakout Game and Reverse Star Tracing Test) were measured in this group (12). Twenty-four-hour urinary magnesium and calcium excretion was measured for participants assigned to the stage I nutrition supplement intervention arms. Blood samples from participants in the nutrition supplement intervention arms were sent to a central laboratory at the University of Pittsburgh for measurement of serum magnesium levels prior to stage I and to a central laboratory at Harvard Medical School for measurement of lipid profiles and plasma fatty acid phospholipid levels prior to entry into stage II of the nutrition supplement intervention. Duplicate laboratory measurements were obtained in a 5% sample of all urine, plasma, and serum specimens. The results of these analyses provided reassuring evidence of our central laboratories’ quality control procedures. For example, the mean difference between replicate urinary sodium values was 4.94 mmol/L and more than 90% of replicate measurements were within 10% of each other. Group data are expressed as a mean value + standard deviation of the mean or as the proportion of study participants with an attribute. To assess the similarity of the treatment groups at baseline, means or proportions for each variable of interest were compared. Statistical procedures used to evaluate the probability that differences in means or proportions could be attributed to chance alone included a two-tailed, nonpaired Student’s t test (means) and a chi-square test of homogeneity (proportions). To identify cross-sectional associations between blood pressure and other baseline variables, univariate relationships were explored initially. Simple linear regression was used to assess the strength of these associations and to test for the presence of significant relationships. Subsequently, multiple linear regression models were used to determine whether particular variables were significantly related to blood pressure after adjusting for the effects of potential confounders. Probability results higher than 0.001 are expressed as exact values.
RESULTS Table 2 provides descriptive data that characterize the overall baseline sociodemographic and medical characteristics of the 2182 participants who met the eligibility criteria for phase I of the TOHP and were randomly assigned to one of the seven active interventions or their corresponding control groups. As can be seen, phase I participants had a mean age of 43 years and almost 50% were in the age range of 40 to 49 years. The majority were white, male, and married. However, 15% were black and almost one-third were women. The vast majority had received a high school education, and a little more than half were college graduates. Over 90% were employed full-time and half of the remainder worked part-time. Thus, the typical participant in TOHP phase I was a well-educated, middle-aged, white male with a full-time job.
Whelton et al. TOHP PHASE I: BASELINE CHARACTERISTICS
AEP Vol. 2, No. 3 May 1992: 295-310
TABLE Trials
2
Baseline
of Hypertension
sociodemographic Prevention
Demographic and social Age (mean, in y) % 30-39 % 40-49 % 50-54 % Male % White % Black % College graduatesb % Employed full-time % Married Health behawor and family history Cigarette smokmg (%) Never Former Current Alcohol intake (g/wk) Vigorous exerctse (ttmesiweek)’ Psychologwal general well-being index score % with history of premature CVD death in parentsd
and medical
characteristics
299
of the 2182 participants in phase 1 of the
(TOHP)” 43.0 ? 6.5 31.3 48.4 20.3 70.1 82.2 15.0 52.5 90.9 76.4
56.4 32.0 11.6 45.3 + 67.6 2.2 ? 2.4 80.6 t_ 11.8
Physical examination Systolic BP (mean, m mm Hg) %
124.9 ” 8.4 1.9 25.7 48.0 19.4 5.0 83.8 2 2.8 45.4 28.6 26.0 74.0 + 9.1 27.6 + 3.4 27.4 + 4.3
9.1
* Data presented as mean k standard dew&on, unless otherwise specified h Four-year degree program. ’ Exercise resulting in persplratlon. d Less than 60 y of age at death. CVD = cardiovascular dwase; BP = blood pressure.
A somewhat lower percentage of TOHP participants were current smokers than is the case for the general population (13). The data presented on alcohol intake and exercise frequency indicate that TOHP phase I participants tended, on average, to consume approximately four 12-02 cans of beer or 4-0.2 glasses of wine per week and participated in only modest amounts of physical activity. The overall baseline psychological well-being score suggests that the phase I cohort tended to exhibit a better than average psychological profile (9, 10). The objective during screening was to enroll persons with an average DBP in the high-normal range. The results in Table 2 demonstrate that this objective was accomplished. The majority had an entry blood pressure between 80 and 85 mm Hg diastolic (74%) and less than 130 mm Hg systolic (76%). The average BMI of 27.6 kg/ m2 for men and 27.4 kg/m2 for women contrasts with corresponding averages of approximately 26 and 25 kg/m2 for their age-matched counterparts in the general population (14). However, the BMI at entry into phase I varied substantially from 18.2 to 36.5 kg/m2 in men and 18.1 to 37.3 kg/m2 in women. The nutrient intake pattern reported at the baseline 24-hour recall (Table 3) is in keeping with what one might expect given the age, race, and sex distribution of the cohort (15). Recognizing that 24-hour recall estimates are generally thought to provide a somewhat more accurate estimate of nutrient intake (7), there was substantial consistency between corresponding results obtained by the 24.hour recall interviews and the food frequency questionnaire. For instance, estimated average total calorie intake (both sexes combined) was 2360 kcal/d by the 24-hour recall method and 2443 kcalld by the food frequency questionnaire (data not shown). The 24-hour urinary
300
Whelton et al. TOHP PHASE I: BASELINE CHARA~ERISTICS
TABLE 3 Baseline nutritional of Hypertension Prevention’
Variable
characteristics
of the 2182 participants
Estimate
24-Hour diet recall (n = 2157) Calories (kca1124 hr) Males Fern&s ‘16Kcal from fat Males
36.8 36.3
% Kcal from saturated fat Males Females % Kcal from monosaturated Males Females o/o Kcal from polyunsaturated Males Females Sodium (mmoli24 hrt Males Females Potassium (mmol124 hr) Males Females Sodium.potassium ratio Males Females Calcium (mgi24 hr) Males Females Magnesium (mgi24 hr) Males Females
12.9 12.6 fat 13.7 13.3 fat 7.1 7.2 180.8 t 122.5 r 87.0 2 61.3 z!z 2.3 t 2.2 t
in phase I of the Trials
Variable
2614.0 2 1029.0 1769.4 t 738.8
Females
AEP Vol. 2, No. 3 May 1992: 295-310
94.7 70.8 35.7 26.2 1.2 1.2
941.8 -c 684.1 c
613.9 469.6
284.7 t 201.3 2
157.9 119.4
Estimate
Food frequency report (n = 2154) Calories (kcal124 hr) Males Females o/ Kcal from fat
2591.0 2095.4
I 1244.1 ” 981.3
Males 39.1 Females 40.7 % Kcal from saturated fat Males 14.2 Females 14.7 Sodium (mmoV24 hr) Males 182.2 I 98.0 Females 145.4 I 84.9 Potassium (mm&24 hr) Males 94.2 ” 40.3 Females 78.4 t 33.8 Calcium (mg/24 hr) Males 960.1 -c 611.3 Females 870.9 + 583.6 Fiber (g/24 hr) Males 17.2 r 8.2 Females 14.2 2 6.9 24sHour urinary excretionb (n = 2045) Sodium (mmoli24 hr) Males 170.4 rt: 68.5 Females 130.6 z!z 53.8 Potassium (mm01124 hr) Males 67.1 4 24.5 Females 50.0 f 18.0 Sodium~potassium ratio Males 2.7 ir I.I Females 2.8 r 1.2
* Data presented as mean 2 standard deviation, unless otherwise specified. h Averageof collections received at screening visit 1 and status review visit
estimates of sodium and potassium intake provided values that were slightly lower than the corresponding estimates by 24-hour recall and food frequency assessments. Given the potential for excretion of sodium and potassium in stool and perspiration, as well as by other nonrenal mechansims, these lower urinary values are to be expected. Our values are similar to those obtained in two previous primary prevention of hypertension trials (16-18). Because of the variability in sodium intake, our baseline average of sodium excretion at SV, and SRV provides limited information regarding an individual’s urinary electrolyte excretion pattern (19). Recognizing this, only 3.7% of our participants had a baseline urinary sodium excretion value lower than our optimal sodium intervention goal of 60 mmol/24 h. Selected baseline characteristics of the three life-style and four nutrition supplement intervention cohorts and their corresponding usual-care or placebo-control groups are presented in Tables 4 and 5. Each of the 12 treatment groups had baseline characteristics that were similar to the overall pattern described in Table 2. More importantly, the seven active intervention groups groups were similar with respect to sociodemographic
and their
corresponding
profile, health
control
status, nutrient
301
Whelton et al. TOHP PHASE I: BASELINE CHARACTERISTICS
AEP Vol. 2, No. 3 May1992: 295-310
intake,
laboratory
Although
an
achieved
values,
comparison
a conventional
imbalance weight
at entry
respectively; Overall By design, Compliance
more
than
with pill-taking
period
that they “consumed”
values
the most
difference
notable
between
the run-in
for stage I was very high.
at least 66% of their more
than
assigned
90% of their
for the 528 stage II participants.
(96%)
the
(73% male versus 63% male,
consumed
was even better
cally, at least 88% of the 507 participants reported
of a gender
had to consume
80% of them
control
instances,
controls
during
into the trial.
corresponding
in several
and their corresponding
all 698 stage 1 participants
scores at entry
and
into the trial was the presence
P = 0.01). compliance with pill-taking
In practice,
well-being
of active
level of significance
loss participants
pills.
and psychological
unadjusted
for whom
information
at least 90% of their assigned
pills.
Specifi-
was available
pills during
the washout
prior to stage II. Prior to both stage I and stage II of the trial, compliance
the pill-taking
regimen
was similar
high rates of compliance mentioned
baseline
ing placebo
run-in
ments
can probably
inclusion phase,
in nonpharmacologic
in the active
in calendar
packs,
be attributed
and exclusion
the intrinsic
intervention
treatment
to a combination
criteria,
motivation
trials,
with
groups.
These
of the previously
the use of a compliance-monitor-
of those who volunteer
our decision
and the attentiveness
and control
to participate
to supply the nutrition
supple-
of the trial staff to the concerns
of the
study participants. .A side effects questionnaire administered
at SV, . A significant
pants
and control
intervention
group,
stools, which
and constipation, the number
which
complaints
and control
in side effect rates between
only for loose/frequent
group. Given
due to chance
a wide range of symptomatic supplement
difference
intervention
intervention
differences
nutrition
groups was noted
in the calcium potassium
covering
to all potential
active
was
partici-
intervention
were more common
was more common
of comparisons,
in the
these probably
reflect
alone.
The cross-sectional relationships of blood pressure with sociodemographic, dietary, and other baseline indicators are presented in Table 6. Among the sociodemographic variables, about
age was significantly
1 mm Hg higher
for women
to have a higher
unemployed,
participants
relationship
consumed
current race-,
cigarette
per week. smokers
and sex-adjusted an average
positive estimated
of borderline
by 24-hour
the 24-hour
dietary
recall,
significance
and SBP. None
recall were significantly
None
questionnaire, was positively 50-mmol significantly
of the
other
nutritional
were significantly associated with
increase related
in urinary
pressure
(P = 0.054), of the other
associated
standard
between nutritional
but
There
there
calorie
was a intake,
variables
from
SBP or DBP. Sodium
was significantly with a 0.2-mm
as measured
Age-,
with DBP,
50 beatsimin. BMI,
with either
questionnaire, was associated
variables,
eight
2 mm Hg, among
associated
and
was
and SBP,
who were nonsmokers,
and significantly
blood
intake, as estimated by the food frequency SBP such that a 50-mmol higher intake SBP.
by almost
trend
There
of alcohol
(approximately
higher,
SBP was
less-educated,
SBP and DBP.
1 mm Hg for every
between
(SBP).
significant
for black,
intake
to their counterparts
of approximately
relationship
relationship,
100 g of alcohol
SBP was significantly
compared
trend
self-reported
heart rate was positively
increase
was no significant
blood pressure was a borderline
to have a higher
between
a rise of 0.8 mm Hg for every
drinks)
with
positive
with systolic
SBP and a nonsignificant
and unmarried
a significant with
associated
for each 5 years of age. There
by the
food
related to Hg higher frequency
associated with either SBP or DBP. Urinary potassium SBP, the latter being almost 1 mm Hg higher for a
potassium
to DBP such
that
excretion.
Urinary
an increase
sodium-potassium
of 2 sodium-potassium
ratio was standard
Baseline characteristics
Demographic and social Age (Y) % Male % White % Black % College graduates % Employed fuii~time % Married Health status Systoltc BP (mm Hg) Diastolic BP (mm Hg) Heart rate (beats/min) Body mass index (kg/m’) Males Females % History of premature CVD death in parentsh Cigarette smoking % Never % Former % Current
TABLE 4
8.5 2.7 9.0 3.6 4.3
124.8 + 83.7 z? 73.2 2
27.2 -c 16.8 2 11.3
55.0 33.3 11.6
6.6
43.4 2 70.9 78.0 19.6 54.1 92.0 77.7
Active (n = 327)
58.9 30.8 10.3
27.2 t 27.0 rt 8.7
125.1 r 83.9 +: 73.8 r
42.6 rt 71.7 76.5 20.9 56.3 91.6 76.2
3.3 4.4
8.1 2.8 9.5
6.5
8.5 2.8 a.7 3.2 4.5
125.3 ? 83.9 +13.9 r 27.3 r 28.4 rt a.7
62.8 29.3 7.9
6.9
43.4 -r 71.1 83.9 13.2 56.6 91.7 77.7
Active (n = 242)
58.8 33. I 8.1
27.4 + 27.4 t 8.2
124.1 t 83.6 t 73.8 L
43.0 + 70.6 83.8 15.0 50.6 89.7 76.9
3.3 4.3
8.4 2.7 9.4
6.6
Control (n = 320)
Stress management
and control groups’
Control (n = 417)
Sodium resttiction
of life-style intervention
50.0 35.7 14.3
29.5 lr 29.6 ? 8.1
124.3 it 83.7 IL 73.7 ?I
43.1 r 72.7 81.8 16.6 50.0 92.5 76.3
Active (n = 308)
2.7 3.4
8.4 2.6 9.0
6.0
55.3 33.3 11.4
29.6 +29.4 III 7.1
124.6 A 84.0 t 74.8 -c
42.4 t 62.9’ 76.6 21.1 43.9 90.2 79.6
Control (n = 256)
Weight loss
2.5 3.3
a.1 3.0 9.2
6.2
AEP Vol. 2, No. 3 May 1992: 29.5-310
Whelton et al. TOHP PHASE I: BASELINE CHARACTERISTICS
303
304
Whelton et a1. TOHP PHASE I: BASELINE CHARACTERISTICS
AEI’Vol. 2, No. 3 295-310
May 1992:
from
fat
Fem& Sc?dium (mmollZ4 hr) Males Females Potassium (mmnl124 hr) Males Females Calcium (mg/24 hr) Males Females Magnesium (rug124 hr) Males Females 24-Hnur urirrary excretion Sodium (mm&24 hr) Potassium (mm&24 hr) Sodium-potassium ratio Calcium (mg/24 hr) Magnesium (mgi24 hr) Serum results Serum magnesium (mg/dL) Omega 3 fatty acidsd Eicosapentaenoic Docosahexaenoic Oleic acidd
Males
% Kcal
155.3 59.5 2.8 183.2 105.7
162.5 -f 77.‘)” 61.5 +‘ 25.0 2.8 I? 1.21 LV2.8 2 101.7 107.7 +- 41.2 0.2
278.4 _t 13&V 201.5 c 133.7
279.6 z 137.2 IVY.4 _t 113.7
1.9 2
858.0 -+ 492.5’ 638.1 ‘- 3V5.3
698.2 438.5’
1.9 2
0.2
i- 64.7 + 23.0 + 1.21 t 81.7 i- 40.6
61.0 t
102O.YI 742.8 r
33.5
28.0
85.1 t
37.0 22.3
87.1 _t 61.3 4
173.6 z!I 88.3 I3il.i + 75.0”
90.6 83.6
37.9 38.2
179.0 i 123.1 +
36.3 36.Y
928.0 676.0 273.9 Lt 129.2 189.0 + 74.9
262.7 t 124.8 201.2 r 90.6 149.9 t 74.6 59.5 Ifi 24.4 2.7 1. 1.2
304.1 ‘- 158.7 198.8 lir 139.1 63.1 16.7 1.0 91.4 42.4
1.Y f
146.4 t 62.7 -t 2.5 t lVO.7 +104.1 t 0.2
649.0 393.8
0.61 rt; 2.07 rt 9.55 I
144.8 t 56.4 ir 2.7 2
0.47 0.82 1.39
47.9 22.8 1.2
zt 550.4 rt 2Y6.1”
32.0 21.7
912.7 i 525.2 667.6 rt 338.1
1013.7 k 605.7 r.
rt 2
86.2 65.1
29.6 21.4
83.2 i 63.9 4
Vi.8 + 60.1 k
38.2 24.0
168.9 2 86.6 i30.8 -+ 56.2
99.8 52.1
166.1 rf: 72.0 121.6 t 6Y.4
183.1 f 103.6 i:
35.7 38.4
36.6 35.5
36.7 35.2
0.60 0.69 1.26
7I.1 23.4 1.3
156.9 i 59.2 _t 2.9 +
0.57 ‘1.95 + 9.51 2
392.5 92.1
302.6 t lY2.5 t
5 520.5 AT 290.0
32.9 22.0
85.3 t 60.Y ? 897.5 537.5
74.1 69.4
159.6 2 112.6 2
36.6 35.4
306
Whelton
et al.
AEP Vol. 2, No. 3 May 1992: 295-310
TOHP PHASE I: BASELINE CHARACTERISTICS
TABLE 6 Cross+ectional relationships of blood pressure with other baseline variables among the overall TOHP study group (n = 2182) Systolic blood pressure (mm Hg) Variable Demographic
Diastolic blood pressure (mm Hg)
p Coefficient
P value
0.218 0.687 0.587 -0.508 -0.797 -0.260
0.013 0.028 0.253 - 0.060 -0.316 -0.014
0.163 0.830 0.128 0.627 0.148 0.923
0.008
0.003
0.001
0.353
1.639 -0.174 0.116 0.002 - 0.030
0.004 0.663 0.121 0.904 0.526
- 0.307 - 0.052 - 0.003 0.021 - 0.004
0.109 0.698 0.904 0.001 0.821
0.364 0.005 0.013 0.012 0.005 0.003 0.007
0.054 0.142 0.148 0.154 0.704 0.113 0.214
-0.027 < -0.001 - 0.003 - 0.001 0.001 -0.001 -0.001
0.666 0.705 0.339 0.699 0.836 0.275 0.526
0.188 0.002 0.008 0.015 0.004
0.217 0.421 0.293 0.508 0.027
0.002
0.641
0.005 0.017 0.034
0.102 0.045 0.841
/? Coefficient
P
value
and social
Age (Y) Female sex Black race College education’ Employed full-time” Married” Health status Alcohol intake (g/wk)’ Smoking status’ Current smoker Past smoker Vigorous exercise (times/wkpb Heart Rate (beatsimin)” Body Mass Index (kg/m2)” 24-Hour diet recall Total calories (1000/d)” Fat (g/24 hr)” Saturated fat (g/24 hr)” Monosaturated fat (g/24 hr)’ Polyunsaturated fat (g/24 hr)” Sodium (mmol/24 hr)” Potassium (mmoU24 hr)”
Food frequency Total calories (1000/d)” Fat (g/24 hr)” Saturated fat (g/24 hr)” Fiber (g/d)” Sodium (mmoU24 hr)” Potassium (mm01124hr)” 24-Hour urinary electrolyte excretion Sodium (mmol/d)” Potassium (mmolid)” Sodium,potassium ratio”
0.069 0.001 0.003 0.006 0.001 < -0.001 0.001 - 0.002 0.150
0.179 0.281 0.245 0.429 0.106 0.811 0.136 0.392 0.009
’ Controlled for age, race, and sex
bExercise
resulting in perspiration.
deviation units (a ratio higher by 2.22) was associated with a DBP that was approximately 0.3 mm Hg higher. In a multivariate model that included age, race, sex, educational status, current smoking status, reported alcohol intake, total caloric intake, and 24-hour urinary sodium and potassium excretion, SBP remained positively and significantly associated with age, female gender, current smoking status, alcohol consumption, and 24-hour urinary potassium excretion (P < 0.001, P = 0.033, 2’ = 0.004, P = 0.006, and P = 0.054, respectively). In a corresponding multivariate analysis of the cross-sectional associations between baseline variables and DBP, only 24-hour urinary excretion of sodium achieved a conventional level of significance (P = 0.030).
AEP Vol. 2, No. 3 May 1992: 295-310
Whelton et al. TOHP PHASE I: BASELINECHARACTERISTICS
307
DISCUSSION Random allocation of participants to treatment groups and blinding of outcome assessment are among the most important design features of the modem clinical trial (20, 2 1). Given an adequate sample size and proper implementation of the process, random assignment should not only prevent bias in allocating study subjects to the various treatment arms but also provide groups that are quite similar at entry into the trial. This applies not only to similarity in variables that are known to influence the outcome of interest but also to unrecognized factors, which frequently account for a majority of variance in the dependent variable. Documentation of strict adherence to a formal randomization process and demonstration of similarity for variables measured at baseline provide the best assurance that the objectives of randomization have been achieved (20). In this respect, our results are very gratifying. Despite the complexity of the trial, there was no evidence of any important deviation from the protocol-approved randomization process at any of the ten clinical centers. Given this and our large sample size, we saw the expected degree of similarity for baseline observations in the active and control treatment groups. Specifically, only 14 of the more than 200 comparisons of baseline observations revealed a statistically significant (I’ < 0.05) difference. This is about the number that one would expect to occur by chance alone. The most noteworthy baseline discrepancy was a slight difference in gender distribution between participants in the weight loss active intervention and control groups. By design, enrollment in the trial was confined to compliant, middle-aged volunteers with a high-normal DBP. Since phase I of the TOHP and other similar trials were conducted in highly selected samples of the general population, their results should be generalized with caution (16-18). However, the pathophysiologic mechanisms of hypertension are probably quite similar for participants in these selected samples and their counterparts in more representative samples of the general population. Thus, the selective nature of our study sample is unlikely to have interfered with the primary purpose of TOHP phase I, which was to test the hypothesis that selected nonpharmacologic treatments lower the DBP in persons with a high-normal DBP at baseline. All of the cross-sectional associations noted between baseline blood pressure and sociodemographic and behavioral variables have been noted in previous reports (22). A striking finding was the fact that all but two (heart rate, and 24-hour urinary sodiumpotassium ratio) of the significant associations were noted for SBP but not for DBP. Presumably, this reflects the fact that the range of SBP, at entry into the trial, was relatively wide whereas the trial entry criteria ensured that the corresponding range of DBP was quite restricted. In addition to having been reported by other authors, the associations of higher blood pressures with older age, less education, increased alcohol consumption, a faster heart rate, and a higher urinary sodium-potassium excretion ratio are all plausible from a biologic standpoint. These associations also tend both to validate our measurement instruments and to reflect the precision of our phase I data collection methods. The finding of a positive relationship between current cigarette smoking and blood pressure was somewhat surprising as an inverse association has been noted in most cross-sectional studies. However, a positive association between cigarette smoking and blood pressure was also noted in the primary prevention of hypertension trial (17). The lack of an association between BMI and blood pressure was also unexpected but may merely reflect the fact that the study design resulted in a somewhat restricted distribution of both variables at entry into the trial. Finally, the positive
308
Whelton ec al. TOHP PHASE I: BASELINE CHARACTERISTICS
AEP Vol. 2. No. 3 May 1992: 295-310
relationship between urinary potassium excretion and SBP was unanticipated (23, 24). Given the large number of probability tests that were performed, this association may reflect a chance finding.
PARTICIPATING
INSTITUTIONS
Clinical
Centers-The
Health,
Baltimore,
AND
PRINCIPAL
Johns Hopkins
STAFF
University
School of Hygiene
and Public
MD: Paul K. Whelton,
MD (Principal Investigator), Lawrence Appel, MD, Jeanne Charleston, RN, Arlene Taylor Dalcin, RD, Craig Ewart, PhD, Linda Fried, MD, Delores Kaidy, Michael J. Klag, MD, Shiriki Kumanyika, PhD, Lyn Steffen, MPH, and W. Gordon Walker, MD; University of Alabama at Birmingham, Birmingham, AL: Albert Oberman, MD (Principal Investigator), Karen Copeland, RD, Heidi Hataway, MS, James Raczynski, PhD, Neil Rappaport, PhD, Mildred Sehn, and Roland Weinsier, MD; University of California at Davis, Davis, CA: Nemat 0. Borhani, MD (Principal Investigator), Edmund Bernauer, PhD, Patricia Borhani, Carlos de la Cruz, Andrew Ertl, Doug Heustis, Marshall Lee, MD, Wade Lovelace, Ellen O’Connor, Liz Peel, and Carolyn Sugars, RD; East Boston Neighborhood Health Center, East Boston, MA: James 0. Taylor, MD (Principal Investigator), Beth Walker Corkery, MPH, Denis A. Evans, MD, Mary Ellen Keough, MPH, Martha Clare Morris, MPH, Eleanor Pistorino, RN, and Frank Sacks, MD; University of Mississippi, Jackson, MS: Herbert G. Langford, MD (Principal Investigator), Mary Cameron, MS, Dianne Chantanop, RN, Sheila Corrigan, PhD, Stephanie Jennings, MS, John Kiley, MD, Judy Mahalak, and Nancy King Wright; University of Tennessee, Memphis, TN: William B. Applegate, MD (Principal Investigator), Amy Brewer, RD, Laretha Goodwin, RN, Stephen Miller, MD, Joe Murphy, PhD, Judy Randle, Jay Sullivan, MD, and Shirley Vossberg, RD; New Jersey Medical School, Newark, NJ: Norman L. Lasser, MD (Principal Investigator), David M. Batey, PhD, Lee Dolan, Sheila Hamill, Pat Kennedy, RD, and Vera 1. Lasser, MA; University of Pittsburgh, Pittsburgh, PA: Lewis H. Kuller, MD (Principal Investigator), Arlene W. Caggiula, PhD, N. Carole Milas, MS, and Monica E. Yamamoto, DrPH; Kaiser Permanente Center for Health Research, Portland, OR: Thomas M. Vogt, MD (Principal Investigator), Charles Coultrera, Denise Ernst, MA, John Givi, PhD, Merwyn R. Greenlick, PhD, Stephanie Hertert, Jack Hollis, PhD, Marlene McKenzie, RN, Margaret Raker, RD, Steve Smith, MA, Victor Stevens, PhD, and Betsy Wagner, MA; St. Louis University School of Medicine, St. Louis, MO: Jerome D. Cohen, MD (Principal Investigator), Mildred Mattfeldt-Beman, RD, Connie Brinkmann, RN, Katherine Roth, RD, and Lana Shepek, RD. Coordinating School,
Boston,
Center-Brigham MA:
and Women’s
Hospital
and Harvard
Medical
Charles H. Hennekens, MD (Principal Investigator), Julie Buring, ScD, Cristina Cann, Nancy Cook, ScD, Ellie Danielson, MIA, Kim Eberlein, MPH, David Gordon, MAT, Patricia Hebert, PhD, Janet Lang, PhD, Jean MacFadyen, Sherry Mayrent, PhD, Elizabeth Reilly, Bernard Rosner, PhD, Suzanne Satterfield, MD, Heather Tosteson, PhD, and Martin Van Denburgh. Project Office-National Heart, Lung, and Blood Institute: Jeffrey A. Cutler, MD (Scientific Project Officer), Erlca Brittain, PhD, Marilyn Farrand, RD, Peter Kaufmann, PhD, Ed Lakatos, PhD, and Eva Obarzanek, PhD. Central Laboratory-University of Minnesota, Minneapolis, MN: John Belcher, PhD (Project Director), Andrea Dommeyer, Ivan Mills, and Peggy Neibling.
AEP Vol. 2, No. 3 May 1992: 295-310
Whelton et al. TOHP PHASE I: BASELINE CHARACTERISTICS
309
University, Boston, MA: Margo Woods, ScD Nutrient Data Center-Tufts (Project Director), B. J. Kremen Goldman, RD, and Elaine Blethen, RD. Lipid Laboratory-Channing Laboratory, Brigham and Women’s Hospital, Bose ton, MA: Frank Sacks, MD (Director of Lipid Laboratory). Data and Safety Monitoring Committee-Jeremiah Stamler, MD (Chairperson), W. Stewart Agras, MD, Marianna Fordyce-Baum, PhD, C. Morton Hawkins, ScD, Theodore Kotchen, MD, Laurence McCullough, PhD, and Ronald Prineas, MB, PhD.
This work was supported by National Heart, Lung, and Blood Institute research grants H637852, HL367853, HL37854, HL37872, HL37884, HL37899, HL37904, HL37906, HL37907, and HL37924.
REEERENCES 1. Langford HG, Blaufox MD, Oberman A, et al. Dietary therapy slows the return of hypertension after stopping prolonged medication, JAMA. 1985;253:657-64. 2. Stamler R, Stamler J, Grimm R, et al. Nutritional therapy for high blood pressure: Final report of a four-year randomized controlled trial-the hypertension control program, JAMA. 1987;257:1484-91. 3. Satterfield S, Cutler ]A, Langford HG, et al, for the Trials of Hypertension Prevention Collaborative Research Group. Trials of Hypertension Prevention phase I design, Ann Epidemiol. 1991;1:455-71. 4. Metropolitan Life Insurance C!ompany. Metropolitan height and weight tables, Stat Bull. 1983;64:4. Lancet. 1970;1:337-8. 5. Wright BM, Dore LF. A random-zero sphygmomanometer, 6. Pennington JA. Associations between diet and health: The use of food consumption measurements, nutrient databases, and dietary guidelines, ] Am Diet Assoc. 1988;88:1221-4. 7. Anderson SA. Guidelines for use of dietary intake data, J Am Diet Assoc. 1988;88:1258-60. 8. Washburn RA, Montoye H]. The assessment of physical activity by questionnaire, Am J Epidemiol. 1986;123:563-76. 9. Dupuy HJ. The Psychological General Well-Being (PGWB) Index. In: Wenger NK, Mattson ME, Furberg CD, Elinson J, eds. Assessment of Quality of Life in Clinical Trials of Cardiovascular Therapies. New York: Le Jacq Publishing; 1984:170-83. therapy on the 10. Croog SH, Levine S, Testa MH. The effects of antihypertensive quality of life, N Engl J Med. 1986;314:1657-64. 11. Kanner AD, Coyne JC, Schaefer C, Lazarus RS. Comparison of two modes of stress measurement: Daily hassles and uplifts versus major life events, J Behav Med. 1981;4:1-39. 12. Rosner BA, Appel LA, Raczynski JM, et al, for The Trials of Hypertension Prevention Collaborative Research Group: A comparison of two automated monitors in the measurement of blood pressure reactivity, Ann Epidemiol. 1990;1:57-69. 13. The Health Benefits of Smoking Cessation. A report of the Surgeon General 1990. Rockville, MD: US Department of Health and Human Services, Public Health Service, Centers for Disease Control, Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health; 1990:269. DHSS publication no. (CDC) 90-8416. reference data and prevalence of over14. Najjar MF, Rowland M. Anthropometric weight, United States, 1976-1980, Vital Health Stat. [ll]. 1987; 238. DHHS publication no. (PHS) 87-1688. 15. Carrel MD, Abraham S, Dresser CM. Dietary intake source data: United States, 1976-1980, Vital Health Stat. [ll] 1983; 231. DHHS publication no. (PHS) 83-1681. 16, Stamler R, Stamler J, Gosch KC, et al. Primary prevention of hypertension by nutritional-hygienic means, JAMA. 1989;262:1801-7.
310
Whelton et al. TOHP PHASE I: BASELINE CHARACTERISTICS
AEP Vol. 2, No. 3 May 1992: 295-310
17. The Hypertension Prevention Trial Research Group. The Hypertension Prevention Trial (HPT): Three-year effects of dietary changes in blood pressure, Arch Intern Med. 1990;150:153-62. 18. Brown KM, Oberman A, VanNatta ML, Forster JL. Baseline characteristics in the Hypertension Prevention Trial, Controlled Clin Trials. 1989;1O(suppl):4OS-64s. 19. Liu K, Cooper R, McKeever J, et al. Assessment of the association between habitual salt intake and high blood pressure: Methodological problems, Am J Epidemiol. 1979;110:219-26. 20. Meinert CL. Randomization and the mechanics of treatment masking. In: Meinert CL, ed. Clinical Trials. Design, Conduct, and Analysis. New York: Oxford University Press; 1986:90-91. 21. Hulley SB, Feigal D, Martin M, Cummings SR. Designing a new study: IV. Experiments. In: Hulley SB, Cummings SR, eds. Designing Clinical Research. Baltimore: Williams & Wilkins; 1988: 110-27. 22. Committee on Diet and Health, Food and Nutrition Board, Commission on Life Sciences, and National Research Council. Diet and health. Implications for reducing chronic disease risk. In: Hypertension. Washington, DC: National Academy Press; 1989:549-62. 23. Whelton PK, Klag MJ. Potassium in the homeostasis and reduction of blood pressure, Clin. Nutr. 1987;6:76-85. 24. Walker WC, Whelton PK, Russell RP, Saito H, Hermann J. Relation between blood pressure, renin, renin substrate, angiotensin II, aldosterone, and urinary sodium and potassium in 574 ambulatory subjects, Hypertension. 1979;1:287-91.