Premier: A Clinical Trial of Comprehensive Lifestyle Modification for

Blood Pressure Control: Rationale, Design and Baseline Characteristics

LAURA P. SVETKEY, MD, DAVID W. HARSHA, PHD, WILLIAM M. VOLLMER, PHD,
VICTOR J. STEVENS, PHD, EVA OBARZANEK, PHD, PATRICIA J. ELMER, PHD,
PAO-HWA LIN, PHD, CATHERINE CHAMPAGNE, PHD,
DENISE G. SIMONS-MORTON, MD, PHD, MIKEL AICKIN, PHD,
MICHAEL A. PROSCHAN, PHD, AND LAWRENCE J. APPEL, MD

PURPOSE: To describe PREMIER, a randomized trial to determine the effects of multi-component

lifestyle interventions on blood pressure (BP).
METHODS: Participants with above optimal BP through stage 1 hypertension were randomized to:
1) a behavioral lifestyle (BLS) intervention that implements established recommendations, 2) a BLS
intervention that implements established recommendations plus the DASH diet, or 3) an advice only
standard of care group. The two BLS interventions consist of group and individual counseling sessions
for 18 months. The primary outcome is systolic BP at 6 months. Additional outcomes include diastolic
BP and homocysteine at 6 months; systolic and diastolic BP at 18 months; fasting lipids, glucose and
insulin at 6 and 18 months; and effects in subgroup.
CONCLUSION: Results from the PREMIER trial will provide scientific rationale for implementing
multi-component behavioral lifestyle intervention programs to control BP and prevent CVD.
Ann Epidemiol 2003;13:462–471. 쑕 2003 Elsevier Inc. All rights reserved.
KEY WORDS: Diet, Blood Pressure, Weight Loss, Lifestyle.

INTRODUCTION (1). Less than half of adults have optimal BP, defined as
SBP ⬍ 120 mmHg and DBP ⬍ 80 mmHg. With increasing
Efforts to prevent morbidity and mortality from chronic age, the prevalence of hypertension rises such that over
diseases have largely focused on clinical management (pri- 50% of adults age 60 years and older have hypertension. A
marily pharmacologic) of individuals with existing disease. compelling body of evidence has documented that drug
Such an approach does not address the potentially revers- therapy can effectively lower BP and thereby prevent stroke
ible causes of these conditions, that is, adverse lifestyles and heart disease (2).
associated with suboptimal dietary habits and low levels of Still, reliance on drug therapy has well-recognized
physical activity.
shortcomings. First, the risk of CVD rises progressively
Elevated blood pressure (BP) is prototypic of these
throughout the BP range, including levels considered non-
chronic conditions. It is a common, powerful, and indepen-
hypertensive (3, 4). In fact, almost a third of BP-
dent risk factor for atherosclerotic cardiovascular disease
related deaths from coronary heart disease are estimated to
(ASCVD). Approximately 25% of the adult US population,
occur in individuals with SBP between 110 and 139 mmHg
about 43 million persons, has hypertension, defined as sys-
(5). Second, drug therapy requires an endless cycle of detec-
tolic BP (SBP) ⭓140 mmHg, diastolic BP (DBP) ⭓90
tion, treatment, and maintenance. The clinical resources
mmHg, and/or current use of antihypertensive medication
for these efforts are considerable. Furthermore, problems such
as lack of awareness (present in nearly 1/3 of hypertensives
(6)) and lack of access to health care in some population
groups reduce the potential effectiveness of drug therapy.
From Duke Hypertension Center and Sarah W. Stedman Center for Third, drug therapy can be expensive, can cause side effects,
Nutritional Studies, Duke University Medical Center, Durham, NC
(L.P.S., P.-H.L.); Pennington Biomedical Research Center, Baton Rouge, and effectively controls BP in only about 50% of those
LA (D.W.H., C.C.); Kaiser Permanente Center for Health Research, Port- treated in the US (6). Fourth, as highlighted above, drug
land, OR (W.M.V., V.J.S., P.J.E., M.A.); DECA, NHLBI, NIH, Bethesda,
MD (E.O., D.G.S.-M., M.A.P.); and Welch Prevention Center, Johns
therapy does not address the major underlying (non-genetic)
Hopkins Medical Center, Baltimore, MD (L.J.A.). causes of elevated BP, each of which can be addressed
Address correspondence to: Laura P. Svetkey, M.D., Duke Hypertension through lifestyle modification. Importantly, in contrast to
Center, 3020 Pickett Road, Durham, North Carolina 27516. Tel.: ⫹(919)-
419-5841. E-mail: [email protected] most drug therapies, lifestyle modifications that reduce BP
Received February 4, 2002; accepted October 14, 2002. can also prevent or control other chronic conditions.

쑕 2003 Elsevier Inc. All rights reserved. 1047-2797/03/$–see front matter

360 Park Avenue South, New York, NY 10010 doi:10.1016/S1047-2797(03)00006-1
AEP Vol. 13, No. 6 Svetkey et al. 463
July 2003: 462–471 PREMIER STUDY DESIGN

Selected Abbreviations and Acronyms

behavioral lifestyle interventions in persons with above-
optimal BP through stage 1 hypertension. One behavioral
ASCVD ⫽ atherosclerotic cardiovascular disease
BLS ⫽ behavioral lifestyle lifestyle intervention includes established clinical practice
BP ⫽ blood pressure guidelines articulated in JNC-V (weight loss if overweight,
BMI ⫽ body mass index reduced sodium intake, increased physical activity, and lim-
CVD ⫽ cardiovascular disease
DASH ⫽ dietary approaches to stop hypertension ited alcohol intake), while the other in addition incorpo-
DBP ⫽ diastolic BP rates the DASH diet. Each is compared with an advice only
JNC ⫽ Joint National Committee on Detection, Evaluation, Prevention standard of care group. It is anticipated that results from
and Treatment of High Blood Pressure
SBP ⫽ systolic BP the PREMIER trial will provide the scientific rationale and
guidance for the implementation of multi-component life-
style intervention programs designed to control BP and
Since 1993, the Joint National Committee on the Detec- ultimately prevent morbidity and mortality from cardiovas-
tion, Evaluation and Treatment of High BP (JNC) (7, 9) cular disease.
and the Working Group Report on Primary Prevention of
Hypertension (8) have recommended four lifestyle modifi-
cations for BP control: 1) weight loss if overweight, 2)
reduced sodium intake, 3) increased physical activity, and
4) limited alcohol consumption. Based on the results of the METHODS
Dietary Approaches to Stop Hypertension (DASH) feeding PREMIER is a multicenter randomized trial. Participating
study (9), the Sixth Report of the Joint National Committee institutions include the NHLBI Project Office, the Coordi-
(JNC-VI) also recommends the DASH diet, which is rich nating Center (Kaiser Permanente Center for Health Re-
in fruits, vegetables, and low-fat dairy products, and reduced search in Portland, Oregon), and four clinical centers (Johns
in saturated fat, total fat, and cholesterol (10). Each of Hopkins University in Baltimore, MD; Pennington Biomed-
these lifestyle modifications can lower BP. However, little ical Research Center in Baton Rouge, LA; Duke University
is known about the effectiveness of simultaneously im- Medical Center in Durham, NC; and a clinical center also
plementing all current recommendations. located at the Kaiser Permanente Center for Health Re-
To date, most trials that studied the effects of behav- search in Portland, OR). The Diet Assessment Center of
ioral lifestyle interventions on BP tested the impact of just the Pennsylvania State University performs diet recalls, and
one or two lifestyle changes (11, 12). Few trials tested the a central laboratory (the Core Laboratory for Clinical
effects of multi-component interventions that simulta- Studies at Washington University, St. Louis, MO) performs
neously implement all 1993 recommendations, and no trial analyses of blood lipids and homocysteine and of urine
has tested a multi-component intervention that implements specimens. The Nutritional Biochemistry Branch of the
all JNC-VI recommendations, including the DASH diet. Centers for Disease Control and Prevention performs analy-
In feeding studies, the DASH diet substantially lowered ses of blood micronutrients including folate and carotenoids.
blood pressure (9, 13); still, no trial has tested the impact Study funding began in September, 1998. Recruitment
of the DASH diet in free-living persons selecting their own began in September, 1999, and ended in June 2001. Data
diet. Another limitation of previous trials has been the lack collection was completed in December, 2002.
of emphasis on increased physical activity; to avoid the
potential for confounding by the BP-lowering effects of
exercise, weight loss trials often focused on reduced calorie Study Population
intake without emphasizing increased calorie expenditure. The target population consists of generally healthy adults
Lastly, because recent trials studied the impact of non- who met JNC-VI criteria for a 6 to 12-month trial of non-
pharmacologic interventions in persons with above optimal pharmacologic therapy (10). Persons were eligible if they
and high normal BP, even less is known about the effects of had above-optimal through high-normal BP (SBP 120–139
multi-component lifestyle interventions in hypertensives. and DBP 80–89 mmHg) or stage 1 hypertension (SBP 140–
Given the public health burden of elevated blood pressure 159 or DBP 90–95 mmHg), were not taking antihyperten-
and the potential to prevent and treat this condition, it is sive medication, and did not meet criteria for JNC-VI risk
important to test the effects of multi-component interven- category C (i.e., target organ damage, clinical cardiovascular
tions on BP and other CVD risk factors, both in persons at disease, and/or diabetes). Individuals with high-normal
risk for developing hypertension and in those with estab- blood pressure were of particular interest because of the
lished hypertension. potential for preventing the development of hypertension
In this setting, the overall aim of the PREMIER trial is to and the excess CVD risk associated with BP in this range
determine the BP-lowering effects of two multi-component (4). Individuals with stage 1 hypertension were included
464 Svetkey et al. AEP Vol. 13, No. 6
PREMIER STUDY DESIGN July 2003: 462–471

because of the potential for non-pharmacologic control of Board at each site. Each participant provided written in-
high blood pressure. formed consent.
Participants were required to be able and willing to partic-
ipate in the intervention programs and to complete follow- Randomization
up data collection. Specific inclusion and exclusion criteria Eligible participants were randomly assigned to one of three
are summarized in Table 1. groups: 1) Advice Only (Standard of Care); 2) a behavioral
intervention that includes established lifestyle recommen-
Recruitment
dations, i.e., weight loss if overweight, reduced sodium
At the time of this report, all study participants have been intake, increased physical activity, and limited intake of
enrolled and randomized. The PREMIER centers used a alcohol (Established); or 3) a behavioral intervention that
combination of mass mailing, community-based screening, includes the established recommendations in #2, plus
and mass-media announcements to recruit participants. Tar- DASH (Established Plus DASH) (Figure 1).
geted strategies to increase minority enrollment included Eligible participants attended a Randomization/Interven-
a) ethnically appropriate mailings targeted to certain zip tion visit at which baseline measurements were collected
codes, b) special community-based screening events, and c) and final consent obtained. At this visit, the randomization
advertisements, articles, and public service announcements assignment was generated by a computer program developed
on radio stations and in newspapers serving minority popula- by the Coordinating Center, and the first intervention
tions. The study was approved by the Institutional Review session, an individual visit, also occurred. Group assignments
were stratified by clinic and baseline BP category (non-
TABLE 1. Eligibility Criteria hypertensive vs. hypertensive) and within these categories
were blocked to provide a balance in treatment assign-
Inclusion criteria
Baseline SBP 120–159 mmHg and DBP 80–95 mmHg ments over time. For logistical purposes, participants were
Age 25 or older randomized in 4 cohorts over time.
BMI 18.5–45 kg/m2
Willing and able to participate fully in all aspects of the intervention Masking
Informed consent
Access to telephone Clinical center staff involved in follow-up data collection
Medication exclusions are masked to participant treatment assignments, while
Regular use of anti-hypertensive drugs or other drugs that raise intervention staff are masked to all outcome data. Partici-
or lower BP pants are told their baseline blood pressure measurements and
Current use of insulin or oral hypoglycemic agents
Current use of medications for treatment of psychosis or also receive a summary of their blood pressure measurements
manic-depressive illness following data collection at the 6-month visit. Provision of
Use of weight-loss medications in the 3 months prior to first such information is appropriate in view of the fact that
screening visit many participants will have stage 1 hypertension requiring
Medical history exclusions
referral for possible BP medication if their BP remains ele-
Cardiovascular event
Congestive heart failure
vated after 6 months of non-pharmacologic therapy (JNC-
Current symptoms of angina or peripheral vascular disease VI). (10)
Cancer diagnosis (except for non-melanoma skin cancer) or treatment
in past two years Interventions
Renal insufficiency (GFR ⬍ 60 ml/min as estimated using Cockroft-
Gault formula) The theoretical basis of the PREMIER behavioral interven-
Random glucose ⭓ 160 mg/dL or Fasting Blood Sugar ⭓ 126 mg/dl tions derives from social cognitive theory (14), self-applied
Psychiatric hospitalization within the last 2 years behavior modification techniques i.e., “behavioral self-man-
Other exclusions agement” (15), and the transtheoretical “stages of change”
Inability to provide acceptable BP measurements
Consumption of more than 21 alcoholic drinks per week or model (16,17). These approaches emphasize the importance
binge drinking of an individual’s ability to regulate behavior by setting
Planning to leave the area prior to the anticipated end of participation goals, monitoring progress towards the goals, and at-
Body weight change ⬎ 15 pounds in the 3 months prior to taining skills necessary to reach the goals. These approaches
first screening visit also seek to increase self-efficacy (one’s confidence in per-
Pregnant, breast feeding, or planning pregnancy prior to
the end of participation forming a given behavior) and outcome expectancies (one’s
Current participation in another clinical trial expectations concerning the outcome of that behavior),
Investigator discretion for safety or adherence reasons both of which are critical mediators of behavior change
Household member of another PREMIER participant or of a (14,18). The transtheoretical model also recognizes that
PREMIER staff member
behavior change is a dynamic process of moving through
AEP Vol. 13, No. 6 Svetkey et al. 465
July 2003: 462–471 PREMIER STUDY DESIGN

FIGURE 1. PREMIER design.

different motivational stages of readiness for change. Differ- intake, and eating the DASH diet. The Advice Only pro-
ent behavioral strategies may be emphasized depending on gram is provided at two 30-minute individual visits, one at
the individual’s stage of change. In addition, the PREMIER randomization and the second after 6-month data collec-
behavioral lifestyle interventions emphasize motivational tion. At these visits the interventionist reviews the recom-
counseling techniques (19, 20, 21). The intervention staff mended guidelines, gives advice, and provides printed
includes nutritionists or health educators who have been educational materials, but does not provide behavioral coun-
trained, through the PREMIER protocol, in behavioral seling. At the end of 18 months, upon completion of all data
techniques. collection, the Advice Only participants receive additional
This theoretical basis of PREMIER contrasts sharply with advice and counseling.
the delivery of lifestyle modifications in the usual clinic Behavioral Lifestyle Interventions. The Established
setting. The typical model involves physician advice, occa- and Established Plus DASH interventions differ from each
sionally with referral to a dietitian who takes a diet history other in two areas: dietary goals and approach to weight loss.
and who then provides educational materials and advice Participants randomized to the Established intervention do
during a small number of individual counseling sessions. not receive instruction and counseling regarding DASH
The typical clinical approach, which is lacking the theoreti- dietary recommendations, whereas participants in the Estab-
cal basis and behavioral tools noted above, but which reflects lished Plus DASH intervention do receive this instruction
current clinical experience, is the basis of the Advice Only and counseling. The dietary goals of the DASH diet include
Standard of Care group in PREMIER. increased consumption of fruits, vegetables and low-fat dairy
Advice Only Standard of Care. Participants assigned products and reduced intake of saturated and total fat (9,22).
to this group receive advice to follow the National High Weight loss in the Established Plus DASH intervention
Blood Pressure Education Program recommendations (10). is achieved by reducing calorie intake, increasing physical
Advice includes weight loss if overweight, reducing sodium activity, and by substituting fruits and vegetables for high-
intake, increasing regular physical activity, limiting alcohol fat, high-calorie foods. In contrast, weight loss in the
466 Svetkey et al. AEP Vol. 13, No. 6
PREMIER STUDY DESIGN July 2003: 462–471

Established intervention achieves weight loss primarily by defined by age, sex and race. Other measurements listed in
reducing calorie intake and increasing physical activity. In Table 3 are used to assess eligibility and adherence to the
both active intervention groups, participants are counseled interventions. Primary outcome assessment occurs at 6
to reduce sodium intake to less than 2400 mg/day. They months for several reasons. First, clinical practice guidelines
are instructed on ways to identify the sodium content of (10) recommend that those with persistent SBP ⭓140 or
food, to select and substitute lower sodium choices, and to DBP ⭓90 mmHg be referred for possible medication treat-
alter sodium content of recipes. ment. Second, because blood pressure data are censored
The structure, basic format, and contact pattern of the when a participant initiates drug therapy, at 6 months we will
Established and Established Plus DASH interventions are have maximum BP data for analysis. And third, adherence
identical. During the initial 14 weeks, group intervention to lifestyle interventions commonly peaks at 6 months. Al-
sessions for both behavioral interventions occur approxi- though the primary outcome assessment occurs at 6 months,
mately weekly. During the next 14 weeks, group interven- analysis of the 18-month data is a critical element of the
tion sessions occur every other week (total of six sessions); design of PREMIER. Previous trials suggest that intensive
a single individual session also occurs during this period. lifestyle interventions achieve their greatest effects on blood
Thereafter, the intervention includes monthly group meet- pressure early on, and that these effects tend to diminish
ings and three quarterly individual counseling sessions. Par- over time. Hence, from a public health point of view it is
ticipants assigned to either behavioral intervention keep important to evaluate the durability of the effects.
food diaries and records of their physical activity, and they All BP measurements are performed by trained, certified
monitor energy, fat gram and sodium intake. Participants in observers using a random zero sphygmomanometer (23) with
the Established Plus DASH intervention also keep records the participant seated, following procedures that are similar
of the number of servings of fruits, vegetables, and dairy to those used in prior studies (22, 24, 25, 26). Each time
products. These records are used for self-monitoring, and blood pressure is measured, a pair of measurements are taken
are also used to provide individualized feedback on the 30 seconds apart. At both the 6-month and the 18-month
participant’s progress. Goals and measures of adherence of assessments, three to four pairs of BP measurements are
the two behavioral interventions are displayed in Table 2. taken over a three-month interval. A single pair of BP
measurements is obtained at the three- and twelve-month
Study Measurements visits for safety monitoring. A participant is considered hy-
pertensive at the six- or 18-month measurement visits if
Table 3 displays the types and schedule of measurements.
the cumulative average over that set of visits is ⭓140
Clusters of measurements occur at 6 and 18 months. The
mmHg systolic, ⭓90 mm/Hg diastolic, or the participant is
primary outcome is systolic BP measured at 6 months. Addi- taking anti-hypertensive medication.
tional outcomes include diastolic BP at 6 months, and
systolic and diastolic BP at 18 months; incidence and
prevalence of hypertension at 6 and 18 months; blood homo- Analysis and Power
cysteine at 6 months; fasting lipids, glucose and insulin at Analysis. The following linear regression model will be
6 and 18 months; and effects in demographic subgroups employed to assess the effects of treatment on change in

TABLE 2. Goals and measures of adherence for PREMIER lifestyle interventions

Established Established ⫹ DASH
Lifestyle factor Goal Goal Adherence measures
Physical activity 180 min/wk 180 min/wk 7-day activity recall, fitness test
Sodium ⭐ 100 mmol/d ⭐ 100 mmol/d 24-hr urine Na; 24-hr recall
Alcohol ⭐ 1 oz/d (men), ⭐ 1 oz/d (men), Alcohol questionnaire
⭐ 0.5 oz/d (women) ⭐ 0.5 oz/d (women)
Weight* ⭓ 15 lb loss ⭓ 15 lb loss Weight
Total calorie intake Individualized Individualized 24-hr diet recall
%kcal fat ⭐ 30% ⭐ 25%** 24-hr diet recall
%kcal sat fat ⭐ 10% ⭐ 7%** 24-hr diet recall
Fruits & vegetables Not specified 9–12 servings/d¶ 24-hr diet recall
24-hr urinary K
Low-fat dairy Not specified 2–3 servings/d¶ 24-hr diet recall
24-hr urinary phosphorus
*Weight loss recommended only for those with BMI ⭓ 25.
¶
Daily number of servings adjusted for individual caloric intake.
**Aspects of DASH dietary pattern.
AEP Vol. 13, No. 6 Svetkey et al. 467
July 2003: 462–471 PREMIER STUDY DESIGN

TABLE 3. Schedule of PREMIER measurements Power Considerations

Pre- Based on data from TOHP-I (30), DASH (9), and DASH-
Months post randomization
randomization
Sodium (13), we estimated that the standard deviation of
Measure Baseline 3 6 12 18
change in SBP would be 5.32–5.92 mmHg, providing 90%
Demographics & height X
power for a sample size of 800 participants (267/group) to
Weight X X X X X
Blood pressure1 X X X X X detect pair-wise between-group differences in SBP of 1.6–
Rose angina questionnaire X X X X X 1.8 mmHg, 3.2–3.6 mmHg among hypertensives (30% of
Rose peripheral vascular the sample), and 1.7–1.9 mmHg among non-hypertensives.
disease questionnaire X Differences of 2.6–2.9 mmHg would be detectable with 90%
Current medication use X X X X X power among African Americans (40% of the sample). In
Symptom questionnaire X X X X X
24-hr diet recall2 X X X each case, effects of this magnitude are considered important
7-day physical activity recall X X X for public health (31).
Sub-maximal treadmill test X X X
Waist circumference X X X
Fasting blood3 X X X
Alcohol intake X X X BASELINE CHARACTERISTICS
Psychosocial questionnaires4 X X X
24-hour urine5 X X X Eight hundred eleven individuals were randomized into the
1
BP (one pair) measured four times at baseline, 3–4 times at 6 and 18 month visits, trial, 11 more than the goal of 800. After randomization,
and once at 3 and 12 month visits. At each time point, measurements are obtained one participant subsequently reported that he was on antihy-
over a three-month period.
2
Diet recall performed twice at each measurement point. pertensive medication at entry into the study; data from
3
Total-cholesterol, LDL-cholesterol, HDL-cholesterol, triglycerides, homocysteine this individual will be excluded. Table 4 displays baseline
(6 months only), folate, carotenoids, vitamin E, and retinol.
4
Self-efficacy and social support for physical activity and dietary behavior, quality of characteristics of the randomized participants, overall and
5
life, perceived stress, and body image. by gender and race-ethnicity. Approximately 34% of partici-
Sodium, potassium, phosphorus, creatinine and urea nitrogen.
pants are African Americans and 62% women. Among Af-
rican Americans, women predominate. The PREMIER
SBP from baseline to 6 months post-randomization: participants are generally overweight (95% with BMI ⭓ 25
kg/m2). Thirty-seven percent have stage 1 hypertension, and
∆y ⫽ βA ⫹ β0y0 ⫹ βExE ⫹ βDxD ⫹ 兺α s ⫹ 兺γ c ⫹ e
i
i i
j
j j
plasma cholesterol is slightly above average for American
adults (32).

Here ∆y is the change (6 months minus baseline) in SBP,

y0 is the baseline SBP, xE and xD are indicator variables of
the Established (E) and Established plus DASH (D) inter- DISCUSSION
vention groups. The si terms are indicators of the The health benefits of the PREMIER lifestyle interventions
performance sites, and the cj terms are indicators of the could be substantial. In hypertensive patients, each compo-
cohorts. The e-term is assumed to be independent of all nent of the PREMIER interventions (weight loss, sodium
explanatory factors. reduction, physical activity, limited alcohol intake, and the
The three primary null hypotheses are that the Estab- DASH diet) has been shown separately to reduce BP, often
lished intervention is not different from the Advice Only by 5–10 mmHg in efficacy studies. While corresponding BP
Standard of Care intervention (H1: βE ⫽ 0), that the Estab- reductions tend to be smaller in non-hypertensive individu-
lished Plus DASH intervention is not different from Advice als, the net effect of multiple small effects is the prevention
Only (H2: βD ⫽ 0), and that the Established and Established of hypertension (11). The effects of the lifestyle inter-
Plus DASH interventions are not different from each other ventions on cardiovascular outcomes could likewise be im-
(H3: βE ⫽ βD). Each of these hypotheses will be tested using pressive. Population-wide adoption of the DASH dietary
conventional regression methods. To control the multiple pattern could reduce the incidence of coronary heart disease
testing type I error, H1 and H2 will be tested using Holm’s by 15% and stroke by 27% in US adults, just from BP
method adjusting for two comparisons (27,28). H3 will be reduction (2). In addition to BP reduction, the lifestyle
tested at the 0.05 level, but only if either of H1 or H2 interventions should also reduce cardiovascular risk through
are rejected. As shown by Proschan (29), this procedure effects on other traditional risk factors, such as plasma lipids
bounds the multiple testing type I error at 0.05. Analyses (33) and on non-traditional risk factors, such as homocyste-
of secondary outcomes and analyses in hypertensives and ine (34). Hence, the above estimated reductions in coronary
non-hypertensives separately will use the same model pre- heart disease and stroke risk likely underestimate the poten-
sented above. tial benefit of the PREMIER interventions.
468 Svetkey et al. AEP Vol. 13, No. 6
PREMIER STUDY DESIGN July 2003: 462–471

TABLE 4. Baseline characteristics of the PREMIER study population, mean (SD) or percent
Men Women
Measure Total Af Am Non- Af Am Af Am Non-Af Am
N randomized N ⫽ 811 N ⫽ 72 N ⫽ 239 N ⫽ 205 N ⫽ 295
SBP(mmHg) 134.9 (9.6) 133.3 (9.0) 133.5 (9.1) 135.5 (9.3) 135.9 (10.1)
DBP(mmHg) 84.8 (4.2) 85.6 (4.1) 85.9 (4.1) 84.4 (4.3) 83.9 (3.9)
% hypertensive 37% 31% 37% 40% 38%
BMI (Kg/m2) 33.6 (5.5) 34.3 (6.0) 32.5 (5.1) 34.2 (5.4) 34.0 (5.5)
% overweight (BMI ⭓ 25) 95% 94% 95% 99% 92%
Age(year) 50.0 (8.9) 48.2 (9.9) 50.8 (9.2) 48.5 (9.2) 50.7 (8.0)
Education
Grade school 0% 0% 0% 0% 0%
Some high school 1% 1% 0% 1% 1%
Complete high school 8% 10% 4% 9% 9%
Some college 34% 32% 23% 42% 37%
Complete college degree 25% 29% 26% 25% 23%
Postgrad work 32% 28% 46% 21% 29%
Income
No answer 4% 4% 3% 4% 3%
⬍$29,999 10% 11% 4% 23% 6%
$30,000–$59,999 32% 33% 21% 36% 37%
$60,000 and higher 55% 51% 72% 37% 54%
Urinary sodium (mg/24 hrs) 4014 (1673) 4543 (1829) 4651 (1758) 3566 (1382) 3685 (1560)
Urinary potassium (mg/24 hrs) 2600 (1016) 2528 (928) 3109 (1030) 2029 (799) 2610 (949)
Total cholesterol (mg/dl)1 212 (38) 210 (37) 213 (38) 204 (38) 217 (38)
LDL-cholesterol (mg/dl)1 135 (34) 141 (30) 138 (35) 133 (35) 134 (33)
HDL-cholesterol (mg/dl)1 48 (13) 45 (10) 40 (8) 53 (13) 53 (13)
Triglycerides (mg/dl)1,2 134 (73) 113 (60) 169 (80) 91 (46) 143 (70)
1
Lipid data set to missing for participants on lipid lowering drugs.
2
Triglyceridesset to missing for participants with value more than 400 mg/dl.
WL4020102
Request # 402-01

Several aspects of the trial design are noteworthy. First, that can compare the effects on blood pressure of interven-
PREMIER is testing the recommendations of policy-making tions with different dietary approaches to weight loss.
organizations. One group receives a behavioral lifestyle mod- The contrasts between each pair of the PREMIER inter-
ification intervention that implements the established set ventions have considerable scientific and public health in-
of lifestyle recommendations (7). The other group receives terest. The comparisons of each of the behavioral inter-
a behavioral intervention that implements JNC-VI recom- ventions (Established and Established Plus DASH) vs. the
mendations (10), that is, the established recommendations Advice Only intervention are of primary interest and will
as well as the DASH dietary pattern. Second, to date, few likely determine the upper limits of BP reduction that is
trials have tested the effects of comprehensive lifestyle modi- achievable from these multicomponent behavioral lifestyle
fication, and none has tested the impact of the DASH interventions among free-living individuals. Also of interest
dietary pattern. Most previous trials have attempted to iso- is the comparison of the Established Plus DASH interven-
late the effects on blood pressure of single factors, such tion vs. the Established intervention. While a significant
as weight loss alone, sodium reduction alone and physical difference between these two arms, for instance superiority
activity alone, or of two factors, typically sodium reduc- of the Established Plus DASH intervention over the Estab-
tion and weight loss (11, 12). The efficacy of the DASH lished intervention, would provide support for a public
diet has been demonstrated in controlled feeding studies health policy that advocates implementation of the DASH
(9, 13) but not in trials of free-living individuals. PREMIER dietary pattern in the setting of established lifestyle recom-
offers an opportunity to determine the extent to which mendations, a finding of no significant difference between
free-living people can adhere to a multi-component lifestyle these two intervention arms would also provide important
intervention including the DASH dietary pattern and to information. Such a result may reflect either no true differ-
the DASH diet itself. Third, the emphasis on physical activ- ence, subadditivity of the effects of lifestyle changes, or the
ity in PREMIER is unusual among trials of comprehensive inability of participants to make all of the changes we are
interventions. Finally, PREMIER is one of the only studies asking them to make. Regardless of the outcome, additional
AEP Vol. 13, No. 6 Svetkey et al. 469
July 2003: 462–471 PREMIER STUDY DESIGN

insight into the observed results will come from secondary to PREMIER once efficacy is documented. Examples of
analyses that document the extent to which participants existing behavioral intervention programs that are offered
were able to achieve the various intervention targets and by health care delivery systems include: smoking cessation
from post hoc multiple regression analyses that assess the programs, some of which include 10 to 15 group sessions
relative effects of the various intervention components on (39,40); diabetes education programs; and cardiac rehabili-
blood pressure. tation. These programs are currently reimbursed by health
Other important design considerations include the selec- insurance and are provided in both fee-for-service settings
tion of the blood pressure measurement technique and the as well as in managed care organizations. Hence there
choice of study population. We chose the random-zero is precedent for implementation of lifestyle interventions
sphygmomanometer because we had excellent information for control of chronic conditions using a similar intervention
about sources of variance. The random-zero sphygmoma- delivery approach.
nometer also reduces observer bias, a critical issue when In summary, results from the PREMIER trial will provide
recruiting participants within a relatively narrow range of the scientific rationale for routinely implementing multi-
blood pressure and when ascertaining a clinical diagnosis component lifestyle intervention programs designed to con-
of hypertension during follow-up. We considered 24 h ambu- trol BP and ultimately prevent BP-related cardiovascular
latory blood pressure monitoring, but were concerned about disease.
the participant burden and the potential for missing data.
We also considered automated devices. However, few de-
vices have been validated (35), and the most common The authors thank Jeffrey A. Cutler, MD and Lawton S. Cooper, MD at
devices have a substantial systematic measurement error NHLBI for their valuable contributions to the design of PREMIER. We also
thank Christine Pfeiffer PhD, Chief, NHANES and Global Micronutrient
and digit preference. Laboratory; Elaine W. Gunter MT, Deputy Director, Division of Laboratory
The study population includes both hypertensive and Sciences, Centers for Disease Control and Prevention; Thomas G. Cole,
non-hypertensive individuals. We decided to include stage PhD, Director, Core Laboratory for Clinical Studies, Washington Univer-
1 hypertensives as this is a clinically meaningful population sity in St. Louis; Helen S. Wright, PhD, Principal Investigator, Pennsylva-
that should receive a trial of lifestyle modification prior nia State University Diet Assessment Center; Diane C. Mitchell,
Investigator, Pennsylvania State University Diet Assessment Center; Shir-
to drug therapy. However, the inclusion of hypertensive iki Kumanyika, PhD, MPH, Center for Clinical Epidemiology and Biostatis-
individuals necessitates using blood pressure measurements tics, University of Pennsylvania School of Medicine; Jerome Williams, PhD,
at 6 months as our primary outcome: according to national Howard University; Leslie Pruitt, PhD, Stanford Center for Research in
guidelines, we are committed to refer individuals with persis- Disease Prevention, Stanford University School of Medicine; and Abby
King, PhD. We also thank the members of the Data, Safety and Monitor-
tently elevated blood pressure after 6 months of lifestyle
ing Board that included Jerome D Cohen (chair), Nancy R Cook, ScD,
treatment for possible drug therapy. In addition, we in- Patricia Dubbert, PhD, Keith C Ferdinand, MD, Jim Raczynski, PhD, and
clude persons with above-optimal BP because this is a popu- Linda Van Horn, PhD.
lation at risk of developing hypertension and, within this In addition, we thank the following individuals at each of the participat-
group, those with high-normal BP (130–139/85–59 mmHg) ing sites:
are at risk of BP-related CVD morbidity and mortality (4). From the Coordinating Center, Center for Health Research, Port-
land Oregon: Jack Hollis, PhD, Njeri Karanja, PhD, Fran Heinith, BSN,
In this population non-pharmacologic interventions offer Kristy Funk, MS, RD, Michael Allison, BS, Gayle Meltesen, MS, Carrie
the potential to prevent hypertension. Souvanlasy.
Results from PREMIER should influence policy per- From Duke University Medical Center, Durham, NC: Colleen
taining to implementation of lifestyle modification in the McBride, PhD, Jamy Ard, MD, Kathleen Aicher, Blondeaner Brown, Denise
contemporary management of patients with above-optimal Ernst, Jeanne Gresko, Madhuri Kesari, Femke Lamers, LaTonya Nealon,
Tori Phelps, LaVerne Pruden, LaChanda Reams, Patrice Reams, Benja-
blood pressure through stage 1 hypertension. Despite re- min Reese, PhD, Fran Rukenbrod, Sonia Steele, Natalie Thorpe,
peated, widely distributed guidelines on lifestyle modifica- Olaunda Williams, Chenghua (Cherry) Yang.
tions, implementation of such guidelines in routine medical From Pennington Biomedical Research Center, Baton Rouge, LA:
practice is suboptimal. At present, implementation relies Philip Brantley, Allison Worthen, Betty Kennedy, Emily Griffin, Erma
on physicians who provide information on lifestyle modifi- Levy, Terri Keller, Shantell Jones, Katherine Lastor.
From Johns Hopkins Medical Center, Baltimore, MD: Barbara Bailey,
cation, often without the assistance of health educators or MS, RD, Jeanne Charleston, RN, MSN, Sharrone Cypress, Arlene Dalcin,
dietitians. Unfortunately, even when referral to a dietitian MS, RD, Maura Deeley, Charalett Diggs, RN, Thomas P. Erlinger, MD,
is part of the treatment plan, the frequency and intensity MPH, Ann Fouts, RN, Angela Hall, Charles Harris, Tara Harrison, Megan
of these contacts are insufficient to effect much change (36, Jehn, Shirley Kritt, Estelle Levitas, Phyllis McCarron, MS, RD, Edgar
Miller, MD, PhD, Pauline Patrick, LD, Charles Powell, Thomas Shields,
37). However, contemporary health care systems, such
LeeLana Thomas, MS, RD, Letitia Thomas, Bobbie Weiss, Deborah
as managed care organizations, increasingly offer specialized Young, PhD.
health education and prevention programs (38) and could From the Center for Health Research clinical site, Portland, OR: Adri-
potentially implement behavioral interventions similar anne C. Feldstein, MD, MS, Daniel S. Laferriere, RN, MSN, Shirley R.
470 Svetkey et al. AEP Vol. 13, No. 6
PREMIER STUDY DESIGN July 2003: 462–471

Craddick, RD MHA, Dana R. Larson RD, MS, Diane J. Cook, RD, 15. Watson DL, Tharp RG. Self-Directed Behavior: Self-Modification for
MPH, Carol L. Young, Susan D. Arnold, RN, BSN, Donna L, Stanley Personal Adjustment. 5th Ed. Pacific Grove, California: Brooks/Cole,; 1989.
B. Postlethwaite, Titza Suvalcu-Constantin, Carol M. Maul, Donna M. 16. Prochaska JO, DiClemente CC. Stages and process of self-change in smok-
Gleason, Cheryl A. Johnson, MEd, Pamela G. McNeal, Debra D. Burch. ing: Towards an integrative model of change. J Consult Clin Psychol.
This work is supported by NIH grants UO1 HL60570, UO1 HL60571, 1983;51:390–395.
UO1 HL60573, UO1 HL60574, UO1 HL62828. 17. Marcus BH, Banspach SW, Lefebvre RC, Rossi JS, Carleton RA, Abrams
DB. Using the stages of change model to increase the adoption of physical
activity among community participants. Am J Health Promot. 1992;
6:424–429.

REFERENCES 18. Bandura A. Self-efficacy: The exercise of control. New York: W. H. Free-
man,; 1997.
1. Burt VL, Whelton P, Roccella EJ, Brown C, Cutler JA, Higgins M, et al. 19. Miller WR, Rollnick S. Motivational Interviewing: Preparing People to
Prevalence of hypertension in the US adult population: Results from the Change Addictive Behavior. New York: The Guilford Press; 1991.
Third National Health and Nutrition Examination Survey, 1988–1991.
Hypertension. 1995;25:305–313. 20. Rollnick S, Miller W. What is motivational interviewing? Behavioral
Cognitive Psychotherapy. 1995;23:325–334.
2. Cutler JA, Psaty BM, MacMahon S, Furberg CD. Public health issues in
hypertension control: what has been learned from clinical trials. In: Laragh 21. Rollnick S, Mason P, Butler C. Health Behavior Change: A Guide for
JH, Brenner BM, eds. Hypertension: pathophysiology, diagnosis, and man- Practitioners. New York: Churchill Livingstone; 1999.
agement. 2nd ed. v. 1. New York: Raven Press.; 1995:253–270. 22. Sacks FM, Obarzanek E, Windhauser MM, Svetkey LP, Vollmer WM,
3. MacMahon S, Peto R, Cutler J, Collins R, Sorlie P, Neaton J, et al. Blood McCullough M, Karanja N, Lin P-H, Proschan MA, Appel LJ, Bray
pressure, stroke, and coronary heart disease. Part 1, prolonged differences GA, Vogt TM, Moore TJ, for the DASH investigators, et al. Rationale
in blood pressure: prospective observational studies corrected for the re- and design of the Dietary Approaches to Stop Hypertension trial (DASH).
gression dilution bias. Lancet. 1990;335:765–774. Ann Epidemiol. 1995;5:108–118.
23. Wright BM, Dore CF. A random-zero sphygmomanometer. Lancet. 1970;
4. Vasan RS, Larson MG, Leip EP, Evans JC, O’Donnell CJ, Kannel WB,
1:337–338.
et al. Impact of high-normal blood pressure on the risk of cardiovascular
disease. N Engl J Med. 2001;345:1291–1297. 24. SHEP Cooperative Research Group. Prevention of stroke by antihyper-
tensive drug treatment of older persons with isolated systolic hypertension:
5. Stamler J, Stamler R, Neaton JD. Blood pressure, systolic and diastolic,
Final results of the Systolic Hypertension in the Elderly Program (SHEP).
and cardiovascular risks: US population data. Arch Intern Med. 1993;
JAMA. 1991;265:3255–3264.
153:598–615.
25. Satterfield S, Cutler JA, Langford HG, Applegate WB, Borhani NO, Brit-
6. Burt VL, Cutler JA, Higgins M, Horan MJ, Labarthe D, Whelton P,
tain E, et al. Trials of Hypertension Prevention: Phase I design. Ann
et al. Trends in the prevalence, awareness, treatment, and control of hyper-
Epidemiol. 1991;1:455–471.
tension in the adult US population. Data from the Health Examinations
Surveys, 1960 to 1991. Hypertension. 1995;26:60–69. 26. Svetkey LP, Sacks FM, Obarzanek E, Vollmer WM, Appel LJ, Lin P-H,
et al. The DASH diet, Sodium Intake and Blood Pressure Trial (DASH-
7. Joint National Committee on Detection. Evaluation, and Treatment of
Sodium): rationale and design. J Am Diet Assoc. 1999;99(Suppl):S96–
High Blood Pressure: The Fifth Report of the Joint National Committee
S104.
on the Detection, Evaluation, and Treatment of High Blood Pressure (JNC-
V). Arch Intern Med. 1993;153:154–183. 27. Holm S. A simple sequentially rejective multiple test procedure. Scand J
Stat. 1979;6:65–70.
8. Working Group on Primary Prevention of Hypertension. National High
Blood Pressure Education Program: Working Group Report on Primary 28. Aickin M, Gensler H. An improvement in the reporting of results while
Prevention of Hypertension; 1993 (NIH Publication No. 93–2669). adjusting for multiple testing. Am J Public Health. 1996;86:726–728.
9. Appel LJ, Moore TJ, Obarzanek E, Vollmer WM, Svetkey LP, Sacks FM, 29. Proschan MA. A multiple comparison procedure for three- and four-armed
et al. A clinical trial of the effects of dietary patterns on blood pressure. controlled trials. Stat Med. 1999;18:787–798.
N Engl J Med. 1997;336:1117–1124. 30. The Trials of Hypertension Prevention Collaborative Research Group. The
10. Joint National Committee on Prevention, Detection, Evaluation, and effects of nonpharmacologic interventions on blood pressure of persons
Treatment of High Blood Pressure. The Sixth Report of the Joint with high-normal levels: Results of the Trials of Hypertension Prevention,
National Committee on the Prevention, Detection, Evaluation, and Phase I (TOHP-I). JAMA. 1992;267:1213–1220.
Treatment of High Blood Pressure (JNC-VI). Arch Intern Med. 31. Cook NR, Cohen J, Hebert PR, Taylor JO, Hennekens CH. Implications
1997;157:2413–2446. of small reductions in diastolic blood pressure for primary prevention. Arch
Intern Med. 1995;155:701–709.
11. The Trials of Hypertension Prevention Collaborative Research Group.
Effects of weight loss and sodium reduction intervention on blood pressure 32. Johnson CL, Rifkind BM, Sempos CT, Carroll MD, Bachorik PS, Briefel
and hypertension incidence in overweight people with high-normal blood RR, et al. Declining serum total cholesterol levels among US adults.
pressure: The Trials of Hypertension Prevention, Phase II (TOHP-II). Arch JAMA. 1993;269:3002–3008.
Intern Med. 1997;157:657–667. 33. Obarzanek E, Sacks FM, Vollmer WM, Bray GA, Miller ER, Lin P-H,
12. Whelton PK, Appel LJ, Espeland MA, Applegate WB, Ettinger WH, Kostis et al. Effects of a blood pressure lowering diet on blood lipids: the Dietary
JB, et al. Sodium reduction and weight loss in the treatment Approaches to Stop Hypertension (DASH) trial. Am J Clin Nutr. 2001;
of hypertension in older persons: A randomized controlled trial of non- 74:80–89.
pharmacologic interventions in the elderly (TONE). JAMA. 1998; 34. Appel LJ, Miller ER III, Jee SH, Stolzenberg-Solomon R, Lin P-H, Erlinger
279:839–846. T, et al. Effect of dietary pattern on serum homocysteine: results of a
13. Sacks FM, Svetkey LP, Vollmer WM, Appel LJ, Bray GA, Harsha D, randomized, controlled feeding study. Circulation. 2000;102:852–857.
et al. DASH-Sodium Collaborative Research Group. Effects on blood pres- 35. O’Brien E, Coats A, Owens P, Petrie J, Padfield PL, Littler WA, et al. Use
sure of reduced dietary sodium and the Dietary Approaches to Stop Hyper- and interpretation of ambulatory blood pressure monitoring: recommenda-
tension (DASH) diet. DASH-Sodium Collaborative Research Group. tions of the British Hypertension Society. BMJ. 2000;320:1128–1134.
N Engl J Med. 2001;344:3–10. 36. Solberg LI, Kottke TE, Brekke ML, Calomeni CA, Conn SA, Davidson
14. Bandura A. Social Foundations of Thought and Action. A Social Cognitive G. Using continuous quality improvement to increase preventive services
Theory. Englewood Cliffs: Prentice Hall; 1986. in clinical practice—going beyond guidelines. Prev Med. 1996;25:259–267.
AEP Vol. 13, No. 6 Svetkey et al. 471
July 2003: 462–471 PREMIER STUDY DESIGN

37. Ammerman A, Caggiula A, Elmer PJ, Kris-Etherton P, Keyserling T, Lewis 39. Zapka JG, Merriam P, Ockene J. Smoking cessation benefits in HMOs.
C, et al. Putting medical practice guidelines into practice: The cholesterol HMO Pract. 1997;11:27–33.
model. Am J Prev Med. 1994;10:209–216. 40. McAfee T, Wilson J, Dacey S, Sofian N, Curry S, Wagener B. Awaking
38. Budd MA, Gruman JC. Behavioral medicine: Taking its place in the the sleeping giant: Mainstreaming efforts to decrease tobacco use in an
mainstream of primary care. HMO Pract. 1995;9:51–52. HMO. HMO Pract. 1995;9:138–143.