The design and methods for MRFIT have been reported in detail
elsewhere (MRFIT 1977 JChronDis, MRFIT 1981 PrevMed, MRFIT 1982
JAMA). Briefly, MRFIT was a randomized controlled trial on the
primary prevention of CHD mortality among men ages 35 to 57 years
who were at increased risk but without definitive clinical evidence
of CHD (coronary heart disease) at baseline. From 1973 to 1975, 22 
clinical centers in 18 cities across the U.S. screened 361,662 men 
for participation.

Men were excluded at Screen 1 if they were at too low risk (their
cigarette smoking, serum cholesterol, and blood pressure measurements
did not place them in the upper 10-15% of a risk score distribution
based on the Framingham Study), or if they were at too high risk (had
a history of hospitalization for heart attack, diabetes mellitus
requiring medication, serum cholesterol 350 mg/dL or higher, or
diastolic blood pressure 115 mm Hg or higher.

Screen 2 excluded men for obesity, clinical evidence of CVD, untreated
symptomatic diabetes, diets incompatible with the MRFIT intervention
(including excessive alcohol intake), treatment with certain
medications (e.g. hypoglycemic or lipid-lowering agents), illnesses or
disabilities likely to restrict full trial participation, and
diastolic blood pressure 120 mm Hg or higher.

At Screen 3, informed consent was obtained and the intervention was
randomly assigned (stratified by clinic). The randomized intervention
was comprised of dietary counseling to lower cholesterol, smoking
cessation counseling, and stepped-care hypertension medication
(Caggiula 1981 PrevMed, Hughes 1981 PrevMed, Cohen 1981 PrevMed)
(called "special intervention" or SI). The placebo group (called
"usual care" or UC) remained under the care of their primary
physician.

On or about each anniversary of randomization, participants returned
to their clinical center for an examination by a MRFIT physician and
to complete behavioral, medical history, and 24 hour dietary recall
questionnaires. Fasting serum samples were collected for each
participant and were analyzed at a central laboratory (Widdowson 1986
CCT) for total serum cholesterol, lipid fractions, and fasting
glucose. Blood pressure was measured according to a standard protocol
by certified personnel (Cohen 1981 PrevMed). Systolic and diastolic
readings were recorded from the first and fifth Korotkoff phase
readings obtained with a random-zero sphygmomanometer while the
participant was seated. Blood pressure at baseline was defined as the
average of 2 measurements at Screen 2 and two measurements at Screen
3. Use of hypertension medication was recorded at each annual visit.

Age at randomization was determined from the participant's recorded
birthdate. Race (African-American vs. not) was determined by
self-report at Screen 1.  Disrobed height and weight were recorded at
Screen 2; we use Body Mass Index (BMI) in kg/m^2. Alcohol use was
determined at Screen 2 by self-report of the number of drinks (beer,
wine, whiskey, or other liquor) usually consumed per week. Cigarette
use as self-reported number smoked per day was recorded at Screen
1. Randomization group is coded as 1 for special intervention and 0
for usual care. Income was given by self-report at Screen 2. All
baseline values were collected prior to the beginning of the
intervention.

These observations reflect a sample of size 1000 of the randomized men
taken proportional to clinic size:

PROC FREQ data = dat ;
  TABLES clinic / out = freqcount ;
DATA new ;
  MERGE dat freqcount ;
  BY clinic ;
PROC SURVEYSELECT DATA = dat OUT = dat 
  METHOD = PPS SEED = 7162002 SAMPSIZE = 1000 ;
  SIZE count ;


Variables are as follows:

ID: PARTICIPANT ID
clinic: CLINIC
sbpbl: BASELINE SYSTOLIC BLOOD PRESSURE (mm Hg)