Multiple risk factor intervention trial revisited
ACP J Club. 1991 May-June;114:71. doi:10.7326/ACPJC-1991-114-3-071
Multiple Risk Factor Intervention Trial Research Group. Mortality after years for hypertensive participants in the Multiple Risk Factor Intervention Trial. Circulation. 1990;82:1616-28.
To assess mortality from coronary heart disease and from all causes in hypertensive men, during and after participation in a multiple risk factor, primary coronary heart disease prevention trial.
Randomized, unblinded trial, averaging 7 years follow-up, followed by surveillance of mortality only for 3.5 years on average.
22 clinical centers in 18 cities in the United States.
361 662 men were screened to find 12 866 who were at increased risk for coronary heart disease but did not have it overtly; 8012 of these were hypertensive and were the subjects of this report.
4019 received a special intervention (SI) program to lower cardiovascular risk, and 3993 were referred to their personal physicians for usual care (UC), starting in 1975. Blood pressure medications began with hydrochlorothiazide or chlorthalidone, up to 100 mg/d. In 1980, because of unfavorable mortality trends among patients receiving hydrochlorothiazide, this medication was replaced by chlorthalidone, at a lowered maximal dose of 50 mg/d.
Main outcome measure
Cause-specific mortality was based on death certificates. After the end of the trial in 1982, vital status was determined from national records.
After 6 years of intervention, systolic (SBP) and diastolic blood pressures (DBP) averaged 7.2 and 4.5 mm Hg lower in the SI group, plasma total cholesterol was 5.6 mg/dL lower in the SI group, and cigarette-smoking rates were 26% in the SI group and 40% in the UC group. For SI men, through 10.5 years, coronary heart disease mortality was 15% less (P = 0.19) and all-cause mortality was 11% less (P = 0.13). These rates appeared to represent the continuation of a favorable trend for men with initial DBP ≥ 100 mm Hg and for those without baseline resting ECG abnormalities, and reversal of an adverse trend for those with DBP < 100 mm Hg and those with abnormal baseline resting ECG abnormalities.
Among men in a multiple risk-factor intervention program, a favorable trend in mortality after the intervention period was associated with a change in the diuretic treatment protocol replacement of hydrochlorothiazide with chlorthalidone and a favorable effect of intervention on nonfatal cardiovascular events during the trial years.
Source of funding: National Heart, Lung, and Blood Institute.
Address for article reprint: Dr. M.O. Kjelsberg, Biostatistics Division, University of Minnesota, School of Public Health, Room 200, 2221 University Avenue SE, Minneapolis, MN 55414.
Strictly speaking, this report does not meet the criteria for inclusion in ACP Journal Club: The report focuses on the period after a randomized trial. However, the authors claim that the trial aftermath findings "support the inference" that multiple risk-factor intervention is beneficial for hypertensive patients. Clinical readers should always be wary of phrases such as "support the inference." In this study, despite an excellent research team, a large number of participants, an enthusiastically applied risk-factor intervention program, and superb follow-up, the study results were inconclusive.
Despite the large number of participants, these results may have been indeterminate because the trial was too small to detect a benefit. The estimates for risk from Framingham that were used in designing the study were out of date and excessive. Risk factors in the control group improved much more than intended. As a result, the differences in risk-factor profiles for the two groups were too small for much of a benefit to be seen despite the large numbers of participants.
Whatever the explanation, there were unexpected mortality trends observed during the trial, and this report shows that these trends seemed to right themselves following the study. This reversal may be caused by mid-trial changes in medication, as the investigators conclude. More likely, however, the reversal is simply an illustration of the phenomenon of "regression to the mean," that is, the tendency of unusual results in natural measurements (blood pressure, cholesterol, mortality, and so on) to return to normal on subsequent assessments. Unfortunately, the findings do not provide any clear guidance for clinical practice.
R. Brian Haynes, MD, PhD
McMaster University Hamilton, Ontario