Background A mortality follow-up of 12 866 men was conducted 16 years after randomization to special intervention (SI) or usual care (UC) groups of the Multiple Risk Factor Intervention Trial to assess the long-term effect of cardiovascular risk factor intervention on coronary heart disease (CHD), cardiovascular death (CVD), and total mortality.
Methods and Results During the 7-year active-intervention phase of the trial, 6428 of the men were given dietary recommendations to lower blood cholesterol, antihypertensive drugs to lower blood pressure, and counseling for cigarette smoking cessation. The remaining 6438 men were referred to their usual source of medical care. After 16 years, 370 SI and 417 UC men had died from CHD, which represents an 11.4% lower mortality rate for SI versus UC men (95% CI, −23% to 1.9%). Results for total mortality followed a similar pattern; 991 SI and 1050 UC men had died by the end of follow-up (relative difference, −5.7%; 95% CI, −13% to 2.8%). For acute myocardial infarction, a subcategory of CHD, the relative difference was −20.4% (95% CI, −34.4% to −3.4%). Differences between SI and UC men in mortality rates from acute myocardial infarction, CHD, and all causes were greater during the posttrial follow-up period than during the trial.
Conclusions Results of a 7-year multifactor intervention program aimed at lowering blood pressure and serum cholesterol and at cigarette smoking cessation among high-risk men give additional evidence of a long-term, continuing mortality benefit from the program.
Although numerous reports exist showing the short-term results of intervention on risk factors for CHD, few exist that outline the long-term effects of efforts to change multiple major risk factors. The Multiple Risk Factor Intervention Trial (MRFIT), designed to test the effectiveness of a multifactor risk-modification program in reducing the death rate from CHD, provides an opportunity to assess long-term efficacy. At the conclusion of the trial (average, 7 years after randomization), little difference between the intervention and the nonintervention groups was shown in mortality rates from either CHD or all causes.1 At that time, MRFIT researchers concluded that the trial had insufficient statistical power as a result of lower event rates in the nonintervention UC group than were predicted during the design phase and the less-than-expected differences between the UC and SI groups in major risk-factor reduction. Other possible reasons for the absence of a clear mortality benefit were a time lag for maximum benefit in mortality reduction and harmful effects of the diuretic-based stepped-care treatment for hypertension in men assigned to the SI group.
Although 7-year mortality rates for the two randomized groups differed little, rates for several nonfatal cardiovascular events (including angina pectoris, congestive heart failure, and peripheral arterial occlusive disease) were considerably lower for SI than UC men.2 Since nonfatal cardiovascular events are associated with an increased risk of subsequent CVD, a reasonable conjecture is that over the next several years death rates from CHD, CVD, and all causes would be lower among SI than UC men.
To ascertain whether any benefit would emerge after the trial ended in February 1982, long-term mortality follow-up was implemented. After 10.5 years of follow-up, the ratio of SI to UC mortality was lower in the 3.8 years after trial than during the trial, which supports the hypothesis of delayed benefit for SI men.3 4 Researchers also noted that a higher mortality rate was observed after the trial among men who had experienced nonfatal cardiovascular events during the trial than among those who did not experience such events. Since incidence of nonfatal events was lower among SI than UC men, this observation supported the inference that intervention-related reduction in incidence of nonfatal events led to subsequent reduction in fatal events, an occurrence that was not observed during the early years after randomization.
This work extends the follow-up period by 5 years (through December 1990) for a total of ≈16 years from randomization and 9 years from the end of the trial. It explores whether the trends observed after 3.8 years of posttrial follow-up persist, concordant with the proposition that intervention among middle-aged men on three major risk factors for CHD mortality confers a long-term benefit.
Design of the MRFIT
An extensive screening effort involving 361 662 middle-aged men 35 to 57 years of age at 22 clinical centers in 18 US cities led to the recruitment of 12 866 men who were without definite evidence of clinical CHD but who had an above average risk of death from CHD because of high blood pressure, elevated blood cholesterol level, and/or cigarette smoking.5 6 7 Half (6428) of these men were randomized to an SI program, and half (6438) were referred to their usual source of medical care (UC). SI men attended intervention visits, and men in both groups visited their local MRFIT center annually for assessment of risk factor and morbidity status. The trial ended February 28, 1982, 6 years after the last man recruited was randomized. The average length of time from randomization to the end of the trial was 7 years.
At intervention visits, men in the SI were instructed on how to alter their eating patterns to obtain substantial reductions in intake of saturated fats and cholesterol, moderate reductions in total fat, and modest increases in polyunsaturated fats. A weight reduction program was instituted for men who weighed >115% of their desirable weight. An educational program with the goal of smoking cessation was implemented for SI men who smoked cigarettes. Men in the SI who had mean diastolic blood pressure ≥90 mm Hg on two successive visits or who reported receiving antihypertensive medication from their personal physicians began stepped-care treatment for hypertension. For overweight hypertensive men, an initial effort was made to control blood pressure through dietary means. Details of the multifactor intervention have been described elsewhere.8 9 10 11
Since February 1982, vital status has been ascertained by matching identifying information reported by participants at the time of enrollment with the National Death Index.12 To obtain cause-specific mortality data, death certificates were collected and coded independently by two nosologists using the ICD-9.13 Disagreements between the two nosologists were adjudicated by a third nosologist. As in earlier reports,3 4 a CHD death is one coded to ICD-9 rubric Nos. 410-414 or 429.2.
A comprehensive search of the National Death Index database has been implemented twice since the trial ended. The first search was for all deaths through December 1985, and results have been reported.3 4 The second search was for all deaths through December 1990, and results are reported here. Mortality ascertainment through 1990 is considered to be essentially 100% complete.12
Comparisons of SI and UC participants for total and cause-specific mortality are based on time-to-event methods including proportional-hazards regression models.14 15 Percentage changes in risk of death are estimated as (RR−1)×100, where RR is the estimated risk of death in the SI group relative to the UC group. RR estimates and CIs were determined with use of a proportional-hazards model.
Mortality comparisons are presented for the trial period (baseline through February 1982), the posttrial period (March 1982 through December 1990), and the entire time period (baseline through December 1990).
CIs (95%) are given to guide the interpretation of results. No adjustment has been made for multiple comparisons made over time or for multiple end points.
A proportional-hazards regression model also was used to estimate the mortality risk associated with nonfatal cardiovascular events that occurred before March 1982. For this analysis, the first occurrence of a nonfatal event during the trial was treated as a time-dependent covariate. Such an analysis might be affected by a substantial amount of missing data arising from nonattendance at annual visits. As previously reported, the attendance rates (number of men alive at the time of the specified annual visit who attended the visit divided by the number of men randomized) were quite high, ≈95% at 12 months and >90% through the six complete cycles of annual follow-up visits for both groups.1
Risk Factor Change
Changes in the three major risk factors were ascertained during the trial and reported.1 For all men at baseline, plasma cholesterol concentration averaged 240 mg/dL (6.20 mmol/L) and diastolic blood pressure averaged 91 mm Hg, and 64% of the men reported that they smoked cigarettes. Six years after randomization, plasma cholesterol concentration averaged 4.6 mg/dL (0.13 mmol/L) lower and diastolic blood pressure averaged 3.1 mm Hg lower for SI than UC men. Antihypertensive drugs, which were prescribed for 19% of all participants at baseline, were prescribed for 57% of SI men and 46% of UC men. Also, cigarette smoking was reported by 32% of SI and 45% of UC men.
Information on levels of risk factors was obtained from participants at 4 of 22 clinical centers ≈3 years after the trial ended.16 For these participants, differences between groups persisted for measurements of total cholesterol and for diastolic blood pressure; however, the difference in percentages of men who smoked cigarettes was no longer statistically significant. (The reduced difference in smoking prevalence was attributed both to increased smoking cessation among UC men and to recidivism in SI smokers who had quit during the trial.) Virtually no difference existed between the study groups in the proportion of men using antihypertensive medication. Except for this limited investigation, no data were collected on risk factor changes after the termination of active intervention in 1982.
Cause-Specific Mortality at 16 Years
The median length of follow-up from randomization through December 1990 was ≈16 years (15.8 years; range, 14.8 to 17.1 years). During this time, 370 CHD deaths occurred among SI men (5.8%) compared with 417 among UC men (6.5%), a relative difference of −11.4% (95% CI, −23.0% to +1.9%) (Table 1).⇓ Cardiovascular mortality was 7.9% lower for SI than UC men (95% CI, −18.4% to +3.9%). Deaths from all causes numbered 991 among SI men (15.4%) and 1050 among UC men (16.3%), which indicates a 5.7% lower risk for SI than UC men (95% CI, −13.6% to +2.8%).
Detailed cause-specific mortality rates are given in Table 2.⇓ Much of the difference in CHD mortality rates between SI and UC men (47 deaths) after 16 years can be attributed to acute myocardial infarction (ICD-9 No. 410); mortality ascribed to this cause (185 SI and 232 UC deaths) was 20.4% lower for SI than UC men (95% CI, −34.4% to −3.4%). Mortality from noncardiovascular causes was 3.3% lower for SI than UC men; mortality from neoplastic disease, 1.8% lower for SI men.
Comparison of Trial and Posttrial Mortality
For CHD, CVD, and total mortality, the relative rate reductions for SI compared with UC men were greater during the posttrial period (March 1982 through December 1990) than during the trial. For CHD mortality, the SI rate was 13.5% lower than the UC rate during the posttrial period, whereas it was only 6% lower during the trial (Table 1). The relative differences between groups in acute MI death rates were similar for the two time periods: 19.6% lower after the trial versus 21.9% during the trial for SI than UC men. Thus, the greater relative difference seen after the trial between SI and UC men for all CHD mortality resulted from a greater posttrial compared with trial relative difference for CHD deaths other than those from acute MI (coded as ICD-9 Nos. 411-414 and 429.2). Noncardiovascular disease mortality followed a pattern similar to that of CHD—relative differences were more favorable for SI than UC men during the posttrial period than during the trial. As a result of favorable trends for both CVD and non-CVD causes of death, total mortality was lower for SI than UC men by 8.1% during the posttrial period, whereas it had been 1.5% higher for SI men during the trial.
Nonfatal Events and Subsequent Mortality
During the trial, 1266 SI (19.7%) and 1502 UC (23.3%) men experienced at least one nonfatal cardiovascular event (see footnote to Table 3⇓ for definition of nonfatal event). Using proportional-hazards models for time until nonfatal event, the RR of a nonfatal event during the trial was .83 for SI versus UC men (95% CI, 0.77 to 0.89).
Occurrence of a nonfatal cardiovascular event was strongly associated with subsequent cardiovascular mortality. In an analysis restricted to trial survivors, the posttrial CHD, CVD, and all-cause death rates were ≈2 to 3 times greater for those who experienced a cardiovascular event during the trial than for those who did not (Table 3).⇑ In a related analysis that used the occurrence of a nonfatal cardiovascular event as a time-dependent covariate in proportional-hazards models, RRs of death among all SI men who experienced a nonfatal cardiovascular event compared with those who did not were 1.57, 1.75, and 1.21 for CHD, CVD, and all causes, respectively. For UC men, these RRs were even higher: 2.70, 2.14, and 1.60, respectively. The excess mortality associated with nonfatal events was limited to deaths from cardiovascular causes.
The roles of high blood pressure, elevated blood cholesterol, and cigarette smoking as risk factors for CHD mortality have been reconfirmed by several recent epidemiological reports.17 18 19 20 21 22 23 The effect of modification of risk factors is more difficult to ascertain. Continuing mortality follow-up for participants in the MRFIT provides one of relatively few opportunities to evaluate long-term efficacy of cardiovascular risk factor modification on mortality.
As of December 1990, an average of 16 years after randomization into the MRFIT, the mortality rates for CHD, CVD, and all causes were, respectively, 11.4%, 7.9%, and 5.7% lower for SI men than for those randomized to the UC. These differences, based on a total of 2041 deaths, are substantially greater than the corresponding figures from the end of the trial in 1982. At that time, CHD mortality was 7.1% lower, CVD mortality was 4.1% lower, and all-causes mortality was 2.1% higher for SI than UC men. Deaths from acute myocardial infarction accounted for much of the difference between SI and UC men in CHD, CVD, and all-causes mortality at 16 years after randomization. There were 20.4% (P=.02) fewer deaths ascribed to acute myocardial infarction in the SI than the UC (185 SI and 232 UC deaths). The relative differences for acute myocardial infarction mortality between SI and UC were large and approximately equal for both the trial and posttrial time periods. Thus, the larger relative reduction over time for CHD mortality is at least in part due to a change in relative difference for CHD death from causes other than acute myocardial infarction; relative difference was higher for SI than UC men during the trial and slightly lower for SI than UC men after the trial.
An apparent posttrial benefit of intervention in the MRFIT was noted in the follow-up report at 10.5 years, and a possible mechanism for this benefit was suggested.3 This concept was explored further in analyses for this report and is still a tenable one. During the trial, UC men were at a significantly higher risk for a nonfatal cardiovascular event. Risk of subsequent mortality from cardiovascular causes was significantly higher among men who experienced a nonfatal event than among those who did not. In addition, UC men who experienced nonfatal events tended to have slightly higher mortality rates than SI men with similar experiences. These observations may also help to explain the higher posttrial CHD and total mortality rates among UC versus SI men and why a difference in mortality might require additional years of follow-up before it becomes detectable.
Results of a multiple risk factor modification trial on cardiovascular morbidity and mortality that showed efficacy both at the end of the trial and after extended follow-up have been reported for participants in the Oslo Study.24 In that 5-year trial (n=1232), the primary end point was first major CHD event (fatal or nonfatal), and the intervention consisted of dietary change and smoking cessation. After follow-up of almost 10 years, the intervention group had had only 25 events compared with 45 in the equal-sized control group; total deaths were 19 and 31, respectively. After 15 years of follow-up, CHD mortality was reported to be 48% lower in the intervention compared with the control group and all-cause mortality to be 26% lower.25
The Finnish multifactor trial had a different outcome. Fifteen years after randomization to the 5-year intervention trial, mortality among the 1222 high-risk Helsinki businessmen studied was significantly greater in the intervention group.26 27 Contributing to total mortality (67 deaths) in the intervention group were 13 deaths due to accidents or suicide versus 1 accidental death and 0 suicides among control group men (46 total deaths). Even so, 34 and 14 cardiac deaths occurred among intervention and control groups, respectively. By comparing the Finnish study with the MRFIT, one can identify major differences in eligibility criteria and intervention methods, including the use of lipid-lowering drugs, different hypertension treatment protocols, elevated blood glucose level as an entry criterion, and a specific drug for treatment of diabetes.
Two European multifactor trials that enrolled participants irrespective of risk factor status have reported long-term follow-up results. The Go¨teborg study (n=30 022) reported no difference between the intervention group and either of two control groups after 12 years with respect to CHD mortality.28 Although men at all risk levels were enrolled in the trial, the intervention was directed primarily at a high-risk subgroup of men. The investigators reported not being successful at reducing risk-factor levels for the entire intervention group to significantly less than the reductions that were taking place in the population at large. The initially favorable intervention and mortality results reported from the 6-year factory trial in Belgium (n=19 409), one of several European countries that participated in the WHO Collaborative Group, declined to no mortality differences between groups at 10 years of follow-up.29 The intervention efforts were the most intensive during the first 2 years of the trial and no major maintenance efforts were developed during the later years of the trial. WHO researchers concluded that risk reduction needs to be maintained to achieve a long-term preventive effect.
The MRFIT confirms the results of the Oslo Study,24 a much smaller trial with similar entry criteria but with an intervention protocol that did not target blood pressure. Results from the MRFIT are divergent from those of the Helsinki study, another small trial in which high-risk men were enrolled but that had broader entry criteria and an intervention protocol that included drug treatment for lowering cholesterol. The MRFIT shows more-favorable long-term results than do multifactor trials of equivalent or larger size that did not restrict entry to high-risk individuals and did not continue vigorous intervention for as long a period. Differences in entry criteria and intervention methodology may explain divergence between MRFIT results and those from the Helsinki study (note also that the MRFIT was 10 times larger than the Helsinki study, which implies greater precision of mortality rates for MRFIT) and also the more-favorable long-term results compared with the larger multifactor trials; however, no analyses have identified specific factors responsible for apparent discrepancies.27 30
Whether the continuing reduction in mortality seen for men in the MRFIT at 16 years can be attributed in part to posttrial maintenance of risk factor reduction is not answerable without more current data than now exists on risk factor levels among the participants. Nevertheless, examination of mortality data accumulated over the 16 years from randomization into the MRFIT to the end of 1990 strengthens earlier indications of continuing benefit, particularly from acute myocardial infarction, for those men randomized to intervention on cigarette smoking, high blood pressure, and elevated blood cholesterol. Given the conflicting reports of efficacy among the few multifactor intervention trials, further follow-up seems desirable to help to determine the stability and pattern of these results over time.
Selected Abbreviations and Acronyms
|CHD||=||coronary heart disease|
|ICD-9||=||International Classification of Diseases, Ninth Revision, Clinical Modification|
|MRFIT||=||Multiple Risk Factor Intervention Trial|
|SI||=||special intervention group|
|UC||=||usual care group|
|WHO||=||World Health Organization|
The Multiple Risk Factor Intervention Trial was conducted under contract with the National Heart, Lung, and Blood Institute, Bethesda, Md. The mortality follow-up and data analysis for this report were supported by NIH research grant No. R01-HL-43232.
- Received December 19, 1995.
- Revision received February 19, 1996.
- Accepted March 4, 1996.
- Copyright © 1996 by American Heart Association
Multiple Risk Factor Intervention Trial Research Group. Mortality after 10 1/2 years for hypertensive participants in the Multiple Risk Factor Intervention Trial. Circulation.. 1990;82:1616-1628.
International Classification of Diseases, North American Clinical Modification. Ann Arbor, Mich: Edwards; 1981;1.
Mantel N. Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemoth Rep.. 1966;50:163-170.
Cox DR. Regression models and life tables. J R Stat Soc B.. 1972;34:187-220.
Selmer R. Blood pressure and twenty-year mortality in the city of Bergen, Norway. Am J Epidemiol.. 1992;136:428-440.
Pekkanen J, Nissinen A, Punsar S, Karvonen MJ. Short- and long-term association of serum cholesterol with mortality: the 25-year follow-up of the Finnish cohorts of the seven countries study. Am J Epidemiol.. 1992;135:1251-1258.
Hjermann I. Non-pharmacological prevention of coronary heart disease. Presented at the 61st Meeting of the European Atherosclerosis Society; May 18, 1993; Capri, Italy.
Strandberg TE, Salomaa VV, Naukkarinen VA, Vanhanen HT, Sarna SJ, Miettinen TA. Cardiovascular morbidity and multifactorial primary prevention: fifteen-year follow-up of the Helsinki Businessmen Study. Nutr Metab Cardiovasc Dis.. 1995;5:7-15.
Wilhelmsen L, Berglund G, Elmfeldt D, Tibblin G, Wedel H, Pennert K, Vedin A, Wilhelmsson C, Werko¨ L. The multifactor primary prevention trial in Go¨teborg, Sweden. Eur Heart J.. 1986;7:279-288.
De Backer G, Kornitzer M, Dramaix M, Kittel F, Thilly C, Graffar M, Vuylsteek K. The Belgian Heart Disease Prevention Project: 10-year mortality follow-up. Eur Heart J.. 1988;9:238-242.
Stamler J. Lessons from the Helsinki Multifactorial Primary Prevention Trial. Nutr Metab Cardiovasc Dis.. 1995;5:1-5. Editorial.