Critiques Multiple Risk Factor Intervention Trial (MRFIT) "designed to test the hypothesis that reductions in cigarette smoking, high blood pressure and elevated serum cholesterol reduce the risk of dying from coronary heart disease". Examines study design and quotes other researchers regarding study shortcomings. Observes "results do not support the hypothesis that...cigarette smoking [is] causally related to coronary heart disease mortality". Includes references.
- Named Organization
- British Medical Journal
- Framingham Heart Study
- Journal of Epidemiology and Community Health
- Journal of the American Medical Assocation
- Multiple Risk Factor Intervention Trial
- Blood pressure
- Cigarette smoking
- Coronary heart disease
- Risk factors
- Risk functions
- Serum cholesterol levels
- Smoking histories
- Special Intervention Group
- Usual Care Group
- Named Person
- Hjermann, I.
- Kolata, G.
- Lundberg, G.
- Oliver, M.
- Rose, G.
- United Kingdom
- Report- Scientific
- Human subjects
- Research studies
Page 1: 00000067
Re: Multiple Risk Factor Intervention Trial
The results of the ten-year, $115 million Multiple Risk
Factor Intervention Trial, known as MRFIT, were reported in the
September 24, 1982 issue of the Journal of the American Medical
Association.1 MRFIT was designed to test the hypothesis that
reductions in cigarette smoking, high blood pressure and elevated
serum cholesterol reduce the risk of dying from coronary heart
disease (CHD). However, the trial failed to confirm this hypothesis.
In the words of an editorial accompanying the report:
"Unfortunately, the fundamental question facing the investigators
at the beginning of the experiment remains unanswered."2 As
discussed in Science: "Briefly, the results are inconclusive."3
An editorial in the British Medical Journal described the project
as ending "with inconclusive results -- and leaving many questions
unanswered.''4 Similarly, another major British medical publica-
tion, The Lancet, editorialized that the MRFIT results "prove
nothing."5 In short, MRFIT failed to show that reducing levels
of smoking, cholesterol and blood pressure, reduces risk of death
due to heart disease.
The study involved over 12,000 men who were believed to
be athigh risk for heart disease on the basis of their smoking
histories, and blood pressure and serum cholesterol levels.
Page 2: 00000068
Measurements of these variables were made on three screening
visits, after which the men were randomly assigned to one of two
groups. Half of the men were assigned to a "special intervention"
(SI) group and received intensive counseling to change smoking
and dietary habits and were treated for high blood pressure. The
other men were assigned to a "usual care" (UC) group, were told
of their risk status and received annual medical assessments at
MRFIT clinics. However, the investigators made no efforts to
change their risk factor levels, and instead referred the UC men
to their usual sources of health care.
During several years of follow-up, men in the SI group
showed a substantial reduction in their risk factor levels com-
pared to the UC group. The largest changes occurred in cigarette
For men who reported smoking at the first screening
visit, the reported quit rates in the SI group were over 70%
greater than in the UC group. However, the difference in the
reported smoking quit rates was even larger when based on smoking
"at baseline," defined by the investigators as the average reported
smoking rates at the second and third screening visits. In this
case, the SI group showed quit rates over 100% greater than the
UC group. In fact, the investigators stated that the SI-UC
difference in reported smoking "exceeded design goals by 122%."
Compared to the UC group, men in tie SI
showed reductions in blood pressure and serum cholesterol.
Page 3: 00000069
However, these reductions were more modest, and generally did not
meet the expectations of the investigators. Despite this, MRFIT
investigators noted that the .overall differences between the two
groups were substantial. In fact, they computed that they had
achieved "83% of the SI-UC risk factor difference initially
assumed in the design."
During the seven year follow-up period, men in the SI
group sh~wed a CHD mortality rate of 17.9 deaths per 1,000 com-
pared to 19 3 per 1 000 in t~e UC group, a small and "statis-
tically nonsignificant difference." These results do not support
the hypothesis that the risk factors studied in MRFIT, especially
cigarette smoking, are causally related to coronary heart disease
Thus, even though the study was successful in reducing
risk factor levels, the investigators were forced to conclude
that "the overall results do not show a beneficial effect on CHD
or total mortality from this multifactor intervention."
Despite reporting this lack of effect, MRFIT researchers
did not discount the value of risk factor intervention in reducing
CHD mortality. They justified their continued belief in the
benefits of risk factor intervention by noting that the possi-
bility that intervention was ineffective "seems inconsistent with
most published scientific data." However, risk factor intervention
Page 4: 00000070
trials from both Norway6 and England7 have failed to demonstrate
statistically significant effects on CHD mortality attributable
to smoking reduction. Moreover, if previous data were sufficient
to establish whether there are benefits of risk factor reduction,
then MRFIT would never have been necessary. Even the authors
emphasized that "the trial was of course initiated to test this
~So faced with findings that failed to support the
~ ~ intervention hypothesis, the investigators offered various
~~ ~ ~ explanations. They pointed to limitations of the statistical
~.~ ~ tests used in the study. In addition, they noted that the major
~ ~ reductions in risk factor levels and mortality rates in the UC
~. ~ ~ group were une.xpected. Finally, they suggested that there may
have been certain subgroups of men in the SI group that were
affected unfavorably by the therapeutic treatment for hypertension
~. ~ ~ used in the intervention.
In regard to the UC group, the MRFIT researchers reported
~ ~ more than expected reductions in risk factor levels during the
study and mortality rates lower than expected. However, it is
questionable whether these developments led to the failure to
demonstrate an overall beneficial effect of intervention.
To illustrate, so long as a
substantial difference in
risk factor levels between the UC and SI groups was obtained,
Page 5: 00000071
reductions in risk factor levels in the UC group should not
matter, unless it was assumed that such decreases led to decreased
mortality for the UC group. Even the authors were hesitant to
entertain, seriously this suggestion, since it assumed what MRFIT
was designed to prove.
On the other hand, they did imply that the lower than
expected mortality in the UC group made possible effects of risk
factor reductions in the SI group more difficult to observe. It
should be noted in this context that expected death rates were
calculated using "risk functions" which were based on data from.
the well-~own Framingham Heart Study and which attribute major
roles to smoking, cholesterol and hypertension in CHD. That the
predicted death rates in the UC group based on these functions
~ ~were~ inaccurate seems to call into question the validity of
Framingham data. In other words, the failure of MRFIT to demon-
strata an effect of intervention may have been because initial
assumptions about the relation of risk factors to coronary heart
disease were wrong.
As noted earlier, MRFIT investigators also focused on
possible "unfavorable" effects of intervention on certain sub-
groups of men in the SI group. It was suggested that these
effects may have offset an overall beneficial effect of interven-
tion on mortality.
5 TIMN 0019380
Page 6: 00000072
However, due to the complex subject selection pro-
cedures, analyses of data from subgroups are not scientifically
valid. This was recognized by the authors who noted that "it
must be remembered that subgroups defined by the presence or
absence of one of the three major risk factors are not otherwise
comparable." Editorial comments regarding MRFIT in other scientific
and medical journals have also noted that such subgroup analyses
are "statistically and are "worrying features''4 of
trial report. Nevertheless, the MRFIT investigators made extensive
~comparisons using such data. They focused on one particular
~subgroup, namely hypertensive men who had electrocardiographic
~ ;~abnormalities. Such men in the SI group were reported to be at
increased risk of death from CHD compared to those in the UC
group. However, as noted in an editorial in the British Medical
"~Journal, "it seems unlikely that the small numbers with such an
adverse response, less than a tenth of the total study, could
~.;~ have diluted any important positive overall effect.
Putting aside questions of validity, the various
reasons offered by the MRFIT investigators to explain away their
findings are peripheral to the central purpose and results of
MRFIT. Thus, they serve merely to divert attention from the
study's overall failure to observe a favorable effect of risk
factor reduction on mortality.
In general, therefore, the MRFIT investigators have
Page 7: 00000073
apparently refused to believe their own findings. They justify
this disbelief through the use of assumptions which the trial
was designed to test and by using admittedly invalid reanalyses
of their data. Nevertheless, an objective view of the study
shows that it was highly successful in meeting its projected
intervention goals, in that there was a substantial difference in
risk factor levels between the SI and UC groups. Yet, it failed
to demonstrate a favorable effect of such intervention on CHD
Shook, Hardy & Bacon
Page 8: 00000074
i. Multiple Risk Factor Intervention Trial Research Group,
"Multiple Risk Factor Intervention Trial: Risk Factor
Changes and Mortality Results," JAMA 248(12): 1465-1477,
September 24, 1982.
2. Lundberg, G., "MRFIT and the Goals of The Journal," JAMA
248(12): 1501, September 24, 1982.
3. Kolata, G., "Heart Study Produces a Surprise Result," Science
218: 31-32, October i, 1982.
4. Oliver, M., "Does Control of Risk Factors Prevent Coronary
Heart Disease?," Br Med J II: 1065-1066, October 16, 1982.
5. Editorial, "Trials of Coronary Heart Disease Prevention,"
Lancet II: 803-804, October 9, 1982.
6. Bjermann, I., et al., "Effect of Diet and Smoking Intervention
on the Incidence of Coronary Heart Disease: Report from the
Oslo Study Group of a Randomised Trial in Healthy Men,"
Lancet II: 1303-1310, December 12, 1981.
7. Rose, G., et al., "A Randomised Controlled Trial of Anti-
Smoking Advice: 10-Year Results," J Epidemiol Community
Health 36: 102-108, 1982.
8 TIMN 0019383