Primary prevention of heart disease and stroke
[response]
Online posting: December 17, 1997
Published in print: January 13, 1998 (CMAJ 1998;158:25)
Re: Primary prevention of heart disease and stroke: a
simplified approach to estimating risk of events and making drug
treatment decisions, James P. McCormack and colleagues, Can
Med Assoc J 1997;157(4):422-8 [full text / résumé]
In response to: H.R. Strasberg; F.A. McAlister
We agree with Dr. Strasberg that computers have the potential to
be useful clinical tools. Although many physicians have computers
in their offices for billing and scheduling purposes, not many
computers are available at the point of care. Our method of
estimating a patient's risk is simple enough that a computer is
not required. Furthermore, a computer program that provides
"black box" estimates of risk and benefit may not allow an
appreciation of the basis for the estimates or an understanding
of the concepts of relative and absolute risk reduction.
That said, we have become aware of a web site (www.hbroussais.fr/scientific/fram.html) where the clinician can
enter a patient's risk factors and obtain an estimate of the
absolute risk for heart disease and stroke based on the
Framingham data. The site offers no discussion of the limitations
and assumptions of this process, and the clinician and patient
are therefore unable to appreciate the impact and interplay of
the different risk factors or the benefits of treatment regimens.
Dr. McAlister points to longitudinal data suggesting that the
benefit of hypertension treatment (reduction of death from
cardiovascular disease) extends beyond the 3- to 6-year period
typical of randomized controlled trials.[1] However, these data
are derived from uncontrolled, nonrandomized observational
studies and address only the issue of death, not morbidity from
cardiovascular disease. The magnitude of that long-term benefit
remains uncertain and may never be evaluated in a randomized
controlled trial.
McAlister also notes a number of reasons why the design of most
3- to 6-year randomized controlled trials has led to a systematic
underestimation of the benefit.
The recruitment of low-risk patients is not an issue because
these individuals represent a large proportion of patients who
receive therapy for hypertension and elevated cholesterol. This
factor is taken into account by our nomograms, which help the
clinician to estimate the baseline risk.
The argument that many trials lack the statistical power to show
a difference is not an issue because the trials and meta-analyses
we included all had statistically significant results.
Loss of patients to follow-up would be an issue only if the loss
were greater in the placebo group than in the treatment group. We
are unaware of any studies affected by this phenomenon.
The suggestion that the benefit was achieved despite only modest
reductions (5 mm Hg diastolic and 10 mm Hg systolic) in blood
pressure may not be entirely correct. Although the difference
between the drug and placebo groups was 10/5 mm Hg, the actual
reduction in the drug group (from baseline) was greater,
approximately 20/10 mm Hg; this magnitude of change may be
important, particularly if one believes that the reduction in the
placebo group was due primarily to nonspecific effects rather
than the placebo effect.[2]
The overall effect of switching some patients in the placebo
group to active treatment at some point during a trial is
difficult to estimate, because it would depend on when and to
what degree it occurred. We estimate that if 20% of the placebo
group received active treatment for half of the trial, there
would be a 3% to 4% decrease in the overall reduction of relative
risk. This decrease would have only a small impact on the
estimates of absolute reduction and number needed to treat.
Although long-term information would be useful, a patient's
decision to adhere to treatment for 5 years is, in our belief, an
important commitment. We therefore continue to feel that
treatment decisions based on a 5-year risk estimate and the
relative risk reduction from randomized controlled clinical
trials is the most reasonable approach when deciding on drug
therapy.
James P. McCormack, PharmD
Associate Professor
Clinical Division Chair
Faculty of Pharmaceutical Sciences
University of British Columbia
Pharmacy Department
St. Paul's Hospital
Vancouver, BC
jmccorma@unixg.ubc.ca
Marc Levine, PhD
Associate Professor
Faculty of Pharmaceutical Sciences
University of British Columbia
Vancouver, BC
Robert E. Rangno, Msc, MD
Associate Professor
Departments of Medicine and Pharmacology
University of British Columbia
Clinical Pharmacology
St. Paul's Hospital
Vancouver, BC
References
- Sytkowski PA, D'Agostino RB, Belanger AJ, Kannel WB.
Secular trends in long-term sustained hypertension, long-term
treatment, and cardiovascular mortality. The Framingham Heart
Study 1950 to 1990. Circulation 1996;93:697-703.
- Ernst E, Resch KL. Concept of true and perceived
placebo effects. BMJ 1995;311:551-3.
CMAJ email
CMAJ Readers' Forum