Why We Need to Know the Limitations of Evidence-Based Medicine

August 1, 2006

Primary care clinicians are encouraged to use the latest evidence-based diagnostic and therapeutic strategies from prospective randomized, controlled trials in their practice. But are the participants in these trials typical patients?

Primary care clinicians are constantly exhorted to apply the latest evidence-based diagnostic and therapeutic strategies. The jewel in the evidence-based crown is the prospective randomized, controlled trial (RCT). But are the participants in these trials typical of patients seen in primary care practices?

Most patients in primary care practices have multiple comorbidities. In fact, in one representative population of nearly 1000 patients, 90% had more than one chronic condition; among those older than 65 years, almost all had 2 or more disorders.1 The most common chronic diseases were hyperlipidemia, heart disease, rheumatologic problems, and kidney and lung diseases.

Can the findings of RCTs be extrapolated to patients who have multiple diseases? And do the therapeutic recommendations take comorbid conditions into account?


The investigators who studied the prevalence of comorbidities in the primary care population set out to answer these questions.2 They randomly chose RCTs that targeted hypertension treatment, including the PREMIER study,3 the HOT study,4 the ALLHAT study,5 and the DASH trial.6 In the populations recruited for the individual studies, a percentage of patients-which was similar to the percentage in the primary care database-had multiple chronic conditions. In fact, depending on the study, anywhere from about 6 to 12 comorbid conditions were present in the study patients.

However, none of the trials provided data on how many patients with comorbidities were excluded from the final analysis. Some of the studies excluded patients with heart disease, cancer, "any life-threatening illness," dementia, hyperlipidemia, insulin-dependent diabetes, and renal disease. In the computations leading to treatment recommendations, it was not clear whether patients with multiple comorbidities were included in the analysis-and if they were, whether there was a difference in their treatment or response.


We know that common comorbidities and other conditions (such as depression) can affect the treatment regimen and the response to therapy. This means that future RCTs have to include and explicitly report the treatment results for "typical" patients-that is, those with multiple comorbidities. More patients will need to be recruited for each study to maintain statistical power-and expenses for each trial will be higher. But if the data are to be relevant to primary care, I am not sure that we have a choice.

I am interested in your thoughts about the relevance of RCT results to your practice. Please e-mail your comments to me at consultantedit@cmpmedica.com.