CMEonHIV.com is dedicated to providing online CME presentations (slides with voiceover) on HIV/AIDS for healthcare professionals given by local and international experts to keep you up-to-date on the ongoing developments in the field.
Conference
"Strengths and pitfalls of using observational studies to determine therapeutic benefits/harm" Dr. Marina Klein (biography) English - 2008-04-25 - 25 minutes
(23 slides)
Summary : The reason that randomized control trials are the gold standard in study design for assessing efficacy is that they present the advantages of eliminating conscious bias in physician selection and patient self-selection, and of balancing unknown confoundings and biases that affect the outcome. However, Dr. Klein points out the disadvantages of randomized clinical trials. Having to follow such large numbers of patients over a significant period of time is an obstacle, not only to the ease of conduct of the study, but to the feasibility of the trial and the generalization of the results.
An interesting alternative that permits the study of rare events and toxicities is the use of observational studies with large cohorts and cohort collaborations, such as the North American AIDS Cohort Collaboration on Research and Design (NA-ACCORD). This allows the accumulation of very large numbers of patients without incurring significant additional costs.
Dr. Klein discusses the problems associated with this method of investigation, and the strategies that can be used to managed them.
One of the major issues is called confounding, which is the result of naturally occurring associations between different risk factors. This confuses the relationship we are interested in with the effect of something else. When this concept is applied to the prescription of different medications amongst patients with the same medical condition, it is referred to as confounding by indication. Similarly, drug channeling is the preferential prescription of drug therapies with similar indications to patients according to their baseline prognoses.
Another problem that must be monitored is the determination of causality. A drug does not necessarily cause the outcome to which it is associated by observation. Thus, Dr. Klein outlines a set of criteria against which to measure the causal association.
USER ACKNOWLEDGES AND AGREES THAT ALL DECISIONS MADE WITH THE ASSISTANCE OR USE OF THE SOFTWARE AND/OR THE WEBSITE AND/OR BASED ON CONTENT FOUND HEREIN WILL BE EXCLUSIVELY THE RESPONSIBILITY OF THE USER.
