Clinical trials are the keystone of evidence-based medicine and are necessary to show the safety and efficacy of medications. This presentation will discuss the limitations of observational studies when used for the assessment of drug efficacy and safety using the CVD Real study as a case study.
The CVD Real study is a retrospective observational cohort study that examined cardiovascular outcomes in propensity matched groups of patients started on a Sodium GLucose co-Transporter- 2 inhibitor (SGLT2i) or another glucose lowering agent (OLGD). Over a period of approximately 9 months, the patients receiving the SGLT2i had an all cause mortality that was 50% of the patients initiated on another OGLD.
In the US cohort of CVD Real, 75.9% of patients received canagliflozin with a 62% reduction of mortality, yet in the CANVAS RCT, canagliflozin had no significant impact on CV mortality. Likewise in the CVD Nordic most patients received dapagliflozin with a 47% reduction of mortality, but in the RCT DECLARE study with dapagliflozin, mortality was not reduced. However heart failure hospitalisation in CVD Real was reduced by 30-40% similar to that observed in the RCTs. Why the differences of mortality rates between CVD Real and the RCTs were observed, but not of heart failure will be discussed. However it seems likely that the large reduction of death rates observed in the CVD Real study was exaggerated by confounding.
Observational studies provide supportive information, especially in a wider population, when RCT outcomes are known, but should not be used to make primary guideline recommendations about a drug or a class of agents. However they are of value in recognizing potential and infrequent adverse effects of a drug, which may not be detected in the limited population of an RCT.