- © 2004 Canadian Medical Association or its licensors
Studying a population of adults seen in clinic for biliary colic, Boris Sobolev and associates1 documented an association between longer waiting times and admissions for emergency cholecystectomy.
However, because the patients were not randomly assigned to the waiting list, readers should entertain the possibility that the findings were driven by an association with a so-called “third variable.” For example, the patients who ended up on the waiting list might have been sicker. If so, the observed association between waiting times and emergency admissions was actually driven by an unobserved association between health status and emergency admissions. Sobolev and associates1 acknowledge the possibility of confounding by patient morbidity, and they do attempt rudimentary adjustment for other potential confounding variables. However, even if they had had access to better data on patients' health status, the criticism of potential confounding would remain.
Prior studies, none of which were cited by Sobolev and associates,1 have addressed this problem by means of econometric methodology.2,3,4 Hamilton and colleagues2 used an estimation strategy that accounted for unmeasured health differences and found no effect of waiting times on death rates for patients waiting for hip fracture surgery. Subsequent comparisons of patients with hip fracture in the United States and Canada3,4 arrived at a similar conclusion.
Policy-makers seeking to draw conclusions from the findings of Sobolev and associates1 would be well advised to consider these more sophisticated econometric analyses in their deliberations.
Alexander C. Tsai Case Western Reserve University School of Medicine Cleveland Heights, Ohio