Cebul et al. (Sept. 1 issue) report positive associations between the use of electronic health records (EHRs) and the quality of diabetes care, findings remarkably different from those of rigorous systematic reviews. Most researchers, moreover, would exclude from synthesis designs similar to that used by Cebul et al. — designs that fail to account for preexisting secular trends and selection bias. The investigators used an uncontrolled, cross-sectional, post-test–only design, comparing patients at 21 sites with and without EHRs. Without several years of baseline observations, however, pre-EHR trajectories of outcomes and the role of the interventions themselves are indistinguishable.
Koppel R, Majumdar SR, Soumerai SB. Electronic Health Records and Quality of Diabetes Care. N Engl J Med. 2011 Nov 15;365(24):2338–9.