To the Editor: In their randomized controlled trial of computerized decision support in an emergency department (ED) setting, Dr. Terrell and colleagues concluded that “computerized physician order entry with decision support significantly reduced prescribing of potentially inappropriate medications for seniors,”1 but a question regarding the study methodology exists. The trial aimed to compare the efficacy of computer-assisted decision support (experimental maneuver) with that of usual care (control maneuver) on the proportion of ED visits that resulted in prescriptions for nine targeted medications (outcome). The critical concern is whether sufficiently equal performance of the experimental and control maneuvers was achieved in the two groups so that the differences in outcome may be attributed to the effect of the experimental maneuver. Human (physician) factors cannot be overlooked as a co-intervention of computerized decision support.2 As described in the text, adverse drug events are a frequent cause of delirium in older adults and lead to more than 10% of ED visits.3,4 Falls, another geriatric syndrome, also account for 10% of visits to the ED and are potentially caused by medication use.5,6 For example, emergency physicians' prescription pattern for sedative agents might be intuitively different for frail patients presenting to the ED with delirium or falls and otherwise healthy patients presenting with chest pain. In the former case, a more-explicit causal relationship between the medication and the reason for ED visit might prompt emergency physicians to adjust medications regardless of the use of computerized decision support. In the trial, physicians, but not the ED visits, were used as the unit of randomization. Because geriatric patients are markedly heterogenous, the failure to adjust for underlying reasons for the ED visits and patients' health status (e.g., comorbidity, disability, and cognitive status) may cause performance bias and lead to conclusions that depart systematically from reality. Further consideration of these potential confounders would provide additional credibility to their efficacy data.
Conflict of Interest: The editor in chief has reviewed the conflict of interest checklist provided by the author and has determined that the author has no financial or any other kind of personal conflicts with this paper.
Author Contribution: Kouta Ito: preparation of letter.
Sponsor's Role: None.