The recent article by Beers and coworkers [1] and the accompanying editorial by Jones [2] illustrate a disturbing trend in data reporting and, more subtly, in patient care. Adopting predetermined criteria for inappropriate medication prescribing and then using a computer print-out of a physician order sheet certainly simplify a study. But how do investigators know whether any of the numerous drugs were beneficial or harmful? Studying computer print-outs rather than patients raises serious concerns.

Furthermore, neither the investigator nor the editorialist commented that, on average, each nursing home resident had 7.2 medications ordered! Because these patients are penned in, largely helpless, and often incapable of giving informed consent, they might even have been taking all of them—a feat few rational ambulatory patients could manage. To what end are these drugs being administered? These patients, if similar to nursing-home patients elsewhere, were mainly demented and bed-ridden. Most could be managed with few, if any, drugs; moreover, no study has ever shown the need for so much medicine. Just manipulating seven different drugs per patient must place a considerable burden on nurses, pharmacists, data entry clerks, and patients, not to mention the largely unnecessary financial burden on the ultimate payer, the public.

Rather than calling for expanding data entry at a time of increasing health care costs, the editorialist might have better expressed outrage that such expensive technology could substitute for the necessary bedside care. Nursing homes already spend too much time recording data; they do not need more opportunities to do so. Part of the fault stems from imposed government regulations, but part belongs to us. We order the drugs and we write the articles. Annals might well more critically appraise articles based on computer data.