Effect of computer-based alerts on the treatment and outcomes of hospitalized patients

BACKGROUND: Hospital computing systems play an important part in the communication of clinical information to physicians. We sought to determine whether computer-based alerts for hospitalized patients can affect physicians' behavior and improve patients' outcomes. METHODS: We performed a prospective time-series study to determine whether computerized alerts to physicians about rising creatinine levels in hospitalized patients receiving nephrotoxic or renally excreted medications led to more rapid adjustment or discontinuation of those medications, and to determine whether such alerts protected renal function. RESULTS: Laboratory data were observed for 20,228 hospitalizations, with documentation of 1573 events (instances of rising creatinine levels during treatment with a nephrotoxic or renally excreted drug). During the intervention period, doses were adjusted or medications discontinued an average of 21.6 hours sooner after such an event (P < .0001). For patients receiving nephrotoxic medications during the intervention period, the relative risk of serious renal impairment was 0.45 (95% confidence interval, 0.22 to 0.94) as compared with the control period, and the mean serum creatinine level was 14.1 mumol/L (0.16 mg/dL) lower on day 3 (P < .01) and 25.6 mumol/L (0.29 mg/dL) lower on day 7 (P < .05) after an event. Forty-four percent of physicians who responded to a questionnaire said that the alerts had been helpful in the care of their patients, whereas 28% found them annoying. Sixty-five percent wished to continue receiving alerts. CONCLUSIONS: Computer-based alerts regarding patients with rising creatinine levels affect physician behavior, prevent serious renal impairment, preserve renal function, and are accepted by clinicians.