Quality improvement opportunities in prescriber alert programs.

Battles may be won, but the war against threats to patient safety continues, and time will tell if escalation of the war will reduce casualties. Brent James, MD, the well-known advocate of process improvement, several years ago described himself as a “terrorist” for health care safety. Despite Dr. James’ commitment to escalate the war in an attempt to reduce medical errors, particularly errors that harm, it will be difficult to eliminate adverse drug events (ADEs) entirely because of the myriad of factors that pose risks to patient safety including drug-drug, drug-disease, and drug-patient nuances. But, the certainty of ADEs does not diminish the importance of the mission to reduce patient harm from avoidable ADEs. This mission includes strategic actions, such as proactive avoidance of unnecessary and potentially harmful drug therapy, downward dose adjustment, and even discontinuation of drug therapy as part of high-quality medication therapy management. However, the size of the target is vague because the actual rate of patient harm associated with potentially avoidable ADEs is both controversial and difficult to estimate accurately. Some have argued that even the most frequently cited figures, such as the Institute of Medicine’s estimate that between 44,000 and 98,000 inpatients in the United States die annually as a result of preventable medical errors including ADEs, are either exaggerated by failure to account for patients’ baseline health status or “not well substantiated” because of their reliance on the “highly subjective” opinions of study investigators about which events are avoidable. Additionally, these estimates are widely publicized, represent a large portion of the research on the effects of ADEs, and are commonly cited as evidence of the need for automated prescribing techniques to reduce ADEs throughout the health care system; yet, they do little to inform managed care decision makers about risks to patients in ambulatory care, who represent the vast majority of beneficiaries with public or private insurance. Gurwitz et al. (2003) used a rigorous and labor-intensive methodology, including thorough reviews of patient charts and incident reports, systematic evaluation of all potential ADEs by multiple raters, and assessments of inter-rater reliability, to estimate the incidence, severity, and preventability of ADEs in a cohort of approximately 30,000 predominantly Medicare + Choice (managed care) enrollees who received care in a large multispecialty group practice. During 30,397 person-years of observation over the 12-month period from July 1999 through June 2000, the investigators identified 1,523 events (rate of 50.1 per 1,000 person-years), more than 70% of which resulted in symptoms that persisted for more than 1 day. Of the 1,523 ADEs, 431 (28.3%) were deemed to be “serious” (e.g., fall with fracture, hemorrhage requiring transfusion or hospitalization, delirium, urticaria); 136 (8.9%) were “life-threatening” (e.g., hemorrhage with hypotension, hypoglycemic encephalopathy, acute renal failure); and 11 (0.7%) were fatal, including bleeding, drug toxicity events, anaphylaxis, peptic ulcer, neutropenia, hypoglycemia, and antibiotic-associated diarrhea. Less serious but “significant” events (e.g., nonurticarial skin rash, hemorrhage not requiring transfusion or hospitalization, fall without fracture, and oversedation) constituted 945 (62.0%) of all ADEs. Of 421 ADEs judged as preventable (27.6% of ADEs), 167 (39.7%) were serious, 72 (17.1%) were life-threatening, and 5 (1.2%) were fatal. Thus, of approximately 30,000 elderly enrollees, a total of 244 (0.8%) experienced ADEs that were both preventable and severe during the 12-month follow-up. Still, the overall rate of preventable ADEs identified by Gurwitz et al., 13.8 per 1,000 person-years (approximately 1.4% of enrollees), suggests a problem that should be addressed by managed care. Studies in nonelderly populations support this assessment, albeit not always with a methodology as rigorous of that of Gurwitz et al., usually concluding that there is a small but troublesome rate of potentially preventable ADEs in ambulatory care. Using a pre-implementation versus postimplementation design and multivariate analyses, Devine et al. (2010) studied prescriptions written in a large multispecialty clinic, finding that prior to the implementation of computerized provider order entry (CPOE), 911 of 5,016 (18.2%) prescriptions contained errors, although only 8 of these errors (0.9%, or 0.16% over all prescriptions) caused harm to the patient. After implementation of CPOE, the error rate declined to 423 of 5,153 (8.2%), with 5 of these (1.2%, or 0.10% over all prescriptions) causing harm.