Effect of Clinical Decision Support on Psychotropic Medication Prescribing in the Long‐term Care Setting

To the Editor: Psychotropic medications are commonly associated with preventable adverse drug events (ADEs) in long-term care (LTC) settings.1, 2 Errors associated with preventable ADEs occur most often at the ordering stage of pharmacotherapy. It has been reported that a “geriatric decision support system” for psychotropic medications can increase the prescription of recommended doses and reduce the prescription of non-recommended drugs in the hospital setting.3 It was hypothesized that implementation of a computerized clinical decision support system (CDSS) providing specific recommendations relating to dosing and choice of psychotropic drugs would have a similar effect in the LTC setting. The institutional review boards at the University of Massachusetts Medical School and the participating facility approved this study. The study was a randomized trial conducted at a large, academically affiliated LTC facility equipped with an integrated electronic medical record and computerized prescriber order entry.4, 5 Randomization was within blocks according to resident unit type, which included Alzheimer's disease, complex medical conditions, behavioral and mental health problems, physical support, mental health support, stroke, and cognition. A team of physicians, pharmacists, and informatics professionals developed the CDSS for 22 psychotropic medications from the following drug classes: typical and atypical antipsychotics, benzodiazepines, and hypnotics and sedatives. There were two broad alert categories: dosing and avoid. Dosing alerts were triggered if the medication dose ordered exceeded the maximum recommended total daily dose, multiple orders existed for the same medication, or total daily dose could not be calculated. Avoid alerts were triggered to discourage the use of typical antipsychotics in managing behavioral symptoms associated with dementia and long-half-life benzodiazepines. Forty alerts were developed and displayed to prescribers in the intervention units and hidden but tracked in control units of the facility. Alerts were not generated during renewals and were informational only. The percentages of psychotropic medication orders that were modified in response to an alert on intervention units and control units were calculated and compared by calculating the relative risk (RR) and 95% confidence interval (CI). The overall rates of inappropriate orders were calculated based on total resident days and compared using a rate ratio (RR) and 95% CI. The percentages of unnecessary alerts that were displayed on the intervention units were also calculated. There were 813 different residents on the study units over the 12-month observation period, with 3,665 and 3,666 resident-months of observation time on the intervention and control units, respectively. There were no differences in average age (85.7±8.2 vs 85.8±8.0; P=.88) or percentage of women (72.3% vs 70.2%; P=.52) between units. Physicians prescribing medications on the intervention units triggered 176 alerts, for a rate of 4.80 per 100 resident-months, compared with 139 hidden alerts, for a rate of 3.79 per 100 resident-months, in the control group (RR=1.27, 95% CI=1.01–1.59). The percentage of psychotropic medication orders that were modified in response to an alert on intervention units was 8%; on the control units, 2% of orders were modified after being initiated under circumstances in which an alert would have been displayed (RR=3.69, 95% CI=1.08–12.57) (Table 1). The overall rate of inappropriate orders was 1.58 per 100 resident-months on the intervention units and 1.61 per 100 resident-months on the control units (RR=0.98, 95% CI=0.67–1.44). In a post hoc analysis, the total number of unnecessary alerts was 113 out of 176 on the intervention units (64%). Various nontechnological strategies have been employed to improve the quality of psychotropic medication prescribing in the LTC setting, such as direct education of prescribers and staff employing “academic detailing” techniques.6, 7 Recently, there have been promising efforts to reduce inappropriate use or excessive dosing of psychotropic medications by employing computerized clinical decision support, but these interventions have been limited primarily to the hospital setting.3 The current study found that the CDSS provided to physicians ordering psychotropic medications in the LTC setting influenced prescribing decisions, although no overall improvement in prescribing quality was demonstrated. This may have been attributable to a low overall rate of inappropriate prescribing in the study facility. In addition, the CDSS did not provide alternative orders within alerts that prescribers could directly accept.8, 9 Furthermore, more than half of the alerts displayed were unnecessary. Last, the evidence relating to some of the medications changed over the course of the study.10 Formal, rigorous evaluations of computerized CDSSs implemented in the LTC setting are essential so that they can be improved upon and promoted with increased confidence. The authors thank Peter Anderson, Yen Mee Low Sin, and Janet Erramuspe-Mainard for their assistance with technical aspects of this study. Conflict of Interest: Jennifer Donovan received funding from Millennium Pharmaceuticals and GlaxoSmithKline and served on the speakers bureau for Sanofi-Aventis. Abir O. Kanaan received funding from Genzyme and GlaxoSmithKline and served on a speakers bureau for GlaxoSmithKline. The other authors have no conflicts of interest to declare. This paper was supported by grants from the Agency for Healthcare Research and Quality (HS010481 and HS15430). Author Contributions: Dr. Gurwitz had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the analysis. Study concept and design: Gurwitz, Field, Rochon. Acquisition of data: Rochon, Lee, Gavendo, Baril, Field, Gurwitz. Analysis and interpretation of data: Donovan, Kanaan, Zhao, Field. Drafting of the manuscript: Donovan, Kanaan. Critical revision of the manuscript for important intellectual content: Donovan, Kanaan, Thomson, Rochon, Lee, Gavendo, Zhao, Baril, Field, Gurwitz. Statistical analysis: Zhao, Field. Administrative, technical, or material support: Thomson, Lee, Gavendo, Baril. Study supervision: Gurwitz. Other (intervention implementation): Rochon, Lee, Gavendo, Baril, Field, Gurwitz. Sponsor's Role: The funding agencies did not contribute to the study design; data collection, analysis, or interpretation; or the decision to submit the manuscript for publication.