Clinical decision support to improve medication safety: improving medication prescribing in outpatient clinics

Brittany L Melton, Purdue University

Abstract

An estimated 40-51 percent of elderly Americans receive potentially inappropriate medications (PIMs). The Beers' List, which was developed to identify PIMs for the elderly, has been the gold standard reference, but it is still not commonly utilized during medication prescribing. Several studies have assessed the effectiveness of the Beers' List in reducing inappropriate prescribing in the elderly, but fewer studies have applied a computerized clinical decision support system (CDSS) to the Beers' List in an outpatient setting. This study sought to address one main research question; is a Beers' List CDSS able to reduce inappropriate prescribing in outpatient clinics? The objectives were to: 1) explore the effectiveness of a CDSS for Beers' List medications in reducing inappropriate prescribing for the elderly, 2) assess differences in prescriber acceptance of a Beers' CDSS recommendations by patient demographic characteristics, 3) assess differences in prescriber acceptance of Beers' CDSS recommendations by clinic type, and 4) assess differences in prescriber acceptance of Beers' CDSS recommendations by medication type. To address these objectives, eight Beers' List medications were chosen and a CDSS was developed using human factors principles to guide the display design. The CDSS was designed to activate when one of the study Beers' List medications was prescribed for a patient who was at least 65 years old. The CDSS was then implemented in the outpatient clinics associated with Wishard Health Services in a pre-/post- intervention observational design. Data were collected for one year prior to CDSS implementation and for seven months afterward. Data extracted included patient age, race, and gender, the medication, date, prescriber, and clinic location. Statistical analyses to assess the data included means, frequencies, linear regression, logistic regression, and logistic regression category contrasts. Prescription data were available between the pre- and post- intervention periods for 485 patients with matched prescriptions. Of the 485 patients included in the study, 74.7 percent were female, 47.7 percent were Black, and the mean age was 71.4±5.5 years. The majority of matched prescriptions written during the study period remained the same, where the medication did not change, between the pre- and post- intervention periods (93.5%). There was no significant difference between the pre- and post- intervention periods in total prescriptions (p=0.552) or proportion of Beers' List medications prescribed (p=0.983) when controlling for month of the year. Logistic regression was used to assess the effect of patient demographics, clinic type, and medication in the CDSS acceptance, defined as the number of prescribed Beers' List medications in the pre- period which were changed to an alternative medication in the post- period. There was no significant difference in acceptance when considering patient age, race, or gender (p=0.488, 0.357, and 0.516 respectively). Beers' List medication type was significant in logistic regression models (p= 0.001). There was no significant difference between primary care clinics and specialty clinics (p=0.144). Beers' List medication prescriptions changed were significantly different for diphenhydramine and clonidine when compared to most of the other Beers' List medications. While other studies saw a significant reduction in inappropriate prescribing, this study only saw a significant difference with specific medications. These results indicate the acceptance of a CDSS for Beers' List medications may be primarily dependent upon the medication rather than the other clinical factors examined in this study when considering prescription renewals. This information may be used to develop CDSS which are more selectively used for specific medications rather than for all Beers' List medications. This may lead to a more effective intervention which reduces the rate of inappropriate prescribing in the elderly.

Degree

Ph.D.

Advisors

Zillich, Purdue University.

Subject Area

Health care management

COinS