Use of computerized data to assess adherence to schizophrenia practice guidelines

Hudson TJ, Owen RR, Mason BL, Lancaster AE. Use of computerized data to assess adherence to schizophrenia practice guidelines. Presented at: VA HSR&D National Meeting; 1999 Sep 15; Washington, DC.

STUDY OBJECTIVE: Implementation of practice guidelines is an increasingly common technique to improve quality of care. Medical record review, the traditional quality assessment method, is labor intensive and typically only allows assessment of a sample of patients. A preferable approach would allow evaluation of all patients without intensive use of resources. We conducted a pilot study to compare the ability of two data sources to identify patients receiving guideline concordant care. METHOD: After obtaining informed consent, demographic and medication data were extracted from written medical records and computerized data of patients with schizophrenia who were hospitalized at a VAMC between 9/93 and 9/94. Data extraction covered the 120-day period following hospital discharge. Medication name, dose, and supply were identified from each data source along with evidence of a follow-up visit within 30 days of discharge. RESULTS: Of the 28 patients evaluated (27 male), 7 were excluded because no chart was available (n = 2) or because the patient was participating in a blinded medication trial (n = 5). The antipsychotic drug names and doses for 10 of 21 patients (48%) matched when data sources were compared. The agreement between computerized and chart data in identifying patients receiving antipsychotic medications outside the guideline-recommended range was good (kappa = 0.6). Computerized data identified a 120-day supply of medication in 62% of patients vs. 24% using chart data. A follow-up appointment was made and kept in 38.1% of patients using computerized data vs. 5% using chart data. CONCLUSION: We concluded that computerized data can be used to evaluate adherence to practice guidelines. Ths poster will discuss the feasibility of using computerized records as well as reasons for disagreement between data sources and methods to resolve these issues. (Abstract by: Author)