Risk-adjusted outcome models for public mental health outpatient programs

Objective. To develop and test risk-adjustment outcome models in publicly f unded mental health outpatient settings. We developed prospective risk mode ls that used demographic and diagnostic variables; client-reported function ing, satisfaction, and quality of life; and case manager clinical ratings t o predict subsequent client functional status, health-related quality of li fe, and satisfaction with services. Data Sources/Study Setting. Data collected from 289 adult clients at five- and ten-month intervals, from six community mental health agencies in Washi ngton state located primarily in suburban and rural areas. Data sources inc luded client self-report, case manager ratings, and management information system data. Study Design. Model specifications were tested using prospective linear reg ression analyses. Models were validated in a separate sample and comparativ e agency performance examined. Principal Findings. Presence of severe diagnoses, substance abuse, client a ge, and baseline functional status and quality of life were predictive of m ental health outcomes. Unadjusted versus risk-adjusted scores resulted in d ifferently ranked agency performance. Conclusions. Risk-adjusted functional status and patient satisfaction outco me models can be developed for public mental health outpatient programs. Re search is needed to improve the predictive accuracy of the outcome models d eveloped in this study, and to develop techniques for use in applied settin gs. The finding that risk adjustment changes comparative agency performance has important consequences for quality monitoring and improvement. Issues in public mental health risk adjustment are discussed, including static ver sus dynamic risk models, utilization versus outcome models, choice and timi ng of measures, and access and quality improvement incentives.