Adjusting for patient characteristics and selection effects in assessment of community mental health centers

Partha Deb, Ann M. Holmes, Richard N. Deliberty

Research output: Contribution to journalArticlepeer-review

2 Scopus citations


Objective: The objective of this study was to determine the effect of patient socioeconomic characteristics and center selection of patients on measured performance of community mental health centers. Data Source/Study Setting: Data were taken from the administrative records of Indiana's public mental health system for 16,516 adults with severe, persistent mental illness treated in 30 community mental health centers. Center performance was compared using longitudinal information on patient functioning. Methods: A mixed random-effects model that is suitable for fitting data with a hierarchical structure was used to assess relative performance. Principal Findings: Measured performance was found to depend significantly on patient education, income, marital status, race, ethnicity, and baseline health (P <0.05). Results also indicated centers that were more successful at maintaining patients in treatment were unfairly underranked by unadjusted performance scores. Conclusions: Both the socioeconomic background of patients and patient selection by centers impact apparent performance in community mental health care. If observational data are used to evaluate community-based providers, analysts might need to account for both effects to ensure comparisons of relative performance are accurate.

Original languageEnglish (US)
Pages (from-to)251-258
Number of pages8
JournalMedical care
Issue number3
StatePublished - Mar 1 2004


  • Mental health outcomes
  • Provider profiling
  • Socioeconomic factors

ASJC Scopus subject areas

  • Public Health, Environmental and Occupational Health

Fingerprint Dive into the research topics of 'Adjusting for patient characteristics and selection effects in assessment of community mental health centers'. Together they form a unique fingerprint.

Cite this