Validity and reliability of an inpatient severity of psychiatric illness measure

Bentson H. McFarland, Anne E. Kovas, Shelby L. Haugan, David A. Pollack, Jo M. Mahler

Research output: Contribution to journalReview article

Abstract

Inpatient psychiatric severity measures are often used but few psychometric data are available. This study evaluated the psychometric properties (reliability and validity) of a measure used to assess severity of psychiatric illness among inpatients. Using the severity measure, minimally trained raters conducted retrospective patient record reviews to assess medical necessity for psychiatric hospitalization. The data analysis compared 135 civilly committed psychiatric inpatients with a heterogeneous group of 248 psychiatric inpatients at a general hospital. The severity measure showed acceptable inter-rater reliability in both populations. Two-way analysis of variance showed that the intra-class correlation coefficient for the total score was 0.65 for general hospital subjects and 0.63 for civilly committed subjects. Differences in mean scores were substantial (15 out of a possible 75 points for general hospital subjects versus 42 for civilly committed subjects, Mann-Whitney U = 562, p < 0.001). As expected, all civilly committed subjects were well above admission cut-off score of 12, versus only 64% of the general hospital patients. The measure is appropriate for retrospective severity assessment and may also be useful for pre-admission screening.

Original languageEnglish (US)
Pages (from-to)102-108
Number of pages7
JournalInternational Journal of Methods in Psychiatric Research
Volume14
Issue number2
DOIs
StatePublished - Jul 11 2005

Keywords

  • Health status indicators
  • Hospitalizations
  • Mental health
  • Psychometrics
  • Severity of illness

ASJC Scopus subject areas

  • Psychiatry and Mental health

Fingerprint Dive into the research topics of 'Validity and reliability of an inpatient severity of psychiatric illness measure'. Together they form a unique fingerprint.

  • Cite this