Looking inside the black box of comprehensive geriatric assessment: a classification system for problems, recommendations, and implementation strategies.

Abstract

To develop and test the inter-rater reliability of a coding system for geriatric problems identified through Comprehensive Geriatric Assessment (CGA) of hospitalized older persons, recommendations generated by the assessment, and implementation strategies for these recommendations. Validation study. A health maintenance organization and a geriatrics academic program. A total of 49 hospitalized older persons, who met at least 1 of 13 inclusionary "targeting" criteria, two geriatricians, and one social worker who coded forms. Standardized coding of CGA consultation sheets into (1) geriatric problems identified, (2) recommendations, and (3) implementation strategies; inter-rater reliability testing of coding system using two physicians and a social worker. On average, each assessed patient had 4.8 recommendations. The largest percentages of recommendations were for non-physician referrals (18.2%), advance directives (13.4%), medication adjustments (11.5%), diagnostic evaluation/monitoring (11.5%), and community services (10.9%). The proportions of agreement between raters in coding problems ranged from 0.77 to 0.90, in coding recommendations from 0.69 to 0.86, and in coding implementation strategies from 0.68 to 0.83. A classification system for measuring some components of the process of care of CGA has satisfactory inter-rater reliability, can be adapted for other settings, and may provide valuable insight into determining which components of CGA confer health benefits.