Development and testing of a process measure of nursing home quality of care.

Abstract

To develop and test quality of care process measures for three medical conditions of nursing home patients: fever, shortness of breath, and chest pain. Flowsheets designed to capture the critical elements of care for the above conditions were developed by an expert panel. Nursing home residents charts were reviewed retrospectively using the flow sheets. The reviews were translated into clinical scenarios, and the quality of care the scenarios represented was rated by an expert panel. All nursing homes in Hennepin County, MN, that care for Medicaid patients. A random sample of 1405 Medicaid nursing home residents from 1984 and 1988. Measures of quality of physician assessment and intervention, quality of nurse assessment and intervention, and global quality were developed and the intra- and interrater reliability were tested. The measures' validity was assessed by their ability to predict resident death. Intrarater reliability was measured as the correlation of the ratings of blinded duplicates. The correlation for the global scale and the four subscales ranged from .74 to .88 (P < .001 for all). Interrater reliability was tested by examining what percentage of the quality ratings were within one unit (1-5 scale) for all three raters. All three raters were within one unit for more than 72% scenarios for all scales. The subscale of quality of physician assessment was able to predict resident death when the worst episode of care (OR = .47, 95% CI(.31-.74)) or the mean episode of care (OR .54, 95% CI(.30-.99)) was used. None of the other subscales or the global measure predicted death. Through the use of an expert panel, measures of nursing home quality of care were developed for shortness of breath, fever, and chest pain. These measures have reasonable reliability and significant face validity. Their validity is supported further by the ability of one of the measures to predict resident death.