Development of a functioning metric for the ageing population using data from the survey of health, ageing and retirement in Europe (SHARE)

Abstract

Background Beyond mortality and morbidity, health statistics would benefit from reporting information on functioning, the third health indicator. The objective of this article is to use data from the Swiss Survey of Health, Ageing and Retirement in Europe (SHARE) to exemplarily create a psychometrically sound and valid metric of functioning for the ageing population living in Switzerland. Methods Partial Credit Model (PCM) analysis, including analysis of targeting, item fit, local item dependencies (LID), unidimensionality, and differential item functioning (DIF), tested the psychometric properties of selected items. The DIF analysis investigated the invariance of item difficulties across sex and age groups, country, language, and the assessment Wave. Results Data from 34,092 individuals aged 50 years and older was selected across assessment Waves of SHARE. The analysis showed that a functioning metric can be constructed with a total of 33 functioning items. Items showed LID and multidimensionality initially, which was solved with a testlet approach. Aggregation into testlets resulted in good fit, unidimensionality, no LID, and no DIF for sex, country, language, and the assessment Wave. Some DIF is found for age groups. The analysis also showed that the selected items target higher levels of problems in functioning than observed in the study population. Conclusions A functioning metric can be constructed from selected functioning items of SHARE. The metric provides a sound interval-scaled score that can be used for longitudinal analyses of ageing in Switzerland and neighboring countries or as an indicator of the level of functioning in an ageing population.