# Psychometric validation of the Chinese version of the PROMIS-29 profile in community-dwelling older adults with multimorbidities

## TL;DR

This study validates the Chinese version of the PROMIS-29 questionnaire for measuring health in older adults with multiple health conditions.

## Contribution

The study provides psychometric validation of the Chinese version of PROMIS-29 for use in older adults with multimorbidities.

## Key findings

- The Chinese version of PROMIS-29 showed acceptable structural validity and reliability.
- Items demonstrated strong discriminatory power and minimal bias.
- The questionnaire is suitable for assessing symptoms and function in older adults with multiple health conditions.

## Abstract

The Patient-Reported Outcomes Measurement Information System 29-item (PROMIS-29) profile is commonly used to measure patients’ self-reported health status. This study examined the psychometric properties of the Chinese version of the PROMIS-29 (v2.1) in older individuals.

Cognitive interviews and psychometric evaluations were conducted between January and August 2023. Cognitive interviews were conducted in accordance with the Cognitive Interviewing–Reporting Framework. The Chinese version of the PROMIS-29 was revised based on feedback from respondents and experts. The structural validity of the PROMIS-29 was evaluated using confirmatory factor analysis (CFA). The convergent and discriminant validities were assessed by calculating Spearman’s rank correlation and comparing known group differences. Reliability was assessed with Cronbach’s α, and item response theory (IRT)-based psychometric assessment was performed using Rasch models for unidimensionality, local independence, item characteristic curve (ICC) matrices, and model fit. Differential item functioning (DIF) was used to examine demographic bias.

A total of 606 cases with a mean age of (72.88 ± 7.2) were included. The CFA showed acceptable convergence. PROMIS-29 was significantly correlated with comparable domains in the legacy questionnaires. Cronbach’s α of the instrument ranged from 0.82 to 0.95. The Rasch models explained 44.5 and 73.2% of the variance, respectively. Local independence analysis showed that the maximum standardized residual correlation coefficients between items within the short form ranged from 0 to 0.69 in absolute value. All items demonstrated excellent discriminatory power. A good Rasch model fit was revealed in terms of the outfit MNSQ, infit MNSQ, and overall outfit MNSQ. Most items showed acceptable item characteristic curve matrices and did not have statistically significant bias (DIF).

The Chinese version of the PROMIS-29 showed acceptable psychometric properties in community-dwelling older adults with multimorbidities. These findings suggest that this questionnaire can be beneficial in the assessment of symptoms and function in the older population.