Evaluating pain outcomes in Chinese ophthalmology patients using the APS-POQ-R-C: a Rasch analysis.

BACKGROUND

This study aimed to evaluate the psychometric properties of the Chinese version of the Revised American Pain Society Patient Outcome Questionnaire (APS-POQ-R-C) using Rasch analysis, to optimize the APS-POQ-R-C for effective pain assessment in Chinese postoperative ophthalmic patients.

METHODS

The polytomous analysis approach of the Rasch model was used to comprehensively evaluate the applicability of the APS-POQ-R-C scale in postoperative ophthalmic patients. Using a sample of 294 valid questionnaires, multiple aspects of the scale were tested, including unidimensionality, local independence of items, reliability and separation, item fit, person-item mapping, test information function, and differential item functioning (DIF) analysis.

RESULTS

Principal component analysis of residuals, explained common variance (0.61) and omega hierarchical (0.72) of the APS-POQ-R-C scale demonstrates essential unidimensionality. The reliability and separation of person were 0.93 and 3.64, item were 0.99 and 10.32, indicating high reliability and separation. The standardized residual correlations between items were all below 0.7, suggesting local independence. The response category functioning results recommended merging categories 8, 9, and 10. Except for item P10, most items had infit and outfit mean square (MNSQ) values within acceptable ranges, indicating good fit to the Rasch model. Item P10's MNSQ values exceeded 1.50. The person-item map indicating that item difficulty was generally higher than the mean ability of the population. The test information curve showed that the scale was most informative for individuals with higher levels of the latent traits. DIF analysis revealed slight gender-related differential functioning in items P5a, P5b, P5c, and P5d, with absolute DIF contrast greater than 0.5.

CONCLUSION

The APS-POQ-R-C can be used to assess postoperative pain management effectively in the study sample, with overall good psychometric properties. Further optimization is suggested, including reducing item redundancy, incorporating more simple items and considering the potential influence of gender differences on responses to the scale.