Purpose: Current patient-reported outcomes for female urinary incontinence (UI) are limited by their inability to be tailored. Our objective is to describe the development and field testing of seven item banks designed to measure domains identified as important UI in females (UIf). We also describe the calibration and validation properties of the UIf-item banks, which allow for more efficient computerized adaptive testing (CAT) in the future.
Methods: The UIf-measures included 168 items covering seven domains: Stress UI (SUI), overactive bladder (OAB), urinary frequency, physical, social and emotional health impact and adaptation. Items underwent rigorous qualitative development and psychometric testing across two sites. Items were calibrated using item response theory and evaluated for internal consistency, construct validity and responsiveness.
Results: A total of 750 women (249 SUI, 249 OAB and 252 mixed UI) participated. Mean age was 55 ± 14 years, and 23 % were Hispanic and 80 % white. In addition to face and content validity, the measures demonstrated good internal consistency (coefficient alpha 0.92-0.98) and unidimensionality. There was evidence for construct validity with moderate-to-strong correlations with the UDI (r's ≥ 0.6) and IIQ (r's = ≥0.6) scales. The measures were responsive to change for SUI treatment (paired t test p < .001, ES range 1.3-2.9; SRM range 1.3-2.5) and OAB treatment (paired t test p < .05 for all domains except social health impact and adaptation, ES range 0.3-1.5, SRM range 0.4-1.0). The measures were responsive based on concurrent changes with the UDI and IIQ (p < 0.05). CAT versions were developed and pilot-tested.
Conclusions: The UIf-item banks demonstrate good psychometric characteristics and are a sufficiently valid set of customizable tools for measuring UI symptoms and life impact.
Keywords: Computerized adaptive test; Female; Item response theory; Patient outcome assessments; Patient-centered outcomes research; Psychometrics; Quality of life; Urinary incontinence.