Living with Cochlear Implants (LivCI): Development and validation of a new patient-reported outcome measure (PROM) of personal factors associated with living with cochlear implants

Background
There is a paucity of research that considers personal factors that have recently been shown to be important to those who use cochlear implants (CI). This study aimed to develop the Living with Cochlear Implants (LivCI) questionnaire, a patient-reported outcome measure (PROM) of personal factors important to living with cochlear implants, and evaluate its psychometric properties using Rasch and classical test theory (CTT) analyses.
Methods
Multi-phase, prospective observational design. CI candidates and recipients (n = 267) completed a 54-item draft LivCI. The item pool was assessed for Rasch model fit, including overall scale and item fit, local dependency, rating scale functioning, differential item functioning (DIF), reliability and unidimensionality. Results guided iterative refinements to produce a definitive LivCI. LivCI scales were evaluated in a second sample (n = 145) using CTT for internal consistency reliability (Cronbach’s alpha) and criterion and construct validity. A hypothesis testing approach, comparing LivCI scores with participants’ scores on five comparator PROMs, was used.
Results
We constructed a 22-item version of the LivCI comprising four unidimensional subscales: (1) function and participation; 2) psychosocial and wellbeing; 3) stigma; and 4) device aesthetics and management. Scales represented domains of a co-developed conceptual framework. All met the Rasch model requirements of fit, unidimensionality, local item independence and item invariance for age and gender (DIF), except the Device Aesthetics and Management scale with one dependent item pair. A 4-point response category structure was adopted to address threshold disordering. Person separation reliability (range = 0.57 to 0.85) and internal consistency reliability (Cronbach’s α = 0.64 to 0.94) were acceptable with exception of the Device Aesthetics and Management Scale. The LivCI was highly correlated (r > 0.6) with PROMs considered to measure similar constructs, sufficient evidence of criterion validity (Pearson correlation value range = 0.62 to 0.75, 100% of a priori hypotheses met). Construct validity was partially supported with 54% (15/28) of predefined hypotheses confirmed.
Conclusions
The 22-item LivCI, comprising four standalone scales, met Rasch model requirements for unidimensionality, invariance, local independence and item fit. With further validation, the LivCI has potential for implementation in routine care and as a research tool.