JMIR Publications

ABSTRACT

Background:

PROMs are important in insomnia, where diagnosis and treatment outcomes rely on patients’ subjective experience of sleep difficulties and daytime consequences. The 10-item Comprehensive Report of Insomnia Symptoms (CRIS) was developed as a brief digital PROM to capture these dimensions.

Objective:

To evaluate the structural and external validity, responsiveness, and score interpretation of the CRIS and the feasibility of a sequential psychometric validation strategy based on open online recruitment.

Methods:

The CRIS was developed through a literature review, multidisciplinary expert consultation, and two rounds of cognitive interviews. Structural validation used a sequential two-wave design, with confirmatory factor analysis in an initial online sample (N=568), followed by item revision and independent validation of the modified model (N=457), including measurement invariance across sex and age. External validation against the ISI was conducted in a separate clinical sample (N=115 at baseline; N=61 at Week 4), assessing convergent and known-groups validity, responsiveness, and exploratory ISI-anchored score ranges.

Results:

In the first wave, Q2 and Q8 showed the weakest factor loadings and helped guide revision of the scale. In Wave 2, standardized loadings ranged from 0.436 to 0.834 and the two factors were moderately correlated (r=0.525). Fit of the prespecified model was mixed (CFI=0.924, TLI=0.900, SRMR=0.0588, RMSEA=0.0943), whereas a post hoc sensitivity model allowing residual covariance between Q3 and Q5 showed improved fit (CFI=0.964, TLI=0.951, SRMR=0.0462, RMSEA=0.0659). Metric invariance across sex and age groups was supported in sensitivity analyses. CRIS scores correlated with ISI scores at baseline (r=−0.34), with a stronger association at Week 4 (r=−0.66). CRIS scores varied systematically across predefined ISI strata, supporting convergent and known-groups validity. Among participants with longitudinal data, changes in CRIS and ISI scores were correlated. Exploratory ROC analyses identified four descriptive CRIS score ranges (≥67, 45–66, 17–44, and ≤16), with 66.5% exact agreement and 98.9% agreement within the same or an adjacent ISI category. An increase of approximately 24 points in the CRIS score showed 77% sensitivity and 77% specificity for correspondence with an improvement of at least 7 points on the ISI (AUC=0.79).

Conclusions:

The CRIS is a brief digital PROM with a coherent two-factor structure, evidence of convergent validity, and evidence of responsiveness to patient-reported change. Sequential evaluation in two independently recruited online samples, together with metric invariance across sex and age, supports the feasibility of structured psychometric validation using this type of recruitment. Its brief visual-analogue format and suitability for repeated administration make the CRIS particularly relevant to digital clinical settings, including longitudinal monitoring during CBT-I and digital CBT-I.