# Psychometric Evaluation of the Insomnia Severity Index in Chronic Disease Patients Using Three Complementary Approaches

## TL;DR

The study validates a Bangla version of the Insomnia Severity Index for chronic disease patients in Bangladesh, confirming its reliability and usefulness for clinical and research purposes.

## Contribution

The study provides a comprehensive psychometric validation of the ISI-Bangla using three complementary methods in a chronic disease population.

## Key findings

- The two-factor model (Night Symptoms, Daytime Impact) showed the best balance of fit and interpretability.
- Network analysis identified central and bridging symptoms, such as difficulty falling asleep and satisfaction with sleep pattern.
- The ISI-Bangla demonstrated strong reliability and measurement invariance across genders.

## Abstract

Insomnia is highly prevalent among individuals with chronic diseases and often exacerbates disease progression while adversely impacting mental health and quality of life. However, validated tools to assess insomnia in this vulnerable population remain limited in emerging economies. This study aimed to evaluate the psychometric properties of the Bangla version of the Insomnia Severity Index (ISI) among Bangladeshi adults with chronic diseases using confirmatory factor analysis (CFA), Rasch modeling, and network analysis.

A face‐to‐face survey was conducted among adults with clinically diagnosed chronic illnesses. CFA was used to test one‐, two‐, and three‐factor structures and gender‐based measurement invariance. Rasch analysis examined item fit, reliability, and response category functioning. Network analysis estimated symptom interrelations, centrality, bridge metrics, and gender‐based network invariance.

The two‐factor model (Night Symptoms, Daytime Impact) provided the most favorable balance between statistical fit and theoretical coherence. Although the three‐factor model yielded marginally better indices, its reliance on a single‐indicator factor and high inter‐factor correlations limited its interpretability. All items showed strong factor loadings and internal consistency. Measurement invariance was supported across gender. Rasch modeling confirmed item fit, category functioning, and reliability, though some items exhibited moderate gender‐based DIF. Network analysis identified ISI_1 (difficulty falling asleep) and ISI_6 (noticeability of sleep problems) as central symptoms, while ISI_4 (satisfaction with sleep pattern) emerged as a key bridge symptom. Predictability was high, and network structure was invariant across gender.

The ISI‐Bangla demonstrates strong psychometric validity in a chronic disease population, supporting its use for clinical and research purposes in Bangladesh. Testing the alternative factor structures confirmed the robustness of the two‐factor specification. The integration of CFA, Rasch, and network analysis provides a comprehensive validation framework.

The Bangla version of the ISI was validated among adults with chronic diseases using confirmatory factor analysis, Rasch modeling, and network analysis. Findings supported a two‐factor structure (nighttime symptoms, daytime impact) with strong reliability, measurement invariance across gender, and coherent symptom networks. Key central and bridging symptoms highlight clinically actionable targets, establishing the ISI‐Bangla as a robust tool for both clinical and research use in Bangladesh.