Mapping the central symptom structure of the insomnia severity index (ISI-7) and sleep duration: a network analysis

Insomnia symptoms and abnormal sleep durations adversely affect mental health and daily life, yet the complex relationships among individual symptoms remain understudied. There is a notable lack of item-level network analyses integrating the Insomnia Severity Index (ISI-7) with sleep duration categories. This study addressed this gap by using network analysis to identify central symptoms and assess gender differences within the insomnia-sleep duration structure.

A cross-sectional study was conducted among 466 job-seeking graduates from two major public universities in Bangladesh. Insomnia symptoms were assessed using the validated Bangla ISI-7, and sleep duration was categorized according to National Sleep Foundation guidelines. Network analysis (EBICglasso) was used to estimate the symptom network, compute node centrality and predictability, and test network invariance by gender using Network Comparison Tests.

Network analysis revealed a highly interconnected structure, with the strongest association between short and long sleep duration (regularized partial correlation = -0.80), which primarily reflects their mutually exclusive dummy coding rather than a substantive clinical antagonism. Dense positive connections were observed among ISI symptoms, especially between "distress caused by sleep difficulties" (ISI7) and "noticeability of sleep problem" (ISI6; edge weight = 0.48). Centrality analysis identified short sleep duration (strength = 1.26; predictability index = 0.69), ISI7 (strength = 1.20), and ISI6 (strength = 1.15) as the most central nodes. The network showed acceptable stability (CS-coefficient = 0.52). No statistically significant gender differences were found in network structure or centrality (global strength: males = 3.61, females = 3.56; p = 0.86), indicating structural invariance.

These findings suggest that symptoms such as distress about sleep and short sleep duration may represent promising targets for future intervention research. Interventions focusing on these symptoms may help improve sleep health. Future research should utilize longitudinal designs and include broader mental health measures to better understand the dynamics of insomnia.