Sleep stage transitions in patients with insomnia disorder: Implications of Markov chain analysis on sleep misperception

Introduction: Recent studies suggest that sleep misperception in Insomnia Disorder (ID) is associated with sleep instability and fragmentation, characterised by higher probabilities of stage-switching and transitioning to lighter sleep stages. We applied the Markov chains methodology to analyse sleep stage transition probabilities across the First Non-Rapid Eye Movement (NREM) Sleep Cycle (FC) and Total Sleep Time (TST) in ID patients and Healthy Controls (HC), with the aim to compare the two groups and predict sleep misperception levels. Method: Twenty IDs (mean age: 43.5 ± 12.7; 7M/13F) and eighteen sex and age-matched HCs (mean age: 41.6 ± 11.9; 8M/10F) were recruited for the study and underwent a nocturnal polysomnography recording, followed by the completion of sleep diaries the next morning. Insomnia Severity Index (ISI) questionnaire and its subscales were used to investigate clinical variables related to the disorder. Sleep stage transition probabilities were calculated following the Markovian methodology, during FC and TST. Sleep Latency misperception (SLm) and TST misperception (TSTm) indices were calculated by comparing polysomnography data with subjective sleep diary variables. These indices were used to identify HC normoestimators and ID misperceptors. The relationship between sleep stage transition probabilities and sleep misperception was analysed through non-parametric tests and correlations. Results: Considering the TST, patients with ID exhibited a higher probability of transitioning from NREM sleep stage 2 (N2) to wake, than HCs (U = 83; p = 0.005). Considering FC, IDs' symptom severity on the ISI subscale positively correlated with the probability of transit from N2 to NREM sleep stage 1 (N1) (rs = 0.534; p < 0.001). Moreover, an increasing probability of staying in N1 was correlated with reduced SLm (rs =  0.488; p = 0.002). Among HCs normoestimators, increasing probabilities of staying in N2 (rs =  0.704; p = 0.005) and NREM sleep Stage 3 (N3) during FC (rs =  0.665; p = 0.013) were significantly correlated with reduced SLm. In contrast, higher transitions between N3 and N2 (rs = 0.712; p = 0.004) and vice-versa (rs = 0.715; p = 0.004) were associated with increased SLm. Conclusion: The present findings suggest that higher sleep misperception positively correlates with increased sleep stage transitions and decreases with greater stability in the same sleep stage, both during FC and TST. Furthermore, patients with ID tend to transit more frequently from deeper sleep to lighter stages, and this probability is significantly related to higher insomnia severity.