Sleep Disturbances as Early Predictors of Depression

Introduction

Depressive disorders represent a category of mood affective disorders that include major depressive disorder, seasonal affective disorder, persistent depressive disorder, and others. According to the DSM-5 (American Psychiatric Association, 2013), these disorders centre around disturbances in an individual’s mood state — such as the presence of sadness, emptiness, or irritability — accompanied by specific cognitive and somatic changes that cause distress and hinder everyday functioning.

While there is considerable variation across depressive disorders in terms of duration, timing, and presumed causes, one symptom is remarkably consistent: disrupted sleep. Changes in sleep-wake mechanisms, including the manifestation of insomnia and hypersomnia, are present across multiple depressive disorders (American Psychiatric Association, 2013). Critically, these disturbances often appear well before the onset of a diagnosable depressive episode, making them a potential early warning signal.

This review examines the relationship between sleep disturbances and depression, the meta-analytic evidence for sleep as a predictive marker, and the emerging potential of digital sleep monitoring for early detection.

Prevalence of Sleep Disturbances in Depression

Sleep disturbances are among the most common symptoms reported by individuals with depression. Previous studies have found that sleep disturbances are present in approximately 90% of individuals with depression (Soehner et al., 2014). Nutt et al. (2008) go further, arguing that sleep disturbances should be considered core symptoms of depression rather than merely associated or comorbid conditions, given their prevalence, their impact on quality of life, and their role as risk factors for relapse and suicidality.

The nature of these disturbances is well-documented. Tsuno et al. (2005) found that individuals with depression experience decreased deep sleep (slow-wave sleep), reduced rapid-eye movement (REM) sleep latency, and an increased proportion of time spent in REM during the early half of the night. Additional findings indicate impaired sleep continuity and duration (Tsuno et al., 2005; Franzen & Buysse, 2008), difficulty initiating sleep (Franzen & Buysse, 2008), and early morning awakenings (Franzen & Buysse, 2008).

Insomnia and Depression

Insomnia — difficulty falling asleep, staying asleep, or experiencing non-restorative sleep — is the most prevalent sleep disturbance in depression. Approximately 75% of individuals diagnosed with some form of depression suffer from insomnia (Nutt et al., 2008), and insomnia complaints are one of the most common at the time of major depressive disorder diagnosis (Geoffroy et al., 2018).

The relationship between insomnia and depression is bidirectional: individuals with insomnia are more likely to develop depression, and individuals with depression are more likely to develop insomnia. However, the predictive direction — from insomnia to depression — carries particularly important clinical implications.

Meta-analytic evidence

Two large-scale meta-analyses have quantified the risk. Baglioni et al. (2011) conducted a meta-analytic evaluation of 21 longitudinal epidemiological studies and found that non-depressed individuals with insomnia had a twofold risk of developing depression compared to those without sleep difficulties, with an adjusted odds ratio of 2.10 (95% CI: 1.86–2.38). The authors concluded that early treatment of insomnia could be “a helpful general preventive strategy in the area of mental health care.”

Li et al. (2016) expanded on this with a meta-analysis of 34 prospective cohort studies encompassing 172,077 participants. The pooled relative risk was 2.27 (95% CI: 1.89–2.71), confirming that insomnia carries more than double the risk of subsequent depression. This association held across geographic regions, though the risk was more pronounced among participants from the United States compared to European cohorts.

These findings establish insomnia not merely as a symptom of depression, but as an independent risk factor for its development.

Hypersomnia and Depression

Hypersomnia — characterised by excessive daytime sleepiness despite adequate nighttime sleep, excessive daytime napping, and increased nighttime wakefulness — affects a smaller but clinically significant proportion of depressed individuals. Approximately 16–20% of individuals with depression (Tsuno et al., 2005) and 10–50% of individuals with broader mood disorders suffer from hypersomnia (Kaplan & Harvey, 2009; Geoffroy et al., 2018).

The profile of hypersomnia varies across specific depressive disorders:

• Major depressive disorder: Insomnia is the predominant sleep complaint, though hypersomnia can co-occur, particularly in presentations with atypical features (Tsuno et al., 2005).
• Bipolar disorder: Hypersomnia appears to be more prevalent, and the co-occurrence of both hypersomnia and insomnia is more strongly associated with bipolar disorder, panic disorders, and PTSD (Murru et al., 2019; Geoffroy et al., 2018).
• Seasonal affective disorder: As many as 68–80% of individuals with SAD experience hypersomnia during winter, with a higher prevalence in females (Kaplan & Harvey, 2009).

These distinct profiles suggest that the type of sleep disturbance may help differentiate between depressive subtypes, which has implications for both diagnosis and targeted intervention.

The Predictive Window: Sleep Before Depression

Perhaps the most important finding in this literature is the temporal relationship between sleep disturbances and depressive onset. Multiple studies have shown that insomnia and hypersomnia can manifest prior to the emergence of depressive disorders (Riemann & Voderholzer, 2003; Taylor et al., 2003; Lustberg & Reynolds, 2000; Fava, 2004; Vargas & Perlis, 2020). This predictive window has been estimated at one to three years before depressive disorder onset.

This temporal gap creates a meaningful opportunity for intervention. Early diagnosis and treatment of insomnia may prevent the onset of major depressive disorder in at-risk individuals (Franzen & Buysse, 2008; Vargas & Perlis, 2020).

The predictive value of sleep extends beyond initial onset. During treatment, improvement in sleep positively indicates improvement in overall prognosis (Vargas & Perlis, 2020). Conversely, sleep problems — particularly insomnia — tend to re-emerge prior to depressive relapse (Lustberg & Reynolds, 2000; Murru et al., 2019; Kaplan & Harvey, 2009; Vargas & Perlis, 2020). Monitoring sleep throughout treatment and during follow-up periods may therefore serve as a valuable indicator of worsening symptoms or impending relapse for both the individual and the clinician.

Digital Sleep Monitoring for Early Detection

Given that sleep disturbances are reliable predictors of both onset and relapse, continuous sleep monitoring has clear clinical value. Traditionally, sleep assessment has relied on polysomnography (laboratory-based) or self-report questionnaires — methods that are either expensive and impractical for long-term monitoring, or prone to recall bias and inaccuracy.

The proliferation of smartphones and consumer wearables offers a scalable alternative. These devices can passively collect behavioural sleep data — including sleep duration, sleep onset and offset times, nighttime awakenings, and sleep regularity — without requiring active user input.

Zhang et al. (2021) investigated this approach in a multicentre longitudinal study, examining the relationship between 18 sleep features measured by a wrist-worn wearable and depression severity (assessed via PHQ-8) in participants with major depressive disorder. The findings confirmed that several wearable-derived sleep features were significantly associated with depression severity:

• Insomnia indicators were significantly related to depression severity.
• Hypersomnia prevalence was elevated in depressed participants.
• Weekend catch-up sleep was significantly positively correlated with depression severity.
• Sleep efficiency and frequency of awakenings after sleep onset were associated with sleep quality and depression.

These findings demonstrate that passively collected digital sleep data can capture clinically meaningful variation in depression severity, validating the use of wearable and smartphone sensors for ongoing sleep monitoring.

Future Directions

Several directions warrant further exploration:

1. Differentiating disorders through sleep profiles. Given the distinct sleep disturbance patterns across major depressive disorder, bipolar disorder, seasonal affective disorder, and other psychiatric conditions, future work should explore whether passive sleep sensor data can reliably distinguish between these conditions. Such differentiation could improve early detection methods and enable more targeted intervention strategies.

2. Longitudinal population monitoring. Large-scale longitudinal studies using smartphone and wearable sleep data could validate the one-to-three-year predictive window in naturalistic settings, moving beyond retrospective clinical assessments.

3. Post-treatment monitoring. Incorporating passive sleep monitoring into post-treatment follow-up care plans could provide early warning of relapse, potentially reducing the burden of recurrent depressive episodes.

4. Integration with other behavioural markers. While sleep disturbances alone carry significant predictive value, combining sleep data with other passively collected behavioural signals — such as physical activity patterns, device usage, and social interaction metrics — may yield even stronger predictive models.

Conclusion

Sleep disturbances are not merely symptoms of depression — they are among its strongest predictive markers. Insomnia doubles the risk of developing depression and can appear one to three years before a depressive episode. Different types of sleep disturbances map onto different depressive subtypes, suggesting diagnostic value beyond simple risk prediction.

The emergence of passive digital sleep monitoring through smartphones and wearable devices creates a practical pathway for continuous, non-intrusive tracking of sleep patterns at population scale. By detecting sleep abnormalities early, digital health tools can support preventive intervention, improve treatment monitoring, and identify relapse risk — ultimately shifting the approach to depression from reactive treatment toward proactive prevention.