Abstract
Purpose
Sleep disturbances are common in individuals with psychiatric diseases and may emerge years before diagnosis. It remains unclear whether they reflect modifiable risk factors or early manifestations of psychiatric illness. We aimed to investigate how different sleep dimensions relate to the risk of subsequent psychiatric illness, and how sleep symptoms evolve before and after diagnosis.
Methods
We analyzed data from 24,038 participants in the Swedish Longitudinal Occupational Survey of Health (2010–2018), linked to national health registers. Associations between baseline sleep characteristics and incident psychiatric diagnoses were estimated using Cox proportional hazards models. Symptom trajectories were modelled using non-linear mixed-effects models on a backward timescale centered at diagnosis.
Results
A total of 716 participants (2.9%) were diagnosed with a psychiatric disorder during follow-up. There was a U-shaped association between sleep duration and incident psychiatric illness. All sleep-related symptoms were associated with increased risk, with the strongest association observed for mental fatigue (HR 2.09, 95% CI 1.77–2.48). Insufficient sleep and sleep-disordered breathing were more common among those who later developed psychiatric illness, with differences remaining relatively stable over time. In contrast, other sleep complaints, such as insomnia symptoms and fatigue, progressively worsened prior to diagnosis and improved thereafter, although levels remained elevated compared to unaffected individuals. Long sleep duration and delayed wake-up times were exceptions, showing continued worsening post-diagnosis.
Conclusion
Gradually intensifying sleep disturbances may reflect emerging psychiatric vulnerability consistent with a prodromal phase. Where repeated monitoring is feasible, these symptoms could serve as early warning signs to guide preventive efforts.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1007/s00406-026-02332-9.
Keywords: Sleep disturbances, Insomnia, Fatigue, Incident psychiatric disease, Longitudinal
Background
Sleep disturbances are common in individuals with psychiatric disorders and often precede clinical diagnosis by months or even years [1–4]. Epidemiological studies have linked sleep problems, including insomnia, fatigue, circadian disruption, and sleep-disordered breathing, to a broad range of psychiatric outcomes, such as depression, anxiety, and psychotic disorders [1–4]. However, it remains unclear whether these disturbances serve as modifiable risk factors or represent early symptoms of an evolving psychiatric condition [5–7].
While much of the existing literature has focused on isolated sleep measures or specific diagnostic categories, psychiatric disorders frequently share overlapping features such as mood instability, altered arousal regulation, and impaired stress responses, all of which can influence or be influenced by sleep [8–10]. Furthermore, sleep–wake dysregulation and impaired daytime functioning, manifested as persistent fatigue or mental fatigue, may reflect early neurobiological alterations that precede overt psychiatric symptoms [11, 12].
Prospective data on how multiple dimensions of sleep evolve in the lead-up to psychiatric illness remain scarce. In particular, little is known about whether long-term patterns of sleep disturbance, changes in sleep timing, or daytime symptoms differ in individuals who later develop psychiatric disorders compared to those who remain unaffected. Understanding these trajectories could help clarify the temporal dynamics of sleep–psychiatric relationships and improve early detection strategies.
In this study, we used repeated measurements from a large Swedish cohort to investigate if long-term patterns of sleep characteristics differ in individuals who develop psychiatric illness compared to those who remain unaffected.
Methods
This study utilized data from the Swedish Longitudinal Occupational Survey of Health (SLOSH) [13], an ongoing cohort study initiated in 2006. The cohort comprises individuals sampled through the Swedish Work Environment Surveys (SWES), which target employed persons aged 16–64 years. Follow-up data have been collected biennially using postal questionnaires up to and including 2020. Participants received one of two questionnaire formats depending on employment status, either for those currently employed or for individuals working part-time or not at all. Both forms included consistent core items covering sleep habits, health indicators, and sociodemographic characteristics.
For this study, we focused on data from waves 3 through 7 (2010–2018), as these waves provided detailed and consistent assessments of sleep-related variables. At each wave, participants self-reported on a range of health, lifestyle, and sleep behaviors. Participation in the study was considered to reflect informed consent, as per Swedish ethical standards, through the return of completed questionnaires. The response rate decreased slightly over time, from 57% in 2010 to 48% in 2018. Compared with non-respondents, participants were more often women, born in Sweden, older, and more likely to hold a university degree.
Through linkage with Swedish national health registers using unique personal identity numbers, we obtained diagnostic data for psychiatric disorders (ICD codes F00-F99) from the National Patient Register and the Cause of Death Register. Among the 29,651 respondents who participated in any of the waves 3–7, we excluded individuals with an existing psychiatric diagnosis prior to baseline (n = 2781) and those lacking responses to all sleep-related items (n = 1832). The resulting sample comprised 24,038 participants (eFigure S1).
To examine longitudinal changes in sleep variables, we restricted trajectory-based analyses to individuals who had provided data at three or more waves between 2010 and 2018. A total of 16,845 participants met this inclusion criterion. Due to staggered introduction of certain sleep items and variable response rates across waves, the number of participants available for specific longitudinal models ranged from 15,035 (physical exhaustion) to 16,748 (sleep sufficiency). The study was approved by the Regional Ethics Review Board in Stockholm.
Definition of sleep related items
Participants reported their typical bedtimes and wake-up times separately for workdays and days off. These data were used to estimate average weekly sleep duration, which was then analyzed as a continuous variable as well as categorized as short (< 7 h), normal (7–8 h), or long sleep (> 8 h). Sleep timing patterns were analyzed as continuous variables. To assess intra-week variability, we calculated social jetlag as the difference in mid-time sleep between workdays and non-workdays, an indicator of circadian misalignment.
Sleep-related symptoms were measured using items from the Karolinska Sleep Questionnaire [14], which assess both nighttime disturbances and daytime impairments. Participants rated the frequency of symptoms including difficulties falling asleep, repeated awakenings, premature awakenings, difficulty waking up, non-restorative sleep, snoring, sleep apnea, as well as persistent fatigue, physical exhaustion, and mental fatigue. Each symptom was scored on a six-point ordinal scale: (1) never, (2) rarely/single occasion, (3) sometimes/a few times a month, (4) frequently/1–2 times/week, (5) mostly/3–4 times/week, and (6) always/5 times or more per week. Nighttime symptoms were considered present if reported at least three times per week (responses 5–6), and daytime symptoms were coded as present if they occurred twice a week or more. Nighttime insomnia was defined as the presence of at least one of the following occurring three or more times per week: difficulties falling asleep, repeated awakenings, or premature awakenings. In addition, participants rated their subjective sleep sufficiency on a five-point scale, which was dichotomized into sufficient (scores 1–3) or insufficient (scores 4–5) sleep.
Statistical analysis
Descriptive statistics were used to characterize the study population. Categorical variables were described with counts and proportions, while continuous variables were summarized using means and standard deviations (SD).
To investigate associations between sleep characteristics and the development of incident psychiatric disorder, we employed Cox proportional hazards regression models. Participants were followed from the date of their baseline questionnaire until the earliest occurrence of psychiatric diagnosis, death, loss to follow-up, or end of follow-up in 2018. The assumption of proportional hazard was tested using Schoenfeld residuals and found to be satisfied in all models. All Cox models were adjusted for a range of baseline covariates, selected based on prior research and clinical relevance. These included age, sex, educational attainment (university vs. no university), smoking status (current vs. non-smoker), physical activity level (low vs. high), and body mass index categorized as underweight, normal weight, overweight, or obese according to WHO standards. Additional adjustments were made for regular use of sleep medication (yes/no), habitual sleep duration (< 7, 7–8, > 8 h) and co-morbidities, including (cardiovascular disease ICD I00-I99, cancer ICD C00-C97, and neurological conditions ICD G00-G99).
To examine longitudinal patterns in sleep symptoms before and after psychiatric diagnosis, we applied linear mixed-effects models among participants with at least three repeated sleep assessments. Time was modeled on a backward scale, where 0 represented the year of diagnosis for those with psychiatric disease, or the end of follow-up for those without. The models included fixed effects for psychiatric disease status, time, and their interaction, enabling us to estimate diverging sleep trajectories and symptom evolution relative to diagnosis. Random effects included a random intercept and slope for time at the individual level. We compared rates of change in sleep variables between participants who developed a psychiatric disease and those who did not and further examined within-person changes pre- and post-diagnosis. Annual differences in sleep-related symptoms between participants with and without incident psychiatric disease were estimated, with positive values indicating worse symptom scores among those who developed psychiatric disease. These models were adjusted for age at year 0, sex, and education attainment. All analyses were performed using SAS software, version 9.4 (SAS Institute Inc, Cary, NC).
Results
Among 25,038 participants without psychiatric diagnosis at baseline, 716 individuals (2.9%) developed a psychiatric disorder during follow-up (median [IQR] duration, 4 [4–8] years). Compared to those who remained free of psychiatric illness, individuals who later developed a psychiatric condition were younger, more likely to be female, and more often current smokers. They also reported lower levels of physical activity and more frequent use of hypnotic medication. Furthermore, participants with incident psychiatric illness more often had a pre-existing neurological disease. Baseline characteristics for the total population and stratified by incident psychiatric disease are presented in Table 1.
Table 1.
Characteristics of study population
| Characteristics | Total | Incident psychiatry disease | P value | |
|---|---|---|---|---|
| Yes | No | |||
| N | 25,038 | 716 | 24,322 | |
| Mean age (SD) | 51.5 ± 11.5 | 49.1 ± 11.9 | 51.6 ± 11.5 | <0.0001 |
| Sex | ||||
| Female, n (%) | 13,750 (54.9) | 446 (62.3) | 13,304(54.7) | <0.0001 |
| Male, n (%) | 11,288 (45.1) | 270 (37.7) | 11,018(45.3) | |
| Higher education, n (%) | 10,041(40.2) | 277 (38.7) | 9764 (40.2) | 0.4009 |
| Smoking status | ||||
| Non-smokers, n (%) | 21,830 (88.0) | 587 (82.7) | 21,243 (88.2) | <0.0001 |
| Current smokers, n (%) | 2970 (12.0) | 123 (17.3) | 2847 (11.8) | |
| Physical activity | ||||
| Low n, (%) | 4756 (19.1) | 169 (23.7) | 4587 (19.0) | 0.002 |
| High n, (%) | 20,094 (80.9) | 543 (76.3) | 19,551 (81.0) | |
| Body mass index | ||||
| Underweight, n (%) | 221 (0.9) | 7 (1.0) | 214 (0.9) | 0.713 |
| Normal weight, n (%) | 11,446 (47.2) | 317 (45.5) | 11,129 (47.2) | |
| Overweight, n (%) | 9414 (38.8) | 272 (39.1) | 9142 (38.8) | |
| Obesity, n (%) | 3178 (13.1) | 100 (14.4) | 3078 (13.1) | |
| Co-morbidities | ||||
| Neurological disease n (%) | 2325 (9.3) | 91 (12.7) | 2234 (9.2) | 0.001 |
| Cardiovascular, n (%) | 2254 (9.0) | 78 (10.9) | 2176 (8.9) | 0.073 |
| Cancer, n (%) | 1521 (6.1) | 40 (5.6) | 1481 (6.1) | 0.579 |
| Hypnotic use | ||||
| No, (%) | 23,657 (94.5) | 619 (86.5) | 23,038 (94.7) | <0.0001 |
| Yes, n (%) | 1381 (5.5) | 97 (13.5) | 1284 (5.3) | |
| Sleep duration | ||||
| < 7 h/night, n (%) | 4212 (16.8) | 128 (17.9) | 4084 (16.8) | <0.0001 |
| 7–8 h/night, n (%) | 10,887 (43.5) | 256 (35.8) | 10,631 (43.7) | |
| > 8 h/night, n (%) | 9939 (39.7) | 332 (46.4) | 9607 (39.5) | |
| Difficulties falling asleep, n (%) | 1809 (7.3) | 107 (15.1) | 1702 (7.1) | <0.0001 |
| Repeated awakenings, n (%) | 2472 (10.0) | 123 (17.4) | 2349 (9.8) | <0.0001 |
| Premature awakening, n (%) | 2449 (9.9) | 137 (19.4) | 2312 (9.6) | <0.0001 |
| Snoring, n (%) | 2075 (8.4) | 81 (11.4) | 1994 (8.3) | 0.003 |
| Apnea, n (%) | 644 (2.6) | 34 (4.9) | 610 (2.6) | 0.000 |
| Difficulties awakening, n (%) | 1442 (5.8) | 84 (11.8) | 1358 (5.6) | <0.0001 |
| Non-refreshing sleep, n (%) | 3881 (15.7) | 198 (28.0) | 3683 (15.3) | <0.0001 |
| Insufficient sleep, n (%) | 2713 (10.9) | 137 (19.2) | 2576 (10.6) | <0.0001 |
| Persistent fatigue, n (%) | 4362 (17.7) | 232 (33.0) | 4130 (17.2) | <0.0001 |
| Physical exhaustion, n (%) | 2549 (10.3) | 155 (21.9) | 2394 (10.0) | <0.0001 |
| Mental fatigue, n (%) | 4161 (16.8) | 231 (32.6) | 3930 (16.3) | <0.0001 |
Missing data (n): education (63); smoking status (238); physical activity (188); BMI (779); difficulties falling sleep (199); repeated awakenings (261); premature awakening (362); snoring (399); apnea (580); difficulties awakening (250); non-refreshing sleep (309); insufficient sleep (120); persistent fatigue (327); physical exhaustion (304); mental fatigue (270)
Associations between individual sleep variables and risk of psychiatric disease
Most baseline sleep characteristics were significantly associated with increased risk of incident psychiatric disease. We observed a U-shaped association between average sleep duration and risk of psychiatric disease (p for overall < 0.001, p for non-linearity = 0.0007) (eFigure S2). Compared to 7–8 h of sleep per night, both short (< 7 h/night) and long (> 8 h/night) sleep durations were associated with increased risk (HR 1.30, 95% CI 1.04–1.62 and HR 1.40, 95% CI 1.19–1.66, respectively) (Table 2).
Table 2.
Associations of sleep variables at baseline with incident psychiatry disease
| Sleep variable | N | Person-years | Outcome | HR (95% CI)a | HR (95% CI)b |
|---|---|---|---|---|---|
| Sleep duration | |||||
| < 7 h | 4212 | 22,661 | 128 | 1.37 (1.11–1.70) | 1.30 (1.04–1.62) |
| 7–8 h | 10,887 | 61,219 | 256 | 1.0 (reference) | 1.0 (reference) |
| > 8 h | 9939 | 55,483 | 332 | 1.41 (1.20–1.67) | 1.40 (1.19–1.66) |
| Difficulties falling asleep | |||||
| No | 23,030 | 128,303 | 603 | 1.0 (reference) | 1.0 (reference) |
| Yes | 1809 | 10,013 | 107 | 2.24 (1.82–2.75) | 1.63 (1.30–2.04) |
| Repeated awakenings | |||||
| No | 22,305 | 124,236 | 584 | 1.0 (reference) | 1.0 (reference) |
| Yes | 2472 | 13,701 | 123 | 1.93 (1.58–2.34) | 1.55 (1.26–1.91) |
| Premature awakening | |||||
| No | 22,227 | 123,377 | 570 | 1.0 (reference) | 1.0 (reference) |
| Yes | 2449 | 14,000 | 137 | 2.05 (1.70–2.47) | 1.75 (1.44–2.13) |
| Snoring | |||||
| No | 22,564 | 125,506 | 627 | 1.0 (reference) | 1.0 (reference) |
| Yes | 2075 | 11,637 | 81 | 1.56 (1.23–1.97) | 1.27 (0.99–1.63) |
| Apnea | |||||
| No | 23,814 | 132,394 | 666 | 1.0 (reference) | 1.0 (reference) |
| Yes | 644 | 3637 | 34 | 2.17 (1.53–3.08) | 1.69 (1.17–2.45) |
| Difficulties awakening | |||||
| No | 23,346 | 129,848 | 625 | 1.0 (reference) | 1.0 (reference) |
| Yes | 1442 | 8179 | 84 | 1.90 (1.51–2.40) | 1.60 (1.26–2.03) |
| Non-refreshing sleep | |||||
| No | 20,848 | 115,858 | 510 | 1.0 (reference) | 1.0 (reference) |
| Yes | 3881 | 21,808 | 198 | 1.91 (1.62–2.26) | 1.65 (1.39–1.97) |
| Insufficient sleep | |||||
| No | 22,205 | 123,427 | 577 | 1.0 (reference) | 1.0 (reference) |
| Yes | 2713 | 15,362 | 137 | 1.79 (1.48–2.16) | 1.53 (1.26–1.87) |
| Persistent fatigue | |||||
| No | 20,349 | 112,810 | 472 | 1.0 (reference) | 1.0 (reference) |
| Yes | 4362 | 24,729 | 232 | 2.12 (1.80–2.48) | 1.81 (1.53–2.14) |
| Physical exhaustion | |||||
| No | 22,185 | 123,265 | 554 | 1.0 (reference) | 1.0 (reference) |
| Yes | 2549 | 14,454 | 155 | 2.23 (1.87–2.67) | 1.89 (1.56–2.28) |
| Mental fatigue | |||||
| No | 20,607 | 115,961 | 478 | 1.0 (reference) | 1.0 (reference) |
| Yes | 4161 | 21,913 | 231 | 2.39 (2.03–2.80) | 2.09 (1.77–2.48) |
aAdjusted for age, sex, and education
bAdjusted for age, sex, education, smoking status, physical activity, body mass index, hypnotic use, average sleep duration, and comorbidities (cancer, cardiovascular disease, neurological disease); model for sleep duration were not adjusted for average sleep duration
Nighttime insomnia symptoms, including difficulties falling asleep, repeated or premature awakenings, were all associated with elevated risk of incident psychiatric illness, as were non-restorative sleep indicators such as difficulties awakening and non-refreshing sleep. Some of the strongest associations were found for daytime symptoms, including persistent fatigue, physical and mental exhaustion (Table 2). These findings remained largely consistent after excluding the first two years of follow-up (eTable S1) and after stratification by age (eTable S2).
In the basic-adjusted model, we observed a J-shaped association between social jetlag and the risk of incident psychiatric disease, with the lowest risk around 0–1 h of discrepancy between workday and free-day sleep timing. The association remained similar but lost statistical significance after full adjustment for lifestyle and health-related covariates (p for overall = 0.149) (eFigure S3).
Mental fatigue emerged as the strongest independent sleep-related predictor of psychiatric disease. In fully adjusted models including all sleep variables, mental fatigue and long sleep duration remained significantly associated with higher risk of incident psychiatric disease (HR 1.66, 95% CI: 1.33–2.09 and HR 1.34, 95% CI 1.14–1.58, respectively) (eTable S3).
Combined sleep characteristics
Nighttime insomnia at baseline, defined as difficulties falling asleep, repeated awakenings, or premature awakenings occurring at least three times per week, was associated with an increased risk of developing psychiatric disease (HR 1.72, 95% CI 1.45–2.04) (Table 3). Participants reporting either nighttime insomnia or persistent fatigue alone had an elevated risk compared to those without symptoms, while those experiencing both had more than twice the risk of psychiatric disease (HR 2.19, 95% CI 1.77–2.71). A similar pattern was observed when daytime symptoms were defined more broadly as persistent fatigue, physical exhaustion, or mental fatigue.
Table 3.
Associations between nighttime insomnia, with and without daytime consequences, and incident psychiatry disease
| Nighttime insomnia | N | Person-years | Outcome | HR (95% CI)a | HR (95% CI)b | |
|---|---|---|---|---|---|---|
| No | 20,409 | 113,458 | 487 | 1.0 (reference) | 1.0 (reference) | |
| Yes | 4257 | 23,857 | 219 | 2.09 (1.78–2.46) | 1.72 (1.45–2.04) | |
| Nighttime insomnia with or without persistent fatigue | ||||||
|---|---|---|---|---|---|---|
| Nighttime insomnia | Persistent fatigue | N | Person-years | Outcome | HR (95% CI)a | HR (95% CI)b |
| No | No | 18,121 | 100,415 | 382 | 1.0 (reference) | 1.0 (reference) |
| No | Yes | 2160 | 12,356 | 100 | 1.98 (1.58–2.47) | 1.77 (1.41–2.22) |
| Yes | No | 2052 | 11,460 | 84 | 1.92 (1.51–2.43) | 1.60 (1.24–2.05) |
| Yes | Yes | 2156 | 12,133 | 132 | 2.73 (2.24–3.33) | 2.19 (1.77–2.71) |
| Nighttime insomnia with or without daytime symptoms | ||||||
|---|---|---|---|---|---|---|
| Nighttime insomnia | Daytime symptom | N | Person-years | Outcome | HR (95% CI)a | HR (95% CI)b |
| No | No | 16,797 | 93,228 | 350 | 1.0 (reference) | 1.0 (reference) |
| No | Yes | 3443 | 19,339 | 135 | 1.72 (1.41–2.10) | 1.59 (1.29–1.95) |
| Yes | No | 1587 | 8911 | 56 | 1.67 (1.26–2.21) | 1.39 (1.03–1.87) |
| Yes | Yes | 2621 | 14,696 | 160 | 2.76 (2.29–3.33) | 2.24 (1.83–2.74) |
| Sleep duration and nighttime insomnia | ||||||
|---|---|---|---|---|---|---|
| Sleep duration | Insomnia | N | Person-years | Outcome | HR (95% CI)a | HR (95% CI)b |
| < 7 h | No | 3203 | 17,195 | 72 | 1.15 (0.87–1.50) | 1.09 (0.83–1.44) |
| < 7 h | Yes | 866 | 4736 | 52 | 2.93 (2.16–3.99) | 2.38 (1.72–3.27) |
| ≥ 7, ≤ 8 h | No | 9133 | 51,176 | 189 | 1.0 (reference) | 1.0 (reference) |
| ≥ 7, ≤ 8 h | Yes | 1652 | 9454 | 67 | 1.90 (1.43–2.51) | 1.61 (1.21–2.16) |
| > 8 h | No | 8073 | 45,087 | 226 | 1.36 (1.12–1.65) | 1.38 (1.13–1.68) |
| > 8 h | Yes | 1739 | 9667 | 100 | 2.74 (2.14–3.50) | 2.24 (1.73–2.90) |
| Sleep duration and persistent fatigue | ||||||
|---|---|---|---|---|---|---|
| Sleep duration | Persistent fatigue | N | Person-years | Outcome | HR (95% CI)a | HR (95% CI)b |
| < 7 h | No | 3213 | 17,121 | 72 | 1.18 (0.90–1.55) | 1.13 (0.86–1.50) |
| < 7 h | Yes | 915 | 5050 | 52 | 2.75 (2.02–3.74) | 2.29 (1.66–3.16) |
| ≥ 7, ≤ 8 h | No | 9037 | 50,512 | 181 | 1.0 (reference) | 1.0 (reference) |
| ≥ 7, ≤ 8 h | Yes | 1729 | 10,055 | 70 | 1.83 (1.38–2.41) | 1.60 (1.20–2.13) |
| > 8 h | No | 8099 | 45,177 | 219 | 1.34 (1.10–1.64) | 1.35 (1.10–1.65) |
| > 8 h | Yes | 1718 | 9624 | 110 | 3.00 (2.36–3.81) | 2.55 (1.99–3.27) |
| Sleep duration and daytime symptoms | ||||||
|---|---|---|---|---|---|---|
| Sleep duration | Daytime symptom | N | Person-years | Outcome | HR (95% CI)a | HR (95% CI)b |
| < 7 h | No | 2886 | 15,335 | 63 | 1.22 (0.91–1.63) | 1.16 (0.86–1.57) |
| < 7 h | Yes | 1230 | 6785 | 62 | 2.57 (1.91–3.45) | 2.19 (1.62–2.98) |
| ≥ 7, ≤ 8 h | No | 8255 | 46,220 | 157 | 1.0 (reference) | 1.0 (reference) |
| ≥ 7, ≤ 8 h | Yes | 2490 | 14,247 | 95 | 1.85 (1.43–2.39) | 1.67 (1.28–2.17) |
| > 8 h | No | 7401 | 41,427 | 191 | 1.35 (1.09–1.67) | 1.35 (1.09–1.68) |
| > 8 h | Yes | 2404 | 13,309 | 139 | 2.86 (2.27–3.61) | 2.50 (1.97–3.17) |
Nighttime insomnia difficulties falling asleep, repeated awakenings and/or premature awakening at least three times per week, daytime symptoms persistent fatigue, physical exhaustion, or mental fatigue, HR hazard ratio, CI confidence interval
aAdjusted for age, sex, and education
bAdjusted for age, sex, education, smoking status, physical activity, body mass index, hypnotic use, average sleep duration, and comorbidities
Bold values indicate statistical significance
Insomnia and fatigue were consistently associated with elevated risk across all sleep duration groups, with the strongest associations observed in individuals sleeping more than eight hours and reporting either insomnia (HR 2.24, 95% CI 1.73–2.90), persistent fatigue (HR 2.55, 95% CI 1.99–3.27), or daytime symptoms more broadly (HR 2.50, 95% CI 1.97–3.17) (Table 3).
Longitudinal trajectories of sleep variables before and after incident neurological disease
Pre-diagnostic differences between unaffected individuals and participants who developed psychiatric disease were evident in multiple sleep dimensions (Figs. 1, 2, eTables S4 and S5). Insufficient sleep, snoring, and sleep apnea were more prevalent among those with incident psychiatric disease, with differences remaining relatively stable over time (Fig. 1, eTables S4 and S5). Difficulties falling asleep began to diverge at least eight years prior to diagnosis, while repeated or premature awakenings, difficulties awakening, and non-refreshing sleep began to diverge around 6–7 years prior to diagnosis, with progressively widening differences over time (Fig. 1 and eTable S5). Daytime symptoms, including persistent fatigue, physical exhaustion, and mental fatigue, also differed by psychiatric diseases status from at least eight years prior to diagnosis, with the largest contrasts observed near the time of diagnosis (Fig. 2, eTables S4 and S5).
Fig. 1.
Trajectories of nighttime sleep symptoms before and after psychiatric diagnosis, by disease status. Mean trajectories (solid lines) and 95% confidence intervals (shaded areas) for snoring (A), apnea (B), premature awakening (C), difficulties falling asleep (D), difficulties awakening (E), number of awakenings (F), non-refreshing sleep (G), insufficient sleep (H), and sleep duration (I) were predicted by linear mixed-effect models, adjusted for age at time 0, sex, and education. The retrospective timeline captures the eight years prior to psychiatric diagnosis or end of follow-up. The prospective timeline captures the six years post-diagnosis. The blue star indicates a significant difference in slope before year 0 between those with and without psychiatric disease. The red star indicates a significant difference in slope before and after diagnosis among those with incident psychiatric illness. Details about coefficients are presented in eTable S4
Fig. 2.
Trajectories of daytime symptoms before and after psychiatric diagnosis, by disease status. Mean trajectories (solid lines) and 95% confidence intervals (shaded areas) for persistent fatigue (A), physical exhaustion (B), and mental fatigue (C) were predicted by linear mixed-effect models, adjusted for age at time 0, sex, and education. The retrospective timeline captures the eight years prior to psychiatric diagnosis or end of follow-up. The prospective timeline captures the six years post-diagnosis. The blue star indicates a significant difference in slope before year 0 between those with and without psychiatric disease. The red star indicates a significant difference in slope before and after diagnosis among those with incident psychiatric illness. Details about coefficients are presented in eTable S4
Most symptom trajectories peaked around the time of diagnosis and showed partial or complete normalization thereafter. For example, the slope of difficulties falling asleep was significantly steeper among participants who developed psychiatric disease compared to unaffected individuals (β = 0.065, 95% CI 0.043–0.086, p < 0.0001), followed by a reversal post-diagnosis (Δ slope = −0.105, 95% CI −0.156 to −0.054) (Fig. 1 and eTable S4).
Sleep duration differed modestly between groups, with significant but small increases emerging from year –5 (Fig. 1, eTables S4 and S5) among those who developed psychiatric disease. However, long sleep duration did not show evidence of post-diagnostic improvement (Fig. 1 and eTable S4).
Changes in sleep timing were subtle but detectable in the 5 years leading up to psychiatric diagnosis, with individuals who later developed psychiatric illness gradually waking up later (β difference = 0.036, 95% CI 0.013–0.059, p = 0.003) (eTable S5, eFigure S4). We observed no significant differences in social jetlag trajectories between groups (eFigure S5).
Discussion
In this prospective study, we found that a wide range of sleep-related symptoms, including nighttime insomnia symptoms, sleep-disordered breathing, and daytime dysfunction, were associated with increased risk of subsequent psychiatric disease. These associations were evident several years prior to clinical diagnosis and generally intensified over time. While many symptoms improved after diagnosis, they did not return to the baseline levels observed among unaffected individuals. Long sleep duration and delayed wake-up times continued to worsen post-diagnosis.
These findings indicate that alterations in sleep may not merely reflect secondary effects of emerging psychiatric symptoms but could represent early warning signs of an unfolding disease process. The progressive divergence in sleep-related symptoms between affected and unaffected individuals strengthens the argument for sleep disturbances as part of a prodromal phase, a subtle clinical state that precedes overt psychiatric illness [15–17]. However, the persistent nature of several symptoms following diagnosis complicates a purely prodromal interpretation.
One possible explanation for this pattern is that psychiatric disorders and sleep disturbances share underlying biological mechanisms. Chronic sleep disruption has been associated with increased activity in inflammatory pathways [18], altered hypothalamic–pituitary–adrenal axis function [19], altered monoaminergic signaling [20], and reduced neuroplasticity [21], all of which have also been implicated in the pathophysiology of depression, anxiety, and other psychiatric conditions [22–24]. Shifts in sleep timing, such as progressively later wake-up times, may reflect subtle alterations in circadian regulation. Although we did not observe significant differences in social jetlag, variation in circadian preference and genetic differences in clock genes have been implicated in psychiatric vulnerability [25, 26] and could underlie some of the observed changes.
The strong association between fatigue and subsequent psychiatric illness may also reflect shared biological pathways. Notably, mental fatigue emerged as a stronger predictor of psychiatric disease than several traditional insomnia symptoms. This observation may suggest that impaired daytime functioning reflects emerging psychiatric vulnerability more closely than sleep disturbance itself, although the two are likely closely interconnected. This is supported by longitudinal data and emerging genetic evidence from twin studies suggesting overlapping genetic and environmental factors that predispose individuals to both fatigue and psychiatric disorders [27, 28]. Taken together, the progressive emergence and persistence of symptoms such as fatigue and extended sleep duration in our data may thus reflect neurobiological changes that unfold over time, potentially interacting with psychosocial stressors, genetic predisposition, and emerging psychiatric symptoms in a bidirectional manner.
Not all sleep complaints showed the same temporal patterns. Long sleep duration and mental fatigue were consistently associated with psychiatric onset across subgroups and analytical models and remained elevated post-diagnosis, potentially reflecting enduring neurobiological changes such as persistent inflammation, circadian dysregulation, or medication effects, but could also represent compensatory responses to physiological or emotional strain. In contrast, classical insomnia symptoms were more sensitive to age, follow-up length, and co-occurring symptoms, suggesting heterogeneity in their predictive value. The partial normalization post-diagnosis likely reflects improved symptom management rather than a true resolution of the underlying dysfunction.
From a clinical perspective, the progressive deterioration in sleep before diagnosis could offer opportunities for early detection and intervention. Individuals with cumulative multi-dimensional sleep complaints, particularly when accompanied by fatigue or sleep timing shifts, could be at elevated risk of psychiatric illness. Monitoring such patterns over time might help identify individuals in the early stages of disease development, potentially enabling targeted support before the onset of full clinical symptoms.
Strengths of this study include its large sample, repeated assessments over a long follow-up period, and the wide range of sleep variables examined. The use of a backward timescale allowed us to examine changes in sleep relative to the timing of diagnosis, improving temporal resolution. However, several limitations should be considered. Respondents were more often women, older, born in Sweden, and more likely to hold a university degree, limiting generalizability beyond the Swedish working-age population. Trajectory analyses required participation in three or more waves, which further selects for more stably engaged participants. However, the baseline prevalence of sleep-related complaints was similar in the total eligible sample and in the subset contributing three or more waves, which mitigates the risk of selection bias. All sleep variables were self-reported, and although this enhances ecological validity, it may introduce differential misclassification that would dilute associations. Information regarding objective sleep apnea diagnosis, severity, and treatment was not available. Future studies incorporating clinical sleep assessments and treatment data are needed to determine whether effective treatment of sleep-disordered breathing influences the subsequent risk of psychiatric illness. Diagnostic classification relied on national registers covering inpatient and specialized and outpatient care and the Cause of Death Register and may not capture subthreshold or undiagnosed psychiatric conditions. Although we captured diagnoses from 1997 onward and excluded individuals with a recorded psychiatric diagnosis at baseline, some participants may have had diagnoses prior to 1997, with subsequent remission, leading to misclassification of prior cases as incident. Information on specific psychiatric diagnoses was not available in the linked dataset. Consequently, we were unable to investigate whether the observed associations differed across psychiatric diagnostic categories. Finally, despite adjustment for a broad set of covariates, residual and time-varying confounding remain possible.
In conclusion, changes in sleep patterns and daytime functioning appear to emerge several years prior to psychiatric diagnosis, with many symptoms stabilizing or partially improving thereafter. While a causal role for sleep disturbances cannot be excluded, the overall temporal pattern aligns more closely with a prodromal phase of psychiatric disease. In settings where repeated assessment is feasible, sleep-related symptoms may serve as valuable early warning signs.
Supplementary Information
Below is the link to the electronic supplementary material.
Author contributions
Conception and design of the study: all authors Acquisition of data: LMH Statistical analysis: JG Drafting of the manuscript: AKH All authors commented on the draft and approved the final version of the manuscript.
Funding
Open access funding provided by Karolinska Institute. This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Data availability
The SLOSH data generated and analyzed during the current study is not publicly available due to ethical restrictions and considering that sensitive personal data are involved. Access to the data may be provided to other researchers in line with Swedish law and after consultation with the Stockholm University legal department. Requests for SLOSH data, stored at the Department of Psychology, Stockholm University should be sent to [data@slosh.se](mailto:data@slosh.se).
Declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Ethical approval
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
References
- 1.Hertenstein E, Feige B, Gmeiner T et al (2019) Insomnia as a predictor of mental disorders: a systematic review and meta-analysis. Sleep Med Rev 43:96–105 [DOI] [PubMed] [Google Scholar]
- 2.Wulff K, Gatti S, Wettstein JG, Foster RG (2010) Sleep and circadian rhythm disruption in psychiatric and neurodegenerative disease. Nat Rev Neurosci 11:589–599 [DOI] [PubMed] [Google Scholar]
- 3.Peppard PE, Szklo-Coxe M, Hla KM, Young T (2006) Longitudinal association of sleep-related breathing disorder and depression. Arch Intern Med 166:1709–1715 [DOI] [PubMed] [Google Scholar]
- 4.Skapinakis P, Lewis G, Mavreas V (2004) Temporal relationships between unexplained fatigue and depression: longitudinal data from an international study in primary care. Psychosom Med 66:330–335 [DOI] [PubMed] [Google Scholar]
- 5.Freeman D, Sheaves B, Waite F et al (2020) Sleep disturbance and psychiatric disorders. Lancet Psychiatry 7:628–637 [DOI] [PubMed] [Google Scholar]
- 6.Fang H, Tu S, Shen J, Shao A (2019) Depression in sleep disturbance: a review on a bidirectional relationship, mechanisms and treatment. J Cell Mol Med 23:2324–2332 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Alvaro PK, Roberts RM, Harris JK (2013) A systematic review assessing bidirectionality between sleep disturbancies, anxiety, and depression. Sleep 36:1059–1068 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Riemann D, Krone LB, Wulff K, Nissen C (2020) Sleep, insomnia, and depression. Neuropsychopharmacology 45:74–89 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Palagini L, Moretto U, Novi M et al (2018) Lack of resilience is related to stress-related sleep reactivity, hyperarousal, and emotion dysregulation in insomnia disorder. J Clin Sleep Med 14:759–766 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Baglioni C, Spiegelhalder K, Lombardo C, Riemann D (2010) Sleep and emotions: a focus on insomnia. Sleep Med Rev 14:227–238 [DOI] [PubMed] [Google Scholar]
- 11.Germain A, Kupfer DJ (2008) Circadian rhythm disturbancies in depression. Hum Psychopharmacol 23:571–585 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Harvey AG (2008) Sleep and circadian rhythms in bipolar disorder: seeking synchrony, harmony, and regulation. Am J Psychiatry 165:820–229 [DOI] [PubMed] [Google Scholar]
- 13.Magnusson Hanson LL, Leineweber C, Persson V et al (2018) Cohort profile: the Swedish longitudinal occupational survey of health (SLOSH). Int J Epidemiol 47:1014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Nordin M, Åkerstedt T, Nordin S (2013) Psychometric evaluation and normative data for the Karolinska sleep questionnaire. Sleep Biol Rhythms 11:2016–2226 [Google Scholar]
- 15.Van Meter AR, Burke C, Youngstrom EA et al (2016) The bipolar prodrome: meta-analysis of symptom prevalence prior to initial or recurrent mood episodes. J Am Acad Child Adolesc Psychiatry 55:543–555 [DOI] [PubMed] [Google Scholar]
- 16.Basquin L, Maruani J, Leseur J et al (2024) Study of the different sleep disturbances during the prodromal phase of depression and mania in bipolar disorder. Bipolar Disord 26:454–467 [DOI] [PubMed] [Google Scholar]
- 17.Pede VB, Jaiswal SV, Sawant VA (2017) Study of prodromal and residual symptoms of depression. Ind Psychiatry J 26:121–127 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Irwin MR, Olmstead R, Carroll JE (2016) Sleep disturbance, sleep duration, and inflammation: a systematic review and meta-analysis of cohort studies and experimental sleep deprivation. Biol Psychiatry 80:40–52 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Nicolaides NC, Vgontzas AN, Kritikou I, et al. (2020) HPA axis and sleep. 2020 Nov 24. In: Endotext. South Dartmouth: MDText.com, Inc. Accessed 2025 Jul 21.
- 20.Menon JM, Nolten C, Achterberg EJM et al (2019) Brain microdialysate monoamines in relation to circadian rhythms, sleep, and sleep deprivation – a systematic review, network meta-analysis, and new primary data. J Circadian Rhythms 17:1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Weiss JT, Donlea JM (2022) Roles for sleep in neural and behavioral plasticity: reviewing variation in the consequences of sleep loss. Front Behav Neurosci 15:777799 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Miller AH, Raison CL (2016) The role of inflammation in depression: from evolutionary imperative to modern treatment target. Nat Rev Immunol 16:22–34 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Lei AA, Phang VWX, Lee YZ et al (2025) Chronic stress-associated depressive disorders: the impact of HPA axis dysregulation and neuroinflammation on the hippocampus – a mini review. Int J Mol Sci 26:2940 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Krishnan V, Nestler EJ (2008) The molecular neurobiology of depression. Nature 455:894–902 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Charrier A, Olliac B, Roubertoux P et al (2017) Clock genes and altered sleep-wake rhythms: their role in the development of psychiatric disorders. Int J Mol Sci 18:938 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Walker WH, Walton JC, DeVries AC et al (2020) Circadian rhythm disruption and mental health. Transl Psychiatry 10:28 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Addington AM, Gallo JJ, Ford DE et al (2001) Epidemiology of unexplained fatigue and major depression in the community: the Baltimore ECA follow-up, 1981–1994. Psychol Med 31:1037–1044 [DOI] [PubMed] [Google Scholar]
- 28.Vassend O, Roysamb E, Nielsen CS et al (2018) Fatigue symptoms in relation to neuroticism, anxiety-depression, and musculoskeletal pain. A longitudinal twin study. PLoS ONE 13:e0198594 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The SLOSH data generated and analyzed during the current study is not publicly available due to ethical restrictions and considering that sensitive personal data are involved. Access to the data may be provided to other researchers in line with Swedish law and after consultation with the Stockholm University legal department. Requests for SLOSH data, stored at the Department of Psychology, Stockholm University should be sent to [data@slosh.se](mailto:data@slosh.se).