Circadian rhythm dysfunction and psychopathology in the offspring of parents with bipolar disorder: a high-risk study in the Chinese population

Participants

This is a case–control high-risk offspring study of bipolar disorder conducted in the Greater Bay Area (Hong Kong Special Administrative Region (SAR) and Guangdong) in China. Parents with bipolar disorder who served as case probands were recruited from psychiatric clinics, while parents without bipolar disorder who served as control probands were recruited from other research projects and communities. The inclusion criteria for case offspring were: (1) aged 6–21 years and (2) having at least one biological parent with bipolar disorder-I/II. The inclusion criteria for control offspring were: (1) aged 6–21 years, (2) having no biological parent with bipolar disorder-I/II and (3) age/sex matched with case offspring. The exclusion criteria for offspring participants were: (1) having any severe medical diseases or neuropsychiatric disorders, (2) being unable to accomplish the measurements or (3) being shift workers. All eligible offspring participants from the same family were recruited. Informed consent was obtained from participants. Assent was also obtained from the parents of offspring under 18 years old. Offspring participants were recruited from 11 February 2017 to 16 January 2020. Statistical analyses were conducted from March 2022 to July 2022.

Measurements and procedures

Probands and their spouses were invited to a clinical interview based on the Diagnostic and Statistical Manual of Mental Disorders15—Fourth Edition (DSM-IV)16 conducted by psychiatrists and psychologists. The families with at least one biological parent diagnosed with bipolar disorder were regarded as case families, while the families without any biological parent with bipolar disorder were regarded as control families (figure 1).

Request permission

The diagnoses and symptoms of sleep and circadian rhythm dysfunction of offspring were ascertained by face-to-face interviews using the Chinese Version of the Diagnostic Interview for Sleep Patterns and Disorders (DISP).17 The original DISP was developed according to the International Classification of Sleep Disorders, Second Edition criteria, while the DISP applied in the current study was further adapted to align with the International Classification of Sleep Disorders, Third Edition (ICSD-3).18 In addition, we have further adapted the items of DISP to meet the diagnostic criteria of sleep and circadian disorders among children and adolescents. Compared with the clinician diagnosis, DISP has classification accuracies of 0.63–0.96 and κ coefficients of 0.20–0.79.17 In this study, data were excluded if they were not collected in accordance with ICSD-3 criteria. In order to capture the emergence of psychopathologies in childhood and adolescence comprehensively, the diagnoses and symptoms of mental disorders among the offspring were assessed by the Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime Version (K-SADSPL), with high inter-rater agreements ranging from 0.93 to 1.00 and high κ coefficients of 0.63 to 1.00.19 Mental, sleep and circadian rhythm disorders were defined according to DSM-IV criteria and ICSD-3 criteria, respectively. In other words, the disorder would need to fulfil the symptom criteria, duration criteria and functional impairments. On the other hand, the study also elicited sleep/circadian dysfunction and psychopathologies at the symptom level. Participants answering in the affirmative on the symptom-related items were considered to have the specific symptoms in their lifetime. Mental symptoms were defined by the key symptoms of the clusters of related disorders, based on K-SADSPL according to DSM-IV. For instance, we defined mood symptoms by key symptoms of depression and bipolar disorder, while anxiety symptoms by key symptoms of panic disorder, agoraphobia, separation anxiety disorder, social anxiety, generalised anxiety disorder, specific phobia and obsessive-compulsive disorder. In terms of sleep and circadian dysfunction, the offspring with the key symptoms of delayed sleep phase disorder, advanced sleep phase disorder, insomnia disorder, restless legs syndrome and nightmares would be considered as suffering from the respective symptoms based on the DISP interview according to ICSD-3.

The uncertain responses in the clinical interviews were defined as missing data. When there was disagreement between the child and parent about the presence of symptoms, the parental report of observed behaviour and the offspring’s report of subjective symptoms were adopted. By combining the clinical interviews and medical records, the best estimate strategy was applied.

For the offspring, pubertal status was assessed by the self-reported Tanner Scale which has been validated locally.20 In addition, the Montgomery-Asberg Depression Rating Scale and the Young Mania Rating Scale were used to measure the severity of depressive and manic symptoms of probands, respectively. The Montgomery-Asberg Depression Rating Scale has a mean intraclass correlation of 0.95 and Cronbach’s α of 0.96,21 while the Young Mania Rating Scale has inter-rater reliability ranging from 0.66 to 0.95 and high validity, reaching up to 0.89.22 Demographic data, including age, sex and region of recruitment, were collected for all the participants.

Quality control and inter-rater reliability

All interviewers had at least 3 years of clinical experience in psychiatry. The training and standardisation of diagnostic interviews were conducted in a workshop prior to the collection of data. Each interviewer was supervised by experienced interviewers for at least five cases before he/she could independently conduct the clinical interviews. Kendall’s coefficient of concordance values ranged from 0.65 to 0.73, which suggested a high level of consistency among interviewers.

Sample size estimation

To achieve a power of 90%, a type I error of 0.05, setting the proportions of 52.1% and 29.1% in the two groups, referring to the largest offspring cohort of bipolar disorder,23 we estimated that the minimum sample size required for this study would be 199 offspring to detect significant differences between the two groups. The sample size was estimated by G*Power V.3.1.9.7.

Statistical analysis

The demographics of all the participants have been presented as the mean (SD) for continuous variables and percentages for categorical variables. Generalised estimating equations (GEEs) were applied to compare the demographics between O-BD and O-control. The prevalence of sleep/circadian rhythm dysfunction and psychopathologies was presented as percentages, and the differences between groups were calculated by GEE, logistic regression and Mann-Whitney U test where appropriate. Because of the data issue related to the inclusion of siblings from the same families, GEE was deemed to be the suitable statistical method for the main analysis.24 Only when the number of cases was insufficient for the use of GEE, logistic regression and Mann-Whitney U test would be used. Survival analysis was used to estimate the relationship between parental history of bipolar disorder and the emergence of sleep/circadian rhythm dysfunction and psychopathologies, including sleep and circadian rhythm disorders, mental disorders and their key symptoms. Two shared frailty proportional hazards models were performed with years from birth as the time-varying covariate. Model 1 is unadjusted. Age at recruitment, sex and region of recruitment were adjusted in model 2. Kaplan-Meier survival plots were used to depict the cumulative events. Subgroup analysis stratified by age was conducted. All the analyses were conducted with IBM SPSS Statistics for Windows, V.26.0 (IBM Corp) and the R project (V.3.6.1, R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-project.org/).

Mental disorders and symptoms in O-BD

As shown in table 1, O-BD had a higher lifetime prevalence of mental disorders and symptoms compared with their control counterparts. After adjustment for age, sex and region of recruitment, O-BD had a nearly twofold higher rate of any mental disorders (40.31% vs 26.24%, adjusted OR: 1.87, 95% CI 1.14 to 3.10), a nearly fivefold higher rate of mood disorders (11.70% vs 3.47%, adjusted OR: 4.68, 95% CI 1.30 to 16.86), a fourfold higher rate of depressive disorders (11.17% vs 3.47%, adjusted OR: 3.99, 95% CI 1.11 to 14.34) and a nearly fivefold higher rate of social anxiety (6.28% vs 1.49%, adjusted OR: 4.70, 95% CI 1.21 to 18.18), when compared with their counterparts. Besides, O-BD had a marginally insignificant 2.5 times higher rate of anxiety disorders (12.63% vs 4.95%, adjusted OR: 2.45, 95% CI 0.97 to 6.23). Likewise, there were significantly higher risks of mental symptoms among O-BD than their counterparts. O-BD had higher rates of any mental symptoms (60.73% vs 43.07%, adjusted OR: 2.14, 95% CI 1.37 to 3.33) and mood symptoms (20.74% vs 10.40%, adjusted OR: 2.59, 95% CI 1.24 to 5.44), when compared with O-control. In addition, O-BD had a marginally significant higher rate of behavioural symptoms (36.96% vs 27.86%, adjusted OR: 1.60, 95% CI 0.97 to 2.63). The two groups showed comparable rates of other mental disorders and symptoms.

We further divided the offspring into two subgroups: those under 12 years old (childhood) and those aged 12 years and over (adolescence). In childhood, O-BD had a 1.91-fold higher risk of any mental disorders when compared with O-control. Also, O-BD had nearly twofold higher risks of having any mental symptoms (58.25% vs 41.18%, adjusted OR: 2.12, 95% CI 1.21 to 3.72) and behavioural symptoms (48.54% vs 30.69%, adjusted OR: 2.13, 95% CI 1.14 to 3.95) than O-control after further adjustment. In adolescence, there were 88 O-BD (47.73% female) who were slightly older (15.94 (2.68) vs 14.32 (2.05), Wald χ²=21.72, p<0.001) and more mature in pubertal maturation (pre/early puberty: 29.89% vs 54.00%, Wald χ²=10.59, p=0.001) than the 100 O-control (52.00% female). The O-BD group had a 7.49-fold higher risk of social anxiety (11.36% vs 2.00%, adjusted OR: 7.49, 95% CI 1.33 to 42.20) compared with the O-control group. In addition, there were increased mood symptoms (36.05% vs 16.00%, adjusted OR: 3.00, 95% CI 1.14 to 7.92) in the O-BD group. Four O-BD adolescent subjects but none of the O-control reported substance use problems (z=–1.86, p=0.063) (table 2).

Sleep and circadian rhythm disorders and symptoms in O-BD

A total of 179 O-BD (age: 11.85 years (4.39); 45.51% female) and 194 O-control (age: 11.63 years (3.34); 49.74% female) were included in the analyses, with comparable pubertal maturation. Table 3 shows that O-BD had a significantly higher prevalence of delayed sleep phase disorder than O-control in the unadjusted model (5.62% vs 1.03%, OR: 5.71, 95% CI 1.21 to 27.04), but the significance was attenuated after adjustment (adjusted OR: 3.06, 95% CI 0.44 to 21.47). There was no difference in other sleep and circadian rhythm disorders including insomnia between the two groups. On the other hand, there was a significantly higher risk of delayed sleep phase symptoms (9.55% vs 2.58%, adjusted OR: 4.04, 95% CI 1.23 to 13.28) in O-BD than in the controls, which persisted after further adjustment.

In terms of age, there were different circadian patterns between children and adolescents. In childhood, O-BD was similar to O-control in both circadian disorder and symptom levels. When O-BD reached adolescence, they reported a higher risk of delayed sleep phase disorder (12.50% vs 2.02%, OR: 6.93, 95% CI 1.44 to 33.42) and delayed sleep phase symptoms (20.00% vs 5.05%, OR: 4.70, 95% CI 1.63 to 13.56) than the O-control group. After adjustment for age at recruitment, sex and region, the statistical significance of delayed sleep phase symptoms remained (adjusted OR: 3.83, 95% CI 1.11 to 13.16), but not at the disorder level. Interestingly, both groups had similar risks of insomnia (disorder and symptom level) in both childhood and adolescence (table 4).

Survival analyses

Survival analysis was used to estimate the longitudinal associations between the parental history of bipolar disorder and the emergence of psychopathologies and sleep/circadian rhythm dysfunction. The findings of shared frailty proportional hazards models indicated that the parental history of bipolar disorder was associated with increased risks of any mental disorders (adjusted HR: 1.83, 95% CI 1.15 to 2.92), mood disorders (adjusted HR: 14.97, 95% CI 2.68 to 83.67), depressive disorders (adjusted HR: 11.39, 95% CI 2.18 to 59.37) and social anxiety disorder (adjusted HR: 6.17, 95% CI 1.07 to 35.58), after adjusting for age at recruitment, sex and region. Besides, the parental history of bipolar disorder increased the lifetime prevalence of anxiety disorders, but the HR was attenuated to marginal significance after adjustment (HR: 3.21, 95% CI 1.12 to 9.19; adjusted HR: 2.88, 95% CI 0.99 to 8.32) (see online supplemental table 1 and figure 2).

Request permission

In terms of the emergence of symptoms, there was a marginally significant association of parental history of bipolar disorder with delayed sleep phase symptoms (adjusted HR: 4.97, 95% CI 0.97 to 25.45) but not delayed sleep phase disorder (adjusted HR: 6.74, 95% CI 0.71 to 64.15) (see online supplemental table 1 and figure 2).

Main findings

To our knowledge, this is the first case–control study of the offspring of parents with and without bipolar disorder, encompassing a comprehensive assessment of sleep and circadian rhythm disorders and symptoms, as well as the empirical evidence of mental disorders and symptoms in the Chinese population. We found further evidence for the specificity of familial patterns of mood disorders/symptoms and circadian rhythm dysfunction in O-BD compared with O-control. The age-related developmental pattern of progressive occurrence of psychopathologies was reflected in the differential occurrence of behavioural symptoms in O-BD under age 12 years and social anxiety disorder, mood symptoms and delayed sleep phase symptoms in O-BD adolescents aged 12 years and over.

While sleep alteration has been frequently reported in patients with bipolar disorder, only a few studies of sleep and circadian rhythm dysfunction suggested that O-BD have higher rates of sleep disorders and evening chronotype than controls.7 25 In this study, there was a higher risk of delayed sleep phase symptoms among O-BD which started to emerge in adolescence, whereas there was no increased risk of insomnia in both O-BD children and adolescents. Thus, it is probable that circadian rhythm dysfunction rather than insomnia may underlie the core sleep–wake pathologies among the high-risk O-BD population.11 Previous studies have indicated that eveningness and insomnia symptoms were independently associated with mood disorders and other psychopathologies.26 27 Hence, there may be an interesting postulation that delayed sleep phase symptoms could be the major prodrome of bipolar disorder, while insomnia might reflect subsequent effects of depression, circadian or other comorbidities.

Earlier findings in both patients with bipolar disorder and O-BD suggested that a broader dysregulation of circadian rhythm may underlie this condition.11 Our previous report of dysregulation of melatonin secretion further suggested that suppressed melatonin secretion might reflect genetic susceptibility conferring increased sensitivity of melatonin suppression to environmental light exposure.11 Excessive electronic media use, especially before bedtime (which may enhance light exposure at night), has been associated with sleep and circadian problems and poor mental health.28 Thus, appropriate guidelines and interventions for the appropriate use of electronic media are needed among adolescents, especially in high-risk adolescents with a family history of bipolar disorder as suggested by our study.

Another contribution of this study is the collection of information on symptoms and subthreshold manifestations of mental and sleep disorders beyond the diagnostic criteria. Our findings replicate the results of earlier bipolar disorder high-risk studies showing that O-BD tend to have an earlier onset of mood and anxiety disorders than the controls.29 Furthermore, the findings suggest that there is a developmental trajectory reflected by an increased risk of behavioural symptoms in childhood, followed by the emergence of social anxiety and mood symptoms in adolescence. The significant effects of symptoms rather than disorders for some conditions suggest that dimensional symptom scales might be a more sensitive indicator of the evolving risk of bipolar disorder, especially when most of the offspring are in their childhood and adolescence.30

The elevated rates of mood, anxiety and delayed sleep phase in O-BD youth are likely to reflect genetic susceptibility to bipolar disorder, which is known for its high heritability among major mental disorders.31 There is a convergence of evidence that bipolar disorder may reflect dysregulation of motor activity and sleep, with shifts to later day peak activity and later sleep onset time.32 Our compelling data on delayed sleep symptoms and disorders confirm these adult data and further demonstrate that these shifts may manifest as early as in adolescence. The emergence of behavioural, mood, anxiety and sleep/circadian disturbances in late childhood and adolescence could further interact with environmental exposure or adverse events,33 resulting in increased stress that will lead to greater impairment in functioning.34

Some discrepant results of our study are also noteworthy. First, the lack of differences in the rates of ADHD in the two risk groups is inconsistent with prior research that reported an elevated risk of ADHD in O-BD.23 Nevertheless, O-BD had an increased risk of behavioural symptoms. The findings suggest that behavioural dysfunction may emerge in childhood among high-risk offspring. Thus, clinicians should focus not only on the categorical ADHD diagnosis but also on the general behavioural symptoms. Interestingly, we found a substantially lower rate of substance use disorders than other similar studies in Canada and the USA.7 35 This might reflect the stringent drug-control policy in China and the relatively young age of the cohort (mean age of 12 years in the current study). Nevertheless, there was an emergence of substance use in the adolescent O-BD group, although the reported prevalence rate was still low.

The inclusion of a wide age range of high-risk subjects from children to adolescents with a comprehensive assessment of mental disorders as well as sleep/circadian assessment at both the symptom and disorder levels among the high-risk offspring allowed us to confirm and extend the preclinical staging model of bipolar disorder among O-BD, as exemplified by the Canadian High-Risk Offspring Cohort.7 In summary, behavioural symptoms start to occur in childhood followed by the emergence of a delayed sleep phase, social anxiety and mood symptoms in early adolescence. We predicted that major mood disorders may emerge in middle/late adolescence and early adulthood before the full-blown development of bipolar disorder. Prospective follow-up is necessary to elucidate the trajectories of these early manifestations of the bipolar spectrum and their transition to mood disorders in late adolescence and early adulthood.

Limitations

The current study has several limitations. First, despite being one of the largest studies of offspring of parents with and without bipolar disorder, the sample size for younger people with certain specific disorders, such as post-traumatic stress disorder, is still insufficient for multivariate data analysis. Second, the wide age range of offspring may introduce potential heterogeneity in psychopathologies and circadian rhythm assessment because of the age-specific development of these systems. Nevertheless, investigating the emergence of both symptoms and disorders across this developmental transition is critical for understanding the evolution and variations in phenotypical patterns associated with bipolar disorder. Hence, the subgroup analyses provided an opportunity to investigate the specific patterns in childhood and adolescence, respectively. Third, although there were some apparent deviations in the age appropriateness of the diagnostic assessments used in this study, K-SADSPL and DISP could still be regarded as the most appropriate diagnostic assessments in this study. K-SADSPL is commonly applied to determine the lifetime profiles of mental problems in the offspring aged between 6 and 18 years, but the study has extended the use of the instrument to late adolescent/early adult subjects aged between 18 and 21 years. In order to better understand lifetime diagnosis, other international bipolar offspring studies in Canada, the Netherlands and Australia have used similar K-SADS instruments for this group of older adolescents/young adults. Additionally, we have adapted the items of DISP according to ICSD-3 criteria to meet the diagnostic criteria of sleep and circadian disorders among children and adolescents. Moreover, as some offspring have not yet reached the typical onset age for the conditions studied, follow-up of this cohort will be essential to collecting information on the full trajectory of risk of psychopathologies and sleep/circadian rhythm dysfunction with regard to the emergence of bipolar disorder. Finally, there were missing data in the current study. The most common missing data included items related to suicide (3.56%) and nightmares (2.68%). As the proportion of missing data was relatively small (less than 5%), it was deemed that it would not lead to significant bias, and the statistical significance of the main analysis remained the same even after multiple imputations.

Implications

The findings of the current study have several important clinical implications for further understanding bipolar disorder. First, the study identified several clinical markers in the high-risk population, especially delayed sleep phase symptoms, which were first reported as the early core manifestation and marker of bipolar disorder. Second, the findings enhanced the understanding of the preclinical staging model of bipolar disorder. Hence, different early clinical intervention and prevention strategies could be developed according to the ages and stages of the high-risk offspring of bipolar disorder. Third, the strong familial association of bipolar disorder and circadian rhythm dysfunction in offspring may provide evidence for the novel prevention of bipolar disorder. Thus, family education and family-based therapy should be prioritised to reduce the risk of bipolar disorder, and our study further highlights the importance of chronotherapeutic management among high-risk offspring.11 In conclusion, the identification of novel prodromal markers and preclinical stages of bipolar offspring emphasises the importance of identification and age-specific precision prevention of bipolar disorder.