Longitudinal Course and Outcome of Chronic Insomnia in Hong Kong Chinese Children: A 5-Year Follow-Up Study of a Community-Based Cohort

Abstract

Objectives:

There are limited data on the long-term outcome of childhood insomnia. We explored the longitudinal course, predictors, and impact of childhood insomnia in a community-based cohort.

Design:

5-year prospective follow-up.

Setting:

Community-based.

Participants:

611 children (49% boys) aged 9.0 ± 1.8 years at baseline; 13.7 ± 1.8 years at follow-up.

Intervention:

NA.

Main Exposures:

Chronic insomnia was defined as difficulty initiating sleep, difficulty maintaining sleep and/or early morning awakening ≥ 3 times/week in the past 12 months.

Outcome Measures:

General health, upper airway inflammatory diseases, and behavioral problems in recent one year were assessed at both time points, while mental health and lifestyle practice were assessed at follow-up study. The questionnaires at baseline and follow-up were reported by parents/caretakers and adolescents themselves, respectively.

Results:

The prevalence of chronic insomnia was 4.2% and 6.6% for baseline and follow-up, respectively. The incidence and persistence rates of chronic insomnia were 6.2% and 14.9%, respectively. New incidence of insomnia was associated with lower paternal education level, baseline factors of frequent temper outbursts and daytime fatigue as well as alcohol use and poor mental health at follow-up. Baseline chronic medical disorders, frequent temper outbursts, and poor mental health at follow-up were associated with the persistence of insomnia in adolescents. Baseline insomnia was associated with frequent episodes of laryngopharyngitis and lifestyle practice (coffee and smoking) at follow-up.

Conclusions:

Chronic insomnia is a common problem with moderate persistent rate in children. The associations of adverse physical and mental health consequences with maladaptive lifestyle coping (smoking and alcohol) argue for rigorous intervention of childhood insomnia.

Citation:

Zhang J; Lam SP; Li SX; Li AM; Lai KYC; Wing YK. Longitudinal course and outcome of chronic insomnia in Hong Kong Chinese children: a 5-year follow-up study of a community-based cohort. SLEEP 2011;34 (10):1395-1402.

INTRODUCTION

Insomnia is commonly found in childhood general population with a prevalence rate of 4% to 9% (by restrictive diagnostic criteria of 3 times/week or often).^1–3 Most of the epidemiologic studies on childhood insomnia are, however, of cross-sectional design.^1–3 Among the few available prospective studies of the long-term course of insomnia in children and adolescents, insomnia or sleep disturbances tend to run a persistent course, at least for a proportion of individuals.^4–11 Nonetheless, most of these studies,^5–8,11,12 especially the earlier ones, used nonspecific terms such as “sleep disturbances” or “sleep problems” rather than operational definition of insomnia. Overall, these prospective studies revealed certain longitudinal patterns of childhood insomnia. First, the usage of stricter criteria leads to lower persistence and incidence rate of insomnia.⁴ Second, persistence of insomnia seems to be age-dependent. Persistence of sleep difficulty is not found between infants and young children¹² but becomes significant in older children and adolescents.⁶

The clinical significance of childhood insomnia is reflected by its correlation with medical conditions,^2,13 mental dysfunction, and poor academic performance.¹⁴ However, the cross-sectional nature of most studies could not disentangle the interactive relationship between insomnia and various dysfunctions.¹⁴ A few prospective studies suggest that insomnia or sleep problems at baseline might be associated with poor somatic health,¹³ substance use¹⁵ and behavioral problems in adolescents.^8,15,16

In brief, previous studies have found that childhood insomnia could be a persistent sleep problem with adverse health consequences. However, several aspects are still unclear. First, all longitudinal studies were conducted in western countries. Little is known about the long-term course of insomnia in non-western populations. Second, adolescents experience significant developmental changes both physically and psychologically,¹⁷ but little is known about the changes of insomnia symptoms of children when they are entering adolescence. Finally, the long-term association of insomnia with physical and mental problems is still inadequately investigated. In this community-based cohort, we aimed to: (1) determine the longitudinal course of chronic insomnia and its correlates in Hong Kong Chinese children; (2) examine the long-term association of childhood chronic insomnia with physical and mental health in adolescents.

METHOD

Subject Selection

The study was conducted and reported as according to the recommendations of the STROBE guideline.¹⁸ This study is part of an ongoing epidemiologic study about sleep problems among Hong Kong Chinese children and their parents, which started since 2003–2004 (baseline).^2,19,20 Follow-up was conducted during the period of 2008–2010. The protocols of this study at both baseline and follow-up were approved by the institutional ethics review committee. Parent (s) or caregiver (s) of the subjects gave written consents to participate this study, and the adolescents gave written assents to the study. Figure 1 delineates the recruitment of the subjects. A total of 6447 children were recruited at baseline. Among 5872 families who left their telephone numbers, a total of 3416 were contactable at the time of follow-up. While 55 of them refused to participate, 3361 (98.4%) gave verbal consents and agreed to complete the study questionnaires. The package of questionnaires for adolescents, fathers and mothers as well as their sibling (s) aged > 6 years old (if any) was mailed to the address provided. The data about the adolescents was analyzed in the current study. The families were reminded at 2–4 week intervals about returning the questionnaires. If the questionnaires were not returned after 2 months, a reminder phone call was made again to the families. If the questionnaires were not returned after another 2 months, a final reminder letter was sent. Finally, a total of 1611 adolescents out of 3416 contactable children (47.2%) returned their packages of questionnaires at the follow-up study.

Figure 1.

Flow diagram in the recruitments of subjects

Measurements

The measures were mainly reported by parents at baseline but were answered by adolescents themselves at follow-up. In both studies, the informants (parents or adolescents) were encouraged to talk with relevant family members while completing the questionnaires. At baseline study, the parents/caretakers were chosen as informants, as there was information that would need the input from the parents, for example, presence of specific medical disorders of their young children. Furthermore, the rate of insomnia in children under 12 years old as reported by parents would be more precise when evaluating a longer term course of insomnia, such as the current study about sleep problems over 12 months.^9,21 At follow-up study, the subjects were asked to complete the questionnaires themselves, as we believed that the adolescents would be able to give a more reliable account of their own sleep and medical problems, especially over some sensitive questions on lifestyle habits (such as smoking and drinking).

Environmental and Sociodemographic Risk Factors

Assessment of socioeconomic status included several binary or ordinal questions,² such as parental educational level (< 12 years or ≥ 12 years), parental employment (employed vs. unemployed), housing type (private vs. public), parental marital status (married or cohabited vs. devoiced or separated), and family income (≤ HK$15,000/month vs. more than HK$15,000/month; HK$7.8 = US$1).

Sleep Questionnaire for Children

Our baseline questionnaire consisted of 54 items on demographics, sleep environment, sleep habits and problems, and family information.^2,19,20 In brief, there were 25 items about childhood sleep problems. The psychometric property of the questionnaire was found to be satisfactory, with Cronbach α coefficient of 0.868 and 5-factor structure corresponding to major childhood sleep problems by factor analysis.² The questionnaire in the follow-up study was slightly modified, but the items about insomnia were the same as previous one. Difficulty initiating sleep (DIS), difficulty maintaining sleep (DMS), and early morning awakening (EMA) were assessed by the following question: “during the past 12 months, how often has your child (or “you,” in follow-up questionnaire) had: (1) difficulty initiating sleep? (2) sudden awakening during sleep and difficulty in returning to sleep? (3) early morning awakening, and could not fall asleep again?” respectively. Subjects were instructed to respond on a 5-point Likert scale (0 = never, 1 = less than once per month, 2 = 1–3 times per month, 3 = 1–2 times per week, 4 = ≥ 3 times per week). Subjects reporting any subtype of insomnia ≥ 3 times/week over the past 12 months were defined as suffering from chronic insomnia. As the major aim of our study was to explore the long-term course and consequences of sleep problems including chronic insomnia, a period of 12 months was chosen. The report of sleep symptoms over the past year could allow the estimation of insomnia problem at the conservative end.

Daytime Symptoms

Different daytime symptoms in the past 12 months, including difficulty getting up in the morning, feeling unrefreshed in the morning, morning headache, and daytime fatigue were also asked.²² We did not incorporate these daytime symptoms to define insomnia disorder,^23–26 but instead we treated these daytime symptoms as correlates of insomnia symptoms in the current study. Studying insomnia symptoms rather than syndrome (symptoms plus impairment) would allow us to estimate the repercussion of insomnia at the conservative end.

General Health, Chronic Medical Diseases, and Upper Airway Inflammatory Diseases (Baseline and Follow-Up)

To measure subjects' general health, the following questions were asked: “how is your child's (your) general health condition in recent one year?” The responses could be: (1 = very poor; 2 = poor; 3 = fair; 4 = good; 5 = very good. Those with responses of 1 to 2 and 3 to 5 were grouped together. “Does (Do) your child (you) need to take medication regularly?” The responses could be: “yes” or “no.” They were also asked whether they had any chronic medical disorders that were diagnosed and required treatment (altogether 14 disorders, including hypertension, eye diseases, hypercholesterolemia, arthritis, epilepsy, cardiovascular disorders, diabetes mellitus, lung disorders, psychiatric/mood disorders, renal disorders, chronic pain, eczema, gastroesophageal reflux disease [GERD], and others). However, due to limited number of cases for each disorder, they were not analyzed individually.

The participants were also asked whether they had the following 5 common upper airway inflammatory diseases in recent one year: allergic rhinitis, nasosinusitis, asthma, adeno-tonsillitis, and laryngopharyngitis. The responses could be: 1 = never; 2 = sometimes; 3 = frequently). Those with “frequent” upper airway diseases were considered abnormal.

Behavioral Problems and Academic Performance during the Past 12 Months (Baseline and Follow-Up)

Hyperactivity and frequent temper outbursts of children (adolescents) were rated as “yes” or “no.” Academic performance was rated as poor if the subjects rated “very worst or worst than past one year.”

Mental Health Assessment during Past One Month (Follow-Up Only)

The Chinese version of General Health Questionnaire (GHQ-12) was employed to evaluate mental health status of the adolescents at follow-up. A GHQ-12 score ≥ 4 was considered poor mental health.²⁷

Lifestyle Assessments during Past 12 Months (Follow-Up Only)

Four types of lifestyle habits, including regular tea, coffee, alcohol, and smoking were asked at follow-up. All of the questions about lifestyle were rated as 0 = never/seldom or 1 = sometimes/often.

Outcomes, Exposures, and Covariates

All subjects were divided into 2 groups on the basis of their insomniac status at baseline. Those subjects without chronic insomnia at baseline were analyzed for the incidence rate of insomnia at follow-up, while those subjects with insomnia at baseline would be analyzed for the persistent rate of insomnia at follow-up. The exposures (risk factors) for the course of insomnia were based on the baseline information, including sociodemographics, behavioral problems, and medical conditions. Another series of major outcomes included general health condition, 5 common upper airway inflammatory diseases, behavioral problems, and lifestyle practice at follow-up. Their exposures were the courses of chronic insomnia (i.e., baseline, incident, and persistent insomnia). The covariates for these major outcomes included age, gender, parental educational level, and family income.

Statistics

Descriptive statistics were presented as percentages for discrete variables and as means (standard deviation) for continuous variables. The differences in sample characteristics were tested by independent t-test or χ² to explore the potential sample attrition between recruited and drop-out subjects whenever appropriate.

Long-Term Course of Chronic Insomnia and Its Predictors

Chi-square statistic was used to compare the prevalence of chronic insomnia at baseline and that of follow-up in the study cohort. As the response rate (47.2%) was less than ideal, the prevalence, persistence, and incidence rate of insomnia at follow-up were further weighted for sociodemographic variables (parental education level, family income and housing type) which differed between those recruited and drop-out subjects. The incidence and persistence rate of chronic insomnia and their predictors were analyzed. The risk factors were extracted from baseline data. Logistic regression (enter method) analysis was used to determine the association between chronic insomnia (incidence and persistence) and risk factors after adjusting for age and gender. Those independent variables found to be significantly associated with the progress of chronic insomnia were further tested by multivariate logistic regression with forward likelihood method.

Consequences of Chronic Insomnia

Chi-square statistic or Fisher exact test was used to explore the potential association between the progress of chronic insomnia and various consequences. Multiple logistic regression (enter method) was used to examine the strength of these associations after controlling for age, gender, parental education level, family income, and corresponding consequences at baseline. For example, the strength of association between hyperactivity at follow-up and chronic insomnia was explored after controlling for age, gender, parental education level, family income, and hyperactivity at baseline. As GHQ and lifestyle practice were only assessed at follow-up, these dependent variables were only adjusted for age, gender, parental education level and family income. P-values < 0.05 were considered statistically significant. SPSS 16.0 for Windows (SPSS Inc, Chicago, IL) was used for all statistical tests.

RESULTS

Sample Characteristics

Table 1 delineates the characteristics of subjects at baseline and follow-up. The mean age of study subjects (n = 1611) was 9.0 ± 1.8 years old at baseline and 13.7 ± 1.8 at follow-up, with a mean follow-up duration of 4.7 years. Those children who participated in the follow-up study had slightly higher socioeconomic status (higher parental education, housing type, and family income, P < 0.05) and were slightly younger (9.0 ± 1.8 vs. 9.2 ± 1.8 years, P < 0.05), but had similar prevalence of chronic insomnia when compared to the dropouts. Similar boy/girl ratios were found between those recruited (boys 49.1%) and drop-out subjects (50.6%) (P > 0.05).

Table 1.

Sample characteristics of subjects at baseline and follow-up

	Baseline sample N = 6447	Baseline cohort N = 1611	Follow-up cohort N = 1611
Children's age (years)*, mean ± SD	9.2 ± 1.8	9.0 ± 1.8	13.7 ± 1.8
Gender, male %	50.6	49.1	49.1
Insomnia rate, %	4.0	4.2	6.6
Paternal occupation (employed),%	96.2	97.5	97.3
Paternal educational level, tertiary or above, %*	15.8	19.3	19.3
Maternal occupation (employed), %	41.4	40.8	50.2
Maternal educational level, tertiary or above, %*	10.0	12.7	12.7
Housing type (public), %*	59.9	54.2	53.9
Family income > HK$15,000/month, %*	46.9	56.4	67.5
Parental marital status (married or cohabited), %	92.5	93.3	91.2

Wave 1 sample (W1 sample) included all recruited subjects at baseline; Wave 1 cohort (W1 cohort) and Wave 2 cohort (W2 cohort) only included recruited subjects at both baseline and follow-up study.

^*P < 0.05 between recruited and drop-out subjects at follow-up. HK $7.8 = US $1.

Course of Chronic Insomnia and Its Predictors

Table 2 reports the 5-year progress of chronic insomnia in our study cohort. The prevalence of chronic insomnia was slightly higher at follow-up than that of baseline (6.6% vs 4.2%, P = 0.001). The incidence and persistence rates of chronic insomnia were 6.2% and 14.9%, respectively. DIS had the highest persistence rate (19.0%), followed by DMS (5.3%). Subjects with EMA at baseline all remitted at follow-up, which indicated that EMA was a rather unstable condition in children. When using a looser criterion of sleep symptoms (≥ once per week), the incidence and persistence rates of chronic insomnia increased to 18.0% and 27.4%, respectively. Similar results were found upon further weighted analysis with adjustment to sociodemographic variables.

Table 2.

Insomnia symptoms of the study cohort at baseline and follow-up

Children	Prevalence Rates			Cohort Change
	Baseline sample N = 6447	Baseline cohort N = 1611	Follow-up cohort N = 1611	Incidence	Persistence	Remission
DIS (%)	2.5	2.6	5.6 (5.5)	5.2 (5.2)	19.0 (17.8)	81.0 (82.2)
DMS (%)	1.3	1.2	0.8 (0.7)	0.7 (0.6)	5.3 (5.0)	94.7 (95.0)
EMA (%)	1.2	0.9	1.4 (1.5)	1.4 (1.5)	0 (0)	100 (100)
Overall insomnia (%)	4.0	4.2	6.6 (6.6)	6.2 (6.2)	14.9 (14.1)	85.1 (85.9)
Insomnia ≥ 1 time/week (%)	11.1	12.9	19.2 (19.4)	18.0 (18.1)	27.4 (28.1)	72.6 (71.9)

DIS, Difficulty initiating sleep; DMS, Difficulty maintaining sleep; EMA, Early morning awakening; Overall insomnia was defined as any subtype of insomnia ≥ 3 times/week. Values in the bracket were weighted for paternal education level, maternal education level, family income, and housing type.

Table 3 presents the factors associated with the incidence and persistence rate of insomnia. Several baseline risk factors were associated with new incidence of insomnia in adolescents, including lower paternal education level (OR[95%CI] = 2.49[1.13-5.51]), frequent temper outbursts (OR[ (95%CI] = 1.85[1.16-2.97]) and daytime fatigue (OR[95%CI] = 2.17 ([1.24-3.77]). Other daytime symptoms, including difficulty getting up, dry mouth, morning headache, and feeling unrefreshed in the morning, were not associated with either persistence or incidence of insomnia in the final models. Chronic medical conditions (OR[95%CI] = 10.2[1.99-52.54]) at baseline were significantly associated with persistence of chronic insomnia in adolescents. Upper airway inflammatory diseases and parent-reported poor health condition at baseline could not predict the course of chronic insomnia (data not shown, P > 0.05).

Table 3.

Baseline factors associated with the progress of insomnia in the study cohort

Baseline Information	Incidence vs Non-Insomnia		Persistence vs Remission
	Crude OR (95%CI)	Adjusted OR† (95%CI)	Crude OR (95%CI)	Adjusted OR† (95%CI)
Gender, female vs male	1.09 (0.72–1.65)	—	0.51 (0.12–2.20)	—
Family income (monthly)
≤ 15,000 vs > 15,000 (HK$)	1.00 (0.66–1.54)	—	1.61 (0.67–3.89)	—
Paternal education
Non-tertiary level vs tertiary level	2.56 (1.24–5.41*	2.49 (1.13–5.51)*	0.63 (0.10–4.05)	—
Maternal education
Non-tertiary level vs tertiary level	1.57 (0.75–3.31)	—	3.50 (0.58–21.08)	—
Children's Information
Hyperactivity	1.89 (1.15–3.11)	—	0.83 (0.17–4.21)	—
Frequent temper outbursts	1.81 (1.17–2.80)*	1.85 (1.16–2.97)*	3.45 (0.81–14.74)	—
Chronic medical disorders	1.42 (0.83–2.41)	—	6.53 (1.49–28.23)*	10.2 (1.99–52.5)*
Feeling tired during daytime, ≥ 3 times/week	2.18 (1.25–3.79)*	2.17 (1.24–3.77)*	1.90 (0.49–7.37)	—
Feeling unrefreshed after waking up, ≥ 3 times/week	1.15 (0.41–2.24)	—	4.56 (0.97–21.43)	—
Feeling headache after waking up, ≥ 3 times/week	0.99 (0.99–1.00)	—	2.60 (0.43–15.7	—
Difficulty getting up in the morning, ≥ 3 times/week	1.84 (1.18–2.85)*	—	2.58 (0.61–10.90)	—
Moring dry mouth, ≥ 3 times/week	1.01 (0.36–2.83)	—	1.97 (0.43–8.97)	—

^†Adjusted for age, gender, parental education, and family income. — Without significance in adjusted model.

^*P < 0.05.

Association of Medical, Behavioral, Mental, and Lifestyle Problems at 5-Year Follow-Up with Chronic Insomnia

Table 4 shows the association between the longitudinal course of chronic insomnia and various outcome measures (i.e., general health, upper airway inflammatory diseases, mental health, behavioral problems, and lifestyle practice) at follow-up. Baseline insomnia was associated with frequent episodes of laryngopharyngitis and lifestyle practice (coffee drinking and smoking) but not mental health and behavioral problems at 5-year follow-up. Although baseline insomnia was associated with poor health condition and chronic medical conditions at follow-up initially, these associations did not persist after controlling for the potential confounding factors and corresponding problems. New incidence of insomnia was associated with poor health condition, frequent episodes of asthma and laryngopharyngitis, frequent alcohol and smoking, behavioral problems, and poor academic performance at follow-up (P < 0.05). Persistence of chronic insomnia was associated with presence of chronic medical disorders at baseline and poor mental health, frequent temper outbursts, and coffee drinking at follow-up. Poor mental health as defined by GHQ-12 score ≥ 4 was associated with incidence and persistence of chronic insomnia, with OR (95%CI) of 4.11 (2.51-6.74) and 5.0 (1.09-22.1) respectively, but not baseline chronic insomnia.

Table 4.

Association between the longitudinal course of insomnia and major outcome measures in the study cohort

	Baseline Insomnia vs Non-Insomnia		Incidence vs Non-Insomnia		Persistence vs Remission
	Crude OR (95%CI)	Adjusted OR† (95%CI)	Crude OR (95%CI)	Adjusted OR† (95%CI)	Crude OR (95%CI)	Adjusted OR† (95%CI)
Overall Health
Poor health condition (5.3%)	2.55 (1.18–5.52)*	–	2.65 (1.35–5.19**)	2.37 (1.12–5.01)*	2.04 (0.35–11.9)	–
Chronic use of medication (4.7%)	2.53 (1.12–5.75)*	–	1.20 (0.47–3.05)	–	2.50 (0.41–15.1)	–
Upper Airway Diseases
Frequent allergic rhinitis (22.8%)	1.36 (0.79–2.34)	–	1.68 (1.08–2.62)*	–	1.78 (0.44–7.19)	–
Frequent nasosinusitis (0.2%)	—	–	7.55 (0.68–84.0)	–	—	—
Frequent asthma (0.6%)	6.76 (1.38–33.2)*	–	6.09 (1.17–31.8)*	14.5 (2.02–104.2)**	—	—
Frequent adeno-tonsillitis (0.5%)	3.33 (0.40–27.4)	–	6.09 (1.17–31.8)*	–	—	—
Frequent laryngopharyngitis (1.8%)	3.86 (1.30–11.4)**	3.41 (1.11–10.57)**	4.98 (1.94–12.8)**	4.25 (1.50–12.0)**	23.1 (2.11–254)**	–
Behavioral Problems
Hyperactivity (8.5%)	0.86 (0.34–2.18)	–	3.59 (2.15–5.99)**	3.63 (2.09–6.32)**	1.42 (0.14–14.2)	–
Frequent temper outburst (22.5%)	1.61 (0.95–2.74)	–	2.45 (1.60–3.76)*	2.27 (1.42–3.63)**	12.9 (2.43–68.6)*	15.8 (2.25–111.6)**
Poor academic performance (12.3%)	1.25 (0.67–2.33)	–	2.81 (1.80–4.40)*	2.71 (1.68–4.37)**	—	—
Mental Health Condition
GHQ–12 score ≥ 4 (12.5%)	1.32 (0.66–2.63)	–	3.86 (2.44–6.12)**	4.11 (2.51–6.74)**	5.33 (1.16–24.5)*	5.00 (1.09–22.1)*
LIfestyle Practice
Tea drinking ≥ 3 times/week (43.9%)	1.11 (0.68–1.81)	–	1.35 (0.89–2.04)	–	1.12 (0.29–4.29)	–
Coffee drinking ≥ 3 times/week (11.1%)	2.43 (1.34–4.41)**	5.04 (1.63–15.6)**	1.47 (0.81–2.65)	–	7.67 (1.80–32.8)**	41.1 (3.68–460)*
Alcohol ≥ 3 times/week (2.4%)	1.25 (0.30–5.31)	–	3.69 (1.58–8.63)**	2.77 (1.02–7.57*	6.00 (0.34–104)	–
Smoking ≥ 3 times/week (1.1%)	5.12 (1.44–18.3)**	4.24 (1.14–15.8)*	4.18 (1.15–15.2)**	3.97 (1.04–15.2)*	2.94 (0.24–2.41)	–

^†Adjusted for age, gender, parental education, family income, and parent-reported poor health condition at baseline.

^*P < 0.05;

^**P < 0.01. — Could not be analyzed due to a lack of cases; – Without statistical significance in adjusted model. The percentages in the bracket of outcomes were rates of positive cases of outcomes in overall sample.

DISCUSSION

Longitudinal Course of Chronic Insomnia from Childhood to Adolescence

Although insomnia has been generally suggested to be chronic and persistent in adults, limited information is known about the long-term course of insomnia in children and adolescents. Our study found that the prevalence of chronic insomnia was 4.2% and 6.6% at baseline and 5-year follow-up respectively. The lack of gender difference in the prevalence of insomnia in this study was consistent with our baseline data³ and a previous adolescent study.⁴ However, it differed sharply from adult insomnia, for which there is a definite female predisposition.^28,29 Perhaps, the young age of our adolescent cohort could not capture the emerging gender difference of insomnia in the older adolescents and young adults.^3,28,29

The persistence rate of chronic insomnia symptoms was 14.9%, which was slightly lower than the figures (21% to 60%) reported in previous studies.^4–10 The slight disparity of the persistence rate between ours and other studies might be related to different study designs. The longitudinal follow-up period (nearly 5 years) in our study was longer than previous studies, which usually involved 1-2 years of follow-up.^4–9,12 The second reason might be related to the use of relatively more strict criteria in defining insomnia in our study (a frequency ≥ 3 times/week over past 12 months). When using a looser criterion (i.e., once a week), the persistence rate of chronic insomnia in the current study rose accordingly (from 14.9% to 27.4%). Another two studies similarly suggested that stricter criteria, for example, insomnia symptom (s) plus daytime fatigue or sleepiness, decreased the persistence rate of insomnia from 45.8% to 22.8% and from 52% to 21%, respectively.^4,6 As the insomnia symptoms were assessed over the past 12 months, the current study could not differentiate whether using a looser criterion with shorter period, such as the past month, would lead to higher prevalence, incidence, and persistence of insomnia. Further studies are warranted to address the potential disparity in the prevalence and course of insomnia across different criteria of insomnia.

Our logistic regression showed that lower paternal education level, frequent temper outbursts, and daytime fatigue in children would predict the new incidence of insomnia during adolescence. In other words, our data suggest that new onset of insomnia may be related to baseline sociodemographic characteristics of the family and daytime functional impairments (fatigue and temper outbursts).

Course of Chronic Insomnia, Physical Health, and Upper Airway Diseases

Previous cross-sectional studies suggested that insomnia represents a common symptom of physical disorders and is generally regarded as the distress or consequences of the medical conditions.^30–33 However, our study indicates a reciprocal relationship between insomnia and medical conditions. We found that presence of insomnia was associated with poor perceived health and frequent episodes of asthma and laryngopharyngitis over the course of 5 years, while the presence of chronic medical conditions could predict the persistence of insomnia. In other words, there is an interactive relationship between insomnia and physical health. While chronic medical disorders predisposed and perpetuated chronic insomnia, some medical conditions were also the outcomes of insomnia.¹³

In addition, insomniac adolescents had increased predisposition towards frequent upper airway inflammatory diseases. In particular, both baseline and new incidence of insomnia could predict frequent attacks of laryngopharyngitis, while new incident insomnia was associated with frequent asthmatic attacks in adolescents. The temporal association between chronic insomnia and upper airway inflammatory diseases was consistent with findings of the middle-aged adults cohort.³⁴ In other words, the predisposition to upper airway inflammatory diseases suggested that alteration of immune functioning is a possible distinct consequence of insomnia. It has been increasingly recognized that sleep plays a vital role in immune functioning.³⁵ Subjects with primary insomnia were reported to have lower levels of CD3+, CD4+, CD8+ cells, and natural killer cell responses than good sleepers.^36,37 A recent experimental study found that subjects with poor sleep over past 2 weeks were 5.5 times more likely to develop a common cold than good sleepers when they both received a nasal drop containing rhinovirus.³⁸ Thus, immune dysfunction might play a critical role in the relationship between chronic insomnia and upper airway inflammatory diseases.

Association between the Course of Chronic Insomnia and Behavioral, Mental, and Lifestyle Problems

Our baseline cross-sectional study found that both hyperactivity and frequent temper outbursts were correlated with chronic insomnia.² In the current follow-up study, the presence of frequent temper outbursts at baseline could predict new incidence of insomnia (Table 3), while incidence of insomnia, but not baseline insomnia, was associated with these two behavioral problems at follow-up. Previous studies, however, suggested that sleep problems were a risk factor for future behavioral problems.^8,15,16 A possible reason might be that these studies measured sleep problems only at baseline and treated sleep problems as exposures and behavioral problems as outcomes. Hence, they could not analyze the effects of behavioral problems on the development of chronic insomnia. On the contrary, our study, by measuring both chronic insomnia and behavioral problems at baseline and follow-up, could allow us to investigate the reciprocal association between chronic insomnia and behavioral problems. These results, thus, suggest that the associations between behavioral problems and chronic insomnia are likely complex and reciprocal. Nonetheless, further studies with more detailed behavioral assessments are warranted to disentangle the causal relationship between chronic insomnia and behavioral problems in adolescents.

Although most studies suggest that chronic insomnia is an important risk factor for future mental disorders in adults,^26,39 little is known about the longitudinal association between chronic insomnia and mental disorders in children and adolescents.⁴⁰ In a cross-sectional study relying on retrospective recall for the presence of lifetime insomnia and psychiatric disorders, Johnson et al. found that prior insomnia was significantly associated with the onset of depression but not anxiety in adolescents.⁴¹ By employing a six-wave survey, Buysse et al. reported that pure insomnia (without depression) was prospectively inter-correlated with insomnia comorbid with depression (insomnia + depression) but not pure depression (without insomnia) in young adults.¹⁰ Altogether, these studies suggest that insomnia might predict depression in adolescents, especially those comorbid with insomnia. Our study found that poor mental health (as measured by GHQ-12) at follow-up was associated with the persistence and incidence of chronic insomnia but not baseline chronic insomnia. This might suggest that there is only state association between insomnia and poor mental health in adolescents. Nonetheless, further prospective study with structured interview for psychiatric disorders will be needed to delineate the complex interactive relationship between chronic insomnia and mental disorders in our cohort.

The association between insomnia and substance misuse has been increasingly recognized in adolescents.⁴² A previous study found that smoking was a risk factor for future insomnia in adolescents.⁶ On the other hand, our study found that childhood insomnia predisposed to adolescents' smoking behavior 5 years later, which was consistent with a recent finding in children having a strong family history of smoking.¹⁵ Thus, there was likely a reciprocal association between insomnia and smoking behaviors in adolescents. In addition, new incidence of insomnia was associated with enhanced risk of alcohol use. In other words, smoking and coffee consumption might be the two major methods being employed by adolescents in coping with daytime consequences (such as fatigue) of insomnia, while new onset insomniac adolescents might consume alcohol to combat their sleep problems.^6,13

Strengths and Limitations

The merit of this study builds on the longitudinal design in a large community-based cohort. Secondly, a series of medical and mental health, behavioral consequences, and lifestyle habits were comprehensively investigated. Some of these outcomes were assessed at both baseline and follow-up, which could allow us to explore the reciprocal relationship between chronic insomnia and various dysfunctions. However, there were several limitations. Firstly, the current study was based on self-reported data without further assessment by structured interview and clinical examination. In this regard, some of the potential comorbid sleep disorders such as OSA and PLMS could not be completely excluded.^43,44 Secondly, parents were the key informants at baseline, which might lead to bias, as the rates of insomnia were slightly higher when reported by children or adolescents themselves than by their parents or teachers.^9,21 The higher rate of insomnia at follow-up than baseline in our study might be related to the aging effect but the potential reporters' bias (parents vs adolescents) could not be completely excluded. Although the adolescents were encouraged to talk with relevant family members for higher accuracy of their reports (for example, medical disorders), it remains unclear who did or did not consult their family members. Thirdly, insomnia has been found to have relatively high remission (around 40% in a period of 1 year) and relapse rates (e.g., around 30% in adults).⁴⁵ Our study might miss those insomniac subjects with waxing and waning course within the 5-year period. Fourthly, in view of the balance between enhancement of valid response and that of over-detailing of the questionnaire in a large epidemiological survey, a parsimonious choice of single item rather than detailed questionnaires (for example, child behavioral checklist) to assess the perceived health and behavioral problems was chosen at both baseline and follow-up. Fifthly, despite our intensive tracking of the subjects, the response rate was rather suboptimal. Nonetheless, the sample attrition in this study might not have prominently altered the outcome, as similar results in the course of chronic insomnia were found with a weighted analysis.

Clinical Implications and Conclusion

To our knowledge, this is the first study investigating the long-term course of childhood insomnia and its long-term association with mental, behavioral, and physical problems in Chinese childhood general population. Several findings are worth noting in our study. First, there was a moderate persistent rate of insomnia in Chinese children over the course of five years. Second, the findings of risk factors accounting for the persistence and incidence of chronic insomnia have important implications for the prevention and treatment of insomnia. Third, the temporal relationship between insomnia and their correlates shed light on their potential complex relationship. Finally, the pervasive impairments in terms of medical, mental, behavioral, and maladaptive lifestyle at 5-year follow-up argue for rigorous intervention of childhood insomnia.

DISCLOSURE STATEMENT

This was not an industry supported study. The authors have indicated no financial conflicts of interest.