Point Prevalence and Risk Factors for Insomnia in Children and Adolescents

Abstract

Background

Insomnia in children and adolescents can be associated with poorer cognitive, emotional, social, and academic development. At present, no estimates of the prevalence of insomnia and its risk factors among adolescents in Germany are available from which the potential need for treatment could be assessed.

Methods

We conducted an online survey of a representative German sample of 1128 children and adolescents (age 10–17) and one parent for each. Levels of severity of insomnia as defined by the ICD-11 criteria were assessed by means of standardized self-reporting with use of the internationally established Insomnia Severity Index. The point prevalences of the levels of severity were calculated via relative frequencies. Potential risk factors for insomnia (sociodemographic factors, obesity, media consumption time, depression, anxiety, parental insomnia) were assessed with validated screening questionnaires and investigated in a multinomial regression model for the prediction of insomnia in childhood.

Results

The following point prevalences were determined: mild insomnia, 26.6; moderate insomnia, 21.4; severe insomnia, 1.6. The most important risk factors for moderate and severe insomnia were existing anxiety (odds ratio and 95 confidence interval 4.54 [2.09; 9.88] and 7.96 [1.72; 36.94], respectively) and parental insomnia (2.49 [1.66; 3.72] and 3.30 [1.06; 10.30], respectively). The most important risk factor for mild and moderate insomnia was depression (1.83 [1.49; 2.24]), while older age (adolescents versus 10– to 13-year-olds) was protective ([0.51; 1.00]).

Conclusion

Many children and adolescents meet the ICD-11 criteria for insomnia according to their self-assessment. Critical life events and stressful experiences were not found to have any significant association with insomnia. A primarily non-pharmacological treatment approach involving the child and parents is indicated to alleviate the, often considerable, psychological strain on the family and prevent chronification of insomnia with adverse effects on the development of the child.

The impact of disturbed nighttime sleep on cognitive, psychological and physiological processes has been the subject of much debate: According to studies, reduced sleep duration and quality are associated with reduced attention (1), aggression inhibition (2) and emotional reactivity (3), as well as increased peripheral insulin resistance, elevated fasting fatty acid and cortisol levels (e1), and dysregulated inflammatory processes (4).

The specific spectrum of sleep problems seen in children and adolescents is due to the developmental psychological, (neuro-)biological, and social peculiarities of this age group. In the young age group, these are associated with externalizing and internalizing behavioral problems (5), learning difficulties (6), school refusal behavior (7), and obesity. A meta-analysis of cohort studies identified disturbed sleep as a risk factor for depression in young people (8). At the same time, a high level of media consumption is associated with poor sleep quality (9). In addition, sleep disorders in children have a negative impact on parental sleep quality (e2). Differences in sleep behavior between the sexes are observed from the onset of puberty onwards as the result of the accompanying hormonal changes (10).

The often synonymous use of the terms “sleep problems”, “sleep disorders” and “insomnia” in the literature without standardized and differentiated data collection impairs the comparability and generalizability of findings (e3). Unlike other sleep–wake disorders and subclinical sleep problems, insomnia is characterized by ongoing impairments in everyday life as the result of subjective problems falling and staying asleep (e4). Table 1 provides an overview of the diagnostic criteria for chronic (7A00) and acute insomnia (7A01) in the most recent edition of the International Statistical Classification of Diseases and Related Health Problems (ICD-11). They are listed in the chapter of sleep–wake disorders and follow the structure and terminology of the third edition of the International Classification of Sleep disorders (ICSD-3) (e4, e5). The previous distinction between psychogenic (nonorganic) insomnia and (organic) insomnia due to a neurological condition of the present ICD-10 is thus no longer applied due to its lack of discriminatory power (e5). According to the German S3-level clinical practice guideline “Non-Restorative Sleep/Sleep Disorders“ (11), the diagnosis is usually established based on clinical findings and a stepped diagnostic procedure. It should include a comprehensive medical history, investigation of physical or mental illnesses and a physical examination as well as an evaluation of self-observations and self-reported information obtained from sleep questionnaires and sleep diaries. The use of the Insomnia Severity Index (ISI) is recommended for screening purposes and for the documentation of treatment effects. This questionnaire is designed to assess the level of insomnia severity (11, 12).

Table 1. Classification of insomnia according to ICD-11.

Diagnosis	Insomnia, chronic	short-term
Code	7A00	7A01
Differentiation	XS5W mild XS0T moderate XS25 severe
Symptoms	• Regular or persistent problems falling asleep or staying asleep despite adequate sleeping opportunity and circumstances for sleep • Resulting dissatisfaction with sleep • Daytime impairment (exhaustion, depressed mood, irritability, general malaise, cognitive impairment) • Possibility of episodic course
Exclusion	• Absence of daytime impairment • Insomnia is the result of: - another sleep–wake disorder - mental illness - other medical conditions - substance or medication use
Time criterion	At least several times a week for at least 3 months	<3 months

Unspecific insomnia can be coded as 7A0Z (XT5R acute, XT8W chronic).

ICD-11, International Statistical Classification of Diseases and Related Health Problems, 11^th Revision

There is a lack of current data on the prevalence of insomnia in children and adolescents and the related health care situation in Germany. The 2002 Cologne Children‘s Sleep Study estimated the prevalence of insomnia symptoms in children at 15 (13). A more pronounced sleep disorder was found in 5–10 of school beginners (14). In a large German cohort study covering the period from 2011 to 2015, approximately 20 of the surveyed children and adolescents had sleep-related difficulties (15), including parasomnias, which are non-specific for insomnia, in addition to problems falling asleep and staying asleep.

Cognitive behavioral interventions have proven effective in the treatment of insomnia (11). However, despite evidence to the contrary, the use of medication as the initial treatment is common in everyday clinical practice (16). In the adult population, a gap in care for patients with chronic insomnia is noted (17).

The aim of our study was to investigate health care-related needs and practical implications, using the following research questions:

• What is the point prevalence of insomnia, taking into account the ICD-11 symptoms as well as the differentiation into mild, moderate and severe insomnia in a population-based sample of children and adolescents in Germany?

• Which protective and risk factors can be identified for the respective levels of insomnia severity?

Methods

A detailed description of the methods used is provided in the eMethods section.

eMethods. Data collection.

We conducted this representative, population-based study as an online survey in cooperation with the German market research and opinion polling company Forsa Institute (forsa) in May/June 2021. It is part of a large study on parent-child health with a special focus on the use of digital media which has been conducted annually since 2019, using a combined cross-sectional and longitudinal design (18). The participants were drawn from the forsa.omninet panel of randomly selected adults and adolescents from the age of 14 (e13). Forsa.onminet is a continuously growing panel of approximately 75 000 participants which were initially recruited exclusively offline, i.e. via phone (landline and mobile) (19). With this approach, the panel reaches both regular internet users and persons who are partially offline, e.g., persons who only use email but no other online services. Persons who do not use the internet are not included in the panel. In 2021, 92 of all households had an internet connection (e14). In that year, internet use among the under-50s was reported at 100 (e15). The composition of the panel is continuously reviewed using representative key characteristics, such as region of residence, age and gender.

In total, 16 464 households from the forsa.omninet panel met the selection criterion: adults aged 28 to 75 years. These were contacted by e-mail. Of the selected adults, 1759 had already participated in preliminary investigations of the larger parent study. Of 6595 respondents (40.1), 1499 (22.7) stated that they had children between the ages of 10 and 20. Ultimately, 1250 parents with their children provided the required information and gave their informed consent prior to participation. Of the qualified households, 83.4 met the inclusion criteria.

The recruitment process is shown in the supplementary eFigure. In households with more than one child of the relevant age, the child with the most recent birthday was invited to ensure a balanced ratio of age and gender. The representativeness of the sample with regard to age, gender and region of residence was sought via the underlying random sampling procedure, using the current microcensus data as a guide.

Parents and adolescents were asked to complete the online questionnaire independently of each other. If the adolescents needed help, the parents were instructed to support them in questions of understanding, but without suggesting answers. Completing the entire questionnaire took about 20–30 minutes. The respondents were able to collect points for their participation in the survey which they could redeem, for example, for donation vouchers.

The study was approved by the local Psychological Ethics Committee of the Center for Psychosocial Medicine at the University Medical Center Hamburg-Eppendorf (UKE) and conducted in accordance with national and institutional ethical guidelines and in compliance with the Declaration of Helsinki.

Survey methods

In addition to sociodemographic information, height and weight, the following variables were captured, using standardized screening questionnaires established for epidemiological surveys. Symptoms of parental and children’s insomnia were captured using the Insomnia Severity Index (ISI), which is recommended in the S3-level German clinical practice guidelines and also validated for children and adolescents (11). ISI is a questionnaire instrument used to assess the severity of disorders of initiating and maintaining sleep. It is suitable for both clinical and research-related purposes. A clinical validation according to ICD-11 and ICSD-3 criteria is available (e16). ISI comprises 7 items relating to the last month. These items address:

• Difficulty falling asleep

• Difficulty staying asleep

• Waking up too early

• Satisfaction with sleep

• Impairments in daily life caused by sleep problems

• Noticeable effects on others, e.g., mood, performance

• Worry/distress because of sleep problems.

Each item is rated on a 5-point Likert scale from 0 to 4 (0 = no problem, 4 = very severe problem). The severity and clinical significance of insomnia can be determined from the total scores of the scale: 0–7 points = no clinically significant insomnia; 8–14 = subthreshold (mild) insomnia; 15–21 = clinical moderate insomnia; and 22–28 = clinical severe insomnia. In our sample, the internal consistency was acceptable with a Cronbach‘s alpha of 0.71 for both the children/adolescents and their parents.

The Patient Health Questionnaire (PHQ)-9 is one of the most frequently used instruments that, based on self-reports related to the last two weeks, assesses the frequency of nine DSM-IV criteria for depressive disorders; PHQ-9 is reliable and also valid for young target groups (e19, e20). Each question is answered on a 4-point Likert scale, ranging from 0 (“not at all”) to 3 (“nearly every day”). The answer scores are aggregated to produce an overall score, which is graded as follows: 0–4 points = no depression; 5–9 points = mild depression; 10–14 points = moderate depression; 15–19 points = moderately severe depression; > 20 points = severe depression. The internal consistency of our sample of children and adolescents was good (Cronbach‘s alpha = 0.89).

The Generalized Anxiety Disorder (GAD-2) scale was used as a self-reported instrument to detect anxiety in children and adolescents. This very short screening tool consists of two items for capturing generalized anxiety with good sensitivity and specificity. On a 4-point Likert scale, questions on feeling nervous/on edge and uncontrollable worrying in the past two weeks were answered (scores range from 0 = not at all to 3 = nearly every day). A total score of ≥ 3 indicated the possibility of a generalized anxiety disorder, requiring further investigation (e21). Our study’s internal consistency was good (Cronbach’s alpha = 0.83). Media use times of children and adolescents during leisure time (days of use per week and duration of use per day in minutes) was systematically surveyed for digital games, social media, and online video platforms (e22) and aggregated into a weighted weekly average (in minutes).

Total scores of perceived stress, measured using the Perceived Stress Scale (PSS-4) (e23), and critical life events, captured using the Life Events Questionnaire (LEQ) (e24), served as covariates to cover possible environmental influences on sleep behavior at the time of the survey (18). PSS-4 is a short questionnaire to capture subjectively perceived stress during the previous two weeks. On a 5-point Likert scale, ranging from 0 (“never”) to 4 (“very often”), statements on feeling overwhelmed, loss of control, dealing with personal problems, and confidence were rated. Higher total scores indicate more severe perceived stress. LEQ captures personal and family-related critical life events, e.g., serious illness or death of a close relative. Responses to the questions were provided in a dichotomous format (yes [1] or no [0]) and summed up to total scores. Higher scores indicated more critical life events experienced. The responses of parents and children were combined as they reflected critical life events in the family context.

Data analysis

The statistical software R (version 4.3.2) was used for the statistical analysis (e25). Response patterns, which were incomplete by at least one third, were not included in the analysis. Remaining missing values (<5) were imputed according to the Fully Conditional Specification (FCS) method, using the R software package “mice” (19). A separate regression model was used to predict each incomplete variable. With this approach, we were able to include data of 1100 respondents in the final analyses. Point prevalences for insomnia were calculated using relative frequencies and 95 confidence intervals for the total sample as well as stratified by sex and age group (children aged 10–13 versus adolescents aged 14–17 [e26]).

The dependent variables age group, sex, obesity, media use duration, level of depression severity, anxiety, and parental insomnia were included together with the above-mentioned covariates in a joint multinomial logistic regression model to predict milder, moderate or severe symptoms of insomnia (R package “nnet”). The model included age group (children versus adolescents), sex, obesity (body mass index [BMI] ≤ 97^th percentile versus BMI >97^th percentile [e27]), anxiety (according to GAD-2 categorization, abnormal versus normal), and parental insomnia (categorization according to ISI, no versus at least moderate) in a dichotomous format, the severity of depression (according to PHQ-9 categorization, no to severe depression) ordinally, and media use time (weighted average per week in minutes) metrically. The z-standardized total values of the covariates were supplemented in the model. The regression method chosen is an extension of the “classical” logistic regression to more than two groups/categories of the dependent variable (e28). The ”no insomnia” group was used as the reference group for the “mild“, “moderate“ and “severe insomnia“ groups; their allocation was estimated via the above-mentioned predictors, using odds ratios and their statistical relevance. When performing the multinomial logistic regression analysis, we did not include a correction for multiple testing, in accordance with current statistical recommendations (e29). In complex multivariate models, it is often inadequate to use classical corrections, such as the Bonferroni correction, because they require that tests be independent of each other. This does not fully apply to our regression-based model with correlated predictors. Moreover, the aim of the analysis was not to test numerous individual hypotheses in isolation, but rather to understand a coherent, theoretically motivated model. According to Gelman et al. (2020) (e29), focusing on effect sizes, confidence intervals, and the totality of evidence is preferable to a dichotomous decision based on significance levels. For this reason, the current interpretation was centered on the relevance of the content and the plausibility of the findings as well as the direction and size of the estimated effects, and not on p-values alone (eTable).

Data collection

Our study is part of a large population-based, representative online survey of parent-child households on psychological family health which was conducted in cooperation with Forsa Institute (forsa), a German market research and opinion polling company. A total of 1128 children and adolescents aged 10–17 and one parent of each consented to participate in the research. The study was approved by the local Psychological Ethics Committee of the Center for Psychosocial Medicine at the University Medical Center Hamburg-Eppendorf (UKE) and conducted in accordance with national and institutional ethical guidelines and in compliance with the Declaration of Helsinki.

Survey methods

In addition to sociodemographic information, height and weight, the following variables were captured, using standardized screening questionnaires established for epidemiological surveys. Symptoms of parental and children’s insomnia were surveyed with the Insomnia Severity Index (ISI) which is recommended as a screening instrument in the S3-level German clinical practice guidelines mentioned above and also validated for children and adolescents (11). The severity and clinical significance of insomnia could be determined from the total scores of the scale: 0–7 points = no clinically significant insomnia; 8–14 = subthreshold (mild) insomnia; 15–21 = moderate clinical insomnia; and 22–28 = severe clinical insomnia.

The Patient Health Questionnaire (PHQ)-9 is an instrument which reliably and also validly for young target groups assesses the self-reported frequencies of the nine criteria of the Diagnostic and Statistical Manual of Mental Disorders (DSM) IV for depressive disorders, using the following severity classification: 0–4 points = no depression; 5–9 points = mild depression; 10–14 points = moderate depression; 15–19 points = moderately severe depression; >20 points = severe depression. The two-item Generalized Anxiety Disorder (GAD-2) scale was used to measure anxiety levels. This short screening tool indicates the presence of a generalized (non-specific) anxiety disorder if the total score is 3 or higher. Media use during leisure time (days of use per week and duration of use per day in minutes) was systematically surveyed for digital games, social media, and online video platforms and aggregated into a weighted weekly average.

Total scores of perceived stress, measured using the Perceived Stress Scale (PSS-4), and critical life events, captured using the Life Events Questionnaire (LEQ), served as covariates to cover possible environmental influences on sleep behavior at the time of survey (18).

Data analysis

The statistical software R 4.3.2 was used for the statistical analysis. Response patterns, which were incomplete by at least one third, were not included in the analysis. Remaining missing values of a questionnaire were estimated by imputation (19). With this approach, data of 1100 persons were included in the final analyses. The point prevalences of the levels of severity of insomnia were calculated as relative frequencies. Next, 95 confidence intervals were determined for the total survey sample and stratified by sex and age group (children [10–13 years] versus adolescents [14–17 years]).

The effects of the categorical variables age group (children versus adolescents), sex, obesity (body mass index [BMI] ≤ 97^th percentile versus BMI < 97^th percentile), severity of depression (no depression to severe depression), symptoms of anxiety (yes versus no), and parental insomnia (no versus at least moderate) were estimated in a joint multinomial logistic regression model together with the metric variables of the weekly media use duration and the above-mentioned covariates to predict mild, moderate or severe symptoms of insomnia.

Results

Sociodemographic characteristics and point prevalences

Table 2 lists the sociodemographic characteristics and point prevalences of insomnia of the parent-child sample. More detailed information is provided in the eTable.

Table 2. Characteristics of the parent-child dyads¹ and insomnia prevalences.

	Children and adolescentsN ( [95 CI])/ mean (sd; range)	ParentsN ( [95 CI])/ mean (sd; range)
Absolute frequency	1100	1100
Sex
Male	581 (52.82 [49.87; 55.77])	570 (51.82 [48.87; 54.77])
Female	519 (47.18 [44.23; 50.13])	530 (48.18 [45.23; 51.13])
Age (in years)	13.42 (2.17; 10–17)	45.62 (7.49; 28–76)
Age groups2	Children 591 (53.73 [50.78; 56.67])
	Adolescents 509 (46.27 [43.33; 49.22])
Insomnia
No	554 (50.36 [47.41; 53.33])	448 (40.73 [37.82; 43.63])
Mild	293 (26.64 [24.02; 29.25])	390 (35.45 [32.63; 38.28])
Moderate	235 (21.36 [18.94; 23.79])	229 (20.82 [18.42; 23.22])
Severe	18 (1.64 [0.89; 2.39])	33 (3.00 [1.99; 4.01])

A complementary description of the sample is provided in eTable 1.

¹ The parent-child dyads included in the analyses are shown.

² Children: 10–13 years of age; adolescents: 14–17years of age

CI, confidence interval; N, absolute frequency; sd, standard deviation

eTable. Supplementary characteristics of the parent-child dyads and insomnia prevalences.

Variables/ categories	Children1 and adolescents N ( [95 CI])/ mean (sd; range)	ParentsN ( [95 CI])/ mean (sd; range)
Parent-child relationship status
Biological child	1003 (91.18 [89.51; 92.86])
Adopted child	10 (0.91 [0.35; 1.47])
Stepchild	51 (4.64 [3.39; 5.88])
Other2	36 (3.27 [2.22; 4.32])
Level of education3, 4
High	574 (55.84 [52,8; 58.87])	261 (23.75 [18.63; 28.88])
Moderate	370 (35.99 [33.06; 38.93])	736 (66.97 [60.06; 73.88])
Low	84 (8.17 [5.24; 11.1])	83 (7.55 [5.76; 9.35])
Employment
School attendance/full-time employment	1046 (95.09 [94.4; 96.83])	655 (59.55 [56.65; 62.45])
Training/part-time employment	35 (3.18 [2.16; 4.24])	334 (30.36 [27.65; 33.08])
Other5. 6	19 (1.72 [0.01; 2.59])	111 (10.09 [5.22; 14.96])
Living environment
Urban environment7	877 (79.73 [77.35; 82.10])
Rural environment	233 (20.27 [17.90; 22.65])
Insomnia
• None
- Girls	250 (48.17 [43.87; 52.47])
- Boys	304 (52.32 [48.26; 56.38])
- Children	291 (49.24 [45.21; 53.27])
- Adolescents	263 (51.67 [47.33; 56.01])
• Mild
- Girls	148 (28.52[24.63; 32.40])
- Boys	145 (24.96 [21.44; 28.48])
- Children	160 (27.07 [23.49; 30.66])
- Adolescents	133 (26.13 [22.31; 29.95])
• Moderate
- Girls	111 (21.39 [17.86; 24.91])
- Boys	124 (21.34 [18.01; 24.67])
- Children	133 (22.50 [19.14; 25.87])
- Adolescents	102 (20.04 [16.56; 23.52])
• Severe
- Girls	10 (1.93 [0.74; 3.11])
- Boys	8 (1.38 [0.43; 2.32])
- Children	7 (1.18 [0.31; 2.06])
- Adolescents	11 (2.16 [0.90; 3.42])

For further information, please refer to Table 2 in the main text.

¹ Children = persons up to the age of 14

² Foster child /not specified

³ Adolescents: (expected) school leaving certificate (based on own assessments)—high = university entrance qualification/A-levels, moderate = intermediate secondary school leaving certificate; low = none/ special school leaving certificate/general secondary school leaving certificate

⁴ Parents: highest educational achievement—high = bachelor/master to PhD; moderate = completion of secondary education (intermediate secondary school leaving certificate /university entrance qualification/completed vocational training); low = no or general secondary school leaving certificate

⁵ Adolescents: students, voluntary service, military service, other employment, or unemployed

⁶ Parents: job seekers, welfare recipients, pensioners, persons with disabilities, trainees, students, no specification

⁷ Areas with more than ≥ 5000 inhabitants (German Federal Institute for Research on Building, Urban Planning, and Spatial Development)

CI, confidence interval; N, absolute frequency; sd, standard deviation

Altogether, 293 of the 1100 children and adolescents, corresponding to about one quarter of the total sample, met the criteria for mild insomnia, 235 children and adolescents for moderate insomnia (about one-fifth) and 18 respondents for severe insomnia (1.6). The confidence intervals for the prevalence rates, stratified by sex and age group, overlap and thus do not indicate the presence of group differences (eTable).

Risk and protective factors

Table 3 presents the findings of the multinomial model, explaining 21.5 of the total variance. Compared to children and adolescents without insomnia, adolescents had a 29 lower chance of suffering from mild insomnia than children, while for moderate insomnia, the chance was 34 lower. With each additional level of severity of depression, children and adolescents had a 1.8-fold increased risk of mild insomnia and a 1.3-fold increased risk of moderate insomnia.

Table 3. Multinomial regression model¹ for the prediction of levels of insomnia severity².

	Mild insomnia3		Moderate insomnia3		Severe insomnia3
	OR (p-value)	95 CI	OR (p-value)	95 CI	OR (p-value)	95 CI
Adolescents4	0.71 (0.050)	[0.51; 1.00]	0.66 (0.027)	[0.46; 0.95]	1.26 (0.684)	[0.41; 3.84]
Female sex5	1.15 (0.409)	[0.83; 1.58]	0.94 (0.720)	[0.66; 1.34]	1.60 (0.393)	[0.54; 4.69]
Media use time6	1.18 (0.052)	[1.00; 1.39]	1.03 (0.737)	[0.85; 1.26]	1.18 (0.503)	[0.73; 1.91]
Obesity7	1.12 (0.696)	[0.63; 2.02]	0.48 (0.082)	[0.21; 1.10]	3.20 (0.081)	[0.87; 11.77]
Depression8	1.83 (< 0.001)	[1.49; 2.24]	1.29 (0.030)	[1.02; 1.61]	1.60 (0.091)	[0.93; 2.75]
Anxiety9	1.46 (0.312)	[0.70; 3.05]	4.54 (< 0.001)	[2.09; 9.88]	7.96 (0.008)	[1.72; 36.94]
Parental insomnia10	0.87 (0.522)	[0.57; 1.33]	2.49 (< 0.001)	[1.66; 3.72]	3.30 (0.040)	[1.06; 10.30]
Critical live events11	0.93 (0.384)	[0.79; 1.10]	1.06 (0.535)	[0.89; 125]	1.05 (0.848)	[0.66; 1.67]
Perceived stress12	0.90 (0.246)	[0.75; 1.08]	0.74 (0.004)	[0.61; 0.91]	1.10 (0.754)	[0.61; 1.98]

¹ The variables mentioned above can explain 22 of the model variance (R2 = 0.22).

² Multinomial regression is a statistical method for predicting a categorical dependent variable with more than two levels using one or more independent variables. Interpretation example: If the odds ratio for the prediction of moderate insomnia compared to no insomnia by anxiety is 4.54, this means: In contrast to persons without anxiety, persons with anxiety have a 4.5-times greater odds to report moderate insomnia compared to no insomnia.

³ In comparison to no insomnia: The Insomnia Severity Index, which is recommended as a screening instrument in the German S3-level clinical practice guideline “Non-Restorative Sleep/Sleep Disorders“, was used to measure the various levels of insomnia severity.

⁴ Aged 14 to 17 years (versus children aged 10 to 13 years)

⁵ Compared to the male sex

⁶ The weekly media use time is shown after z-standardization, i.e. mean and standard score normalization, to enable comparability of the values regardless of units of measurement and scales.

⁷ BMI percentile >97^th percentile compared to BMI percentile ≤ 97^th percentile

⁸ Using the Patient Health Questionnaire (PHQ-9), five categories, ranging from no depression to severe depression, were formed.

⁹ Using the Generalized Anxiety Disorder (GAD-2), two categories (generalized anxiety present versus no generalized anxiety present) were formed.

¹⁰ One corresponding parent was tested using ISI. Two categories were created (at least moderate insomnia vs. no insomnia).

¹¹ The Life Events Questionnaire (LEQ), which was entered into the analysis as a z-scaled total score, was included as a covariate.

¹² The z-scaled total score of the Perceived Stress Scale (PSS-4) was included in the model as a covariate.

BMI, Body Mass Index; OR, odds ratio; CI, confidence interval

The association between risk factor and greater severity of insomnia is stronger in the presence of existing anxiety (4.5-fold increased chance of moderate insomnia, 8-fold increased chance of severe insomnia) and parental (clinically significant) insomnia (2.5-fold increased chance of moderate insomnia, 3.3-fold increased chance of severe insomnia). Sex has no influence. Contrary effects were observed for obesity; with p-values of 0.08, however, they did not reach the significance threshold: While respondents with obesity had a 52 lower chance of having moderate insomnia, their chance of having severe insomnia was increased by factor 3.2. Longer media use times were associated with an 18 increased chance only with regard to mild insomnia. However, with a p-value of 0.052, this result also narrowly fell short of the threshold for statistical significance.

Discussion

Our study is the first to systematically and in a standardized way record the ICD-11 criteria of insomnia in a large German sample of parent-child dyads to estimate the point prevalence. Aside from identifying age group-related and psychological risk and protective factors, relevant associations between parental and children’s insomnia were revealed.

Half of the 10– to 17-year-olds reported experiencing symptoms of mild to moderate insomnia in the previous month. This is consistent with the proportion of children and adolescents suffering from sleep disorders according to the findings of a meta-analysis (20). Approximately one-fifth meet the criteria for moderate insomnia. This proportion corresponds to the results of a meta-analysis on adolescent insomnia (21); however, it is higher than the proportions found in meta-analyses using the same instrument (ISI) that primarily looked at adult study populations (22). It approximately corresponds to the proportion of children and adolescents in Germany who experience emotional and behavioral problems (23). The proportion of severe insomnia (1.6) does not differ from the findings in adult populations (22). It is comparable to the prevalence of (severe) depression in children and adolescents (24).

In a meta-analysis on sleep disturbances, Jahrami et al. (20) identified children and adolescents as one of the most vulnerable groups. In the current study, belonging to the adolescent age group was identified as a protective factor for mild and moderate insomnia. This may be attributable to the fact that behavior-related insomnia is more prevalent in the younger age group; this type of insomnia is commonly associated with resistance to going to bed (25) and is also more strongly influenced by parental interaction and genetic factors (26). This finding stands in contrast to representative results on chronic lack of sleep, a condition that is significantly more common among adolescents than children (27). Compared to boys, girls suffered more frequently from chronic lack of sleep (27) and other overt sleep problems (28). However, in our survey sample, female sex was not identified as a risk factor for insomnia. The findings of studies on adults are inconclusive: While a systematic review reported appreciable sex differences for sleep–wake disorders (29), a meta-analysis of international studies on ISI found no differences (22). This may be attributable to the use of different measuring instruments and varying concepts of insomnia and sleep problems.

In our study, depression was found to be a risk factor mainly for less severe forms of insomnia. We identified anxiety as the most significant risk factor (up to 8 times higher odds ratios) for more severe forms of insomnia. In children and adolescents, the relationships between sleep disorders on the one hand and depression and anxiety on the other are bidirectional (30, 31). According to a prospective meta-analysis, sleep disorders can predict depressive symptoms in children and adolescents (8). Genetic causes have been identified in this regard (32). At the same time, depression and anxiety disorders as well as developmental disorders (attention deficit hyperactivity disorder, autism, epilepsy) frequently occurred as comorbidities or in association with insomnia in children and adolescents (25, e7). In adolescents, the relationship between mental illness and insomnia is complex as it is influenced by biological, psychological and social factors, including cytokine and cortisol levels, altered sleep architecture, dysfunctional beliefs, cognitive distortions, and reduced social interactions (33).

Parental insomnia of at least moderate severity was found to be a further important risk factor for moderate to severe insomnia among their children. This finding is consistent with the findings of a recent Chinese study on a random sample of 68 751 parent-child dyads (34). Insomnia is a polygenic, stress-associated illness that is most likely caused by an interaction between genetic and environmental factors (e8). Possible environmental factors include family functionality (35), features of the surroundings, such as green spaces and air pollution (e9), as well as parental monitoring (e10) and handling of children’s sleep problems (e11).

Obesity as a risk factor for severe insomnia failed to reach the threshold for statistical significance. According to a meta-analysis of prospective studies, a shorter sleep duration, which can be due to insomnia, can increase the risk of obesity in children (36). There is, however, a global lack of studies focusing on sleep quality rather than sleep quantity (e6).

High media consumption is widely discussed as a factor contributing to disturbed nighttime sleep (37). In our study, media use narrowly missed the significance threshold as a risk factor for mild insomnia. Previous studies, lacking a uniform conceptualization of sleep problems, have shown considerable heterogeneity also with regard to media consumption (38). This is compounded by the fact that long media use times during the COVID-19-Pandemie have become the norm and correlate less with use quality, for example problematic use patterns (39). These, in turn, seem to have a greater impact on sleep quantity and quality than media use time alone (40).

The treatment of children and adolescents with insomnia consists primarily of psychoeducation for those affected and their caregivers. It includes sleep hygiene and age-adapted cognitive behavioral therapy techniques with exercises for cognitive restructuring as well as teaching relaxation techniques (e12). If patients with chronic insomnia fail to respond to this treatment, the temporary use of melatonin may be considered (e28). At a higher level, holistic treatment with improvement of stress tolerance appears to be a promising approach (26). Clinical management recommendations for Germany are expected to become available once the development of the S2e-level clinical practice guideline “Non-Restorative Sleep/Sleep Disorders —Insomnia in Children and Adolescents” is completed (AWMF register no. 028–012).

Strengths and limitations

The strengths of our study include the large parent-child survey sample and the standardized collection of data on insomnia at the various levels of severity. The following limitations apply: Data collection was carried out during the COVID-19 pandemic. It cannot be ruled out that the prevalence estimate was influenced by the pandemic situation in combination with increased psychological stress (e9). To take potential environmental influences at the time the survey was conducted into account, critical life events and perceived stress were included in the analysis as covariates (18). The meta-analysis by AlRasheed et al. (22) showed that only subthreshold (mild) insomnia cases increased during the COVID-19 pandemic, while the prevalence of more severe forms of insomnia did not change. Furthermore, our findings are comparable with the results of international pre-pandemic meta-analyses (21). The data collected is entirely based on self-reports with limited numbers of items. It is not possible to establish a clinical diagnosis solely on the basis of the screening questionnaires used. Data on mental and physical comorbidities as well as objective parameters could not be collected. This is a general limitation of epidemiological studies. As a consequence, an over- or underestimation of the reported findings may have occurred due to self-misjudgments or selection bias, with particularly affected groups being underrepresented. To address these limitations, standardized validated and reliable survey instruments were used in combination with a complex sampling method.

Follow-up studies should aim for oversampling of particular groups, such as children with obesity, in order to increase the power to also detect smaller effects. In addition, future research should collect data on chronic and short-term insomnia using differentiated time criteria and prospective study designs should be implemented to allow for a causal interpretation of risk and protective factors. In our study, only one parent could be included in the survey for reasons of practicability. Further research should include genetic testing of entire families. The described strong association between parental and children’s insomnia can be taken into account in everyday clinical practice: If parents report sleep problems, the possibility that the child may have a sleep disorder that requires treatment should be given consideration, and vice versa.

Conclusion

The presented epidemiological findings have implications for everyday clinical practice. Almost one in four children aged between 10 and 17 reported symptoms of at least moderate insomnia. Thus, a low-threshold systematic insomnia assessment is warranted. A holistic, primarily non-pharmacological treatment approach involving the child and parents is indicated to alleviate the often considerable psychological strain on the family and prevent adverse effects on the development of the child and chronification of insomnia. There is a need for prospective research focusing on biological and environmental influences on insomnia in children and adolescents.

eFigure.

Recruitment of sample

N, sample size