Inclusion criteria

Children and adolescents between 12 and 18 years are eligible for participation. A sufficient understanding of the German language, as well as informed consent of both participants and legal guardians, is required to participate in the study.

For the index group, potential participants are those with depressive disorders according to the diagnostic criteria of the ICD-10.³³ More precisely, we include the ICD-10 codes F32.0 (mild depressive episode), F32.1 (moderate depressive episode), F32.2 (severe depressive episode without psychotic symptoms), F33.0 (major depressive disorder, recurrent, mild), F33.1 (major depressive disorder, recurrent, moderate), F33.2 (major depressive disorder, recurrent, severe without psychotic features) and F43.21 (adjustment disorder with prolonged depressive reaction) in our index group.

For the control group, we include healthy adolescents matched by gender, age and pubertal status according to the Pubertal Development Scale (PDS). This shall align age- and gender-specific differences.^{Reference Baglioni, Nanovska, Regen, Spiegelhalder, Feige and Nissen34}

Exclusion criteria

Potential participants who suffer from psychotic or acute suicidal conditions are excluded from study participation. Pregnancy, antidepressant or glucocorticoid medication, and neurological or endocrinological illnesses that may affect brain physiology or relevant endocrinological parameters are additional exclusion criteria. Moreover, those with cannabis consumption within the 3 months before study participation are excluded.

Furthermore, most psychiatric comorbidities constitute exclusion criteria. For the index group, we include anxiety disorders, as they appear cumulatively comorbid with depressive disorders.^{Reference Fava, Alpert, Carmin, Wisniewski, Trivedi and Biggs35,Reference Choi, Kim and Jeon36} All other psychiatric comorbidities lead to exclusion.

For the control group, we exclude participants with any psychiatric disease, symptoms indicating a current psychiatric disease, a history of psychotherapeutic or psychiatric intervention, or a condition requiring therapeutic support. Potential participants excluded because of the latter criterion are offered therapeutic help if the indication is given.

Assessment of depressive symptoms, psychiatric comorbidities and stress responsiveness

Initially, all participants are screened and interviewed with the Mini-International Neuropsychiatric Interview for Children and Adolescents (M.I.N.I. KID 6.0)^{Reference Sheehan, Lecrubier, Sheehan, Amorim, Janavs and Weiller37} for psychiatric disorders (controls) or comorbidities (patients) (T1, see Fig. 1). The M.I.N.I. KID 6.0 interview is structured into 24 diagnostic modules assessing potential symptoms of psychiatric disorders according to the DSM-IV and ICD-10.

Depressive symptoms are assessed via the Beck Depression Inventory-II,^{Reference Hautzinger, Keller and Kühner38} which is one of the most commonly used self-rating scales to measure depressive symptoms^{Reference Richter, Werner, Heerlein, Kraus and Sauer39} from the age of 13 years (T1, see Fig. 1). Frequently observed symptoms are inquired for the past 2 weeks in 21 items, with four possible response options covering increasing symptom severity. As an adjustment for the younger study participants, we excluded item 21 (loss of interest in sexuality) in our assessment. Internal consistency for the German version is good (α ≥ 0.84), retest reliability is r ≥ 0.75 and discriminative validity is confirmed.^{Reference Kühner, Bürger, Keller and Hautzinger40} The applicability to an adolescent sample was checked and validated.^{Reference Dolle, Schulte-Körne, O'Leary, von Hofacker, Izat and Allgaier41,Reference Pietsch, Hoyler, Frühe, Kruse, Schulte-Körne and Allgaier42}

As symptoms of anxiety and anxiety disorders occur often concurrently with depressive disorders,^{Reference Fava, Alpert, Carmin, Wisniewski, Trivedi and Biggs35,Reference Choi, Kim and Jeon36} we examine anxiety via the Beck Anxiety Inventory (BAI)^{Reference Margraf and Ehlers43} (T1, see Fig. 1). The BAI consists of 21 items focusing on somatic symptoms of anxiety experienced within the past week.^{Reference Beck, Epstein, Brown and Steer44} The German BAI showed reliable and valid psychometrics.^{Reference Geissner and Huetteroth45} For adolescent samples, the BAI showed at least acceptable psychometric properties in other language versions.^{Reference Geissner and Huetteroth45,Reference Osman, Hoffman, Barrios, Kopper, Breitenstein and Hahn46}

Stable sleeping habits and needs will be assessed by a short self-designed self-report survey (T1, see Fig. 1). This survey assesses chronotype, the regularity of bedtimes, the duration of sleep needed to feel rested and the frequency of sleeping through school or work. Furthermore, it queries the usual bed and wake-up times for each school/working days and weekend days/holidays. The items used for this survey are based on the Sleep Habit Survey (SHS).^{Reference Hamann, Rusterholz, Studer, Kaess and Tarokh47} For the subscales, the original SHS showed acceptable internal consistency, with α ranging from 0.70 to 0.79.^{Reference Kandel and Davies48}

Participants with a depressive disorder additionally answer a self-designed questionnaire examining potential differences in relevant aspects of sleep quality since the onset of depressive symptoms (T1, see Fig. 1). This questionnaire consists of eight items, in which the participants compare the need to sleep, sleep and dream quality, WASO, mood at awakening, time of awakening and personal relevance of the topic sleep before and after symptom onset, and evaluate changes on a five-point Likert scale.

Sleeping quality of the previous 2 weeks is evaluated with the Sleep Questionnaire B (SF-B/R)^{Reference Görtelmeyer49} (T1, see Fig. 1). The SF-B/R is a validated 31-question survey. Five sleep indices can be calculated: difficulty falling asleep, difficulty staying asleep, premature awakening, general sleep characterisation and total sleep duration. In addition, seven factor scales (sleep quality, feeling of being refreshed after sleep, mental balance before sleep, feeling mentally exhausted before sleep, psychosomatic symptoms during the sleep phase, dream recall and sleep–wake regulation) can be assessed. The SF-B/R is designed for sleep research as well as clinical practice. Satisfactory reliability values >0.70 are shown with regard to the internal consistency of the factor levels and the retest reliability of the SF-B/R (2–4 weeks). With regard to convergent validity, medium correlations are found with construct-related domains, such as measures of well-being, depression and personality factors.^{Reference Görtelmeyer49}

Participants also complete the Stress and Coping Questionnaire for Children and Adolescents (SSKJ 3-8 R)^{Reference Lohaus, Eschenbeck, Kohlmann and Klein-Heßling50} (T1, see Fig. 1). The SSKJ 3-8 R is a self-report questionnaire based on the transactional model of stress and coping developed by Lazarus and Folkman, assessing several strategies to reduce negative emotions or to change conditions in problematic or negatively experienced situations.^{Reference Weis and Heine51} The first part comprises an assessment of vulnerability to different stressors on four-level visual analogue scales. The second part includes five- and three-point Likert scales evaluating the usage frequency of five different coping strategies.^{Reference Weis and Heine51} The German version of the SSKJ 3-8 R showed sufficient reliability and validity has been confirmed.^{Reference Schmidt and Laessle52,Reference Eschenbeck, Kohlmann, Lohaus and Klein-Heßling53}

Pubertal development is assessed with the use of the German adaption^{Reference Watzlawik54} of the Pubertal Development Scale (PDS)^{Reference Petersen, Crockett, Richards and Boxer55} as a self-report questionnaire (T1, see Fig. 1). Female participants asses the current status of breast development, pubic hair growth, growth in height and skin, and estimated level of development compared with peers. Male participants are asked to evaluate the start and level of beard growth, existence of puberty vocal changes, pubic hair growth, growth in height, and skin changes, and to compare their level of development with their peers.^{Reference Watzlawik54,Reference Petersen, Crockett, Richards and Boxer55} Measures of criterion validity as well as internal consistency have achieved acceptable values.^{Reference Watzlawik54}

Accelerometry

For sleep measuring, we use the ActTrust 2 wrist actimeter (Condor Instruments) worn on the non-dominant wrist for seven consecutive nights, starting on Sunday evening (measurement week, see Fig. 2). Wearing and handling of this accelerometer is practiced by the adolescents and their legal guardians during the first study appointment. During the measurement week, participants put on the accelerometer when they go to bed in the evening. By pushing the event button, they record the time they intend to sleep and mark unexpected events or delays (for example, getting up again to go to the bathroom). While sleeping, data is recorded in 1-min intervals. The selected interval is the default of the manufacturer Condor Instruments, and is obligatory for the post-processing of sleep data. In the morning, wake-up time is recorded by pushing the event button again. Having marked the wake-up time, participants take the accelerometer off.

Fig. 2 Study diagram. BAI, Beck Anxiety Inventory; BDI-II, Beck Depression Inventory-II; CAR, cortisol awakening response; M.I.N.I KID 6.0, Mini-International Neuropsychiatric Interview for Children and Adolescents; PDS, Pubertal Development Scale; SF-A/R, Sleep Questionnaire A – revised version; SF-B/R, Sleep Questionnaire B – revised version; SSKJ 3-8 R, Questionnaire for the Measurement Stress and Coping in Children and Adolescents – revised version.

After participants return the accelerometer at the second study appointment, we check the recordings with the ActStudio software (Condor Instruments, São Paulo, Brazil, PC version 1.0.10, for Windows; see https://www.condorinst.com.br/en/actstudio-software/). In accordance with the standard procedure in sleep accelerometry, we manually identify the main sleep period.^{Reference Ruiz, Beijamini, Beale, Gonçalves, Vartanian and Taporoski56} Bed time, get up time, time in bed, total sleep time, onset latency, sleep efficiency, WASO, awakenings and the intensity of blue light are provided by the software. Additionally, activity data during sleep is derived from the data, as increased motor activity is associated with impaired sleep quality.^{Reference Meltzer, Short, Booster, Gradisar, Marco and Wolfson9}

In the validation study, the ActTrust accelerometer achieved satisfactory results with a sensitivity of 95.69%, a specificity of 30.59%, an accuracy of 80.24%, a predictive value for sleep of 81.52% and a predictive value for wakefulness of 68.93%.^{Reference Rodrigues and Eckeli57} Regarding the correlation analysis between accelerometry and polysomnography, r = 0.491 (P < 0.001) was observed for total sleep time, r = 0.499 (P < 0.001) for wake after sleep onset, r = 0.345 (P < 0.001) for sleep efficiency and no statistical significance was observed for sleep onset latency.^{Reference Rodrigues and Eckeli57}

Daily sleep questionnaires

To examine subjective sleep quality, participants complete a short questionnaire including 11 items, each morning after accelerometry measurements (measurement week, see Fig. 2). Participants evaluate the satisfaction with the past night's sleep, current mood and feelings of recreation, difficulties with sleeping in nocturnal awakening and stress level of the past day, on a six-point Likert scale. Information on the presence of nightmares, the consumed quantity of caffeinated beverages and the consumed quantity of alcoholic beverages will be gathered. Additionally, the questionnaire assesses sleeping and wake-up time to double check the ActTrust recordings.

On the last morning of measurements, participants answer the daily questionnaire as well as the Sleep Questionnaire A (SF-A/R)^{Reference Görtelmeyer49} (measurement week, see Fig. 2). The SF-A/R includes 25 questions assessing several qualitative and quantitative aspects of the past night's sleep. Similarly to the SF-B/R, sleep indices and factor scales can be calculated. We use data from both questionnaires to compare measurements of the self-designed daily sleep questionnaire with a standardised sleep survey.

Saliva sampling

To analyse the CAR and the course of alpha-amylase after awakening, saliva samples are collected. We instruct participants to collect saliva samples via Salivette Cortisol (Sarstedt, Nümbrecht, Germany; item number 51.1534.500) devices on three consecutive weekdays (Tuesday to Thursday) at three time points (measurement week, see Fig. 2). The first sampling time is immediately after awakening (base level of CAR), the second sample is collected 30 min after awakening (peak in CAR) and the last sampling time is 45 min after awakening (decrease in CAR). The chosen time intervals conform to the recommended sampling points of the expert consensus guidelines for CAR.^{Reference Stalder, Kirschbaum, Kudielka, Adam, Pruessner and Wüst32}

At predetermined times, participants open a container equipped with a medication event monitoring system cap that records the opening time. The Salivettes are equipped with the fitting label for sampling days and time. Participants are instructed in oral and written form not to eat, drink or brush their teeth for 10 mins before collecting saliva samples.

The Salivettes are coded with weekdays and corresponding colours, and are kept separate for each weekday, to maximise clarity and simplify saliva sampling for participants. Sampling times are recorded and participants are informed about the monitoring before the implementation of measurements.

After sample collection, participants store the samples in their fridge or freezer, and return the samples on the second study appointment (T2, see Fig. 2). Then, the saliva samples are stored transitionally in the institute's freezer at −20°C. After completion of the study, the saliva samples are sent with a pseudonymised labelling to the Laboratory of the Department of Biopsychology, Technical University of Dresden, Germany, for analysis.

Primary outcome

The primary outcome of our study is the difference between the index group and control group on all measures of subjective (daily self-assessment questionnaires) and objective (ActTrust measurements, sleep onset latency, sleep efficiency, WASO and awakenings) sleep quality, the pattern of the CAR and the course of alpha-amylase after awakening.

Secondary outcome

The secondary outcome of our study is the examination of the severity of depressive symptoms, further psychopathologies, and stress responsiveness at the first study appointment and its relation to sleep quality.

A priori power calculation

A priori sample size was calculated with the program G*Power version 3.1.9.4,^{Reference Faul, Erdfelder, Buchner and Lang61} based on a previous study by Schmidt et al^{Reference Schmidt, Laessle and Hellhammer14} with a medium effect size of f = 0.30, an α error probability of 0.05 and two groups. To investigate group differences in the CAR through a repeated measures analysis of covariance (ANCOVA), a total of 28 participants would be required to achieve a power of 80%. In accordance with Selvi et al, we assume a medium effect size of d = 0.40 for correlational analyses of sleep quality and depressive symptoms.^{Reference Selvi, Aydin, Boysan, Atli, Agargun and Besiroglu62} A sample size of N = 44 was deemed necessary for statistical power of 80%. Finally, to enable the calculation of a direction-dependence analysis in case of non-normally distributed and sufficiently skewed data,^{Reference Wiedermann and von Eye63} we will include a sample of N = 60 participants: 30 participants with a depressive disorder and 30 healthy participants as parallel controls.