Study design, data collection and data sources

Data for this cohort study were collected from September 2022 to April 2024 in the five Regions of Denmark. The regions are administrative entities responsible for the public secondary health care system, i.e. psychiatric and somatic hospitals. In Denmark, health care coverage is universal and hospitals are mainly public.

As part of the recruitment and data collection, the research team visited 26 sites (hospitals) across the country (Fig. 1). Each site was visited for 1–4 days. A smaller group of participants visited the research team at The National Research Centre for the Working Environment (Danish: Det Nationale Forskningscenter for Arbejdsmiljø, NFA), Copenhagen, Denmark.

Fig. 1

Overview of geographical location of the data collection in Denmark

The baseline data collection consisted of data from the examination day (“day zero”), blood samples and a background questionnaire. Hereafter, all participants were followed for 14 consecutive days via sleep diaries and objective measurements of sleep with wrist actigraphy. During the first 7 days after the baseline data collection, subsamples were also followed with objective, continuous measurements of light exposure, physical activity, skin temperature, and blood glucose. Saliva samples were collected across one day in a subsample.

Inclusion and exclusion criteria

Women working in one of the five Danish regions were eligible for inclusion in the cohort. The study population was delineated to female employees for three reasons: (i) The cohort supplements existing studies of night work among men; (ii) women are dominant in the health care sector; (iii) hormones and biological markers differ between sexes necessitating sex stratified analyses and hence, a larger study population.

Employees could participate regardless of their work schedule as long as they worked a minimum of 28 h a week. The ambition was that the cohort should consist of 70% night shift workers, 10% permanent night workers, and 20% day or evening workers. Pregnant women or women trying to conceive in the 14 days period were not eligible for inclusion due to potential changes in hormone levels and other biomarkers measured in blood and saliva samples. Otherwise, we did not apply any health-related inclusion or exclusion criteria for the cohort (see Table 1 for further details about inclusion criteria for data collection in subsamples).

Recruitment of hospitals and participants

Participants were recruited through the workplaces (i.e. hospitals). Prior approval from each of the five Regions’ workplace cooperation committees (Danish: MED-udvalg) consisting of representatives for management and employees was obtained. A contact persons in each region communicated information about the project to the management of the hospitals in each region. Decisions about participation at the hospital level was then taken either by management alone or in collaboration between employers’ and employees’ representatives. We did not systematically collect any information about reasons for participation or non-participation at the organizational level, yet, lack of resources was at play. Some hospitals, which did not initially sign up for the project, signed up at a later time as resources became available. The role of the hospitals in the recruitment process was to assign a local contact person, distribute information about the project (received from the research group), and to book rooms for the baseline examination of the participants.

To obtain awareness about the research project, information was communicated through the hospitals’ own communication channels, at the project’s website, via social media and by contact to labor unions. If possible, flyers were distributed at the hospitals (e.g. in the employees’ canteen) a week before or during the data collection.

Recruitment was done at an individual level: Each participant booked a time-slot for participation, typically starting between 7:00 am and 16:30 pm. These “opening hours” allowed employees to participate regardless of their shift type. Enrollment, examination and blood sampling took 30–45 min. Participation could take place during working hours (if approved by the management) or during leisure-time.

Participants

A total of 1075 individuals signed the consent of participation statement and provided data for the study. Participants were nurses, nursing assistants, midwifes, medical doctors, biomedical laboratory scientist, and administrative staff employed either at a somatic or psychiatric hospital. The participants worked in a broad range of departments, e.g. medical, surgical, emergency or outpatient wards. The participants represented various medical specialties, e.g. psychiatry, orthopedic surgery, gynecology and obstetrics, radiology, and oncology.

Table 1 presents an overview of the data sources, the inclusion criteria for each type of data collection, and number of participants in the subsamples contributing with data for each specific data source. An overview of the number of participants with combinations of available data can be found at the project’s webpage (www.nfa.dk/1001nights).

Table 1 Overview of data sources and inclusion criteria for the specific parts of the data collection according to the protocol. The examination, collection of blood sample, and distribution of the background questionnaire were taking place on the day of enrollmentEthics and adherence to general data protection regulation

The study is approved by the Danish National Committee on Health Research Ethics (H-21077744) and follows the regulations of the Danish Data Protection Agency including an internal registration and risk assessment.

On the examination day, the participants received written and oral information about the project, including potential discomfort and complications related to participation, and they signed an informed consent form. Participation was voluntary and participants could withdraw from the project at any time. We emphasized that employers were not informed about who participated in the project and health information from individual participants was treated confidentially.

Individual feedback was given in the case of a high depression score or abnormal levels of blood lipids and glycated hemoglobin. Furthermore, all participants with available data received an individual sleep report (sleep diary data) and a glucose report (continuous glucose measurement (CGM) data).

Measurements on the examination dayBlood pressure

Blood pressure was measured three times with an Omron 3 Comfort blood pressure monitor on the upper right arm after the participants had rested in a sitting position for approximately 15 min (while receiving information about the project). Participants were instructed to sit in an upright position with both feet on the floor and asked not to talk during the measurement. The mean of the two last measurements was calculated and recorded.

Height and weight

The participants were instructed to be barefoot, empty their pockets and take off outerwear such as jackets and thick sweaters. Height was measured up against a wall using a Seca 217 altimeter measuring to nearest 0.1 cm. For weight measurements, we used an electronic Seca 803 Scale measuring to nearest 0.1 kg.

Hip and waist circumference

Hip and waist circumferences were measured with a regular measurement band. Waist was measured between the lower rib margin and the hip approximately at the level of the belly button, while hip circumference was measured where the participants were broadest. Participants were asked to empty their pockets and take of jackets and thick sweaters. Measurements were registered in centimeters and rounded to nearest 0.5 cm.

Blood sample

Up to 50 ml of blood was collected from each participant from the antecubital vein by venipuncture using Vacutainer tubes. The participants were non-fasting prior to sampling. Information on time of sampling, alcohol use and physical activity up to four hours before sampling was recorded. 2 ml of blood from each participant was kept as whole blood. After 15 min at room temperature, the rest of the blood was centrifuged for 15 min at 4000 rpm most often within 60–90 min after sampling. Afterwards, the samples were separated into 1 ml aliquots of serum, EDTA plasma, and cell fractions and stored at -80 ºC at NFA.

Background questionnaire

The background questionnaire was electronically distributed via the participants’ email. Participants were encouraged to complete the background questionnaire as soon as possible; no reminders were sent. The questionnaire contained 57 questions about sociodemographic factors, working hours and working hour preferences, sleep, health behaviors, health, medication use and COVID-19. A comprehensive Morningness-Eveningness Questionnaire (MEQ) was applied, and a MEQ-score was calculated according to the guidelines [14]. Likewise, we used the Major Depression Inventory (MDI) and calculated a scale score according to the guidelines [15]. Sleep questions were inspired by Karolinska Sleepiness Questionnaire [16]. A full overview of the questionnaire is available at the project’s webpage (www.nfa.dk/1001nights).

Sleep diaries

Sleep diaries were distributed via the participants’ email every morning at 5:00 for 14 days, and the participants were instructed to fill out the sleep diary immediately after termination of their primary sleep (Fig. 2). The sleep diary covered timing of working hours, sleep, physical activity, meals, and snacks, sleep quality, psychosocial and physical job demands, and symptoms since the previous primary sleep. The goal was to cover all activities across the last 24 h (sleep diary 1) or since the last sleep diary was filled out (sleep diary 2–14). A full overview of the sleep diaries is available at the project’s webpage (www.nfa.dk/1001nights). In total, 13,768 sleep diaries were completed (hereof 40 partially completed) yielding a response rate of 92.5%; 803 participants completed all 14 sleep diaries.

Fig. 2

Example of timing of collection of diary data after a day and a night shift. Participants were instructed to fill out the sleep diary after their primary sleep covering all activities across the 24 h day (sleep diary 1) or since last sleep diary was filled out (sleep diary 2–14)

Coding of working hours

Based on information on shift start and shift end reported in the sleep diaries, the participants’ working hours were categorized into shift types. Shift types were classified as “day shift” (≥ 3 h between > 06:00 and < 21:00); “evening shift” (≥ 3 h between ≥ 18:00 and < 02:00); and “night shift” (≥ 3 h between ≥ 23:00 and ≤ 06:00) [17]. The definitions were not mutually exclusive, and the categorization was prioritized in the following order: night > evening > day. Shifts with a duration of less than 3 h were not categorized. Some participants worked two shifts within a 24 h day, e.g., a day shift followed by a night shift. Days off included all days without work, including recovery days (i.e. days without work after a night shift), vacation, and sick leave.

During the study period, the 13,768 sleep diaries, provided us with information about 4553 day shifts, 997 evening shifts, 1963 night shifts, 77 short shifts (< 3 h), 14 work periods that could not be categorized due to missing information, and 6458 days without work; 294 days had two periods of work registered (i.e. double shifts).

Saliva samples

In total, 286 participants, stemming from two different subsamples, collected five saliva samples across one 24-h day. Saliva sampling was conducted when waking up (before brushing teeth), 45 min after awakening, 4 h after awakening, 10 h after awakening and right before going to bed (before brushing teeth). The participants were advised not to change their daily routines during the sampling and were instructed not to collect the 10-h sample, if they had already gone to bed.

The first subsample (n = 235, inclusion criteria B, Table 1) were instructed to tip a cotton tampon into the mouth without touching it with the fingers, chew it for approximately two minutes, and then spit it back into the tube. Two tubes (Salivette®, one cotton swab and one synthetic swab, both neutral, SARSTEDT) were used at each measurement point. The second subsample (n = 51, inclusion criteria C, Table 1) were instructed to collect saliva samples by salivating directly into a single tube (conical tube, without swab, SARSTEDT), ensuring that there was at least 1 ml of clear saliva in each tube for all five collection times.

All the participants were asked to note the specific sampling time on a label on the saliva tube and to fill out a log book answering four questions regarding their intake of coffee and cigarette smoking within the past four and past two hours, respectively, leading up to the sampling. Participants were instructed to store the samples in a refrigerator or freezer until all samples were collected and returned via postal mail together with the log book. When received at NFA, saliva samples were stored at -80˚ Celsius. All log books were digitalized and validated by two independent researchers.

Technical measurements of sleep (wrist actigraphy)

999 participants wore an accelerometer (Actigraph xGT3x-BT) for 14 consecutive days following the examination day. The Actigraphs were set up with a sampling rate of 30 Hz, and data were averaged across epochs of 60 s. The participants were instructed to wear the Actigraph on the wrist of their non-dominant hand for as much of the 24-h day as possible and most importantly during sleep. For hygienic reasons most participants took off the Actigraph during working hours as well as when showering.

Technical measurements of physical activity (accelerometers on thigh and back)

We received accelerometer (SENS Motion) data from 698 participants, who wore them for seven consecutive days. The participants had one accelerometer placed in the middle of the upper back and one placed in the middle of the right thigh patched by a researcher on the examination day. The accelerometer had a sample rate of 12.5–25 Hz (depending on the device) allowing physical activities such as walking, running, sitting and forward bending to be estimated. The algorithms are a further development of the algorithms described by Skotte et al. [18]. The SENS device also measured skin temperature.

Light exposure measurements

307 participants wore a device measuring light exposure (HOBO Pendant® Temperature/Light Data Logger) for seven consecutive days following the examination day. The device had a sampling rate of 1 per 120 s. The device was attached to the participants clothing with a clip and participants were instructed to attach it by clipping it on the outermost layer of clothing in shoulder/collar height to approximate retinal light exposure. The participants were instructed to wear the device as much as possible during the examination period including both work and leisure time. When sleeping the device should be placed next to the bed with the measuring side facing up. If the participants did not wear the device for a longer period of time (e.g. forgetting to wear it for more than 15 min), they were instructed to note it in a log book. Upon return, all log books were digitalized and validated by two independent researchers.

Interstitial glucose measurements

51 participants wore a glucose sensor (DEXCOM G6), measuring glucose levels in the interstitial fluid for seven consecutive days following the examination day. The DEXCOM G6 system records the glucose level every 5 min. The sensor was inserted underneath the skin surface on the abdomen or upper arm by trained staff and attached to the participant’s skin using adhesive tape. Data were transmitted to a receiver from the sensor. The receiver should be within 6 m of the sensor to allow transmission and recording of data. The participants were instructed to adhere to their usual daily habits, especially timing and content of meals, while wearing the device. Data collection was non-blinded, and the participants could receive alarms in the case of very low glucose levels. Data was extracted from the receiver using Glooko® software.

Linkage to the Danish Working Hour Database (DAD) and national health registers

The cohort is nested in DAD, which contains payroll data for all employees in the five administrative regions in Denmark (www.nfa.dk/dad) from 2007 and onwards (currently until 2020) and it is regularly being updated [17]. The database contains daily information on working hours, overtime and absence, together with information on workplace, job type, department and date of employment. Linkage to DAD via personal registration numbers implies that data from questionnaires, blood samples and technical measurements will be supplemented with detailed information about future and historical working hours potentially dating more than 15 years back in time and allowing for a precise exposure assessment. Likewise, the cohort will be linked to Danish health registers containing information about somatic and psychiatric diagnoses [19, 20] and redemption of medication [21].

Collaboration

The data collection for at a subsample of 322 participants are part of the EU-project The Exposome Project for Health and Occupational Research (EPHOR, www.ephor-project.eu). Together with data collections in Spain, Sweden and the Netherlands, we followed the study protocol for the shift work part of EPHOR, which implied additional self-reported measures, and collection of saliva samples and light exposure.

The research team welcomes further national and international collaborations. For more information, please visit the project’s webpage (www.nfa.dk/1001nights) or contact the corresponding author.

Statistical analyses

First, sociodemographic and health-related characteristics of the 1001 nights-cohort were compared across four groups based on their self-reported contractual work schedule (i.e. permanent day work; permanent evening work or 2-shift work without night work; permanent night work; 2- and 3-shift work with night work). Percentages and means with their standard deviation (SD) are presented in Table 2.

Table 2 Description of the study population of the 1001 nights-cohort. 994 participants responded to the questionnaire, and participants with missing data are excluded analysis by analysis

Second, for the group of participants with night work, we describe the self-reported characteristics of their work schedules and preferred working hours (Table 3).

Table 3 Description of night shift workers of the 1001 nights-cohort

Third, we compared the study population of the 1001 nights-cohort with the source population based on data from DAD. For this comparison, DAD was delineated to the inclusion criteria of the 1001 nights-cohort (i.e. women, ≥ 18 of age, employed ≥ 28 h a week, no absence due to pregnancy in 2019 or maternity leave in 2020). We used 2019-data to avoid COVID-19-related anomalies in working hours. For this comparison, we extracted information on age, average weekly working hours, job groups and work schedules (Table 4).

SAS 9.4 and R were used for the statistical analyses.

Table 4 Comparison of the 1001 nights-cohort with the source population. The comparison is based on 2019 data, and the same register-based inclusion and exclusion criteria are applied on the two groups for this comparison