Cardiovascular Health Effects of Shift Work with Long Working Hours and Night Shifts: Study Protocol for a Three-Year Prospective Follow-Up Study on Industrial Workers

The regular shift work schedule is a “five-shift plan”. At plant A, it consists of a five-week plan with seven night shifts, in which five are of eight h duration and two are of 12 h duration. Furthermore, the schedule includes five evening shifts lasting eight hours. As for day shifts, of which there are altogether seven, the duration is eight h on weekdays, but 12 h on weekends. The schedule for plant B is similar as to that of A when it comes to the night shift, but for plant B, only four evening shifts lasting 8 h are included. This group of workers have altogether seven day shifts, of which three are of 12 h duration and four are of 8 h duration. Both plants allow workers to take extra shifts and switch shifts between workers. All workers will have five week’s annual leave in addition to days off when changing from one type of shift to another (Table 1). Normally, workers will rotate between work tasks, consisting of supervising the production by watching computer screens, regular controls and maintenance of machines, quality control of products, and driving forklifts. For the nine months leading up to the shutdown in plant B, the work schedule will be intensified for employees at this plant as they are preparing for shutdown. Therefore, the shift work schedule will include 12-h shifts on all working days for these workers in this period, before they return to their normal shift plan after the shutdown (Table 2).

At baseline, subjects will answer a self-administered questionnaire regarding education level, work experience, medication use and history, and lifestyle factors. This will include specific questions on diseases and illness, medicines, with additional focus on those affecting the cardiovascular system, e.g., blood pressure and cholesterol lowering medication, type and quantity of physical activity, smoking habits, type of shiftwork, work history, and previous shift work.

We will collect glycosylated hemoglobin (HbA1c) in Ethylene diamine tetra acetic acid (EDTA) blood. Serum for the investigation of lipids (cholesterol, low-density lipoprotein (LDL), high-density lipoprotein (HDL), and C-reactive protein (CRP) will be collected on gel tubes and centrifuged at 35 × 1000 rpm for 15 min within 60 min after the blood has been drawn from a vein. The Department of Medical Biochemistry Oslo University Hospital will analyze the samples within 48 h. HbA1c EDTA blood will be analyzed with a Tosoh G7 HPLC analyzer (Tosoh Bioscience, Inc. San Francisco, CA, USA) using the “high performance liquid chromatography” separation principle. Cholesterol, LDL and HDL in serum will be analyzed by enzymatic colorimetric method in the Cobas 8000 (Cobas 8000 Modular Analyzer Roche Diagnostics). CRP will be assed in serum using the particle-enhanced immune turbidimetric method on Cobas 8000.

We will sample nonfasting blood (tubes containing K2EDTA or serum Sep Clot Activator) at the same time of the day on all three occasions for the investigation of lipids (cholesterol, LDL, HDL), HbA1c, and CRP. The serum tubes will be centrifuged at the workplace of the participants and the blood samples will be delivered to the Department of Medical Biochemistry, Oslo University Hospital, who will analyze the samples within 24 h. Cholesterol, LDL, and HDL in serum will be analyzed by an enzymatic colorimetric method in the Cobas 8000 c702. HbA1c in EDTA blood will be analyzed with a D100 TM BioRad, which uses high performance liquid chromatography as the separation principle. Furthermore, CRP in serum will be quantified by particle-enhanced immunoturbidimetric method with a Cobas 8000 (Cobas 8000 c702).

Whole blood will be collected in EDTA plasma tubes and inverted five times prior to centrifugation (2000× g 15 min). Plasma will be collected, mixed, and aliquoted into approximately 500 μL volumes in 1.5 mL Eppendorf tubes and stored at −80 °C until further analysis. All samples will be thawed on ice and centrifuged at 10,000× g for ten minutes prior to appropriate dilution. Plasma levels of Interleukin-6 (IL-6), Monocyte chemoattractant protein-1 (MCP-1), Tumor necrosis factor-alpha (TNF-α), P-Selectin, CD40 ligand (CD40L), and Leptin will be determined using a magnetic bead-based Luminex screening six-plex assay, and by a single plex Adiponectin assay. The bead assays will be processed and analyzed with a Luminex IS 200 instrument (Bio Rad, Hercules, CA, USA). An in-house longitudinal control will be used on each plate (n = 6) to determine the intra- and inter-assay percentage coefficient of variation.

Following five minutes of seated rest, we will measure blood pressure and resting heart rate (RHR) from the subjects left arm using BpTRU^® (Bp TRU medical devices, Coquitlam, CO, Canada). For systolic (sBP) and diastolic blood pressure (dBP), the mean reading out of three measurements (for each measure) carried out in intervals of one minute will be used.

The arterial pulse waveform is the sum of the forward pressure wave and the backward propagating wave that is reflected from the periphery afterwards. We will measure central blood pressure, augmented pressure, and its index in a noninvasive manner using a SphygmoCor XCEL (Atcor Medical, New South Wales, Australia). A standard blood pressure cuff will be applied on the left arm on the radial artery to obtain pressure waveforms. The system produces a central aortic waveform from the pressure oscillations in the partially-inflated cuff [25]. In order to measure pulse wave velocity (PWV), we will place a blood pressure cuff on the right thigh approximately 200 mm from the inguinal area. We will measure the distance between the cuff over the femoral artery and the suprasternal notch, and from the suprasternal notch to the pulse of the carotid artery. Aortic PWV will be calculated from the distances between measurements points and the measured delay of time between 10 proximal and distal waveforms. Prior to the test, the participants are requested not to smoke, eat big meals, or consume alcohol, as recommended by the manufacturer [25].

Since atherosclerosis is associated both with an increased cIMT and the development of carotid plaques [26], we will use cIMT as a measure for subclinical atherosclerosis. cIMT is defined as the distance between the lumen-intima interface and the media-adventitia interface and plaque as a focal structure that protrudes into the arterial lumen of at least 0.5 mm or 50% of the surrounding cIMT value, or that demonstrates a thickness of 1.5 mm measured form the media–adventitia interface, as reported in the Mannheim consensus document [27]. Ultrasound scanning will be performed by a GE Vivid E9 system (GE Healthcare, Chicago, IL, USA) using a linear L9 array transducer. A minimum 10 mm length of both common carotid arteries at least 5 mm proximal to the carotid bifurcation will be scanned in an area free of plaques and with a clearly visible lumen–intima interface, using the semi-automated software of the Vivid system (GE Healthcare, Chicago, IL, USA). We will record the mean cIMT value of three end-diastolic images from both sides, as well as the maximal cIMT. In addition, the bifurcation and internal and external carotid arteries will be assessed for the presence of plaques. Data will be analyzed offline with the Echopac PC v12 (GE Healthcare, Chicago IL, USA). The total plaque burden of all eight carotid segments will be calculated as a score ranging from 0 to 24 as defined by ten Kate et al. [28].

Aerobic fitness will be determined by maximal oxygen consumption (VO_2max). We will test all subjects by a standardized graded test using a cycle ergometer (Monark 874 E, Monark Exercise AB, Vansbro, Sweden) while measuring oxygen uptake with a Cosmed K5 metabolism analyzer (Cosmed Srl, Rome, Italy). Starting with an initial load of 70 Watts and a cadence of 70 revolutions per minute (RPM), load will thereafter be increased by 35 Watts every minute. When the subjects fail to keep up with a cadence of minimum 60 RPM despite encouragement, they will be considered exhausted. We will use the 30 s averaged interval with the highest measurements to calculate participant’s VO_2max.

At eight weeks, data will be collected in a similar manner as for baseline, with the exception of cIMT measurement. All participants will additionally answer a self-administered questionnaire focusing on changes in medication (e.g., statins and antihypertensive medication), report their physical activity, and their present work tasks.

At the one-year follow-up, we will carry out a reexamination similar to the one at baseline excluding blood analyses, cIMT, and VO_2max. In addition, we will measure HRV in a subgroup of 15 rotating shift workers and 15 former shift workers presently engaged in day work, during a cross-shift period of five weeks after one year of follow-up. This is a noninvasive procedure that takes approximately 20 min. We will use high definition ECG and beat-to-beat blood pressure measurements and the validated software in the Task Force Monitor^® (CNSystems, Graz, Austria). This will give indices of sympathetic and parasympatetic activity by the high frequency, low frequency, and very low frequency oscillations. We will also assess time domain-derived parameters [33]. Sleep will be measured subjectively by electronically-administered questionnaires [34] sent to the participants every evening at 21:00 during the five-week period. Objective measures of sleep will be actigraphy and radar monitoring. Each participant will wear an Axivity AX3 accelerometer (Axivity Ltd., Newcastle, UK) for two weeks, which, in addition to sleep, will measure the physical activity level for that period. For five weeks, each participant will also have a bedside XeThru x4m200 radar module capable of recording body movements and respiration rate during sleep [35]. A user button on the bedside module will let the participants input exact bedtime and risetime. Parameters of sleep including total sleep time, wake after sleep onset, sleep onset latency, number of awakenings, and sleep efficiency will be inferred from the recorded signals and the user inputs, and will be used to assess overall sleep patterns during the entire shift-period of five weeks.

At this point, we will additionally measure respirable dust. This assessment includes personal sampling using a JS holdings cyclone (FH022) and a custom-built pump with flow 2.2 L/min during a full day shift among six workers at both plants for three consecutive days. The flow rate will be recorded with a rotameter (Aalborg instruments, Denmark) pre- and post-sampling. The respirable cyclone is attached in the breathing zone of the worker. Particles will be collected on 37 mm polyvinyl filters with a 5 µm pore size (Millipore). All filters are weighed before and after exposure using a microbalance Sartorious MC 5 (Sartorius AG, Göttingen, Germany) after the filters have been acclimatized in a temperature- (20 ± 1 °C) and humidity- (40 ± 2%) controlled laboratory for a minimum of one day. Direct measure of particle size distribution and number concentration will be assessed using two particle counters, i.e., an aerodynamic particle sizer spectrometer (TSI 3321) and a scanning mobility particle sizer spectrometer (classifier model TSI 3082 and detector model TSI 3756). These instruments will be placed near the production line, i.e., in the area where the workers are present when they are handling the products manually.

After two years, there will be a three-month shutdown of Plant B due to maintenance, where its workers will be on paid leave. Directly prior to, and immediately after this shutdown, we will carry out examinations of the affected subjects. Measurements will be done in a similar manner as baseline, except from measures of cIMT and VO_2max. At this time point, we will also monitor sleep quality and physical activity as in the one-year follow-up, and collect one-year retrospective detailed working hours from the employer’s register and do a repeated measure of respirable dust after the reopening of the plant.

The final data collection, identical to the one at baseline, will be carried out three years after baseline. The questionnaire will be similar to the previous ones, but will, in addition, include information on any change in lifestyle habits, such as smoking and nutrition, for the previous year.

All electronic files will be placed on an encrypted database protected by two-step verification and a firewall. All physical documentation will be stored in facilities specialized for this purpose, with a physical lock. No unauthorized personnel will have access to these files or facilities.

Based on previous experiences, we expect approximately 60% of the total workforce at the plants to volunteer for participation in the project. The project will be deeply anchored with the plant’s leadership. Participants will be closely followed by the Occupational Health Service after inclusion, and we expect 10–20% loss to follow-up for the whole 3-year period. For the intervention group, 80–90% compliance is expected for the planned and supervised sessions. Outcome will be censored in case of people leaving the employment at the plants due to unemployment, disability pension, or rehabilitate pension. Missing outcome values will be taken into account by the mixed model approach, and no prior imputation is necessary. For the adjustment variables, we do not expect much missingness.

Based on the assumption of 60% participation, we expect approximately 125 participants at baseline. Furthermore, our power calculation assumes a dropout of 10% after one year, and 20% after three years.

We consider the first research question as the primary one, and to evaluate statistical power, we focused on the effect of shift work on change in systolic blood pressure from baseline to last follow-up. We fixed the between-subject and within-subject variability to values estimated in a similar longitudinal study [37], i.e., we assumed a between-worker standard deviation of 9.3 and a within-worker standard deviation of 7.8. Furthermore, we assumed an effect size corresponding to a yearly increase of 1.5–2 units in blood pressure in the shift group compared with the reference group of day workers. Based on these assumptions, we obtained a power of 0.56–0.80 of getting a significant effect of shift work on change in systolic blood pressure. However, to better account for the fact that we have multiple outcomes, we took this power analysis one step further. We assumed that of the variables included in our study, 1/3 had an effect size of the same magnitude as for the systolic blood pressure, while for the other 2/3 of the outcomes, we assumed there were no effects of shift-work exposure. We adjusted the p-values for multiple-testing using a Bonferroni correction and obtained a power of 0.57–0.95 of getting at least one significant finding when adjusted for multiple testing.