The stepwise introduction of a remodelled shift model allowed to design a comparative evaluation to assess the potential effects on work ability, quality of life and self-reported health among police officers. Our results indicate that the remodelled shift schedule may have had positive effects on these outcome parameters in the first year. The follow-up after five years suggests that the initial positive effect slowly disappears as the time working with this schedule increases.
Since the evaluation was a prerequisite for the piloting of the remodelled shift schedule and a basis for decision-making regarding its subsequent implementation, police officers’ attitude towards the remodelled shift schedule might have affected their motivation to participate in the survey and may have also biased their answers. Different response rates at the police department level may indicate this as well as the results of comparing piloting versus non-piloting police departments in the controlled before-after analysis. Indeed, the proportion of police officers willing to change the shift system was 81.3% in the group piloting the remodelled shift, compared to 57.6% in the group not piloting. The baseline scores of the officers willing to change to the remodelled shift schedule were significantly lower at baseline and increased more in the first year than those not willing to change the schedule (see supplementary Table 3). Initial positive effects after the awaited piloting may have faded away over time in the daily work routine.
Although we cannot draw conclusions on separate aspects of the shift schedule, it is plausible that the large number of over-length shifts (12-hour shifts) within the schedule may have a relevant impact in the vanishing of the initial positive effects over time. Working 12-hours or longer has been associated with higher levels of fatigue among police officers [15, 16]. Long working hours has been identified as a one of the organisational stressors that may increase the risk of emotional exhaustion and psychological distress among law enforcement officers [17]. The number of 12-hour shifts was the most frequently mentioned disadvantage of the evaluated shift schedule by the participants in the survey [18]. These long shifts were described by the police officers in qualitative interviews as being highly demanding, particularly the 12-hour overnight shifts (from 06:00 p.m. to 06:00 a.m.) [19]. However, the multivariate analysis adjusting for age (among other factors) allowed us to conclude that there is no statistically significant association between long-term work with the remodelled shift schedule and detrimental effects to self-reported health or quality of life. This might be an effect of the more frequent and longer recovery periods within this schedule. Findings indicate that schedules preventing shift worker from summing up chronic sleep deficits may reduce adverse effects of shift work irrespective of shift duration [20].
Overall, the police officers in our study showed a good work ability [21], comparable to that of a nationwide representative sample of employees aged 31–60 years in Germany (M = 40.22; SD 6.20) [22]. The work ability index improved in the first year among those working with the remodelled schedule. In the long term, we did not observe any relevant decrease in the work ability index – taking into account the effects of age – in relation to working longer with the remodelled schedule, although it is well known that the WAI has a tendency to decline with age [23]. At the individual level, the WAI is able to identify workers at high risk of long-term sickness absence [24], thus good work ability – as seen in our samples – can be considered to be a good indicator of general health. We further addressed self-rated health with a categorical question, as recommended by WHO and additionally with an ad hoc 0–10 scale. Overall self-rated health has been showed to be a strong predictor of mortality, independent of the instrument used [25]. Whereas the short-term evaluation after one year shows a higher risk of poor health among the officers working with the old shift schedule, the associations again disappears over time. The long-term analysis showed no difference in self-rated health neither with the categorical question nor with the scale rating. This results are in accordance with the analysis of routine sickness leave data of the police department, which show a positive development of this parameter over time [Herold et al., in preparation].
Compared to the available WHOQOL-Bref norm values for the whole population in Germany (67.59 SD 17.93) [12], quality of life in our sample was rated lower at the baseline (M = 59.84; SD 19.94) but was comparable in the follow-up survey (M = 67.93; SD 18.18). Considering that the norm values include age groups over 65 years which may have lower levels of quality of life, our results indicate lower levels of quality of life among police officers. A cross-sectional study among criminal police officers recently showed lower health-related quality of life scores compared to results from the general population [26]. A slight decrease in quality of life, as measured with the ad hoc 0–10 scale, was also evident in our analysis in association with longer work with the remodelled shift schedule, which, however, was not statistically significant.
Due to the incremental implementation of the remodelled shift schedule, the data can only reflect effects from a maximum of five and a half years of experience with this shift schedule. Since the observed positive effects on the short term vanished over time, the question for future research is whether working with the remodelled shift schedule over more than 5 years leads to detrimental effects on health and quality of life.
Strengths and limitations
As in other studies on occupational health, one main limitation of our study is that we cannot rule out a healthy-worker effect [27]. The questionnaire was distributed in the police stations among active officers. It is possible that during the 5 year period some officers have left their work due to health problems, which would lead to an underestimation of detrimental effects of the shift schedule. Unfortunately we did not have access to data regarding to illness-related retirements. In general, the staff of the police department has changed over time due to scheduled retirements and recruitment of staff. The average age now is 3.4 years younger than 2015, when the first survey was conducted. In addition, there has been an increase in female police officers. These demographic changes explain the differences reported in Table 3, when comparing the cross-sectional results of the three surveys. We have accounted for these confounders in the multivariate analysis.
Although the study samples at the three time points were representative for the whole target staff regarding age and gender distribution, the participation differed considerably among police stations. There were stations with participation rates under 50% and others with response rates over 70%. In addition, in the third survey we had to exclude 9% of the questionnaires, since the participants did not tick the box providing participation consent.
Finally, the small coefficients of determination (R-squared) calculated for all our multivariate models (see Table 5 and Table 6) indicate that particularly on the long-term factors relvevantly affecting the outcome parameters could not be captured with our data.