Shiftwork, work-family conflict among italian nurses, and prevention efficacy

Chronobiology International, 25(2&3): 425–442, (2008) Copyright # Informa Healthcare ISSN 0742-0528 print/1525-6073 online DOI: 10.1080/07420520802118236 Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. FACTORS AFFECTING WORK ABILITY IN DAY AND SHIFT-WORKING NURSES Donatella Camerino,1 Paul Maurice Conway,1 Samantha Sartori,1 Paolo Campanini,1 Madeleine Estryn-Béhar,2 Beatrice Isabella Johanna Maria van der Heijden,3 and Giovanni Costa1 1 Department of Occupational and Environmental Health, University of Milano, and IRCCS Maggiore Hospital, Mangiagalli and Regina Elena Foundation, Milano, Italy 2 Service Central de Médecine du Travail Hôpitaux Hôtel Dieu AP-HP de Paris, Paris, France 3 Maastricht School of Management, Maastricht, the Netherlands, Open University of the Netherlands, Heerlen, the Netherlands; University of Twente, Enschede, the Netherlands Satisfactory work ability is sustained and promoted by good physical and mental health and by favorable working conditions. This study examined whether favorable and rewarding work-related factors increased the work ability among European nurses. The study sample was drawn from the Nurses’ Early Exit Study and consisted of 7,516 nursing staff from seven European countries working in state-owned and private hospitals. In all, 10.8% were day, 4.2% were permanent night, 20.9% were shift without night shift, and 64.1% were shift workers with night shifts. Participants were administered a composite questionnaire at baseline (Time 0) and 1 yr later (Time 1). The Work Ability Index (WAI) at Time 1 was used as the outcome measure, while work schedule, sleep, rewards (esteem and career), satisfaction with pay, work involvement and motivation, and satisfaction with working hours at Time 0 were included as potential determinants of work ability. Univariate and multivariate analyses were conducted after adjusting for a number of confounders (i.e., country, age, sex, type of employment, family status, and other job opportunities in the same area). Work schedule was not related to Time 1 changes in WAI. Higher sleep quality and quantity and more favorable psychosocial factors significantly increased work ability levels. Higher sleep quality and quantity did not mediate the effect of work schedule on work ability. No relevant interaction effects on work ability were observed between work schedule and the other factors considered at Time 0. As a whole, sleep and satisfaction with working time were gradually reduced from day work to permanent night work. However, scores on work involvement, motivation, Address correspondence to Donatella Camerino, Department of Occupational and Environmental Health, University of Milano, San Barnaba 8, 20122, Milano, Italy. Tel.: þ 39 02 50320159; Fax: þ39 02 50320150; E-mail: donatella.camerino@unimi.it 425 426 D. Camerino et al. and satisfaction with pay and rewards were the highest in permanent night workers and the lowest in rotating shift workers that included night shifts. (Author correspondence: donatella.camerino@unimi.it) Keywords Nursing staff, Work ability index, Shift work, Sleep, Psychosocial factors INTRODUCTION Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. Background Since 1870, there has been a gradual decline in the average annual working hours in Europe, North America, and Australia (Maddison, 1995), with some differences across countries in terms of the speed or intensity of the reduction. The importance of guaranteeing leisure time for workers has been increasingly acknowledged, and in 1919, the principles of the 8 h work day and 48 h work week were adopted in the Hours of Work Convention number 1 (International Labour Office, 1919). Such progress was also accompanied by the recognition of the economic value of leisure. The increased emphasis laid in Europe by trade unions and enlightened employers toward shorter working hours was mainly aimed at protecting workers’ health, preserving and/or creating new jobs, and more recently guaranteeing a better work-life balance. Against this historical trend, the current nursing shortage in Europe often requires healthcare organizations to overlook regulations concerning working hours in an attempt to guarantee adequate coverage of vacant posts. This may lead to increased working hours and non-ergonomic planning of work schedules (e.g., reduction of rest time between shifts, too many consecutive night shifts or weekends worked, low work-time predictability, etc.). Hence, the current nursing shortage may increase difficulties in facing shift-work-related problems, both for the nurses and for those in charge of their health and well-being, including employers. Impact of Working Hours on Work Ability A number of studies have concluded that inadequate work planning and a poorly organized work schedule may impact health. In particular, this may result in a reduced quantity and quality of sleep, a decline in cognitive and physical performance and an associated increased risk for errors and accidents, and interference with family and social engagements (Åkerstedt, 2003; Dorrian et al., 2006; Eriksen & Åkerstedt, 2006; Fitzpatrick et al., 1999; Oginska & Pokorski, 2006; Poissonnet & Véron, 2000). All of these negative consequences may be perceived by a worker as a reduction in his/her capacity to perform his/her work (work ability). For this reason, work ability is seen as a relevant parameter to evaluate the impact of work Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. Work Ability in Day and Shift-Working Nurses 427 schedules on individual health and performance (Takahashi et al., 2006). The disruption of circadian rhythms and sleep homeostasis (which implies both work tasks being carried out during sub-optimal physiological activation and work breaks being taken in periods and in ways not suitable for an adequate recovery) and unfavorable working conditions associated with shift work (such as lighting, temperature, an impoverishment of interpersonal relationships, and a reduction in the number of available services) may negatively influence the perception of one’s own work ability (Costa et al., 2004a; Foster & Kreitzman, 2004). As for studies specifically addressing the relationship between working hours and work ability, Costa et al. (2005) and Capanni et al. (2005) found in their cross-sectional study of healthcare personnel that work ability as assessed by the Work Ability Index (WAI) decreased more steeply with age in shift compared to day workers. Fischer et al. (2006) observed in their nursing study a significant association between sleep problems and WAI and also a dose-response relationship between fatigue and WAI. While in the univariate analysis, working one 12 h night shift followed by 36 h off was associated with a higher risk for inadequate WAI (,37 in this study), in the fully-adjusted model (also including sleep), work ability was not significantly associated with work schedule, suggesting a possible mediation of sleep in the relationship between work schedule and WAI (even if Fisher et al. did not explicitly test for this hypothesis). Psychosocial Characteristics and Shift Work: Interactive Effects on Work Ability Other studies have demonstrated that work ability is also linked to a number of work-related psychosocial factors. These include mental demands, development opportunities, satisfaction with working time, interesting job, management style, and satisfaction with work prospects and salary (Goedhard & Goedhard, 2005; Ilmarinen et al., 2005; Kerkhof et al., 2006; Sjogren-Ronka et al., 2002; Tuomi et al., 1991, 2001). These factors can be considered to be consistent with Ilmarinen’s et al. (2005) “fourth floor of the work ability house” (including physical and psychological working characteristics). Ilmarinen et al. consider good work ability as an interactive process between the ‘fourth floor’ and the three other floors or levels (i.e., health, competence, and values). Within this framework, work ability is determined by the demands of the situation and the perceived recourses available to meet them. Individuals derive a sense of self-mastery (Ben-Zur, 2002) in their “perceived ability to alter events” (Burger, 1989), a personal characteristic that may enhance individual coping skills and thus the capacity to establish a balance between one’s resources and work demands (Ilmarinen, 2006). Accordingly, rewarding characteristics of the job may be considered as 428 D. Camerino et al. Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. buffers against the negative effects on work ability related to shift work. Specifically, this means that for those experiencing higher compared to lower levels of rewarding job characteristics, the adverse impact of shift work on work ability may be reduced. One limitation of the research examining the relationship between work schedule and work ability is that it is mainly explored via cross-sectional designs, which reduces the ability to provide conclusive evidence concerning causal effects. To remedy this, our study used a prospective design to examine causal relationships. Study Hypotheses The first aim of our study was to test the following hypothesis: At the 1 yr follow-up time, day workers, as compared to shift workers, with favorable conditions, such as satisfaction with working hours, overall reward (and more specifically esteem and career reward), satisfaction with pay, and work involvement and motivation, demonstrate an increased work ability (H1). Moreover, drawing upon Fischer et al.’s study (2006), we tested the following hypothesis: At the 1 yr follow-up, the quantity and quality of sleep will mediate the effect of work schedule on work ability (H2). We earlier indicated the possible buffering effects of work that contains rewarding characteristics. Therefore, we also tested the following hypothesis: At the 1 yr follow-up, favorable conditions, such as quantity and quality of sleep, satisfaction with working hours, overall reward (and more specifically esteem and career reward), satisfaction with pay, and work involvement and motivation moderate the adverse effects on work ability potentially associated with shift-work (H3). Finally, because perceived psychosocial factors may vary according to work schedule (Härmä, 2006), we also evaluated if differences between shift patterns may exist in the rewarding components of the job, such as satisfaction with working hours, esteem and career reward, pay, and work involvement and motivation. MATERIALS AND METHODS Sample and Procedure The present study is based on data collected from the longitudinal Nurses’ Early Exit Study (NEXT study). The study was conducted in several European countries between 2002 and 2004 and aimed at identifying reasons to explain the premature departure from the nursing profession. The NEXT study complies with the ethical Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. Work Ability in Day and Shift-Working Nurses 429 requirements of the journal (Touitou et al., 2006) and was approved centrally by the University of Wuppertal (Germany) and locally at each national study center. In each country, participants were selected using a stratified sampling procedure to reflect the national distribution of nursing staff by type of institution (hospital, nursing home, and homecare service), geographical spread, and ownership (public or private). A prospective questionnaire-based design was used for data collection, and a self-administered questionnaire was distributed at baseline (Time 0) and follow-up (Time 1). The first assessment was done between October 2002 and June 2003, depending on the countries’ study planning, while in each case the follow-up assessment was conducted 1 yr after baseline. In most countries, questionnaires were sent to participants via the institution’s internal mailing system. On a few occasions, direct posting to the participants’ home addresses was done; in these cases, a prepaid envelope was supplied for the return of the survey. Participants were assured about their anonymity via a complex individual code. The code was necessary to match cases over the two assessments. More detailed objectives and procedures of the NEXT study have been described elsewhere (Hasselhorn et al., 2003, 2006). The sample consisted of nursing staff only and excluded those holding management positions. The nurses were employed in stateowned and private hospitals in Belgium, Germany, France, Italy, The Netherlands, Poland, and Slovakia. This case selection was done so as to increase homogeneity in working conditions. A total of 18,726 replies were received at Time 0 and response rates varied between 41.3– 75.8% across the countries. At Time 1, some 7,516 replies were received, resulting in an overall response rate at follow-up of 41.7%. Those who did not respond at Time 1 included participants who left their institution, were no longer able to take part in the investigation for whatever reason, or had lost interest in the research. To ascertain if drop-outs from the study was selective, we ran several analyses on a number of baseline variables to compare those who were respondents with those who were non-respondents at follow-up. The proportion of participants who dropped out from the study, compared to those who took part at Time 1, was higher for the age group .45 yrs (20.4% vs. 16.9%), for day workers (39.3% vs. 31.4%), for temporary contract workers (9.8% versus 6.4%), and for those perceiving difficulty in finding a new job (64.2% vs. 56.6%), while it was lower for male workers (11.8% vs. 14.2%). These results suggest a possible selection based on the healthy worker effect, mostly if one considers that older people, women, workers deemed not fit for shift work, temporary contract workers, and workers with low expectations of finding new jobs are usually those having the poorest health status. Finally, it should be noted 430 D. Camerino et al. that in the multivariate analyses, sample size decreased due to missing values for the variables considered. Measures Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. Baseline Predictors Work Schedule. For the present study, the responses to type of work schedules were categorized into day work, shift work without nights, shift work with nights, and night shift only. Sleep. Sleep was assessed using four items. The questions ask the respondents to assess sleep quantity (two items) and sleep quality (two items). An overall sleep scale was created by calculating the mean of the four items. The score of sleep quantity and quality scale ranged from 1 to 5, with higher scores indicative of more restorative sleep. One missing item per participant was tolerated for scale construction. For cases with one missing value, the mean of the remaining three items was substituted. The sleep quality and quantity scale obtained a Cronbach’s alpha of 0.78. Reward. Reward was measured using the reward component of the Effort-Reward Imbalance Questionnaire (ERI-Q, Siegrist & Peter, 1996). The ERI-Q has been previously proven to have optimal (Siegrist et al., 2004) and predictive validity (van Vegchel et al., 2005). Reward was measured by 11 items, covering three components of reward (i.e., esteem reward, five items; career reward including job security, five items; and pay, one item). The overall reward score, ranging from 11 (“lowest reward”) to 55 (“highest reward”), was formed by summing up the scores obtained on the eleven individual items. The scores of the reward components esteem reward and career reward ranged from 5 (“lowest reward”) to 25 (“highest reward”). Two missing items per participant were tolerated for scale construction. The individual mean calculated on the data provided was used to replace missing values. The scales overall reward, esteem reward, and career reward obtained a Cronbach’s alpha of 0.80, 0.76, and 0.64, respectively. Satisfaction with Pay. Satisfaction with pay was assessed using a threeitem measure (e.g., “How satisfied are you with your pay in relation to your need for income?”). Answers were given on a five-point scale ranging from “not at all satisfied” to “very much satisfied.” No missing items were tolerated for scale construction. Cronbach’s alpha for this scale was 0.79. Single-Item Questions. Satisfaction with working hours was assessed with the following item: “All in all, are you satisfied with your working time with respect to your well being?” The response options were “yes” or “no.” Work involvement and motivation were assessed by the item: “Do you feel motivated and involved in your work?” Responses were Work Ability in Day and Shift-Working Nurses 431 made on a five-point scale (from “to a very small extent” to “to a large extent”). Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. Outcome Measure Perceived work ability was measured using the WAI (Ilmarinen & Tuomi, 2004). The total score was calculated by summing the scores across the seven items of the WAI (Tuomi et al., 1998). The range for the total score was 7– 49 points, with higher scores indicative of higher perceived work ability. Work ability was categorized using the original cut-off points: poor (7 –27), moderate (28– 36), good (37– 43), and excellent (44– 49). Internal validity of the WAI has been demonstrated (Eskelinen et al., 1991; Nygård et al., 1991), and the instrument has shown stable test – retest reliability (De Zwart et al., 2002). In this study, however, we adopted a short version of the WAI. This version differs from the original WAI instrument in that item three contains only 15 of the 51 medical conditions contained in the full WAI, thus improving face validity. Nübling et al. (2004) developed an algorithm to allow comparability of the data obtained by the two versions and found a good convergent validity. Confounders Based on the existing WAI literature (e.g., Tuomi et al., 1991) and also on other analyses of the data from the NEXT study (Hasselhorn, 2003), the following confounders were included: country, age (continuous), gender, type of employment contract (fixed/temporary), family status (living alone, living as the only adult with child/children, living with another adult, living with another adult and child/children), and availability of job alternatives (five response categories). For the present study, the variable “availability of job alternatives” was dichotomized into “difficult finding job alternatives” and “easy finding job alternatives.” Age was entered in the regression model as a continuous variable. Using age as a category and in a quadratic form did not provide a better fit to the data. Statistical Analysis To test univariate associations between WAI at Time 0 and the other study variables, AVOVA tests or Pearson’s bivariate correlations were completed, depending on the scale of measurement. ANOVA or chi-square tests were also conducted to evaluate the associations of work schedule with sociodemographic and work-related factors. In the case of the ANOVAs, we conducted Tukey’s post-hoc tests to test for homogeneous subsets of means. Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. 432 D. Camerino et al. A hierarchical linear regression model was fitted to test the multivariate effects of the study variables on WAI at Time 1. Variables were entered into the regression model in four sequential steps. In Step 1, we introduced WAI at Time 0 and only the confounders (i.e., country, gender, age, family status, employment contract, and availability of job alternatives; job seniority was not included for its high collinearity with age) in order to assess the independent effect of predictors on WAI change over the two time points. In Step 2, the study predictors measured at baseline (work schedule, satisfaction with working hours, overall reward [and esteem and career reward separately], satisfaction with pay, work involvement and motivation) were then entered into the regression model (H1). In Step 3, sleep quality and quantity was then entered to control for a possible mediation effect of this variable in the relationship between work schedule and WAI at Time 1 (H2). In the final step (Step 4), the interaction terms were added to the model (H3). Each interaction was tested in separate regression models (in Table 3, interactions are displayed together for space reasons). F-change tests were calculated to assess the contribution given to the explanation of the outcome by each specific group of variables included in the sequential steps. All analyses were conducted using the statistical package SPSS 14.0 (SPSS, Inc., Chicago, Illinois, USA). RESULTS Descriptive Analysis As shown in Table 1, nearly one-third of the sample consisted of nursing staff from Italy, while the other countries contributed between 6.6% and 16.8% to the total sample. The overall sample was mainly women (85.8%), with age ranging from 18 to 63 yrs (20.1% were .45 yrs of age) and job seniority in nursing ranging from 1 to 43 yrs. More than one-half of the nursing staff lived with another adult plus child/children (56.9%). The majority held a fixed job post (93.6%). As for work schedule, 10.8% were day, 4.2% permanent night, 20.9% shift without nights, and 64.1% shift with night work. According to Table 1, WAI at Time 0 was significantly associated with all the baseline variables. Specifically, WAI was significantly lower in nurses
45 yrs than in those 45 yrs (poor: 6% vs. 3%; moderate 28% vs. 23%; good 44.4% vs. 50.7%; excellent 21.7% vs. 23.3%) and lower in permanent night workers compared to nursing staff involved in the other shift patterns, even after adjustment for age (F(3,6430) ¼ 3.0, p ¼ .03). Moreover, we did not find any significant interaction between age and work schedule on the WAI. WAI means at Time 0 and at Time 1 were 39.0 (SD ¼ 5.7) and 39.2 (SD ¼ 5.5) and Pearson’s bivariate correlation 0.61 ( p , .001). We combined the four WAI categories into two by TABLE 1 Socio-demographic and Work-Related Characteristics of the Study Sample and Associations With the Work Ability Index (WAI) at Time 0 Sample† Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. Variables (T0) N % or mean (SD) WAI (T0) mean or Pearson’s bivariate correlation Socio-demographic characteristics Country (%) Belgium Germany France Italy The Netherlands Poland Slovakia 589 1025 953 2460 730 1260 499 7.8% 13.6% 12.7% 32.7% 9.7% 16.8% 6.6% 39.4 (4.8) 38.1 (6.1) 38.6 (5.3) 39.8 (5.3) 41.8 (4.8) 37.6 (5.8) 40.0 (4.6) F(6,6494) ¼ 58.84 ( p . .001) Gender (%) Female Male 6448 1063 85.8% 14.2% Age (mean) 7479 37.3 (8.1) Seniority in nursing (mean) 7465 14.5 (8.3) 39.0 (5.5) 40.4 (5.4) F(1,6496) ¼ 48.99 ( p . .001) r ¼ 2.13 ( p , .001) r ¼ 2.14 ( p , .001) Family status (%) Alone Only adult together with child/children With another adult With another adult and child/children 920 404 12.5% 5.5% 39.1 (5.7) 38.1 (6.2) 1853 4199 25.1% 56.9% 39.3 (5.6) 39.3 (5.4) F(3,6412) ¼ 5.16 ( p ¼ .001) Work schedule (%) Day work Shift work without nights Shift work with nights Only night shift 802 1560 4775 311 10.8% 20.9% 64.1% 4.2% 39.2 (5.8) 38.8 (5.7) 39.4 (5.4) 38.4 (5.5) F(3,6451) ¼ 4.94 ( p ¼ .002) Employment contract Fixed Temporary 6996 481 93.6% 6.4% 39.1 (5.5) 40.1 (5.4) F(1,6466) ¼ 12.04 ( p ¼ .001) Availability of job alternatives (%) Difficult finding job alternatives Easy finding job alternatives 4208 3221 56.1% 43.9% 38.6 (5.7) 39.9 (5.3) F(1,6427) ¼ 80.02 ( p ¼ .001) (continued) 433 434 D. Camerino et al. TABLE 1 Continued Sample† Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. Variables (T0) N % or mean (SD) Study predictors Satisfaction with working hours (%) Yes No 4853 2473 66.2% 33.8% Sleep (1–5) Reward (11–55) Esteem reward (5– 25) Career reward (5–25) Satisfaction with pay (1–5) Work involvement and motivation (1– 5) 7414 7332 7394 7333 7136 7425 3.4 (0.8) 43.8 (8.1) 212 (4.3) 20.0 (4.2) 2.2 (0.9) 3.9 (1.1) WAI (T0) mean or Pearson’s bivariate correlation 40.1 (5.10) 37.3 (5.9) F(1,6367) ¼ 405.68 ( p . .001) r ¼ .33 ( p , .001) r ¼ .31 ( p , .001) r ¼ .27 ( p , .001) r ¼ .26 ( p , .001) r ¼ .15 ( p , .001) r ¼ .23 ( p , .001) Equal number of asterisks indicates homogeneity of means among groups according to Tukey’s posthoc test (by country, the following homogeneous subsets were found: Poland–Germany, Germany– France, France–Belgium, Belgium– Italy–Slovakia, The Netherlands). † Total number of cases may vary across variables owing to different number of missing values. adding the poor and moderate categories, and the good and excellent categories. This showed that 23.1% of the sample experienced a change in WAI after the 1 yr time span; 10.8% improved and 12.1% decreased their WAI. As shown in Table 2, at baseline a slightly higher percentage of men than women were involved in night work. Day and permanent night workers reported higher age and seniority in nursing. All study predictors were associated with work schedule. In general, all predictor scores were significantly less favorable among the nursing staff working shifts, including nights. Interestingly, staff working permanent night shifts, while showing the poorest sleep (as expected) and the second lowest satisfaction with working hours, reported the highest scores for reward (both for overall reward and for esteem and career reward separately) and satisfaction with pay. Multivariate Analysis Table 3 reports the results of the hierarchical linear regression. Step 1 included all confounders plus WAI at Time 0. This model significantly improved the explanation of WAI at Time 1 compared to a null model with no independent variables included ( p , .001). Working as a nurse in The Netherlands, being younger, and living as the only adult with 435 Work Ability in Day and Shift-Working Nurses TABLE 2 Prevalence (%) or Means (and Standard Deviation in Brackets) of Work-Related Variables by Work Schedule Work schedule Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. Socio-demographic and work-related variables N (overall) Gender Female Male Day work 802 Shift work without nights Shift work with nights Only night shift 1560 4775 311 11.3% 7.8% 21.3% 18.7% 63.0% 70.9% Age (yrs) 40.4 (8.1) 38.4 (8.6) 36.1 (7.7) Seniority in nursing (yrs) 17.6 (8.1) 15.5 (8.8) 13.5 (8.0) 4.4% 2.6% x2 ¼ 29.47 ( p , .001) 40.5 (7.8) F(3,7410) ¼ 102.68 ( p , .001) 17.6 (7.9) Unsatisfied with working hours 16.0% 25.3% 39.7% F(3,7393) ¼ 86.57 ( p , .001) 31.2% Satisfaction with pay (1–5) 2.5 (0.91) 2.4 (0.97) 2.1 (0.88) x2 ¼ 232.14 ( p , .001) 3.2 (0.81) F(3,7351) ¼ 41.95 ( p , .001) 46.3 (6.53) F(3,7274) ¼ 38.06 ( p , .001) 21.8 (3.5) F(3,7333) ¼ 83.27 ( p ¼ .004) 21.2 (3.4) F(3,7279) ¼ 36.29 ( p , .001) 2.7 (0.86) Work involvement and motivation (1–5) 4.0 (1.04) 4.0 (1.03) 3.8 (1.07) F(3,7082) ¼ 94.44 ( p , .001) 4.2 (0.87) Sleep (1–5) 3.7 (0.76) 3.5 (0.77) 3.4 (0.78) Reward (11–55) 451 (7.86) 44.9 (7.93) 43.1 (8.14) Esteem reward (5–25) 21.5 (4.1) 21.1 (4.4) Career reward (5–25) 20.4 (4.2) 20.6 (4.0) / 19.6 (4.3) 21.3 (4.3) / F(3,7358) ¼ 29.62 ( p , .001) Equal number of asterisks across columns indicates homogeneity of means among groups according to Tukey’s post-hoc test. children or with another adult with children (both compared to living alone) were significantly related to an increase in WAI 1 yr later. Overall, predictors included in step 2 significantly improved the regression model ( p , .001) compared to step 1, which only included the confounders. A 1 yr increase in WAI was significantly predicted (after adjustment for baseline WAI and confounders) by a higher work involvement and motivation, a higher satisfaction with pay, higher overall reward, and higher esteem and career reward at Time 0. However, in the multivariate model, work schedule was not related to changes in WAI scores over the 1 yr interval considered. Note that work schedule was not related to changes in WAI even in a model with no adjustment for all the other variables entered in step 2 (analysis not shown in Table 3). These results provide some support for the first hypothesis, with the only exception being work schedule and satisfaction with working hours. TABLE 3 Hierarchical Linear Regression Analysis for the Effects of Work-Related Variables (Time 0) on Work Ability Index (WAI) at Time 1 Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. Variables Step 1 WAI T0 Country (ref: The Netherlands) Germany France Italy Poland Slovakia Belgium Gender (ref: male) Female Age Family status (ref: alone) Only adult together with child/children With another adult With another adult and child/children Employment contract (ref: fixed) Temporary Availability of job alternatives (ref: easy finding job alternatives) Difficult finding job alternatives Step 2 Work schedule (WS) (ref: day work) Shift work without nights Shift work with nights Only night shift Satisfaction with working hours (ref: unsatisfied) Reward (11–55)† Esteem reward† Career reward† Satisfaction with pay (1 –5) Work involvement and motivation (1–5) Step 3 Sleep (1 –5) Step 4 (interaction)‡ Age  WS Satisfaction with working hours  WS Sleep  WS Reward  WS Esteem reward  WS Career reward  WS Satisfaction with pay  WS Work involvement and motivation  WS  b t p Model F change ( p) .60 51.01 ,.001 21.94 21.78 2.89 22.20 2.94 21.40 27.82 27.09 23.90 28.22 22.93 25.10 ,.001 ,.001 ,.001 ,.001 ,.001 .003 2.22 2.04 21.11 25.40 Ns ,.001 .59 .22 .50 2.14 1.05 2.57 Ns 2.09 2.35 Ns 2.26 21.71 Ns .35 .15 2.23 2.13 1.47 0.70 2.67 2.95 Ns Ns Ns Ns .03 .03 .04 .31 .18 2.88 1.96 2.34 3.79 2.73 .01 .05 .02 ,.001 .006 F(7,5334) ¼ 7.82 ( p , .001) .39 4.29 ,.001 F(1,5333) ¼ 18.36 ( p , .001) .03 .01 F(14,5341) ¼ 246.05 ( p , .001) F(3,5330) ¼ 0.20 (ns) F(3,5330) ¼ 1.92 (ns) F(3,5330) ¼ 1.91 (ns) F(3,5330) ¼ 3.76 ( p ¼ .01) F(3,5305) ¼ 2.08 ( p ¼ .04) F(3,5280) ¼ 2.89 ( p ¼ .04) F(3,5330) ¼ 0.92 (ns) F(3,5330) ¼ 1.22 (ns)  Adjusted unstandardized regression coefficient. Reward, esteem reward, and career reward were entered separately into the analysis: Step 2 model with esteem reward: F-change(7,5308) ¼ 9.36 ( p , .001); Step 2 model with career reward: F-change(7,5283) ¼ 9.43 ( p , .001). ‡ Each product term in Step 4 entered in separate regression analyses. † 436 Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. Work Ability in Day and Shift-Working Nurses 437 The inclusion of sleep quality and quantity at Time 0 in step 3 significantly improved model fit ( p , .001), with a better sleep predicting an increased WAI at Time 1 ( p , .001). Sleep did not play the hypothesized mediation role between work schedule and WAI (second study hypothesis), but this is mainly attributable to the fact that work schedule, itself, did not show any significant impact on WAI. According to step 4, only overall reward and the two reward components separately (esteem and career) significantly interacted with work schedule in predicting WAI at Time 1 ( p ¼ .01), indicating that the third hypothesis claiming an interactive effects between work schedule, on the one hand, and psychosocial factors and sleep, on the other hand, did not find consistent support in the present study. DISCUSSION The first hypothesis of our study stated that among hospital nursing staff working in seven European countries, work schedule and workrelated psychosocial factors are related to changes in work ability over a 1 yr time span. In our prospective analysis, work schedule did not significantly predict changes in WAI at Time 1. The lack of association between work schedule and work ability 1 yr later may be partly attributed to the high inter-individual variability in tolerance to shift work (and consequently in its effects on health) due to both endogenous and exogenous factors (Costa, 2003). However, it should be noted that as far as the cross-sectional data of our study are concerned, night work was linked with lower work ability even when age was adjusted. This may suggest a possible cumulative effect of night work on work ability that cannot be seen within a 1 yr interval. Notwithstanding, work ability did not show any steeper age-related decrease among shift compared to day workers, thus not confirming results of previous cross-sectional studies that found shift workers to report lower work ability compared to day workers, with increased discrepancies occurring with age (Costa et al., 2000, 2004b, 2005). This result may be accounted for, at least partly, by the “healthy worker effect” acting at two possible levels: early retirement and selection of healthier nursing staff into less favorable shift schemes. In our study, higher levels of work involvement and motivation, satisfaction with pay, esteem reward, and career reward were found to significantly increase work ability of the nursing staff at Time 1, though effect sizes were small. As for the psychosocial characteristics of the job, our findings complement results of previous studies involving non-nursing occupations in which work ability was found to be associated with psychosocial aspects of the job, such as work stress, time pressure, job control, and leadership (Ilmarinen et al., 2005; Pohjonen, 2001; Salonen Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. 438 D. Camerino et al. et al., 2003; Sjögren-Rönkä et al, 2002). In addition, our prospective study suggests that increasing rewards may promote a better work ability even in a short time frame of 1 yr. Also, restorative sleep (both in terms of quality and quantity) was found to be predictive of increased work ability, corroborating in a longitudinal perspective the association between good sleep and high WAI observed by Fischer et al. (2006) in their cross-sectional study. However, the hypothesized mediation effect of sleep on the relationship between work schedule and work ability could not be confirmed, but this finding should be considered to be inconclusive because, as previously stated, no main effects of work schedule, itself, were observed. As a whole, the third hypothesis concerning interactions was not supported in our study. In fact, apart from overall reward (and also esteem and career reward separately), there was no consistent evidence for the hypothesized modification effect of sleep and psychosocial work characteristics on the relationship between work schedule and work ability. Though statistically significant, the interaction between work schedule and reward is not discussed, as it may be a chance finding. Finally, our study also explored if levels of the potential predictors of work ability differed among nursing staff working distinct shift patterns. While sleep and satisfaction with working hours were lower in shift compared to day workers, nursing staff working permanent night shifts reported the highest rewards. This is probably related to specific incentives given by the organization that may have stimulated nursing staff to choose this particular shift pattern. In contrast, personnel involved in rotating shift schedules that included night work, which is the most usual shift pattern among healthy nursing staff, reported the least rewards. The nursing staff may suffer from a lack of voluntary choice of the type of shift scheme they can work, but it may be also that they are not offered adequate incentives for their night duties. Study Limitations The main advantage of our study resides in its longitudinal design. This makes it possible to exclude the hypothesis that the causal relationship between WAI and the psychosocial factors is reversed; that is, that work ability determines the variation in psychosocial factors and not the opposite, as is commonly hypothesized (de Lange et al., 2005). This study has some limitations, which should be taken into account when interpreting results. The interval between the two data collections may be too short to observe significant relationships between work characteristics and work ability. The obtained mean scores for the WAI were highly similar across the 1 yr period. This would partly explain the lack of prospective association in our study between work schedule and WAI. Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. Work Ability in Day and Shift-Working Nurses 439 The short time lag may also explain the reduced size of the regression coefficients due to the high stability of the outcome variable (Zapf et al., 1996). A second limitation is that the work schedule was measured only at baseline, which prevented us from tracking any changes to the work schedule at Time 1. It may be possible that some nurses changed their work schedule at Time 1, and the reported WAI score at this point reflected the new, rather than the original, schedule. Because the transition to day work may be associated with increased work ability and transition to shift work to decreased work ability, differences in the outcome may be reduced if changes in work schedule occurred during follow-up. Finally, the little evidence about interactions in our study may be partly attributable to the fact that no main effects of work schedule on WAI were observed. The above discussed limitation concerning inadequate time interval between measurements may also apply to interactions, with the additional limitation of measurement error that may increase in the presence of interaction terms, with consequences on association strength. CONCLUSIONS AND PRACTICAL IMPLICATIONS These results support the emphasis posed by European policies on increasing resources available to workers, such as sufficient time for adequate recovery, in order to protect their health and well-being. Moreover, they confirm the relevance of working conditions and adequate balances between human resources and work for sustaining and promoting good work ability, and also the need to take career stages of the workers into account and look at the different needs arising through age (Ilmarinen, 2006). Despite the current nursing shortage and its consequences on the planning of work schedules, our study suggests that healthcare organizations may find ways to support the intention of nursing staff to stay at work and to continue working until pension age. This can be achieved by activating a process aimed at guaranteeing a balance between demands and resources, mainly through dynamic, continuous, and suitable changes in working conditions (Ilmarinen, 2006). Careful attention to restorative sleep, provision of job alternatives, and career rewards (also by means of horizontal career steps), plus an organizational climate supporting personal recognition and more satisfactory pay (“fourth floor” factors according to the “work ability house”), are the main resources that, along with job involvement and motivation (“third floor”), seem to be effective, also in a short- to medium-term, for improving work ability among nursing staff. As job involvement and motivation could also be the result of favorable working conditions, further studies with longer and repeated follow-up assessments could help in clarifying the relationship between working conditions, job involvement/motivation, and work ability. 440 D. Camerino et al. ACKNOWLEDGMENTS The funding for NEXT was provided by SALTSA and the European Commission within the Fifth Framework, Project ID: QLK-6-CT-200100475, and was academically coordinated by Dr. Hans-Martin Hasselhorn from the University of Wüppertal, Germany. Web site: http://www.NEXT uni-wuppertal.de. Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. REFERENCES Åkerstedt T. (2003). Shift work and disturbed sleep/wakefulness. Occup. Med. 53:89– 94. Ben-Zur H. (2002). Coping affect and aging: the roles of mastery and self-esteem. Per. Ind. Diff. 32: 357–372. Burger JM. (1989). Negative reactions to increases in perceived control. J. Per. Soc. Psych. 56:246– 256. Capanni C, Sartori S, Carpentiero G, Costa G. (2005). Work ability index in a cohort of railway construction workers. In Costa G, Goedhard W, Ilmarinen J (eds.) Assessment and Promotion of Work Ability, Health and Well-Being of Ageing Workers. Amsterdam: Elsevier, pp. 253–257. Costa G. (2003). Factors influencing health of workers and tolerance to shift work. Theoretical Issues in Ergon. Sci. 4:263–288. Costa G, Olivato D, Antonacci G, Ciuffa V. (2000). Evaluation of functional working capacity by the work ability index in Italian workers. In Goedhard WJA (ed.) Healthy and Productive Aging of Older Employers, Aging and Work. The Netherlands: The Hague 4, pp. 53–61. Costa G, Åkerstedt T, Nachreiner F, Baltieri F, Carvalhais J, Folkard S, Dresen MF, Gadbois C, Gartner J, Sukalo HG, Härmä M, Kandolin I, Sartori S, Silverio J. (2004a). Flexible working hours, health, and well-being in Europe: Some considerations from a SALTSA project. Chronobiol. Int. 21:831–884. Costa G, Antonacci G, Olivato D, Bertoldo B, Ciuffa V. (2004b). Aging and Work Ability Index in Italian workers. In Ilmarinen J, Lehtinen S (eds.) Past, Present and Future of Work Ability. Helsinki: Finnish People and Work Research Report 65, pp. 33–40. Costa G, Sartori S, Bertoldo B, Olivato D, Antonacci G, Ciuffa V, Mauli F. (2005). Work ability in health care workers. In Costa G, Goedhard W, Ilmarinen J (eds.) Assessment and Promotion of Work Ability, Health and Well-Being of Ageing Workers. Amsterdam: Elsevier, pp. 264–269. de Lange AH, Taris TW, Kompier MAJ, Houtman ILD, Bongers PM. (2005). Different mechanisms to explain the reversed effects of mental health on work characteristics. Scand. J. Work Environ. Health 31:3– 14. De Zwart BCH, Frings-Dresen MHW, Van Duivenbooden JC. (2002). Test-retest reliability of the Work Ability Index questionnaire. Occup. Med. 52:177–181. Dorrian J, Lamond N, Van den Hervel C, Pincombe J, Rogers AE, Dawson D. (2006). A pilot study of the safety implications of Australian nurses’ sleep and work hours. Chronobiol. Int. 23:1149–1163. Eriksen CA, Akerstedt T. (2006). Aircrew fatigue in trans-Atlantic morning and evening flights. Chronobiol. Int. 23:843– 858. Eskelinen L, Kohvakka A, Merisalo T, Hurri H, Wägar G. (1991). Relationship between the self-assessment and clinical assessment of health status and work ability. Scand. J. Work Environ. Health 17(Suppl.1):40–47. Fischer FM, Borges FN, Rotenberg L, Latorre MRDO, Soares NS, Rosa PL, Teixeira LR, Nagai R, Steluti J, Landsbergis P. (2006). Work ability of health care shift workers: What matters? Chronobiol. Int. 23:1165–1179. Fitzpatrick JM, While AE, Roberts JD. (1999). Shift work and its impact upon nurse performance: current knowledge and research issues. J. Adv. Nurs. 29:18–27. Foster R, Kreitzman L. (2004). Rhythms of Life. The Biological Clocks that Control the Daily Lives of Every Living Thing. London: Profile Books, p. 315. Goedhard RG, Goedhard WJA. (2005). Work ability and perceived work stress. In Costa G, Goedhard W, Ilmarinen J (eds.) Assessment and Promotion of Work Ability, Health and Well-being of Ageing Workers. Amsterdam: Elsevier, pp. 79– 83. Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. Work Ability in Day and Shift-Working Nurses 441 Härmä M. (2006). Workhours in relation to work stress, recovery and health. Scand. J. Work Environ. Health 32:502–514. Hasselhorn HM, Tackenberg P, Mueller B. (2003). Working conditions and intent to leave the profession among nursing staff in Europe. Working Life Research Report 7. Stockholm: National Institute for Working Life, p. 258. Hasselhorn HM, Tackenberg P, Kuemmerling A, Wittenberg J, Simon M, Conway PM, Bertazzi PA, Beermann B, Buscher A, Camerino D, Caillard JF, D’Hoore W, Estryn-Behar M, Fontenla M, Gould D, van der Heijden B, Josephson M, Kiss P, Kovarova M, Kuhn K, Laine M, Le Nezet O, Lindberg P, Oginska H, Pokorski J, Pokorska J, Radkiewicz P, Rimarcik M, van der Schoot E, Stelzig S, Stordeur S, Wickstroem G, Widerszal-Bazyl M, Mueller BH. (2006). Nurses’ health, age and the wish to leave the profession—findings from the European NEXT Study. La Medicina del Lavoro 97:207–214. Ilmarinen J. (2006). The ageing workforce—challenges for occupational health. Occup. Med. 56: 362–364. Ilmarinen J, Tuomi K. (2004). Past, present, and future of work ability. In Ilmarinen J, Lehtinen S (eds.) Past, Present, and Future of Work Ability. Helsinki: Finnish Institute of Occupational Health, pp. 1–25People and Work Research Report 65. Ilmarinen J, Tuomi K, Seitsamo J. (2005). New dimensions of work ability. In Costa G, Goedhard W, Ilmarinen J (eds.) Assessment and Promotion of Work Ability, Health and Well-being of Ageing Workers. Amsterdam: Elsevier, pp. 3– 7. International Labour Office (ILO). (1919). The Hours of Work Industry Convention (1st ed). ILO, Geneva, 220 pp. Kerkhof GA, Jansen B, Van Amelsvoort LGPM. (2006). Vital working hour schemes: The dynamic balance between various interests. Chronobiol. Int. 23:1099–1104. Maddison A. (1995). Monitoring the World Economy 1820– 1992. Organization for Economic Cooperation and Development: Washington, DC, p. 255. Nübling M, Hasselhorn H-M, Seitsamo J, Ilmarinen J. (2004). Comparing the use of the short and the long disease list in the Work Ability Index Questionnaire. In Costa G, Goedhard W, Ilmarinen J (eds.) Assessment and Promotion of Work Ability, Health and Well-Being of Ageing Workers. Amsterdam: Elsevier, pp. 292–295. Nygård CH, Eskelinen L, Suvanto S, Ilmarinen J. (1991). Association between functional capacity and work ability among elderly municipal employees. Scand. J. Work Environ. Health 17(Suppl.1): 122–127. Oginska H, Pokorski J. (2006). Fatigue and mood correlates of sleep length in three age-social groups: School children, students, and employees. Chronobiol. Int. 23:1317–1328. Pohjonen T. (2001). Perceived work ability of home care workers in relation to individual and workrelated factors in different age groups. Occup. Med. 51:209–217. Poissonnet CM, Véron M. (2000). Health effects of work schedules in healthcare professions. J. Clin. Nurs. 9:13– 23. Salonen P, Arola H, Nygård C-H, Huhtala H, Koivisto A-M. (2003). Factors associated with premature departure from working life among ageing food industry employees. Occup. Med. 53:65–68. Siegrist J, Peter R. (1996). Threat to occupational status control and cardiovascular risk. Israel J. Med. Sci. 32:179– 184. Siegrist J, Starke D, Chandola T, Godin I, Marmot M, Niedhammer I, Peter R. (2004). The measurement of effort–reward imbalance at work. European comparison. Soc. Sci. Med. 58:1483–1499. Sjogren-Ronka T, Ojanen MT, Leskinen EK, Tmustalampi S, Malkia EA. (2002). Physical and psychosocial prerequisites of functioning in relation to work ability and general subjective well-being among office workers. Scand. J. Work Environ. Health 28:184–190. Takahashi M, Nakata A, Haratani T, Otsuka Y, Kaida K, Fukasawa K. (2006). Psychosocial work characteristics predicting daytime sleepiness in day and shift workers. Chronobiol. Int. 23: 1409–1422. Touitou Y, Smolensky MH, Portaluppi F. (2006). Ethics, standards, and procedures of animal and human chronobiology research. Chronobiol. Int. 23:1083–1096. Tuomi K, Eskelinen L, Toikkanen J, Jarvinen E, Ilmarinen J, Klockars M. (1991). Work load and individual factors affecting work ability among aging municipal employees. Scand. J. Work, Environ. Health 17(Suppl.1):128–134. 442 D. Camerino et al. Chronobiol Int Downloaded from informahealthcare.com by Nyu Medical Center on 03/11/14 For personal use only. Tuomi K, Ilmarinen J, Jahkola A, Katajarinne L, Tulkki A. (1998). Work Ability Index, (2nd edn). Helsinki: Finnish Institute of Occupational Health, Helinski, 37pp. Tuomi K, Huuhtanen P, Nykyri E, Ilmarinen J. (2001). Promotion of work ability, the quality of work and retirement. Occup. Med. 51:318– 324. van Vegchel N, de Jonge J, Bosma H, Schaufeli W. (2005). Reviewing the effort-reward imbalance model: drawing up the balance of 45 empirical studies. Soc. Sci. Med. 60:1117–1131. Zapf D, Dormann C, Frese M. (1996). Longitudinal studies in organizational stress research: A review of the literature with reference to methodological issues. J. Occup. Health Psychol. 1:145–169.