A growing body of evidence demonstrates that work-family conflict, defined as a bidirectional conflict between work and family roles, is an important risk factor for workers' health and well-being [Amstad, et al. 2011, Allen and Armstrong 2006]. Previous studies have focused on the impacts of work-family conflicts on workers' well-being (e.g. job satisfaction[Cortese, et al. 2010], intention to turnover[Fuss, et al. 2008], and health satisfaction[Knecht, et al. 2011]) and mental health outcomes (e.g. depression [Hammer, et al. 2005, Frone, et al. 1996, Okechukwu, et al. 2012] , substance dependence disorder[Frone 2000], and mood/anxiety disorders[Wang, et al. 2007]).

Recently, several studies have suggested that work-family conflict could be detrimental to workers' physical health outcomes as well [Clays, et al. 2009, Jansen, et al. 2006, Berkman, et al. 2010, Sabbath, et al. 2012, Ohta, et al. 2011]. For example, in a prospective study of over 6000 employees, Jansen et al showed that workers with high work-family conflict were more likely to have long duration of sickness absence than workers with low-medium work-family conflict after adjusting for confounders including work schedule and having dependent children [Jansen, et al. 2006]. Also, Berkman et al found that working with managers who are supportive about work-family need is associated with reduced cardiovascular disease risk [Berkman, et al. 2010].

Work-family conflict is of increasing concern for workers who provide direct patient care in hospitals [Grzywacz, et al. 2006] . For example, one U.S. survey of nurses showed that more than 90% experienced work-family conflict more than once over the past six months and fully half reported experiencing work-family conflict more than one day per week [Grzywacz, et al. 2006]. Previous studies have also found that work-family conflict is associated with various negative outcomes such as job dissatisfaction [Cortese, et al. 2010], lower sense of coherence [Takeuchi and Yamazaki 2010], and increase of perceived work demand[Camerino, et al. 2010] among patient care workers in hospitals.

However, little is known about how work-family conflict is related to physical health outcomes such as musculoskeletal disorders among hospital patient care workers despite their high prevalence of musculoskeletal disorders [Engels, et al. 1996]. In two studies, Hammig and colleagues investigated the association between work-family conflict and musculoskeletal pain [Hammig and Bauer 2009, Hammig, et al. 2011]. They found an association between high work-family conflict and self-reported neck/shoulder pain and back pain among workers across four different industries including the healthcare sector [Hammig, et al. 2011]. However, this study did not adjust for working conditions such as night shift work and number of work hours. These factors are particularly critical for hospital health care workers in relation to both their experiences of work-family conflict and their potential exposure to musculoskeletal pain [Camerino, et al. 2010, Trinkoff, et al. 2006, Dex and Bond 2005, Eriksen, et al. 2004]. In addition, the focus on neck, shoulder and back pain may miss other types of musculoskeletal pain common among hospital patient care workers, who are known to experience a high rate of musculoskeletal pain in other body areas such as arm and low extremities [Ando, et al. 2000, Smith, et al. 2003].

In this study, we selected potential confounders that have been associated with work-family conflict and musculoskeletal pain based on previous studies. After adjusting for those confounders including six physical work factors and four work-related psychosocial factors, we examined whether work-family conflict is associated with self-reported musculoskeletal pain (i.e. neck or shoulder pain, arm pain, low back pain, leg pain, any musculoskeletal pain, and a number of body areas in pain) in the past 3 months among hospital patient care workers.

Table I shows the distribution of the study population and the prevalence of any self-reported musculoskeletal pain in the past 3 months and the mean score of work-family conflict by key covariates. Among 1,199 hospital patient care workers, 89.9% were female, 73.9% were staff nurse, 28.4% worked some night shift. Female workers were more likely to report any musculoskeletal pain in the past 3 months compared to male workers. Workers who had a second job or who worked night shift reported significantly higher levels of work-family conflict compared to their counterparts. For the prevalence of musculoskeletal pain in the past 3 months, the highest prevalence was for back pain (55.2%) while the lowest prevalence was for arm pain (10.5%) (Table II). Overall, 76.0% of hospital patient care workers reported pain in at least one body part and 3.8% reported pain in all of four body parts (i.e. neck/shoulder, arm, low back, and lower extremity pain).

As shown in Table III, after adjusting for confounders included in the analyses, work-family conflict remained statistically significantly associated with all indicators of musculoskeletal pain in the past 3 months except low-back pain. We found the strong associations of high work-family conflict with neck or shoulder pain (OR: 2.34, 95% CI: 1.64 - 3.34), arm pain (OR: 2.79, 95% CI: 1.64 - 4.75), lower extremity pain (OR: 2.20, 95% CI: 1.54 - 3.15) and any musculoskeletal pain (OR: 2.45, 95% CI: 1.56 - 3.85), and a number of body areas in pain (OR: 2.47, 95% CI: 1.82 - 3.36) in the past 3 months in the fully adjusted model. However, the association between high work-family conflict and low back pain was attenuated and became non-significant after adjustment for potential confounders (OR: 1.22, 95% CI: 0.86 - 1.74); whereas, the association with intermediate work-family conflict remained significant (OR: 1.49, 95% CI: 1.12 - 1.98).

We found consistent associations between work-family conflict and self-reported musculoskeletal pain (i.e. neck or shoulder pain, arm pain, leg pain and any musculoskeletal pain, and co-morbidity) in the past 3 months among hospital patient care workers. This analysis adjusted for relevant confounders, including physical work factors and work-related psychosocial factors. These results extend the findings of previous research reporting that work-family conflict is associated with neck/shoulder pain and low back pain, and are consistent with prior findings that work-family conflict is associated with a variety of negative consequences including depression [Hammig and Bauer 2009], substance dependence disorder [Frone 2000], cardiovascular disease risk [Berkman, et al. 2010] and sickness absence [Clays, et al. 2009, Sabbath, et al. 2012].

Although we found that high work-family conflict was associated with most indicators of musculoskeletal pain, the association between high levels of work-family conflict and self-reported low back pain was attenuated and became non-significant in the adjusted model, whereas the association between intermediate levels of work-family conflict and low back pain remained significant. This finding is different from findings from previous studies that reported an association between work-family conflict and low back pain, and may reflect previous studies' lack of full adjustment for working conditions such as shift work, worked hours per week, job title [Hammig, et al. 2011] or other relevant psychosocial factors such as job demand, decision latitude, and supervisor support [Hammig, et al. 2009]. In addition, future research needs to examine why the relationship of work-family conflict with low-back pain could be potentially different from the relationships with other MSDs.

There could be several explanations of pathways linking work-family conflict and musculoskeletal pain. When workers experience work-family conflict, workers could be preoccupied at the workplace with home-related duties. Accordingly, it could influence psychosocial working conditions, such as job demand, which could then impact musculoskeletal pain [Lang, et al. 2012, Bongers, et al. 2002]. However, the adjustment of potential confounders including work-related psychosocial stressors (i.e. job demand, job control, supervisor support, and co-worker support), work physical factors (i.e. sitting, standing, walking, lifting and carrying, pushing and pulling), and work characteristics (i.e. job title, having a second job or not, only day shift or not, worked hours per week) attenuated less than 20% of the crude associations for most of musculoskeletal outcomes in this study, this suggests that the association would not be wholly explained by psychosocial and other working conditions.

It is also possible that psychological stress play a role in the pathway because experiences of work-family conflict could increase physiological stress reaction such as increased muscle tension which can lead to musculoskeletal pain[Lundberg, et al. 2002]. Hammig et al examined a hypothesis that the association between work-family conflict and musculoskeletal pain is mediated by general stress, showing that the association was substantially attenuated when they adjusted for general stress in addition to other covariates [Hammig, et al. 2011]. However, in a post-hoc analysis when we adjusted for non-specific psychological distress assessed by the six-item Kessler scale [Kessler, et al. 2002] in addition to other covariates, we found only slight attenuation in the associations between work-family conflict and musculoskeletal pain. Because there could be remained exposure to stress which could not be captured self-reported questionnaire, future research is required to examine the role of stress in the association between work-family conflict and musculoskeletal pain as a mediator.

This study also found that workers who worked more than 45 hours per week or who has a second job or who worked at night shift were more likely to experience work-family conflict than their counterparts, which is consistent with the findings of previous literature [Camerino, et al. 2010, Dex and Bond 2005]. We also found that staff nurse reported higher level of work-family conflict than others. The difference of work-family conflict level by job title may be explained by our finding that the prevalence of ‘having only a day shift’ is much lower among staff nurse (22.6%) compared to patient care associate (34.2%) and others (49.2%). Future studies with larger samples of different types of workers are needed to understand joint differences in work-family conflict and health-related associations by job title and other work characteristics. Future studies are also needed to investigate the impact of hospital-level policies, such as access to paid leave, on the association between work-family conflict and injury outcomes [Asfaw, et al. 2012].

Several limitations to this study should be noted. First, temporal precedence cannot be established due to the cross sectional design of the study. We cannot rule out the possibility of reverse causation in which workers with musculoskeletal pain reported more work-family conflict. Second, although we had a strong adjustment of potential confounders related to working conditions, there could be residual confounding pertaining to worker's health condition because workers with poor health conditions may be more likely to report musculoskeletal pain and also more likely to report high work-family conflict. This is particularly possible because both of exposure and outcome variables were assessed using a self-reported questionnaire.

Another limitation would be that, because of the high prevalence of several outcome variables such as low back pain and any musculoskeletal pain, the estimated odds ratio could overestimate the ratio that would have been observed had adjusted prevalence ratios been computed. However, we found strong relationship of work-family conflict with relatively low prevalent outcomes like arm pain as well as other highly prevalent outcomes , the observed strong association could not be wholly explained by the potential overestimation. Finally, work-life conflict consists of two reciprocal components: work-family conflict that work interferes with family life and family-work conflict that family interferes with work, this study only assessed work-family conflict. A previous study, which measured both of these components and examined their associations with musculoskeletal pain, found that family-work conflict is not associated with musculoskeletal pain whereas work-family conflict is a strong predictor of musculoskeletal pain[Hammig, et al. 2011], as we have observed.

There are also several strengths in this study. First, we could adjust for confounders including working conditions, other work-related psychosocial constructs, organizational practice, physical work factors, and the number of children under 5 years old at home. Second, we applied multilevel logistic regression with a random intercept at the work unit level to take into account the hierarchical data structure.

Our study found that work-family conflict could be a potential risk factor of musculoskeletal pain among hospital health care workers. These findings remained significant after adjustments for working conditions (i.e. job title, having a second job or not, working only day shift or not, hours worked per week), psycho-social factors (i.e. demand, decision latitude, co-worker support, and supervisor support), organizational ergonomic practice, and physical work factors. Work-family conflict could be proposed as an important domain for health promotion strategy among hospital patient care workers. However, it will be important for future studies to first replicate these findings with a prospective study design.