ORIGINAL ARTICLE
Predictors of occupational burnout among nurses: a dominance analysis of job stressors Ji-Wei Sun, Hua-Yu Bai, Jia-Huan Li, Ping-Zhen Lin, Hui-Hui Zhang and Feng-Lin Cao
Aims and objectives. To quantitatively compare dimensions of job stressors’ effects on nurses’ burnout. Background. Nurses, a key group of health service providers, often experience stressors at work. Extensive research has examined the relationship between job stressors and burnout; however, less has specifically compared the effects of job stressor domains on nurses’ burnout. Design. A quantitative cross-sectional survey examined three general hospitals in Jinan, China. Method. Participants were 602 nurses. We compared five potential stressors’ ability to predict nurses’ burnout using dominance analysis and assuming that each stressor was intercorrelated. Results. Strong positive correlations were found between all five job stressors and burnout. Interpersonal relationships and management issues most strongly predicted participants’ burnout (113% of average variance). Conclusion. Job stressors, and particularly interpersonal relationships and management issues, significantly predict nurses’ job burnout. Relevance to clinical practice. Understanding the relative effect of job stressors may help identify fruitful areas for intervention and improve nurse recruitment and retention.
Key words: burnout, dominance analysis, job stressors, nurses, relative weight
What does this paper contribute to the wider global clinical community?
• Strong, positive correlations were
•
•
found between all examined dimensions of job stressors and burnout. Dominance analysis was used to examine the core research question. Interpersonal relationships and management issues most strongly predicted nurses’ burnout. A greater understanding of these stressors may lead to interventions that effectively reduce nurse burnout. Hospital management should be strengthened: organisational leaders are encouraged to create a positive working environment, help nurses reduce their perceived workload and remove organisational constraints.
Accepted for publication: 21 January 2017
Introduction Burnout is a psychological syndrome associated with job stressors and personal factors (Garrosa et al. 2008, Wang et al. 2015). Despite an emerging interest in personal factors Authors: Ji-Wei Sun, MB, RN, MMed Candidate, School of Nursing, Shandong University, Jinan; Hua-Yu Bai, BSN, RN, MMed Candidate, School of Nursing, Shandong University, Jinan; Jia-Huan Li, BS, School of Nursing, Shandong University, Jinan; Ping-Zhen Lin, MB, RN, MMed Candidate, School of Nursing, Shandong University, Jinan; Hui-Hui Zhang, MB, RN, MMed Candidate, School of Nursing, Shandong University, Jinan; FengLin Cao, PhD, RN, Associate Dean and Professor, School of Nursing, Shandong University, Jinan, China
© 2017 John Wiley & Sons Ltd Journal of Clinical Nursing, doi: 10.1111/jocn.13754
influencing burnout, environmental predictors (e.g. stressor) also play a contributing role. Work-related stress affects emotional health and work–life balance. Most research examining burnout has used multiple regression analysis to assess unique variance contributions to observed burnout. Correspondence: Feng-Lin Cao, School of Nursing, Shandong University, No. 44 Wenhua Xi Road, Jinan, China. Telephone: +86 531 88382291. E-mail:
[email protected]
1
J-W Sun et al.
Regarding interrelated predictive factors (which were assumed in this), single factors’ predictive power may be less than combinations of predictors. Diverging analytical concerns may underlie conflicting findings regarding stressors’ effect on nurses’ burnout (Hawley 1992, Kohler 1992, Li & Liu 2000). For example, overload is the most frequently cited workplace stressor among critical care nurses (Li & Lambert 2008); however, patient care is the most stressful (Wang et al. 2015). Little research has quantitatively compared several job stressors’ effect on nurse burnout. This research therefore compared the relative weight of job stressors on nurse burnout to inform interventions that effectively prevent nurse burnout and thereby improve nurse retention.
Background Burnout includes a loss of enthusiasm for work (emotional exhaustion, EE), negative feelings and cynical attitudes (depersonalisation, DP), and low sense of personal accomplishment (PA; Maslach et al. 2001, Maslach & Leiter 2008). Unlike depression, burnout refers specifically to a person’s relationship to his or her occupation and usually results from long-term exposure to occupational stress (Folkman & Greer 2000, Ruotsalainen et al. 2016). Burnout is particularly prevalent among human services professionals whose positions involve interactions with people (Felton 1998). In addition, burnout may lead to adverse outcomes such as medical errors, suicide, depression and absenteeism (Dimou et al. 2016). Nurses are a key group of health service providers; nursing involves experiencing job stressors that may cause exhaustion and thereby affect nurses’ mental health (Purcell et al. 2011). Nurses accordingly experience considerable burnout (Skinner et al. 2012). Nurse burnout rates range from 32% in Scotland to 54% in the USA (Kravits et al. 2010). Nurse burnout affects nurse retention rates; additionally, nursing shortages exist globally, including in the United States (AHCA/NCAL 2013), Japan (Maruyama et al. 2016) and China (Wang et al. 2015). Nurses’ working environment includes stressors such as shift work, role stress, heavy workload, clinical duties, managing critically ill patients, low support and recognition from family members, and conflicts with managers and colleagues (Shamali et al. 2015, Wang et al. 2015). Five commonly recognised job stressors affecting nurses exist: professional and career issues (PC), workload and time pressure (WT), resource and environmental problems (RE), patient care and interaction (PI), and interpersonal relationships and management issues (IM) (Dolan 1987, Gray-Toft & Anderson 1994, Li & Liu 2000, Zhou & Gong 2015). PC reflects low social status and salary, limited promotion opportunities and frequent
2
shift changes (Han et al. 2011). WT reflects heavy workloads and excessive written work (Weigl et al. 2016). RE reflects poor work environments, crowded wards and insufficient medical equipment (Kanai-Pak et al. 2008, Wu et al. 2010). PI reflects noncooperative patients, error anxiety, patients’ condition severity and impolite patient caregivers (McQueen 2000). IM reflects limited understanding and support from nursing managers or other colleagues, conflicts with doctors and nursing leaders, and mutual distrust among colleagues (Han et al. 2015, Weigl et al. 2016). These stressors are not mutually independent; to some extent, they can be aroused by a common underlying factor or they may causally relate to each other. For instance, nurses who take on more patients will likely experience greater workload, constrained time and increased interpersonal conflict. Additionally, nurses who experience shift overload are likely to also experience emotional instability and mental and physical exhaustion (Aiken et al. 2002); this may increase interpersonal conflict. Dominance analysis is therefore necessary to determine the unique and combined effects of these job stressors on nurses’ burnout. Numerous studies have used multiple regression analysis; however, the coefficients generated represent the amount of unique variance predicted by any given predictor. If the analysed predictors are correlated (which is expected regarding occupational stressors affecting nurses), regression analysis results tend towards existing model dependence; that is, the relative weights of predictive variables in a complete model will vary depending on variations in derived subset models. Dominance analysis calculates the average direct effect of each variable (only considering the variable itself), the overall effect (depending on all predictive variables in the full model) and the partial effect (depending on all other predictive variables in the subset model). Importantly, this method explains the predictive variables’ relative weights when compared all possible subset models. Each predictive variable’s relative weight may be expressed as the average percentage of the total contribution of known variance; accordingly, each predictive variable’s relative weight is more accurate and intuitive. Additionally, the sequence of relative weights is not exaggerated or reduced. In this context, this study compared job stressors’ correlation with nurses’ burnout using dominance analysis.
Methods Aim This study compared job stressors’ correlation with nurses’ burnout. © 2017 John Wiley & Sons Ltd Journal of Clinical Nursing
Original article
Design
Predictors of occupational burnout Table 1 Participant demographics
A cross-sectional design was used.
Participants Nurses were recruited from three hospitals in Ji Nan, Shandong Province, China. Purposive sampling was used. Three general hospitals were selected; we subsequently selected six units (medical ward, surgical ward, obstetrics and gynaecology ward, paediatric ward, emergency ward and intensive care ward), in proportion to the total number of nurses at each hospital.
Data collection The full sample contained 633 nurses; however, only 602 participants provided complete, usable responses (response rate: 951%). Table 1 presents participants’ demographics. Participants completed a battery of questionnaires in a fixed order. Participants self-administered a demographics questionnaire that examined age, gender, working times, highest level of education, monthly salary and type of ward. The Nurse Job Stressor Scale (NJSS, Li & Liu 2000) examines stressors affecting nurses; it is a well-established selfreport measure of occupational stressors among Chinese nurses that contains 35 items (Gu et al. 2009, Wang et al. 2015, Zhou & Gong 2015); scores use a 4-point Likert scale (1 = never have this feeling, 4 = this kind of feeling is very strong). The NJSS contains five subscales: PC (seven items, e.g. ‘I got low social status and salary’); WT (five items, e.g. ‘I have heavy workloads and excessive written work’); RE (three items, e.g. ‘there are poor work environments’); PI (11 items, e.g. ‘patients are not cooperative and patient caregivers are impolite’); and IM (nine items, e.g. ‘understanding and support from nursing managers or other colleagues are limited’) (Li & Liu 2000, Zhou & Gong 2015). The original scale’s Cronbach’s a was 098 for all items and ranged from 083–095 in each subscale (Li & Liu 2000, Zhou & Gong 2015). In this study, the NJSS’ overall Cronbach’s a was 0931; subscales’ Cronbach’s a was as follows: PC, 0819; WT, 0823; RE, 0764; PI, 0811; and IM, 0875. Scores on all of the NJSS’ items were summed to give a total score; higher scores indicated more stress. The Maslach Burnout Inventory–Human Services Survey (MBI-HSS; Leiter & Harvie 1996) is a well-established measure of burnout that contains 22 items and measures three factors of burnout; scores use a 7-point Likert scale (0 = never, 6 = every day). The Chinese translation of the
© 2017 John Wiley & Sons Ltd Journal of Clinical Nursing
n Age 602 20–29 years 448 30–39 years 131 40+ years 23 Years practicing 602 as a RN Beds/nurses 602 Sex Male 18 Female 584 Highest level of education* Less than a bachelor 141 A bachelor of nursing 457 or higher Monthly salary (dollar)*
