Work stress associated cool down reactions among nurses and hospital physicians and their relation to burnout symptoms

Conceptually, ‘Cool Down’ reactions imply two steps: 1) a person’s own perception of an emotional exhaustion in response to work stress, and 2) active attempts to find adaptive strategies in terms of emotional distancing to retain ‘functionality’ in the job [23, 24] instead of dysfunctional exhaustion as assumed for burnout. Despite conceptual similarities between Cool Down and burnout in the first phases (Fig. 1), the individual reactions may result in different end-points, i.e., either times of sickness and finally inability to work (burnout), or finding active strategies to keep ‘functionality’ in the job (cool down) which requires a reflection process and a decision to emotionally withdraw from the patients as a coping strategy. Cool down is clearly seen as a strategy to continue performing the work duties, but with a negative impact on emotional care for the patients [23, 24]. Affected work behavior (“Personal accomplishment”) is not part of the Cool Down concept, but an integral part of Maslach’s burnout concept [9, 25].

During their education nurses, medical students and physicians are often advised by experienced superiors - or they learn it on role models - to distance themselves from the suffering of their patients to protect the own emotional stability and functionality [24]. They “learn to distance themselves from recipients in order to help them better”, as others have argued, too [21]. Insofar, this strategy is ‘useful’ only for the health professionals. However, the consequence of this emotional distancing will be care as a ‘duty’, and probably lower quality of care, too. Besides a good technical performance, which is the prerequisite of professional care, the interpersonal performance is of outstanding relevance for the quality of care, i.e., the empathic and helping relationship, being focused on the patients’ needs and problems, the unseparated and attending time spend with them, respectful communication, etc. [26]. Thus, Cool Down reactions will address specifically this point of impaired personal relationship which is a consequence of a person’s strategic decision to actively cope with perceived emotional exhaustion, resulting in emotional distancing. While these individual (and relation-oriented) reactions are crucial for the Cool Down concept, burnout is mostly due to environmental factors which belong to the work setting and the inability to cope with those stressors of daily life [25]. These would result in depressive symptoms, substance abuse and potentially loss of job (Fig. 1). Apart from cynicism, adaptive strategies are not addressed in the Maslach’s Burnout Inventory (MBI).

With respect to a modified Job Demand – Resource model, Cool Down reactions – as measured with the Cool Down Index [24] – may have their causes in specific demands (i.e., aggravating factors such as work burden, stress perception, and duration of work) and are probably buffered by individual resources (i.e., protective factors such as self-efficacy expectation, situational awareness / mindfulness, team support and partnership) (Fig. 2).

Among the protective variables we see health professionals’ self-efficacy expectation which refers to a person’s ability to reach specific goals, aims or tasks, or more precisely to the beliefs about the own capabilities and conviction that one can successfully deal with / manage their responses. It is expected that a person copes better with adversive experiences or challenging tasks (such as work load or work stress) when their self-efficacy expectation is high [27]. Therefore, one may suppose that Cool Down reactions should be lower when health professionals have higher expectations of personal efficacy.

A further resource might be a persons’ situational awareness (‘mindfulness’), because it is regarded as helpful to build teamwork and to promote compassion. In healthy (non-clinical) individuals, a recent meta-analysis has confirmed “large effects on stress, moderate effects on anxiety, depression, distress, and quality of life, and small effects on burnout” for mindfulness programs [28]. In nurses, mindfulness-based stress reduction applied as an intervention program improved their coping with work stress (i.e., lower stress, burnout and anxiety, higher empathy, focus and mood) [29]. In line with this, we assume that health professionals’ (‘natural’) situational awareness (which is different from the ‘trained’ mindfulness of respective programs) can buffer their stress related Cool Down reactions.

Health care professionals’ work engagement and also life satisfaction are seen as ambivalent variables. They can be either positive variables with respect to Cool Down reactions when work is perceived as a fulfilling activity - characterized by vigor, dedication and absorption [30] - and when life in general is perceived as satisfying; or a negative variable when work engagement is low because of high work burden, stress and duration of work or when life in general is seen as less satisfying. It has been shown that work engagement is negatively associated with burnout [31], and positively related to work and life satisfaction and self-rated health [30]. In this context we see work engagement and life satisfaction as positive variables. Satisfaction with the hospital team (‘team support’) is regarded as a Cool Down buffering variable, because van der Doef and Maes [14] reported that social support may buffer some of the negative influence of given stressors. In the same vein we see partner support as a buffering variable.

One may assume that the hospital staff is aware of their work stress, and that they are able to reflect on their emotional exhaustion; some of them may decide to distance themselves from their patients. One may also assume that most are not satisfied with these emotional withdrawal reactions because they do know that this will affect the caring relation with the patient and their satisfaction with own work, too. Therefore, we intended to measure these perceptions, particularly to identify which persons among the hospital staff experience Cool Down, and further to analyze Cool Down aggravating and buffering variables as depicted in Fig. 2. We assume that 1) all hospital staff (differentiated with respect to profession and gender) experience Cool Down to a similar degree, but 2) that the perception of stress and work burden (incl. burnout symptoms) might be different between nurses and physicians and women and men, but not their Cool down reactions which are seen as an adaptive strategy. Further, we assume that 3) work stress associated variables (i.e., perceived stress and work burden, duration of work per week, and burnout symptoms) are positively associated with Cool down reactions, while internal and external resources (i.e., self-efficacy expectation, situational awareness, partner support, team satisfaction) are negatively associated and might function as a buffer. Age was tested as a further variable because one may assume that persons with higher age may have found strategies to cope with work stress, while younger persons might thus be more vulnerable to work stress.

This anonymous cross-sectional study (survey) was performed with two sample populations, i.e., nurses and hospital physicians (Table 1), and approved by the IRB of the Witten/Herdecke University (#59/2013 and #25/2014). We provided an information sheet which described the background of the study, guaranteed anonymity and assured confidentiality. All had the option to respond anonymously via an internet based questionnaire or conventional paper-pencil questionnaire. Printed questionnaires were stored at the Witten/Herdecke University. None of the hospital managers has access to these questionnaires. All data were pooled and analyzed in more general categories (women versus men; nurses versus physicians), and were not analyzed for the recruiting sources (i.e., hospitals).

Hospital physicians were informed about this study with information sheets administered to regional hospitals, or via website announcements of the Medical Association North-Rhine (Ärztekammer Nordrhein) and the medical union “Marburger Bund”. Among the physicians, 48% filled out the online questionnaire. Nurses of 24 predominantly regional hospitals and geriatric care units were informed with similar information sheets (85% from various regional hospitals and 15% from geriatric care units). Only a few completed the online questionnaire, while most completed the paper-pencil version. Inclusion criteria were profession as a nurse or hospital physician, being active at work, and willing to participate. We had no access to staff away sick.

All respondents were free to decide participating; there was neither an obligatory order by hospital managers to participate nor a control of completeness in requested participants. The sample should thus be regarded as a ‘convenience’ sample. To clarify whether the sample of hospital physicians may be representative for a communal hospital setting or not, we compared the data of physicians from a prevalence sample (regional hospital with a response rate of 47.5%) with those of the whole sample. Both groups did not significantly differ with respect to gender, age, family status, employment status, years of employment, duration of work per week, and perceived work burden (data not shown). Thus, with the exception of the specialization pattern we can regard the whole sample as representative for hospital physicians working in local hospitals. Because most nurses filled the paper questionnaire, we were able to calculate their response rates which ranged from 11 to 57% in the different hospitals (in average 28.5%). Nurses working for temporary employment agencies showed a response rate of 46.9%. All in all we enrolled 1384 persons (66% nurses, 34% physicians).

To describe the enrolled health professionals, we analyzed their profession, gender, age, family / partner status, time of employment, and duration of work per week.

To measures the Cool Down and burnout symptoms and to relate these with health professional’ demands and resources (Fig. 2), the following standardized questionnaires were administered in their German language versions:

Descriptive statistics, reliability (Cronbach’s coefficient α) and factor analyses (principal component analysis using Varimax rotation with Kaiser’s normalization) of Cool Down items, as well as analyses of variance (ANOVA) of Cool Down and burnout scores within the sample, first order correlations between Cool Down and demand and resource variables, and stepwise regression analyses to identify significant Cool Down predictors were computed with SPSS 22.0. To handle missing data, we chose the MissForest method (software R) - nonparametric missing value estimation for mixed data [38], because the dataset contains both continuous and categorical variables (mixed).

Given the exploratory character of this study, significance level was set at p < .05. With respect to classifying the strength of the observed correlations, we regarded r > .5 as a strong correlation, an r between .3 and .5 as a moderate correlation, an r between .2 and .3 as a weak correlation, and r < .2 as a negligible correlation.

Among the 1384 enrolled persons (66% nurses, 34% physicians), 69.5% were women and 30.5% men (Table 1). Among the nurses, 80.9% were female and 19.1% male, while among the physicians, 47.5% were female and 52.5% male (p < .0001; Chi²). The mean age of nurses and physicians was similar (40.6 ± 11.6 and 40.9 ± 10.7, not significant). Duration of work per week was in average 39 ± 13 h. Mean score of self-perceived work burden was 61 ± 21, indicating moderate to stronger work burden. All further data are given in Table 1.

First we intended to confirm the validity of the 9-items of the CDI (since its development, item #6 is not used at all) which in fact had a good internal consistency also in this sample (Cronbach’s alpha = 0.87) (Table 2). Exploratory factor analysis pointed to two sub-constructs which accounted for 60% of variance (Table 2), i.e., the 6-item factor “Perceived Emotional Distance” (PED; alpha = .83), and the 3-item factor “Emotional Withdrawal as a Strategy” (EWS; alpha = .80). Although the instrument is intended to be used as an index, data with the respective factors (which may add specific information for the interpretations) were addressed in the further analyses.

Because the CDI was strongly associated with the MBI (r = .54), we performed an explorative factor analysis to clarify whether both instruments cover the same topics or differential aspects. Therefore, we used both instruments’ items measuring the frequency of perceived feelings (MBI’s “Personal Accomplishment” items were not used). Exploratory factor analysis indicated 4 main factors (58% of explained variance): Factor 1 representing MBI’s “Emotional Exhaustion” subscale and two CDI items (CDI9 - I myself increasingly go short; CDI7 - I increasingly think how nice it would be to pack it all in). Factor 2 was MBI’s “Depersonalization” subscale with two of MBI’s “Emotional Exhaustion” items (MBI6 and MBI16). Factor 3 was represented by CDI’s “Perceived Emotional Distance” items with one of MBI’s “Depersonalization” item (MBI11 – I worry that this job is hardening me emotionally). Factor 4 comprises the CDI’s “Emotional Withdrawal as a Strategy” items only. Thus, the respective CDI and MBI subscales do not strongly intermix and they were differentiated as more or less independent factors.

To answer whether or not Cool down and burnout symptoms are related to health professionals’ demand and resource variables (Fig. 2), correlation analyses were performed (Table 3).

The CDI and its sub-construct “Perceived Emotional Distance” correlated moderately to strongly with MBI’s “Emotional Exhaustion” and “Depersonalization” component on the one hand, and with stress perception and work burden on the other hand. The CDI sub-construct “Emotional Withdrawal as a Strategy” correlated much weaker and moderately with these stress-associated variables (Table 3). In contrast, MBI’s “Personal Accomplishment” correlated only marginally or weakly with the CDI and its subscales. Moreover, the CDI correlated moderately negative with health professionals’ resource variables (i.e., general life satisfaction and satisfaction with team support, work engagement, self-efficacy expectation and Conscious presence and Self-Control).

There were no significant associations between CDI and age, duration of employment or duration of work per week, while particularly duration of work per week was moderately associated with burnout symptoms (Table 3).

In the sample of nurses, we also measured their satisfaction with the own ‘quality of care’ provided. This variable was moderately negative related to CDI’s “Perceived Emotional Distance” (r = −.37) and only marginally with “Emotional Withdrawal as a Strategy” (r = −.18), with burnout symptoms (r = −.42) and stress perception (r = −.36), and moderately positive with work engagement (r = .42) and best with nurses’ satisfaction with the team support (r = .49).

Next, we intended to describe the prevalence of CDI scores in the sample. The scores of the perceived frequency of Cool Down reactions showed a left skewness (mean: 21 ± 9; 25% percentile at 14, 75% percentile at 27; maximal score 54), while the perceived intensity was rather normally distributed (mean 27 ± 10; 25% percentile at 18, 75% percentile at 35; maximal score 54). Thus, the CDI (which combines frequency and intensity) showed a left skewed bulk of scores (mean: 48 ± 18; 25% percentile at 33, 75% percentile at 61; maximal score 107).

To analyze the influence of profession, gender and partner status, we performed variance analyses (ANOVA) (Table 4). While burnout, stress perception and perceived work burden scores were significantly different between women and men, gender associated differences were not found for CDI and its sub-constructs (only the intensity was marginally lower in men) (Table 4). Interestingly, MBI’s “Emotional Exhaustion” component did not significantly differ between women and men in general, but for profession. For profession, the strongest differences were found for MBI’s “Personal Accomplishment” component (Table 4). While the CDI did not significantly differ between nurses and physicians, instead burnout symptoms and perceived work burden were significantly higher in physicians, but not the stress perception itself (Table 4).

With respect to the partner status (i.e., living with or without a partner), which might be seen as a resource, the CDI scored weakly higher in persons living alone, while in contrast the MBI scores did not differ between both partner status groups (Table 4). Instead, “Personal Accomplishment” was marginally lower in persons without a partner, and their stress perception was slightly higher, too.

In an attempt to identify which variables would predict health professionals’ Cool Down reactions best, we performed stepwise regression analysis for nurses and physicians separately (Table 5). All variables were included in the regression models which were either significantly related to the CDI or differed significantly in the aforementioned analyses, or due to theoretical considerations (i.e., influence of duration of work as a putative stressor).

For nurses, “Emotional Exhaustion” was CDI’s best predictor and explains 51% of variance. Three further variables would add 6% percent of explained variance, i.e., “Depersonalization”, stress perception and work burden. Two further included variables (i.e. situational awareness and team satisfaction) explain together <0.1% of variance and are thus not of strong relevance in this model.

For physicians, “Depersonalization” (44% of variance) was the best predictor. Team satisfaction would add further 6% of variance. Four further variables were significant in this model too, and would add additional 6% of explained variance, i.e. “Emotional exhaustion”, low general life satisfaction, living without a partner and work burden.

However, work engagement, self-efficacy expectation and gender were not among the significant variables in the respective models of nurses or physicians.

A limitation of this study is its cross-sectional design which does not allow causal interpretations. So far we do not know whether the perception of an inner withdrawal in terms of Cool Down precedes the dysfunctional burnout trias, or whether they both may develop concordantly. Conceptually, their end-points are different. Future studies may rely on mixed methods and longitudinal design to further explore the differences between organizational and individual factors. So far it is also unclear whether persons suffering from dysfunctional burnout may finally chose distancing strategies when they return to their job; here, longitudinal studies are needed, too.

Moreover, we are aware that those with clinical relevant burnout symptoms may already have left their job or are at least not at work, and are thus not in the sample. We also have to state the possibility of omitted covariate bias; there might be further variables to explain the described variance in the regression models which remain to be identified in future studies.

A further limitation is that the sample of this study might not be representative for all nurses and hospital physicians, and should thus be regarded as a large convenience sample. Nevertheless, this sample seems appropriate to address the main research questions.

The above described data would support the notion that health care professionals require a work environment which facilitates maintaining their motivation for an optimal and compassionate care of their patients. A structural approach of support by the hospital management would require first to control and eliminate work stressors (including work structure), and second a multifaceted approach to strengthen and support hospital staff’s resources in terms of team building, communication skills, resilience training, compassion training, etc. In nurses, both organizational and intrapersonal resources were identified as important buffers [42]. Resilience trainings for nurses to handle their stressful work environments were already tested [43], but remain to be implemented. So far it is unclear whether resilience and leadership programs that focus on individual rather than organization factors are indeed effective in the long term.