Burnout in ICU Caregivers

A Multicenter Study of Factors Associated to Centers

lanie Verdon1, Adrian Businger2, Guido Domenighetti3, Hans Pargger2, Paolo Merlani1, Me Bara Ricou1, and the STRESI1 Group*
1 Service of Intensive Care, Department of Anesthesiology, Pharmacology, and Intensive Care, Geneva University Hospitals and the University of Geneva, Geneva; 2Division of Operative Intensive Care, Department of Anesthesia and Intensive Care, University Hospital of Basel, Basel; and 3 Multidisciplinary Intensive Care, Regional Hospital of Locarno, Locarno, Switzerland

Rationale: The stressful work environment of ICUs can lead to burnout. Burnout can impact on the welfare and performance of caregivers, and may lead them to resign their job. The shortage of ICU caregivers is becoming a real threat for health care leaders. Objectives: To investigate the factors associated with burnout on a national level in order to determine potential important factors. Methods: Prospective, multicenter, observational survey of all caregivers from 74 of the 92 Swiss ICUs, measuring the prevalence of burnout among the caregivers and the pre-specied center-, patientand caregiver-related factors inuencing its prevalence. Measurements and Main Results: Out of the 4322 questionnaires distributed from March 2006 to April 2007, 3052 (71%) were returned, with a response rate of 72% by center, 69% from nurseassistants, 73% from nurses and 69% from physicians. A high proportion of female nurses among the team was associated with a decreased individual risk of high burnout (OR 0.98, 95% CI:0.97 0.99 for every %). The caregiver-related factors associated with a high risk of burnout were being a nurse-assistant, being a male, having no children and being under 40 years old. Conclusions: The ndings of this study seem to open a new frontier concerning burnout in ICUs, highlighting the importance of team composition. Our results should be conrmed in a prospective multicenter, multinational study. Whether our results can be exported to other medical settings where team-working is pivotal remains to be investigated.
Keywords: critical care; job satisfaction; multicenter; intensive care unit management; intensive care unit organization

AT A GLANCE COMMENTARY
Scientic Knowledge on the Subject

The burnout syndrome can affect up to 45% of intensive care unit (ICU) nurses and physicians. Burnout impacts on quality of care and may lead to a shortage of ICU caregivers in the future. Therefore it is important that health care leaders and hospital managers be aware of possible risk factors for burnout.
What This Study Adds to the Field

The proportion of female nurses impacts on the prevalence of burnout among the whole ICU team. The ndings of this study seem to open a new frontier concerning burnout in ICUs, highlighting the importance of team composition. The results should be conrmed in a prospective multicenter, multinational study analyzing all the known risk factors for burnout.

I found myself looking at the body as a wonderful machine, but not as a creature with a soulthat worried me a bit. What in fact I had to do was consciously unlearn that sort of thing and look at human beings as human beings.R. Richardson (1)

(Received in original form January 13, 2011; accepted in nal form August 5, 2011) Supported by the Swiss Society of Intensive Care Medicine and by departmental funds from the Department of Anesthesiology, Pharmacology, and Intensive Care of Geneva University Hospitals. *Members of the STRESI1 Group are listed at the end of the article. Author Contributions: P.M. participated in the design of the study, performed the statistical analysis, interpreted the data, and wrote the article. M.V. designed the study, collected data, and helped writing the manuscript. A.B., G.D., and H.P. collected data and helped write the manuscript. B.R. conceived the study, participated in the design of the study, had full access to and interpreted the data, and wrote the article. All authors read and approved the nal manuscript. This work was presented partially at the annual congress of the Swiss Society of Intensive Care Medicine in November 2007 and at the Annual Congress of the European Society of Intensive Care Medicine (ESICM) in Lisbon in 2008. Correspondence and requests for reprints should be addressed to Bara Ricou, M.D., Intensive Care, Department APSI, Geneva University Hospitals, Rue Gabrielle-Perret-Gentil 4, 1211 Geneva 14, Switzerland. E-mail: bara.ricou@ hcuge.ch This article has an online supplement, which is available from this issues table of contents at www.atsjournals.org
Copyright 2011 by the American Thoracic Society Am J Respir Crit Care Med Vol 184. pp 11401146, 2011 Originally Published in Press as DOI: 10.1164/rccm.201101-0068OC on August 18, 2011 Internet address: www.atsjournals.org

ICU caregivers work in a stressful environment that can lead to burnout syndrome. Stress occurs when there is a discrepancy between the demands and the resources of the individual (2). Burnout is a psychological syndrome that occurs in response to chronic emotional and interpersonal stressors at work (3). It can lead to emotional instability, difculties to commit, a feeling of failure, and an urge to leave ones job (4, 5). Burnout can affect up to 45% of the medical and nursing staff (6, 7). Consequences to caregivers health include insomnia, irritability, and depressive symptoms (7), which in turn can impact on the quality of the care they provide (8). From an institutional point of view, burnout is a target for improvement, as it can lead to absenteeism and resignation from jobs (9, 10). A shortage of ICU caregivers is already anticipated (4, 11), whereas the training of this highly specialized staff is costly. Therefore, every effort should be made to better understand what leads to burnout. The known factors associated with burnout, such as job dissatisfaction (12), end-of-life issues, conicts (7), lack of recognition, or responsibility overload (6, 10), are difcult to correct. In a previous single-center study, we demonstrated that the degree of stress and of burnout were great in a Swiss surgical ICU (13). In this multicenter study led at a Swiss national level, we sought to investigate specically whether the characteristics related to centers, which we chose to name center-related

Merlani, Verdon, Businger, et al.: Factors Related to Burnout in ICU

1141

factors, can be associated with burnout in ICUs, to determine potential targets for corrective action in the future. These factors were distinctly addressed as patient-related factors (mortality, length of stay) and as caregiver-related factors (the professional category, their sex, the number of children, whether they live alone or not, their age, and so on). Our secondary objective was to assess the distribution of stress and burnout in a multicultural country such as Switzerland, and to detect possible differences among the professional categories regarding the factors associated with burnout. Some results of this study have been previously reported in abstracts presented at the Annual Congress of the European Society of Intensive Care Medicine (ESICM) in Lisbon in 2008 (14) and at the Annual Congress of the Swiss Society of Intensive Care Medicine in November 2007 (15, 16).

or stress were then performed, considering all individuals to obtain adjusted estimates of the ORs and to identify factors independently associated with the outcomes. All predictors at a 0.05 a level in the univariate analysis of burnout were entered into both (stress and burnout) mixed-effect logistic regression models of the multivariate analysis. Variables are expressed as mean 6 standard deviation (SD) if not specied otherwise. A P value less than 0.05 was considered statistically signicant.

RESULTS
Seventy-four of the 92 ICUs (80%) listed by the Swiss Society of Intensive Care Medicine participated in the survey. Of the 4,322 questionnaires distributed, 3,052 (71%) were fully completed and returned. The response rate by center was 72 6 12% (mean 6 SD) and differed slightly according to the regions studied. The CONSORT (Consolidated Standards of Reporting Trials) diagram (Figure 1) summarizes the distribution of responses. Characteristics of the centers, patient population, and caregivers are summarized in Table 1. Overall, 2,996 of the 3,052 (98.2%) caregivers answered the questionnaire about burnout. The response rate of nurse-assistants was 46% (137 of 299), for nurses it was 73% (2,450 of 3,345) and for physicians it was 69% (465 of 678). Among the 2,996 respondents, 874 (29%) showed a high degree, 995 (33%) a medium and 1,167 (39%) a low degree of burnout. The mean MBI score was 15.6 6 15 (mean 6 SD), ranging from 50 to 73, with a median score of 17.0 (interquartile range, 20). High burnout was present in 48 of 117 nurse-assistants (41%), 683 of 2,415 nurses (28%), and 143 of 459 physicians (31%). The rate of high degree of burnout among the ICU centers ranged from 5 to 62%, with a mean of 28%. The overview of the percentage of burnout in the different caregivers in the three regions is shown in Table E2. Among the 3,052 questionnaires, 3,047 answers about stress were analyzable (99.8%). Overall, 1,117 of the 3,047 (37%) caregivers working in Swiss ICUs who responded felt highly stressed. The list of variables analyzed and the univariate results concerning burnout and stress are shown in Table E1. Results of the uni- and multivariate analysis of the factors associated with burnout are shown in Table 2. An additional multivariate model that also includes the variable feeling stressed is shown in Table E3 and a model analyzing only female nurses is shown in the Table E4. Results of the uni- and multivariate analysis of the factors associated with stress are shown in Table E5.

METHODS
The setting comprised all certied ICUs in Switzerland. Certication is provided by the Swiss Society of Intensive Care according to national regulation (17).The directors of all Swiss ICUs (three language regions: French-, Italian-, and German-speaking parts) were invited to participate in the study. The centers data were directly collected from the ICU directors who agreed to participate (see Table E1 in the online supplement), separately from the questionnaires distributed to the caregivers. Once the director had given consent, all caregivers, including physicians, nurses, and nurse-assistants, of the center received the questionnaire between March 2006 and April 2007. The nurse team was dened as the sum of the nurses working at a given center, without correcting for the percentage of work activity. In Switzerland, nurseassistants are caregivers who help the nurses in the basic care of patients (washing, feeding, and voiding). They do not administer medication or set up mechanical ventilation. The questionnaire, a simplied version of the questionnaire described in detail elsewhere (13; and in Table E1) was selfadministered and anonymous. It comprised demographic data, personal characteristics, the risk of burnout according to the Maslach Burnout Inventory (MBI) in its Fontaine French version (18), and an evaluation of stress based on a national survey of Swiss workers led by the Swiss State Secretariat of Economic Affairs (19), based on a fourpoint Likert Scale (I feel stressed: never, sometimes, often, very often) (forced-choice Likert Scale) to minimize a central tendency bias (20). To conrm the previous assertion that repeated stress can lead to burnout, we included the evaluation of stress in our questionnaire. Because the burnout depends on the individuals personal resources as well as his or her capacity to deal with stress, we assessed perceived stress. Objective stress is not pertinent in this eld. A high degree of stress was dened as self-reported by the caregivers according to the national survey. A high degree of burnout was dened as the cumulated score of MBI of greater than 9 according to previous reports (6, 7). The coordination center in Geneva collected and analyzed the data anonymously. The protocol was approved by the Ethics Committee of Geneva University Hospitals, and informed consent was assumed if the questionnaire was returned.

Center-related Factors

Power Size and Statistical Analysis

A detailed description is available on the online data supplement (Table E2). In brief, Stata statistical software (release 11.0; Stata Corporation, College Station, TX) and StatView (SAS institute Inc, Cary, NC) were used. In the univariate analysis, risk factors for both high degree of burnout and stress were evaluated by two-tailed Fisher exact tests for dichotomous variables, larger chi-squared test for independence, and unpaired t tests or Mann-Whitney U tests for continuous variables, as appropriate. Odds ratios (ORs) with 95% condence intervals (CIs) were calculated, considering all participants to estimate the effect size of risk factors associated with a high degree of burnout and stress. Multivariate mixed-effect logistic regressions with random effects on the intercept accounting for centers concerning a high degree of burnout

By univariate analysis (Table 2), a higher proportion of female caregivers among the nurses and among the physicians decreased the risk of burnout. By multivariate analysis a higher proportion of female nurses in the team was associated with decreased risk of high burnout (OR for every percentage, 0.985; 95% CI, 0.9730.997), whereas caregivers working in German-speaking ICUs were associated with a high degree of burnout. The association between the proportion of female nurses and burnout in the multivariate model, performed only in female nurses (Table E4), was even stronger (OR for every percentage more of female nurses, 0.979; 95% CI, 0.9640.994). Furthermore, this association was almost linear and no signicant interaction between the type of caregiver or the sex and the female proportion was identiable. This association could not be veried in specic models for physicians or for nurse-assistants only because of the small size of the two cohorts.

1142

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

VOL 184

2011

Figure 1. CONSORT (Consolidated Standards of Reporting Trials) diagram.

By univariate analysis (Table E5), the factors associated with increased risk of high stress were university hospitals, pediatric ICUs, the number of beds per unit, and a higher proportion of females among physicians whereas working in the Germanspeaking part of Switzerland was associated with decreased risk. By multivariate analysis (Table E5), working in a university hospital was independently associated with a high degree of stress. There was no association between the stress and the proportion of female nurses or physicians in a team.

No signicant random center effect was identied in any of the mixed-effect logistic regression models (for burnout) with random effects on the intercept accounting for centers, whereas in the model analyzing stress there was a signicant effect (Table E5). The multivariate logistic regression analyzing centers with burnout prevalence higher than the median (28.6%) value (Table E7) showed that only German-speaking centers (OR, 4.71; 95% CI, 1.09120.363; P 0.04) were associated with an

TABLE 1. CHARACTERISTICS OF CENTERS, PATIENT POPULATIONS, AND CAREGIVERS

French-speaking Centers (n 17; 23%) University centers, n (%) Pediatric centers, n (%) Number of beds, m 6 SD Nurse/patient (daytime), mean 6 SD Nurse/patient (nighttime), mean 6 SD Nurse working hours, 8 h/12 h/mixed: n/n/n (%/%/%) Intern rotation (mo), mean 6 SD Percentage of females in the NA team (%), mean 6 SD Percentage of females in the nurse team (%), mean 6 SD Percentage of females in the medical team (%), mean 6 SD Patient populations Mortality rate, % (min-max) Length of ICU stay (d), mean 6 SD Number patients/yr, mean 6 SD 8 (47)* 5 (29)* 12.2 6 6.4 0.67 6 0.19 0.51 6 0.14 3/10/4 (18/59/23)* 4.1 6 2.5 91.7 6 12.7 82.2 6 12.8 40.5 6 24.3 5 (110) 4.9 6 3.7 934 6 533 Italian-speaking Centers (n 5; 7%) 0 (0) 0 (0) 7.8 6 2.5 0.65 6 0.12 0.48 6 0.06 5/0/0 (100/0/0) 5.4 6 3.9 88.7 6 13.1 61.9 6 7.5 35.8 6 15.1 6 (58) 2.9 6 0.3 780 6 199 Caregivers Nurse-assistants (n 137; 5%) Female sex, n (%) Age , 40 yr, n (%) Full time job, n (%) Years of experience in ICU, median [IQR] No child, n (%) Living alone, n (%) 119 (87) 75 (55)* 66 (48) 5.5 [8] 58/136 (43)* 46 (34)* Nurses (n 2,450; 80%) 2,018 (82) 1,740 (71) 1,054 (43) 8 [9.8] 1,536/2,419 (63)* 687/2,446 (28) Physicians (n 465; 15%) 193 (42)* 321 (69) 400 (86)* 2.4 [7.3] 273/463 (59) 92 (20)* All Respondents (n 3,052; 100%) 2,331 (76) 2,136 (70) 1,526 (50) 7.5 [10] 1,868/3,020 (62) 826/3,050 (27) German-speaking Centers (n 52; 70%) 9 (17) 4 (8) 9.5 6 5.0 0.71 6 0.13 0.61 6 0.14 52/0/0 (100/0/0)* 5.4 6 3.1 80.9 6 36.5 83.5 6 10.8 38.9 6 28.7 4 (110) 3.1 6 1.5 1,105 6 537 All Centers (n 74; 100%) 17 (23) 9 (12) 10.0 6 5.5 0.70 6 0.14 0.57 6 0.15 60/10/4 (81/14/5) 5.1 6 3.0 84.8 6 30.0 81.7 6 12.3 39.0 6 26.8 4 (110) 3.5 6 2.9 1,044 6 521

Denition of abbreviations: ICU intensive care unit; NA nurse-assistant. * P , 0.05, Fishers exact test or multiple chi-square test, compared with both other categories. y P , 0.05, t test, compared with both other categories. z P , 0.05, Mann-Whitney U test, compared with both other categories.

Merlani, Verdon, Businger, et al.: Factors Related to Burnout in ICU

1143

TABLE 2. UNI- AND MULTIVARIATE MIXED-EFFECTS LOGISTIC REGRESSION FOR FACTORS ASSOCIATED WITH BURNOUT, WITH RANDOM EFFECT BY CENTERS
Univariate Analysis Center-related factors Universitynonuniversity Pediatric ICUsadult ICUs Number of beds (per bed) German-, French-, Italian-speaking part of Switzerland Proportion of females in the nurse team (per %) Proportion of females in the medical team (per %) Patient-related factors Mortality rate (per %) Length of stay (per day) Caregiver-related factors Nurse-assistant(nurse, physician) Sex: male No child Living alone Age < 40 yr OR 0.917 0.619 1.007 1.125 0.983 0.996 1.063 0.946 1.725 1.484 1.390 1.130 1.340 95% CI 0.7811.076 0.4830.793 0.9961.019 0.9521.330 0.9760.990 0.9920.999 1.0301.097 0.9140.980 1.1832.515 1.2411.774 1.1771.643 0.9481.347 1.1221.599 OR 0.969 1.290 0.998 1.505 0.985 0.996 1.060 0.967 2.126 1.522 1.438 1.036 1.305 Multivariate Analysis 95% CI 0.6771.385 0.7252.296 0.9741.022 1.1531.964 0.9730.997 0.9921.001 1.0031.120 0.9051.032 1.4303.160 1.2541.849 1.1841.747 0.8541.257 1.0721.589 P Value 0.86 0.39 0.88 0.003 0.01 0.16 0.04 0.31 ,0.001 ,0.001 ,0.001 0.72 0.008

Denition of abbreviations: CI condence interval; ICU intensive care unit; OR odds ratio. Notes: n 2,914. Random-effect parameter (center): estimate, 0.026; standard error, 0.013. Likelihood ratio versus logistic regression model: P 0.08.

increased risk and that centers with a high percentage of female nurses were instead associated with a reduced risk of a high prevalence of burnout (OR, 0.943; 95% CI, 0.9681.011; P 0.04), reecting the increased risk of males to have burnout at the individual level.
Patient-related Factors

By univariate analysis, the mortality rate and the length of ICU stay increased the risk of high burnout and high stress. By multivariate analysis, only the mortality remained signicantly associated with high burnout (Table 2).
Caregiver-related Factors

Of 117 nurse-assistants, 48 (41%) had a high degree of burnout. This proportion was signicantly higher (P 0.004) compared with the 683 of 2,415 nurses (28%) and 143 of 459 physicians (31%). The total numbers differed slightly because of six caregivers who did not declare their professional specicity. Table E2 shows the details regarding the burnout of the three professional categories and of the three linguistic regions. The distribution and characteristics of the various professional categories is summarized in Table 1. The proportion of nurse-assistants who felt highly stressed was not signicantly greater than for physicians and nurses (41 vs. 36%; P 0.27). Nurses were not signicantly more stressed than the others (37 vs. 33%; P 0.07). Fewer physicians said they were stressed compared with nurse-assistants and nurses (31 vs. 38%; P 0.009). Most caregivers (64%) had difculties reconciling work and private life. In the univariate analysis (Table 2 and Table E5), having no child and being under 40 years old was associated with increased risk of high burnout and high stress. Age was rst considered as a continuous variable and then in 5- and 10-year intervals. The nonlinear association of age with stress and burnout ultimately required its analysis as a dichotomized variable (more and less 40 yr). Being a male increased the risk of burnout and decreased the risk of high stress. The activity rate of work inuenced neither burnout nor stress. Being a nurse-assistant compared with being a nurse or a physician was associated with increased risk of high burnout. The number of years of experience in ICU was

inversely correlated with burnout and was collinear with age. The multivariate analysis (Table 2) showed that being a male, a nurse-assistant, having no children, and being under 40 years old was each independently associated with high burnout. Being a female, having no children, and not living alone were independently associated with high stress (Table E5). We did not identify signicant interactions between variables. When the factor Feeling stressed was added to the burnout multivariate analysis, it became the predominant independent factor that increased the risk of high burnout (OR, 3.72; 95% CI, 3.124.43; P , 0.001) (Table E3). Most factors determined in the previous model changed very little, except that the mortality rate stopped being signicant. An additional multivariate model (Table E4) considering only female nurses showed results that were almost identical to those in Table 2, except for the center mortality rate, which ceased being signicant (OR, 1.06; 95% CI, 0.991.13; P 0.07). Table E6 compares the various multivariate models. Among the caregivers who answered the questions about both burnout and stress (2,992), high burnout and high stress were differently distributed depending on the linguistic regions. Caregivers from the French-speaking region showed high stress (359 of 830; 40%), among whom 142 of 359 (40%) showed high burnout, whereas from the Italian- and German-speaking regions, 87 of 199 (44%) and 652 of 1,963 (33%), respectively, showed high stress, among whom 44 of 87 (51%) and 316 of 652 (48%), respectively, showed high burnout (P , 0.0001 for high stress, P 0.02 for high burnout; both test for independence). High stress and high burnout were also differently distributed depending on the caregivers sex (overall, 2,987 caregivers responded to all questions). High stress was present in 229 of 715 (32%) male caregivers, among whom 131 of 229 (57%) showed high burnout, whereas 868 of 2,272 (38%) female caregivers showed high stress, among whom 371 of 868 (43%) showed high burnout (P 0.0003 for high stress, P 0.0001 for high burnout).

DISCUSSION
The major nding of this multicenter study was that a potentially important center-related factor, namely the increased proportion of female nurses among the caregivers, was associated with

1144

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

VOL 184

2011

decreased risk of high burnout. This could open a new era in managerial point of view for heads of ICUs. Indeed, we showed that the relation was almost linear, meaning that the risk of burnout among the whole ICU team may be affected by the composition of the team. The positive impact of the female proportion was even more pronounced among the nurses. Unfortunately, the effect could not be conrmed specically in separate models including only physicians or nurse-assistants because of the small size of the two cohorts. Interestingly, the female proportion of nurses had the same effect on males and females in the global model concerning burnout. Burnout is a preoccupying condition both for the employee and the manager: the employee suffering from burnout feels professionally exhausted, isolated, as having failed overall, and is less motivated to work, and the manager faces inefciency, conicting interaction, and absenteeism (3). Burnt-out physicians tend to care less for patients (8) and the number of medical errors increases with burnout (21). Burnout can lead to depressive symptoms and to premature resignation from the ICU (22). These consequences present a major threat from a managerial point of view in the present context of increasing decit in critical care physicians (23) and nurses (24). For the rst time, we have indications that factors related to the center could be involved in the prevalence of high burnout. The high response rate from each center and the high participating rate of the centers (74 among 92; 80%) should have minimized the selection bias and allowed a fair analysis of the factors linked to the centers and the caregivers. The strengths of this work also reside in the inclusion of ICU caregivers from different backgrounds, and who speak different languages, and the high response rate that enhances the relevance of our ndings as well as their generalizing quality for other ICUs. Many publications show that the burnout phenomenon is common throughout the world and is not exclusively found in wealthy countries (2527). Furthermore, our results are applicable to ICUs of other countries because a number of multicenter European studies such as the Ethicus Study regarding end-oflife practices (28), the Conicus Study regarding conicts in ICUs (29), and the SEE Study, a multinational Sentinel Events Evaluation (30) included Switzerland and the results showed that Switzerland is comparable to many surrounding countries. Results from Switzerland are also often similar to those found in North America, that is, SAPS (Simplied Acute Physiology Score) studies (31, 32). The originality of this study resides in several other points. First, we took into account the feeling of stress and burnout together because burnout is the result of chronic stress (3). The fact that high stress was not always associated with high burnout in our study must be commented on. One explanation might be that being burned out may lessen the resistance to stress, and thus might contribute to a vicious cycle wherein the role of each factor might be confounding. Also, the apparent paradox that female subjects are more stressed yet less prone to burnout compared with the male subjects could be due to methodological bias. Indeed, female caregivers may have found it easier to admit their distress than the male caregivers did. The interesting nding in this context was that male caregivers were actually more in burnout than females, independent of their professional category. Men may be less inclined to express their distress and may be helped by being taught how to manage their feelings (33). Second, the difference between French- and Germanspeaking regions may be attributed to cultural differences. As a rule, caregivers speak the regional language to communicate with the patients and their families. The different languages clearly indicate singular cultures in Switzerland. Depending on cultural background, the feeling of stress and burnout may vary

(34). Our study design does not allow explaining the relationship between the feeling of stress and burnout or the potential cultural impact. A qualitative approach of such questions might be of great interest. In addition, the types of ICU, that is, university or pediatric ICU, were examined. Stress was felt more often in university and pediatric units, but it was not associated with burnout. The size of the units, the nurse-to-patient ratio, the 8- or 12-hour shifts, the size of the medical or nurse team, the number of patients per year, and the time spent working at the ICU had no impact on the prevalence of burnout in the multivariate analysis. Last, the three professional categories (nurse-assistants, nurses, and physicians) were investigated simultaneously in each center. To our knowledge, the data regarding the nurse-assistants are new. A number of papers about stress or burnout of caregivers have been published throughout the world (10, 35, 36), but up to now none has given a whole picture of the subject in an entire country and allowed comparison of professional categories. The professional category at the highest risk of burnout was that of the nurse-assistants. Because the participation in end-of-life and postmortem care increases the psychological burden and the risk of burnout (7), and because these caregivers are usually less numerous in ICUs, the unavoidable consequence is that they are more frequently confronted with these difcult situations than are others. Our ndings may prompt heads of ICUs to take particular care of this professional category (37). The increased mortality rate in centers associated with high burnout is consistent with the ndings of Embriaco and colleagues (6) and Poncet and colleagues (7), who demonstrated that the prevalence of burnout increased among physicians and nurses who had often been confronted with death or who had participated in decisions of foregoing life-sustaining therapy. In contrast, our results do not sustain the international and European guidelines proposing that ICUs should not exceed 16 beds (38, 39). These guidelines were based mostly on experts opinion and no data were available regarding the specic issue of the occurrence of burnout among caregivers and the size of the centers. The mortality rate was in no way related to the size of the ICUs. Our study has some limitations. First, Switzerland is a small country with a unique organization of 26 cantons with 92 certied ICUs of 6 to 36 beds. However, the rate of burnout was overall similar to previous reports from other countries and the ndings regarding the related factors may still be generalizable. Second, the number of nurses exceeded by far the number of physicians and nurse-assistants, which may have biased the results. However, the multivariate analysis allowed us to analyze the three professions at the same time accounting for their differences. Moreover, we did not identify signicant interactions in our multivariate model indicating that the analysis or the models should be split. Third, a few of the variables in the analysis, such as the 12-hour shifts or the mixed shifts, were sparse and therefore the results are less dependable. Thus we cannot exclude that some other factors may be related to stress or burnout. Fourth, there was 29% nonrespondents. These may either be people who do not experience high rates of distress and are consequently less interested in completing a questionnaire about stress and burnout, or they could be those who were too affected by burnout to be interested in lling out a questionnaire. We have no way to obtain more information about this population. The results could therefore be modied if all caregivers had responded. On the other hand, we were able to include 80% of all Swiss ICUs (compared with less than 60% of centers participating in other studies until now). This means that the data concerning

Merlani, Verdon, Businger, et al.: Factors Related to Burnout in ICU

1145

center characteristics should be less prone to bias selection (having included centers with high and low burnout) than the studies performed until now. Therefore, 71% of caregivers participating to the study representing 80% of centers seems acceptable, even if not totally free from potential bias. Fifth, we did not attempt to compare the degree of stress or burnout with other professions, because the reasons leading to burnout may differ. The aim of our study was to focus more on nding possibly modiable factors that lead ICU caregivers to burnout. Besides the modication of team composition, promising solutions for caregivers would be team building (40), staff empowerment (41), development of a culture aimed at preventing conicts (42), and modication of the inner self of the caregivers (43). Sixth, we did not evaluate all the possible confounders, in particular regarding the higher risk of burnout of younger caregivers. It is possible that caregivers with the lowest seniority were less able to schedule days off, were not invited to participate in the research group, took conicts more personally, felt more inhibited in establishing working relationships with the physicians and head nurses, and were asked to work more night shifts and had less discretion over holiday time. Each of these relationships could be confounded by age and each of these established risk factors was not collected or evaluated as part of the multivariable modeling in our study. Seventh, the questionnaire used was a simplied form of the previously published questionnaire (13). Because the aim of our study was to investigate the potential factors related to the centers, we decided to simplify the questionnaire to increase the response rate to our questionnaire. This fact did not allow us to investigate all the previously detected concerns of the different caregivers. Last, we acknowledge that several known factors related to burnout such as conicts, ability to schedule days off according to personal wishes, and relationships with physicians were not investigated in our study. In fact, our study was done during 20062007, and at that time the now well-known data from Embriaco and colleagues (6) and Poncet and colleagues (7) about burnout and about conicts from Azoulay and colleagues (29) were not available. In conclusion, the ndings of this study suggest for the rst time that the composition of the team, in our case the percentage of female nurses, could be one of the major determinants of burnout in ICUs. Even if these new results seem appealing, we must be careful before social reengineering of the workplace be undertaken. The main message of our study may be that we must now prospectively investigate and conrm in a multicenter multinational study team compositions and their impact on burnout and stress in ICUs. In the future it would be interesting to better analyze qualitatively ICU team composition. The second message of this work could be that there is a subset of caregivers at higher risk of burnout such as young women alone, without children. Further studies into the possible reasons for burnout in this subset of caregivers are needed to optimally address the problem. In the meantime, the ICU heads should ascertain that personnel at higher risk are given particular care, and that resources should be provided to afford psychological support and promote a team culture. Although not proven yet, it may well be that the attitude of the leaders toward their collaborators concerning an acknowledgment of the high risk of burnout will induce a global cultural modication in the unit. Ideally, institutional approaches should be complemented by personcentered interventions, such as mindfulness meditation, selfawareness exercise, or narrative medicine (43, 44). This study seems to open a new frontier concerning burnout in ICUs, highlighting the importance of team composition. Our results should be conrmed in a prospective multicenter, multinational study. Whether the results can be exported to other medical settings

where team-working is pivotal remains for the moment an interesting question to be investigated.
Author Disclosure: M.P.s institution has received travel support from the Swiss Society of Intensive Medicine. M.V.s institution has received travel support from the Swiss Society of Intensive Medicine. A.B., G.D., and H.P. have no nancial relationship with a commercial entity that has an interest in the subject of this manuscript. B.R.s institution has received travel support from the Swiss Society of Intensive Medicine. STRESI1 Group: Centers referees (in alphabetical order of the physician referee/nurse): Ba chli E, Baenziger O, Baumgartner JD, Berner M, Bu hrer C, Cas s C, Chevrolet JC, Chiole ro R, Schaller MD, sina T/Rusca P, Chevalley F/Lapre Cotting J/Gonzalez A, Domenighetti G/Bonetti L/Bomio C, Eckert P/Theytaz C, Felleiter P, Fournier F/Bezencon A, Genet Donati P/Savin C, Giambarba C, Heim C, Hirlinger A, Holy D, Kocher A/ Croce N, Lang M, Malacrida R/Sperandio I, Mangold K, Marquardt K, Marsch S, Moessinger A /Lilla M, Pagani JL/Vannay C, Paganoni R, Pagnamenta A /Bo gli L, Perren A/Sommaca C, Quartenoud B/ Sluga M/Racine V, Schmidlin D, Tagan D/Wallef S, Thorens JB/Mathieu D, von Dreden G, Wiederkehr P, Zender H/Collin C, Zuercher Zenklusen R/Cortinovis P. Centers (in alphabetical order): Bu rgerspital Solothurn, Cardiocentro Lugano, pital de la Chaux de Fonds, Ho pital de Dele mont, Ho pital de Clara Spital, Ho pital de La Tour, Ho pital de Morges, Ho pital de Neucha tel, Ho pital Fribourg, Ho pital de Vevey, Ho pital d9Yverdon, Inselspital Bern, Kantonsspital de Sion, Ho Aarau (Chirurgie und Medizin), Kantonsspital Baden, Kantonsspital Bruderholz, Kantonsspital Frauenfeld Spital Thurgau, Kantonsspital Glarus, Kantonsspital Graubu nden, Kantonsspital Luzern (Chirurgie und Medizin), Kantonsspital Mu nsterlingen Spital Thurgau, Kantonsspital Olten, Kantonsspital Schaffhausen, Kantonsspital St. Gallen (Chirurgie und Medizin), Kantonsspital Uri, Kantonsspital Winterthur, Kantonsspital Zug, Kinderspital Zu rich, Klinik Beausite, Klinik Hirslanden, Klinik im Park, Klinik im Schachen, Ospedale Beata Vergine Mendrisio, Locarno, Ospedale San Giovanni Ospedale Civico Lugano, Ospedale La Carita Bellinzona, Ost-Schweitzer Kinderklinik, Schweizer Paraplegikerzentrum, Spital Bern Tiefenau, Spital Bu ler Frutigen Meiringen Interlaken lach, Spital Davos, Spita AG, Spital Lachen, Spital Liestal, Spital Limmatal, Spital Ma nnedorf, Spital Oberengadin, Spital Schwyz, Spital Sursee-Wolhusen, Spital Uster, Spital Zollikerberg, Spitalzentrum Biel, SRO Langenthal, Stadtspital Triemli, Stadtspital Waid, Universita ts-Kinderspital beider Basel, Universita tsspital Basel MIPS, Universita tsspital Basel OIB, Universita tspital Zu rich (cardiovascular, medical, and surgical ICUs), University Hospital of Geneva (medical, surgical, neonatology and pediatric ICUs), University Hospital of Vaud (medical, surgical, neonatology and pediatric ICUs), Zieglerspital Bern. Acknowledgment: The authors are grateful to Ms. Fabienne Gigon for help and careful reading of the manuscript. Methodological statistical advice was provided by the Clinical Research Center, University of Geneva and Geneva University Hospitals (C. Combescure). All the authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

References
1. Richardson R. A necessary inhumanity? Med Humanit 2000;26:104106. 2. Folkman S, Lazarus RS, Gruen RJ, DeLongis A. Appraisal, coping, health status, and psychological symptoms. J Pers Soc Psychol 1986; 50:571579. 3. Maslach C, Schaufeli WB, Leiter MP. Job burnout. Annu Rev Psychol 2001;52:397422. 4. Mealer ML, Shelton A, Berg B, Rothbaum B, Moss M. Increased prevalence of post-traumatic stress disorder symptoms in critical care nurses. Am J Respir Crit Care Med 2007;175:693697. 5. Steinbrook R. Nursing in the crossre. N Engl J Med 2002;346: 17571766. 6. Embriaco N, Azoulay E, Barrau K, Kentish N, Pochard F, Loundou A, Papazian L. High level of burnout in intensivists: prevalence and associated factors. Am J Respir Crit Care Med 2007;175:686692. 7. Poncet MC, Toullic P, Papazian L, Kentish-Barnes N, Timsit JF, Pochard F, Chevret S, Schlemmer B, Azoulay E. Burnout syndrome in critical care nursing staff. Am J Respir Crit Care Med 2007;175:698704. 8. Shanafelt TD, Bradley KA, Wipf JE, Back AL. Burnout and selfreported patient care in an internal medicine residency program. Ann Intern Med 2002;136:358367. 9. Vahey DC, Aiken LH, Sloane DM, Clarke SP, Vargas D. Nurse burnout and patient satisfaction. Med Care 2004;42(2, Suppl)II57II66. 10. Coomber S, Todd C, Park G, Baxter P, Firth-Cozens J, Shore S. Stress in UK intensive care unit doctors. Br J Anaesth 2002;89:873881. 11. Health Resources and Services Administrations (HSRA). The Critical Care Workforce: A Study of the Supply and Demand for Critical Care Physicians. Report of the Congress. Senate Report:108-81. Washington, DC: U.S. Government Printing Ofce; 2003.

1146

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE

VOL 184

2011

12. Aiken LH, Clarke SP, Sloane DM, Sochalski J, Silber JH. Hospital nurse stafng and patient mortality, nurse burnout, and job dissatisfaction. JAMA 2002;288:19871993. 13. Verdon M, Merlani P, Perneger T, Ricou B. Burnout in a surgical ICU team. Intensive Care Med 2008;34:152156. 14. Verdon M, Businger A, Merlani P, Domenighetti G, Pargger H, Ricou B; STRESI1 Group. Stress and burnout in Swiss ICU caregivers: a nationwide multicenter study [abstract]. Intensive Care Medicine 2008;34(Suppl 1):A0012. 15. Verdon M, Businger A, Merlani P, Domenighetti G, Pargger H, Ricou B; STRESI1 Group. The STRESI Study: burnout and stress in ICU [abstract]. Swiss Med Wkly 2007;137:A11S. 16. Verdon M, Merlani P, Domenighetti G, Perneger T, Ricou B; STRESI Group. Professional characteristics from 22 Swiss ICUs [abstract]. Swiss Med Wkly 2007;137:A49. r die Anerkennung der Intensivstationen, Schweizerische 17. Kommission fu r Intensivmedizin. Directives pour la reconnaissance Gesellschaft fu s de soins intensifs (USI) par la Socie te Suisse de Me decine des unite Intensive (SSMI). Released September 2, 2010. Available from: http:// www.sgi-ssmi.ch/index.php?eIDtx_nawsecuredl&u0&leleadmin/ user_upload/Downloads/PDF/KAI_Richtlinien_100902_F_neu.pdf&t 1317368372&hashb8884d672eba53d95c4512d0476dec60 puisement professionnel des P, Mauranges A. Le syndrome de 18. Canou ponses. Paris: Masson; 2001. soignants: de lanalyse du burn-out aux re tariat dEtat a ` lEconomie, SECOSUVAPromotion Sante Suisse. 19. Secre venir le stress au travail destine Un programme pour diminuer et pre aux collaboratrices et collaborateurs et aux cadres, 2010. Available from: http://www.stressnostress.ch// 20. Likert R. A technique for the measurement of attitudes. Arch Psychol 1932;140:155. 21. Fahrenkopf AM, Sectish TC, Barger LK, Sharek PJ, Lewin D, Chiang VW, Edwards S, Wiedermann BL, Landrigan CP. Rates of medication errors among depressed and burnt out residents: prospective cohort study. BMJ 2008;336:488491. 22. Elpern EH, Silver MR. Improving outcomes: focus on workplace issues. Curr Opin Crit Care 2006;12:395398. 23. Ewart GW, Marcus L, Gaba MM, Bradner RH, Medina JL, Chandler EB. The critical care medicine crisis: a call for federal action: a white paper from the critical care professional societies. Chest 2004; 125:15181521. 24. Buerhaus PI, Staiger DO, Auerbach DI. Is the current shortage of hospital nurses ending? Health Aff (Millwood) 2003;22:191198. 25. Li J, Lambert VA. Workplace stressors, coping, demographics and job satisfaction in Chinese intensive care nurses. Nurs Crit Care 2008;13:1224. 26. Moreira deSouza D, Magnago RF, Sakae TM, Magajewski FR. Prevalence of burnout syndrome in nursing staff in a large hospital in south of Brazil. Cad Saude Publica 2009;25:15591568. 27. Santana Cabrera L, Hernandez Medina E, Eugenio Robaina P, SanchezPalacios M, Perez Sanchez R, Falcon Moreno R. Burnout syndrome among nurses and nurses aides in an intensive care unit and admission wards. Enferm Clin 2009;19:3134. 28. Sprung CL, Cohen SL, Sjokvist P, Baras M, Bulow HH, Hovilehto S, Ledoux D, Lippert A, Maia P, Phelan D, et al. End-of-life practices in European intensive care units: the Ethicus Study. JAMA 2003;290: 790797.

29. Azoulay E, Timsit JF, Sprung CL, Soares M, Rusinova K, Lafabrie A, Abizanda R, Svantesson M, Rubulotta F, Ricou B, et al. Prevalence and factors of intensive care unit conicts: the Conicus Study. Am J Respir Crit Care Med 2009;180:853860. 30. Valentin A, Capuzzo M, Guidet B, Moreno RP, Dolanski L, Bauer P, Metnitz PG. Patient safety in intensive care: results from the multinational Sentinel Events Evaluation (SEE) Study. Intensive Care Med 2006;32:15911598. 31. Metnitz PG, Moreno RP, Almeida E, Jordan B, Bauer P, Campos RA, Iapichino G, Edbrooke D, Capuzzo M, Le Gall SAPS Jr. 3From evaluation of the patient to evaluation of the intensive care unit. 1. Objectives, methods and cohort description. Intensive Care Med 2005; 31:13361344. 32. Moreno RP, Metnitz PG, Almeida E, Jordan B, Bauer P, Campos RA, Iapichino G, Edbrooke D, Capuzzo M, Le Gall SAPS Jr. 3From evaluation of the patient to evaluation of the intensive care unit. 2. Development of a prognostic model for hospital mortality at ICU admission. Intensive Care Med 2005;31:13451355. 33. Butterworth T, Carson J, White E, Jeacock J, Bishop V, Clements A. It is good to talk: an evaluation study in England and Scotland. Manchester, UK: University of Manchester Press; 1997. 34. Estryn-Behar M, Van der Heijden BI, Oginska H, Camerino D, Le Nezet O, Conway PM, Fry C, Hasselhorn HM. The impact of social work environment, teamwork characteristics, burnout, and personal factors upon intent to leave among European nurses. Med Care 2007;45:939950. 35. Chen SM, McMurray A. Burnout in intensive care nurses. J Nurs Res 2001;9:152164. 36. Kinzl JF, Traweger C, Biebl W, Lederer W. Burnout and stress disorders in intensive care doctors. Dtsch Med Wochenschr 2006; 131:24612464. 37. Levy MM. Caring for the caregiver. Crit Care Clin 2004;20:541547; xi. 38. Guidelines/Practice Parameters Committee of the American College of Critical Care Medicine, Society of Critical Care Medicine. Guidelines for intensive care unit design. Crit Care Med 1995; 23:582588. 39. Ferdinande P; Members of the Task Force of the European Society of Intensive Care Medicine. Recommendations on minimal requirements for intensive care departments. Intensive Care Med 1997; 23:226232. 40. Reader TW, Flin R, Mearns K, Cuthbertson BH. Developing a team performance framework for the intensive care unit. Crit Care Med 2009;37:17871793. 41. Wahlin I, Ek AC, Idvall E. Staff empowerment in intensive care: nurses and physicians lived experiences. Intensive Crit Care Nurs 2010;26:262269. 42. Danjoux Meth N, Lawless B, Hawryluck L. Conicts in the ICU: perspectives of administrators and clinicians. Intensive Care Med 2009; 35:20682077. 43. Krasner MS, Epstein RM, Beckman H, Suchman AL, Chapman B, Mooney CJ, Quill TE. Association of an educational program in mindful communication with burnout, empathy, and attitudes among primary care physicians. JAMA 2009;302:12841293. 44. Connelly JE. Narrative possibilities: using mindfulness in clinical practice. Perspect Biol Med 2005;48:8494.