Work stress, burnout, and perceived quality of care: A cross-sectional study among hospital pediatricians

1 Original Research Article This is a post-review version of the manuscript that in print with the European Journal of Pediatrics. Please refer to the European Journal of Pediatrics for the final in-print version (DOI 10.1007/s00431-015-2529-1). 2 3 Work stress, burnout, and perceived quality of care: A cross-sectional study among hospital pediatricians 4 5 Weigl, M. (1), Schneider, A. (1), Hoffmann, F. (2), Angerer, P. (3) 6 7 8 9 10 11 12 13 (1) Institute and Outpatient Clinic for Occupational, Social, and Environmental Medicine, Ludwig-Maximilians-University Munich, Munich, Germany (matthias.weigl@med.lmu.de, anna2.schneider@med.lmu.de) (2) Dr. von Hauner University Children's Hospital, Ludwig-Maximilians-University Munich, Munich, Germany (florian.hoffmann@med.lmu.de) (3) Institute for Occupational and Social Medicine, Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany (peter.angerer@uni-duesseldorf.de) 14 15 16 17 18 Corresponding author: Matthias Weigl, Dr. Institute and Outpatient Clinic for Occupational, Social, and Environmental Medicine, Ludwig-Maximilians-University Munich, Ziemssenstrasse 1, D-80336 Munich, Germany, Phone: ++49-89-4400-55311, matthias.weigl@med.lmu.de. 19 20 Keywords: Hospital, pediatricians, work life, burnout, quality of care, survey, pediatric care 21 1

22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 ABSTRACT Poor hospital work environments affect physicians work stress. With a focus on hospital pediatricians we sought to investigate associations between work stress, burnout, and quality of care. A cross-sectional study was conducted in N = 96 pediatricians of a German academic children s hospital (response rate: 73.8%). All variables were assessed with standardized questionnaires. Multivariate regression analyses were applied to investigate associations after adjusting for potential confounders. Critically high work stress (effortreward ratio, ERR > 1.0) was reported by N = 25 (28.4%) participants. Pediatricians in inpatient wards had significantly more work stress than their colleagues in intensive care units and outpatient wards. 10.2% of surveyed pediatricians reported critically high burnout. Again, inpatient ward staff reported significantly increased emotional exhaustion. After controlling for several confounders, we found that pediatricians with high work stress and emotional exhaustion reported reduced quality of care. Mediation analyses revealed that especially pediatricians emotional exhaustion partially mediated the effect of work stress on quality of care. Conclusions: Results demonstrate close relationships between increased work stress and burnout as well as diminished quality of care. High work stress environments in pediatric care influence mental health of pediatricians as well as quality of patient care. 40 41 42 43 44 45 46 47 WHAT IS NEW Our study reveals for the first time associations between work stress and burnout specifically for hospital pediatricians. Pediatricians with increased work stress and burnout reported mitigated quality of care. Beyond indirect effect of work stress through emotional exhaustion on quality of care we also observed direct detrimental effects of pediatricians work stress on mitigated care quality. Detrimental effects of the hospital work environment need to be constrained in order to promote pediatricians work life and care quality. 48 2

49 50 51 52 53 54 55 LIST OF ABBREVIATIONS (in alphabetical order) ERI Effort Reward Imbalance ERR Effort Reward Ratio ICU Intensive Care Unit OR Odd Ratio M Mean SD Standard Deviation 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 INTRODUCTION Physician distress has received growing attention in research - particularly due to various consequences for their individual health, productivity, and clinical safety [25]. Growing evidence suggests that mental well-being and efficacy of physicians are closely related to the characteristics of their work environment [10]. Physician distress and psychiatric disorders were linked to poor quality of health care, including medical errors, hospital acquired infections, and unsatisfied patients [8,15,24,5,12]. Particular focus has been put upon physician burnout due to its pervasiveness and detrimental effects on physician health as well as productivity [17,22,10]. Therefore, investigations of the interplay between physicians work environment, work stress, and quality of care are needed to inform respective preventive efforts [25]. Although burnout in hospital pediatricians is considered to be widespread, the evidence base is inconsistent [19]. Reported figures on affected pediatricians vary, particularly with respect to (1) career stages and (2) clinical work environment. Concerning (1) career stages, high burnout prevalence was reported for junior pediatricians. Fahrenkopf et al. [5] observed that 74% of residents in three US children s hospitals were classified as burned-out. In a one-year observational study on the effects of work hour restrictions, 75% and 57% of residents reported critical burnout scores over time [9]. In another two-year observational study, 17% to 46% of residents were classified as 3

76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 burned-out [14]. Hence research on the prevalence of burnout in hospital pediatricians in different career stages is still limited. A further factor is the actual (2) clinical work environment of hospital pediatricians. In comparison to outpatient care, risk for increased burnout was found to be higher in pediatric intensive care units (ICUs) [6]. Thus, a major risk factor for physician ill health is associated with the very nature of the clinical environment in hospitals [23]. Although a variety of upstream and personal factors contribute to burnout, psychosocial work stress has been identified as a predominant predictor of mental strain in physicians [20,10]. One prominent approach to evaluate the individual psychosocial work environment is the Effort-Reward-Imbalance (ERI) model [20,21]. It emphasizes potentially harmful effects of an imbalance between efforts spent at work and rewards received in return (i.e., high effort/low reward constellations). Among others, efforts encompass items that address time pressure, workflow interruptions, overtime, and work overload. Whereas rewards address salary, promotion prospects, esteem, and job security. Regarding physicians, evidence suggests that the ERI model provides valuable information on workrelated risk factors for impaired health and mental well-being [3,20]. To the best of our knowledge, there is no research investigating the interplay between work stress and burnout in hospital pediatricians to date. There is an inconsistent knowledge base on differences in work stress and burnout between junior and senior pediatricians, as well as among pediatricians in different clinical work environments. Finally, there is currently no study available on the interplay of work stress, burnout, and quality of care in hospital pediatricians. This study thus aimed to identify associations between work stress and burnout as well as their individual and joint contribution to pediatricians self-perceived quality of care. We further hypothesized that pediatrician burnout mediates the relationship between work stress and self-perceived quality of care. 101 102 103 MATERIALS AND METHODS Study sample and procedure 4

104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 Data for this cross-sectional survey study was collected at an academic children s hospital in Germany. It is a well-renowned university pediatric clinic with 150 patient beds. With its highly specialized services covering all fields of pediatric medicine, it is one of the largest pediatric hospitals in Central Europe. It includes three intensive care units (ICUs) with 40 beds as well as 5 inpatient wards. All pediatric specialties are represented in outpatient departments. The clinic serves 41.500 outpatients and 6.000 inpatients per year. The questionnaire-based study was advertised by e-mail, intranet, and through announcements during morning lectures. All pediatricians employed at the clinic were eligible for participation, except those on leave (e.g., maternal leave). 130 physicians were invited based on registration data from hospital administration. Enveloped questionnaires were sent to all eligible pediatricians. Completed surveys were directly returned to the external study team. The data collection period amounted to two months and included several reminders. A total of 96 questionnaires were returned (response rate: 73.8%). Four pediatricians reported to work solely in research labs with no patient contact. The final sample of N = 88 pediatricians was working in patient care. The study was approved by the Committee on Ethics of Human Research of the Medical Faculty, Ludwig-Maximilians-University Munich (124/07). All participants signed a letter of informed consent. 123 124 125 Measurement instruments Standardized questionnaires were used, in which pediatricians were asked to report on their work stress, burnout, self-perceived quality of care, and various demographic characteristics. 126 127 128 129 130 Work stress The 23-item Effort-Reward-Imbalance (ERI) questionnaire was applied to measure work stress [21]. This well-established questionnaire consists of two scales termed effort (six items, e.g., Over the past few years, my job has become more and more demanding ), and reward (11 items), including subscales for promotion (four items, e.g., Considering all my 5

131 132 133 134 135 136 137 138 139 140 141 142 efforts and achievements, my work prospects are adequate ), esteem (five items, e.g., Considering all my efforts and achievements, I receive the respect and prestige I deserve at work ), and job security (two items, e.g., My job security is poor ). Responses are scored on a five-point scale where a value of 1 indicates no stressful experiences, and a value of 5 indicates very high stressful experiences. Consequently, the range for the effort scale is 6 to 30, and 11 to 55 for the reward scale. Thus, higher scores reflect higher levels of effort and reward. According to a predefined algorithm, a ratio of the two scales effort and reward (weighted by item numbers) is calculated to quantify the degree of mismatch between high efforts (i.e., invested costs ) and low rewards (i.e., lack of gains ) at work (ERR: effortreward ratio). Thus, values close to zero indicate favorable conditions (relatively low effort, relatively high reward), whereas values beyond 1 indicate a high amount of effort spent that is not met by rewards received or expected in return [21]. 143 144 145 146 147 148 149 150 151 152 Burnout Burnout, as measured through its core components emotional exhaustion and depersonalization, was assessed with the German version of the Maslach Burnout Inventory [4]. Emotional exhaustion was assessed with a nine-item scale, where an example item is I feel burned out from my work. A six-point frequency scale was applied (0 = never/ occasionally to 5 = very often). Depersonalization was measured with a five-item scale (e.g., I have become more callous toward people since I took this job ). Thresholds to categorize pediatricians as having no/low or high burnout are based on normative classifications, where values above the scale means of emotional exhaustion (M > 3.5) and depersonalization (M > 2.5) are indicative of high burnout [7]. 153 154 155 156 157 Perceived quality of care To measure perceived quality of care, a two-item measure from a well-established German questionnaire for work analysis in hospitals was used [4]. The two items were My workload frequently leads to reduced quality of work and Adverse work conditions frequently lead to a loss of quality. A five-point scale was applied (1 = not at all to 5 = to a very great extent). 6

158 159 160 161 162 163 164 Socio-demographic characteristics Additionally, personal and work-related information was assessed: gender (1 = male, 2 = female), partnership/marriage (1 = yes, 2 = no), professional tenure (in years), average weekly working hours, shift work (1 = yes, 2 = no), type of contract (1 = full time, 2 = part time), current work environment (1 = inpatient ward, 2 = Intensive Care Unit, 3 = outpatient ward), and career stage/position (1 = head or senior physician, 2 = specialist, 3 = junior physician in specialty training). 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 Statistical Analyses Firstly, prevalence of socio-demographic and study variables was calculated (Table 1). Four pediatricians working solely in research labs were excluded from analyses. Secondly, we used analyses of variance (ANOVA) to examine differences in regard to the clinical work environment (inpatient ward, ICU, outpatient ward) and career stage (head or senior physician, specialist, junior physician) of pediatricians (Table 2). Additional post-hoc Scheffétests were applied to account for differences between group means. Thirdly, multivariate regression analyses were applied to detect associations between work stress and burnout (Table 3) as well as between work stress, burnout, and self-perceived quality of care (Table 4). In all multivariate analyses, we report crude and adjusted associations which account for gender, professional tenure, clinical work environment, and position (see Tables 2-4). Pediatricians with high work stress (i.e., more effort invested than reward received = Effort Reward ratio, ERR > 1.0) were compared to pediatricians with low work stress (ERR < 1.0). Additionally, we classified all participants into different subgroups according to their data on the components of the ERI measurement instrument. Respective thresholds were mean effort sum > 22, and mean reward sum > 34, with the subscales job promotion (M > 11), esteem (M > 16), and job security (M > 5). Predictor and outcome variables were standardized to represent an increase by one SD. Finally, an established approach for mediated regression analyses was applied to examine the proposed mediation effects [16]. First, main effects and subsequently indirect (mediator) effects were computed (Table 5). Mediation was formally tested with bootstrapped 7

186 187 188 189 190 191 estimates of the indirect effect of work stress on perceived quality of care through burnout. This approach was chosen because bootstrapped estimates are robust against violations of the assumption of normal distribution of variables [18]. Covariates (i.e., gender, professional tenure, clinical work environment, and position) were controlled for in testing the associations between the independent variable and mediator and between the mediator and dependent variable. All analyses were performed with SPSS 22.0. 192 193 RESULTS 194 195 196 197 198 199 Table 1 reports characteristics of the pediatrician sample. The majority worked in outpatient wards (40.9%), followed by inpatient wards (29.5%), and ICUs (29.5%). In regard to clinical facilities and age, both distributions were representative of the actual distribution in the examined hospital. Pediatricians in inpatient wards tended to be younger, were more frequently in postgraduate training positions, and worked significantly longer hours per week. ICU pediatricians were working more frequently in shift work schedules (see Table 1). 200 201 ** Table 1 about here ** 202 203 204 205 206 207 208 Pediatricians work stress, burnout, and self-perceived quality of care Table 2 reports pediatricians work stress, respective ERI-subscales, burnout, and selfperceived quality of care. All measures showed consistent reliabilities: Scales Cronbach s alpha were.76 for effort,.82 for reward,.90 for emotional exhaustion,.75 for depersonalization, and.89 for quality of care. Table 2 also reports comparisons of scale means between pediatricians in different clinical work environments and career stages/positions. 209 210 ** Table 2 about here ** 211 8

212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 The overall mean work stress score was ERR = 0.88 (SD = 0.33). Critically high work stress (ERR > 1.0) was reported by N = 25 (28.4%) pediatricians. Regarding the ERI-components, high effort was reported by N = 20 (22.7%), whereas low reward was observed in N = 18 pediatricians (20.5%). Pediatricians in inpatient wards reported significantly more work stress than their colleagues in ICUs and outpatient wards. They reported significantly more effort and less reward, including less esteem and job security (see Table 2). In regard to overall burnout, pediatricians reported higher emotional exhaustion than depersonalization (see Table 2). N = 9 pediatricians reported exhaustion above the critical threshold (M > 3.5, 10.2%), whereas N = 11 pediatricians reported critically high depersonalization (M > 2.5, 12.5%). None of the pediatricians scored above the critical threshold on both burnout dimensions. Concerning the clinical work environments, inpatient ward pediatricians reported higher emotional exhaustion as compared to ICU-based pediatricians. Junior pediatricians experienced significantly more feelings of depersonalization compared to head and senior pediatricians. A difference in self-perceived quality of care was observed: ICU pediatricians reported significantly higher quality of care as compared to their colleagues in inpatient wards. In the next step, multivariate analyses were conducted to investigate crude and adjusted associations between pediatricians work stress and burnout. We found that higher work stress was associated with increased emotional exhaustion (see Table 3). Regarding the ERI components, higher effort was significantly associated with increased exhaustion whereas higher reward was related to lower exhaustion. These associations were consistent for almost all reward subscales, which were related to less feelings of emotional exhaustion. We found no significant associations between work stress and depersonalization among the surveyed pediatricians (see Table 3). 237 238 ** Table 3 about here ** 239 9

240 241 242 243 244 In the following step, we accounted for relationships between pediatricians work stress, burnout, and their perceptions of their quality of care. Table 4 depicts crude and adjusted associations. Increased work stress was significantly associated with lower perceived quality of care; increased effort was significantly related with mitigated care whereas high reward was associated with increased quality of care. 245 246 ** Table 4 about here ** 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 Additionally we found that pediatricians with elevated emotional exhaustion reported significantly reduced quality of care. However, this association was not observed between depersonalization and quality of care (see Table 4). In a final step, we tested the proposed mediation model. The results of the mediated regression analyses are summarized in Table 5. All effects were estimated adjusting for gender, professional tenure, clinical work environment, and position. Since we found no significant associations between work stress and depersonalization (cf., Table 3), we exclusively focused on emotional exhaustion as a possible mediator. We thus investigated the effects of work stress (effort-reward imbalance, effort, reward) and emotional exhaustion (as a mediator) on self-perceived quality of care. Direct effects between the discrete variables replicate the above reported associations (cf., Table 5). Concerning general work stress (i.e., ERR), we found an indirect mediation effect (B = -.21, 95% CI -.39 - -.06) through emotional exhaustion on perceived quality of care. Furthermore, indirect, partial mediation effects for effort (B = -.21, 95% CI -.39 - -.07) as well as for reward (B =.22, 95 % CI.12 -.37) were observed. Notwithstanding, there were direct effects of work stress on perceived quality of care that were not mediated by emotional exhaustion. 264 ** Table 5 about here ** 265 266 DISCUSSION 10

267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 To the best of our knowledge, this is the first investigation examining the associations between Effort-Reward imbalance (ERI), burnout, and quality of care in hospital pediatricians. Our findings demonstrate that work stress and burnout are prevalent among pediatricians. Furthermore, we show that hospital pediatricians with increased effort-rewardimbalance were significantly more likely to report emotional exhaustion. This association is particularly pronounced in the face of high demands, i.e., effortful work environments in pediatric care. Moreover, our study contributes to growing evidence on the single and joint effects of detrimental work environments on poor physician well-being and associated effects on self-perceived quality of care. In our study sample, the prevalence of burnout was around 10% in participating pediatricians. Due to the limited generalizability of our sample, comparisons concerning burnout rates in pediatricians need to be treated very cautiously. Notwithstanding, our results show that there is a substantial amount of pediatricians who suffer from burnout. Concerning the impact of seniority, we found that junior pediatricians reported more depersonalization than their senior colleagues. This result may indicate that junior pediatricians have more difficulty developing adequate emotional strategies to deal with patients, e.g., empathy and concern for patients feelings. In the long term, declines in empathy are associated with inferior medical performance [28]. Our observations that pediatricians with increased exhaustion perceive mitigated quality of care are thus in line with previous investigations on the effects of increased distress and self-perceived medical errors among physicians [28]. However, potential influences of postgraduate training conditions as well as workload may explain further variance in the observed differences between junior and senior pediatricians. 290 291 292 293 294 In regard to pediatricians work stress, we investigated the overall imbalance between effort and reward as well as subcomponents of the work stress measure, i.e., three subcomponents of reward, which have not yet been addressed in previous research. Overall, around 20% of pediatricians reported critically high work stress, which was indicated by a 11

295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 substantial imbalance between effort and rewards received, i.e., ERR > 1.0 [21]. These results are in accordance with a nationwide study on German surgeons, in which 25% of the participants reported ERR above 1.0 [8]. Pediatricians in inpatient wards reported significantly higher work stress and burnout than their colleagues in intensive and outpatient care. Specifically, inpatient ward pediatricians reported increased effort as well as less reward at work, particularly in regard to esteem and job security. Potential post-hoc explanations are that inpatient ward pediatricians dealt with increased job demands (i.e., high work load and low staffing, high degree of administrative and coordinative duties, short-term contracts, workload sums up if ward pediatricians are not present, like during night shifts) as well as lack of respective reward and support (i.e., limited administrative support, low presence of senior physicians). This chronic work stress might have triggered increased exhaustion and reduced vigor in daily clinical work. In comparison, pediatricians in outpatient wards have more autonomy over their workload and work schedules. In ICUs, senior physicians are constantly present, the staffing is more balanced, and continuous presence of senior physicians, particularly during night shifts, allows compensating for times of excessive workload. Additionally, increased work time reduces the opportunities for recovery. Surveyed ICU physicians had shorter weekly work hours compared to inpatient ward pediatricians (cf., Table 1). Our results thus emphasize that the nature of organizational and contextual characteristics of the immediate clinical work environment can have a substantial impact on pediatricians work stress. Notwithstanding, potential differences between clinical units may also be attributed to further individual and organizational characteristics that were not included in this study. We found significant differences in average working time, contract, and shift work schedules between pediatricians of the units under study (see Table 1). Although we adjusted for these characteristics in multivariate analyses, we cannot exclude that further factors may have caused these differences, i.e., expertise or self-selection bias. We observed that increased work stress was linked to higher levels of burnout, particularly to emotional exhaustion [10]. While prospective evidence on the impact of work 12

323 324 325 326 327 328 329 330 331 332 333 334 335 stress on burnout in pediatricians is not available to date, similar studies in other medical specialties corroborate the impact of effort-reward imbalance and elevated risks of mental disorders and poor functioning at work [11]. Thus, promoting well-being of pediatricians includes reduction of the burden of psychosocial stress at work. In order to foster pediatricians functioning on the job, work environments should provide good working conditions in terms of fair effort and work demands, as well as sufficient rewards, such as recognition and appreciation from supervisors and hospital management. Lastly, our findings revealed a clear relationship between elevated work stress and mitigated perceived quality of care. Our results are in line with similar findings in German surgeons, where enhanced psychosocial work stress was linked to lower quality of care [12,8]. The observed associations contribute to increasing evidence of adverse work conditions in the hospital and their impact on medical performance and suboptimal care, which results in reduced patient safety [12]. 336 337 338 339 340 341 342 343 344 345 346 347 348 349 Limitations Although our study addressed various limitations of previous research, it has certain inherent limitations. First and foremost, the generalizability of our findings from a single academic pediatric clinic to other hospitals is unknown. We covered various clinical environments and included pediatricians of all career stages. However, specific characteristics of the academic work environment (e.g., potential work overload through teaching and research duties, complex cases) may limit external validity. Secondly, this study applied a cross-sectional design which does not allow for inferences about causality between study variables. This refers particularly to our mediator model, which cannot be entirely confirmed using cross-sectional data. Hence, our findings only allow a preliminary evaluation of the proposed relationships. Although there is a sound theoretical basis for the suggested directions between study variables, reverse or reciprocal effects cannot be ruled out (i.e., additional causal relationships from perceived quality of care to work stress). This further precludes assessments of the temporal relationships and magnitude of the associations 350 between work stress, burnout, and quality of care. However, there is strong longitudinal 13

351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 evidence for the effect of adverse working conditions on physicians well-being [23,27,3]. Concerning the direction of effects between physician stress and suboptimal performance, previous studies suggest that different causal directions may occur: For example, West al. [28] suggest a vicious cycle where suboptimal performance (i.e., medical errors) leads to personal stress, which then contributes to further deficits in patient care. Consequently, mental well-being of physicians and professional performance could be associated bidirectionally. Thirdly, the classification criteria and respective thresholds for the existence of burnout vary substantially, which limits comparability across studies [28,5]. The burnout measurement tool used in our study cannot diagnose burnout by itself. However, the applied classification approach is likely to detect clinically relevant burnout scores [7]. Since emotional exhaustion and depersonalization are considered as core components of professional burnout, we did not consider personal efficacy in our analyses which is considered as a further important component of burnout [13]. Fourthly, common method variance cannot be ruled out, as both exposure and outcome variables were collected through pediatricians self-reports. The item wordings on the quality of care measure refer to adverse work conditions, i.e., work overload. Hence, common method variance may lead to an inflation of observed associations between work stress and quality of care measures. Thus, multi-method designs are suggested for future research to rule out common method bias. Fifthly, quality of care was assessed via self-reports. Although physicians self-reports on suboptimal performance and medical errors are a feasible way to address these outcomes, there are scholarly discussions on the validity of self-identified care evaluations as well as the overlap between physician-identified performance outcomes and expert-based or objective performance ratings [28]. Our outcome measure consisted of two items what may limit the validity and reliability concerning overall quality of clinical care. Although this measure was derived from a well-established assessment tool, this particular scale has not been empirically associated with medical errors yet. Thus, we cannot infer about potential overlaps with actual quality of care such as medication errors, patient satisfaction, documentation errors, or delays in completing medical records. Sixthly, our findings draw 14

379 380 381 382 383 384 385 386 387 388 389 upon a comparatively limited number of pediatricians (N = 88). Thus, selection bias may have occurred, such that pediatricians with very high or very low burnout as well as increased work stress did not take part in our survey. Multivariate analyses that draw on small sample sizes may suffer from insufficient power in reliably estimating associations that take account for potential confounders of the work stress, burnout, and quality of care relationships. Another potential selection effect is the healthy worker effect, which suggests that participating pediatricians are more resilient to work stress and burnout. Further limitations refer to potential confounding variables that may have affected the investigated relationships between work stress and burnout, e.g., personality traits like self-reflection, self awareness and competence [1]. Because of the exploratory nature of our study, no multiplicity adjustment was applied. 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 Implications of the study Apart from adverse effects on physicians individual well-being, there are various cost implications caused by poorly designed work environments, e.g., through replacement of burned-out physicians or hazardous quality of care. Therefore interventions that mitigate contributing factors on work stress and pediatrician burnout are appropriate. Overall, our results support the notion that limiting work stress could be beneficial to pediatrician wellbeing as well as to quality of care. The improvement of hospital work environments with the objective of improving well-being could be a promising strategy to enhance performance and quality of pediatric care. We suggest system-based, i.e., organizational and work-design interventions that address the reciprocity of effort and reward in the workplace. This is of particular interest within the clinical context, where high job demands and detrimental working conditions lead to increasing resignation in physicians and jeopardize patient safety [8,24,11,12]. This could be achieved through reducing effort and work demands, while simultaneously improving reward in the hospital workplace. In regard to physicians working life, work overload and exceeding job demands could be limited, i.e., excessive over-time work, exceeding patient load, and poorly designed shift schedules [11,24]. Regarding rewards, respective measures should focus on improved career prospects based on junior 15

407 408 409 410 411 412 413 414 415 416 physicians training and professional achievements, and ways of providing non-material rewards from supervisors and management, e.g., through recognition, positive feedback, and support. Additionally, strengthening job control and autonomy could contribute to the promotion of physician well-being [11,27]. Our results also suggest that individual interventions and programs are needed to prevent, identify, and treat burnout among pediatricians. Thus, interventions that foster pediatricians individual resources, well-being, and capabilities to deal with high work demands should supplement organizational interventions [2,26]. Such interventions may address pediatricians personal awareness and self-care in order to develop coping strategies to deal with detrimental work conditions in the hospital. 417 418 419 420 421 422 423 424 CONCLUSION Work stress, in terms of effort-reward imbalance, is associated with emotional exhaustion and diminished quality of care in a sample of German pediatricians. Reducing the imbalance in working life, particularly reducing demands and improving reward at work, may have beneficial effects on pediatrician mental well-being. For future research, further longitudinal studies and controlled interventions to improve work environments of pediatricians should be evaluated in regard to their effect on pediatrician well-being and patient care. 425 426 427 428 429 430 Funding and Acknowledgements The study was partly funded by the Munich Center of Health Sciences (MC-Health). We gratefully acknowledge Tina Arenz, Andrea Schwarzer, and Christian Schröter for their support in study preparation and data collection. We also thank Jake Burns for his support in preparing the manuscript. 431 432 Ethics approval 16

433 434 This study was conducted with the approval of the Ethics Committee of the Medical Faculty of the Ludwig-Maximilians-University Munich (No. 124/07). 435 436 Conflict of Interest Statement: The authors declare that they have no conflict of interest. 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 REFERENCES 1. Alarcon G, Eschleman KJ, Bowling NA (2009) Relationships between personality variables and burnout: A meta-analysis. Work & Stress 23 (3):244-263 2. Bourbonnais R, Brisson C, Vezina M (2011) Long-term effects of an intervention on psychosocial work factors among healthcare professionals in a hospital setting. Occup Environ Med 68 (7):479-486. 3. Buddeberg-Fischer B, Klaghofer R, Stamm M, Siegrist J, Buddeberg C (2008) Work stress and reduced health in young physicians: prospective evidence from Swiss residents. Int Arch Occup Environ Health 82 (1):31-38 4. Büssing A, Perrar KM (1992) Measurement of Burnout. The German Version of the Maslach Burnout Inventory (MBI-D). Diagnostica 38:328-353 5. Fahrenkopf AM, Sectish TC, Barger LK, Sharek PJ, Lewin D, Chiang VW, Edwards S, Wiedermann BL, Landrigan CP (2008) Rates of medication errors among depressed and burnt out residents: prospective cohort study. BMJ 336 (7642):488-491 6. Garcia TT, Garcia PC, Molon ME, Piva JP, Tasker RC, Branco RG, Ferreira PE (2014) Prevalence of Burnout in Pediatric Intensivists: An Observational Comparison With General Pediatricians. Pediatr Crit Care Med. 7. Kleijweg JH, Verbraak MJ, Van Dijk MK (2013) The clinical utility of the Maslach Burnout Inventory in a clinical population. Psychol Assess 25 (2):435-441. 8. Klein J, Grosse Frie K, Blum K, von dem Knesebeck O (2011) Psychosocial stress at work and perceived quality of care among clinicians in surgery. BMC Health Serv Res 11:109. 17

460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 9. Landrigan CP, Fahrenkopf AM, Lewin D, Sharek PJ, Barger LK, Eisner M, Edwards S, Chiang VW, Wiedermann BL, Sectish TC (2008) Effects of the accreditation council for graduate medical education duty hour limits on sleep, work hours, and safety. Pediatrics 122 (2):250-258 10. Lee RT, Seo B, Hladkyj S, Lovell BL, Schwartzmann L (2013) Correlates of physician burnout across regions and specialties: a meta-analysis. Hum Resour Health 11:48. 11. Li J, Weigl M, Glaser J, Petru R, Siegrist J, Angerer P (2013) Changes in psychosocial work environment and depressive symptoms: a prospective study in junior physicians. Am J Ind Med 56 (12):1414-1422. 12. Mache S, Vitzthum K, Klapp BF, Groneberg DA (2012) Improving quality of medical treatment and care: are surgeons' working conditions and job satisfaction associated to patient satisfaction? Langenbecks Arch Surg 397 (6):973-982. 13. Maslach C, Schaufeli WB, Leiter MP (2001) Job burnout. Annual Review of Psychology 52:397-422 14. Pantaleoni JL, Augustine EM, Sourkes BM, Bachrach LK (2014) Burnout in pediatric residents over a 2-year period: a longitudinal study. Acad Pediatr 14 (2):167-172. 15. Paquet M, Courcy F, Lavoie-Tremblay M, Gagnon S, Maillet S (2013) Psychosocial work environment and prediction of quality of care indicators in one Canadian health center. Worldviews Evid Based Nurs 10 (2):82-94. 16. Preacher KJ, Hayes AF (2008) Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behav Res Methods 40 (3):879-891. 17. Shanafelt TD, Bradley KA, Wipf JE, Back AL (2002) Burnout and self-reported patient care in an internal medicine residency program. Ann Intern Med 136 (5):358-367 18. Shrout PE, Bolger N (2002) Mediation in experimental and nonexperimental studies: New procedures and recommendations. Psychol Methods 7 (4):422-445. 18

486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 19. Shugerman R, Linzer M, Nelson K, Douglas J, Williams R, Konrad R (2001) Pediatric generalists and subspecialists: determinants of career satisfaction. Pediatrics 108 (3):E40 20. Siegrist J, Shackelton R, Link C, Marceau L, von dem Knesebeck O, McKinlay J (2010) Work stress of primary care physicians in the US, UK and German health care systems. Soc Sci Med 71 (2):298-304. 21. Siegrist J, Starke D, Chandola T, Godin I, Marmot M, Niedhammer I, Peter R (2004) The measurement of effort-reward imbalance at work: European comparisons. Soc Sci Med 58 (8):1483-1499 22. Thomas NK (2004) Resident burnout. JAMA 292 (23):2880-2889 23. Tyssen R, Vaglum P (2002) Mental health problems among young doctors: an updated review of prospective studies. Harv Rev Psychiatry 10 (3):154-165 24. Virtanen M, Kurvinen T, Terho K, Oksanen T, Peltonen R, Vahtera J, Routamaa M, Elovainio M, Kivimaki M (2009) Work hours, work stress, and collaboration among ward staff in relation to risk of hospital-associated infection among patients. Med Care 47 (3):310-318. 25. Wallace JE, Lemaire JB, Ghali WA (2009) Physician wellness: a missing quality indicator. Lancet 374 (9702):1714-1721 26. Weigl M, Hornung S, Angerer P, Siegrist J, Glaser J (2013) The effects of improving hospital physicians working conditions on patient care: a prospective, controlled intervention study. BMC Health Serv Res 13:401. 27. Weigl M, Hornung S, Petru R, Glaser J, Angerer P (2012) Depressive symptoms in junior doctors: a follow-up study on work-related determinants. Int Arch Occ Env Hea 85 (5):559-570. 28. West CP, Huschka MM, Novotny PJ, Sloan JA, Kolars JC, Habermann TM, Shanafelt TD (2006) Association of perceived medical errors with resident distress and empathy: a prospective longitudinal study. JAMA 296 (9):1071-1078 513 19

514 515 Table 1 Characteristics of participating pediatricians Age (years) Professional tenure (years) Average working hours per week Gender Career Stage/ Position Partnership Contract (M ± SD) (M ± SD) (M ± SD) Male Female Head or senior physician Specialist Junior physician Yes No Full time Overall Group N = 88 37.46 ± 8.55 7.85 ± 6.71 53.66 ± 11.55 41 (46.6%) 47 (53.4%) 28 (31.8%) 21 (23.9%) 39 (44.3%) 77 (87.5%) 10 (11.4%) 80 (90.9%) Clinical work environment Outpatient Wards N = 36 (40.9%) 39.86 ± 9.30 9.10 ± 7.24 48.92 ± 11.91 16 (44.4%) 20 (55.6%) Inpatient Wards N = 26 (29.5%) 34.50 ± 8.07 5.48 ± 5.41 59.50 ± 7.41 12 (46.2%) 14 (53.8%) 15 (41.7%) 5 (19.2%) 8 (22.2%) 3 (11.5%) 13 (36.1%) 33 (91.7%) 3 (8.3%) 29 (80.6%) 18 (69.2%) 22 (84.62%) 3 (11.54%) 26 (100%) ICU N = 26 (29.5%) 37.19 ± 7.17 8.42 ± 6.73 54.42 ± 11.94 13 (50.0%) 13 (50.0%) 8 (30.8%) 10 (38.5%) 8 (30.8%) 22 (84.6%) 4 (15.4%) 25 (96.2%) Part time 8 (9.1%) 7 (19.4%) 0 (0.0%) 1 (3.8%) Significance Test ANOVA (F; p) / Chi² 3.09;.05 2.34;.10 7.36; <.01.19;.91 11.68;.02.75;.69 8.14;.02 516 517 518 519 Shift work Yes No 22 (25.0%) 63 (71.6%) 3 (8.33%) 5 (19.2%) 32 (88.89%) 21 (80.8%) 14 (53.85%) 10 (38.46%) 19.24;.00 Note: ICU = Intensive Care Unit, M = Mean, SD = Standard deviation, Significance testing: ANOVA/Chi-Square test, bolded if p <.05. Frequency of missing values is not separately depicted. 20

520 Table 2: Pediatricians burnout, work stress, and self-perceived quality of care Clinical work environment Career Stage/ Position Variables Work Stress Scale Range (very low/never very high/very often) Overall Group Outpatient Wards Inpatient Wards ICUs Significance Head or senior pediatrician s Specialist s Junior physicians Significance N = 88 N = 36 N = 26 N = 26 N = 28 N = 21 N = 39 F; p (M ± SD) (M ± SD) (M ± SD) (M ± SD) (F; p) (M ± SD) (M ± SD) (M ± SD) 521 522 523 Burnout Effort-Reward Ratio.88 ±.33.82 ±.30 1.07 ±.36.76 ±.28 7.29; <.01.86 ±.30.91 ±.34.87 ±.35.16;.85 Effort (sum) (6-30) 19.33 ± 4.18 18.50 ± 3.89 21.62 ± 4.23 18.19 ± 3.74 Reward (sum) (11 55) 43.02 ± 8.33 43.61 ± 8.14 39.12 ± 7.83 46.12 ± 7.82 Reward: Esteem Scale 20.70 ± (5-25) (sum) 3.99 aa 21.53 ± 3.48 aa 18.46 ± 4.07 21.85 ± 3.74 Reward: Job Security Scale (2-10) 7.93 ± 2.43 a 8.14 ± 1.97 a 9.08 ± 6.50 ± 2.89 (sum) 1.76 Reward: Promotion Scale (sum) Emotional Exhaustion (Mean) (4-20) 14.53 ± 3.95 14.33 ± 4.61 14.15 ± 3.47 15.19 ± 3.43 (0 5) 2.52 ±.90 Depersonalization (Mean) (0 5) 1.40 ±.81 Quality of Care Self-perceived quality of care 2.53 ±.95 2.88 ±.82 2.15 ±.79 1.47 ±.81 1.52 ±.73 1.18 ±.88 (1 5) 2.53 ± 1.11 2.49 ± 1.00 2.02 ±.96 3.12 ± 1.13 6.22; <.01 18.50 ± 4.19 20.19 ± 4.04 19.46 ± 4.25 1.02;.37 5.20;.01 41.89 ± 8.09 43.19 ± 8.35 43.74 ± 8.61.40;.67 6.70; <.01 19.89 ± 3.86 a 21.45 ± 3.63 a 20.87 ± 4.24.95;.39 8.93; <.01 8.19 ± 1.88 a 8.24 ± 2.39 7.59 ± 2.78.70;.50.52;.60 14.00 ± 4.37 13.86 ± 3.50 15.28 ± 3.82 1.27;.29 4.61;.01 2.42 ±.81 2.52 ±.82 2.60 ± 1.02.30;.74 1.36;.26 1.16 ±.79 1.28 ±.87 1.64 ±.74 3.25;.04 7.39; <.01 2.63 ± 1.16 2.55 ± 1.12 2.46 ± 1.08.18;.84 Note: ICU = Intensive Care Unit, M = Mean, SD = Standard deviation, Significance testing: ANOVA, bolded if p <.05. Frequency of missing values is not separately depicted. a One missing value, aa Two missing values. 21

524 Table 3: Associations between work stress and (low and high) burnout among pediatricians (results of logistic regression analyses) Burnout Emotional Exhaustion Depersonalization Low High Associations and Significance Testing Low High Associations and Significance Testing Work Stress N = 79 N = 9 Crude Adjusted N = 77 N = 11 Crude Adjusted (M ± SD) (M ± SD) OR [95% CI] OR [95% CI] (M ± SD) (M ± SD) OR [95% CI] OR [95% CI] 525 526 527 Effort-Reward Ratio.82 ±.27 1.41 ±.36 Effort (sum) Reward (sum) Reward: Esteem Scale (sum) Reward: Job Security Scale (sum) Reward: Job Promotion Scale (sum) 18.71 ± 3.77 24.78 ± 3.80 44.11 ± 33.44 ± 7.88.5.88 21.21 ± 3.68 aa 4.06 8.22 ± 2.28 a 2.35 14.86 ± 11.67 ± 3.85 3.81 2.13 [1.28 3.53] 1.79 [1.12 2.87].61 [.39 -.95].62 [.39 -.97] aa.08 [.01 -.45] aaa.64 [.41 1.00] a.26 [.10 -.70] aa.74 [.48 1.13] 22.90 [2.99 1.41 [.74-175.11] a.86 ±.33.98 ±.31 1.17 [.77 1.80] 2.70] a 20.11 [2.81 19.16 ± 20.55 ± 1.37 [.68-144.06] a 1.16 [.76 1.77].4.30 3.14 2.79] a.83 [.02-43.35 ± 40.73 ±.41] a 8.19 9.35.87 [.57 1.33].67 [.34-1.34] a 20.88 ± 19.45 ± 3.87 aa.85 [.55 1.31] aa.63 [.32-4.76 1.25] aaa 8.00 ± 7.45 ± 2.44 a 2.42.90 [.59 1.38] a.84 [.43-1.65] aa.36 [.15-14.64 ± 13.82 ±.85] a 3.89 4.45.91 [.60 1.39].75 [.39-1.46] a Note: M = Mean, SD = Standard deviation; OR = Odd Ratio (reported for standardized predictor variables); Associations were adjusted for confounders including gender, professional tenure, career stage/position, and clinical work environment; bolded if p <.05. a One missing value, aa Two missing values, aaa Three missing values. 22

528 529 530 531 532 533 534 Table 4: Associations between work stress, burnout and pediatricians self-perceived quality of care Work Stress Burnout Effort-Reward Ratio Low (ERR < 1) 63 High (ERR > 1) Perceived Quality of Care Associations and Significance Testing Crude Adjusted N (M ± SD) B [95% CI] B [95% CI] 25 Effort Low 68 High 20 Reward Low 18 Reward: Esteem Scale Reward: Job Security Scale Reward: Job Promotion Scale Emotional Exhaustion High 70 Low 14 High 72 Low 15 High 72 Low 20 High 68 Low 79 High 9 Depersonalization Low 77 High 11 2.90 ± 1.05 1.18 [.78 1.03 [.59 1.47] 1.60 ± 1.58] 0.60 2.82 ± 1.15 [.71 1.06 1.60] 1.04 [.59 1.49] 1.55 ± 0.58 0 0 1.91 ± -.70 [-1.21 - - -.58 [-1.08 - - 0.71.20].07] 2.69 ± 1.14 0 0 1.93 ± -.67 [-1.23 - - 0.76.10] -.39 [-1.0 -.22] 2.66 ± 1.13 0 0 1.77 ± -.83 [-1.37 - - 0.80.29] -.55 [-1.16 -.06] 2.68 ± 1.11 0 0 2.12 ± -.48 [-.98-0.81.02] -.50 [-1.0 - -.02] 2.65 ± 1.16 0 0 2.63 ±.87 [.20 1.11 1.55].75 [.08 1.42] 1.67 ± 0.71 0 0 2.56 ±.22 [-.42-1.09.87].17 [-.45 -.80] 2.31 ± 1.29 0 0 Note: M = Mean, SD = Standard deviation; ERR = Effort-Reward Ratio; B = Regression coefficient; Tests for associations were computed with standardized quality of work scores; Associations were adjusted for confounders including gender, professional tenure, clinical work environment, and career stage/position; Regression coefficients for confounders are not depicted; Pediatricians with missing data were not excluded; bolded if p <.05. 23

535 536 Table 5: Results of mediation analyses of work stress (ERI, effort, or reward) on quality of care through emotional exhaustion (1) Work stress (effort-reward-imbalance) Emotional Exhaustion Perceived quality of care a) Direct effects B SE t p Work stress (effort-reward-imbalance) on quality of care -.52.10-5.39 <.01 Emotional exhaustion on quality of care -.30.12-2.41.02 Work stress (effort-reward-imbalance) on emotional exhaustion.72.09 8.43 <.01 Bootstrapping CI (95%) b) Indirect effect B SE lower upper ERI on quality of care through emotional exhaustion -.21.08 -.39 -.06 Model summary R² =.43, F(df) = 8.43(79), p <.01 (2) Work stress (effort) Emotional Exhaustion Perceived quality of care a) Direct effects B SE t p Work stress (effort) on quality of care -,49.10-5.11 <.01 Emotional exhaustion on quality of care -.33.11-2.88.01 Work stress (effort) on emotional exhaustion.64.09 7.05 <.01 Bootstrapping CI (95%) b) Indirect effect B SE lower upper Work stress (effort) on quality of care through emotional exhaustion -.21.08 -.39 -.07 Model summary R² =.43, F(df) = 8.46(79), p <.01 537 538 539 540 (3) Work stress (reward) Emotional Exhaustion Perceived quality of care a) Direct effects B SE t p Work stress (reward) on quality of care.44.10 4.40 <.01 Emotional exhaustion on quality of care -.38.11-3.46 <.01 Work stress (reward) on emotional exhaustion -.58.10-6.07 <.01 Bootstrapping CI (95%) b) Indirect effect B SE lower upper Work stress (reward) on quality of care through emotional exhaustion.22.06.12.37 Model summary R² =.41, F(df) = 7.98(79), p <.01 Note: Work stress, burnout, and quality of care scores were standardized; Associations adjusted for gender, professional tenure, clinical work environment, and position. Regression coefficients for confounders are not depicted. 24