Doctors' Stress Responses and Poor Communication Performance in Simulated Bad-News Consultations

Abstract

Purpose 

No studies have previously evaluated factors associated with high stress levels and poor communication performance in breaking bad news (BBN) consultations. This study determined factors that were most strongly related to doctors' stress responses and poor communication performance during a simulated BBN task.

Method 

In 2007, the authors recruited 24 doctors comprising 12 novices (i.e., interns/residents with 1–3 years' experience) and 12 experts (i.e., registrars, medical/radiation oncologists, or cancer surgeons, with more than 4 years' experience). Doctors participated in simulated BBN consultations and a number of control tasks. Five-minute-epoch heart rate (HR), HR variability, and communication performance were assessed in all participants. Subjects also completed a short questionnaire asking about their prior experience BBN, perceived stress, psychological distress (i.e., anxiety, depression), fatigue, and burnout.

Results 

High stress responses were related to inexperience with BBN, fatigue, and giving bad versus good news. Poor communication performance in the consultation was related to high burnout and fatigue scores.

Conclusions 

These results suggest that BBN was a stressful experience for doctors even in a simulated encounter, especially for those who were inexperienced and/or fatigued. Poor communication performance was related to burnout and fatigue, but not inexperience with BBN. These results likely indicate that burnout and fatigue contributed to stress and poor work performance in some doctors during the simulated BBN task.

Breaking bad medical news (BBN) to patients and their families is a task that many doctors perform daily.¹ Skillful communication is required to achieve good patient outcomes including improved information recall,² better psychological adjustment to cancer,³ and increased satisfaction with medical care.⁴ A number of best-practice clinical guidelines, including the American SPIKES protocol⁵ and Australian Clinical Practice Guidelines,⁶ are available to assist doctors in BBN. The six steps of the SPIKES protocol comprise the codified principles of Setting, Perception, Invitation, Knowledge, Empathy, and Summary. The Australian Clinical Practice Guidelines are a set of consensus guidelines for BBN endorsed by the American Society of Clinical Oncology. These protocols impose a significant cognitive burden on busy clinicians, trying to remember all of these steps while under the duress of BBN.

The recent focus on doctors' communication performance has likely made BBN transactions more difficult, particularly if doctors perform the task poorly.⁷ Doctors report thinking often about BBN.⁸ They regard it as one of the most difficult tasks they engage in, and they report experiencing anticipatory stress, fear, and anxiety in relation to BBN.^9–11 Such stress, when experienced at a high level or for a prolonged period of time, may lead to adverse outcomes in patients. For example, doctors may delay BBN despite patients wanting to hear it,¹² or doctors may avoid situations in which they must discuss prognosis,¹³ because they feel discomfort in these transactions.¹¹

However, little empirical evidence exists that indicates how doctors respond to these medical situations and the increased pressure to “get it right.” In general, doctors report quite high levels of anxiety, depression,^14–16 and stress,^17,18 although it is not clear whether specific stressors (such as BBN) may accumulate and increase a doctor's stress or distress levels over time. Also unclear is whether stress that physicians perceive during these encounters can adversely impact their clinical and communication performance. However, research shows that task performance generally decreases during high-stress situations,^19,20 so for this study we anticipated impaired communication in some highly stressed doctors.

Extensive literature in psychology has contributed a substantial body of work around the physiological responses to stress within experimental/simulated conditions. Most believe that maximization of hypothalamic-pituitary-adrenal axis (HPAA) activation requires a rather specific setting with ego involvement and social-evaluative judgment by others.²¹ Some real-life situations are reproducible enough to serve as standard stressors. A stressful examination, for example, gives rise to both HPAA and sympathoadrenal stimulation.²²

Most of the BBN literature is based solely on opinion and presents no empirical data,²³ although two studies have recently included evaluations of the impact of simulated BBN consultations on medical students. van Dulmen and colleagues²⁴ found that students showed an anticipatory cortisol response (i.e., cortisol is increased during the body's “fight or flight” response to stress), high autonomic arousal (i.e., high systolic blood pressure [SBP] and heart rate [HR]), and high globally assessed stress and state-anxiety levels during the BBN task. In addition, Cohen and colleagues²⁵ found that breaking both good and bad news resulted in high HR, SBP, and self-reported distress levels compared with controls, although the bad-news task was more stressful than the good-news task. To date, no other publications have evaluated predictors of or associations with students' stress responses in simulated tasks. Such an evaluation might better inform the development of targeted teaching interventions that may improve medical student and doctor performance in these important clinical interactions.

Ptacek and Eberhardt²⁶ suggest that the Transactional Model of Stress and Coping (TMSC)²⁷ has utility in explaining doctors' stress responses to BBN. They propose that doctors' discomfort during the task should vary as a function of their BBN experience, their perception of the severity of the news, and their perception that they were partly responsible for the news. The present study evaluated experience and skill in communicating bad news and the type of medical news given in relation to doctors' stress responses during a simulated BBN task.

We explored relationships among BBN experience, type of news given, intrinsic doctor factors, communication performance, and doctors' stress levels through simulated-patient consultations. (In this study, we did not examine either the severity of the news or the doctors' perception of the news examined.)

The main aim of this study was to determine whether experience and/or consultation or intrinsic doctor factors were related to doctors' stress responses and poor communication performance in a simulated BBN task. In accordance with the literature, we expected that (1) doctors would show higher stress levels breaking bad versus good news, (2) higher mean HR and HR variability (HRV) changes in the BBN consultation would be related to doctors' inexperience with BBN, poor communication performance, and type of news given, and (3) poorer communication performance would be related to doctors' inexperience with BBN and burnout (i.e., high emotional exhaustion and depersonalization, low personal accomplishment).

Method

In 2007, we conducted this study with full approval from the human research ethics committee of the University of New England.

Experimental design

We operationalized experience as experience group (novice or expert), number of previous times BBN (none, 1–4, 5–10, >10 ≤ 20, or >20), and years of BBN experience (none, 1–2 years, 3–5 years, > 5 ≤ 10 years, or >10 years). We operationalized type of news as a good-news scenario (i.e., diagnosis of degenerative bone disease when the patient feared cancer recurrence) or a bad-news scenario (i.e., recurrence of cancer). We rated doctors' communication performance using the brief Communication Rating Scale (CRS). We also evaluated intrinsic doctor factors (i.e., perceived stress, anxiety, depression, fatigue, and burnout) in relation to doctors' stress responses during the BBN task. These last factors have not previously been evaluated in this regard.

We examined doctors' stress responses during the consultations using HR and HRV measures. Previous studies have shown that increased sympathetic cardiac control occurs during stressful situations, as evidenced by HR and SBP.²⁸ HRV has also been used to study the effects of mental stress on physiological arousal.²⁹ Reports show that such measures are sensitive to minor laboratory stressors,^29,30 daily hassles,³¹ and changes in acute stress and anxiety.^32,33 Thus, for our study we used HR and HRV measures (detailed below) as surrogate markers of doctors' stress responses.

Finally, we evaluated correlates of doctors' poor communication performance in the BBN task. Such an evaluation might shed light on potentially modifiable factors that may contribute to poor clinical performance in some doctors. For example, higher than normal levels of psychiatric morbidity are reported in doctors,^17,34 as are burnout^35,36 and significant fatigue.³⁷ The literature has previously linked these states to poor clinical performance and decreased competence in doctors.^18,38,39 No studies, however, have previously evaluated factors that may be related to poor communication performance during specific stressful encounters such as BBN.

We studied the associations of BBN, communication performance, and stress via simulated patient consultation as others had previously.⁴⁰ The approach is also routinely used in medical student training at the University of Sydney, Australia, and other medical schools. Medical educators have suggested that the technique of simulating BBN to standardized patients helps students understand the importance of psychosocial and emotional aspects of patient care.^41,42 Further, working with simulated patients is the closest experience to an actual patient encounter.^43,44

Participants

We recruited novice doctors (interns or residents) indirectly via a single announcement at a junior medical officer training session, in which we detailed the BBN simulation and measured the HRV of an actor; approximately half of the doctors later contacted us about participating in the study. One of us (S.D.) approached experienced doctors directly and invited them to participate; one expert doctor declined because of an acute illness and two because of scheduling difficulties. All 24 participating doctors were in regular contact with hospitalized patients, all described themselves as being in good health without a history of cardiovascular disease, and none were taking medications that may have affected cardiovascular function (e.g., beta-blockers). We asked the participants to follow their usual routines with respect to caffeine intake on the day of participation.

Materials

We used the Polar S-Series s180i monitoring system to assess HR and HRV. This noninvasive, watch-sized device was placed on doctors' wrists, and a transducer band was placed around their chest with the transmitter positioned in the midline, directly below the xiphoid. Error rates for the device were low (<0.05%); polar software replaced missing beats, when identified, with the means of the surrounding values. The apparatus has previously been used to assess the effects of mental stress on HR and HRV, and its measures are highly correlated with values obtained from other commercial systems.²⁹

Physiological measures

We used frequency and time-domain analyses to assess HRV, according to the recommendations for analysis and interpretation of HR and HRV data.⁴⁵Frequency-domain analysis generates three spectral components; only two are indicated for short-interval recordings of up to five minutes: low frequency (0.04–0.15 Hz) and high frequency (0.15–0.4 Hz). We used absolute power and normalized units to describe the frequency components; absolute power provides a recording of power values in milliseconds squared, whereas normalized units report values of each power component in proportion to the total power, minus the variable low-frequency component. An increase in absolute power and a decrease in normalized units components have previously been reported in tasks involving mental stress.²⁸

Time-domain analysis examined raw RR intervals (i.e., time between two consecutive R waves in the electrocardiogram) in time. We derived a number of measures according to European Task Force recommendations⁴⁵ based on the interval between consecutive normal beats reflecting the underlying sinus rhythm. These measures comprised (1) standard deviation of normal to normal intervals (SDNN), (2) standard deviation of the average normal to normal intervals (SDANN), (3) square root of the mean normal to normal interval (RMSS), and (4) percentage of adjacent pairs of normal to normal intervals differing by more than 50 milliseconds (pNN50). RMSS and pNN50 values have previously been shown to change in response to mental stressors.²⁹

Medical scenarios

We used two scenarios, a good-news (i.e., control condition involving a noncancer diagnosis) and a bad-news scenario (i.e., experimental condition involving a diagnosis of cancer recurrence). The good-news scenario involved a 50-year-old woman with a prior history of breast cancer who presented to the emergency department with supraclavicular pain, radiating to the right arm. Tests later indicated that the pain was caused by degenerative bone disease and was not due to a recurrence of the cancer. In the bad-news scenario, the patient presented with the same history, but her test results indicated that her cancer had returned. The doctors' task was to provide the patient with the news and discuss the need to investigate the lesion, consult an oncologist, and manage any pain. A different experienced actress who had rehearsed and received detailed dramaturgical instruction performed each scenario. We asked the doctors to respond to each actress as they would any other patient in this situation.

Psychosocial measures

The Depression Anxiety Stress Scale (DASS) is a self-report scale designed to assess stress, anxiety, and depression.⁴⁶ The DASS is a 42-item scale which asks participants to rate how much each symptom applies to them from does not apply to me at all (0) to applies to me very much (3); higher scores indicate higher levels of the states. This scale has adequate validity and reliability when administered to nonclinical samples.⁴⁷

The Maslach Burnout Inventory–Human Services Survey⁴⁸ is the most widely used measure of burnout. This 22-item scale assesses burnout on three dimensions: emotional exhaustion, depersonalization, and personal accomplishment (reverse scored). We asked doctors to rate statements on six-point Likert-like scales, with higher scores indicating more burnout symptoms. Adequate reliability and validity are reported for the scale amongst physicians.⁴⁹

We assessed subjective impact of fatigue by asking doctors to rate the extent to which fatigue had caused a problem for them over the past month, using a 10-point visual analogue scale, ranging from 0, no problem to 9, extreme problem; higher scores indicate a greater impact of fatigue.

Communication performance

We used the CRS to evaluate doctors' communication performance⁵⁰ in the good- and bad-news tasks. Ratings were made within a dramatic/narrative context on three dimensions: managing the patient's information needs, managing the patient's emotional needs, and the doctor's body language. Ratings for managing emotional needs, for example, were 1 (rude, condescending), 2 (uninterested, blocking), 3 (deflects discussion), 4 (allows minimal discussion), 5 (acknowledges and responds), and 6 (elicits and responds). A single trained rater, one of us (K.B.), rated all of the doctors' performance on each dimension on scales from 1 to 6, with higher scores indicating better performance.⁵¹ Each dimension was rated separately and then summed to give a total CRS score; we used only total CRS scores in this study. The CRS has high internal consistency reliability (Cronbach alpha = 0.86) (Shaw J, University of Sydney, unpublished PhD thesis, 2007⁵¹).

Procedure

We conducted the testing in a quiet room with an ambient temperature of 22°C (about 72°F). Doctors were fitted with the monitoring device and recorded while engaging in a number of tasks:

1. completing (while alone in the room) a short questionnaire about their BBN experience and recent history of stress, anxiety, depression, burnout, and fatigue;
2. participating in scenario 1 (i.e., breaking good or bad news);
3. taking a short medical history (i.e., a filler, low-stress control task allowing HR to return to baseline; six minutes);
4. participating in scenario 2 (i.e., breaking good or bad news); and
5. receiving feedback on their communication performance.

The sequence of tasks was identical for all participants except that the order of the two scenarios was counterbalanced across study participants to control for any order effects.

A wall-mounted video camera recorded the medical scenarios. We used the recordings to rate doctors' communication performance and time-sequence HR data with any movement or other artifacts.

We selected five-minute HR epochs for each of the aforementioned activities (two breaking news scenarios and one control task) and corrected the errors as described. The tachogram was scanned to confirm that all aberrant beats were removed. HRV analyses used interbeat intervals that were generated on a beat-to-beat basis as the time difference in peak voltage between two successive R-waves.

Statistical analyses

We used SPSS (version 15; Chicago, Illinois) to generate descriptive statistics for univariate categorical and continuous measures. A between-groups analysis of variance (ANOVA) evaluated differences between novices and experts with respect to intrinsic doctor factors and communication performance during the BBN task. Paired samples t tests evaluated stress response differences in doctors with regard to the good- and bad-news tasks. Correlations were reported as Pearson product moment correlations r. We checked the data for normality and statistical outliers and applied appropriate corrections. We used simultaneous multiple regression analyses to determine cross-sectional predictors of HR, HRV, and communication performance. Because of the small sample size, we included only three independent variables in the regression models. We selected independent variables to include in the models on the basis of their statistical significance (P < .05). Independent variables were the BBN experience variables (i.e., experience group, number of times BBN, and years of BBN experience), scenario type, and intrinsic doctor factors (i.e., stress, anxiety, depression, burnout, and fatigue). Dependent variables were HR, low frequency, absolute power, high frequency, and normalized units (i.e., frequency-domain measures); SDNN, SDANN, RMSS, and pNN50 (i.e., time-domain measures); and total communication performance score.

Results

Twenty-four doctors participated in the study including 12 novices (5 male, 7 female) and 12 experts (9 male, 3 female). Novices were interns or residents (with 1–3 years of experience) working in a range of specialties (we did not record their current workplaces). Experts were registrants (with more than 4 years of experience) or consultants in oncology (with more than 8 years of experience) working at public hospitals in Sydney, Australia. Experts were older than novices (mean = 42 ± 8 years; [range: 30–54 years] versus mean = 29 ± 6 years; [range: 24–49 years]), and a greater proportion of them were male (75%), whereas more novices were female (58%). Most experts (8/12) had more than 10 years of BBN experience, and the remainder had 3 to 10 years of BBN experience; most experts (11/12) had broken bad news more than 20 times, although one had performed the task 10 to 20 times. Novices had either no BBN experience (4/12) or 1 to 2 years of BBN experience (7/12), although one novice had 3 to 5 years of BBN experience. Most novices had never broken bad news or had done so only 1 to 4 times in their careers, although one had performed the task 5 to 10 times. Experts had more years of BBN experience than novices (χ₄² = 21.333, P = .000) and had broken bad news more times than novices (χ₄² = 24.0, P = .000).

We computed descriptive statistics for intrinsic doctor factors and communication performance for novices and experts separately (Table 1). ANOVA indicated that novices and experts did not differ on intrinsic doctor factors or communication performance.

Table 1: Means and Standard Deviations (SD) of Doctors' Stress, Psychological Distress, Fatigue, Burnout, and Communication Performance Scores Between Novice and Expert Australian Doctors, The University of Sydney Northern Clinical School, 2007

HR and HRV measures peaked during the consultation and quickly returned to baseline after the task. We computed descriptive statistics for five-minute-epoch HR and HRV measures separately for experts and novices in the good- and bad-news tasks (Table 2). Paired samples t tests indicated that doctors' stress responses, as indexed by HR and HRV measures, differed between the good- and bad-news tasks (Table 3). In separate analyses, we evaluated initial 30-second HR and HRV data; these results were similar to those using mean five-minute-epoch data (analyses not shown). Results showed that doctors' HRs were significantly higher during the bad-news task compared with the good-news control condition; HRV components in the bad-news scenario were lower for each of the following: low frequency, SDNN, RMSS, SDANN, and pNN50 (Table 2).

Table 2: Means and Standard Deviations (SD) of Heart Rate and Heart Rate Variability Measures During Good-News and Bad-News Scenarios, Between Novice and Expert Australian Doctors, The University of Sydney Northern Clinical School, 2007

Table 3: Means and Standard Deviations (SD) of Heart Rate and Heart Rate Variability Difference Measures in 24 Australian Doctors During Good-News and Bad-News Scenarios, The University of Sydney Northern Clinical School, 2007

Simultaneous multiple regression analyses indicated that mean five-minute-epoch HR and some HRV measures were associated with BBN experience and fatigue (Table 4). Specifically, HR was related to the number of times BBN; low frequency and pNN50 were related to experience group; and normalized units were related to fatigue.

Table 4: Simultaneous Multiple Regressions of the Association of 24 Australian Doctors' Experience, Communication Performance, Intrinsic Doctor Factors, and Autonomic Arousal, as Indexed by Heart Rate (HR) and Heart Rate Variability (HRV), at The University of Sydney Northern Clinical School, 2007

Multiple regression analyses determined that poor BBN performance was related to high depersonalization and fatigue level (Table 5). This regression model accounted for 36% of the variance in communication performance, with an adjusted R² of 0.358.

Table 5: Simultaneous Multiple Regression Analysis of the Association of 24 Australian Doctors' Experience, Communication Performance, and Intrinsic Doctor Factors to Doctors' Total Communication Rating Scale Scores, The University of Sydney Northern Clinical School, 2007

Discussion

This is the first systematic evaluation of doctors' stress responses and poor communication performance, as related to their intrinsic factors, in simulated BBN consultations. Doctors showed an early stress response which peaked during the consultation and rapidly diminished after the task. A similar trajectory was reported by van Dulmen and colleagues²⁴ using salivary, cortisol, SBP, HR, and stress and state-anxiety levels in medical students BBN to actual patients. Our results suggest that BBN to a simulated patient is not the same as breaking the news to a real patient, with the latter scenario likely resulting in a more substantial and prolonged stress response.

As expected, we recorded higher stress levels in the bad-news consultation, relative to the good-news consultation, as indexed by HR and some HRV measures (e.g., SDNN, SDANN, RMSS, pNN50, and low frequency). This finding is consistent with those of Cohen and colleagues,²⁵ who found that HR and SBP were higher in a simulated bad- versus good-news task. These results are also consistent with previous studies indicating that certain time-domain HRV measures (e.g., RMSS, pNN50) can change in response to mental stressors,²⁹ as can some frequency-domain HRV measures (e.g., absolute power, normalized units).²⁸

Multiple regression analyses indicated that doctors' experience, operationalized as either the number of times they had broken bad news or their experience group (i.e., novice or expert), was related to HR and some HRV measures (i.e., pNN50, low frequency). In addition, doctors' fatigue levels were related to an HRV measure (i.e., normalized units). However, perceived stress, psychological distress (i.e., anxiety, depression, and burnout), and poor communication performance were not related to high stress levels in the BBN task.

Taken together, these results suggest that fatigued and inexperienced doctors were more likely to experience stress during the simulated BBN task than expert and nonfatigued doctors. Novice doctors' long work hours are well documented in the literature, as are their high rates of fatigue and sleep disturbance.^37,52 Novices and experts reported similar levels of fatigue in this study, indicating that junior doctors were no more fatigued than the senior doctors. These results also suggest that doctors' inexperience with BBN rather than competence in the task was related to high autonomic arousal during the BBN task.

On a theoretical level, these results are consistent with Ptacek and Eberhardt's²⁶ predictions using the TMSC.²⁷ Their model posits that doctors' discomfort during the BBN task should vary as a function of their BBN experience, their perceptions of the severity of the news, and their perceptions that they were partly responsible for the news. As expected, we found that inexperience with BBN and breaking bad—rather than good—news were related to high stress responses in doctors, although the last factor (i.e., perceived responsibility) was not assessed in this study. Additionally, we found that doctors' experience of fatigue was also reflected in high stress levels during the BBN task.

Finally, poor communication performance during the BBN task was found to be related to two doctor factors, high burnout (i.e., depersonalization) and higher fatigue level, accounting for slightly more than one third of the variance in communication performance. These results suggest that fatigue and burnout may impact doctors' communication performance in the BBN task. Such an interpretation is consistent with literature reports that doctors typically work long hours and experience high levels of burnout and fatigue^{35–37,52,53} and that doctors affected by burnout and fatigue perform more poorly in the workplace than their nonaffected peers.^38,39

However, in this study, we did not find poor communication performance to be related to doctors' inexperience, perceived stress, or psychological distress. Moreover, novices and experts did not differ on perceived stress, psychological distress, or burnout measures. Similar findings are reported by van Dulmen and colleagues,²⁴ who found that baseline stress levels were not related to BBN performance in medical students during simulated consultations.

The results of this preliminary investigation should be interpreted with caution, given several obvious study limitations. The sample size was small and the multiple analyses likely led to an increase in Type I error rate. The CRS is a new scale that is subject to continuing validity and reliability studies to be reported separately. The similar stress levels between novices and experts and the result that stress seems unrelated to experience of BBN could also be related to the small sample size. In addition, these findings might not be generalizable to real patient encounters because we obtained the data using a simulated-patient approach.

Conclusions

BBN is a psychologically and physiologically stressful experience for doctors, even during simulated encounters. In this study, inexperienced and fatigued doctors were most likely to show high autonomic arousal during the BBN task. However, high autonomic arousal was not related to doctors' poor communication performance, although poor communication performance was related to symptoms of burnout and fatigue. Taken together, these results suggest that inexperience, coupled with burnout and fatigue, may have contributed to high autonomic arousal and poor clinical performance in some doctors. Poor performers of these communication tasks are known to be at an increased risk of future medical litigation.⁵⁴

In 1961, Donald Oken⁵⁵ identified the potential trauma of the first experience of BBN for young doctors. Given the lack of empirical research in BBN, it is hardly surprising that that potential has not diminished 40 years later. In 2001, three out of four U.S. medical trainees reported they first delivered bad news when they were a student or intern.⁵⁶ Most knew the patient for just hours or days, few engaged in any planning, and a senior doctor was rarely present. These young doctors rated the importance of skills in delivering bad news highly; they believed such skills can be improved and thought that more guidance should be offered to them in preparation for such activity. Therefore, more thoroughly evaluating relationships between stress, burnout, fatigue, and poor clinical performance in doctors is important. Such information may better inform the development of targeted teaching interventions that could improve performance in these important clinical interactions.

Acknowledgments

This work was supported by a grant from the Lola Douglas Fund, 2004–2005. The authors wish to thank all the doctors who generously participated in this study.