|Home | About | Journals | Submit | Contact Us | Français|
Disclosing harmful errors to patients is recommended, but appears to be uncommon. Understanding how trainees disclose errors and how those practices evolve during training could help educators design programs to address this gap.
To determine how trainees would disclose medical errors.
A survey of 758 trainees (488 students and 270 residents) in internal medicine at two academic medical centers. Surveys depicted one of two harmful error scenarios that varied by how apparent the error would be to the patient. We measured attitudes and disclosure content using scripted responses.
Trainees reported their intent to disclose the error as “definitely” (43%) “probably” (47%) “only if asked by patient” (9%), and “definitely not” (1%). Trainees were more likely to disclose obvious errors in comparison with ones patients were unlikely to recognize (55% vs. 30%, P<0.01). Respondents varied widely in what information they would disclose. Fifty percent of trainees chose statements explicitly stating an error occurred rather than only an adverse event. Regarding apologies, trainees were split between a general expression of regret (52%) and an explicit apology (46%). Respondents at higher levels of training were less likely to use explicit apologies (Trend P<0.01). Prior disclosure training was associated with increased willingness to disclose errors (OR 1.40, P=0.03).
Trainees may not be prepared to disclose medical errors to patients, and worrisome trends in trainee apology practices were observed across levels of training. Medical educators should intensify efforts to enhance trainees’ skills at meeting patients’ expectations for open disclosure of harmful medical errors.
Academic medical centers face the task of preparing the next generation of physicians to meet patient expectations for error disclosure. Patients uniformly express the desire for health care providers to promptly disclose and apologize for errors in their care[1–4]. Yet, surveys suggest a minority of harmful errors are disclosed to patients and disclosure conversations often fail to meet patient expectations[5–8]. At academic medical centers, trainees are frequently involved in medical errors, but may not know whether or how to disclose the mistake to patients [9–11]. Although trainees want to be open with patients, discussing errors with patients presents substantial challenges for students and residents[9, 12, 13].
Preparing trainees for disclosure is particularly difficult because these delicate conversations require advanced communication skills, and because trainees and patients suffer considerable emotional distress after mistakes. Despite the importance of this topic, only a minority of trainees receive training in error disclosure. Furthermore, few trainees receive feedback about their disclosure skills or know where to seek help after an error[15, 16]. It is unknown if the lack of an organized curriculum to impart error disclosure skills is partly responsible for the fact that current disclosure practices often fail to meet patient and regulatory expectations. In the absence of formal curricula, trainees may otherwise learn disclosure skills through the hidden curriculum and direct observation of senior clinicians[17, 18]. This lack of formal training may lead trainees to struggle with disclosure conversations when they enter practice.
Programs to enhance the disclosure of errors to patients are growing. The Joint Commission’s accreditation standards require healthcare institutions to ensure patients are informed about unanticipated outcomes in their care. Although prior work has examined whether trainees are likely to disclose errors[9, 11], little is known about how they would do so. Understanding what information trainees might communicate in disclosure conversations could help educators identify gaps between the disclosure content patients desire and the information trainees report they would share. Furthermore, a better understanding of how disclosure practices evolve during medical training could help target curricula to different stages of training. Therefore, we undertook a cross-sectional survey at two centers to describe how medical students and residents at different levels of training would disclose hypothetical medical errors to patients.
Between June and October of 2003, we administered questionnaires to 999 U.S. trainees, including 629 medical students (320 second year (MSII), and 309 fourth year (MSIV)), 159 interns (PGY1), and 211 residents (PGY 2–3) in internal medicine from two academic institutions: Washington University School of Medicine/BJC HealthCare (St. Louis, MO), and the University of Washington (Seattle, WA). These institutions differ in that one is private and one is public. Both are located in two of the 18 states whose malpractice climate was reported to be “in crisis” at the time of the survey due to limited availability of affordable malpractice insurance. At the time the anonymous survey was conducted, neither institution required error disclosure training for students or residents. The Institutional Review Boards at both institutions approved this study. The questionnaires were distributed to trainees at orientation sessions, conferences, and by email. Respondents could complete a paper or web-based questionnaire. The results of the same survey of attending physicians in surgery, medicine, and pediatrics have been previously reported[7, 21, 22]. A distinct subset of survey responses regarding trainees’ attitudes and experiences with error have been previously reported and were referenced for statistical analyses, but are not duplicated here.
We utilized the Institute of Medicine’s definitions of “adverse event,” “medical error,” and “near miss.” We developed and pilot tested our own definitions of “serious error” and “minor error.” We defined a serious error as an “Error that causes permanent injury or transient but potentially life threatening harm” and a minor error as an “Error that causes harm which is neither permanent nor potentially life threatening.” Survey definitions were repeated on every page of the questionnaire and key terms were capitalized or bolded throughout.
To explore how differences in errors might affect disclosure, we created two scenarios depicting serious errors for which widespread consensus supports disclosure to patients (Appendix A). The errors were designed to be comparable in severity, but to vary by how apparent the error would likely be to the patient. Respondents randomly received one of the two scenarios. The more apparent error was an insulin overdose due to the physician’s handwritten order for “10 U” being misinterpreted as “100 U,” resulting in severe hypoglycemia. The less apparent error was a hyperkalemic dysrhythmia due to failure to check the results of a potassium level ordered after starting a medicine known to cause hyperkalemia. The patient would likely be unaware of this error unless the physician disclosed it.
For each scenario, respondents received parallel questions asking: (1) how serious the error was, (2) how responsible the respondent was for the error, (3) how upset the respondent would be, (4) how likely the respondent was to be sued, and (5) how likely the respondent would be to disclose this error to the patient. Five questions measured what information physicians would volunteer to the patient about the error: (1) “What would you most likely say about what happened?” (2) “How much detail would you most likely give the patient about the error?” (3) “What most closely resembles what you would most likely say about the cause of the error?” (4) “What would you most likely say regarding an apology?” (5) “What would you most likely say about how the error would be prevented in the future?” For each question, three response scripts represented increasing amounts of information (no disclosure, partial disclosure, and full disclosure). The text of these responses was published previously. Several rounds of pilot testing, including cognitive interviews with practicing physicians, were conducted to ensure that the survey questions were clear, the scenarios were realistic, and the disclosure responses were plausible. Data from the disclosure scenario responses were combined for initial analysis and then analyzed separately.
Descriptive statistics included means and standard deviations for continuous variables and percentages for categorical variables. Questions that used 4-point Likert response scales were dichotomized at the midpoint (agree vs. disagree) or grouped by responses of comparable quality (e.g., responses of “very likely” and “extremely likely” were combined). Categorical variables were compared using Pearson chi-square, chi-square for linear trend, Fisher’s exact tests, and odds ratios as appropriate. All tests were two-tailed and a P-value less than .05 was considered significant. Because physician sense of responsibility for the error, worry about loss of patient trust, and training to perform error disclosure may affect the quality of disclosure[7, 11], we performed bivariate analysis to examine their associations with four disclosure responses of importance to patients: intention to definitely disclose the error, explicitly mentioning that an “error” occurred, specifically describing the cause of the error, and providing a full apology. Because fear of litigation may exacerbate trainees’ emotional responses to error, we performed bivariate analysis to determine if worry about lawsuits was associated with variation in responses to questions regarding feeling responsible, upset or worried about one’s reputation. Comparisons were limited to plausible associations. Analyses were performed using SAS Version 9.1.3 (SAS Institute, Cary, NC).
Surveys were completed by 758 (280 MSII, 208 MSIV, 151 Interns, 119 Residents) of the 999 eligible trainees (76%, Table 1). Of respondents, 332 (44%) were in training at Washington University and 426 (56%) were from the University of Washington. Response rates did not vary significantly by scenario or by institution. Not all respondents completed every question, so denominators may be lower for certain questions. The highest non-response rate for any question in the survey was 1%.
Most trainees (85%) agreed their scenario represented a serious error. Recognition that the scenario represented a serious error increased with level of training (Trend P<0.01, Figure 1). Approximately half (49%) felt a lawsuit would be somewhat or very likely as a result of the error. Concern about the likelihood of legal action declined with higher level of training (Trend P<0.01, Figure 1). Trainees reported their intent to disclose the error as “definitely” (43%) “probably” (47%) “only if asked by patient” (9%), and “definitely not” (1%). Reporting a definite intent to disclose the error rose with level of training (Trend P=0.03, Figure 1).
A majority of trainees (78%) felt that as the doctor they were very or extremely responsible for the error in their scenario. Almost all trainees (96%) reported they would be very or extremely upset about the error, and 42% were very or extremely concerned that their reputation would be damaged. Trainees who felt that the error would trigger a lawsuit were more likely to report the following emotional responses: Feeling responsible: OR 1.83, 95% CI 1.28–2.61; Feeling upset: OR 2.18, 95% CI 1.02–4.66; Worried about reputation: OR 3.15, 95% CI 2.33–4.25.
Respondents varied widely in what information they would disclose to patients. Half (50%, Table 2) of trainees chose statements explicitly stating an error occurred, whereas 48% mentioned the adverse event but not the error. When asked how they would describe the cause of the error, 40% chose a specific description of exactly what the error was, 45% offered non-specific information, and 14% would not volunteer any explanation unless asked. Among the 319 trainees who indicated they would definitely disclose the error, many limited disclosure content (32% described the error as an adverse event, and 52% gave partial or no disclosure of the cause of the error). Nearly all trainees (98%) would offer some form of apology, but trainees were split between a general expression of regret (52%) and an explicit apology (46%). More experienced trainees were less likely to provide an explicit apology (Figure 1, Trend P<0.01). In discussing how to prevent recurrent error, 57% would offer specific steps, 37% chose a general pledge to prevent recurrences, and 6% would not volunteer any information. Senior residents were significantly less likely than all other trainees to offer detailed prevention plans (54/118; 46% vs. 377/634; 59%, P<0.01).
Fewer trainees reported they would definitely disclose the less apparent (hyperkalemia) error scenario versus the more apparent (insulin overdose) scenario (30% vs. 55%, P<0.01). In terms of disclosure content, trainees were less likely to explicitly state an error occurred after the unapparent error (42% vs. 58%, P<0.01). Apology content did not vary between scenarios, but trainees were more likely to communicate specific steps to prevent error recurrences in response to the less apparent error (72% vs. 42%, P<0.01).
Those who had previously received training in error disclosure were more likely to express a definite intent to disclose the error, to provide a specific explanation of the error, and to describe in detail the cause of the error (Table 3). Similarly, trainees who felt responsible for the error in their scenario were more likely to specifically describe the error, to describe its cause in detail, and to provide an explicit apology (Table 3). Respondents who felt that disclosure would decrease patient trust were less likely report a definite intent to disclose the error (OR 0.71, 95% CI 0.53–0.94).
The circumspect disclosure content chosen by the 758 trainees in our study falls short of current disclosure guidelines and would likely not meet patient expectations[2, 3]. Of particular concern, more senior trainees were less likely than their junior colleagues to offer explicit apologies to patients or to address specific steps to prevent error recurrence. Possible explanations for these worrisome trends could include burnout, loss of faith in quality improvement systems, or an overall decline in senior trainees’ empathic attitudes toward patients . Efforts to involve trainees in quality improvement programs, cross-link disclosure training with other teaching sessions on empathic communication, and emphasize the importance of a sincere and explicit apology in disclosure training may help mitigate these trends.
The information trainees would disclose largely mirrored that of attending physicians in our prior work [7, 21]. For example, a majority of both groups did not describe the event as an error, did not offer a full apology, and did not provide specific plans to prevent future recurrences. Both groups also struggled with disclosure of less apparent errors. The similarity between trainee and attending responses suggests that trainees need role models who are more adequately prepared for full disclosure. Curricula that targets both trainees and faculty disclosure skills may be important to develop such role models. Morbidity and Mortality conferences is one example of a natural venue to focus on best practices in error disclosure in an audience of mixed-experience learners.
What can educators do to improve trainee approaches to disclosure? The first step is to start early. We noted an overall increase in willingness to disclose errors between the 2nd and 4th year of medical school. This suggests that educators have an important window of opportunity to inculcate positive disclosure attitudes early in the medical education process[15, 25].
Second, we found that trainees were less likely to disclose an error when they felt it would result in a loss of the patient’s trust. Baseline trust between patients and trainees may be especially fragile due to both parties’ awareness of the trainees’ incomplete expertise. Faculty should openly acknowledge this concern and facilitate the process of rebuilding trust after an error that involves a trainee. Attendings leading disclosure conversations with patients should consider ways to openly support trainees as competent doctors (when appropriate) and facilitate mending the relationship between the trainee and his/her patient.
Third, trainees uniformly anticipated emotional distress after a serious error, highlighting an opportunity for faculty to provide emotional support. Educators can normalize fallibility by sharing their own personal experiences with errors. They should also address trainees’ fears about litigation, which heightens emotional distress. Although well-intentioned faculty might exclude trainees from disclosure conversations to streamline communication or to protect them from distress, this approach hinders opportunities to model error disclosure techniques. Simulated disclosures with actors or video may help prepare trainees for difficult conversations in a less emotionally charged environment.
Finally, we found a positive association between trainees’ perceived responsibility for the error and the likelihood they would offer a more comprehensive disclosure and apology. Although educators should encourage trainees to accept responsibility for errors, guidance is necessary. Policies that require trainees to immediately report errors to their supervising physician are prudent. Transparent clerkship evaluation policies and non-punitive institutional error reporting systems could increase trainees’ willingness to speak up about error. Faculty input in explicitly defining the trainee’s role is critical, especially until standards for trainee responsibilities in disclosure are developed.
While this study represents the largest and most comprehensive study to date describing how trainees would disclose medical errors, it has several limitations. The data came from only two academic centers, which may limit generalizability. The study was cross-sectional rather than longitudinal, limiting our ability to draw conclusions about the effect of training on attitudes over time. Although the response rate was robust, non-response bias may have affected the results. The scenarios were hypothetical, and therefore may not reflect actual behaviors. Responses are subject to social desirability bias. Lastly, the data were collected in 2003, and trainee attitudes and behaviors may have changed during the period between data collection and analysis. However, recent studies continue to document an ongoing gap in trainee disclosures [9, 15]. Therefore, we believe the data can still inform educators who are developing disclosure curricula.
This work highlights several areas for future research. Large scale standardized evaluation of existing disclosure training models could help in the design, implementation, and evaluation of best practices for disclosure training. A better understanding of the intensity and frequency of intervention required to make lasting improvements in trainees’ disclosure skills is also needed. Assessment of the quality of actual disclosures, including direct feedback from patients, is another important next step. A closer examination of effects of disclosure involvement on trainee emotional distress would also be informative. Lastly, studies that further explore the effects of the hidden curriculum on the dynamic between learners and supervising clinicians as related to error disclosure could help prepare clinician-educators to better model ideal disclosure practices.
The current training environment may not encourage an approach to error disclosure that is consistent with patient expectations and national guidelines. Errors represent powerful teaching opportunities, and are an ideal time for faculty to role model effective disclosure techniques. Educators should consider developing disclosure curricula that highlight the content desired by patients and the specific support needed by trainees. Ensuring trainees are skilled at error disclosure before they enter practice should be a core goal across medical education.
We would like to acknowledge Carolyn Prouty for her assistance in preparing this manuscript and Eric Van Eaton for assistance in survey administration.
Funding: Agency for Healthcare Research and Quality (grants # 1U18HS1189801, 1K08HS01401201), the Greenwall Foundation Faculty Scholars Program, and the National Center for Research Resources (grant # UL1 RR024992). The funding sources had no role in the design, conduct, or reporting of the study or in the decision to submit the manuscript for publication.
|Type of Scenario||Description|
(More Apparent Error)
|You have admitted a diabetic patient to the hospital for a COPD exacerbation. You handwrite an order for the patient to receive “10 U” of insulin. The “U” in your order looks like a 0. The following morning, the patient is given 100 U of insulin, 10 times the patient’s normal dose, and is later found unresponsive, with a serum glucose level of 35 mg/dl (1.94 mmol/L). The patient is resuscitated and transferred to the intensive care unit. You expect the patient to make a full recovery.|
(Less Apparent Error)
|You administer a new medicine with a common adverse effect of increasing the potassium level to an outpatient with hypertension. The patient’s baseline potassium level is normal (4.0 mEq/L). You order a repeat blood test to measure the potassium level, to be drawn the next week, but forget to check the laboratory results. Two weeks after the patient begins taking this new medicine, the patient starts feeling palpitations and goes to the emergency department. In the emergency department, the patient experiences an episode of ventricular tachycardia, requiring cardioversion. The patient’s potassium level at this event is 7.5 mEq/L. The patient is hospitalized for 4 days, and makes a full recovery. The patient returns to your office for a follow-up visit. On reviewing the patient’s chart, you see the overlooked laboratory results, which showed the patient’s potassium level had increased substantially from 4.0 to 5.6 mEq/L. Had you seen this elevated potassium level earlier, you would have discontinued the new medicine and treated the hyperkalemia, likely avoiding the life-threatening arrhythmia.|
Contributions: "Andrew White – Data analysis, manuscript drafting, approval for publication Sigall Bell – Data analysis, manuscript drafting, approval for publication Missy Krauss – Data analysis, manuscript revision, approval for publication Jane Garbutt – Study design, data acquisition, manuscript revision, approval for publication Claimore Dunagan - Study design, data acquisition, manuscript revision, approval for publication Victoria Fraser - Study design, data acquisition, manuscript revision, approval for publication Eric Larson – Data analysis, manuscript revision, approval for publication Wendy Levinson - Study design, manuscript revision, approval for publication Tom Gallagher - Study design, data acquisition, manuscript drafting, approval for publication ".
Competing interests: "We have no competing interests to report."
Ethical approval: "The study was reviewed and approved by the Institutional Review Boards of the University of Washington and Washington University School of Medicine."
Financial disclosure: None.
Andrew A. White, Department of Medicine, University of Washington, Seattle. Washington.
Sigall K. Bell, Department of Medicine and Infectious Disease, Harvard University, Boston, MA.
Melissa J Krauss, Division of Biostatistics, Washington University School of Medicine, St. Louis, MO.
Jane Garbutt, Department of Medicine, Washington University School of Medicine, St. Louis, MO.
W. Claiborne Dunagan, Department of Medicine, Washington University School of Medicine, St. Louis, MO.
Victoria J. Fraser, Department of Medicine, Washington University School of Medicine, St. Louis, MO.
Wendy Levinson, Department of Medicine, University of Toronto, Toronto, Ontario.
Eric B. Larson, Center for Health Studies, Group Health Cooperative, Seattle, WA.
Thomas H. Gallagher, Department of Bioethics and Humanities, University of Washington, Seattle, Washington.