Article Text
PDF
Abstract
Purpose Attributes of the organisational culture of residency training programmes may impact patient safety. Training environments are complex, composed of clinical teams, residency programmes, and clinical units. We examined the relationship between residents' perceptions of their training environment and disclosure of or apology for their worst error.
Method Anonymous, self-administered surveys were distributed to Medicine and Surgery residents at Boston Medical Center in 2005. Surveys asked residents to describe their worst medical error, and to answer selected questions from validated surveys measuring elements of working environments that promote learning from error. Subscales measured the microenvironments of the clinical team, residency programme, and clinical unit. Univariate and bivariate statistical analyses examined relationships between trainee characteristics, their perceived learning environment(s), and their responses to the error.
Results Out of 109 surveys distributed to residents, 99 surveys were returned (91% overall response rate), two incomplete surveys were excluded, leaving 97: 61% internal medicine, 39% surgery, 59% male residents. While 31% reported apologising for the situation associated with the error, only 17% reported disclosing the error to patients and/or family. More male residents disclosed the error than female residents (p=0.04). Surgery residents scored higher on the subscales of safety culture pertaining to the residency programme (p=0.02) and managerial commitment to safety (p=0.05). Our Medical Culture Summary score was positively associated with disclosure (p=0.04) and apology (p=0.05).
Conclusion Factors in the learning environments of residents are associated with responses to medical errors. Organisational safety culture can be measured, and used to evaluate environmental attributes of clinical training that are associated with disclosure of, and apology for, medical error.
- Medical error
- medical education
- organisational behaviour
- programme evaluation
- survey research
- Women's health
- Human error
Statistics from Altmetric.com
- Medical error
- medical education
- organisational behaviour
- programme evaluation
- survey research
- Women's health
- Human error
Introduction
Everyone makes mistakes. Over the past decade, the medical profession has started to apply a systems approach to patient safety, recognition that coordination of individual, team, and organisational forces are needed to promote patient safety. Analysis of the root causes of an error can prevent future errors by identifying and correcting problems.1 However, in order to learn from mistakes and develop safer systems, errors must first be identified and reported.
Unfortunately, many errors are never reported. In one study, merely half of the house officers told their attending physicians about the most serious errors they committed.2 Underreporting of adverse events is estimated to range from 50% to 96% annually.1 3 4 Rather than dealing with mistakes constructively by reporting and learning from them, studies indicate that physicians typically respond to their mistakes defensively, blaming the system, other members of the healthcare team, or even the patient.2 5–7 Possible explanations for underreporting medical errors include fear of litigation acting as a deterrent,8–10 and the professional medical culture that limit an individual's willingness to discuss error.5 11
While elements of professional medical culture are hypothesised to lead to widespread underreporting of medical errors, few studies have elucidated and measured aspects of medical culture that are associated with a failure to disclose, particularly in the learning environments of clinical training programmes. In contrast to medical culture, non-medical industries such as aviation and nuclear safety have traditionally valued a professional ‘culture of safety’, which facilitates reporting of errors, so that individuals operating in groups within an organisation can learn how to prevent future errors.12 Medical educators have recently attempted to incorporate system-based thinking into their curriculum, in order to incorporate aspects of a safety culture that, along with enquiry and trust, were previously lacking in residency settings.2 7 12 Although there is recognition of the need to create a learning culture of safety for residents,13 measuring educational culture has proved to be a challenge. In addition to the complexities of quantitatively measuring an abstract concept of ‘organisational culture’,14 residents train in multiple environments which are dynamic and divergent: their individual clinical teams (which often rotate), the academic residency training programme, and the clinical institution(s) (usually a hospital) each contributing to a trainee's overall sense of culture.
This study endeavours to measure a trainee's perception of their training environment as it relates to safety, and their response to committing an error. By collecting data on both these elements, we explore the relationship between the culture of the training environment and individual behaviour. Specifically, we hypothesised that house officers would be more likely to disclose and apologise for an error if they rate their clinical team as having an environment in which they can report errors without fear of punishment or rejection; rate their training programme as having positive attitudes about reporting and coping with errors in the workplace; and rate the hospital as having a high level of commitment to patient safety. Secondary aims of our study were to evaluate the association between individual characteristics of clinical trainees and, first, disclosure of a medical error, and second, apology for the error.
Methods
Study design and survey administration
Anonymous, self-administered surveys were distributed to medicine and surgery residents at Boston Medical Center during educational conferences and department meetings in 2005. The distribution and retrieval procedures of the surveys ensured privacy and anonymity of the residents. The researchers (who were also attending physicians) were blinded as to which residents completed the survey, which ensured no conflict of interest for the researchers if asked to evaluate residents. The residents were assured that their privacy and anonymity would be protected, and that the researchers would remain blinded to their participation status. Specifically, the researchers approached a group of housestaff during an educational conference or meeting, explained the purpose of the survey, and then left the room and building. Residents who choose to participate completed the surveys and returned them in sealed envelopes to a box in the room. Those who chose not to participate returned blank surveys in sealed envelopes. At the end of each conference, a research assistant returned to the room to collect the box with the sealed envelopes. All participants received a $10 honorarium, whether or not they completed the survey. The survey, database, and protocol were de-identified. To further protect participants in the event of an accidental breach of anonymity, a certificate of confidentiality was obtained from the National Institutes of Health. The project was carefully reviewed and approved by the institutional review board.
Survey content
Our survey focused on three levels of environment that had the most face validity of microculture constructs within a resident's learning environment. Questions were selected from three validated surveys of organisational culture, adapted for this study to focus on the organisational environment of housestaff. Since the full survey instruments were deemed too high a respondent burden, the authors carefully considered and then selected items from each survey most relevant to the study. The microenvironment of the immediate clinical team was examined with five of seven questions from the Team Psychological Safety Survey, which assesses the belief that well intentioned actions will not lead to punishment or rejection by the team (see online appendix).15 The macroenvironment of the residency programme was assessed with 10 of the 37 items from the Error Orientation Questionnaire; the items selected assess attitudes to errors and approaches to coping with errors in the workplace (see online appendix).16 Perception of hospital management's commitment to patient safety on the clinical unit was assessed with four of 19 items from the Patient Safety Survey (see online appendix).17 Responses were coded with six-point Likert scales, and summed to derive a total score for each survey. For all three scales (Team Psychological Safety, Residency Programme Error Orientation, and Managerial Safety Commitment), higher scores correlated with more positive aspects of culture. In relation to error, participants were asked to recall the circumstances of and share details regarding their most significant medical or surgical error using open-ended text. In a multiple choice format, participants were specifically asked about the following: consequences for the patient; consequences for the resident; if and to whom they disclosed the error; if they had apologised for the error; and perceived causes for the error. Responses to each question were constructed from the results of a previous survey of residents regarding medical error, conducted by Wu et al.2 In addition, residents were asked to characterise their own level of distress from the error using a 10-point Likert scale.
Analysis
Univariate analysis was used to describe demographics, residency type, reporting rates (to colleagues and friends), apology and disclosure rates (to patients), emotional responses of residents, types of mistakes, and consequences to and responses of both the residents and patients. Errors were classified from the written responses into one of the following categories: procedural, medical management, laboratory test follow-up, delayed diagnosis, or other/not classifiable. We used χ2 to evaluate differences between categorical variables and Wilcoxon rank sum methods for the three organisational culture scales. In order to compare the subscales with each other, the raw score was converted to a scaled score, by dividing each raw score by the maximum possible score of each subscale, and multiplying by 33.3. We then calculated an overall Medical Culture Summary score, by summing the three scaled subscores, that is, each subscale contributes one-third of the overall Medical Culture Summary score.
Results
Surveys were distributed to 109 residents and 99 surveys were returned, making an overall response rate of 91%. Two residents' surveys were excluded because they reported no mistake, leaving a final population of 97 residents, 59 (61%) from internal medicine residents and 38 (39%) from surgical residents. There were 57 (59%) male residents, of which 33 (58%) were internal medicine residents and 24 (42%) were surgical residents. Two surgical residents did not report their gender, and were excluded from analyses which included gender. The most significant medical or surgical error that was the focus of residents' responses typically occurred in an inpatient setting and during the first year of training (table 1). Seventy-five per cent of the residents were extremely distressed by their mistake. While 41 (42%) did not provide an adequate description of their error to be classified, 26 (27%) were classified as medication related, 12 (12%) as procedural, 11 (11%) due to delayed diagnosis, and 9 (9%) due to inadequate follow-up to a laboratory test.
Characteristics of residents and setting where error occurred
Although 20 (21%) of the involved patients had no reported consequences resulting from the errors, common consequences included delayed treatment for 23 (24%), delayed diagnosis for 22 (23%), prolonged hospital stay for 17 (18%), medical complications for 13 (13%), and death for 13 (13%) patients. The errors resulting in patient death were largely errors involving anticoagulants, potassium balance (either not checking blood work or inadequate management of blood potassium level), or insulin. There were no consequences for 60 (62%) of the residents due to the error, but 30 (30%) reported some form of reprimand, 16 (16%) presented the case at a morbidity and mortality conference (which was reported as a consequence), 6 (6%) reported their work and family life was affected, and 1 (1%) was named in a law suit. The most common attributions for the error reported by residents included being too busy (32, 33%) and inexperience (31, 32%). Many residents also attributed their error to having inadequate knowledge, hesitating before acting, or being too tired (table 2). While 30 (31%) reported apologising for the situation associated with the error, only 17 (18%) reported disclosing the error to patients and/or their family. Five residents both disclosed (29% of those who disclosed) and apologised (17% of those who apologised table 3).
Characteristics of worst medical errors*
Individual factors associated with disclosure and apology, by gender* and programme
Correlates of disclosure and apology
The disclosure rate was higher among surgery residents (24%) than internal medicine residents (14%), but this difference was not statistically significant (p=0.2). Of the residents who disclosed their error, 32 (33%) reported that it was unsupervised. Three (3%) residents reported being told by their attending not to discuss the error with the patient. Female internal medicine residents were significantly less likely to disclose their worst medical error to patients or their families than their male counterparts (p=0.03). In contrast, more female surgery residents, 7 (58%), apologised for their error compared with male surgery residents 7 (29%), though the difference did not reach statistical significance (p=0.1). Of the 13 errors that resulted in a patient's death, only 3 (23%) of the residents disclosed the error to the patient and/or the patient's family, but 6 (46%) residents apologised to the patient's family. More residents who made errors in medication management (8 of 26, 31%) disclosed their error than those who made errors with procedures (3 of 12, 25%) or delayed diagnosis (2 of 11, 18%) (p=0.05).
Correlates of organisational culture measures
The Safety Culture Summary score was positively associated with disclosure of medical error to the patient and/or patient's family (p=0.04) and apology for the error (p=0.05). There was a trend of association between disclosure and higher scores on the subscales clinical Team Psychological Safety (p=0.07) and Residency Programme Error Orientation scales (p=0.07), but not for Managerial Safety Commitment (p=0.2). Report of apology to the patient and/or patient's family was not associated with the clinical Team Psychological Safety score (p=1.0) but was positively associated with scores on the Residency Programme Error Orientation (p=0.05) and Managerial Safety Commitment (p=0.01). There were no significant gender differences in scores for each of the subscales as well as the summary measure of safety culture. Surgery residents had higher scores on the Residency Programme Error Orientation (p=0.02) and Managerial Safety Commitment scales (p=0.05) compared with medicine residents, but there was no significant difference between programmes in the Safety Culture Summary score (table 4).
Safety environment factors associated with disclosure and apology, by gender* and programme
Conclusions
Only 17% of the residents we surveyed reported disclosing their most significant error to their patient and/or patient's family, and only 31% of the residents reported apologising for their most significant error. Our results suggest that factors in the learning environments of the clinical team and residency programme are associated with error disclosure and apology among residents. Individual factors, such as gender and type of error, also appear to be associated with error disclosure and apology, and more residents apologised for the error than disclosed it.
Our findings of discordance between apology and disclosure of medical error are consistent with previous studies exhibiting residents may be more willing to apologise for a bad outcome than to reveal that they played a role in causing the bad outcome, resulting in a ‘partial disclosure’.18 19 Collectively, these findings imply that factors that facilitate apologising for an error may differ from influences that facilitate disclosing an error. These findings are reflected in State laws that distinguish different components of conversations with patients about unanticipated outcomes: ‘expression of sympathy’ (apology), ‘explanation’ (disclosure), and ‘admission of fault’, which does not cleanly translate into either category.20 Additional explanations for the discordance may include the social context in which the error occurred. For example, apologising for a systemic error that occurred would likely be easier than disclosing personal responsibility for an error, which could have greater legal and professional consequences.19
The relationships of gender to our outcome measures are complex. More male residents disclosed error (driven mostly by male internal medicine residents) while more women apologised (driven mostly by female surgical residents). With our small sample size, definitive conclusions about the interactions among gender, specialty and disclosure are difficult to ascertain from our data. Previous studies2 9 have also demonstrated that individual attributes, such as gender and emotional response to the error, influence the reporting rate of the error. However, in contrast to our results, women in a previous study were more likely to discuss their errors with their patients and make constructive changes in their practice.2 Although the female residents in our study were less likely to disclose their error, the female surgical residents were more likely to apologise, consistent with past reports of greater empathy among female physicians.21 Our results suggest that there are barriers to disclosure in the learning environments of clinical trainees that affect men and women differently. Further research will be needed to elucidate which barriers to disclosure and apology affect genders differentially. For example, possible barriers to disclosure may be attitudinal— women may feel they have more to lose than men by disclosing in order to be professionally successful, or emotional— women may feel more of a sense of helplessness and loss of control once information is disclosed.
We found surgery residents to have higher scores on the residency programme's Error Orientation Scale and the clinical unit's Managerial Safety Commitment Scale than medicine residents, but not on the Safety Culture Summary score. This is consistent with a previous survey of residents, which found that presentations of errors causing adverse events occurred 18% of the time in internal medicine ground rounds compared with 42% in surgery.22 The differences between these two fields are likely due to divergent regulatory and cultural factors. Historically, morbidity and mortality rounds have served as a forum where surgeons learn from poor outcomes and aspire to identify their errors,23 but this tradition is weaker in medicine training programmes.24 The Accreditation Council for Graduate Medical Education (ACGME) requires that surgery morbidity and mortality conferences present and discuss ‘all deaths and complications that occur on a weekly basis’. Historically, there has been no similar requirement for internal medicine.24 Without a specific requirement to do so, adverse events and errors occurring in the medicine service may not be generally discussed.21
Several medicine residencies have developed programmes to address the current ACGME competency on Systems Based Practice,13 by teaching systems-based thinking using root cause analysis of medical errors,25 which require residents to develop an awareness of working in multidisciplinary teams to enhance patient safety, and participate in identifying system errors and implementing potential solutions.13 Several studies have demonstrated the benefits of such educational interventions,26–29 although to our knowledge, no study has attempted to measure changes in learning environment as a result of such interventions. Although such programmes are helpful, current ACGME guidelines do not require training programmes to address a thorough behavioural process of managing medical mistakes: accepting responsibility; discussing with colleagues; disclosing and apologising to patient; conducting an error analysis; and making changes in a practice setting designed to reduce future errors.1 2 5 6 11 30
The ability of residents to cope with medical error may be dependent on reassurance and learning opportunities provided by medical colleagues and supervisors.26 Our findings would support this assertion, given the positive association between our derived Medical Culture Summary score and reporting of disclosure and apology. However the low frequency of disclosure and apology suggests that more work needs to be done within our training programmes to mitigate the negative effects of error to individuals, and gain potential benefits from more thorough processing of errors for individuals and the healthcare system.
There are several issues germane to housestaff and errors that are not addressed by our study. Some believe that an effective apology includes offering some form of reparation for the mistake.31 We did not examine the issue of reparation. While the literature suggests that resident physicians who accept responsibility for their errors and discuss them are more likely to report other improvements in their medical practice,2 we did not examine this phenomenon. Furthermore, while disclosure in a timely and appropriate manner may influence a patient's decision to pursue legal action,1–3 we did not explore the relationship between our findings and legal action. In addition, though a doctor's emotional reaction to an error can last for years32 and negative emotional responses are associated with increased odds of future self-perceived errors,33 we did not evaluate emotional reactions or predict the future likelihood of error. Lastly, we did not directly examine the extent to which subjects were trained regarding coping with medical error, and hence could not determine if this training influenced their behaviour.
This study has several limitations. First, residents at only one academic medical centre were surveyed, so the results may not be generalisable. In addition, the residents were surveyed during 2004–2005, so it is possible that these results may no longer be accurate. During the past 6 years, the ACGME training requirements have increased their focus on systems-level thinking and training programmes have increasingly focused on reduction of error. However, the authors feel that the key findings of this study are relevant today. The rate of safety culture change is relatively slow, as demonstrated by a recent hospital survey administered by the Agency for Healthcare Research and Quality: average composite change in safety culture to change 1% over 1–2 years.20 An increased focus on reduction in error does not translate directly to increases in individual accountability, apology or disclosure of error. Second, the modest sample size, limits opportunity for multivariable analyses as well as statistical power to detect potential associations. Third, our survey directed residents to consider a single error. We did this to focus respondents' attention on the details of an event that they would remember clearly to gain insight into aspects of organisational culture. This specific error may not be representative of most errors. In addition, as most of these errors occurred during the residents' first year of training, the expectation for disclosure and apology may be different than for the other years of training. However, a prior study that included trainees at our institution suggested that the responsibility of delivering bad news often falls to junior members of the team, including first-year trainees and medical students.34 Fourth, the scales of organisational culture we used have rarely been used in healthcare settings. As such, the clinical significance of our observed score differences are unclear.
Despite the limitations, we successfully adapted survey tools previously used in a business environment to measure aspects of the learning environment of clinical trainees which are associated with disclosure and apology for medical error. This instrument needs to be validated in other institutions before proving its value as a metric in residency programme safety culture. If validated, such an instrument could be a valuable tool to assess changes in learning environments. Measuring culture and providing such feedback to leadership and staff is one of the safe practices recommended by the National Quality Forum to promote patient safety and reduce medical error.35
Measuring culture change requires a multimodal approach, of which this instrument could make a valuable contribution.36 Since the ability to measure medical culture, and changes to it, is immature,37 our study provides baseline measurements to help move the field further along. Developing learning environment metrics will be valuable to other institutions and training programmes in the coming years, as incremental programmatic changes in systems-level thinking and disclosure of medical error continue to impact the learning environments of residents.
Our results suggest a need for training programmes to provide trainees with structured, meaningful ways to cope with errors to prevent negative emotional responses, as well as create learning environments that facilitate disclosure of errors. Attention may need to be paid to explicate potential gender-related differences. All this is particularly important if, as a profession, we are to instil proper values, attitudes and responses to the inevitable occurrence of error in the next generation of physicians. As residency programmes incorporate systems-level thinking into residency education for patient safety and error prevention, it is important not to neglect the humanistic and interpersonal consequences of error for providers and patients. In order to do so, we need to develop measurement tools for learning environments. Further research is needed to identify successful environmental attributes that promote disclosure and healthy processing of medical errors.
Acknowledgments
The authors wish to thank Victoria Brower, MPH, for creating and maintaining the study database, and Ignacio De La Cruz for help with manuscript preparation.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Download Supplementary Data (PDF) - Manuscript file of format pdf
Footnotes
Funding This work was supported by an American Cancer Society Physician Training Award in Preventive Medicine (PT APM 97 185 04, Jane Liebshutz, Principal Investigator) and a Boston University Medical Center Risk Management Grant (Andrea Kronman, Principal Investigator).
Competing interests None.
Ethics approval Boston University Institutional Review Board.
Provenance and peer review Not commissioned; externally peer reviewed.