Objective To (1) develop and test survey items that measure error disclosure culture, (2) examine relationships among error disclosure culture, teamwork culture and safety culture and (3) establish predictive validity for survey items measuring error disclosure culture.
Method All clinical faculty from six health institutions (four medical schools, one cancer centre and one health science centre) in The University of Texas System were invited to anonymously complete an electronic survey containing questions about safety culture and error disclosure.
Results The authors found two factors to measure error disclosure culture: one factor is focused on the general culture of error disclosure and the second factor is focused on trust. Both error disclosure culture factors were unique from safety culture and teamwork culture (correlations were less than r=0.85). Also, error disclosure general culture and error disclosure trust culture predicted intent to disclose a hypothetical error to a patient (r=0.25, p<0.001 and r=0.16, p<0.001, respectively) while teamwork and safety culture did not predict such an intent (r=0.09, p=NS and r=0.12, p=NS). Those who received prior error disclosure training reported significantly higher levels of error disclosure general culture (t=3.7, p<0.05) and error disclosure trust culture (t=2.9, p<0.05).
Conclusions The authors created and validated a new measure of error disclosure culture that predicts intent to disclose an error better than other measures of healthcare culture. This measure fills an existing gap in organisational assessments by assessing transparent communication after medical error, an important aspect of culture.
- health services research
- safety culture
- social sciences
- organisational theory
- continuing education
- continuing professional development
- healthcare quality improvement
- patient safety
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
- health services research
- safety culture
- social sciences
- organisational theory
- continuing education
- continuing professional development
- healthcare quality improvement
- patient safety
The Institute of Medicine noted the importance of having a strong, positive safety culture in ‘To Err is Human’, and the Joint Commission has developed a leadership standard that requires hospital leaders to ‘evaluate the culture on a regular basis’ (Standard: LD.03.01.01). Safety culture can be defined as ‘shared perceptions among managers and staff concerning the importance of safety.’1 Over the past decade, there has been intense interest in identifying key dimensions of patient safety culture surveys most relevant to healthcare settings.2–7 Despite this focus, current safety culture assessments fail to measure an important emerging competency, error disclosure.8
Error disclosure is an important dimension of safety culture because disclosure is central to transparency. Patients want to know about errors,9–11 clinicians want to be truthful with patients about their care, accrediting bodies increasingly require communication with patients about unanticipated outcomes, and healthcare organisations are better equipped to provide safe and effective care when adverse events and errors are systematically discussed, reported and reviewed.
We define error disclosure culture as shared healthcare provider dispositions towards transparency with respect to disclosing errors to patients. Understanding disclosure culture at an institution could be important for several reasons. For example, hospital units with strong error disclosure cultures may also be more likely to report near misses and errors because of an underlying belief that it is important to communicate these events to others and a commitment to do so. Organisational learning is more likely when disclosure, reporting and quality improvement processes are integrated. Healthcare systems can incorporate lessons learnt from the near misses and errors into future policies, practices or remedies designed to prevent the recurrence of these events. Therefore, one reason to study error disclosure culture is that greater organisational learning and patient safety improvements can occur when this culture is strong. Error disclosure culture is also important to understand because patients increasingly expect open communication when things go wrong. Finally, as ‘Disclosure and Offer’ programmes (ie, programmes that couple disclosure with offers of compensation in cases of preventable adverse events) gain attention as a potential way to improve patient safety, reduce malpractice lawsuits and stem extensive litigation costs, healthcare organisations will be increasingly interested in ways to assess disclosure attitudes and performance.12 ,13
Given the importance of error disclosure but the absence of existing measures, we sought to create a scale to assess error disclosure culture. We envisioned an instrument that could be used alone or in combination with other existing safety culture tools (ie, it can be added to the Safety Attitudes Questionnaire,6 AHRQ's Hospital Survey On Patient Safety Culture2 or the Patient Safety Culture in Healthcare Organizations).4 ,5 In addition, we used the scale to measure care giver attitudes about disclosure and to examine relationships between these attitudes and safety culture.
We surveyed clinical faculty from six health institutions (four medical schools, one cancer centre and one health science centre) in The University of Texas System. IRB approval was obtained prior to initiating data collection. All clinical faculty (approximately 5000) from these institutions were sent an email with a link to an electronic survey containing questions about safety culture and error disclosure once a week for 4 weeks. Four hundred and ninety-seven faculty members completed the surveys. Participation was anonymous and no identifying information from the participants was collected. Given the exploratory nature of our work, we were not focused on achieving a high response rate.
The survey contained questions about error disclosure culture, safety culture, teamwork culture, a hypothetical scenario depicting an error14 and demographics. Box 1 contains the new error disclosure culture survey items that are the focus of this study. Three physician experts in error disclosure (TG, SB, EJT) and one psychometrician (JE) created a new survey that focused on the extent to which disclosure occurs, the perceived impact of disclosure on patient's trust of the provider, the general disclosure culture, and colleague and leadership support for disclosure. Some of the items from this survey were adapted from previous work by one of the physician experts (TG).14 The safety and teamwork culture dimensions of the Safety Attitudes Questionnaire6 were also included in this survey to examine the relationship of disclosure culture to these other scales, and appear in online appendix A. All culture survey questions were measured on a 5-point Likert-type scale, where 1=strongly disagree and 5=strongly agree. Each participant was randomly given one of two hypothetical scenarios depicting a medical error of comparable severity and was asked questions about the assigned scenario. For our analysis, we analysed one of the scenario response questions: How likely would you be to disclose this error to the patient? There were five response options, ranging from 1=‘I would definitely not disclose this error to the patient’ to 5=‘I would definitely disclose this error’.14
Error disclosure culture scales
Error disclosure general culture (Cronbach's α=0.87)
1. We routinely disclose MINOR ERRORS to patients/families in my clinical area.
2. We routinely disclose SERIOUS ERRORS to patients/families in my clinical area.
5. The culture in my clinical area makes it easy to disclose MINOR ERRORS.
6. The culture in my clinical area makes it easy to disclose SERIOUS ERRORS.
7. I am encouraged by my colleagues to disclose MINOR ERRORS to patients/families.
8. I am encouraged by my colleagues to disclose SERIOUS ERRORS to patients/families.
9. I am encouraged by hospital leadership to disclose MINOR ERRORS to patients/families.
10. I am encouraged by hospital leadership to disclose SERIOUS ERRORS to patients/families.
Error disclosure trust culture (Cronbach's α=0.68)
3. Disclosing a MEDICAL ERROR in my clinical area damages patient's trust in my competence.*
4. Disclosing a MEDICAL ERROR in my clinical area damages peer's trust in my competence.*
Note: *refers to items that were reverse coded. We included the following definitions in the survey: MEDICAL ERROR = The failure of a planned action to be completed as intended or the use of a wrong plan to achieve an aim. SERIOUS ERROR = Error that causes permanent injury or transient but potentially life-threatening harm. MINOR ERROR = Error that causes harm which is neither permanent nor life-threatening.
We conducted a principal axis factoring analysis with varimax rotation on the 10 items of the error disclosure culture scale to see how many scales we actually needed to measure error disclosure. We computed Cronbach's α to assess reliability of the culture scales and bivariate correlations to examine associations between the different culture scales. We applied the Bonferroni correction when interpreting the significance of the correlational results. We examined via t tests whether error disclosure items focused on serious errors were perceived differently than when those items asked about minor errors. We also estimated per cent positive—the ratio of participants who agreed or strongly agreed with the survey items divided by the total number of participants who responded to the survey item—for our culture dimensions. We reported this information because organisations typically use per cent positive when interpreting the results, with a general guideline of positive per cent <60% indicating an area that needs improvement and a positive per cent of 80% or greater as one indicating ideal performance.
We estimated two types of validity—discriminant and predictive—to examine error disclosure culture. Discriminant validity is established when correlations between variables are not high, indicating that the variables measure different content. To examine discriminant validity, we computed correlations between the safety culture scales to determine whether the scales were highly correlated with each other or not. Correlations between scales <0.85 are indicative of discriminant validity.15 ,16 Predictive validity occurs when the predictor variables are significantly related to important outcome variables. We estimated predictive validity by computing correlations between safety culture scales (ie, predictors) and intent to disclose a hypothetical error (ie, outcome).
We found that error disclosure culture was comprised of two factors. The first factor (herein referred to as error disclosure general culture) consisted of eight items and the second factor (herein referred to as error disclosure trust culture) consisted of two items; box 1 lists the items for these two scales and the Cronbach's α for each scale. Online appendix B contains the rotated factor matrix for the 10 items showing the loadings of these items on these factors.
Table 1 contains demographic information about the participants from each institution. Sample sizes from each institution ranged from 41 to 125. Across institutions, the participants were primarily MDs, which is consistent with the professional composition of the faculty. The participants were split somewhat evenly along gender.
Descriptive statistics for the four types of culture—error disclosure general, error disclosure trust, safety and teamwork—are reported in table 2. For all of the institutions, teamwork culture had the highest mean response. Per cent positive indicated that teamwork culture was the highest culture dimension for four institutions (A, B, E and F) and error disclosure trust culture was the highest culture dimension for the remaining two institutions (C and D).
Table 3 contains bivariate correlations between the different culture scales. All correlations between the different culture scales were <0.85, indicating support for discriminant validity (ie, correlations between variables were not too large). We examined the correlations between the four culture scales and intent to disclose the hypothetical error to determine predictive validity (ie, does type of culture predict intent to disclose). Error disclosure general culture and error disclosure trust culture were the only scales that had predictive validity (r=0.25, p<0.001 and r=0.16, p<0.001, respectively), with safety and teamwork culture not significantly related to intent to disclose (r=0.09, p=NS and r=0.12, p=NS).
We also divided our sample into two groups: those who reported prior education/training about how to disclose medical errors to patients (n=188) and those who did not (n=301). Consistent with our expectations, those who received education/training had higher scores for error disclosure general culture (Mean (M)=3.81, SD=0.75) than those who did not receive such education/training (M=3.56, SD=0.73; t=3.7, p<0.05). Differences were also observed for error disclosure trust culture, with those receiving prior education/training having higher scores (M=3.87, SD=0.90) than those who did not receive education/training (M=3.63, SD=0.89; t=2.9, p<0.05). Given that these items were reverse-coded, these results indicate that those who received previous education perceived disclosure as less damaging to their patients' and peers' perceptions of them than those who did not receive previous education. Taken together, these results reveal that previous disclosure education was linked with more positive perceptions about error disclosure and its effects on patient trust.
Finally, we computed t tests to examine whether significant differences existed between our disclosure culture items when they focused on minor errors as opposed to serious errors. Items 1 (M=3.6) and 2 (M=4.2) focused on disclosing to patients/families (t=13.1, p<0.05), items 7 (M=3.3) and 8 (M=3.8) on support and encouragement by colleagues (t=10.4, p<0.05), and items 9 (M=3.4) and 10 (M=3.7) on support and encouragement by leadership (t=8.4, p<0.05). These items showed significantly higher levels of agreement for serious errors as compared with minor errors.
Our study is the first known attempt to measure error disclosure culture and determine the extent to which it predicts unique information beyond that targeted by other existing measures of culture. We found that our survey measured two important dimensions of error disclosure culture, general and trust. We also discovered that these two types of error disclosure culture predicted participants' intent to disclose a hypothetical error, whereas safety culture and teamwork culture did not. Finally, we found that performance in these two error disclosure culture scales appeared strengthened by prior education about disclosure.
There are three main implications of our study. First, our survey of disclosure culture may help healthcare organisations assess the extent to which the culture of their organisation encourages error disclosure. Such an assessment will allow organisations to understand what needs to be improved so that error disclosure is more likely to occur. For example, our data show that in five of the six institutions surveyed only a minority of respondents had a positive attitude about the general error disclosure culture, and respondents were less likely to disclose minor errors compared with serious errors. Thus, leaders of these organisations can target educational efforts to highlight the importance of and rationale for disclosure in order to improve these attitudes, and potentially improve actual disclosure behaviour. Overall, our results suggest that measuring disclosure culture is worthwhile and reveals important, actionable information institutions could use to enhance response to adverse events and patient safety.
Second, the finding that respondents who had previously received education about disclosure had more positive responses on disclosure culture scales is encouraging for future training efforts. Disclosure training is becoming more common and is being introduced into medical school, residency and faculty level curricula nationwide.17–23 Our study suggests that training can impact attitudes and that the survey items we developed may uniquely detect changes in disclosure attitudes and potentially predict likelihood of transparent communication with patients and families when things go wrong. Future research will need to assess whether training also improves the actual disclosure process.
Finally, we believe that organisations should seriously consider adding survey items about disclosure to annual safety culture assessments. Regularly assessing error disclosure culture will allow organisations to establish baseline measurement initially and then examine changes over time in subsequent years. Disclosure is a professional obligation of care givers and healthcare organisations, and is a marker of patient-centred care.24 It is also reflects the transparency of an organisation, which is believed to be a key component of safe organisations. Care givers and organisations that share information about errors with patients may also be more likely to share the information internally. Such openness to discussing errors is the foundation of creating organisations that can learn from errors and provide safer care.25–29
Future studies can help to better define other potential applications for these error disclosure culture scales. The modest Cronbach's α for the error disclosure trust culture scale suggests that additional survey items that measure similar content could enhance the effectiveness of this scale. Additional studies are also needed to link these error disclosure culture dimensions with safety and quality outcomes, assess whether disclosure training can improve both attitudes and behaviour, and explore the relationship between error disclosure culture and the actual quality of the disclosure process.
A limitation of our study is the low response rate, which was not surprising given that the length of the survey (51 questions) and lack of incentives for completion. Additional possible reasons for this response rate include asking participants about sensitive issues and/or participants lack of belief in the utility of surveys. However, because our study focused on assessing internal validity as opposed to external validity (ie, generalisation), the response rate was adequate for our purposes. We do not make claims about the generalisability of the disclosure attitudes we measured. Future studies with a higher response rate can enable unit level assessments to better understand performance characteristics of disclosure culture in regional work settings. Additionally, as with any self-reported measure, social desirability is always a potential concern. If social desirability did occur, this would mean that the relatively low scores we reported for error disclosure culture are overestimates of disclosure and institutions need to pay more attention to disclosure (ie, reality may be even more concerning than our data suggest).
In conclusion, as error disclosure gains national attention as a potential way to reduce lawsuits, meet patient expectations, fulfil care giver obligations, and increase transparency and quality of care, developing valid methods for assessing disclosure attitudes is critical. We report the first survey that can reliably and validly measure error disclosure culture. Organisations and researchers can use these error disclosure culture scales as a standalone measure, or as an add-on to existing safety culture surveys, to assess disclosure culture and identify educational needs and impact in their organisations. Measuring error disclosure culture is an important step healthcare organisations can take towards enhancing the response to patients affected by adverse events and strengthening patient safety.
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
Data available on request from the first author.
Funding Funding for the first author provided by a K02 award from the Agency for Healthcare Research and Quality (1K02HS017145-02). Funding for the second and fifth authors provided by a R18 award from the Agency for Healthcare Research and Quality (R18-HS019561-01). We have two AHRQ grants that supported efforts by Etchegaray, Gallagher and Thomas.
Competing interests None.
Ethics approval Approval provided byThe University of Texas Health Science Center at Houston's IRB.
Provenance and peer review Not commissioned; externally peer reviewed.