Disclosure of medical error to the patient is frequently discussed in the literature,1^–8 but there are few papers considering the ethical aspects of dealing with medical errors made by others.910 Doctors who notice medical errors made by their peers may find themselves in a conflict between loyalty to colleagues and the obligation to inform the patient. Guidelines and codes of medical ethics generally insist on disclosing one’s medical error to the patient, but there are no definite or strict protocols on how to encounter and handle such errors made by peers.11

When another’s medical error is encountered, the accepted approach is silence, and social norms disapprove reporting.12 Most doctors believe it is best to remain silent, overlook and deny medical error.9 Some doctors may even find disclosing a peer’s medical error to a patient morally unethical. They deem it necessary to defend other doctors, because they think that all doctors are likely to make mistakes and dread dealing with law suits.10

A wide range of approaches have been suggested in the few resources on this issue.691314 In his book, Lo9 gives four approaches to dealing with medical errors made by peers:

• do not mention anything unless you are asked directly by the patient or companions;

• confront the doctor who has made the mistake and ask them to disclose it to the patient;

• arrange a meeting with the patient and your peer;

• directly disclose the error to the patient.

The author does not recommend any of the approaches in particular and only mentions the disadvantages of each approach.9 We did not find any research article on preferred approaches to dealing with peers’ medical errors, even though such studies can help with a better, acceptable norm for doctors. However, acceptability of disclosure of a peer’s error might vary among different cultures. The present study was designed to evaluate doctors’ views on approaches to deal with medical errors made by their peers.

METHODS

This cross-sectional study involved general practitioners and was conducted between April and July 2006. The research ethics committee of Tehran University of Medical Sciences reviewed and approved the study protocol.

We targeted general practitioners because they see more patients compared with specialists and are more likely to encounter peer medical error. We included those general practitioners who had been in medical practice for more than 2 years to ensure at least one encounter with a peer error and so to receive more realistic responses. The doctors were approached through continuing medical education (CME) programmes targeted at general practitioners and held in Tehran: one comprehensive 5-day programme consisting of 50 workshops in all medical disciplines (10 workshops per day) and three other CME programmes on infectious diseases, occupational medicine and general surgery.

Questionnaires were distributed among the general practitioners attending the CME programmes. Participation in the study was voluntary, and participants returned the completed questionnaires on the same day. We offered each participant a small complementary calculator valued at about $1.00 for taking part in the study. The questionnaire contained a hypothetical clinical situation of a medical diagnosis error made by a peer leading to a no-harm event, adverse event or fatal event (box 1). The general practitioners were asked what their approach towards each situation would be:

(A) I would inform the patient or their companions of my peer’s error only if they ask me about the possibility of such error.

(B) I would disclose the error to the patient anyway.

(C) I would take no action in any case.

(D) If the patient or their companions ask me about the possibility of medical error, I would justify my peer’s action.

(E) I would inform my peer of the error he/she has made and ask him/her to disclose it to the patient.

(F) I would report my peer’s medical error to the local medical council.

Participants were free to mark more than one non-conflicting choice for each clinical situation. Also, they were asked what they would prefer their peers to do if they themselves made a medical error.

Since the participants were allowed to make more than one choice for each clinical situation, to make results more informative we grouped the chosen options for the analysis. “Disclosure to patient” included those who chose statements (A), (B) or a combination of other choices with one of these two choices. “Disclosure to the medical council” included those who chose (F) alone or in combination with other choices. “Peer justification” incorporated those who included (D) in their choices. “Disclosure to peer” consisted of participants who chose (E) with or without other choices, and finally, “Take no action” included those who chose (C). Therefore, there was a response overlap between the different groups.

The relative distribution (percentage) and the 95% confidence interval were used in data analysis to define and compare the rate of each approach. We assessed the effect of different categorical variables on the doctors’ approaches with the χ² test, and determined the associations for continuous variables through analysis of variance (ANOVA). The κ coefficient of agreement was calculated to measure the association between doctors’ responses in each scenario.

RESULTS

We distributed 474 questionnaires, and 400 were completed and returned (84.4% response rate). Eight participants were specialists and another five had less than 2 years of practice experience, so their questionnaires were excluded. In all, 387 questionnaires were used in the analyses. With regard to the participants, 113 (29.3%) doctors were women and 273 (70.7%) were men. The mean (SD) age and practice experience was 38.5 (8.3) and 10.7 (8.0) years, respectively.

Table 1 summarises the frequencies of the different doctors’ responses concerning their approach to peers’ medical error for the three clinical scenarios and the approach they expected from their peers when they themselves make a medical error. There was a weak but significant relationship between the approach they chose towards another’s medical error and what they expected from their peers (see table 2 for p values). Table 2 also shows the coefficients of agreements between doctors’ approach towards their peers’ medical error in different outcome events; these are greater than the agreement between their selected approach and their expected approach mentioned above.

For all three outcome events, one-fifth of the participants believed the error must be disclosed to the patient (table 3). About 70% of this group, however, believed it should be disclosed only if the patient or their accompanying person asked (table 1). The frequency of choosing “Disclosure to the medical council” was significantly higher where the error led to death. The highest frequency of choosing “Take no action” was when there was a near-miss event and in this case, it was twice as frequent as for the other two outcome events. The most common approach for all events was “disclosure to peer”. However, participants expected their peers to take this approach more than they would themselves. Among the doctors who expected others to disclose their error to them, 63% stated “Inform me so I wouldn’t repeat it”, 15.7% said “Inform me so I can tell the patient myself”, “Inform me so I can compensate” or “Inform me so I can remedy it”. Others did not give any details about the action they would take.

There was no association between the variables age, gender and years of practice and doctors’ responses concerning their approach to the three clinical outcomes or their expectations from their peers.

In terms of training in approaches to peer medical error, 378 participants responded, of whom 57.7% (n = 218) of the participants claimed they had received no training, 30.7% (n = 116) stated they had received training, and 11.6% (n = 44) claimed they had received some but not enough training. Training only showed a correlation with the doctors’ approach to near-miss errors (p = 0.004; χ² = 28.9). Those with no training (26.2%; 95% CI 25.6% to 26.8%) were more likely to disclose the error to the patient than those who have received training (16.3%; 95% CI 15.2% to 17.4%) or those with insufficient training (13.3%; 95% CI 12.7% to 13.9%) (p = 0.018; χ² = 8.1).

Among the participants, 66.8% (n = 245) had encountered peer medical error in the past 6 months. Of these, only 9.9% (n = 24) had received training on approaches to peer medical error; 60.5% (n = 147) had received no training, and 29.6% (n = 72) had received insufficient training.

DISCUSSION

In this study on general practitioners’ approaches to their peers’ errors, we found that our respondents prefer to report the error to the peer who had made the error so that the peer could disclose it to the patient themselves. They also expect the same approach from their colleague in case of a similar event. According to our results, the least preferred approach to near-miss errors or those leading to harm is reporting them to the medical council, and in cases of errors leading to death, it is justifying the error to the family. In their expectation from their peers, the least preferred approach was doing nothing.

Lawton and Parker15 evaluated incidents reported by doctors, midwives and nurses and found that doctors are reluctant to report their peers’ error to a superior member of staff, even in cases of reverse outcome error (mean 2.97 out of 5), and they are more likely to report a bad outcome error to their colleagues than poor or good outcome errors. Our study shows similar results in terms of reporting to the medical council, although we evaluated other possible approaches to peer error as well.

In agreement with the results of our study, the most frequently expressed opinion on this issue in the literature is that doctors must contact their peers and inform them of their error when they notice one, and encourage them to disclose it to the patient.91316 Wu believes doctors can disclose the error to the patient directly if this method is not efficient,17 while some others think the witnessing doctor can report the error to the authorities when the erring doctor fails to disclose it to the patient.1416 In the present study, the second most preferred approach by doctors was telling the patient. Fost believes it is not necessary to verify the occurrence of the error and doctors are obliged to disclose their logical doubts to the patient.18 If disclosure to the patient occurs in the context of a blame-free culture, it may not ruin the doctor­–patient relationship.

In this study, about 40% of the doctors stated they had received at least some training on approaching peer medical errors. Although not satisfactory, it was more than we expected. It is common in medical schools to teach respect for colleagues as a professional duty. But it is quite unlikely for medical students to receive any specific training regarding how to approach a peer’s medical error. As receiving training was not correlated with any approach options (only reduced disclosure to patients in one type of error outcomes) in our study, it is likely that the participants were referring to the limited education on professional relationships and peer protection.

This study had some limitations. We evaluated doctors’ approaches towards peer medical error by presenting a clinical scenario which might not have reflected their attitudes in real life. Furthermore, most general practitioners practise in offices and they may have different views on errors occurring in ambulatory care. However, they preferred their peers to take a similar approach to their own mistakes.

Only general practitioners were targeted in our study, which could limit the applicability of the results to all specialties. We attempted to compare our study population with the total population of Iranian general practitioners. The only reliable available demographic data on the total Iranian general practitioner population are on the gender proportion (66.5% men), which was similar to that of our study population. As all general practitioners are obliged to obtain credits to revalidate their practice licences, and because we had a good response rate, we are fairly confident there was no remarkable selection bias; however, there may have been a greater number of participants from Tehran than from the provinces.

Notwithstanding these limitations, the findings of this study are worth consideration. Peer error can be encountered quite often, and although the dominating culture in the profession often makes it difficult for us to disclose medical errors to peers, doctors prefer to be informed of their errors. They need to be encouraged to disclose their error to their patients, and so it is important to take a non-judgemental approach towards the peer. Moreover, doctors require education and guidance about how to handle peers’ medical error.