BACKGROUND

Apologies may play a significant role in medical care, especially in the context of patient safety, medical error disclosure, and malpractice risk.1 From a legal and interpersonal perspective, apologies are constituents of expected professional behavior.2 Research demonstrates that when state laws, institutional policies, and individual skills align—including the ability to offer a genuine apology—patients and families benefit 3,4,5. As of 2019, 38 states and the District of Columbia have enacted the so-called apology laws that protect physicians by preventing apology statements from being used as evidence in court proceedings.6 As such, knowing the difference between a genuine apology and a partial, or “non-apology” apology (defined below) is important; failure to do so may jeopardize the well-being of patients and also threaten the reputation of physicians and the organizations in which they practice.

Research shows that when patients who feel insulted or ignored do not receive an apology for harm caused to them, they are more likely to sue for medical malpractice. This is in direct contrast to the actions of those patients and family members who do receive a sincere apology and an offer to “make things right” 7,8. In the face of having caused significant harm to patients, there is evidence that physician apologies, coupled with communication skills training, reduces the risk of medical malpractice suits.3,9,10 Physicians who have been sued for malpractice, irrespective of the outcome, experience negative psychological sequelae, including anger and depression 11. Accordingly, the skills involved in knowing when and how to apologize may be important for emotional, interpersonal, and economic health on both sides of the stethoscope.

Given the importance of apologies in medicine, it is striking that relatively little empirical research, outside of malpractice risk, has been conducted in day-to-day practice or educational settings. While conceptual models and suggestions for effective apologies have been put forward, 12,13,14 these are largely based on clinical experience, anecdote, and/or small samples. After conducting a librarian-assisted search of the literature, we found very few examples of apology training or research in undergraduate medical education. The exceptions were interventions that have focused on the apology process only where significant harm has occurred, or as a sub-component of error disclosure. 15,16 Because error disclosures are distinct from apologies—for example, a person can disclose an error with apologizing for it—it is difficult to disentangle whether, or the extent to which, this training was specific to apology.

One area of medical education that has received substantial attention is the influence of the “informal” or “hidden” curriculum on medical students’ attitudes, behaviors, and actions.17,18 The primary finding from this research is that students learn as much by observing and adopting the behaviors of others (generally from those more senior to themselves) as they do in formal classroom settings. The implications for how students learn about apologies from the informal curriculum are clear. Absent formal instruction, medical students are likely to adopt the behaviors they see other, more senior, faculty engaging in with their patients, colleagues, and peers. When these behaviors include partial or “non-apology” apologies, faculty may inadvertently be modeling suboptimal behaviors.

Non-apology apologies, which appear to be on the rise, are “often offered in the passive (e.g., ‘mistakes were made’), or the conditional (e.g., ‘If I have offended anyone, I am sorry’) and rarely entail an outright admission of wrongdoing on the part of the apologizer.” 19,20 To date, there is ample anecdotal evidence, as well as a plethora of case studies (mostly in the mass media21,22,23), demonstrating the harm that often accompanies non-apology apologies. At least one empirical study of non-apology apologies has shown that they may reduce the propensity to forgive.24

To learn more about what medical students observe, experience, and learn about the apology process, we conducted a qualitative content analysis of 238 unique third-year student professionalism narratives drawn from 7,384 that were collected over a 15-year period at Indiana University School of Medicine (IUSM). Given the relative lack of knowledge in this area, we chose a discovery-oriented qualitative method, immersion crystallization, for our analysis.25

METHODS

From 2003–2018, third-year medical students on their internal medicine rotation at IUSM participated in an IRB approved, AMA-funded project entitled STEP, Strategic Teaching and Evaluation of Professionalism. Students were invited to post a narrative responding to the prompt “write about an experience that taught you something about professionalism.” After removing all identifying information, the narratives were used as the basis for faculty-facilitated discussions about professionalism. (Detailed descriptions of the program have been published elsewhere 26,27,28.)

Our conceptualization of apology was modeled on the work of Aaron Lazare,12 a physician who defined apology as an interpersonal process (i.e., interaction) where the needs of the offended and the offender converge. In Lazare’s conceptualization, the offended party seeks a restoration of dignity and respect, an assurance that they are safe from further harm, an understanding that they have suffered from the offense, and a promise of adequate reparations. In turn, the offender desires a remediation of guilt and shame, an opportunity to rebuild the relationship, and avoidance of further damage. The apology process consists of the following elements: (1) acknowledgement of the offense; (2) an explanation for why the offense occurred; (3) remorse, shame, or humility on the part of the offender; and (4) reparation on behalf of the offended. Lazare’s is one of several models that include the same four basic elements, although they are sometimes arranged in a different order. As a relatively new area of study, there are as yet no validated measures of apology in the literature. We chose Lazare’s approach to develop our coding scheme because it aligns with expert consensus and because his scholarship has been applied in medical contexts.29

In the current study, apologies were classified as incomplete, or non-apology apologies, if they contained only the first or second elements; partially complete, if they contained the first three elements; and complete if they contained all four elements. We only classified apologies as containing the fourth element, reparation, if the offender went above and beyond what was normatively expected to make up for the offense. An example of a single narrative containing all the elements is presented below.

  1. (1)

    “There was an incident on our service where a patient was supposed to be discharged after receiving IV fluids for a certain amount of time…These IV fluids were forgotten about and the patient ended up leaving without ever receiving them. Unfortunately, the patient developed acute renal failure most likely secondary to having not been given those fluids… (Element 1: Acknowledgement)

  2. (2)

    Following this incident, on morning rounds in front of the entire team, the attending sat down, looked the patient in the eyes, and admitted that it was his fault that this occurred. (Element 2: Explanation)

  3. (3)

    The attending said that he was deeply sorry and took full responsibility for the mistake and ensured that the patient knew this. (Element 3: Remorse/shame/humility)

  4. (4)

    After the apology, the attending went on to describe how he would personally take care of the issue and do everything…to return his [the patient’s] renal function to normal…” (Element 4: Reparation)

In this complete apology, the attending took responsibility for the incident, provided an explanation by acknowledging that it was his/her fault, expressed remorse, and offered reparation (i.e., taking over the case and ensuring proper treatment).

Data analysis

We first developed a data dictionary based on words and synonyms that involve apology. The dictionary terms and frequencies of use in the student narratives are summarized in Table 1.

Table 1 Data Dictionary

Next, the statistical program R was used to identify these key words within the database of 7,384 narratives. Potential narratives were then imported into REDCap and reviewed to identify duplicates and false positives. To be included in the final sample, narratives had to include two or more words that appeared in the data dictionary and were either synonyms of apology or suggested that an apology might be present. For example, both the words “sorry” and “regret” would need to be present for the narrative to be included. We chose this particular threshold because several of our target words could be in narratives but not be representative of an apology interaction (e.g., “I am sorry for writing this so hastily.”; “I am sorry this treatment is not working optimally”). We labeled these false positives. Thus, apologies had to have been offered in the context of an offense or mistake. In total, 363 narratives met our initial inclusion criteria. Upon manual review of these narratives, 125 were determined to be false positives, leaving a final sample of 238.

The database of 238 narratives was independently coded by two analysts (ICF and RMF) using the apology coding form (Appendix Table 6). Coding differences were resolved by consensus. In addition, a third analyst (RG) coded 10% of the narratives to ensure the trustworthiness of the coding. There was 100% consensus on the final coding.30

The data were then examined in several ways. First, we analyzed the elements included in each apology. Next, we coded the data by professional status. In addition, we noted when the apologies occurred. This question was of interest because there is variability in the literature on recommendations for the timing of apologies. Some authors suggest that an apology should occur as soon after an offense as possible, 31 while others suggest a waiting period until emotions cool down.32 We also coded whether the apology interactions were positive or negative and labeled them as lessons learned or narrative residues. This was done to assess whether more complete apologies were experienced as more or less positive. We carefully analyzed each interaction for tone and language, with particular attention paid to instances where students explicitly noted that they found the experience to be positive (e.g., inspiring) or negative (e.g., shameful). Finally, we calculated in how many instances proffered apologies were fully, partially, or not accepted by those receiving them.

RESULTS

In total, 238 apology narratives were analyzed. Table 2 presents the distribution of apologies according to the number of elements that were present.

Table 2 Elements of Apology Present in Student Narratives

In virtually all instances, some form of acknowledgement for an offense occurred. This might include statements such as “I’m sorry,” “forgive me,” or “I apologize”—the three most common forms acknowledgement took. Apologies containing only the first two of the four elements—non-apology apologies—occurred less than half the time. These apologies, consisting of an acknowledgement and an explanation, often took the form of, “I apologize, I was tied up with another patient.” A similar number of apologies contained three apology elements, constituting a partially complete apology. Examples included, “I genuinely am sorry for how I responded; I feel terrible. I was tired and took it out on you.” Or, “I apologize. I’m embarrassed to say that I lost track of time and that’s why I’m late.” Complete apologies containing all four elements occurred in fewer than one out of every five cases. When they did occur, there was a statement or offer of reparation such as, “Your time is valuable. I will buy a smart watch with a timer function today to help keep me on time for future visits.”

Who Apologizes

Table 3 describes the types of individuals who initiated apologies.

Table 3 Individuals Involved in the Apology Process

In all, almost three quarters of the apologies came from attending physicians and residents, the two groups primarily responsible for medical student education. The fact that the majority of apologies came from faculty suggests that students were largely observing their teachers and less frequently involved in offering apologies themselves.

Patients and/or family members (165; 69%) were the most frequent recipients of apologies, followed by the students (20; 8%), faculty (20; 8%), or nurses (14; 6%). Others, such as lab techs or other team members, accounted for 19 cases (8%). The person most frequently apologizing to patients and/or family members was the attending physician (96; 58%), followed by the intern or resident (35; 21%), and the student (34; 21%). In a small number of cases (N = 6/238; 3%), students apologized on behalf of the attending as the following excerpt illustrates.

A patient complained to me that the attending was rude. The nurse was in the room and she agreed. I said what a good doctor he is and that I think he takes his job seriously and that may be interpreted as rude. I apologized on his behalf. I am not sure if i should tell him about this.

Apology Timing

Of those apologies for which timing could be identified, the majority (N = 126; 57%) occurred as the event was unfolding or immediately thereafter. In a significant minority of cases, apologies were delayed until the end of a shift or the next day (96; 43%). To determine whether the timing of an apology influenced students’ experiences, we analyzed the impact of immediate vs. delayed apologies. Table 4 presents the results using the chi-square statistic to look for differences.

Table 4 Impact of Immediate or Delayed Timing on Student Experience

The chi-square statistic was 0.67, p value = 0.41, which was not statistically significant, suggesting that timing had little impact on students’ overall experience of the interaction.

Impact on Student Experiences?

The majority of apologies students observed were seen as positive (N = 166; 70%) and were often praised as models of high-quality care. Many students explicitly noted that they hoped to be like the attendings and residents they were observing. For example, one student noted:

She [the resident] was truthful and informative and was able to apologize for the decisions made overnight without placing blame on others. From observing this discussion I really wondered if I could have facilitated such an open discussion where the patient and family could be informed and feel supported… I would hope that I could use what I saw when faced with a similar situation in the future.

At the same time, the experience for some students was negative, especially when faculty or fellow students did not properly apologize for obvious lapses. In almost a quarter of the cases (N = 58/24%), students reported being embarrassed by the actions of their colleagues or superiors, as one author described:

I felt embarrassed that someone who had received similar training to me would act this way towards a patient….I also felt angry that this patient simply had to accept the behavior because she was depending on this physician for a life-saving treatment…. we are all responsible to each other to practice professionalism with our patients and with each other in order to uphold the ideals of medicine and remain proud of our profession.

In a few cases, the faculty’s apology-related behaviors had a profoundly negative impact. These interactions were highlighted by students as examples of ways not to act as a medical professional.

It taught me exactly what I never want to let happen to me as a future physician. You should never get so angry or frustrated as to hinder team cohesiveness and possibly harm patient care.

Apology Completeness and Student Experience

Table 5 describes apology completeness and its effect on whether the student judged the experience as positive or negative.

Table 5 Apology Completeness and Impact on Student Experience

There was a significant effect of apology completeness on student experience (p < 0.001). Specifically, the more complete the apology, the more positive the student rated their overall experience of the interaction.

Apology Reception

In almost two-thirds of cases, the acceptability of the apology was not included in the narratives we analyzed (N = 155; 65%). Where it was possible, apologies were either partially or fully accepted 80% of the time (66/83). We determined a full acceptance to have occurred when the writer noted that the offended party explicitly said something like, “I accept your apology.” Partial apologies occurred when the student observed a calming effect on the offended (e.g., went from angry and closed off to open and communicative). Apologies were not accepted in 20% of interactions (17/83). There were too few cases to conclude anything definitive regarding the effectiveness of apology completeness on apology reception.

DISCUSSION

Taking responsibility for one’s own or others’ actions, clearly explaining what happened, demonstrating remorse, and repairing the damage are critical elements of a genuine apology. How well apologies are constructed and delivered affects both their effectiveness and their impact. In this study, we used students’ narrative descriptions to better understand the apology process as a lived experience and as it occurs in day-to-day practice. In particular, we analyzed the frequency with which attendings and residents engaged in the act of apology. We also assessed whether the quality of these interactions influenced students’ experiences positively or negatively. We found that the majority of faculty and residents engaged in incomplete apologies, mostly immediately following errors or mistakes. We also found that when apologies occurred, even if they were not complete, medical students and patients tended to view them as favorable.

In general, our data suggest that the type of apology observed may have implications for students’ overall learning experiences. It was clear from our analyses that the more complete the apology, the more highly the students rated the interaction. This is encouraging, for it suggests that the more genuine an apology is, the more likely it is to be seen in a positive light. Moreover, because nearly two-thirds of apologies came from an attending physician or resident, it also suggests that students may be witnessing positive behaviors on which they could model future interactions with patients and colleagues. That this did not seem to be influenced by the timing of apologies is also noteworthy. Overall, our data do suggest that students recognize the importance of genuine apologies.

At the same time, it is worth keeping in mind that complete apologies were relatively rare; only one out of five met our criteria, and 80% were enacted by faculty. Thus, while complete apologies were seen by students as positive when they occurred, they were also infrequent. Moreover, not all our results are positive. The fact that the majority of apologies were incomplete or “non-apology” apologies is concerning. For example, one study that measured each of the four elements found that acknowledgement and reparation were the most important when offering a sincere apology. 33 In addition, a large field study found that reparation (a necessary ingredient for a complete apology) had the largest effect on consumer satisfaction and subsequent behavior.34 Few reparations occurred in our data suggesting that students may be regularly observing faculty responses to mistakes and lapses that are suboptimal. More troubling, perhaps, is the fact that, in some instances, apologies were deemed distasteful or unacceptable by students, so much so that several felt obligated to apologize on behalf of a faculty member’s behavior. This is not what should be occurring in a teaching environment. Coupled with the low frequency of complete apologies, this suggests that faculty may be inadequately trained in how to conduct and model the apology process. If left unaddressed, students may assume that this behavior is normal or, more problematically, acceptable.

Our findings have several implications. First, in terms of professional development, greater awareness on the part of faculty and residents regarding their role in shaping the values, attitudes, and skills of trainees cannot be overstated. If faculty provide the basic template from which trainees learn about apology, they need to be both skilled and aware of the impact their behaviors can have on others. The failure (or inability) to demonstrate genuine apologies may leave students without exposure to this desirable clinical skill and its potential benefits. As with any skill, development takes patience, practice, and effective coaching. Nevertheless, given the current focus in medical education on teambuilding and trustworthy interprofessional relationships, knowing when and how to apologize are skills that medical students should be exposed to early on in training. Role play, small group discussion, and asking for feedback to improve when apologies occur are all ways in which students can build their skills and increase the effectiveness of their care.

This study has several limitations. First, the data come from a single institution over a 15-year period and cannot be generalized to other institutions. Second, longitudinal data is subject to secular changes and the content and quality of narratives may have changed over time. We compared word length and anecdotal impressions of the coders and did not see changes to either length or content. Third, the data we analyzed came from an open-ended prompt asking students to describe an experience that taught them something about professionalism; the exercise was not focused specifically on the apology process. Nonetheless, 238 students chose to write about apology interactions, suggesting that it is of some importance. Fourth, the dictionary of terms used to identify apologies may have been incomplete thereby excluding some unknown number of narratives. We note that a number of apology-related words in our search returned relatively few or no results suggesting that the bulk of apologies were, in fact, included in the analysis. Fifth, we had limited data from which to generate useful conclusions about the impact of apologies on patients and family members. Finally, stratifying the results by professional status may have introduced bias into how the results were reported. Nevertheless, from our perspective, this approach made the most practical and methodological sense in terms of the questions we were attempting to address.

CONCLUSION

The role of apology in everyday clinical interactions is understudied and often confused with other socially desirable elements of care, such as being empathic or disclosing harmful errors. More needs to be done in undergraduate medical education to clarify the importance of apology in everyday practice and help students develop mastery of the skills involved in making genuine apologies. In a world where partial and non-apologies have become a seemingly legitimate way to limit risk and responsibility, genuine acknowledgement and apology for wrongdoing, whether it involves a microaggression or operating on the wrong limb, stand out as a prescription for reducing suffering and beginning a process of healing.

Apology has an important role to play in professional identity formation 35,36,37 and contributes to the idea that learning to be a medical professional is more than just demonstrating mastery of a set of mechanical skills. It requires developing a way of being that extends beyond oneself to the unique relationships doctors form with their patients, colleagues, and communities. We believe this important skill is worthy of greater attention and research, especially given the broader social trend toward substituting partial or non-apology apologies for the real thing. All human beings, including doctors, make mistakes and misjudgments. We would all do well to remember a paraphrase of the Bard’s words “To err is human; to (genuinely) apologize is divine.”