Patient Perceptions of Mistakes in Ambulatory Care

Background  Little information exists about current patient perceptions of medical mistakes in ambulatory care within a diverse population. We aimed to learn about the perceptions of medical mistakes, what factors were associated with perceived mistakes, and whether the participants changed physicians because of these perceived mistakes.

Methods  We conducted a cross-sectional survey at 7 primary care practices in North Carolina of English- or Spanish-speaking adults, aged 18 years and older, who saw a health care professional during 2008. Main outcome measures were 4 questions about patient perceptions of medical mistakes in the ambulatory care setting, including (1) overall experience with a medical mistake; type of mistake, such as a (2) diagnostic mistake or (3) treatment mistake, and its associated harm; and (4) effect of this mistake on changing physicians.

Results  Of 1697 participants, 265 (15.6%) responded that a physician had made a mistake, 227 (13.4%) reported a wrong diagnosis, 212 (12.5%) reported a wrong treatment, and 239 (14.1%) reported having changed physicians because of a mistake. Participants perceived mistakes and harm in both diagnostic care and medical treatment. Patients with chronic back pain, higher educational attainment, and poor physical health were at increased odds of perceiving mistakes, whereas African American patients were less likely to perceive mistakes.

Conclusions  Patients perceived mistakes in their diagnostic and treatment care in the ambulatory setting. These perceptions had a concrete effect on the physician-patient relationship, often leading patients to seek another health care professional.

Since the Institute of Medicine's publication of To Err Is Human: Building a Safer Health System,¹ attention to errors and mistakes in health care has increased dramatically. Most research examining medical mistakes has focused on inpatient care, but evidence suggests that patients perceive mistakes and experience errors in their ambulatory care as well.^2-4

By exploring patient perceptions of medical mistakes, health care professionals can better understand patients' satisfaction with the current health care system. Although patient perceptions of mistakes likely encompass broader concepts than the strict definition of medical error, these perceptions have been shown to affect care regardless of whether a true adverse event has occurred.^5-9 Psychological and emotional harm can result from perceived mistakes whether or not they would be defined as errors or adverse events by the medical community.⁸ Perceptions of mistakes have been shown to play a role in patient satisfaction,^10-12 which is linked to physician trust and medication adherence.^13-15 Thus, even though patient perceptions of medical mistakes may not always represent true adverse events, they nonetheless may influence patient satisfaction, regimen adherence, and other outcomes and therefore deserve study.

Among the sparse literature on medical mistakes, most research in ambulatory care has focused on validating medical errors using incident reports or malpractice claims,^16-19 a method that underestimates the prevalence of perceived mistakes. Small qualitative studies also have been published,^8,20 as well as a few larger patient surveys, but these were conducted in fairly homogeneous populations such as health maintenance organization groups or academic settings.^5,13,21 One national telephone survey carried out in 1997 on behalf of the National Patient Safety Foundation found that 22% of 639 participants reported that a mistake had happened to themselves, a family member, or a close friend while in a physician's office.²² The largest and most rigorous study of ambulatory care was conducted more than 10 years ago and focused only on patients' perceptions of adverse drug events rather than on all medical mistakes.²³ Therefore, there is a need to better understand the current perceptions of medical mistakes in ambulatory care within diverse populations.

To address these deficits, we surveyed a large, racially and socioeconomically diverse sample of adults about their perceptions of medical mistakes in ambulatory care. We aimed to learn about their perceptions of mistakes, to what extent the mistakes had caused perceived harm, what factors were associated with perceived mistakes, and whether the respondents changed physicians because of these perceived mistakes.

We conducted a survey in 2008 of 7 primary care medical practices belonging to a practice-based research network²⁴ from geographically representative regions in North Carolina. Two were located in urban settings, 1 in a suburb, 2 in towns, and 2 in rural settings. Three were university-affiliated clinics, 2 were community health centers, and 2 were private practices. The 7 practices were chosen because of their racial and socioeconomic diversity.

We placed 1 or 2 trained research assistants in the waiting room of each practice for 15 days, with instructions to offer study participation to all adults who saw a physician, nurse practitioner, or physician assistant.²⁴ Persons in acute distress or who could not comprehend the consent form were excluded. Research assistants approached eligible patients, obtained written consent, and offered assistance with survey completion using methods approved by the institutional review board of the School of Medicine of The University of North Carolina at Chapel Hill. The written survey was self-administered and was returned to the assistants at the clinic. Bilingual assistants were placed in practices with significant Hispanic populations. Both English and Spanish versions of the consent form and survey were available in all practices. We recruited 1754 patients, with a 63.9% recruitment rate of eligible patients. These analyses involved the 1697 people (96.8% of participants) who responded to the survey's screening question on medical mistakes.

As part of a broader survey of issues affecting primary care patients in North Carolina, 4 questions about medical mistakes were asked to assess patients' perceptions of the prevalence and severity of medical mistakes in the ambulatory care setting. Although these questions were asked in primary care offices, they were designed to focus on the care provided in all types of ambulatory care clinics. The questions were derived from published sources,^{13,17,22,25,26} modified by the research team, and reviewed for content and face validity by 10 primary care researchers. The term mistake was chosen instead of medical error because studies have demonstrated patients' confusion about the latter term.^27,28 Participants could answer yes to any of the 4 questions, which were not mutually exclusive. The questions were as follows: (1) Has a doctor in a doctor's office ever made a mistake in your care? (2) In the past 10 years, has a doctor in a doctor's office made a wrong diagnosis or misdiagnosed you? (If yes, how much harm did this cause you?) (3) In the last 10 years, has a doctor in a doctor's office given you the wrong medical treatment or delayed treatment? (If yes, how much harm did this cause you?) (4) Have you ever changed doctors because of either a wrong diagnosis or a wrong treatment of a medical condition? Harm was quantified using a 5-point Likert scale with categories of “none,” “a little,” “some,” “a lot,” or “severe.”

The 4-page study survey included questions on participants' demographics, medical history, self-reported health, frequency of physician visits, and disability. General demographic information included age, sex, race/ethnicity, marital status, and level of education. Self-reported health was evaluated by asking participants to rate their health on a 5-point scale. Participants were asked to indicate their chronic illnesses from a list of 16 common conditions such as hypertension, cancer, type 2 diabetes mellitus, and chronic back pain. Disability was measured by participant report of days of limitations in activities due to problems with physical or mental health within the past 30 days.²⁹

To explore patients' perceptions of mistakes, a random sample of participants who responded yes to either of the questions about a wrong diagnosis or a wrong treatment were selected for interview (conducted by C.E.K. or a trained research assistant). A written consent form was sent to the participants along with an invitation to participate in the telephone interview. Of 82 participants selected, 59 had interviews completed, of which 52 were eligible for analysis. Of the 23 who did not complete an interview, 9 were unreachable, 7 refused, and 7 asked to be rescheduled but could not be contacted in a timely fashion. Seven interviews were ineligible for analysis because they referenced non–ambulatory care mistakes. The interview comprised key probing questions aimed to elicit the events surrounding the perceived medical mistake, including the general timeline, location, type of mistake, and perceived sequelae. Verbal responses were recorded on individual surveys by the interviewer. Interviews lasted 30 minutes on average, and all open-ended responses were recorded verbatim in written notes by the investigator. Participants reported diagnostic mistakes, treatment mistakes, or both. Seven participants reported the same mistake as both a diagnostic and a treatment mistake because they believed they had received improper treatment from an incorrect diagnosis. The decision was made to categorize these as reports of diagnostic mistakes only. We excluded nonnative English speakers in an effort to minimize interpretation issues. Furthermore, only 5 sites were included in this portion of the study because the other 2 sites had not been sampled at the time the interview cohort was selected. A content analysis was conducted independently by 2 researchers (C.E.K. and P.D.S.), and conflicts were resolved by consensus.

Simple frequencies were used to describe the sample and responses to the 4 medical mistake questions. Bivariate comparisons were computed, relating whether participants had ever perceived a mistake in the ambulatory care setting with factors previously demonstrated to affect health care utilization, including selected chronic comorbid conditions,³⁰ health status, disability, race, age, and sex.^5,31

To identify characteristics independently associated with perceiving a medical mistake in ambulatory care, we estimated a hierarchical multivariate regression model. Given the small number of sites, a fixed hierarchical model that clustered participants by site was used to control for confounding effects of site on our outcome of interest. We entered potential predictors, whose bivariate associations with perceiving medical mistakes were significant at P < .15, into the multivariate model. Significant predictors were found using the backward selection technique. After the model was finalized, assessment of collinearity and interaction was performed. There was an almost significant interaction (P = .051) between chronic back pain and report of any days spent in poor physical health, but controlling for this interaction did not substantively change the findings and was therefore excluded in the final model. Assessment of collinearity yielded no variance inflation factor greater than 1.86. The presence of significant interaction between predictors was assessed. Multiple imputation modeling to control for missing data yielded similar results. All descriptive analyses and hypothesis testing were completed using Stata/SE 10.0 software (StataCorp, College Station, Texas).

Age and sex distributions were similar to overall US averages of ambulatory care clinics.³² The mean (SD) age of participants was 46 (16) years (range, 18-95 years). Of the participants, 1160 (68.4%) were women, 620 (36.5%) were African American, and 329 (19.4%) were Hispanic (Table 1).

Of the 1697 participants, 265 (15.6%) responded that a physician had made a mistake in their care, 227 (13.4%) reported a wrong diagnosis, 212 (12.5%) reported a wrong treatment, and 239 (14.1%) reported having changed physicians because of a mistake. About two-thirds (n = 151) of those who reported having experienced a mistake changed their physicians. Only about 4% (n = 74) said yes to all 4 questions. The severity of perceived harm was similar regardless of the type of mistake (Figure). Of the 218 participants who reported harm from a perceived diagnostic mistake, 91 (41.7%) reported “a lot” [of] or “severe” harm. Of the 208 participants who reported harm from a perceived treatment mistake, 95 (45.7%) believed they had “a lot” [of] or “severe” harm. A list of the 52 participants' interviews shows a range of perceived mistakes and associated types of harm (Table 2). Three primary types of mistakes were mentioned in these interviews: communication/relationship issues, normal diagnostic/treatment challenges, and possible adverse events/near misses.

Physician-patient communication and relationship issues were mentioned across the category of severity of harm. One participant described, “I went to the doctor, and they just told me that there was nothing wrong with me. There was something wrong with my belly, but they thought it was all in my head. They just mistreat people there, and they didn't care. I finally went to a university hospital, and I had endometriosis.” Another participant related the following story: “I’ve had 2 almost identical incidents where the same doctor called me after I’d had a routine stress test, and each time he said he thought I’d had a silent heart attack, but it turns out it was an old heart attack that I’d had like 10 years ago, but he didn't read my chart at all. When it happened the second time, I said that's enough, and I transferred over to another doctor that my wife was seeing.”

Some of the reported mistakes causing “none” or “a little” harm might be considered by health care professionals to be normal diagnostic or therapeutic challenges. For example, one participant reported, “I had a rash all over my legs and stomach and back. The doctor said it was just an allergic reaction to something, but it wasn't getting better. I went to a dermatologist and found out that I had psoriasis.” Another reported, “I had a sinus infection in the spring of 2007, and they gave me some low-strength antibiotic and it did nothing. . . . Well, I was still feeling lousy and so I had to go through 3 iterations and on the third try it finally got better.”

The interviewed participants who reported “a lot” [of] or “severe” harm from diagnostic or treatment mistakes appeared more likely to have had what the medical community would define as true adverse events or near misses. One participant reported, “I had a swollen lymph node under my arm; it was very tender. I went to my main doctor, who sent me to a specialist, and that doctor wanted to take off my breast. I wanted another opinion. I got one, and that doctor sent me for a mammogram and biopsy of the lymph nodes. It turned out I had ‘cat scratch fever.’” Another reported the following: “I had my breast removed because of cancer. I had surgery, and I had saline breast implants afterward. The area got infected. It started as a pinhole and got bigger and bigger and was just rotting away. It got so bad I had to put a towel in my bra because it was oozing so bad, but my doctor who did the surgery said it was going to be okay. Finally, my family persuaded me to get a second opinion. That second doctor took one look at me and took me right in for emergency surgery.”

Of the participants, 239 (14.1%) reported changing physicians because of a perceived mistake. Only 6.3% of participants (88 of 1397) who reported they had never experienced a mistake reported changing physicians. Most participants (66 of 89, or 74.2%) who reported “a lot” [of] or “severe” harm from a diagnostic mistake changed their physician, whereas 39.4% (26 of 66) who reported “none” or “a little” harm did so. Similarly, 75.5% (71 of 94) of those who reported “a lot” [of] or “severe” harm from a treatment mistake changed their physician, whereas only 43.4% (23 of 53) who reported “none” or “a little” harm changed physicians. Often, participants reported that changing physicians was accompanied by a reluctance to discuss the mistake with their former physician. As one participant reported, “I never talked to the same doctor again. If I told him, I don't think he would have taken it very well. I think he would have told me that I was wrong again.”

Several patient characteristics were associated with increased odds of perceiving a mistake. Approximately 19.6% of white patients perceived a mistake vs 13.1% for African American patients and 13.4% for Hispanic patients (Table 3). Educational attainment beyond high school increased the odds of perceiving a mistake 2-fold (P < .001). Participant report of any day in the past month in poor physical health was associated with a 2.2-fold increased odds of perceiving a mistake (P < .001). The only medical condition to be associated with increased odds of perceiving a mistake was self-report of chronic back pain, which was associated with a 1.5-fold increased odds (P = .02). Age and sex were not associated with perception of mistakes but were kept in the model for strong theoretical reasons. Depression, heart disease, report of any days in the past month in “not good” mental health, more than 4 visits to a physician within the past year, and self-reported health of “fair” or “poor” were associated with perception of mistakes in bivariate analyses but were not significant in the final model.

In this diverse primary care sample, 15.6% of participants perceived mistakes in ambulatory care and 14.1% changed their physicians because of these mistakes. Participants perceived mistakes in both diagnostic care and medical treatment. Mistakes were perceived to have caused harm across the spectrum of severity. Around 8% of participants reported “a lot” [of] or “severe” harm due to diagnostic and treatment mistakes. Factors independently associated with a higher likelihood of perceiving a mistake were higher levels of educational attainment, report of poor physical health, and chronic back pain, whereas African Americans were less likely to perceive mistakes.

This study, which included a large geographic area, urban and rural settings, and insured and uninsured groups of patients, found that perceptions of mistakes in ambulatory care are widespread. Approximately 15% reported a medical mistake, similar to the 11% reported in the 2005 study by Solberg et al⁵ of patient perceptions of mistakes in ambulatory care. If a physician sees 30 patients a day, as many as 5 of these patients may believe they have experienced a mistake in their care at some point.

A significant minority who reported mistakes believed they had experienced “a lot” [of] or “severe” harm. This finding conflicts with published reports of adverse events, which concluded that most do not cause serious harm.^33-35 It has been argued that not all mistakes are concerning, only those that cause or have the potential to cause harm.³⁶ However, patients may perceive mistakes and harm, pursuing litigation in regard to known adverse effects of treatment and normal diagnostic and treatment challenges, when patient-physician communication is poor.³⁷ Broader concepts of mistakes and harm, in which mistakes around such issues as communication or normal diagnostic and treatment challenges can be perceived even when standards of care are met, appear to be prevalent among patients.

Approximately 14% of participants changed their physicians because of a perceived mistake. This is somewhat higher than the 10% rate at which respondents to the National Patient Safety Foundation survey reported changing physicians because of perceived mistakes.²² Changing physicians because of a perceived mistake is a valid measure of dissatisfaction and could be a useful measure in efforts to improve patient satisfaction in ambulatory clinics.⁵

Our additional qualitative information supports prior studies demonstrating that perceived mistakes involve communication and relationship problems in addition to diagnostic or therapeutic errors. Similar conclusions have been reported in other qualitative studies.^8,20 Kuzel et al⁸ found that access and relationship issues were more commonly reported as mistakes than technical issues, such as misdiagnosis or improper medical treatment. Given that participants may have trouble distinguishing diagnostic from treatment mistakes, as 7 of our 52 interviews suggest, this broad view of mistakes may mean that prior classifications of errors²⁶ may not be helpful in research on patient perceptions.³⁶ Furthermore, our participants frequently reported events such as medication trials or dermatologic challenges that physicians would consider normal diagnostic or treatment processes. Additional cognitive testing of these frameworks is needed to see whether these divisions in classification are applicable to patients' perceptions. Efforts to prevent true adverse events may not be sufficient to improve public perceptions of mistakes in the ambulatory care setting; the medical system may also need to improve the communication of expectations for care.

Knowing which patients are at increased risk for perceiving mistakes may be useful to physicians so that they can more explicitly set expectations. Frequent users of health care, due to complex disease or multiple comorbidities, are at increased risk for true adverse events and perceived mistakes.^23,38,39 For example, chronic back pain, which may be therapeutically challenging, is a risk factor for perceiving a mistake. Low educational level and minority status typically confer an increased risk of medical errors, a finding that appears to conflict with our results.³⁹ This discrepancy may exist because minority status and lower educational levels are associated with increased patient satisfaction^40,41 and lower rates of complaints.³¹ Perception of mistakes, therefore, may be due to both true adverse events and patient expectation, a conclusion supported by patient satisfaction models.^10,42 Patients of minority background may have lower expectations and be less likely to perceive mistakes, whereas those with poor physical health or chronic conditions are more likely to perceive mistakes owing to their frequent use of health care and the increased opportunity to experience an error.

Although this study was limited to adults in primary care, it represents a large, diverse statewide sample. Because all patients had a primary care physician, these findings may not represent the perceptions of patients without primary care. However, the outcomes did focus on any perceived event in ambulatory care and are not limited to perceptions of primary care. No review of medical records was performed, so no comparison data on actual adverse events were available. Similarly, because this research was patient focused, no assessment at the site level was performed other than recording the site itself. The limitations that apply to in-office survey studies, such as selection bias, apply to this study as well. This bias may cause oversampling of individuals who are more frequent users of primary care, such as those with chronic illnesses. Finally, although the study questions were drawn from previously published sources, cognitive interviewing with these questions was not done in this particular survey.

In conclusion, to our knowledge, this study is the first large, diverse cross-sectional survey of patients' perceptions of medical mistakes in the ambulatory care setting. In addition, it quantifies the influence of these perceptions in terms of perceived harm and decisions to change physicians. Our results indicate that patients with chronic back pain, higher educational attainment, and poor physical health are at increased odds of perceiving a mistake across a wide range of ambulatory care settings. These perceptions have concrete effects on the physician-patient relationship, often leading patients to seek another physician. Therefore, it is important to identify ways to address patients' perceptions of mistakes. Such interventions would likely include attention not only to situations defined by the medical community as adverse events but also to patient expectations of health care encounters and to physician-patient communication regarding diagnostic and therapeutic processes and outcomes.

Correspondence: Christine E. Kistler, MD, Department of Family Medicine, The University of North Carolina at Chapel Hill, 590 Manning Dr, Chapel Hill, NC 27599 (umanohone@yahoo.com).

Accepted for Publication: March 22, 2010.

Author Contributions: Drs Kistler, Walter, and Sloane had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Kistler and Sloane. Acquisition of data: Kistler, Mitchell, and Sloane. Analysis and interpretation of data: Kistler, Walter, and Mitchell. Drafting of the manuscript: Kistler. Critical revision of the manuscript for important intellectual content: Kistler, Walter, Mitchell, and Sloane. Statistical analysis: Kistler. Obtained funding: Sloane. Administrative, technical, and material support: Mitchell. Study supervision: Sloane.

Financial Disclosure: None reported.

Funding/Support: Dr Kistler is supported by grant T32AG000212-16 from the National Institute on Aging. Dr Walter is supported by VA Health Services Research and Development grant IIR 04-427, grant IR01CA134425 from the National Cancer Institute, and grant 2006-0108 from the John A. Hartford Foundation, Inc, as well as the Robert Wood Johnson Foundation Physician Faculty Scholars Program. Data collection was supported by Academic Career Award K07 AG21587 to Dr Sloane from the National Institute on Aging.

Additional Contributions: We thank Hana Masood, BS, a medical student at East Carolina University, for her help with conducting the follow-up interviews, and Kate Kirby, MS, a senior statistician in the University of California–San Francisco Division of Geriatrics, for her assistance with data analysis.