Disclosure of Hospital Adverse Events and Its Association With Patients' Ratings of the Quality of Care

Background  Little is known about how the characteristics of adverse events (AEs) affect the likelihood of disclosure or how the disclosure of an AE relates to patients' perception of quality of care.

Methods  The study included a random sample of medical and surgical acute care adult patients in Massachusetts hospitals between April 1 and October 1, 2003. The unit of analysis was the AE, and multivariable regression analyses accounted for clustering at the patient level.

Results  Overall, 603 patients reported 845 AEs, and 40% of AEs were disclosed. The AEs that required additional treatment (odds ratio [OR], 1.64; 95% confidence interval [CI], 1.16-2.32) or affected patients who reported good health (OR, 2.04; 95% CI, 1.29-3.24) were more likely to be disclosed. Disclosure was less likely if the events were preventable (OR, 0.58; 95% CI, 0.41-0.83) or if the patients were still affected by the AE at the time of survey (OR, 0.49; 95% CI, 0.31-0.78). Higher-quality ratings were associated with disclosure (OR, 2.04; 95% CI, 1.39-2.99) of preventable and nonpreventable events and with patients who felt that they were able to protect themselves from AEs (OR, 1.98; 95% CI, 1.21-3.24). Lower-quality ratings were associated with events that were preventable (OR, 0.55; 95% CI, 0.40-0.76), with events that caused increased discomfort (OR, 0.62; 95% CI, 0.46-0.86), or with events that still adversely affected the patient at the time of the survey (OR, 0.68; 95% CI, 0.46-0.98).

Conclusions  Rates of disclosure of AEs by medical personnel remain low in hospitalized patients. Disclosure of some of these events is associated with higher ratings of quality by patients.

Patient safety is an essential component of the delivery of high-quality care, and the Institute of Medicine's 2000 report, “To Err is Human,” highlighted the need for improvements.¹ In accordance with the Institute of Medicine's report, this study defines an adverse event (AE) as an “injury caused by medical management rather than the underlying condition of the patient. An [AE] attributable to error is a ‘preventable [AE].’” While all AEs result from medical management, not all are preventable nor result in harm. Prior studies in the United States and abroad have identified high rates of harmful AEs.^1-4 Both professional and legislative efforts set an expectation that if an AE occurs health care providers will disclose it to the patient.^4,5

In past surveys,^6-12 patients have consistently reported interest in having AEs disclosed when they occur, including information about the underlying causes, consequences, and steps that have been taken to prevent recurrences. The manner in which the incident is handled has important consequences for the affected patients' (or their families’) decision to take legal action.^13-16 Physicians generally support disclosure; however, many harmful errors are not disclosed to patients.^12,17-22 Fear of lawsuits, shame, embarrassment,lack of disclosure training, fear of losing patient trust, and uncertainty about how to discuss errors with patients are barriers to physician disclosure.^6,17,18,21 Moreover, prior physician surveys document varying opinions regarding what information to disclose and whether disclosure is warranted in near-miss situations as well as in cases involving minor harm to a patient.^6,18-23

Previous research has focused on patients' and providers' attitudes about disclosure. Little is known about how the characteristics of AEs affect the likelihood of disclosure. Also, to our knowledge, the impact of disclosure of an AE on patients' perceived quality of care has not been previously examined. To better understand these issues, we surveyed patients who were discharged from acute care hospitals in Massachusetts about their experiences with AEs.

We analyzed data from a larger study involving 2582 recently hospitalized patients and their experiences during their hospital stay. Previous analyses described patients who did and did not report AEs, and the concordance between patient reports and the medical records.^24,25 The present study was limited to an analysis of AEs reported by patients.

A 2-stage probability sample of patients discharged to home from acute care hospitals in Massachusetts, stratified by the size of the hospital, was selected for study. Based on a desired sample size of 6000 patients, sampling ratios were constructed so that the probability of selection across hospitals was approximately constant for the state as a whole. Details of the sample selection are described elsewhere.²⁴

Of 6003 patients identified by the study hospitals, 4163 were eligible for the study. Eligible subjects were adults, 18 years or older, who were medical or surgical patients who stayed overnight or longer in a Massachusetts hospital between April 1 and October 1, 2003. Telephone interviews were carried out an average of 9 months after discharge. Interviews were completed with 2582 patients, for a response rate of 62% (2582 of 4163) (Figure 1). Detailed interviews were completed with those patients who reported AEs during their hospitalization (n = 749).

A partial HIPAA (Health Insurance Portability and Accountability Act) waiver was obtained to permit hospitals to provide patient names and addresses but no clinical information to the data collection organization. All the protocols for sampling and data collection were approved by the institutional review boards of the institutions with which the investigators were affiliated and by the institutional review boards of the participating hospitals.

The survey instrument underwent multiple stages of development and testing based on focus groups and cognitive interviewing.^26,27 The final interview took 20 minutes on average to administer. The principal measurement objective was to learn about a patient's experience with AEs. We focused on AEs rather than errors for 2 reasons: (1) AEs represent poor outcomes regardless of whether or not they were caused by errors; and (2) patients are more likely to be aware of harm that they sustained rather than of an error that may not have caused them any injury. Patients were asked to describe any “negative effects” or “complications” of their hospitalizations. The first series of questions regarding AEs assessed 4 categories of hospital care: (1) hospital staff administering medicines brought from home, (2) new medicines given in the hospital, (3) surgery, and (4) procedures or tests. These categories of events are well-known sources of AEs. While the list of categories is not exhaustive, it is comprehensive and covers the vast majority of types of events reported in previous studies of AEs in acute care hospitals.^3,28 Any additional events not included in these 4 categories were catalogued as miscellaneous experiences.

Disclosure was defined as a positive answer to the following question: “Did anyone from the hospital explain why the negative effects occurred?” Additional questions were asked about the effects of the AEs (Table 1). All patients were also asked to rate their perception of the quality of the medical treatment they received. Quality ratings were categorized on a 5-point scale (1, excellent; 2, very good; 3, good; 4, fair; and 5, poor).

We also collected information on patient characteristics (sex, age, and race/ethnicity) and asked patients about their self-perceived health status and whether they felt able to protect themselves from AEs. Furthermore, we asked about the effects of AEs on their hospitalization (whether there was increased discomfort, increased length of stay, and additional therapy needed or whether the patient was still affected at the time of the interview).

Sampled patients received an initial mailing that included an introductory letter, a letter from the hospital, and answers to frequently asked questions. A 1-800 number was provided for patients to opt out. Interviews were conducted by professional interviewers from the Center for Survey Research at the University of Massachusetts, Boston. Interviewers called the sampled patients, explained the purposes of the study, answered questions, and attempted to arrange a time to conduct the interview. A minimum of 6 calls were made at different times and on different days to those who did not respond to earlier calls.

After the interviews, the respondents' information about each AE was reviewed independently by 2 physicians (E.C.S. and S.N.W.). The physician reviewers had access only to what the respondents reported in the interview. A reported incident was excluded if the event was likely the result of the underlying condition rather than of medical care, if it occurred before the sampled hospital admission, if no AE was apparent from the patient's report, or if the incident appeared to duplicate other reports. A total of 749 patients reported AEs. Of these 749 patients, 603 reported 845 events that met our definition of an AE (Figure 1).

The physician reviewers then coded the severity and preventability of each patient-reported event using categories adapted from previous studies.^29,30 Based on their clinical experience and judgment, they classified each incident as life threatening, serious, significant, or insignificant. A serious event was one that affected organ function; a significant event resulted in symptoms, prolonged hospitalization, or abnormalities in laboratory test results. Insignificant events were defined as those that were judged to be of little consequence, such as a venipuncture or an infiltrated intravenous catheter site. The 2 physician reviewers judged preventability in the context of average care in US hospitals. Reliabilities were calculated based on 120 cases that the reviewers both reviewed independently, yielding excellent κ scores for the presence of adverse drug events (κ = 0.97) or other AEs (κ = 0.85). The reliability was good for preventability (κ = 0.71) and fair for severity (κ = 0.45).

Preliminary analyses found no effect on estimates of weighting for differential nonresponse across hospitals. Furthermore, because the design effect resulting from the 2-stage sample was equal to 1.01, adjustments of variance estimates were not needed.³¹

The unit of analysis was the AE. Binary logistic regression was used to determine predictors of disclosure. The complete binary model consisted of all patient (Table 2) and event (Table 1) characteristics. A series of ordinal logistic regression models were constructed to assess whether the patient or AE characteristics were individually associated with higher quality, unadjusted for confounders. A complete ordinal logistic multivariate model was then constructed to evaluate the effect of all patient and event characteristics that were associated with patient quality ratings. Because the survey comprised primarily white participants (92%), the other race/ethnicity groups were collapsed into a “nonwhite” category. Another ordinal logistic model, with quality ratings as the dependent variable, was applied to all previously mentioned variables along with a severity-disclosure interaction term. Finally, to assess the impact of preventability of AEs on patient-perceived quality, the previously mentioned ordinal logistic models were analyzed separately for the subset of events that were categorized by the physician reviewers as preventable. Generalized estimating equations (the GENMOD procedure in SAS software, version 9.1; SAS Institute Inc, Cary, North Carolina) were used to account for clustering of multiple AEs at the patient level. Preliminary analyses found no significant clustering at the hospital level.

Our study consisted of 845 AEs reported by 608 patients (Table 2). More than 50% of the participants were white (92%), female (64%), and older than 50 years (66%). Racial and ethnic minorities represented 11% of the sample. The majority of the participants reported being in good health (73%) and felt that they were able to protect themselves from AEs (83%).

A small percentage of the events (7%) were attributable to the way the participants' previous medications were managed in the hospital, and 40% were related to a newly prescribed drug. Surgical procedures accounted for 34% of the events, while tests and procedures accounted for 9% of the total. Most participants reported 1 incident, while 27% reported 2 or more events (maximum reported, 7) (Table 1).

Of 845 AEs, 40% were disclosed (as defined above) to participants. The physician panel ranked 31% (n = 260) of the events as preventable and 75% as severe. Of the preventable events, 30% were disclosed. More than 50% of the patients reported increased discomfort, but only 22% reported still being affected at the time of interview because of the AE. An increased length of stay because of the AE was reported by 24% of the patients.

Patients were less likely to report disclosure if they were older than 50 years, did not report good health, experienced preventable events, or were still affected by the AE at the time of the interview (Table 3). In multivariate analyses, patients were less likely to report disclosure if the event was judged preventable or if they were still affected by the AE at the time of the interview. Disclosure was more likely with AEs that required additional treatment and among patients who reported good health (Table 4).

Overall, the mean (SD) quality rating for patients in this study was 2.09 (1.18) (1, excellent [42%]; 2, very good [27%]; 3, good [18%]; 4, fair [9%]; and 5, poor [5%]). Patients were more likely to rate the quality of their hospitalization higher if there had been disclosure of the AE in both bivariate (Figure 2) and multivariate (Table 5) analyses. These findings remained unchanged in the subset analysis that was restricted only to physician-rated preventable events (n = 260). The adjusted odds ratio (OR) that was associated with disclosure on patients' quality ratings was 2.20 (95% confidence interval [CI], 1.25-3.86) for preventable events. Patients rated the quality of their hospital experience lower if they were still affected, if they had increased discomfort because of the AE, or if they reported being in better health. These analyses controlled for patients' perceived ability to protect themselves from an AE. Other factors associated with lower-quality ratings included whether the event was rated preventable and whether the event was associated with new medicines that were prescribed during the hospitalization. Among preventable events, lower quality was associated with patients reporting increased discomfort because of the AE (OR, 0.46; 95% CI, 0.27-0.78). No statistically significant interaction was found between the effect of the severity of the AE on patient quality ratings by whether the AE was disclosed or not (P = .87).

Our study has 5 major findings. First, we found that fewer than half of the AEs reported by patients were disclosed. Second, an AE was more likely to be disclosed if additional treatment was provided to address the AE during hospitalization. Third, clinicians were less likely to disclose events that were associated with a more prolonged impact on the patient. Fourth, they also disclosed fewer preventable events than nonpreventable ones. Finally, patients who reported an AE and had it disclosed to them gave a higher rating of the quality of their care than patients who reported an AE that was not disclosed. Importantly, high-quality ratings continued to be associated with disclosure whether or not the event was deemed preventable.

Studies suggest that physicians generally agree that disclosure of AEs is an important component of high quality of care.^{6,7,12,17,18,21} However, physicians are also reluctant to fully disclose errors to patients and have varying opinions of when and how much to disclose.^20-22 We found a relatively low rate of disclosure by medical personnel (40%), similar to previous studies documenting a “disclosure gap.”^{14-17,20,21,23} We also found that the probability of disclosure was related to the nature of the AE, suggesting that hospital staff are selective about which events they disclose.²² Disclosure was more likely if the AE required additional treatment and less likely if the AE was preventable. In the former case, the event may have more often been apparent, making disclosure moot. These findings are consistent with prior research demonstrating that for errors that might not be apparent to the patient, physicians are less likely to disclose the errors and are more likely to provide less information about the errors when they do disclose them.²² Our findings on preventability suggest hesitancy in disclosing events that might imply hospital errors. This may be because of a fear of lawsuits, shame, and a lack of disclosure training.

Our results suggest that patients may appreciate being informed about errors in their care and may view disclosure of errors as a signal of high-quality (rather than low-quality) care.^6-8,17 Disclosure of AEs doubled the odds that patients would give higher ratings to the quality of their care. This finding is consistent with previous patient surveys documenting the high value that patients place on disclosure.^6-13 Prior studies have shown that disclosure may strengthen the physician-patient relationship,^11,32,33 and failure to disclose has been associated with decreased patient trust and satisfaction and with an increased chance of a malpractice suit.^9,16,17,34 Our findings extend the literature by showing that disclosure is associated with patient perception of higher-quality care, even among patients who have been harmed by an AE. Our results may reassure physicians who worry about the consequences of disclosure on patients' perceptions of quality.

Our survey has the advantage of eliciting views of quality offered by patients who report having been harmed by an AE. Our study sample consisted of recently discharged medical and surgical patients who reported 1 or more AEs. This approach provides a potentially more accurate perspective on disclosure behavior than previous studies that have surveyed patients about their opinions regarding hypothetical vignettes involving AE disclosure.^{7,11,12,17,23}

Relying solely on medical record review for identifying AEs has proved to be insufficient owing to variable standards of documentation, clinician unawareness or oversight, and concern about liability exposure.^35-37 Several studies suggest that patients are capable of identifying medical errors.^37-39 Patient reports have important advantages in that patients may uncover AEs that are not identified from the medical record.^37,39 There are also disadvantages to patient reporting of AEs, including patients' lack of technical knowledge, poor recall, and social desirability bias, because patients may feel the need to respond favorably, contrary to their actual experience. However, a prior study of our data suggests that even many months after discharge patients are still able to recall AEs that occurred during their hospitalization. We found that agreement between the medical chart review and patient survey was 77% for all events and 94% for serious and life-threatening events.²⁵ These findings are slightly better than those of a single teaching hospital analysis of inpatient AEs that stated that 55% of patient-reported events had confirming information in the medical record.³⁹

Our study had several limitations. First, the average 9-month interval between the patient's discharge and the telephone interview may have led to underreporting of AEs because of recall bias and may have increased the reported incidence of events that had a greater clinical impact. Most of the events were characterized by the physician panel as serious, making the finding of an association between higher perceived quality and disclosure particularly striking. Second, underreporting may have occurred because of our inability to survey those patients who died or who were sickest, 2 groups that may be at highest risk for AEs. Younger and healthier patients may be better able to participate and report AEs than older and sicker patients, potentially overestimating the relationship between quality and disclosure. Clinicians may have preferentially disclosed the AE and their errors to healthier patients: those who were more likely to recover from the incident and to understand what had occurred. Third, disclosure is a complex event or series of events that our data set may not be able to fully capture. It is possible that the AE was disclosed to a family member without the patient's awareness. Fourth, our reliance on a single question for quality limits the ability of our study to characterize the different dimensions of quality that might be related to the disclosure process. Fifth, most respondents were white, and our findings may not generalize to other racial or ethnic communities that may experience and report complications differently or rate the quality of their care differently.⁴⁰ Finally, although our data are from 2003, they are consistent with more recent data on disclosure. A recent study by Iedema et al,⁴¹ using interview data from late 2007, demonstrated that patients injured by AEs value and expect disclosure from their providers.

Disclosure of AEs is an important component of high-quality care. The Institute of Medicine's 2000 report, “To Err Is Human,” defines safety in health care as an essential component of quality health care delivery.¹ The National Quality Forum has defined disclosure of unanticipated outcomes to patients as an essential dimension of high-quality health care and has included disclosure to its list of safe practices.⁴² These standards have been endorsed by major quality-of-care organizations such as the Joint Commission, the Institute for Healthcare Improvement, the Agency for Healthcare Research and Quality, and the Centers for Medicare and Medicaid Services. Our findings demonstrate that patients, even when harmed, consider disclosure of AEs as integral to quality. These findings remain even in cases in which the AEs were deemed preventable (ie, errors) by physician review. Also, disclosure is essential to patient-centered care because it respects patient autonomy and truth telling.⁴³ Our findings suggest that the disclosure gap remains prevalent even though patients rate their care favorably when AEs are disclosed. Interventions to increase disclosure may contribute to patients' perceptions of quality.

Correspondence: Lenny López, MD, MDiv, MPH, Institute for Health Policy, Massachusetts General Hospital, 50 Staniford St, Ninth Floor, Boston, MA 02114 (llopez1@partners.org).

Accepted for Publication: September 5, 2009.

Author Contributions: All authors 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: López, Weissman, Schneider, and Epstein. Acquisition of data: Weissman and Weingart. Analysis and interpretation of data: López, Weissman, Weingart, Cohen, and Epstein. Drafting of the manuscript: López, Weissman, and Epstein. Critical revision of the manuscript for important intellectual content: López, Weissman, Schneider, Weingart, Cohen, and Epstein. Statistical analysis: López. Obtained funding: Weissman, Schneider, and Epstein. Administrative, technical, and material support: Weissman, Weingart, and Cohen. Study supervision: Weissman and Epstein.

Financial Disclosure: None reported.

Funding/Support: This work was supported by a Cooperative Agreement from the Agency for Healthcare Research and Quality to the Massachusetts Department of Public Health (U18 HS11928). Dr López acknowledges the support of an Institutional National Research Service Award (5 T32 HP11001-19).

Role of the Sponsor: The sponsors had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the manuscript.