

* Department of Emergency Medicine
University of North Carolina School of Medicine, Chapel Hill, North Carolina
Department of Epidemiology and Biostatistics, Case Western Reserve University School of Medicine, Cleveland, Ohio
|| Department of Health and Policy Administration, University of North Carolina School of Public Health, Chapel Hill, North Carolina
| ABSTRACT |
|---|
|
|
|---|
Methods. A 4-scenario survey instrument portraying a range of medical error was provided to a convenience sample of parents who presented with children to an emergency department. Parents were asked to categorize the error, express preferences for disclosure and reporting, and then report how they expected to respond with and without disclosure. Basic demographics were collected also. Bivariate analyses of demographics were performed with Fisher's exact tests, analysis of scenario responses was performed with Somers' D, and the independent effects of the study variables were assessed with a generalized estimating equation.
Results. Research assistants approached 661 parents; 499 participated (75% response rate). Of all scenarios presented to the parents, they judged 54% of the scenarios as severe, 99% wanted disclosure, 39% wanted the error reported to a disciplinary body, and 36% were less likely to seek legal action if the error was disclosed by the physician. In multivariate modeling, severity was associated with desire for disclosure, reporting, and change in likelihood of legal action with disclosure.
Conclusions. Regardless of severity, parents want to be informed of error. Educational interventions to improve error disclosure should emphasize the uniformity of parental preferences for disclosure, reporting, and the decreased likelihood of legal action when errors are disclosed than if discovered through other means.
Key Words: emergency department patient safety medical error reporting systems pediatrics
Abbreviations: RR, relative risk CI, confidence interval
Medical error is an area of concern for the American public, with 42% reporting that they had been affected by a medical error either personally or through a friend or relative.1 In pediatrics, medical errors have been noted in emergency care, ambulatory, inpatient, intensive care, and surgical settings.26 Data about parental preferences for disclosure after an error in pediatric patients is lacking. In adult patients, we know that there is an overwhelming desire to be informed of medical error, and the desire for disclosure increases with error severity.711 Although disclosure of error is a moral and ethical obligation, physician disclosure of medical error is uncommon, in part because physicians are uncertain of patient responses to error disclosure and may fear disciplinary or legal action.8, 9, 1215
Another reason that physicians may hesitate to disclose medical errors is the possibility that patient preferences differ by patient race/ethnicity, gender, or socioeconomic status. Indeed, substantial research has focused on the contribution of patient preferences to racial/ethnic disparities in treatment.1618 In general, patient-provider communication has been shown to be less effective when social distance exists between the physician and the patient.1923 Minority patients tend to be less assertive and less active in the physician-patient interaction24 and less satisfied with their ability to communicate with their physician.25 Physicians may interpret these differences as being indicative of less interest in learning of errors.
Effective educational interventions to assist physicians with error disclosure require understanding patient preferences for disclosure, likely patient responses to error disclosure, and how preferences and responses change with different patient populations. The purpose of this study was to characterize parental preferences for error disclosure and reporting about errors in the care of their children; assess responses to error disclosure; and determine how preferences and responses differ by parental race/ethnicity, gender, age, and insurance.
| METHODS |
|---|
|
|
|---|
18 years old presenting at any hour to a tertiary care academic emergency department. Parents who were competent to conduct the interview in English and whose pediatric patients were not triaged into the most severe category were approached by trained research assistants and asked to complete the survey with or without the assistance. The emergency department had an annual patient volume of 60000 visits (with 14000 pediatric visits) and supported residents from all hospital departments, including an emergency medicine residency. This protocol was approved by the University of North Carolina School of Medicine Institutional Review Board.
Survey-Instrument Development
Four vignettes were developed to portray a range of error type and severity: (1) medication error, potentially serious but with no clinical effects; (2) medication error, lifelong disability; (3) failure to diagnose, minor clinical effects; and (4) failure to diagnose, lifelong disability. The second part of the survey included demographic questions. After initial development by the investigators, the survey instrument was presented for comment to 10 practicing physician-faculty members from a range of specialties. After revision based on these comments, the survey instrument was presented to patients and assessed through cognitive interviewing.26 The final version was established after 4 iterations when participants demonstrated understanding of all items in the instrument. A 10% pilot was performed to ensure instrument performance (the survey instrument is shown in Fig 1). To control for possible survey effects, 4 versions of the survey were randomly administered with different scenario orders.
|
Independent Variables
Based on previous work,711 we hypothesized that the desire for disclosure would increase as error severity increased. Although the scenarios were carefully developed to represent a range of severity, we determined severity by asking the participant if the scenario portrayed an error and then asking for categorization of the error as "minor," "moderate," or "severe." Parental and child race/ethnicity, age, and insurance were self-categorized by the parent. Race/ethnicity was categorized as white, African American, American Indian, Hispanic, Asian, or other. We included only participants who responded "white" or "African American" because of sample size. Parent and child race/ethnicity were concordant in 92% of the sample, so we based our analysis on parental race/ethnicity. Parental age was categorized as <21, 21 to 30, 31 to 45, and >45 years; child age was categorized as <1, 1 to 5, 6 to 13, and 14 to 18 years. Insurance was self-categorized by the parent as public, private, or uninsured.
Analyses
All data were entered into a Microsoft Access database (Microsoft Corp, Redmond, WA) and then converted into a Stata 8.0/SE (Stata Corp, College Station, TX) database by using Stat/Transfer (Circle Systems, Seattle, WA).
We used cross-tabulations with Fisher's exact tests to explore bivariate differences in the demographic characteristics and race/ethnicity (Table 1). For consistency across the cell size and to account for correlation within a participant's responses, we used Somers' D clustered on participant to determine the statistical significance of the associations of preferences for disclosure, reporting to a disciplinary organization, and response to error disclosure with parental race/ethnicity, gender, age, and insurance (Tables 2 5).27
|
|
|
| RESULTS |
|---|
|
|
|---|
Questionnaire Characteristics
To determine internal reliability of the items, we asked several paired items addressing desire for information, desire for financial compensation, likelihood of legal action, and desire for reporting of the error. All of the paired items showed acceptable internal validity with pairwise correlations ranging from 0.68 to 0.84. Order effects were explored with
2 tests on all dependent variables, and after the Sidak correction for multiple comparisons, respondents were slightly more likely to seek legal action if they learned of the error from other means in the scenario portraying a medication error with medical consequences if that scenario was the first scenario presented. To ensure that respondents were responding to each item independently without simply checking all of 1 response option, we analyzed the distribution of responses within each participant scenario. We found that the mean SD was 1.2 with an interquartile range of 1.11 to 1.38 on the 5-point item-response scale, which assured us that the "yeah-sayer" bias was sufficiently minimal.29 Because the desire for disclosure, desire to report, and likelihood of litigation may be influenced by the extent to which the parent attributed the error to the physician, we included the item "I believe the physician is the party most responsible for this event." This item had a mean intraparticipant SD of 0.8 (on a 5-point item-response scale) across the 4 scenarios, which assured us that participants' attribution of the error to the physician was consistent across scenarios.
Demographics
The study population was 77% female parents, 44% female children, 34% parents under the age of 31, 46% children aged 6 to 18, and 50% with private insurance (Table 1). African American parents tended more than white parents to be more female (85% vs 72%, respectively; P < .01), have more children <1 year old (23% vs 11%, respectively; P < .01), and be more likely to be publicly insured (69% vs 33%, respectively; P < .01) (Table 1). The 2 groups had no meaningful difference in terms of child gender or parental age.
Bivariate Results
Of all the scenarios to which participants responded, participants identified 54% of the scenarios as severe, 99% warranted disclosure, only 39% wanted the error reported to a disciplinary body, and 36% were less likely to seek legal action if they were informed of the error by the physician (Table 2, second column). Although there were no meaningful differences by parental race/ethnicity in preferences for disclosure or change in likelihood of legal action with disclosure, African American parents were more likely than white parents to rate the event as a severe error (62% vs 49%, respectively; P < .01) and more likely to want the event reported to a disciplinary organization (50% vs 33%, respectively; P < .01). Analysis by gender revealed no important differences in severity assessment, desire for disclosure or reporting, or change in likelihood of legal action (Table 2).
There were no age-related differences in severity assessment, desire for disclosure, and change in likelihood of legal action with disclosure (Table 4). There was a modest but statistically significant trend for younger parents to desire that the error be reported to a disciplinary organization. With respect to insurance status, there were no differences in desire for disclosure or change in likelihood of legal action with disclosure (Table 5). There was a clinically meaningless but statistically significant association with the publicly insured, who were more likely to rate the mistake as severe, than the privately insured (57% vs 50%, respectively; P < .01). There was a definite trend for the publicly insured to be more likely than the privately insured to want the event reported to a disciplinary body (46% vs 30%, respectively; P < .01).
|
| DISCUSSION |
|---|
|
|
|---|
We have demonstrated a previously unexplored parental desire for disclosure of errors in their children's care. Studies of adults711 demonstrate an increased desire for disclosure with increasing error severity; in this study we identified a very modest effect of error severity on desire for disclosure. This undoubtedly is related to the near-universal desire for disclosure expressed by parents. Even in scenarios for which participants rated the error as minor, there was still a strong desire for disclosure. Although there is great activity in exploring differences in patient preferences and patient-provider communication by patient race/ethnicity, gender, age, and socioeconomic status,1625 we have demonstrated that the desire for disclosure is equivalent for these 2 groups.
Physicians may hesitate to disclose errors to patients because of a fear that patients will want the error reported to a disciplinary body.8, 11 We found that this fear may be justified and that the desire for reporting increases as the perceived severity of the error increases. Although parental gender, age, and insurance were unrelated to the desire for reporting, African American parents were more likely to want the error reported to a disciplinary organization even after adjusting for multiple potential confounders. A recent study suggested that adult patients are more likely to forgive physician errors as the patients' level of education increases.30 It is possible that our finding is related to unmeasured educational level that was not addressed sufficiently with the inclusion of insurance status. Although not addressed with these data, it is also possible that parents who feel greater social distance from physicians are more likely to desire physician accountability.1923
Although physicians may fear that the likelihood of legal action increases with disclosure,8, 11 we found the opposite to be true. Fully 36% of the parents reported that they would be less likely to seek legal action with disclosure than if they learned of the error from another source. For the remaining 63%, disclosure did not affect their desire to seek legal action. It should be noted that this sizeable decrease in the likelihood of seeking legal action is predicated on the assumption that the parent would learn of the error through an alternative route if the physician did not disclose. Although parental race/ethnicity, age, gender, or insurance did not change the likelihood of seeking legal action with disclosure, error severity had a substantial effect. Parents seemed to be most likely to reduce their likelihood of seeking legal action with disclosure when the error was moderate. This may be because the moderate error was sufficient to interest the parent in seeking legal action but not so egregious as to convince the parent to seek legal action irrespective of physician behavior. This finding of parental preference is consistent with work done with adult patients who preferred full disclosure, but the disclosure was unlikely to alter the desire for legal counsel if serious harm occurred.31 Although these results suggest that physicians are less likely to face legal action when an error is disclosed, physicians may nonetheless feel that the patient may be unaware that an error occurred if there is no disclosure.
Our study possesses several limitations worth noting. First, the study was performed at a single institution and may not be generalizable to other institutions. It is possible that local culture and attitudes toward medical care varies with geography. It would seem especially likely that the threshold to litigate would vary from country to country and may even have substantial regional variation within the United States. Such a variation in threshold to litigate would provide different baseline responses with respect to litigation and response to disclosure. The influence of race, gender, age, or insurance, however, would differ only if such variation differs systematically from our institution by race, gender, age, or insurance. Nonetheless, these findings should be explored with a more diverse sampling strategy before they are broadly generalized.
Second, this study relied on participants to respond to hypothetical scenarios as they would to real events. Physicians have been demonstrated to respond similarly to patient vignettes and actors portraying patients,3234 but parents may respond differently to abstract clinical vignettes than they would to injury or suffering of their own child from medical error. Parents indicated a reduced likelihood of seeking legal action in the face of disclosure, but meeting a parent's emotional needs may have less sway when the parent faces financial opportunity presented through legal action. Furthermore, the wording, construction, order, and layout of the questionnaire influence the responses,29, 35, 36 and our results may have differed had any of these factors been altered.
Third, we were limited in our ability to adequately address the interaction effects and the association between patient characteristics and likelihood of seeking legal action. The coefficients for the association of patient characteristics with the desire for disclosure and the desire for reporting were sufficiently consistent for us to determine clinical equivalence. For the association between patient characteristics and the change in likelihood of seeking legal action, as well as the interaction effects, however, we were able only to demonstrate that the associations had modest and statistically nonsignificant estimates. Inability to find a difference, however, does not indicate that no meaningful difference exists.
| CONCLUSIONS |
|---|
|
|
|---|
|
| ACKNOWLEDGMENTS |
|---|
| FOOTNOTES |
|---|
Address correspondence to Cherri Hobgood, MD, Department of Emergency Medicine, University of North Carolina, 101 Manning Dr, Chapel Hill, NC 27599. E-mail: hobgood{at}med.unc.edu
C.H. and B.W. conceived the project and developed the study materials; A.E. collected and entered the data; J.H.T.-S. constructed the analytical data set and analyzed the data; C.H., J.H.T.-S., and B.W. interpreted the data; J.H.T.-S. drafted the manuscript; C.H., B.W., and A.E. contributed substantially to revision; and C.H., J.H.T.-S., B.W., and A.E. take responsibility for the paper as a whole.
| REFERENCES |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
M. Levetown and and the Committee on Bioethics Communicating With Children and Families: From Everyday Interactions to Skill in Conveying Distressing Information Pediatrics, May 1, 2008; 121(5): e1441 - e1460. [Abstract] [Full Text] [PDF] |
||||
![]() |
J. R. Clarke, J. C. Lerner, and W. Marella The Role for Leaders of Health Care Organizations in Patient Safety American Journal of Medical Quality, October 1, 2007; 22(5): 311 - 318. [Abstract] [PDF] |
||||
![]() |
J. Garbutt, D. R. Brownstein, E. J. Klein, A. Waterman, M. J. Krauss, E. K. Marcuse, E. Hazel, Wm. C. Dunagan, V. Fraser, and T. H. Gallagher Reporting and Disclosing Medical Errors: Pediatricians' Attitudes and Behaviors Arch Pediatr Adolesc Med, February 1, 2007; 161(2): 179 - 185. [Abstract] [Full Text] [PDF] |
||||
![]() |
T. H. Gallagher, J. M. Garbutt, A. D. Waterman, D. R. Flum, E. B. Larson, B. M. Waterman, W. C. Dunagan, V. J. Fraser, and W. Levinson Choosing Your Words Carefully: How Physicians Would Disclose Harmful Medical Errors to Patients. Arch Intern Med, August 14, 2006; 166(15): 1585 - 1593. [Abstract] [Full Text] [PDF] |
||||
![]() |
L. Berlin Will Saying "I'm Sorry" Prevent a Malpractice Lawsuit? Am. J. Roentgenol., July 1, 2006; 187(1): 10 - 15. [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||