- Email: [email protected]
Reduction in Patient Enrollment in the Veterans Health Administration After Media Coverage of Adverse Medical Events
Joint Commission
Journal on Quality and Safety
Patient Safety
Reduction in Patient Enrollment in the Veterans Health Administration After Media Coverage of Adverse Medical Events
William B. Weeks, M.D., M.B.A., C.H.E. Peter D. Mills, Ph.D., M.S.
nsuring patient safety is recognized as an increasingly important aspect of medical care. Studies have indicated that medical errors are common.1,2 Many medical errors appear to be preventable3 and can be attributed to systems issues.4 Although there has been legitimate debate about the size of the problem,5,6 there is little debate that health care should reduce the frequency of adverse events and medical errors. Errors in health care are associated with costs that patients tend to bear. These include significant iatrogenic morbidity and mortality,1,7 direct costs of additional required care,8,9 and indirect, long-term costs of lost income, increased disability rates, and increased burden on caregivers.10,11 Direct and disability costs associated with adverse medical events in the United States have been estimated to be $37.6 billion per year.10 Less is known about the costs that organizations bear when medical errors occur. We previously outlined a model of organizational costs associated with preventable medical errors, including legal, marketing, and operations costs.12 In that model, one hypothesized cost is lost revenue associated with poor brand image and subsequent patient disenrollment; however, no studies have evaluated whether such a relationship exists. The Veterans Health Administration (VHA) offers an ideal setting in which to determine whether such a relationship exists for three reasons. First, VHA is a national health care system that operates under a single
E
652
December 2003
Article-at-a-Glance Background: Health care organizations may experience costs associated with preventable adverse events in the form of poor brand image and subsequent patient disenrollment. A retrospective cohort design was used to determine whether media coverage of adverse events that occurred in Veterans Health Administration (VHA) hospitals was associated with subsequent veteran disenrollment. Methods: Twenty-four newspaper reports of medical adverse events that occurred between 1994 and 1999 within the VHA system were identified. Regionally adjusted changes in enrollment rates for VHA facilities that had reported adverse events were compared with those that had not one year before and one and three years after publication of the newspaper reports. Results: Facilities that had published reports of adverse events had lower enrollment rates after publication of the report for two groups of veterans. Conclusions: Within the VHA system, health care organizations involved in adverse events that generated publicity suffered a greater rate of patient disenrollment. If safe patient care practices can reduce adverse publicity, they may enhance corporate value by maintaining enrollment of the patient population.
Volume 29 Number 12
Copyright 2003 Joint Commission on Accreditation of Healthcare Organizations
Joint Commission
Journal on Quality and Safety
administration; all VHA medical centers are subject to the same enrollment policy. Second, because VHA medical centers are organized into regional Veterans’ Integrated Service Networks (VISNs) that develop regional strategies regarding enrollment, regional variation due to different regional enrollment patterns can be identified and corrected. Finally, VHA collects enrollment information on two categories of users that differ in their reliance on the VHA system. Therefore, one can test the hypothesis that increased reliance on a health care system may be associated with more tolerance for quality concerns. To determine whether there is an association between media-facilitated damage to brand image and patient disenrollment, we examined the relationship between media coverage of adverse events and veteran disenrollment in the VHA between 1994 and 1999. In particular, we examined relationships between the date of media reporting on the adverse event, veteran category, and veteran disenrollment, controlling for regional changes in enrollment over time.
Methods Identification of Newspaper Articles In August 2002 we searched a computerized database to identify reports of medical adverse events that occurred within the VHA system and were published between October 1, 1993, and September 30, 1999. The database collects local, national, and international newspapers from more than 30,000 news and business sources.13 To identify these adverse events, we searched news sources using the following search terms: fatal mistakes, malpractice, negligence, wrongful death, veteran care, VA hospital, and Veterans Affairs. After eliminating duplicate articles, we identified 24 separate media reports of 36 adverse medical events that were published in the specified time frame.14–37 For each media report, we identified the paper in which the report was published and the date of publication. From the text of the article, we identified the nature of the adverse event, the date of the adverse event, and the VHA medical center in which the adverse event occurred (Table 1, page 654).
Changes in Patient Enrollment From the VHA centralized administrative database, we obtained the number of patients seen each year at
December 2003
each VHA facility between 1994 and 2001 for two groups of veterans. Category A veterans have received disability ratings for their service or have low incomes; they are likely to be more dependent on the VHA health care system than Category C veterans, who are not receiving disability payments and who have higher incomes. To compare regionally adjusted differences in enrollment, we first calculated the rate of change in enrollment for each patient category at each facility for one and three years. For example, the three-year change in enrollment for a facility in 1994 would be the number of veterans enrolled in 1997 minus the number of veterans enrolled in 1994 divided by the number of veterans in 1994. We performed the same calculations for each VISN. We then generated a standardized change in enrollment rate for each facility by dividing the change in enrollment rate for each facility within a geographic VISN by the VISN change in enrollment rate and represented these growth rates as a deviation from one. For example, the standardized three-year change in enrollment rate for a facility is calculated as follows: 1+ ((Facility enrollment in 1997 – Facility enrollment in 1994) / (Facility enrollment in 1994)) 1+ ((VISN enrollment in 1997 – VISN enrollment in 1994)/ (VISN enrollment in 1994))
The standardized change in enrollment rate for a facility represents that facility’s growth, relative to the growth of all the facilities in the region. If a facility’s standardized change in enrollment rate is 1.0, enrollment at the facility is changing at the same rate as enrollment in the rest of the region; if it is less than 1.0, change in enrollment is lower than in the rest of the region, and if greater than 1.0, change in enrollment is faster than in the rest of the region. We calculated one-year and threeyear standardized change in enrollment rates for each category of patient (Category A and Category C veterans) at each facility, for each year of the study.
Relationship Between Reports of Adverse Events in the Media and Changes in Enrollment We were interested in examining the relationship between newspaper media coverage of an adverse event and changes in patient enrollment. We hypothesized that a relative decline in enrollment would follow publication of newspaper reports about adverse events. Therefore,
Volume 29 Number 12
Copyright 2003 Joint Commission on Accreditation of Healthcare Organizations
653
Joint Commission
Journal on Quality and Safety
Table 1. Characteristics of the Media Reports Studied* VA Medical Center Temple, TX Cleveland, OH Beckley, WV Long Beach, CA Long Beach, CA Denver, CO West LA, CA Chillicothe, OH Lexington, KY Canandaiqua, NY Providence, RI Bay Pines, FL Omaha, NE Columbia, MO Boston, MA Miami, FL Little Rock, AR Atlanta, GA San Antonio, TX Omaha, NE Miami, FL Miami, FL Gainesville, FL Muskogee, OK Montrose, NY Philadelphia, PA Denver, CO Little Rock, AR Columbia, SC Battle Creek, MI Northampton, MA Bay Pines, FL Togus, ME Togus, ME Togus, ME
Date of Incident 10/1/93 8/90–1/91 1/91–5/93 1/1/87 7/16/91 1/60–1/85 2/7/95 5/1/90 11/92–1/93 4/1/97 3/1/97 8/13/94 8/3/93 12/1/92 3/1/96 6/22/96 3/13/96 1/23/97 11/1/96 10/1/93 1/94–1/97 12/1/96 6/1/93 2/1/94 1/96–1/99 1/95–1/97 11/1/96 3/1/98 7/1/91 10/95–2/96 1/95–2/96 9/3/98 8/1/97 3/1/98 1/1/99
Adverse Event Description Contaminated oxygen—3 deaths Unnecessary surgery Misdiagnosis resulting in death Failed to grant timely disability Operating room error death Failed to coordinate care Scalding death Failed to protect community Negligent care Suspicious death Suspicious death Misdiagnosis resulting in death Adverse drug events—2 deaths 45 unexplained deaths Blood transfusion death Bleeding death Restraint death Scalding death Blood transfusion death Wrong treatment resulting in death Leader in unusual deaths in Florida Leader in malpractice settlements Misdiagnosis resulting in death Murder by injection Substandard care Improper injections Failed to coordinate care Restraint death Restraint death Misdiagnosis resulting in disability 8 alleged murders Misdiagnosis resulting in death Adverse drug event—death Wrong surgery Lack of warning about side effects
Date of Report 11/9/93
Reference† 14
2/12/94
15
2/18/95 4/2/95 5/6/95 5/11/95 6/29/96 8/25/96
16 17 18 19 20 21
4/2/97
22
4/6/97
23
5/15/97
24
7/26/97 9/7/97 10/13/97 10/26/97
25 26 27 28
1/15/98
29
3/3/98 4/9/98 6/8/98 7/9/98 8/19/98 12/25/98 1/13/99
30 31 32 33 34 35 36
8/27/99
37
* VA, Veterans Administration. † See References, page658.
654
December 2003
Volume 29 Number 12
Copyright 2003 Joint Commission on Accreditation of Healthcare Organizations
Joint Commission
Journal on Quality and Safety
Changes in Enrollment Rates and Media Reports of Incidents
Proximity of the Dates of Publication and Adverse Event and Changes in Enrollment Because some published reports of adverse events were distal to their occurrence, we wanted to determine whether proximal reporting of adverse events was associated with greater enrollment losses. To examine the impact of the proximity of the occurrence of the adverse event to its report on enrollment rates, we compared one-year change in enrollment rates for adverse events that occurred within two years of the reporting thereof to change in enrollment rates for adverse events that occurred two to four years prior to their reporting. We therefore eliminated three reports that were published five or more years after the adverse events they reported occurred.
Results Figure 1. Standardized changes in enrollment rates are shown for facilities with media reports of incidents minus the standardized changes in enrollment rates for facilities without media reports of incidents for the year prior to publication of the article, the year after publication of the article, and three years after publication of the article.
we compared standardized change in enrollment rates for both categories of veterans at facilities that had newspaper coverage of an adverse event to rates in facilities that did not have such media coverage. To examine our hypothesis that disenrollment would follow publication of media reports, we calculated rates at three different time periods: ■ The year before publication of the newspaper report ■ One year after publication of the report ■ Three years after publication of the report Because the number of facilities with reports was so low, we identified quartiles of regionally adjusted change in enrollment rates. In the case of rare events, the population odds ratio (OR) provides a good approximation to the population relative risk.38 We therefore calculated the OR and 95% confidence intervals (CIs) for a facility with a media report of an adverse event having a standardized change in enrollment rate in the lowest quartile of all such calculated rates. Each newspaper report was treated as an independent event, even if a particular facility had multiple reports.
December 2003
Changes in Enrollment The year following publication of the report, the oneyear standardized change in enrollment rate for Category A veterans was 3.2% lower in facilities that had media coverage of an incident than for facilities that did not have incidents (Figure 1, above). Similarly, the change in enrollment rate after an incident for Category A veterans was 4.3% lower three years after the incident for facilities that had media coverage of incidents compared with those that did not. The differences were much more dramatic for Category C veterans (4.25% lower one year after the report and 14.7% lower three years after the report). We did not find evidence that these changes were the continuation of a preexisting disenrollment trend. The year before publication of newspaper reports of adverse events, facilities that had those reports had higher standardized rates of enrollment compared with facilities that did not have media coverage of adverse events: the one-year standardized change in enrollment the year before the report was 0.29% higher for Category A and 11.46% higher for Category C veterans in facilities that had media coverage of an incident than for facilities that did not have media coverage of an incident. Facilities that had media coverage of an adverse incident were 3.8 times more likely to have lower-thanaverage growth rates for all categories of patients in the year after the report (OR = 3.80, CI = 1.51–9.62) and were
Volume 29 Number 12
Copyright 2003 Joint Commission on Accreditation of Healthcare Organizations
655
Joint Commission
Journal on Quality and Safety
Changes in Growth of Enrollment Rates and Time Lag Between Occurrence and Reporting of Incidents
2.5 times more likely to be in the lowest quartile for growth for the years three years after the report (OR = 2.48, CI = 1.05–5.85).
disenrollment than those that were not. We found no indication of a preexisting trend in disenrollment before publication of the report. These results suggest that a causal relationship between bad press and patient disenrollment is possible and that the marketing and brand costs associated with adverse events could be substantial. The finding that patients with less reliance on VHA—Category C veterans—had much more dramatic disenrollment rates than patients with high reliance on the system suggests that persons with little reliance on a particular health care system may switch out of it should quality concerns arise. The time between the adverse event and its report in the media appeared to have less of a dramatic effect on the enrollment patterns of veterans who are likely to have greater reliance on the VHA system than those who are not. Enrollment patterns of the least reliant group appeared to be dramatically influenced by the proximity of the event to its report. From a management perspective, these findings suggest that the worst damage occurs quickly and that the less time between the event and media coverage, the greater the adverse impact on brand.
Proximity to Media Report
Limitations
For Category A veterans, the proximity of the adverse event to the report about it appears to be unrelated to standardized change in enrollment rates (Figure 2, above). For Category C veterans—who are less likely to be reliant on VHA care—media exposure proximal to the adverse event appeared to have a greater impact. Reports within one year of the incident show an 11% difference in growth rate between facilities with and without reports, whereas reports within two to four years of the incident were associated with no difference.
This study has several limitations. First, this was a retrospective, observational study: The findings are associative, not causative. It is possible that other factors caused the disenrollment seen in facilities with media reports of adverse events. However, our ability to consider prior trends and to adjust for regional differences in growth rates as well as the consistent differences, albeit of differing magnitude, across different classes of patients warrant further evaluation of the proposed relationship. In addition, our analytic techniques identified only particularly egregious adverse events, several of them criminal. The same findings may not generalize to adverse events of a less dramatic sort. Second, we limited our analysis to newspaper publication of adverse events. Most patients who use VHA are likely to obtain information through other media, as well, such as television, radio, and the Internet. Although newspaper reports represent a critical aspect of information transfer and may facilitate dispersion of
Figure 1. The relationship between time lag between the occurrence and reporting of incidents and the standardized change in enrollment rates for facilities that had media coverage of an adverse incident is shown.
Conclusions We examined the relationship between media reporting of adverse events and subsequent disenrollment of patients in the VHA system and found evidence that, after correcting for regional differences in growth rates, VHA medical centers that were identified in published reports experienced a greater degree of
656
December 2003
Volume 29 Number 12
Copyright 2003 Joint Commission on Accreditation of Healthcare Organizations
Joint Commission
Journal on Quality and Safety
information to the general population, we were not able to assess the relationship between reports through other media and changes in enrollment rates. In addition, we were not able to distinguish the influence of news powerhouses—such as the Associated Press and Reuters—in more widely distributing these stories. It is probable that wider distribution of content across multiple media will have a more pronounced effect than a single print article. Finally, and most importantly, the VHA system has characteristics that may render the findings nongeneralizable to other health care systems. First, VHA treats a primarily male population that is older, poorer, more sick, and more likely to be disabled than the general population.39 Second, VHA obtains capitated reimbursement for services: Budgets are distributed based on the number of veterans who use the system, and the perenrollee allocation is designed to cover all provided health care services for one year. The large majority of health care in the United States is reimbursed under a fee-for-service mechanism: Reimbursement is based on each service provision. The fee-for-service system may produce bizarre financial incentives to provide unsafe care: Additional revenues might be generated through treatment of adverse events. Therefore, capitated systems such as VHA may have more incentives to provide safer care. Third, it may be more difficult for VHA to hide mistakes. Because it is a government agency, VHA is subject to oversight by a variety of government agencies and operates under the Freedom of Information Act (which allows for journalistic probing of operations) in addition to meeting the standard regulatory and accreditation requirements of the U.S. health care delivery industry. Finally, a powerful patient-centered lobby may lead VHA to be somewhat more patient oriented than other health care delivery systems. For these reasons, VHA might be more highly motivated to provide safer care than the general health care industry. Veterans may therefore react more strongly to press reports that challenge their perception of quality of care provided through VHA. On the other hand, because of their poverty and degree of illness, veterans may react less strongly to press reports that challenge quality perceptions: They simply may have no other place to go.
December 2003
Differences that we found when comparing Category A to Category C veterans suggest that the latter might be more representative of the general population—willing to leave care systems that have negative publicity relating to quality of care issues. Regardless of differences in the way health care is delivered in a particular system, patients who have a choice are likely to avoid a system if they find it provides substandard care. Clearly, additional studies of nonfederal health care delivery systems operating under a fee-for-service mechanism are required to resolve these issues.
Implications Despite these limitations, our results have implications for managers. First, managers might anticipate disenrollment of patients following an egregious adverse event. Such disenrollment is likely to occur even if media coverage of the event is delayed. More broadly, perceived quality is likely to influence patient enrollment decisions. Therefore, when considering costs associated with patient safety improvement, managers should consider the costs associated with damage to brand image that might result from patient safety violations. Managers of health care systems have a moral imperative to provide safe care for their enrolled patient populations. Our findings suggest that there may be a strong business case for providing safe care as well. As managers consider more of the downstream costs associated with adverse medical events and the long-term ramifications of providing unsafe care, the investments required to ensure safe care may become easier and easier to justify. J
William B. Weeks, M.D., M.B.A, C.H.E., is Senior Scholar, Veterans Administration National Quality Scholars Fellowship Program; Director, Field Office of the Veterans Health Administration (VHA) National Center for Patient Safety (NCPS), White River Junction, Vermont; and a member of the Joint Commission Journal on Quality and Safety's Editorial Advisory Board. Peter Mills, Ph.D., M.S., is the Associate Director, Field Office, VHA NCPS. The views expressed in this article do not necessarily represent the views of the Department of Veterans Affairs or of the United States government. Please address requests for reprints to William B. Weeks, M.D., M.B.A., C.H.E, [email protected].
Volume 29 Number 12
Copyright 2003 Joint Commission on Accreditation of Healthcare Organizations
657
Joint Commission
Journal on Quality and Safety
References 1. Brennan T.A., et al.: Incidence of adverse events and negligence in hospitalized patients. Results of the Harvard Medical Practice Study I. N Engl J Med 324:370–376, Feb. 7, 1991. 2. Institute of Medicine: To Err Is Human: Building a Safer Health System. Washington, DC: National Academy Press, 2000. 3. Leape L.L., et al.: The nature of adverse events in hospitalized patients. Results of the Harvard Medical Practice Study II. N Engl J Med 324:377–384, Feb. 7, 1991. 4. Leape L.: Error in medicine. JAMA 272:1851–1857, Dec. 21, 1994. 5. McDonald C., Weiner M., Hui S.: Deaths due to medical errors are exaggerated in Institute of Medicine report. JAMA 284:93–95, Jul. 5, 2000. 6. Leape L.: Institute of Medicine medical error figures are not exaggerated. JAMA 84:95–97, Jul. 5, 2000. 7. Thomas E.J., et al.: Incidence and types of adverse events and negligent care in Utah and Colorado. Medical Care 38:261–271, Mar. 2000. 8. Bates D., et al.: The cost of adverse drug events in hospitalized patients. Adverse Drug Events Prevention Study Group. JAMA 277:307–311, Jan. 22, 1997. 9. Johnson J.A., Bootman J.L.: Drug-related morbidity and mortality: A cost-of-illness model. Arch Intern Med 155:1949–1956, Oct. 9, 1995. 10. Thomas E.J.: Costs of medical injuries in Utah and Colorado. Inquiry 36:255–264, fall 1999. 11. Brahams D.: Medical errors: A costs burden on society. Med Leg J 68(A1):1–2, 2000. 12. Weeks W., et al.: The organizational costs of preventable medical errors. Jt Comm J Qual Improv 27:533–539, Oct. 2001. 13. LexisNexisTM at http://www.nexis.com/research (accessed Aug. 21, 2002). 14. McQuiston J.: Patient dies after falling into a coma at a hospital. The New York Times. Nov. 9, 1993, p.B7. 15. VA malpractice settlement ok’d. The Charleston Gazette, Feb. 12, 1994, p. B1. 16. Helfand D.: Ex-Marine sues U.S. over 8-year medical ordeal. Los Angeles Times, Feb. 18, 1995, p. B7. 17. Cekola A.: War’s aftermath: U.S. concedes liability in Vietnam hero’s death from medical tests but fights family’s claim. Los Angeles Times, Apr. 2, 1995, p. A3. 18. Chronis P.: U.S. judge awards vet $4.6 million. The Denver Post, May 6, 1995, p. B2. 19. Shuit D.: VA cites several errors in patient’s scalding death. Los Angeles Times, May 11, 1995, p. B1.
658
December 2003
20. Hall G.: Judge criticizes lawyers. He cites violation of ethics. Cincinnati Enquirer, Jun. 29, 1996, p. B. 21. Kaufman B.: Court says family can pursue malpractice suit against VA. The Cincinnati Enquirer, Aug. 25, 1996, p. B4. 22. Dahl D.: Suspicious VA deaths investigated. St. Petersburg Times, Apr. 2, 1997, p. A7. 23. Dahl D.: Two more VA center deaths revealed. St. Petersburg Times, Apr. 6, 1997, p. A1. 24. Dahl D.: Special Report on VA Medical Centers/Fatal Mistakes. St. Petersburg Times. May 15, 1997, p. A1. 25. Ruggles R.: Omahan's estate wins wrongful-death lawsuit. Cancer victim had chance of a cure if VA Medical Center had used chemotherapy, a federal judge rules. Omaha World-Herald, Jul. 26, 1997, p. A9. 26. Dahl D.: Deaths haunt Miami VA hospital. St. Petersburg Times, Sep. 7, 1997, p. B1. 27. Dahl D.: VA hospital leads nation in settlements. St. Petersburg Times, Oct. 13, 1997, p. B1. 28. Dahl D.: Deaths, federal probe surface at VA center. St. Petersburg Times, Oct. 26, 1997, p. 1A. 29. Dahl D.: VA doctor is changed with murder. St. Petersburg Times, Jan. 15, 1998, p. 1A. 30. Dahl D.: Whistleblower: VA injections were improper. St. Petersburg Times, Mar. 3, 1998, p. 1A. 31. Chronis P.: Veteran’s family files suit: Improper treatment by hospital alleged. The Denver Post, Apr. 19, 1998, p. B2. 32. Dahl D.: Coverup charged in VA deaths. Press Journal. Vero Beach FL, Jun. 8, 1998, p. A12. 33. Dahl D.: Patient restraints linked to deaths at VA hospitals. St. Petersburg Times, Jul. 9, 1998, p. A1. 34. Singhania L.: Vet sues VA for $3 million, alleges malpractice after leg amputation. Associated Press. Grand Rapids, MI, Aug. 19, 1998. 35. Associated Press: Son claims VA hospital at fault in eighth death. State and regional news. Springfield, MA, Dec. 25, 1999. 36. Ballingrud D.: Veteran did not get proper care. St. Petersburg Times, Jan. 13, 1999, p. A1. 37. Associated Press: VA hospital facing 3rd malpractice suit this year. State and regional news. Augusta, ME, Aug. 27, 1999. 38. Daniel W.W.: Biostatistics: A Foundation for Analysis in the Health Sciences, 7th ed. New York: John Wiley and Sons, Inc., 2000. 39. Wilson N.J., Kizer K.W.: The VA health care system: An unrecognized national safety net. Health Aff (Millwood) 16:200–204, Jul.–Aug. 1997.
Volume 29 Number 12
Copyright 2003 Joint Commission on Accreditation of Healthcare Organizations
Journal on Quality and Safety
Patient Safety
Reduction in Patient Enrollment in the Veterans Health Administration After Media Coverage of Adverse Medical Events
William B. Weeks, M.D., M.B.A., C.H.E. Peter D. Mills, Ph.D., M.S.
nsuring patient safety is recognized as an increasingly important aspect of medical care. Studies have indicated that medical errors are common.1,2 Many medical errors appear to be preventable3 and can be attributed to systems issues.4 Although there has been legitimate debate about the size of the problem,5,6 there is little debate that health care should reduce the frequency of adverse events and medical errors. Errors in health care are associated with costs that patients tend to bear. These include significant iatrogenic morbidity and mortality,1,7 direct costs of additional required care,8,9 and indirect, long-term costs of lost income, increased disability rates, and increased burden on caregivers.10,11 Direct and disability costs associated with adverse medical events in the United States have been estimated to be $37.6 billion per year.10 Less is known about the costs that organizations bear when medical errors occur. We previously outlined a model of organizational costs associated with preventable medical errors, including legal, marketing, and operations costs.12 In that model, one hypothesized cost is lost revenue associated with poor brand image and subsequent patient disenrollment; however, no studies have evaluated whether such a relationship exists. The Veterans Health Administration (VHA) offers an ideal setting in which to determine whether such a relationship exists for three reasons. First, VHA is a national health care system that operates under a single
E
652
December 2003
Article-at-a-Glance Background: Health care organizations may experience costs associated with preventable adverse events in the form of poor brand image and subsequent patient disenrollment. A retrospective cohort design was used to determine whether media coverage of adverse events that occurred in Veterans Health Administration (VHA) hospitals was associated with subsequent veteran disenrollment. Methods: Twenty-four newspaper reports of medical adverse events that occurred between 1994 and 1999 within the VHA system were identified. Regionally adjusted changes in enrollment rates for VHA facilities that had reported adverse events were compared with those that had not one year before and one and three years after publication of the newspaper reports. Results: Facilities that had published reports of adverse events had lower enrollment rates after publication of the report for two groups of veterans. Conclusions: Within the VHA system, health care organizations involved in adverse events that generated publicity suffered a greater rate of patient disenrollment. If safe patient care practices can reduce adverse publicity, they may enhance corporate value by maintaining enrollment of the patient population.
Volume 29 Number 12
Copyright 2003 Joint Commission on Accreditation of Healthcare Organizations
Joint Commission
Journal on Quality and Safety
administration; all VHA medical centers are subject to the same enrollment policy. Second, because VHA medical centers are organized into regional Veterans’ Integrated Service Networks (VISNs) that develop regional strategies regarding enrollment, regional variation due to different regional enrollment patterns can be identified and corrected. Finally, VHA collects enrollment information on two categories of users that differ in their reliance on the VHA system. Therefore, one can test the hypothesis that increased reliance on a health care system may be associated with more tolerance for quality concerns. To determine whether there is an association between media-facilitated damage to brand image and patient disenrollment, we examined the relationship between media coverage of adverse events and veteran disenrollment in the VHA between 1994 and 1999. In particular, we examined relationships between the date of media reporting on the adverse event, veteran category, and veteran disenrollment, controlling for regional changes in enrollment over time.
Methods Identification of Newspaper Articles In August 2002 we searched a computerized database to identify reports of medical adverse events that occurred within the VHA system and were published between October 1, 1993, and September 30, 1999. The database collects local, national, and international newspapers from more than 30,000 news and business sources.13 To identify these adverse events, we searched news sources using the following search terms: fatal mistakes, malpractice, negligence, wrongful death, veteran care, VA hospital, and Veterans Affairs. After eliminating duplicate articles, we identified 24 separate media reports of 36 adverse medical events that were published in the specified time frame.14–37 For each media report, we identified the paper in which the report was published and the date of publication. From the text of the article, we identified the nature of the adverse event, the date of the adverse event, and the VHA medical center in which the adverse event occurred (Table 1, page 654).
Changes in Patient Enrollment From the VHA centralized administrative database, we obtained the number of patients seen each year at
December 2003
each VHA facility between 1994 and 2001 for two groups of veterans. Category A veterans have received disability ratings for their service or have low incomes; they are likely to be more dependent on the VHA health care system than Category C veterans, who are not receiving disability payments and who have higher incomes. To compare regionally adjusted differences in enrollment, we first calculated the rate of change in enrollment for each patient category at each facility for one and three years. For example, the three-year change in enrollment for a facility in 1994 would be the number of veterans enrolled in 1997 minus the number of veterans enrolled in 1994 divided by the number of veterans in 1994. We performed the same calculations for each VISN. We then generated a standardized change in enrollment rate for each facility by dividing the change in enrollment rate for each facility within a geographic VISN by the VISN change in enrollment rate and represented these growth rates as a deviation from one. For example, the standardized three-year change in enrollment rate for a facility is calculated as follows: 1+ ((Facility enrollment in 1997 – Facility enrollment in 1994) / (Facility enrollment in 1994)) 1+ ((VISN enrollment in 1997 – VISN enrollment in 1994)/ (VISN enrollment in 1994))
The standardized change in enrollment rate for a facility represents that facility’s growth, relative to the growth of all the facilities in the region. If a facility’s standardized change in enrollment rate is 1.0, enrollment at the facility is changing at the same rate as enrollment in the rest of the region; if it is less than 1.0, change in enrollment is lower than in the rest of the region, and if greater than 1.0, change in enrollment is faster than in the rest of the region. We calculated one-year and threeyear standardized change in enrollment rates for each category of patient (Category A and Category C veterans) at each facility, for each year of the study.
Relationship Between Reports of Adverse Events in the Media and Changes in Enrollment We were interested in examining the relationship between newspaper media coverage of an adverse event and changes in patient enrollment. We hypothesized that a relative decline in enrollment would follow publication of newspaper reports about adverse events. Therefore,
Volume 29 Number 12
Copyright 2003 Joint Commission on Accreditation of Healthcare Organizations
653
Joint Commission
Journal on Quality and Safety
Table 1. Characteristics of the Media Reports Studied* VA Medical Center Temple, TX Cleveland, OH Beckley, WV Long Beach, CA Long Beach, CA Denver, CO West LA, CA Chillicothe, OH Lexington, KY Canandaiqua, NY Providence, RI Bay Pines, FL Omaha, NE Columbia, MO Boston, MA Miami, FL Little Rock, AR Atlanta, GA San Antonio, TX Omaha, NE Miami, FL Miami, FL Gainesville, FL Muskogee, OK Montrose, NY Philadelphia, PA Denver, CO Little Rock, AR Columbia, SC Battle Creek, MI Northampton, MA Bay Pines, FL Togus, ME Togus, ME Togus, ME
Date of Incident 10/1/93 8/90–1/91 1/91–5/93 1/1/87 7/16/91 1/60–1/85 2/7/95 5/1/90 11/92–1/93 4/1/97 3/1/97 8/13/94 8/3/93 12/1/92 3/1/96 6/22/96 3/13/96 1/23/97 11/1/96 10/1/93 1/94–1/97 12/1/96 6/1/93 2/1/94 1/96–1/99 1/95–1/97 11/1/96 3/1/98 7/1/91 10/95–2/96 1/95–2/96 9/3/98 8/1/97 3/1/98 1/1/99
Adverse Event Description Contaminated oxygen—3 deaths Unnecessary surgery Misdiagnosis resulting in death Failed to grant timely disability Operating room error death Failed to coordinate care Scalding death Failed to protect community Negligent care Suspicious death Suspicious death Misdiagnosis resulting in death Adverse drug events—2 deaths 45 unexplained deaths Blood transfusion death Bleeding death Restraint death Scalding death Blood transfusion death Wrong treatment resulting in death Leader in unusual deaths in Florida Leader in malpractice settlements Misdiagnosis resulting in death Murder by injection Substandard care Improper injections Failed to coordinate care Restraint death Restraint death Misdiagnosis resulting in disability 8 alleged murders Misdiagnosis resulting in death Adverse drug event—death Wrong surgery Lack of warning about side effects
Date of Report 11/9/93
Reference† 14
2/12/94
15
2/18/95 4/2/95 5/6/95 5/11/95 6/29/96 8/25/96
16 17 18 19 20 21
4/2/97
22
4/6/97
23
5/15/97
24
7/26/97 9/7/97 10/13/97 10/26/97
25 26 27 28
1/15/98
29
3/3/98 4/9/98 6/8/98 7/9/98 8/19/98 12/25/98 1/13/99
30 31 32 33 34 35 36
8/27/99
37
* VA, Veterans Administration. † See References, page658.
654
December 2003
Volume 29 Number 12
Copyright 2003 Joint Commission on Accreditation of Healthcare Organizations
Joint Commission
Journal on Quality and Safety
Changes in Enrollment Rates and Media Reports of Incidents
Proximity of the Dates of Publication and Adverse Event and Changes in Enrollment Because some published reports of adverse events were distal to their occurrence, we wanted to determine whether proximal reporting of adverse events was associated with greater enrollment losses. To examine the impact of the proximity of the occurrence of the adverse event to its report on enrollment rates, we compared one-year change in enrollment rates for adverse events that occurred within two years of the reporting thereof to change in enrollment rates for adverse events that occurred two to four years prior to their reporting. We therefore eliminated three reports that were published five or more years after the adverse events they reported occurred.
Results Figure 1. Standardized changes in enrollment rates are shown for facilities with media reports of incidents minus the standardized changes in enrollment rates for facilities without media reports of incidents for the year prior to publication of the article, the year after publication of the article, and three years after publication of the article.
we compared standardized change in enrollment rates for both categories of veterans at facilities that had newspaper coverage of an adverse event to rates in facilities that did not have such media coverage. To examine our hypothesis that disenrollment would follow publication of media reports, we calculated rates at three different time periods: ■ The year before publication of the newspaper report ■ One year after publication of the report ■ Three years after publication of the report Because the number of facilities with reports was so low, we identified quartiles of regionally adjusted change in enrollment rates. In the case of rare events, the population odds ratio (OR) provides a good approximation to the population relative risk.38 We therefore calculated the OR and 95% confidence intervals (CIs) for a facility with a media report of an adverse event having a standardized change in enrollment rate in the lowest quartile of all such calculated rates. Each newspaper report was treated as an independent event, even if a particular facility had multiple reports.
December 2003
Changes in Enrollment The year following publication of the report, the oneyear standardized change in enrollment rate for Category A veterans was 3.2% lower in facilities that had media coverage of an incident than for facilities that did not have incidents (Figure 1, above). Similarly, the change in enrollment rate after an incident for Category A veterans was 4.3% lower three years after the incident for facilities that had media coverage of incidents compared with those that did not. The differences were much more dramatic for Category C veterans (4.25% lower one year after the report and 14.7% lower three years after the report). We did not find evidence that these changes were the continuation of a preexisting disenrollment trend. The year before publication of newspaper reports of adverse events, facilities that had those reports had higher standardized rates of enrollment compared with facilities that did not have media coverage of adverse events: the one-year standardized change in enrollment the year before the report was 0.29% higher for Category A and 11.46% higher for Category C veterans in facilities that had media coverage of an incident than for facilities that did not have media coverage of an incident. Facilities that had media coverage of an adverse incident were 3.8 times more likely to have lower-thanaverage growth rates for all categories of patients in the year after the report (OR = 3.80, CI = 1.51–9.62) and were
Volume 29 Number 12
Copyright 2003 Joint Commission on Accreditation of Healthcare Organizations
655
Joint Commission
Journal on Quality and Safety
Changes in Growth of Enrollment Rates and Time Lag Between Occurrence and Reporting of Incidents
2.5 times more likely to be in the lowest quartile for growth for the years three years after the report (OR = 2.48, CI = 1.05–5.85).
disenrollment than those that were not. We found no indication of a preexisting trend in disenrollment before publication of the report. These results suggest that a causal relationship between bad press and patient disenrollment is possible and that the marketing and brand costs associated with adverse events could be substantial. The finding that patients with less reliance on VHA—Category C veterans—had much more dramatic disenrollment rates than patients with high reliance on the system suggests that persons with little reliance on a particular health care system may switch out of it should quality concerns arise. The time between the adverse event and its report in the media appeared to have less of a dramatic effect on the enrollment patterns of veterans who are likely to have greater reliance on the VHA system than those who are not. Enrollment patterns of the least reliant group appeared to be dramatically influenced by the proximity of the event to its report. From a management perspective, these findings suggest that the worst damage occurs quickly and that the less time between the event and media coverage, the greater the adverse impact on brand.
Proximity to Media Report
Limitations
For Category A veterans, the proximity of the adverse event to the report about it appears to be unrelated to standardized change in enrollment rates (Figure 2, above). For Category C veterans—who are less likely to be reliant on VHA care—media exposure proximal to the adverse event appeared to have a greater impact. Reports within one year of the incident show an 11% difference in growth rate between facilities with and without reports, whereas reports within two to four years of the incident were associated with no difference.
This study has several limitations. First, this was a retrospective, observational study: The findings are associative, not causative. It is possible that other factors caused the disenrollment seen in facilities with media reports of adverse events. However, our ability to consider prior trends and to adjust for regional differences in growth rates as well as the consistent differences, albeit of differing magnitude, across different classes of patients warrant further evaluation of the proposed relationship. In addition, our analytic techniques identified only particularly egregious adverse events, several of them criminal. The same findings may not generalize to adverse events of a less dramatic sort. Second, we limited our analysis to newspaper publication of adverse events. Most patients who use VHA are likely to obtain information through other media, as well, such as television, radio, and the Internet. Although newspaper reports represent a critical aspect of information transfer and may facilitate dispersion of
Figure 1. The relationship between time lag between the occurrence and reporting of incidents and the standardized change in enrollment rates for facilities that had media coverage of an adverse incident is shown.
Conclusions We examined the relationship between media reporting of adverse events and subsequent disenrollment of patients in the VHA system and found evidence that, after correcting for regional differences in growth rates, VHA medical centers that were identified in published reports experienced a greater degree of
656
December 2003
Volume 29 Number 12
Copyright 2003 Joint Commission on Accreditation of Healthcare Organizations
Joint Commission
Journal on Quality and Safety
information to the general population, we were not able to assess the relationship between reports through other media and changes in enrollment rates. In addition, we were not able to distinguish the influence of news powerhouses—such as the Associated Press and Reuters—in more widely distributing these stories. It is probable that wider distribution of content across multiple media will have a more pronounced effect than a single print article. Finally, and most importantly, the VHA system has characteristics that may render the findings nongeneralizable to other health care systems. First, VHA treats a primarily male population that is older, poorer, more sick, and more likely to be disabled than the general population.39 Second, VHA obtains capitated reimbursement for services: Budgets are distributed based on the number of veterans who use the system, and the perenrollee allocation is designed to cover all provided health care services for one year. The large majority of health care in the United States is reimbursed under a fee-for-service mechanism: Reimbursement is based on each service provision. The fee-for-service system may produce bizarre financial incentives to provide unsafe care: Additional revenues might be generated through treatment of adverse events. Therefore, capitated systems such as VHA may have more incentives to provide safer care. Third, it may be more difficult for VHA to hide mistakes. Because it is a government agency, VHA is subject to oversight by a variety of government agencies and operates under the Freedom of Information Act (which allows for journalistic probing of operations) in addition to meeting the standard regulatory and accreditation requirements of the U.S. health care delivery industry. Finally, a powerful patient-centered lobby may lead VHA to be somewhat more patient oriented than other health care delivery systems. For these reasons, VHA might be more highly motivated to provide safer care than the general health care industry. Veterans may therefore react more strongly to press reports that challenge their perception of quality of care provided through VHA. On the other hand, because of their poverty and degree of illness, veterans may react less strongly to press reports that challenge quality perceptions: They simply may have no other place to go.
December 2003
Differences that we found when comparing Category A to Category C veterans suggest that the latter might be more representative of the general population—willing to leave care systems that have negative publicity relating to quality of care issues. Regardless of differences in the way health care is delivered in a particular system, patients who have a choice are likely to avoid a system if they find it provides substandard care. Clearly, additional studies of nonfederal health care delivery systems operating under a fee-for-service mechanism are required to resolve these issues.
Implications Despite these limitations, our results have implications for managers. First, managers might anticipate disenrollment of patients following an egregious adverse event. Such disenrollment is likely to occur even if media coverage of the event is delayed. More broadly, perceived quality is likely to influence patient enrollment decisions. Therefore, when considering costs associated with patient safety improvement, managers should consider the costs associated with damage to brand image that might result from patient safety violations. Managers of health care systems have a moral imperative to provide safe care for their enrolled patient populations. Our findings suggest that there may be a strong business case for providing safe care as well. As managers consider more of the downstream costs associated with adverse medical events and the long-term ramifications of providing unsafe care, the investments required to ensure safe care may become easier and easier to justify. J
William B. Weeks, M.D., M.B.A, C.H.E., is Senior Scholar, Veterans Administration National Quality Scholars Fellowship Program; Director, Field Office of the Veterans Health Administration (VHA) National Center for Patient Safety (NCPS), White River Junction, Vermont; and a member of the Joint Commission Journal on Quality and Safety's Editorial Advisory Board. Peter Mills, Ph.D., M.S., is the Associate Director, Field Office, VHA NCPS. The views expressed in this article do not necessarily represent the views of the Department of Veterans Affairs or of the United States government. Please address requests for reprints to William B. Weeks, M.D., M.B.A., C.H.E, [email protected].
Volume 29 Number 12
Copyright 2003 Joint Commission on Accreditation of Healthcare Organizations
657
Joint Commission
Journal on Quality and Safety
References 1. Brennan T.A., et al.: Incidence of adverse events and negligence in hospitalized patients. Results of the Harvard Medical Practice Study I. N Engl J Med 324:370–376, Feb. 7, 1991. 2. Institute of Medicine: To Err Is Human: Building a Safer Health System. Washington, DC: National Academy Press, 2000. 3. Leape L.L., et al.: The nature of adverse events in hospitalized patients. Results of the Harvard Medical Practice Study II. N Engl J Med 324:377–384, Feb. 7, 1991. 4. Leape L.: Error in medicine. JAMA 272:1851–1857, Dec. 21, 1994. 5. McDonald C., Weiner M., Hui S.: Deaths due to medical errors are exaggerated in Institute of Medicine report. JAMA 284:93–95, Jul. 5, 2000. 6. Leape L.: Institute of Medicine medical error figures are not exaggerated. JAMA 84:95–97, Jul. 5, 2000. 7. Thomas E.J., et al.: Incidence and types of adverse events and negligent care in Utah and Colorado. Medical Care 38:261–271, Mar. 2000. 8. Bates D., et al.: The cost of adverse drug events in hospitalized patients. Adverse Drug Events Prevention Study Group. JAMA 277:307–311, Jan. 22, 1997. 9. Johnson J.A., Bootman J.L.: Drug-related morbidity and mortality: A cost-of-illness model. Arch Intern Med 155:1949–1956, Oct. 9, 1995. 10. Thomas E.J.: Costs of medical injuries in Utah and Colorado. Inquiry 36:255–264, fall 1999. 11. Brahams D.: Medical errors: A costs burden on society. Med Leg J 68(A1):1–2, 2000. 12. Weeks W., et al.: The organizational costs of preventable medical errors. Jt Comm J Qual Improv 27:533–539, Oct. 2001. 13. LexisNexisTM at http://www.nexis.com/research (accessed Aug. 21, 2002). 14. McQuiston J.: Patient dies after falling into a coma at a hospital. The New York Times. Nov. 9, 1993, p.B7. 15. VA malpractice settlement ok’d. The Charleston Gazette, Feb. 12, 1994, p. B1. 16. Helfand D.: Ex-Marine sues U.S. over 8-year medical ordeal. Los Angeles Times, Feb. 18, 1995, p. B7. 17. Cekola A.: War’s aftermath: U.S. concedes liability in Vietnam hero’s death from medical tests but fights family’s claim. Los Angeles Times, Apr. 2, 1995, p. A3. 18. Chronis P.: U.S. judge awards vet $4.6 million. The Denver Post, May 6, 1995, p. B2. 19. Shuit D.: VA cites several errors in patient’s scalding death. Los Angeles Times, May 11, 1995, p. B1.
658
December 2003
20. Hall G.: Judge criticizes lawyers. He cites violation of ethics. Cincinnati Enquirer, Jun. 29, 1996, p. B. 21. Kaufman B.: Court says family can pursue malpractice suit against VA. The Cincinnati Enquirer, Aug. 25, 1996, p. B4. 22. Dahl D.: Suspicious VA deaths investigated. St. Petersburg Times, Apr. 2, 1997, p. A7. 23. Dahl D.: Two more VA center deaths revealed. St. Petersburg Times, Apr. 6, 1997, p. A1. 24. Dahl D.: Special Report on VA Medical Centers/Fatal Mistakes. St. Petersburg Times. May 15, 1997, p. A1. 25. Ruggles R.: Omahan's estate wins wrongful-death lawsuit. Cancer victim had chance of a cure if VA Medical Center had used chemotherapy, a federal judge rules. Omaha World-Herald, Jul. 26, 1997, p. A9. 26. Dahl D.: Deaths haunt Miami VA hospital. St. Petersburg Times, Sep. 7, 1997, p. B1. 27. Dahl D.: VA hospital leads nation in settlements. St. Petersburg Times, Oct. 13, 1997, p. B1. 28. Dahl D.: Deaths, federal probe surface at VA center. St. Petersburg Times, Oct. 26, 1997, p. 1A. 29. Dahl D.: VA doctor is changed with murder. St. Petersburg Times, Jan. 15, 1998, p. 1A. 30. Dahl D.: Whistleblower: VA injections were improper. St. Petersburg Times, Mar. 3, 1998, p. 1A. 31. Chronis P.: Veteran’s family files suit: Improper treatment by hospital alleged. The Denver Post, Apr. 19, 1998, p. B2. 32. Dahl D.: Coverup charged in VA deaths. Press Journal. Vero Beach FL, Jun. 8, 1998, p. A12. 33. Dahl D.: Patient restraints linked to deaths at VA hospitals. St. Petersburg Times, Jul. 9, 1998, p. A1. 34. Singhania L.: Vet sues VA for $3 million, alleges malpractice after leg amputation. Associated Press. Grand Rapids, MI, Aug. 19, 1998. 35. Associated Press: Son claims VA hospital at fault in eighth death. State and regional news. Springfield, MA, Dec. 25, 1999. 36. Ballingrud D.: Veteran did not get proper care. St. Petersburg Times, Jan. 13, 1999, p. A1. 37. Associated Press: VA hospital facing 3rd malpractice suit this year. State and regional news. Augusta, ME, Aug. 27, 1999. 38. Daniel W.W.: Biostatistics: A Foundation for Analysis in the Health Sciences, 7th ed. New York: John Wiley and Sons, Inc., 2000. 39. Wilson N.J., Kizer K.W.: The VA health care system: An unrecognized national safety net. Health Aff (Millwood) 16:200–204, Jul.–Aug. 1997.
Volume 29 Number 12
Copyright 2003 Joint Commission on Accreditation of Healthcare Organizations