Investigating Geographic Variation in Mortality ... - Wiley Online Library

American Journal of Transplantation 2012; 12: 1598–1602 Wiley Periodicals Inc.

C

Copyright 2012 The American Society of Transplantation and the American Society of Transplant Surgeons doi: 10.1111/j.1600-6143.2011.03981.x

Brief Communication

Investigating Geographic Variation in Mortality in the Context of Organ Donation E. Sheehya,∗ , K. J. O’Connorb , R. S. Luskina , R. J. Howardc , D. Cornellc , J. Finnd , T. Monee , F. W. Selckf and F. L. Delmonicog a

New England Organ Bank, Waltham, MA LifeCenter Northwest, Bellevue, WA c LifeQuest Organ Recovery Services, Gainesville, FL d Midwest Transplant Network, Westwood, KS e OneLegacy, Los Angeles, CA f Johns Hopkins Bloomberg School of Public Health, Baltimore, MD g New England Organ Bank and Massachusetts General Hospital, Boston, MA *Corresponding author: Ellen Sheehy, [email protected] b

Organ procurement organizations (OPOs) report a nearly fourfold difference in donor availability as measured by eligible deaths per million population (PMP) based on hospital referrals. We analyzed whether mortality data help explain geographic variation in organ supply as measured by the number of eligible deaths for organ donation. Using the 2007 National Center for Health Statistics’ mortality data, we analyzed deaths occurring in acute care hospitals, aged ≤ 70 years from cerebrovascular accidents and trauma. These deaths were mapped at the county level and compared to eligible deaths reported by OPOs. In 2007, there were 2 428 343 deaths reported in the United States with 42 339 in-hospital deaths ≤ 70 years from cerebrovascular accidents (CVA) or trauma that were correlated with eligible deaths PMP (r2 = 0.79.) Analysis revealed a broad range in the death rate across OPOs: trauma deaths: 44–118 PMP; deaths from CVA: 34–118 PMP; and combined CVA and trauma: 91–229 PMP. Mortality data demonstrate that deaths by neurologic criteria of people who are likely to be suitable deceased donors are not evenly distributed across the nation. These deaths are correlated with eligible deaths for organ donation. Regional availability of organs is affected by deaths which should be accounted for in the organ allocation system. Key words: Donation rate, donation service area, organ allocation, organ donation Abbreviations: CMS, Centers for Medicaid and Medicare Services; CVA, cerebrovascular accident; FIPS, Federal Information Processing Standard; ICD, Inter-

1598

national Statistical Classification of Disease; NCHS, National Center for Health Statistics; OPO, Organ Procurement Organization; OPTN, Organ Procurement and Transplantation Network; PMP, per million population; SRTR, Scientific Registry of Transplant Recipients. Received 02 August 2011, revised 02 December 2011 and accepted for publication 26 December 2011

Introduction Waiting times for organs vary dramatically across the United States. Factors affecting local waiting times include incidence and prevalence of end-stage organ disease, organ supply and clinical practice related to patient listing and organ acceptance. Geographic variation in underlying organ supply has been reported by Organ Procurement Organizations (OPOs) based on referrals of patient deaths from hospitals. Organ supply is a function of many variables including causes of death, health status of the population, death rate and OPO efficiency. Since 2002, OPOs have been required by the Centers for Medicaid and Medicare Services (CMS) to report data to the Organ Procurement and Transplantation Network (OPTN) on the number of referrals of eligible deaths for organ donation by each hospital in the OPO’s geographic service area with the aim of measuring OPO efficiency in converting potential to actual donors. An eligible death for organ donation is reportable by the OPO to the OPTN if the death is of a person aged ≤70 years who is legally declared brain dead, without any absolute contraindication(s) to organ donation as defined by a standardized list of conditions from CMS such as those for metastatic cancer and seropositivity for human immunodeficiency virus. Recovered organ donors include some patients who do not meet the criteria of “eligible death for organ donation” including those over the age of 70 years and donors after circulation death (not declared brain dead). Defining eligible death for organ donation was a deliberate and collaborative process which required consideration and compromise due to state and regional differences in brain death laws as well as clinical practice related to organ acceptance for transplantation. OPO staff received training, including case studies, to better ensure consistent reporting of eligible deaths across service areas. The number

Geographic Variation in Mortality

of eligible deaths per million population (PMP) reported by each OPO has consistently shown wide variation. For 2009, the number of eligible deaths PMP at the OPO level ranged from a low of 15.7 to a high of 61.1 with a national mean of 31.7 eligible deaths PMP. The variation observed in eligible deaths PMP in 2009 was seen in all previous years for which the data were reported (2002–2008). The more than threefold difference in number of eligible deaths for organ donation as reported by OPOs mirrors the variation previously seen in brain dead organ donor potential as assessed by hospital chart review (1). Explanations offered for the variation include underlying differences in service area demographics and death rates, health status, access to treatment, clinical practice and OPO reporting practices (2–9). To date, there has been no systematic, data-based analysis of reasons for the difference in population-adjusted number of eligible deaths across the United States. This study investigates whether mortality data can help to explain the variation in underlying supply of organs for transplantation across US OPO service areas.

Methods

not included in this data file, the OPO serving Puerto Rico was excluded from our analysis leaving 57 OPOs for the analysis covering the 50 US States. Thirty-six of the 3141 counties were shared by two OPOs. For these counties, the associated deaths were divided evenly between the two OPOs. Although our study focused on 2005 and 2007 data, we also examined 2001 data from the NCHS Multiple Cause of Death Mortality file to determine whether the earlier time period would produce similar regional differences in death rates. CVA and trauma Deaths were calculated for 2001, 2005 and 2007 for OPOs and rankings were compared to assess whether organizations’ rankings were similar across the three time periods.

Eligible death for organ donation The term “eligible death” and its definition were codified and adopted by the OPTN (a person aged ≤70 years declared brain dead, without any absolute contraindication(s) to organ donation as defined by a standardized list of conditions from CMS.) Training on the definition and reporting according to the definition was provided to OPOs. OPOs began submitting eligible death data to the OPTN on a monthly basis in 2002. The data are reviewed and published every 6 months at www.srtr.org. Eligible death data for 2005 and 2007 were compared with NCHS mortality data from the same years. Eligible death data for our study were obtained from the Scientific Registry of Transplant Recipients website (accessed at www.srtr.org, OPO Data, Table 3).

Mortality data The study sample consisted of all reported, in-hospital deaths in the United States. Information on the decedent’s age, underlying cause of death (coded from the Manual of the International Statistical Classification of Diseases, 10th Revision (ICD-10) (9)), death location (hospital, nursing home, residence, etc.), county and state were obtained from the National Center for Health Statistics’ (NCHS) Multiple Cause of Death Mortality file (10). This database contains comprehensive death data for all US deaths and is compiled from death certificates. The 2007 US mortality file, the latest data available from the National Center for Health Statistics, contained approximately 2.4 million deaths. Our study examined only deaths of patients 70 years and younger occurring in acute-care hospitals with a cause of death of cerebrovascular accident (CVA) or trauma to align with the causes of death of most brain-dead organ donors. This approach mirrors the methodology of Ojo et al. (11) who used the same NCHS mortality data files for the 1989–1993 time period and determined that 88.5% of brain-dead donors died from causes of deaths captured by CVA or trauma: stroke, motor vehicle accident, homicide (gunshot or stabbing) or head trauma (11). Based on the reported ICD-10 code(s) from the death certificate, NCHS divides deaths into 42 discrete disease categories. We limited our analysis to deaths ≤70 years old occurring in the hospital from either CVA or trauma. CVA deaths were drawn exclusively from one of the 42 categories, “cerebrovascular diseases” and trauma deaths resulted from combining five categories: motor vehicle accidents, assault, intentional self-harm, all other and unspecified accidents and adverse effects, and all other external causes. Patients dying from CVA or trauma who had additional causes of death listed which represented contraindications to organ donation, in accordance with the definition of eligible death, were eliminated from our study. Each of the 58 OPOs is federally designated to serve hospitals within a given set of counties. Using the list of counties served by each OPO and the FIPS code—Federal Information Processing Standard—contained in the NCHS data file, we aggregated CVA and Trauma deaths at the county level to the service area of each OPO. Because deaths from Puerto Rico are

American Journal of Transplantation 2012; 12: 1598–1602

Death rates per million population To permit comparisons across OPOs, we standardized all in-hospital deaths ≤70 years by million population since OPOs are diverse in the number of people and hospitals that they serve. US Census Bureau data by county were used to determine the population of each OPO service area. The number of deaths from CVA and trauma (assembled from the NCHS data base as described above) as well as the eligible deaths for each OPO service area (reported by the OPOs to the OPTN) were divided by the number of live population in the service area to determine the rate of CVA and trauma deaths per million population and eligible deaths per million population within each OPO. Deaths per million population (≤ 70 years old and occurring in hospital from a cerebrovascular accident or trauma) were mapped to display geographic differences in death rates across the United States. Eligible death data were compared with CVA and trauma PMP by OPO as well.

Statistical analysis All data analysis was conducted using Stata version 10 (Stata Corp., College Station, TX, USA). Statistical tests were performed to evaluate associations and differences in ranking. Associations between variables were evaluated using a coefficient of determination and reported using the r-square test. OPO ranking changes between time periods were tested using Spearman’s rank correlation.

Results Composition of deaths There was significant variation across OPO service areas in the composition of inpatient deaths according to age and cause of death. In 2007 there were 2 428 343 deaths in the United States with 873 589 occurring in hospitals. By eliminating deaths over the age of 70 years, those with causes that prevent organ donation (e.g. metastatic cancer) and selecting deaths from CVA and trauma (i.e. causes that are most often associated with brain death), 42 339 (4.8% of 2007 in-hospital deaths) remained. These deaths were 1599

Sheehy et al.

split between CVA (19 005) and trauma (23 334.) Analysis of the 2001 data yields similar results with 2 416 425 total deaths, 963 725 in-hospital deaths and 41 393 deaths (4.3% of 2001 in-hospital deaths) from causes that could lead to organ donation without absolute contraindication. Total deaths during this time period remained stable (1% increase from 2001 to 2005 and 0.8% decrease from 2005 to 2007).

Differences in death rates across OPOs An examination of national mortality data reveals that death rates vary across the United States among OPO service areas. On a per million live population basis, there was nearly a threefold variation on each of the following measures: inhospital deaths, deaths of individuals aged 70 years and younger, deaths from traumatic causes and deaths from cerebrovascular accidents (see Table 1). To understand whether there are geographic patterns to the distribution of CVA and trauma deaths, we constructed a map of the 57 organizations classified into quartiles of CVA and trauma deaths per million population. The result is a striking picture of areas of the country with high death rates, most notably the Southeast, and those with much lower death rates, the Northeast and Upper Midwest (Figure 1).

Table 1: Range of in-hospital deaths per million population across donation service areas (2007)

High Third quartile Median First quartile Low

In-hospital deaths

In-hospital ≤ 70 years old

In-hospital, ≤70 years from CVA and trauma

4215 3513 3050 2563 1439

1711 1314 1106 949 616

229 175 146 120 91

Stability of OPO rankings from 2001 to 2007 The 2001, 2005 and 2007 data indicate that OPOs with more CVA and trauma deaths in 2007 also had high rates of CVA and trauma deaths in 2001 and 2005. OPO rankings with respect to CVA and trauma deaths failed to change significantly between 2001 and 2005 (rho = 0.931; p < 0.001) or from 2005 to 2007 (rho = 0.940; p < 0.001). Comparison of eligible deaths and donors per million population versus per CVA and trauma deaths 2007 NCHS data reveal that CVA and trauma deaths per million population ranged from 91 to 229 across OPO service areas with a national mean of 144. Eligible deaths

Figure 1: Variation in CVA and trauma death rate by donation service area.

1600


Geographic Variation in Mortality 70 60 50

Eligible Deaths PMP

40 30 20 10 0 0

50

100

150

200

250

eligible deaths could explain some of the difference in the rates of eligible deaths PMP across OPO service areas. The definition, however, is clear and easy to apply in that it requires legal declaration of brain death and includes a short list of medical conditions which eliminate a hospitalreported death as an eligible death. In addition, several national trainings for OPO personnel were held to ensure consistent application of the definition of eligible death and individual OPO data are available online for inspection and review.

CVA and Trauma Deaths PMP R-square = 0.79 Figure 2: 2007 CVA and trauma death rates versus eligible death rates.

PMP ranged from 15.6 to 59.3 with a national mean of 34.8. Deaths PMP from trauma and CVAs were correlated with reported eligible deaths PMP by OPO, with an r-square of 0.79. (Figure 2) This result suggests that mortality data alone can explain 79% of the difference observed and reported in eligible deaths across OPO service areas. The correlation between eligible deaths per CVA and trauma deaths and eligible deaths PMP was not strong (r-square = 0.51) suggesting that living population is not a good indication of organ donor potential.

Discussion The regional differences in organ donor potential which persist over time reflect the broader regional disparities observed in healthcare. Geographic differences in incidence and prevalence of diseases such as cancer and stroke are well documented along with regional variation in motor vehicle fatalities, obesity and tobacco use (6–8). How people die as well as their health status at the time of death determines suitability for organ donation. In addition, clinical practice surrounding trauma care, declaration of brain death and end-of-life care can influence whether or not patients qualify as suitable organ donors (2). Our findings are consistent with previously published studies (1,3,12–15) and suggest that using living population to estimate organ donor potential should be reevaluated. Differences in organ availability by region appear to be due at least in part to varying death rates of patients 70 years and younger from CVA and traumatic causes. Allocating organs based on the assumption that organ potential is the same across allocation regions should be questioned. Differences in OPO performance may exacerbate regional differences in patient waiting time for transplant. One potential limitation of our study is the reliance on OPOreported data for eligible deaths. OPOs base their reports to the OPTN on the referrals that they receive from hospitals and OPO-specific methods for counting and reporting American Journal of Transplantation 2012; 12: 1598–1602

National mortality data are based on individual death certificates and subject to criticism for the lack of uniformity in coding cause of death across areas, clinicians and institutions (16). Efforts have been made by the federal government to improve education and feedback to clinicians and institutions to improve the accuracy of death certificate data (17). Despite potential for qualitative inconsistencies in reporting, the number of deaths is likely to be reasonably accurate as death certificates are essential for survivors to settle decedent estates and other legal matters and federal requirements mandate that states report all deaths. Finally, the subset of CVA and trauma deaths we studied is broader than the pool of eligible deaths reported by OPOs. NCHS mortality data are derived from hospital death certificates and have no information as to the ventilatory status of patients during the hospitalization. Therefore, we were unable to limit our analysis to ventilated patients. All conditions or diseases that would exclude patients as suitable organ donors may not be captured on the death certificate. Due to the inability to exclude nonventilated patients and those with medical conditions contraindicated for organ donation, our study sample includes nearly four times as many deaths as are reported as eligible deaths for organ donation by OPOs. Future studies could refine the methodology to identify and quantify ICD-10 codes of actual donors and exclude ICD-10 cause of death that are not associated with organ donors based on national experience. One interesting future study would be to collect the ICD-10 codes for eligible deaths and organ donors from OPO records and establish the list of inclusion codes based on actual OPO data. The variation observed in death rates across US regions, however, is unlikely explained by differences in death reporting or variation in death certificate completion alone. Not only does our study confirm that living population is not a good indicator of organ availability, but it also calls for a reevaluation of the current system of organ allocation which is based on an assumption that underlying supply of organ donors is similar across the country. Organ allocation in the United States favors the use of recovered organs first within an OPO’s service area and next within the region. This system is justified by shorter ischemic times for the organs as well as by encouraging and reinforcing the community’s commitment to donation by transplanting 1601

Sheehy et al.

local residents. Unfortunately, given substantial differences in underlying organ donor potential by OPO service area coupled with widely varying and somewhat arbitrary OPO service areas, the result of the current allocation system is uneven waiting times for organs across the United States. More research is needed to understand the multiple and complex reasons for discrepancies in waiting times, but our study finds evidence to suggest some of the variation is due to variation in mortality by region.

Conclusion New organ allocation approaches must be considered to equalize the differences in underlying organ supply to ensure that access to transplantation is equitable. To achieve the goal of equal access to transplantation for all Americans, it is imperative that the proposals currently being discussed for a new national kidney allocation system (18,19) are modified to acknowledge and account for geographic disparities in organ supply and that the existing allocation policy for liver transplantation account for organ availability by eligible deaths as well (20).

Disclosure E.S. and F.S. have had full access to the data and take responsibility for integrity of the data and the accuracy of the data analysis. Analyses, interpretations and conclusions are those of the authors, not the National Center for Health Statistics. The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.

References 1. Sheehy E, Conrad SL, Brigham LE, et al. Estimating the number of potential organ donors in the United States. N Engl J Med 2003; 349: 667–674. 2. Prendergast TJ, Claessens MT, Luce JM. A national survey of end-of-life care for critically ill patients. Am J Respir Crit Care Med 1998; 158:1163–1167. 3. Hauptman PJ, O’Connor KJ. Procurement and allocation of solid organs for transplantation. N Engl J Med 1997; 336: 422–431.

1602

4. Gortmaker SL, Beasley CL, Brigham LE, et al. Organ donor potential and performance: Size and nature of the organ donor shortfall. Crit Care Med 1996; 24: 432–439. 5. Luskin RS, Buckley CA, Bradley JW, et al. An alternative approach to evaluating organ procurement organization performance. Transplant Proc 1999; 31: 353–355. 6. Devesa SS, Grauman DJ, Blot WJ, et al. Atlas of Cancer Mortality in the United States, 1950–1994. NIH Publication No. 99-4564, 1999. 7. Baker SP, Whitfield RA, O’Neill B. Geographic variations in mortality from motor vehicle crashes. N Engl J Med 1987; 316: 1384– 1387. 8. State-specific prevalence and trends in adult cigarette smoking— United States,1998–2007, Morb Mortal Wkly Rep 2009; 58: 221– 226. 9. The international classification of diseases, 10th revision. Commission on Professional and Hospital Activities, Ann Arbor. 1993. 10. National Center for Health Statistics. Detailed Mortality File, 1999– 2007 (machine readable data file and documentation, CD-ROM Series 20, No. 2K). Hyattsville, Maryland 2010. 11. Ojo AO, Wolfe RA, Leichtman AB et al. A practical approach to evaluate the potential donor pool and trends in cadaveric kidney donation. Transplantation 1999; 67: 548–556. 12. Evan RW, Orians CE, Ascher NL. The potential supply of organ donors: An assessment of the efficacy of organ procurement efforts in the United States. JAMA 1992; 267: 239–246. 13. Organ Procurement Organziations. Alternatives being developed to more accurately assess performance. Washington, DC: General Accounting Office, November 1997. (GAO/HEHS98-26.) 14. Luskin R, Delmonico FL. Assessing organ donation from the dead should not be done by reporting a census of the living. Am J Transplant 2003; 3: 1185–1187. 15. United Network for Organ Sharing data on organ donors and transplants. Available at: http://optn.transplant.hrsa.gov/latestData/ rptData.asp. Accessed on January 18, 2011. 16. Carter JR. The problematic death certificate. N Engl J Med 1985; 313: 1285–1286. 17. Rosenberg HM. Improving cause of death statistics. Am J Publ Health 1989; 79: 563–564. 18. Hippen BE, Thistlethwaite R, Ross LF. Risk, prognosis, and unintended consequences in kidney allocation. Available at: nejm.org: 10.1056/NEJMp1102583. Accessed on April 4, 2011. 19. Leichtman AB, McCullough KP, Wolfe RW. Improving the allocation system for deceased-donor kidneys. Available at: nejm.org: 10.1056/NEJMp1102728. Accessed on April 4, 2011. 20. Yeh H, Smoot E, Schoenfeld DA, Markmann JF. Geographic inequity in access to livers for transplantation. Transplantation 2011; 91: 479–786.
