The
n e w e ng l a n d j o u r na l
of
m e dic i n e
original article
Racial Variation in Medical Outcomes among Living Kidney Donors Krista L. Lentine, M.D., Mark A. Schnitzler, Ph.D., Huiling Xiao, M.S., Georges Saab, M.D., Paolo R. Salvalaggio, M.D., Ph.D., David Axelrod, M.D., Connie L. Davis, M.D., Kevin C. Abbott, M.D., M.P.H., and Daniel C. Brennan, M.D.
A bs t r ac t Background From the Center for Outcomes Research (K.L.L., M.A.S., H.X.) and the Division of Nephrology (K.L.L.), Saint Louis University School of Medicine; and the Division of Nephrology, Washington University School of Medicine (G.S., D.C.B.) — both in St. Louis; the Kidney and Pancreas Transplant Program, University of Washington, Seattle (P.R.S., C.L.D.); the Department of Surgery, Dartmouth–Hitchcock Medical Center, Hanover, NH (D.A.); and the Nephrology Service, Walter Reed Army Medical Center, Washington, DC (K.C.A.). Address reprint requests to Dr. Lentine at Saint Louis University Center for Outcomes Research, Salus Center, 4th Fl., 3545 Lafayette Ave., St. Louis, MO 63104, or at
[email protected]. N Engl J Med 2010;363:724-32.
Data regarding health outcomes among living kidney donors are lacking, especially among nonwhite persons. Methods
We linked identifiers from the Organ Procurement and Transplantation Network (OPTN) with administrative data of a private U.S. health insurer and performed a retrospective study of 4650 persons who had been living kidney donors from October 1987 through July 2007 and who had post-donation nephrectomy benefits with this insurer at some point from 2000 through 2007. We ascertained post-nephrectomy medical diagnoses and conditions requiring medical treatment from billing claims. Cox regression analyses with left and right censoring to account for observed periods of insurance benefits were used to estimate absolute prevalence and prevalence ratios for diagnoses after nephrectomy. We then compared prevalence patterns with those in the 2005–2006 National Health and Nutrition Examination Survey (NHANES) for the general population.
Copyright © 2010 Massachusetts Medical Society.
Results
Among the donors, 76.3% were white, 13.1% black, 8.2% Hispanic, and 2.4% another race or ethnic group. The median time from donation to the end of insurance benefits was 7.7 years. After kidney donation, black donors, as compared with white donors, had an increased risk of hypertension (adjusted hazard ratio, 1.52; 95% confidence interval [CI], 1.23 to 1.88), diabetes mellitus requiring drug therapy (adjusted hazard ratio, 2.31; 95% CI, 1.33 to 3.98), and chronic kidney disease (adjusted hazard ratio, 2.32; 95% CI, 1.48 to 3.62); findings were similar for Hispanic donors. The absolute prevalence of diabetes among all donors did not exceed that in the general population, but the prevalence of hypertension exceeded NHANES estimates in some subgroups. End-stage renal disease was identified in less than 1% of donors but was more common among black donors than among white donors. Conclusions
As in the general U.S. population, racial disparities in medical conditions occur among living kidney donors. Increased attention to health outcomes among demographically diverse kidney donors is needed. (Funded by the National Institute of Diabetes and Digestive and Kidney Diseases and others.)
724
n engl j med 363;8 nejm.org august 19, 2010
The New England Journal of Medicine Downloaded from nejm.org on November 5, 2012. For personal use only. No other uses without permission. Copyright © 2010 Massachusetts Medical Society. All rights reserved.
R acial Variation in Outcomes among Kidney Donors
L
iving kidney transplantation is considered to offer patients with end-stage renal disease the best opportunity for dialysis-free survival.1 In 2006, approximately 27,000 transplantations from registered living kidney donors were performed worldwide,2 and living donors supplied nearly 40% of kidney transplants in the United States.3 Most evidence concerning the safety of living kidney donation for donors derives from single-center studies with limited statistical power and few nonwhite donors.4 In a recent study, investigators at the University of Minnesota achieved high ascertainment of longterm patient and renal survival and reported no adverse effects of living kidney donation on life span or risk of end-stage renal disease, as compared with survey data from the general U.S. population.5 Notably, in the Minnesota cohort, 98.8% of the patients were white. Linkage of records from the Organ Procurement and Transplantation Network (OPTN) (as supplied by the United Network for Organ Sharing) with the Social Security Administration’s Death Master File recently indicated that although surgical and long-term mortality were higher among black donors than among white donors, the long-term rate of death did not exceed that of corresponding control subjects in the National Health and Nutrition Evaluation Survey (NHANES).6 Although racial disparities in the burden and consequences of diabetes mellitus, hypertension, and chronic kidney disease in the general population have been extensively documented,7-10 few data exist concerning long-term medical outcomes among nonwhite kidney donors. Currently, the OPTN collects data on living donors for only 2 years of follow-up,11 and incomplete reporting and donor loss to follow-up are common,12 owing in part to compliance barriers, such as cost and inconvenience.13 Thus, addition al methods for capturing health outcomes among racially diverse living kidney donors are needed. To determine longer-term postdonation medical outcomes independent of a donor’s interaction with the transplantation center, we linked administrative data from a private insurance provider with OPTN-supplied identifiers for living donors. Using these data, we identified postdonation diagnoses of hypertension, diabetes mellitus, chronic kidney disease, and cardiovascular disease; investigated variation in the risk of postdonation medical diagnoses, according to socio-
demographic traits; and estimated the prevalence of these diagnoses in demographic subgroups. We also compared relative and absolute prevalence estimates with those in recent NHANES data.
Me thods Data Sources and Participant Selection
We assembled our study data by linking OPTN records for living kidney donors with administrative data from a national private U.S. health insurer. OPTN data include information on all donors and transplant recipients in the United States, as submitted by OPTN members.14 The Health Resources and Services Administration (HRSA) provides oversight on the activities of the OPTN. After approval by the institutional review board at Saint Louis University, we linked beneficiary identifier numbers from the insurer’s electronic databases, using names and birthdates, with unique OPTN identifiers for living kidney donors. Analyses were performed with the use of limited data sets in compliance with the provisions of the Health Insurance Portability and Accountability Act with all direct identifiers removed. Study participants were eligible if they had an OPTN record of having served as a living kidney donor from October 1987 through July 2007 and were eligible for benefits under the participating insurer after donor nephrectomy at some point during the period from May 2000 through December 2007, the period of available claims data. All participants were simultaneously enrolled in medical and pharmacy benefits with this company exclusively during the study window. U.S. Census data were incorporated according to residential ZIP Code at the time of donor nephrectomy. Outcome Measures
We ascertained medical diagnoses of hypertension, diabetes mellitus, chronic kidney disease, and cardiovascular disease among living kidney donors, using billing claims with corresponding diagnosis codes as listed in the International Classification of Diseases, Ninth Revision, Clinical Modification, similar to algorithms described previously.15-19 We also examined drug-treated hypertension and diabetes (with either insulin or oral agents) in pharmacy claims, using drug-category codes. Stage-specific coding for chronic kidney disease was introduced in October 2005. Therefore, we
n engl j med 363;8 nejm.org august 19, 2010
The New England Journal of Medicine Downloaded from nejm.org on November 5, 2012. For personal use only. No other uses without permission. Copyright © 2010 Massachusetts Medical Society. All rights reserved.
725
The
n e w e ng l a n d j o u r na l
Left censoring
Donor nephrectomy, OPTN registry data, U.S. Census data
Right censoring
of
m e dic i n e
fined according to the participant’s report of these diagnoses on the basis of encounters with a doctor or other health care professional. Statistical Analysis
Window of insurance benefits, claims data with diagnoses
Figure 1. Schematic Diagram of Linkage of Study Data Sources. Identifiers from the Organ Procurement and Transplantation Network (OPTN) were linked to the administrative data of a private U.S. health insurer for 4650 living kidney donors from October 1987 through July 2007. Post-nephrectomy medical diagnoses and conditions requiring medical treatment were ascertained from billing claims. Cox regression analyses with left and right censoring to account for observed periods of insurance benefits were used to estimate absolute prevalence and prevalence ratios for diagnoses after nephrectomy. Prevalence patterns were then compared with those in the 2005–2006 National Health and Nutrition Examination Survey (NHANES) for the general population. U.S. Census data were incorporated according to residential ZIP Code at the time of donor nephrectomy.
examined diagnoses of chronic kidney disease of stage 3 to 5 or end-stage renal disease (i.e., chronic kidney disease requiring dialysis) in a prespecified subgroup with insurance eligibility ending June 2006 or later.
Data sets were merged and analyzed with the use of SAS for Windows software, version 9.2 (SAS Institute). Since windows of insurance benefits varied across the sample, we used Cox regression with left and right censoring to account for observed periods of insurance benefits to model the frequency with 95% confidence intervals and correlates with adjusted hazards ratios of prevalent diagnoses after donor nephrectomy (Fig. 1). The prevalence of diagnoses 5 years after donor nephrectomy in the full cohort and in prespecified subgroups was estimated from outcome-specific Cox models. We estimated correlates of prevalent diagnoses in the general population using SAS Proc Survey logistic software to correct for unequal selection probabilities and response rates in NHANES. The prevalence of medical conditions in subgroups in the general population was estimated by transforming the logistic-regression equation. A P value of less than 0.05 was considered to indicate statistical significance.
Baseline Demographic Variables
Demographic data from the OPTN at the time of donor nephrectomy included age, sex, and race or ethnic group, as reported by the donor to the transplantation center. Because the OPTN began collecting information on predonation hypertension in June 2004, we examined baseline hypertension status in a secondary analysis. An index of neighborhood socioeconomic status at the time of nephrectomy was computed from U.S. Census data linked by ZIP Code, according to methods used by the Agency for Healthcare Research and Quality20 (for details, see the Methods section in the Supplementary Appendix, available with the full text of this article at NEJM.org).
R e sult s Demographic Characteristics of Donors
Among 4650 kidney donors in the study cohort, 76.3% were white, 13.1% black, 8.2% Hispanic, and 2.4% another race or ethnic group (Table 1). White donors were significantly older at the time of donation than were nonwhite donors in the study sample and nationally (P
