Part1: Sex Disparity in Deceased-Donor Kidney Transplant Access By Cause Of Kidney Disease
May 25, 2022
For more info. contact tina.xiang@wecistanche.com
Abstract
Background and objectives Women with kidney failure have lower access to kidney transplantation compared with men, but the magnitude of this disparity may not be uniform across all kidney diseases. We hypothesized that the attributed cause of kidney failure may modify the magnitude of the disparities in transplant access by sex.
Design, Setting, participants,&measurements We performed a retrospective cohort study of adults who developed kidney failure between 2005 and 2017 according to the United States Renal Data System. We used adjusted Cox models to examine the association between sex and either access to waitlist registration or deceased-donor kidney transplantation and tested for interaction between sex and the attributed cause of kidney failure using adjusted models.
Results Among a total of 1478,037 patients,271111 were registered on the waitlist and 89.574 underwent deceased-donor transplantation. The rate of waitlisting was 6.5 per 100 person-years for women and 8.3 per 100 person-years for men. In adjusted analysis, women had lower access to the waitlist (hazard ratio, 0.89;95%confidence interval,0.89 to 0.90) and to deceased-donor transplantation after waitlisting (hazard ratio, 0.96;95%confidence interval, 0.94 to 0.98). However, there was an interaction between sex and the attributed cause of kidney disease in adjusted models(P<0.001). Women with kidney failure due to type 2 diabetes had 27%lower access to the kidney transplant waitlist (hazard ratio,0.73;95% confidence interval, 0.72 to 0.74) and 11%lower access to deceased-donor transplantation after waitlisting compared with men(hazard ratio,0.89;95% confidence interval, 0.86 to 0.92).In contrast, sex disparities in access to either the waitlist or transplantation were not observed in kidney failure secondary to cystic disease.
Conclusions The disparity in transplant access by sex is not consistent across all causes of kidney failure. Lower deceased-donor transplantation rates in women compared with men are especially notable among patients with kidney failure attributed to diabetes.
Introduction
Kidney transplantation improves both survival and quality of life for patients who develop the need for maintenance KRT compared with dialysis (1-4), but women are less likely to receive kidney transplantation than men (5-12). Historically, women have had lower access to the transplant waitlist (5,7), but the reasons for this disparity have not been fully delineated. Although studies have examined differences in access to the transplant waitlist or transplantation itself on the basis of sex-related differences in body size (8), human leukocyte antibody sensitization (12,13), age (7), or other factors (7,14), few studies have determined whether there is variability in access to transplantation by sex depending on the attributed cause of kidney failure. Understanding whether the underlying attributed cause of kidney failure is associated with differential access to transplantation by sex could enhance our understanding of potential reasons for these disparities.
Some conditions that cause kidney failure may not affect men and women equally and therefore partially contribute to the observed sex disparities in access to transplantation. In addition, some causes of kidney failure may elevate the risk for concurrent cardiovascular risk more than others. Because it is known in the general population that women with cardiovascular disease are less likely to receive and achieve aggressive risk reduction measures (such as aspirin) compared with men, differential cardiovascular risk profiles across different causes of kidney failure could also contribute to delayed or lower access to transplantation in women compared with men (15-17).
In this study, we examined whether the observed sex-based disparities in access to the waitlist and deceased-donor kidney transplantation differed by attributed cause of kidney failure using data from the United States Renal Data System (USRDS).
Materials and Methods
Study Population
We included adults aged ≥18 years who started maintenance KRT between January 1, 2005, and December 31, 2017, according to the USRDS. We excluded (1)patients considered for multiorgan transplants (except simultaneous pancreas-kidney transplants)and (2)patients with missing data on the cause of kidney failure (Figure 1).
Factors of Interest
We abstracted the primary predictor, patient sex, from the Patients file in the USRDS. We abstracted the cause of kidney failure from data reported on the CMS-2728 Medical Evidence (MEDEVID) form (filed for all patients with kidney failure at maintenance KRT initiation)and categorized these causes as type 2 diabetes mellitus, type 1 diabetes mellitus, hypertension, cystic disease, GN, or other. The most common causes of kidney failure in the "other" category included unknown etiology (29%), tubular necrosis without recovery(18%), and multiple myeloma (7%). The attributed cause of kidney failure served as the effect modifier of interest in all primary analyses.
Covariates
We adjusted for other covariates that could be associated with a patient's likelihood of waitlisting or transplantation (7,14,18-20). These include age at kidney failure onset (categorized as<35, 35 to <50, 50 to<65, and ≥65 years), race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, and other), initial calendar year of maintenance KRT(to account for temporal trends), and zip code from the USRDS Patients file. For analyses from time of waitlist to transplantation, age at waitlisting (categorized as above)and year of waitlisting were abstracted from the Waitlist file. We used zip codes to determine Organ Procurement and Transplantation Network(OPTN)region and median neighborhood household income as a continuous variable by matching the patient's zip code to the 2006-2010 American Community Survey (21). We abstracted body mass index(BMI)(categorized as<18.5, 18.5 to <25, 25 to <30， 30 to <35， 35 to<40， and ≥40 kg/m² according to Centers for Disease Control definitions)(22), insurance type (none, Medicaid, Medicare, or other), and comorbidities(diabetes, congestive heart failure, peripheral arterial disease, stroke, myocardial infarction, tobacco use, chronic obstructive pulmonary disease, cancer) at kidney failure onset from the MEDEVAC file of the USRDS.Panel reactive antibody (PRA)was abstracted from the Waitlist file in the USRDSand categorized as0,>0to 20,>20-80, and>80%.
Missing Covariates
Because 3% of income data were missing from the total cohort (which we believed to be an important predictor), we used chained multiple imputations to account for missingness. We imputed missing income data using models that included age, sex, race/ethnicity, insurance type, calendar year of maintenance KRT, cause of disease, OPTN region, and transplantation as an outcome with 20 repetitions.
Outcomes
Outcomes of interest were (1)time to kidney transplant waitlisting from first maintenance KRT date;(2) time to first deceased-donor kidney transplant from first maintenance KRT date; and (3) time to first deceased-donor kidney transplant starting from waitlisting for patients registered on the waiting list on or after first maintenance KRT date. Follow-up for outcomes was censored when living-donor kidney transplantation occurred.
Dates of waitlist registration were abstracted from the Waitlist file. Dates of transplantation and first transplant donor source(living versus deceased) were determined using the USRDS Patients file. Follow-up for outcomes was administratively censored on December 31, 2017.
Statistical Analyses
We compared baseline characteristics by sex and attributed the cause of kidney failure using a t-test, Mann-Whitney U test, and chi-square test as appropriate. We also compared differences in the burden of comorbid conditions at the time of maintenance of KRT initiation by sex and attributed the cause to kidney failure.
Association between Sex and Waitlisting
We assessed the association between sex and access to the waitlist using an unadjusted Cox proportional hazards model. Time to waitlisting (in years)was measured from the first date of kidney failure until waitlist registration (either active or inactive status) and censored for death or end of study follow-up. For those whose waitlist date was equal to or preceded the first maintenance KRT date, the time to waitlisting was set to 0.1 days after the first KRT date.
Next, we added covariates to the model in a stepwise fashion, first including a limited set of characteristics including age at kidney failure, race/ethnicity, BMI, and attributed cause of kidney disease (Model 1). We next included median neighborhood income by zip code, insurance status, calendar year of maintenance KRT initiation, OPTNregion, and comorbid conditions (Model 2).To test whether disparities in waitlisting by sex were more pronounced in those with different attributed causes of kidney failure, we tested for interactions between sex and the attributed cause of kidney failure in our fully adjusted Cox models.
To account for changes to the Kidney Allocation System (KAS), which was implemented on December 4, 2014, we performed additional stratified analyses where we examined access to the waitlist before versus after this date.
Association between Sex and Deceased-Donor Kidney Transplantation among all Patients with Kidney Failure
We examined the association between sex and hazard of deceased-donor kidney transplantation among all patients who started maintenance KRT using an unadjusted Cox model and censoring for living-donor transplantation or death. We included covariates in a stepwise fashion as described and tested for interaction between sex and attributed the cause of kidney failure in the fully adjusted models. For those whose first kidney transplant date was the same as the first maintenance KRT date, the time to transplantation was set to 0.1 days after the first KRT date.
Association between Sex and Deceased-Donor Kidney Transplantation in Patients Waitlisted after Starting KRT
We also examined the association between sex and hazard of deceased-donor kidney transplantation among waitlisted patients who registered on the waitlist on or after their first KRT using an unadjusted Cox model. Time in these models started from the date of first waitlist registration (on the day of or after KRT initiation) to deceased-donor kidney transplantation, and follow-up was censored for living-donor transplantation and death. We then serially adjusted models in the same fashion as above, except using age at waitlisting and year of waitlisting (instead of age at KRT and year of KRTinitiation) and the addition of PRA to both Models 1 and 2. All covariates except age, year of waitlisting, and PRA were abstracted at the time of kidney failure onset. Finally, we tested for interaction between sex and attributed the cause of kidney failure in the fully adjusted model.
We did not analyze transplantation as an outcome in stratified analysis by the era of first maintenance KRT given that our post-KAS follow-up period was brief (ending in 2017) for this outcome and fewer deceased-donor
kidney transplants occurred during this short follow-up period.
For all interactions, we considered P<0.05 to be statistically significant. Plots of scaled Schoenfeld residuals were used to assess the assumption of proportional hazards.
Sensitivity Analysis
In sensitivity analysis, we performed competing risk analyses using Fine-Gray models adjusted for the same covariates as above but treating the death as a competing risk for all outcomes. We also repeated Fine-Gray models for the outcome of deceased-donor kidney transplantation, treating living-donor transplantation as a competing risk. For these sensitivity analyses, we used a 5%-20% random data sampling approach and tested for interaction in this smaller subset due to infeasible run times.
Given a prior study that showed an interaction between age and sex for access to kidney transplantation (7), we tested for the presence of such interaction in our fully adjusted models and performed stratified analysis by age for outcomes in which a statistically significant interaction (P<0.05) was detected.
Finally, we performed additional analyses for access to the waitlist and transplantation after the first maintenance KRT, where we excluded those who were waitlisted before the date of first maintenance KRT to understand whether sex disparities were present in those who did not have the advantage of pre-emptive waitlisting.
Data analyses were conducted using STATA 15 (College Station, TX) and SAS Software version 9.4 (Cary, NC). The University of California San Francisco Institutional Review Board considered this study non-human subject research, and the study was also approved by the Stanford University Institutional Review Board (Protocol #51697).
