Association of Race and Socioeconomic Position with Outcomes in Pediatric Heart Transplant Recipients

Subjects

We assessed the association of patient SE position with posttransplant outcomes in patients from four pediatric heart transplant centers (Children's Hospital Boston, MA, Loma Linda University Children's Hospital, CA, Atlanta Children's Hospital, GA, and Children's Hospital of New York, Columbia University, NY) participating in the Pediatric Heart Transplant Study (PHTS). The PHTS is an event-driven, prospective, multi-institutional database established in January 1993 that collects outcome data in children listed for heart transplantation at participating institutions. All PHTS centers have obtained Institutional Review Board approval for participation in PHTS.

All patients who underwent their first heart transplant at one of the four participating centers between January 1, 1993 and December 31, 2005 and participated in the PHTS were eligible for this study. We excluded non-US residents (international patients) who came to the US for heart transplant. For the purpose of this analysis, the four participating centers provided the block group of residence of their patients already enrolled in the PHTS study, which were then used to collect their baseline SE data.

Assessment of SE position

Each patient's home address at the time of heart transplant was used to determine the block group of residence from the US census web site. Based on a previously described measure using six SE variables for the block group of residence, a summary SE score was derived for each patient and used to determine the SE position (16,18). The SE variables selected for the summary score were first described by Diez Roux et al. (16) using factor analysis, a statistical method to select variables out of a large set that can be meaningfully combined into a composite score.

The summary SE scores used in this study were determined from data on six SE variables for each patient's block group of residence, which were extracted from the US Census 2000 report. These variables represent three dimensions of wealth and income (log of the median household income, log of the median value of housing units, and the percentage of households receiving interest, dividend, or net rental income), two dimensions of education (the percentage of adults 25 years of age or older who had completed high school and the percentage of adults 25 years of age or older who had completed college) and one dimension of occupation (the percentage of employed persons 16 years of age or older in executive, managerial or professional specialty occupations) for the residents of the block group. For each variable, a z-score for each block group was calculated by subtracting the overall mean of that variable (across all block groups in the sample) from the value of the variable for that block group and dividing by the standard deviation. The summary SE score for each patient was obtained by adding the six z-scores (one for each of the six variables) for that patient. Data on the proportion of individuals living in the block group who were below the federally defined poverty level were also collected (18). Race/ethnicity was defined as reported by the transplant center to the PHTS.

Outcome variables

The primary outcome variable was graft loss (death or retransplantation) and was analyzed for both early (first year posttransplant) and late phase (conditional to surviving the first posttransplant year). Patients were followed until their graft loss (event) or censored on December 31, 2007. To further understand the association of SE position with graft loss, we also evaluated the association of SE position with the following secondary outcome variables: (a) cumulative rejection episodes (number of rejection episodes/patient over time), (b) time to first rejection episode, (c) time to first rejection episode with hemodynamic compromise (7) and (d) time to first late rejection episode (defined as first rejection episode more than 1 year after primary transplant) (5). As in previous PHTS studies (5,6), we defined a rejection episode as an event that resulted in acute augmentation of immune suppression because of endomyocardial biopsy findings or due to clinical and/or echocardiography findings of graft dysfunction.

Statistical analysis

Patients were divided into four equal-size groups based on their ranks (quartiles) on summary SE scores (group 1 being the lowest and group 4 the highest SE group). The groups were compared for the distribution of demographic, clinical, and SE variables using analysis of variance or Wilcoxon rank-sum test, as appropriate. Kaplan–Meier survival curves (with log-rank test) were used to compare time to graft loss and other time-to-event outcomes. Parametric multivariable proportional hazard modeling was used to assess the association of SE position (SE quartile) with risk of early and late graft loss and with secondary outcomes when found significant by log-rank test. Models were constructed using a forward selection procedure retaining variables significant at the 0.10 level based on a likelihood ratio test. These models were adjusted for transplant centers (irrespective of statistical significance) to control for institutional differences in distribution of patient diagnosis, race, SE position, clinical practice patterns and in cost of living in different US regions. All tests were two-sided. The data were analyzed using statistical software SAS version 8.2 (SAS Institute Inc, Cary, NC). A p-value less than 0.05 was considered statistically significant.

The authors had full access to the data and take responsibility for its integrity. All authors have read and agree to the manuscript as written.

Study population

Five hundred and twenty eight children who participated in the PHTS during 1993–2005 underwent their first heart transplantation at one of four participating institutions in this study. Of these, 38 children (7%) did not have a US home address in their medical records (because they were either non-US residents or had a US postoffice box address) to determine their block group of residence. The remaining 490 children formed the analytic cohort. Their mean age at transplant was 6.1 years, 57% were male, 68% were white, 16% were black and 23% were Hispanic (including white and black Hispanics). The distribution of demographic and clinical characteristics of children by their SE groups is shown in Table 1. The differences in distribution of racial groups across SE groups were statistically significant; blacks and Hispanics were more likely to be in the lower SE groups and whites in the higher SE groups (Table 1). The groups did not differ with respect to age, cardiac diagnoses, listing status, year of transplant, pre-transplant management with invasive support (inotrope, ventilator, extracorporeal membrane oxygenation, ventricular assist device) or percent with renal insufficiency.

Neighborhood SE characteristics

The SE characteristics of block groups of residence of the four SE groups are compared in Table 1. As expected, the differences between patient groups were statistically significant for all SE variables (p < 0.001 for all comparisons). The higher SE group patients lived in neighborhoods with progressively higher median household income and housing value, higher proportion of adults who had completed high school or college education, more workers in managerial, professional or executive profession and greater number of households that had rental, interest or dividends as source of income. The percentage of individuals living below the poverty level declined progressively with higher SE neighborhoods.

Graft loss

Overall, graft loss occurred in 152 children (122 deaths, 30 retransplant) during the study period. Graft loss occurred in 39%, 30%, 24% and 31% patients, respectively, in the four SE groups. Figure 1 compares graft loss during the first posttransplant year in the four SE groups using Kaplan–Meier curves. One-year freedom from death or re-transplant was 85%, 89%, 89% and 79% in the four groups, respectively (p = 0.10, log rank test for comparison among groups censored at 1-year). Race (black vs. non black, p = 0.2) or ethnicity (Hispanic vs. non-Hispanic) were not associated with first-year graft loss (p = 0.9). In a multivariable analysis adjusted for institutions and race, first-year graft loss was independently associated with ventilator support at transplant (hazard ratio, HR 2.68, 95% CI 1.38, 5.22, p = 0.003) and had a borderline association with SE group 4, the highest SE group (HR 1.94, 95% CI 0.98, 3.9, p = 0.05, Table 2). There was no association of first-year graft loss with age at transplant, patient gender, year of transplant, listing status at transplant and ECMO or VAD at transplant.

Figure 2 illustrates graft loss conditional to surviving the first year (n = 397) by SE groups (panel 1) and by race (panel 2). There was a significant difference among SE groups in late graft loss (p = 0.007) with worse outcome in SE Group 1 (lowest SE group) compared to other groups. Black children were at higher risk of death or re-transplant compared to non-black children among first years survivors (p < 0.001). The difference in risk of graft loss between Hispanic and non-Hispanic children was not significant (p = 0.2). In a multivariable analysis, late-phase graft loss was independently associated with black race (HR 1.81, 95% CI 1.07, 3.07, p = 0.02) and the lowest SE group (HR 1.77, 95% CI 1.13, 2.80, p = 0.01, Table 2).

Risk of rejection

A total of 851 rejection episodes were diagnosed in the study cohort. At least one rejection episode occurred in 358 (73%) of 490 children. Late rejection episodes (more than 1-year posttransplant) occurred in 172 (43%) of 397 1-year survivors of transplant.

Figure 3 illustrates a comparison of cumulative rejection episodes per patient with duration since transplant in the four SE groups (panel 1) and between black and non-black children (panel 2). The incidence of cumulative rejection episodes per patient was almost identical among the SE and racial groups during the first year posttransplant. However, SE group 1 and black race both appeared to be associated with higher cumulative rejection episodes with increasing duration since transplant.

There was no difference among SE groups in time to first rejection episode. The median time to first rejection episode (in months) was 2.3, 2.3, 1.4 and 1.5 in the four SE groups, respectively (p = 0.6, log-rank test). Furthermore, time to first rejection episodes was not different between black and non-black children (median time 2.9 vs. 1.7 months, p = 0.8) and between Hispanic and non-Hispanic children (median time 2.3 vs. 1.7 months, p = 0.30). Black children were at increased risk (shorter time) for first rejection with hemodynamic compromise compared to non-black children p = 0.008); however no difference was noted among the four SE groups for this outcome (p = 0.1).

Black patients were at higher risk of late rejection compared to non-blacks (p = 0.02, Figure 4, panel 1). The difference among SE groups in time to first late rejection was also statistically significant, SE Group 1 being the highest risk group (p = 0.001, Figure 4, panel 2). In a multivariable model, risk factors for late rejection included black race (HR 1.64, 95% CI 1.11, 2.43, p = 0.001), older age, and history of rejection during the first year, whereas patients on ECMO at transplant and SE group 3 patients (HR 0.48, 95% CI 0.32, 0.73, p = 0.0004) were at decreased risk of late rejection (Table 3).