We found that the odds of not being accepted for lung transplant were 1.56-fold higher for Medicaid compared with non-Medicaid patients. This key finding was independent of differences in demographic factors, disease severity indicators, potential contraindications to lung transplant, and before or after use of the LAS between Medicaid and non-Medicaid patients. To our knowledge, this is the first study to examine disparities with respect to access to lung transplant early in the evaluation process (i.e., before lung transplant wait listing). Although the United Network of Organ Sharing (UNOS) oversees organ allocation in the United States and has a mandate to ensure access will not be based on “political influence, race, gender, religion, or financial or social status,” this governance is primarily limited to after wait listing (10
). Our study also demonstrates that other indices of low SES, including not graduating from high school and residing in lower-income ZIP codes, were independently associated with not being accepted after lung transplant evaluation.
The underlying reasons for disparities in access to lung transplant for patients with CF of low SES are likely complex and multifactorial. Inadequate social support and noncompliance with medical regimens represent two key contraindications to lung transplant candidacy. Although we used the best available proxies to adjust for these important mediators in our analysis, these two factors are difficult to measure comprehensively based on data available in the CFFPR. Therefore, Medicaid and our other indicators of low SES might cluster with these incompletely measured factors. Not graduating from high school was the SES indicator most strongly associated with not being accepted for transplant, suggesting that educational attainment may influence acceptance beyond its role as an SES indicator. Transplant physicians might be less willing to wait-list patients with inadequate health literacy due to concerns about post-transplant compliance with treatment regimens, as has been suggested to be the case in the renal transplant literature (11
). Although we have implied that disparities in access to transplant are largely physician driven, it is possible that the apparent disparity is due to patient choice. For example, individuals of low SES might prefer not to be transplanted due to lack of perceived benefit. Patients living farther away from a lung transplant center tended to be of lower SES, and therefore geographical barriers might have influenced their decision not to proceed with transplant. Interestingly, driving time from residence to closest lung transplant center (by quartile) was not independently associated with not being accepted for lung transplant when adjusted for other SES indicators. This lack of association supports a previous finding from the renal transplant literature, which failed to demonstrate reduced access to kidney transplant for patients residing farther from the nearest transplant center (12
Another major observation from our study is that Medicaid patients were younger and sicker than their non-Medicaid counterparts at the time of initial lung transplant evaluation. First, this suggests that Medicaid patients are referred later than non-Medicaid patients, as they are sicker at the time of evaluation. Second, this suggests that nutrition, medical management (including adherence to therapies), and/or access to care might be worse for patients of low socioeconomic position, as they are younger at the time of needing a transplant. This is consistent with previous studies that have demonstrated differences with respect to health outcomes by SES. However, this observation needs to be interpreted with some caution, as sicker patients might be more likely to qualify for Medicaid, thus leading to reverse causation.
Our study is subject to a few important limitations. First, we used Medicaid as a proxy for low socioeconomic position. This proxy has been criticized in previous studies (8
); patients with more severe disease are more likely to qualify for Medicaid, as medical expenses are considered when evaluating eligibility (13
). Although this can be problematic when studying the association between SES and health outcomes, this may not be a significant limitation when studying access to care, particularly when disease severity is accounted for. As there is no single accepted measure for SES (14
), we examined four alternate indicators of low socioeconomic position, each of which has inherent limitations. Median household income by ZIP code is an ecologic measure and therefore requires relative homogeneity of household income within a geographic region for it to be reliable measure of individual-level SES in nonlinear models (15
). Therefore, these results should not be interpreted at the individual level; rather, these results should be seen as an area level effect that could represent a larger, multifactorial effect resulting in less access to care. Estimation of distance from residence to closest lung transplant center was also an ecologic measure that required a few assumptions. We assumed that patients chose their transplant center based on proximity alone (e.g., patients did not travel longer to another center of choice) without respect to interstate boundaries and that comorbidities (e.g., colonization with Burkholderia cepacia
) did not influence selection of transplant center. We chose driving time as opposed to road distance, as travel times for identical road distance can vary based on highway versus city versus rural driving. High school graduation also has limitations, as graduation may be affected by disease severity, with more severely diseased patients missing more schooling due to illness.
Our analysis focused on the decision of the initial lung transplant evaluation. A potential concern is that approximately one-third of patients were deferred and thus classified as not accepted. We chose to classify patients initially deferred as not accepted, as initial deferral might have similar implications to being rejected. For example, deferral may lead to critical delays, such that these patients might be more likely to die before wait listing, thus reducing their opportunity for transplant. To support this reasoning and classification, 167 of 370 patients (45%) initially deferred died while still being classified as deferred. Nevertheless, patients initially deferred could have been accepted during repeat evaluation, especially if they were referred too early, required further work-up, and/or needed time for medical optimization. To evaluate the effect of this potential misclassification, we performed a sensitivity analysis comparing patients who were ultimately accepted to those who were ultimately declined for transplant at the end of cohort follow-up. This resulted in a slightly stronger association between low socioeconomic position and reduced access to transplant. Furthermore, we conducted a sensitivity analysis restricting to patients with a FEV1 of less than 30% and without potential contraindications in an attempt to focus our analysis on those individuals most likely to be considered for transplantation and had results similar to our primary analysis.
An additional limitation is that we presented our analysis using ORs and not risk ratios. ORs tend to overestimate effect sizes for nonrare outcomes, thus limiting the interpretation of absolute values. However, our study was designed to explore potential associations between SES indicators and access to transplant rather than to highlight or compare the absolute strengths of any associations.
Our study findings are concerning, since virtually all patients with end-stage CF have health insurance and are followed at an accredited CF care center. Despite this, there was still a strong differential access to lung transplant by SES. The differences observed in CF are likely exaggerated in other pre–lung transplant populations, such as chronic obstructive pulmonary disease and idiopathic pulmonary fibrosis, whereby a wider spectrum exists with respect to quality of care and health insurance coverage. It remains unclear as to why patients with CF of low socioeconomic position experience differential access to lung transplant, but it does not appear to be related to underlying differences in disease severity or potential contraindications. It is possible that residual confounding and/or mediation were introduced. For example, patient frailty and poor functional status are not measured in the CFFPR but may represent reasons that patients of low socioeconomic position have higher odds of not being accepted for transplant. Furthermore, we were unable to account for social habits such as alcohol or other substance abuse that are likely more common among patients of low SES and may also represent contraindications to lung transplant. We believe the overall influence of residual confounding/mediation on our results is likely minimal, as we have adjusted for a number of variables with little or no impact on the strength of the association between Medicaid insurance and not being accepted for lung transplant.
In conclusion, we found evidence that acceptance for lung transplantation was lower for patients with CF of low socioeconomic position. We in the medical community need to support an urgent call for studies and evaluations to (1) explore factors associated with Medicaid status that preclude lung transplant listing to improve access to lung transplantation for adult patients with CF of low socioeconomic position, and (2) investigate whether this differential access to care is present in other populations in which evaluation for lung transplant occurs.