Role of Socioeconomic Status in Kidney Transplant Outcome

Materials and Methods

Results

This study was approved by the University of Utah’s Institutional Review Board. A total of 92,844 patients with kidney transplants were identified and studied.

Survival Analysis

Although the models described here were not adjusted for the year of the transplantation because of a relatively short study period, we realize that the time of the transplantation might still confound the results. Therefore, additional analysis was performed after adjustment for the model for the year of the transplantation. Also, because using the year of the transplantation as an indicator of transplant era may introduce bias in the statistical analysis, we selected the use of mycophenolate mofetil, that came on the market in mid-1990s, as a surrogate for the transplant era. Using both approaches revealed results very similar to our original analysis and are not presented here.

Role of Education Level.

Kaplan-Meier plots suggested incremental improvement in the outcome with increased education level (Figure 1). The proportional hazard model demonstrated improved outcome associated with more advanced education. Compared with patients who receive grade to high school (0 to 12) education, those with some college to bachelor’s degree had significantly better graft (hazard ratio [HR] 0.93, P < 0.005) and recipient (HR 0.90, P < 0.005) survival. Furthermore, recipients with postcollege graduate degrees had even better outcome, with HR of 0.85 (P < 0.005) for the graft and HR of 0.88 (P = 0.09) for recipient failure (Table 3). When different racial groups were analyzed separately, black and white patients (but not Asian patients) had a similar trend for the graft survival. A similar trend for the recipient survival was observed only in white patients (Tables 4 and 5).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Kaplan-Meier analysis illustrates the association of recipients’ education level with graft (A) and recipient (B) survival (for both analyses, P < 0.001 by log-rank test).

View this table:

• View inline
• View popup

Table 3.

Proportional hazard model of the graft and recipient survival for the whole group of patients^a

View this table:

• View inline
• View popup

Table 4.

Proportional hazard model of the graft survival stratified by recipient race^a

View this table:

• View inline
• View popup

Table 5.

Proportional hazard model of the recipient survival stratified by recipient race^a

Primary Source of Pay.

On the basis of Kaplan-Meier plots, recipients with private insurance had better outcome than those with Medicare, Medicaid, or other sources of payment. As far as recipient survival, patients who have Medicare seem to have worse outcome than other groups, which could be confounded by older age (data not shown). In the entire patient population, using Medicare as a reference group, the proportional hazard model demonstrated statistically significant benefit to graft survival from having private insurance (HR 0.87, P < 0.001; Table 3). This effect was not observed in Asian recipients but remained in the black patients (HR 0.8, P < 0.001) and white patients (HR 0.89, P < 0.001). Private insurance in comparison with Medicare also had an advantage for recipient survival in the entire group of patients (HR 0.8, P < 0.001) as well as in Asian (HR 0.66, P < 0.05), black (HR 0.71, P < 0.001), and white patients (HR 0.83, P < 0.001), when analyzed separately. HMO/PPO was associated with significantly higher risk for graft failure (HR 1.2, P < 0.05) but improved patient survival (HR 0.52, P < 0.001; Tables 4 and 5). Having Medicaid, US/state government agency, or other sources of payment for medical services did not show any significant association with outcome as compared with Medicare either in the whole patient population or in the subgroups divided by race.

Role of Citizenship.

Kaplan-Meier plots suggested that resident and nonresident alien recipients had the best graft outcome and patient survival as compared with US citizens (Figure 2). Using the Cox model in the entire patient group with US citizens as the reference group, resident aliens had significantly better graft outcome (HR 0.81, P < 0.001; Table 3). When analysis was stratified by race, only in white patients did this association reach statistical significance (HR 0.823, P < 0.001; Table 4). Similar results were observed for recipient survival: when compared with US citizens, legal aliens had a survival advantage (HR 0.7, P < 0.001; Table 2). This effect was observed in Asian patients (HR 0.66, P < 0.05) and white patients (HR 0.7, P < 0.001) but not in black patients (Table 5). In addition, similar to resident aliens, nonresident aliens have better outcome than the US citizens, but this finding did not reach statistical significance, likely because of the very small sample size of this group.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Kaplan-Meier analysis illustrates the association of recipients’ citizenship and with graft (A) and recipient (B) survival (for both analyses, P < 0.001 by log-rank test).

Subgroup Analysis: Adult Patients with the First Transplant.

We reanalyzed the subset of patients who were older than 18 yr and had the first transplant (n = 78,181). In this subset of the data, we found associations similar to those in the whole data set. Compared with patients with high school attendance, those with college education had a trend toward better graft (HR 0.96, 95% confidence interval [CI] 0.91 to 1.01, P = 0.0965) and recipient (HR 0.90, 95% CI 0.84 to 0.97, P < 0.01) outcome. Those with more advanced education had even better outcome for the graft (HR 0.88, 95% CI 0.78 to 1.00, P < 0.05) and recipient (HR 0.90, 95% CI 0.77 to 1.05, P = 0.173) survival. Compared with patients with Medicare (n = 28,882), recipients who had private insurance (n = 15,339) had lower risk for the long-term graft failure (HR 0.86, 95% CI 0.82 to 0.90, P < 0.001) and recipient death (HR 0.80, 95% CI 0.76 to 0.85, P < 0.001). Having HMO/PPO (n = 618) was associated with worse graft survival (HR 1.27, 95% CI 1.07 to 1.51, P < 0.01) but better recipient survival (HR 0.54, 95% CI 0.37 to 0.78, P < 0.005). Also, resident aliens (n = 1739) had an advantage over US citizens for graft (HR 0.81, 95% CI 0.74 to 0.90, P < 0.005) and recipient (HR 0.70, 95% CI 0.62 to 0.81, P < 0.001) survival.

Discussion

Socioeconomic factors have been shown to affect health care outcomes. Poverty, unemployment, and low education level have been listed among the factors that adversely affect health (19). SES has been suggested to play a significant role in kidney transplant outcome. Among others, the following socioeconomic factors have been listed as risk factors for posttransplantation noncompliance: occupational status, educational level, language or cultural barriers, and ethnic background (20). Race and income have substantial effects on mortality and use of services among Medicare beneficiaries (21). Poor individuals are less likely than wealthy individuals to be medically suitable, to be interested in transplantation, and to complete the pretransplantation workup (22).

In a study similar in design to our project and based on UNOS data, in patients with liver transplant, it has been shown that neighborhood income had no effect on graft or patient survival, education had only marginal influence on the outcome (survival was lower in those with a high school education than in those with graduate education), and patients with Medicaid and Medicare had lower survival when compared with those with private insurance (23). The results of our study that was done in kidney transplant recipients are similar. In the entire patient group, there is a statistically significant benefit to graft and patient survival from having private insurance compared with Medicare. This effect was observed across almost all racial groups (except for Asian patients, for whom there was no significant association between private insurance and graft survival). HMO/PPO was associated with significantly higher risk for graft failure but improved patient survival. These results are similar to those reported in liver transplant recipients (23).

Compliance with regard to immunosuppressive medication use is one of the key factors in prolonging graft survival. Until 1993, Medicare regulations allowed for coverage of immunosuppressive medications only for 1 yr posttransplantation unless the recipients maintain their Medicare beneficiary status through disability or age. Between 1993 and 1995, that duration was gradually extended to 3 yr posttransplantation, which was extended further by 8 mo in 2000. Woodward et al. (24) have shown that extending the coverage from 1 yr to 3 yr posttransplantation has eliminated the 4.5% difference in graft survival between low-income and high-income recipients. In a follow-up analysis (25), the same authors estimated that if Medicare provided life-long immunosuppressive medications to all of the recipients, then graft failure would be reduced by 1.2% annually beginning in the fourth year posttransplantation. Furthermore, Medicare beneficiaries who are eligible to receive the immunosuppressive medication coverage must still pay 20% of the cost of these medications as Medicare covers only 80% of the total cost. With many HMO/PPO or private insurance carriers, subscribers may be required to pay substantially lower copayments. This could also contribute to better compliance with immunosuppressive medications and, subsequently, better graft outcome among HMO/PPO or private insurance subscribers compared with Medicare beneficiaries. Medicare beneficiaries who undergo kidney transplantation do not have coverage for nonimmunosuppressive prescription drugs. This could explain the poor recipient survival seen in our analysis. Medicare beneficiaries were shown to be more likely to take required prescription medications if they had prescription drug coverage (26). Finally, even though we included age as an independent variable in our analysis, there may be a residual confounding effect of age on the outcome. As Medicare patients tend to be older compared with HMO/PPO or private insurance patients, Medicare beneficiaries may have poor outcome compared with non-Medicare recipients.

A possible reason for inferior outcome in Medicaid beneficiaries is as follows: Previous studies have shown that Medicaid beneficiaries are less likely to receive optimal treatment, and their outcome is worse compared with privately insured patients for common conditions such as myocardial infarction and bronchial asthma (27,28). Restricted access to medical care because of lower reimbursement by Medicaid and highly variable coverage benefits between different states are some of the possible explanations for such poor outcomes. Low income and poverty indicated by Medicaid may contribute directly or indirectly to worse health outcomes in general. Indeed, cost-related skipping of medications has been shown to be associated with the level of drug coverage and the income level (29). A potential explanation for better outcomes among those with private insurance could be better quality of care. Furthermore, these patients might be either healthy enough to be employed or have high personal income to be able to afford private insurance, both of which may potentially influence the outcomes.

Our study demonstrated that better recipient but worse graft survival is associated with HMO/PPO coverage. Better graft survival does not always translate into improved recipient survival. In this particular case, we hypothesized that the HMO population is likely to consist of relatively young employed patients. The residual confounding effect of age might explain longer recipient survival as compared with Medicare beneficiaries. Poor graft survival among HMO/PPO recipients is intriguing and difficult to explain. It is interesting that in another study, no difference in clinical outcome was demonstrated between HMO and fee-for-service patients (30). We speculate that higher copays and deductibles in these plans for specialist physician visits and expensive immunosuppressive medications compared either with private insurance plans or with Medicare/Medicaid may be a hindrance for the patients to comply with required posttransplantation treatment. Indeed, the HMO membership was associated with higher degree of cost-related skipping of medications as compared with Medicare beneficiaries (29).

Also, in our analysis, recipients with higher education level had better graft and patient survival. There is a clear trend in incremental lowering of the HR for both graft failure and recipient survival with advanced education level. Theoretically, people with a higher level of education are more likely to be well informed and have better awareness of posttransplantation care, which could potentially improve outcomes. The correlation between the better education status and compliance is arguable. We also contemplated the possibility of the association between education level and insurance status so that the effect of these variables on the outcome is not independent. It seems logical that the insurance status would be associated with education level of the recipients. To address this question, we evaluated the potential association between these variables in a bivariate (χ²) analysis. We found a significant association between education level and insurance status (P < 0.001). In general, people with a higher level of education tend to earn higher income. For example, according to US Census Bureau data (31), annual average earnings of workers with a bachelor’s degree was $45,678 in 1999 compared with $24,572 for those with only a high school diploma. Higher income may translate into better medical care and greater ability to pay for medication, which in turn may translate in to better graft and recipient survival. This effect of the higher education on the outcome might or might not be independent of the insurance status. To address this potential confounding effect of the education level on the insurance status, we constructed two separate models with each one having either the level of education or the type of insurance coverage and found that the results were very similar to the main analysis that included both the variables. This similarity suggests an independent association between the outcomes and the level of education and the type of insurance coverage.

Finally, in our analysis, resident aliens seem to have a significantly better outcome than US citizens in terms of both graft and recipient survival. This effect was observed in white but not in black patients and may be explained by a number of factors. These results are somewhat counterintuitive, as previously the negative association between foreign immigration status and the health outcome has been suggested (19). Low income, unemployment, low level of education, lack of health insurance and access to quality health care, and anti-immigrant sentiment and discrimination in health care (19) as well as language barrier (32) have been listed as potential reasons. However, because of the selection process, the transplantation population might not reflect the general trends described in immigrants. Also, it is conceivable that people who recently arrived in the United States might have certain differences in environmental factors compared with the people who were born in the United States. In a recent study, women from Poland and more recent migrants had generally more nutritious intakes, compared with US-born women or earlier migrants (33). In the year 2000, people who were born outside the United States composed an estimated 11.1% of the US population (34). In the report published by the Centers for Disease Control and Prevention, women who were born outside the United States had better birth outcomes than their racial/ethnic US-born counterparts (35). In addition, we speculate that the fraction of the foreign population that immigrates to the United States might not be simply a random slice of the foreign society but rather the motivated and active part of it. Therefore, one can hypothesize that there is a degree of selection bias when the outcome of the American citizens is compared with foreign nationals who live in the United States. The attitude toward medical care (e.g., commodity rather than active involved process), such as the physician’s authority and subsequent compliance with recommendations, might be different between Americans and foreigners. Although most of these potential reasons are merely speculative, some cultural differences in regard to specific aspects of medical care between Americans and foreign nationals have indeed been described in the literature (36).

In general, on the basis of our results, it seems that some of the socioeconomic factors are significant predictors of the outcome across racial groups. That supports the independent role of these factors from the racial characteristics. Also, in the studies that analyzed the role of race in the transplantation outcome, potentially uneven distribution of the socioeconomic factors (as demonstrated in Table 2) in the different racial groups may confound the results of the analysis and hypothetically explain some of the differences in the outcome between the races.

This study is a retrospective analysis that used data that were reported to the USRDS. There are limitations as well advantages to large renal transplant database analyses. Database analyses can show long-term differences in outcomes and provide the statistical power to help determine the differences between the primary variables of interest. However, the quality of data is always a concern because of the significant amount of missing and potentially erroneous information (misclassification bias). In particular, SES is difficult to quantify, and subjects might be unwilling to share certain information (e.g., immigration status, education level). For example, in our data set, the education level variable is missing 61% of the data. The analysis of the role of the education status therefore is based on the remaining 39% of the data (>32,000 patients). It would be optimal not to have any missing values, but because of the very large sample size of the remaining observations, the results are still valid. One can assume random pattern of missingness in the data and analyze only records with nonmissing information. Although the random distribution of the missing values is likely, we decided not to discard the records with values missing but rather code and analyze them separately as reported above. Misclassification bias is difficult to address in the retrospective study and remains one of the shortcomings of the data registry analyses. Other potential problems with retrospective data analysis should be considered. As in other retrospective studies, although this analysis establishes the association between independent variables and the outcome, it cannot ascertain causative relationships. In addition, as in most retrospective studies, the results could be distorted by reverse causality described elsewhere (37). Finally, as our multivariate models depend on the variables that are available in the data set, certain potential confounders may not be included in the analysis (e.g., annual income, employment status, geographic location, IQ level, marital status). Analysis of the role of these variables may present a new and exciting opportunity for future research.

Conclusion

Recipients with higher education level, resident aliens (as compared with US citizens), and patients with private insurance have an advantage in the graft and recipient outcomes independent of racial differences.