|Home | About | Journals | Submit | Contact Us | Français|
Studies have demonstrated that African American race is a strong predictor of non-donation. However, it is often and correctly argued that African American race is a crude explanatory variable that is a surrogate marker of socioeconomic status (SES), education and access to health care. We hypothesized that when controlling for these factors, African American race would cease to be a predictor of organ donation.
A retrospective review was performed of 1292 Alabama decedents approached for organ donation between 2006 and 2009. Multivariable logistic regression models were constructed to identify the most parsimonious model that could explain the variation in the log-odds of obtaining consent.
Consent for donation was obtained from 49% of the decedent's families. Household income was a predictor of organ donor consent only in Caucasians. Surprisingly, household income was not statistically different between consented and non-consented African American decedents ($25,147 vs. $26,137; p=0.90). On multivariable analysis, education, urban residence and shorter distance between the decedent residence and donor hospital were significantly associated with obtaining consent for organ donation. On univariate analysis, the odds of donor consent in Caucasians compared to African Americans was 2.76 (95% CI 2.17 – 3.57). When controlling for SES and access to healthcare variables, the odds of donor consent increased to 4.36 (95% CI 2.88 – 6.61).
We interpret this result to indicate that there remains unknown but important factor(s) associated with both race and obtaining organ donor consent. Further studies are required to isolate and determine whether this factor(s) is modifiable.
Consent rates for organ procurement from deceased donors varies across the United States. Several studies have been performed to identify barriers for organ donor consent. Certain demographics have consistently been associated with non-donation. African American race is one of the most significant prognostic variables associated with non-donation. Multiple studies have demonstrated that African American race is a strong predictor of non-donation.(1-5) However, it is often and correctly argued that African American race is a crude explanatory variable that may be a surrogate marker of income, education and access to health care.(6-9) Studies suggest that the decreased socioeconomic status (SES) observed in African Americans compared to Caucasians accounts for much of the observed racial disparities in health.(10-12) However, even after adjustment for SES, there remains a clear decrease in access to medical care observed for African Americans as evidenced by geographic residential segregation and a decrease in the delivery of a broad range of medical procedures compared to Caucasians.(13-17) Therefore, it is not clear that African American race itself is an independent predictor of non-donation, or merely a surrogate marker of socioeconomic status and measures of access to health care. The impact of household income, education level, residence (urban vs. rural) and distance to donor hospital on deceased donor consent rates have not been fully evaluated. Our hypothesis is that lower household income, lower education level, rural residence and longer distance to the donor hospital will be strong predictors of non-donation. Furthermore, we hypothesize that when controlling for these SES variables and measures of access to healthcare, African American race will cease to be a significant predictor of organ donation.
Compared to the general Alabama population, there were higher percentages of males and African Americans among potential donors whose families were approached for organ donation consent. Potential donors were also older than the general population and less likely to be married. Consent was obtained from 49% (637/1292) of the decedent's families over the 4-year study period. Although 38% of the 4.7 million people living in the state of Alabama are registered organ donors(18), only 18% (229/1292) of the decedents approached for consent were registered organ donors. (Table 1)
Increasing household income was associated with consent for organ donation (Figure 1). For every $10,000 increase in household income, the odds ratio of obtaining donor consent was 1.33 (95% CI 1.19 – 1.48, p<0.001). Stratified by race, however, household income was a predictor of organ donor consent only in Caucasians. There is a clear stepwise increase in organ donation as a function of increasing household income in Caucasians although a similar trend was not observed in African Americans. (Figure 1) Household income was higher in Caucasian families in which consent was obtained compared to decedents in which consent was not obtained ($35,305 vs. $33,058; p=0.001). In contrast, household income was not significantly different between African American families in which consent was obtained compared to those in which consent was not obtained ($25,147 vs. $26,137; p=0.90).
A similar number of potential donors were high school graduates (74.9% vs. 75.3%, p=0.28) but fewer potential donors had obtained a bachelors degree (12.0% vs. 19%, p<0.0001; Table 1) compared to the general Alabama population. Stratified by race, there was no significant difference in the proportion of African Americans compared to Caucasians with a high school diploma (71.6% vs. 77.2%, p=0.06) or who had completed at least some college (26.9% vs. 32.1%, p=0.09).
Increased education was a significant predictor of organ donation. There was no significant difference between those decedents who had completed high school compared to those without a high school diploma whereas the odds ratio of organ donation for decedents who had completed some college compared to those without a high school diploma was 1.76 (95% CI 1.31 – 2.37, p<0.001). (Figure 2)
The highest consent rates were observed in the never married (55.5%) and divorced groups (54.5%), followed by the married group (45.5%), while the lowest consent rates were observed in the widowed group (35.9%). There was a negative relationship between married decedents and organ donor consent. Compared to decedents who were not married (including never married, widowed and divorced), the odds ratio of organ donor consent in married decedents was 0.75 (95% CI 0.56 – 0.92, p=0.013). The negative association between married decedents and organ donor consent was similar between African Americans and Caucasians.
Significantly fewer decedents whose families were approached for organ donation resided in a rural location (34.7% vs. 45.0%, p<0.0001; Table 1). Organ donor consent was significantly increased in decedents with an urban residential home address compared to a rural residential home address (OR 1.55, 95% CI 1.16 - 2.08, p=0.0019; Figure 3). The association between urban residence and organ donation was similar between Caucasians and African Americans. However, only 16.1% of African American decedents approached for organ donation consent had a rural address compared to 45.1% of Caucasians.
Consent for organ donation was strongly correlated with the distance between the decedent home residence and donor hospital. The distance to the donor hospital was less than 100 miles in 143 decedents, between 100 and 200 miles for 771 decedents, and over 200 miles in 279 decedents. For every 100-mile interval further away from the donor hospital, the odds ratio of donor consent was 0.59 (95% CI 0.47 – 0.75, p<0.0001). The impact of distance between home address and the donor hospital on organ donor consent was similar between African Americans and Caucasians. However, African American decedents approached for organ donation were much more likely to live closer to the donor hospital (p<0.0001). For example, 15% of African Americans lived within 100 miles of the donor hospital compared to only 10% of Caucasians. In contrast, only 15% of African Americans lived over 200 miles from the donor hospital compared to 28% of Caucasians.
A multivariable analysis was performed to measure the strength of effect of race on organ donation when controlling for SES, education, marital status, residence (urban vs. rural) and distance to donor hospital. The model also controlled for other decedent demographics and organ center practices that previous work had demonstrated to be significantly associated with organ donor consent.(21) Caucasian race, urban residence, education beyond a high school diploma and shorter distance to hospital were significantly associated with organ donor consent on multivariable analysis. (Table 2) Surprisingly, household income was not significant when controlling for other variables. Models were developed examining income as a categorical predictor (quintiles and $10,000 increments) as well as a continuous variable, with no change in association.
On univariate analysis, the odds of donor consent in Caucasians compared to African Americans was 2.76 (95% CI 2.17 – 3.57, p<0.001). When controlling for SES and access to healthcare variables, the odds of donor consent in Caucasians compared to African Americans increased to 4.36 (95% CI 2.88 – 6.61, p<0.001). (Table 2)
There have been nationally sponsored, large-scale efforts to disseminate “best practice” organ donor approaches to decedents and their families, which have manifested in significant increases in donor consent and recovered organs in the past decade.(22) Unfortunately, organ donation recently has peaked in the US and even slightly decreased in the past 2 years.(23) Examination of nationally available organ procurement data suggests that organ donation goals are being met in most Caucasian decedents but a large disparity in organ donation exists in African American decedents.(23) One of the largest sources of unrealized potential organ donors is African American decedents.
Large qualitative research efforts have investigated the underlying reasons for why African American race is associated with non-donation.(24) Certain attitudes and beliefs are prevalent among African Americans unwilling to donate organs. A cultural mistrust of the medical system has been demonstrated to be a dominant non-donation attitude,(5, 25, 26) especially in older African Americans(27). Central to this concern is a fear that physicians will not do all they can to save the life of those identified as potential organ donors.(25, 28, 29) Other common beliefs impacting donation is that organ allocation is unfair(4, 29) and that there is a black market for organ donation(4). Such attitudes and beliefs may be strongly influenced in African Americans living in the South with knowledge of the Tuskegee Syphilis Study, arguably the most infamous biomedical research study in US history.(30)
Although African American race is statistically a strong predictor of non-donation,(1-5) African American race is also a well-established predictor of lower SES and poorer access to health care.(6-17) We hypothesized that when controlling for SES and access to health care, decedent African American race would fail to be a significant predictor of organ donation. Much to our surprise, the data seems to suggest exactly the opposite. SES and access to healthcare variables were actually masking the magnitude of effect of decedent race on organ donor consent. When controlling for SES and access to healthcare, the odds ratio of donor consent in Caucasians compared to African Americans increased from 2.76 to 4.36. Our multivariable model (Table 2) demonstrates that decedent race was the most highly weighted variable. This observation is partially explained by African Americans being more likely to have an urban residence address and live closer to the donor hospital compared to Caucasians. However, there remains unknown but important factor(s) associated with both African American race and obtaining organ donor consent. Contrary to our hypothesis, this factor does not appear to be household income, marital status, education level or disparities in healthcare access. Perhaps most surprising is the nearly identical household income in African American donors ($25,147) and non-donors ($26,137). The association between increasing household income and donor consent was observed only in Caucasians. (Figure 1) One interesting phenomena observed in the lowest income quintile group of Caucasians was a higher donation rate than in the next 2 quintile groups (Figure 1). Unfortunately, this study does not provide insights as to why this is the case although this interesting phenomena deserves further study.
As with all retrospective database studies, this investigation has limitations. First, the study population consists of decedents who died in Alabama and there were not significant number of Hispanics or Asians for analysis. Second, only 4% of the decedent families were approached by an African American requesting consent limiting the ability to measure the effect of requestor race on donor consent. Third, SES was calculated from the US census tract data and is limited by the data collection practices and assumptions used by the US Census Bureau(19). Fourth, some of the variables considered may be statistically similar such as shorter distance to donor hospital and urban residence. Finally, obtaining organ donor consent from the decedent's family is a complex multistep psychosocial process; the experiences, attitudes and decisions made by the families represented in this study may or may not be generalizable.
In conclusion, certain decedent SES factors and measures of access to healthcare variables are significantly associated with organ donor consent. Controlling for statistically significant SES and access to healthcare variables markedly increased the strength of effect of race on organ donor consent. We interpret this result to indicate that there remains unknown but important factor(s) associated with both race and obtaining consent that must be isolated and determined whether this factor(s) is modifiable.
Consistent with the referral practices of The Routine Death Notification Legislation (42 CFR Part 482)(31), hospitals that receive Medicare funding are mandated to notify the designated organ procurement organization of every death or imminent death. Organ donor referral data was obtained from the Alabama Organ Center electronic database. Ethics approval for this study was obtained from the University of Alabama Institutional Review Board Protocol #N100301001.
Between January 1, 2006 and December 31, 2009, the Alabama Organ Center received 3423 notifications. After a standardized telephone screening by an organ procurement specialist, the decision was made to request consent for organ donation from families of 1292 decedents (38%), who form the study population. The reasons for not approaching families of the remaining 2131 decedents for consent were that the potential candidate was not brain dead (including those failing to satisfy locally established guidelines for donation after cardiac arrest opportunities - 32%), sepsis/ multisystem organ failure (15%), medical concerns (14%), malignancy (13%), advanced age (10%), decedent death prior to referral process (4%), communication difficulties with referring hospital (4%), and other (8%). During this study period, the Alabama Organ Center did not practice first person consent. Written consent was required from the decedent's family in all cases.
We examined the following measures of SES and access to healthcare: household income, urban vs. rural residence as defined by the Alabama Rural Health Association(32), education and distance to donor hospital. Death certificates were obtained from the Alabama Department of Public Health Records(33) for the 1292 decedents for whom familial consent was requested. Specific data obtained include decedent's home address, education level, marital status, cause of death and hospital in which the patient was pronounced dead. Using the decedent's residence address and US Census Data(19), the median household income for the census block containing the decedent's address was determined. Finally, the distance between the decedent's home residence and death hospital was calculated.
The dependent variable of interest in this study was familial consent for organ donation, as identified in the Alabama Organ Center electronic organ donor database. In addition to decedent race, independent variables examined in this study include household income, residence (urban vs. rural), education level and distance to the donor hospital. Examination of the data began by examining measures of central tendency (sample mean and median) as well as measures of dispersion (variance, standard deviations). The primary outcome variable, obtaining familial consent for organ donation, is dichotomous and, therefore, the analytic approaches focused upon both Pearson Chi-square analyses and logistic regression. Crude unadjusted odds ratios were initially calculated for each variable and compared between Caucasians and African Americans. To identify the most parsimonious model that could explain the variation in the log-odds of obtaining consent, multivariable logistic regression models were constructed. Measure of association is summarized with odds ratios with corresponding 95% confidence intervals. For all inferences, the probability of a Type I error (α) was set to 0.05. All analyses were conducted using the SAS 9.2 (Cary, NC).
Special thanks to Walt Montgomery from the Alabama Organ Center for donor data retrieval.
1Division of Abdominal Transplant Department of Surgery University of Alabama at Birmingham, DuBay-Design, Research, Writing Gray-Design, Writing Eckhoff-Design, Writing
2Biostatistics Division School of Public Health University of Alabama at Birmingham, Redden-Analysis
3Graduate Studies in Public Administration Department of Government University of Alabama at Birmingham, Akhlaque-Analysis
4Preventive Medicine Division Department of Medicine University of Alabama at Birmingham, Fouad-Design, Writing
5Department of Social and Behavioral Health School of Medicine Virginia Commonwealth University, Siminoff-Design, Writing
6Dept. of Behavioral and Community Health School of Public Health University of Maryland, College Park, Holt-Design, Writing
7Health Behavior Division School of Public Health University of Alabama at Birmingham, Kohler-Design, Writing
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.