Chemotherapy provided after surgical resection is beneficial for many with non-localized colon cancer. Survival benefits are large for many with advanced, but non-metastasized, stage III disease and smaller, but still significant for some with stage II disease [1-3]. There are probably even small survival benefits of chemotherapy for people with metastasized, stage IV colon cancer, in addition to its clear palliative benefits [4]. Yet nearly half of all such people in the United States do not receive chemotherapy with the intention to either cure or comfort [5]. Moreover, chemotherapy access may be affected by social and economic characteristics such as race or ethnicity, and income and health insurance adequacy. African Americans (AA), who are also poorer and less adequately insured, on average, than non-Hispanic white Americans (NHWA), seem particularly disadvantaged [6,7]. They are less likely to receive adjuvant chemotherapy and this can explain much of their survival disadvantage [5,8-10].

Social and health care systemic factors seem most implicated in this racial divide for the following reasons. First, clinical studies of equal-access health care systems within the US have consistently observed similar chemotherapy and survival rates among AAs and NHWAs with colon cancer [11-14]. Second, colon cancer chemotherapy rates seem to be much higher in Canada than in the US [15,16]. Third, recent population-based studies of colon cancer care in California and Ontario suggested that people of color, including black people of various ethnic backgrounds, with colon cancer are much better served in California [16-18]. Moreover, health insurance inadequacies in the US versus universal coverage of medically necessary care in Canada largely explained the between-country divide. Being uninsured or insured by Medicaid has been consistently found to be less adequate than being insured by Medicare or by a private insurer [5,17,18]. Fourth, this field’s research syntheses allow for the conclusion that socioeconomic status (SES) and treatment differences largely explain observed AA-NHWA survival differences [5,8-10]. The fact that they have largely, but not completely accounted for racial differences with social-systemic factors suggests the plausibility of at least one of their causes being biological. For example, AA and NHWA patients may differ on oncogene-based tumor characteristics that affect survival directly or indirectly, through their effects upon clinicians’ decisions to prescribe chemotherapies. Modest equivocal differences between racialized groups on colon tumor grade [8], however, suggest that, though possible, this explanation for the differential uptake of chemotherapy and survival differences between AAs and NHWAs is improbable.

In attempting to account for SES, this field’s population-based studies have typically used census data to define low-income neighborhoods. However, their lowest income neighborhoods only ranged from 10% to 20% poor. They therefore had limited power to study colon cancer care among the “truly disadvantaged” [19] who live, for example, in America’s poorest neighborhoods where 30% to 40% or more of households have incomes below the poverty line [19-21]. Consequently, they probably did not account for residual confounding. That is, their AA participants probably had substantially lower incomes than their NHWA counterparts, even in the lowest income neighborhoods studied [22,23]. This field also seems limited by its focus on the mere main effects of race/ethnicity, SES and health insurance. Recognizing that it is important to analyze different racial/ethnic minority groups uniquely, our previous research on breast cancer care among Mexican American women leads us to anticipate complex interactions of ethnicity, poverty and health insurance status [24-26]. We were recently presented with a serendipitous opportunity to systematically replicate that notion among AAs with colon cancer.

We oversampled people with colon or breast cancer in the highest poverty neighborhoods of California for other primary studies [17,18,24]. This necessarily meant that we also oversampled AAs. Secondary to our original study’s intentions we had the opportunity to compare the chemotherapy and survival experiences of AAs and NHWAs with colon cancer while providing more control for residual confounding by SES than previous studies had. This field’s historical-theoretical context strongly suggests multiplicative, rather than additive disadvantages of being poor and inadequately insured among AAs. That is, the combined effects of being poor and of being uninsured or underinsured on colon cancer care are probably worse for AAs than for NHWAs. Therefore, we hypothesize a 3-way ethnicity by poverty by health insurance interaction on the receipt of chemotherapy. Secondarily, we hypothesize that this complex interaction, along with the main effects of poverty, health insurance adequacy and treatment access, will now completely explain any observed AA-NHWA survival differences.

Six thousand, three hundred people who were diagnosed with colon cancer between 1996 and 2000 were randomly selected from the California Cancer Registry (CCR) that was joined to the 2000 census by census tracts [17,18,27]. The original sample was also stratified by SES (extremely poor neighborhoods where 30% or more of the households were poor, 5% to 29% or less than 5% poor) and place (large or smaller urban or rural). This secondary analysis excluded localized, stage I cancers for which chemotherapy is not indicated as well as people of other racial or ethnic backgrounds [28,29]. This study then analyzed colon cancer care among 459 AAs (also non-Hispanic) and 3,001 NHWAs with stage II to IV disease.

A logistic regression model was used to test the 3-way ethnicity (AA vs. NHWA) by poverty (extremely poor or not) by health insurance adequacy (uninsured or Medicaid insured vs. privately or Medicare insured) interaction in predicting binary chemotherapy receipt [30]. Odds ratios (OR) and 95% confidence intervals (CI) were estimated from regression statistics. AA study participants were younger (M = 67.4, SD = 13.9), on average, than the NHWA participants (M = 70.9, SD = 13.0), F (1, 3,458) = 28.39, p < .05. Therefore, all rates were internally age-adjusted and reported as percentages (rates per 100). Chemotherapy rates were also stage-adjusted as clinical indication and prescription rates differ significantly for stage II, III and IV disease. Then standardized rate ratios (RR) were reported for critical between-group comparisons with 95% CIs derived from the Mantel-Haenszel χ² test. Standardized rate difference (RD) indices were also used to further aid the interpretation of practical-clinical significance. The hypothesis that the main and interacting effects of ethnicity, poverty, health insurance and chemotherapy access would completely explain any observed AA-NHWA survival difference was tested with unadjusted and adjusted Cox proportional hazards regression models [31]. All participants were minimally followed for 10 years, from the date of their diagnosis until January 1, 2011. Hazard ratios (HR) and 95% CIs were estimated from regression statistics. Other methodological details have been reported [17,18,27]. This study was reviewed and cleared by the University of Windsor research ethics board.

First, this field’s established ethnicity-outcome relationships were systematically replicated among historical cohorts of AAs and NHWAs with non-localized colon cancer. This study’s AA participants were about 25% less likely to receive chemotherapy and about 25% more likely to die over the 10 years following their diagnosis than were its NHWA participants. Next, this field’s well-known main predictive effects were replicated. AAs, those who lived in extremely poor neighborhoods and those who were inadequately insured were all significantly less likely than NHWAs, the less poor or the adequately insured to receive chemotherapy and so more likely to die sooner. Then an ethnicity by poverty by health insurance interaction on chemotherapy receipt was discovered. This study’s central original finding, the 3-way interaction is evidence of the multiplicative disadvantage experienced by extremely poor and inadequately insured AAs seeking colon cancer care. Among relatively advantaged people, AA and NHWA chemotherapy rates differ little, if at all (< 2%). However, among the multiply disadvantaged extremely poor and inadequately insured, the AA-NHWA chemotherapy RD (> 20%) could, we think, be fairly characterized as huge in terms of both its clinical and human significance. Finally, controlling for residual SES confounding by oversampling from extremely poor neighborhoods and further controlling for demographic and biological differences through mathematical modeling, we discovered that the main and interacting effects of poverty, health insurance adequacy and treatment access can completely account for the typically observed AA-NHWA survival differences. After we accounted for such social forces, the survival experiences of the AA and NHWA colon cancer patients we studied were essentially identical.

Two-way poverty by health insurance interactions have previously been observed in studies of breast and colon cancer care among NHWAs [18,24,25]. The beneficial effects of health insurance were observed to be strongest in low poverty neighborhoods. It was theorized that the effectiveness of health insurance programs was positively impacted by the availability of other key resources. In more affluent neighborhoods, where social and economic capital abounds, most people with cancer seemed able to absorb the indirect (time lost from work, recuperation, transportation and others) and direct, but additional uncovered, out-of-pocket costs of care (co-insurance and co-payments). Within high poverty neighborhoods on the other hand, relatively lacking in such capital reserves, health insurance programs seemed to be much less effective. Such extremely poor people were probably much less able to pick-up the co-insurance costs and co-payments that are prevalent in American cancer care. The 3-way ethnicity-poverty-health insurance interaction observed in this study may be thought of as an extension of the previously observed 2-way poverty-health insurance interaction. It strongly suggests that the 2-way interaction’s effect is stronger for AAs. Their relative lack of capital reserves could explain this differential ethnic effect [33,34]. We saw that this study’s oversampling from extremely poor neighborhoods seemed to fairly well control AA-NHWA differences on depth of impoverishment [22,23], at least in terms of their annual household incomes that only differed by about $1,000. However, a contemporaneous analysis found that among the poorest households in America, NHWAs typically held $10,000 worth of equity, while AAs typically had no assets at all [7]. It seems probable that their lack of capital reserves operates to further potentiate the disadvantages already experienced by extremely poor and inadequately insured AAs as they try to purchase chemotherapeutic and other colon cancer care services.

Overall, AAs with non-localized colon cancer are 25% less likely to receive indicated chemotherapy than are NHWAs with similar colon cancers. Among the extremely poor and inadequately insured AAs are 60% less likely than NHWAs to receive such care. Ethnicity, poverty and health insurance status appear to interact in such a way that the multiplicative barriers to care of being extremely poor and inadequately insured are worse for AAs than they are for NHWAs. AAs are more prevalently poor and inadequately insured, but even when depth of income-based poverty is controlled, they still have substantially fewer assets than NHWAs so they are probably much less able to absorb the indirect and direct, but uncovered, costs of colon cancer care. Policy makers ought to be cognizant of these factors as they roll out Obamacare and consider future reforms of health care in America.

The authors declare that they have no competing interests.

KMG conceptualized the study, led the study design and writing and supervised the analysis. SH performed the analysis and assisted with study design, data interpretation and writing. INL, EB, FCW and SMK assisted with study design, data interpretation and writing. KMG and INL also obtained funding. All authors read and approved the final manuscript.

The pre-publication history for this paper can be accessed here:

http://www.biomedcentral.com/1472-6963/14/133/prepub