In this large population-based series of classical HL patients in California, we found that survival was poorer for patients living in lower-SES neighborhoods at diagnosis, and for black or Hispanic patients even after adjustment for neighborhood SES. These findings of social class-associated survival disparities for this highly curable cancer underscore the importance of determining and ameliorating the underlying causes of such disparities so that all patients can benefit from the well-established and successful treatments.
Our findings are supported by previous studies that considered SES and racial/ethnic differences in survival separately (9
) and by our prior study (based on 922 subjects also included in these analyses) that considered neighborhood SES and race/ethnicity simultaneously (23
). HL patients diagnosed in the US Surveillance, Epidemiology and End Results (SEER) regions in 1987-1992 had a 5% increased risk of cancer-specific death with each quintile decrease in area-level median household income (9
); in a Brazilian HL clinical series (2001-2005), higher SES was associated with better 2-year overall survival (93% vs. 79%) (10
). By contrast, in an Austrian clinical series of HL patients (1969-2002), those of higher SES had worse failure-free survival (13
), and in a population-based Danish HL cohort (1994-2003), there was no association between socioeconomic or demographic measures and 5-year relative survival (24
). In our prior study, lower neighborhood SES and non-white race/ethnicity were associated with poorer survival in patients 15-44, but not 45-96, years of age (23
). Some of these differences in findings may be attributable to different health care systems. For example, Austria and Denmark have well-established universal health insurance systems, whereas the US provides government-supported health insurance only to those aged 65 and over, and Brazil’s universal health care system was implemented relatively recently.
At least two studies have considered HL survival in non-white racial/ethnic groups. A US survey of HL patients (diagnosed in 1970-1981), 74% of whom were under 55 years of age at diagnosis, found that blacks had a 44% higher risk of death than whites five to 10 years after diagnosis (12
). Our study of SEER data for HL diagnoses in 1983 through 1995 found that non-white young and older adult HL patients had a 40-50% increased risk of HL-specific death, depending on age and symptoms, compared to white patients (11
). However, no previous US studies have considered HL survival in Hispanics, and HL survival studies of APIs have been limited to small series (n < 50) of Taiwanese patients (25
Social disparities in survival may occur because patients with lower-SES or non-white race/ethnicity are diagnosed at a later stage of disease (27
), as found in one HL study (28
), or receive poorer cancer treatment (27
). In our study, the percentages of HL patients with lower neighborhood SES and diagnosed with stage III/IV disease were higher in blacks and Hispanics than in Asians/Pacific Islanders and whites. However, we found neighborhood SES and racial/ethnic differences in survival in patients with all stages of disease, suggesting that factors over and above stage at diagnosis influence the disparities we observed. Improvements in standards for HL treatment likely contributed to the better survival observed in more recent diagnostic years (29
). However, we did not have information to evaluate the quality of staging, which largely determines treatment (20
), or the completeness of treatment received by study patients. Thus, we could not assess how these factors may have varied by neighborhood SES or race/ethnicity and thereby contributed to our findings. Young-adult black and Hispanic patients with unknown stage of disease had markedly worse survival, suggesting that lower quality of staging in these groups may be related to poorer treatment and, consequently, worse survival. Furthermore, SES is related to having adequate health insurance, a variable not available in this study, and being uninsured or Medicaid-insured has been found to be associated with poorer survival after cancer (30
Other explanations for our observations of HL outcome disparities by neighborhood SES and race/ethnicity include differences in medical follow-up or integrated care after diagnosis, comorbidities, health behaviors, and other host factors (27
). If poor health behaviors and comorbid conditions are more prevalent in lower-SES and/or non-white HL patients, as found for patients with other cancers (27
), then these factors could increase post-treatment complications and reduce survival. Furthermore, inadequate long-term follow-up in patients could result in a delay in diagnosing and treating complications (10
). Finally, host genetic factors also may contribute to the observed survival differences by race/ethnicity. For example, racial/ethnic variation in certain immune-function genes may impact survival after HL (34
In our study, relative survival was worse in the lower neighborhood SES group than the higher-SES group in all racial/ethnic categories. With the exception of older cases at 5 and 10 years after diagnosis, we found better relative survival in females than in males, consistent with previously reported five-year relative survival estimates (1
). We also noted better overall survival for young-adult females than males, as found previously, including prior to the era of successful HL treatments (11
). While there was some variation in the magnitude of neighborhood SES disparities in relative survival over the duration of follow-up, we found that the gaps persisted for at least 15 years after diagnosis in groups defined by age, gender, and stage of disease.
Our study had a number of strengths. It used a large, population-based series of patients, including all HL patients diagnosed in California over an 19-year time period; thus, our findings are likely more generalizable than those from some prior studies of disparities in HL, as survival has been shown to be different in patients in population-based cancer registries than in patients in clinical trials (40
) or treated in comprehensive cancer centers (41
). Another strength is that survival time for our study was uniformly collected, which minimizes bias due to differential follow-up. We also used a measure of neighborhood SES shown to have adequate sensitivity for demonstrating SES gradients in HL incidence (17
) and survival (23
). Our study has the advantage of using customized SES- and race/ethnicity-specific life tables, which likely estimate the relative survival after HL more accurately than unadjusted/general life tables. Because HL incidence varies by neighborhood SES at diagnosis (17
) and higher-SES populations have longer survival, the use of general population mortality tables to determine the expected mortality for calculations of relative survival would likely overestimate the relative survival rates of these patients.
Our study also was subject to some limitations. We did not have individual-level measures of SES to consider along with our measure of neighborhood SES. While individual and neighborhood SES are associated, neighborhood SES has been shown to underestimate associations observed with individual-level SES (42
). Furthermore, area-based SES measures may reveal differences in health outcomes and risk factors not captured by individual-level SES alone, as 23 of 25 studies found a significant association between area SES measures and health outcomes after adjusting for individual-level SES (43
). Because cancer registry data lack information on potentially relevant clinical data such as prognostic serum measures (44
) or tumor characteristics (e.g., presence of Epstein-Barr virus in tumor cells (23
)), our analyses could not examine the impact of these factors on survival. Finally, our study is also subject to the effects of some potential misclassification in cancer registry data, including for race/ethnicity, although we have detected excellent overall agreement with self-reported race/ethnicity for whites and blacks, and intermediate agreement for Hispanics and Asians (45
); and for the underlying cause of death code, although this previously has been found to be between 84% and 90% accurate (47
While over 75% of all HL patients are now considered cured of their disease (2
), our data show that this remarkable clinical accomplishment has not been distributed evenly across socioeconomic and racial/ethnic groups. Rather, lower neighborhood SES and non-white patients are less likely to benefit from optimal treatment and clinical care, even among young adults, an age group for which a cancer diagnosis might be expected to provoke particularly close medical attention and follow-up care. Our findings of similar results for overall and HL-specific survival, as well as the persistence of differences in relative survival by neighborhood SES over time, suggest that neighborhood SES and racial/ethnic disparities in HL survival stem more from differences in the initial treatment and management than in the response to late complications resulting from HL. Therefore, targeted efforts to expand access to high-quality staging and treatment for lower-SES and black and Hispanic HL patients should help to bring survival in these groups up to the standard enjoyed by more privileged patients. In the meantime, efforts are needed to identify additional reasons for these marked survival differences.