Race, Ethnicity, and Socioeconomic Status Are Associated With Prolonged Time to Treatment After a Diagnosis of Colorectal Cancer: A Large Population-Based Study

Abstract

See editorial on page 1022. See editorial on page 1022. There has been a downward trend in colorectal cancer (CRC) incidence and mortality in the United States since the 1980s; however, racial, ethnic, and socioeconomic disparities persist.1Siegel R.L. et al.CA Cancer J Clin. 2020; 70: 145-164Crossref PubMed Scopus (761) Google Scholar These disparities are the result of differences in CRC risk factors, screening, diagnosis, and quality of care.2Edwards B.K. et al.Cancer. 2010; 116: 544-573Crossref PubMed Scopus (1381) Google Scholar,3Lansdorp-Vogelaar I. et al.Cancer Epidemiol Biomarkers Prev. 2012; 21: 728-736Crossref PubMed Scopus (101) Google Scholar Although many studies have demonstrated a role of race, ethnicity, and socioeconomic status (SES) in CRC screening and diagnosis, the impact of these factors on time to treatment after CRC diagnosis is understudied. Time to treatment after CRC diagnosis impacts mortality, survival, and quality of life.4Lee Y.H. et al.PLoS One. 2019; 14e0210465Crossref PubMed Scopus (0) Google Scholar,5Visser M.R. et al.J Surg Oncol. 2006; 93: 571-577Crossref PubMed Scopus (55) Google Scholar Thus, we aimed to use national cancer registry data to study the impact of race, ethnicity, and SES on time to treatment among individuals with CRC in the United States. Because access to timely care is challenging for the medically underserved, exploring this relationship may help reduce disparities in CRC outcomes. We performed a cross-sectional analysis using data from the National Cancer Institute Surveillance, Epidemiology, and End Results (SEER) database. Our study population included all individuals in SEER 18 (18 US regions) ages 20 to 79 years with a histologically confirmed diagnosis of colon or rectal cancer between January 1, 2000 and December 31, 2016. Individuals were excluded as detailed in Supplementary Figure 1. The primary outcome was the diagnosis-to-treatment interval (DTI), defined as the number of months between CRC diagnosis and initiation of first CRC treatment (surgery, radiation, or chemotherapy). We created 2 DTI categories: ≤1 month and >1 month. DTI > 1 month was considered prolonged DTI, consistent with data that a delay of over 30 days between a confirmed diagnosis of CRC and initiation of treatment is associated with 1.5-fold higher risk of death.4Lee Y.H. et al.PLoS One. 2019; 14e0210465Crossref PubMed Scopus (0) Google Scholar We used multivariable logistic regression to determine factors associated with prolonged DTI and assessed the interaction between SES and race/ethnicity on DTI, controlling for relevant confounders. Additional details are provided in the Supplementary Materials. Supplementary Table 1 summarizes the demographic and clinical characteristics of the 330,988 CRC cases included in the study. Average DTI was 0.6 months (SD, 1.0). The results of the multivariable logistic regression analyses are shown in the top part of Table 1. There was a significant interaction between race/ethnicity and SES (P < .001). Within each SES strata, racial/ethnic minority subgroups were significantly more likely to have prolonged DTI than non-Hispanic whites (NHW; P < .05) (Table 1, bottom). The only exceptions were medium and high SES non-Hispanic American Indian/Alaska Natives (NHAIAN). Differences from NHW were greatest in the lowest SES groups. For example, among Hispanics with low SES, the odds of prolonged DTI were 78% higher than in NHW with low SES (adjusted odds ratio, 1.78; 95% confidence interval, 1.69–1.88). However, there was a 26% higher odds of prolonged DTI for high SES Hispanics compared with high SES NHW (adjusted odds ratio, 1.26; 95% confidence interval, 1.17–1.36). There was a similar pattern for non-Hispanic Asian/Pacific Islanders (NHAPI) and NHAIAN but not for non-Hispanic blacks (NHB).Table 1Results of Main Effects Model and Interaction Model Evaluating Associations With Diagnosis-to-Treatment Interval (n = 330,988)Main effects modelUnadjusted odds ratio (95% CI)Adjusted odds ratio (95% CI) main effects modelAdjusted odds ratio (95% CI) interaction modelaThe multivariable main effects model and interaction models were controlled for sex, age, year of diagnosis, stage at diagnosis, urban/rural location, race/ethnicity, SES, and race/ethnicity × SES.Sex Male1.001.001.00 Female0.81 (0.79, 0.83)0.81 (0.79, 0.83)0.81 (0.79, 0.83)Age, y 20–290.55 (0.48, 0.64)0.48 (0.41, 0.56)0.48 (0.41, 0.56) 30–390.68 (0.63, 0.73)0.63 (0.59, 0.68)0.63 (0.59, 0.68) 40–490.78 (0.75, 0.81)0.74 (0.71, 0.77)0.74 (0.71, 0.77) 50–590.96 (0.93, 0.98)0.93 (0.90, 0.96)0.93 (0.90, 0.96) 60–69 (ref.)1.001.001.00 70–790.91 (0.88, 0.93)1.00 (0.97, 1.03)1.00 (0.97, 1.03)Year of diagnosis Per 5-y increments1.33 (1.32, 1.35)1.32 (1.30, 1.33)1.32 (1.30, 1.33)Stage at diagnosis Localized (ref.)1.001.001.00 Regional1.15 (1.12, 1.18)1.16 (1.13, 1.19)1.16 (1.13, 1.19) Distant1.30 (1.26, 1.33)1.30 (1.26, 1.34)1.30 (1.26, 1.34)Setting of residence Rural1.001.001.00 Urban1.42 (1.36, 1.48)1.35 (1.29, 1.41)1.33 (1.27, 1.39)Race/ethnicity NHW (ref.)1.001.00---- NHB1.39 (1.35, 1.44)1.31 (1.26, 1.35) Hispanic (all races)1.66 (1.61, 1.71)1.55 (1.50, 1.60) NHAPI1.32 (1.28, 1.38)1.29 (1.24, 1.34) NHAIAN1.62 (1.41, 1.85)1.39 (1.18, 1.65)SES Low (ref.)1.001.00---- Medium0.86 (0.84, 0.89)0.90 (0.87, 0.92) High0.82 (0.80, 0.84)0.87 (0.84, 0.89)Interaction model for SES and race/ethnicityLow SESMedium SESHigh SESOdds ratios for race/ethnicity within strata of SES NHWRefRefRef NHB1.35 (1.29, 1.42)1.31 (1.22, 1.40)1.37 (1.25, 1.49) Hispanic1.78 (1.69, 1.88)1.50 (1.42, 1.59)1.26 (1.17, 1.36) NHAPI1.43 (1.31, 1.56)1.38 (1.29, 1.48)1.17 (1.10, 1.24) NHAIAN1.68 (1.31, 2.16)1.30 (0.98, 1.73)1.11 (0.75, 1.64)Odds ratios for SES within strata of race/ethnicity NHWRef0.93 (0.90, 0.97)0.94 (0.90, 0.97) NHBRef0.91 (0.84, 0.97)0.95 (0.87, 1.04) HispanicRef0.79 (0.74, 0.84)0.66 (0.61, 0.72) NHAPIRef0.90 (0.81, 1.00)0.77 (0.69, 0.84) NHAIANRef0.72 (0.49, 1.06)0.62 (0.39, 0.98)Odd ratios that are statistically different from 1 have P < .05. SES level is defined as YOST, a composite score provided by National Cancer Institute SEER that is constructed from 7 variables (median household income, median house value, median rent, percent below 150% of poverty line, education index, percent working class, and percent unemployed) to measure different aspects of the SES of a census tract. Then census tracts are categorized into SES tertiles (low, medium, and high) based on the YOST score. CI, confidence interval.a The multivariable main effects model and interaction models were controlled for sex, age, year of diagnosis, stage at diagnosis, urban/rural location, race/ethnicity, SES, and race/ethnicity × SES. Open table in a new tab Odd ratios that are statistically different from 1 have P < .05. SES level is defined as YOST, a composite score provided by National Cancer Institute SEER that is constructed from 7 variables (median household income, median house value, median rent, percent below 150% of poverty line, education index, percent working class, and percent unemployed) to measure different aspects of the SES of a census tract. Then census tracts are categorized into SES tertiles (low, medium, and high) based on the YOST score. CI, confidence interval. When we examined the association between SES and DTI stratified by race/ethnicity (Table 1, bottom), findings were similar for Hispanics, NHAPI, and NHAIAN—as SES increased, the likelihood of prolonged DTI decreased. High SES Hispanics and NHAPI had 34% and 23% lower odds, respectively, of prolonged DTI compared with low SES Hispanics and NHAPI. For NHB, SES tier did not significantly impact DTI. Overall, differences in prolonged DTI by race/ethnicity and SES were most pronounced among NHAIAN, Hispanics, and NHAPI. Older age, male sex, advanced stage at diagnosis, later year of diagnosis, and urban residence were also associated with prolonged DTI (Table 1, top). We used national cancer registry data to demonstrate that race/ethnicity and SES are significantly associated with time to CRC treatment in the United States. Our findings are consistent with a well-documented synergistic relationship between race/ethnicity and SES in health outcomes.6Williams D.R. Ann N Y Acad Sci. 1999; 896: 173-188Crossref PubMed Scopus (973) Google Scholar When racial/ethnic disparities in health are adjusted for SES, differences often diminish, indicating that SES accounts for some differences by race/ethnicity.6Williams D.R. Ann N Y Acad Sci. 1999; 896: 173-188Crossref PubMed Scopus (973) Google Scholar There is also an independent role of race/ethnicity on the primary outcome, however. Even at the same SES level, minorities received treatment later than NHW. This interaction was strongest for Hispanics, NHAPI, and NHAIAN, suggesting that improving SES disparities (eg, income, education) in these groups may improve time to treatment. In contrast, race/ethnicity seems to be more influential than SES for NHB, suggesting we must focus on something inherent to how NHB are medically managed to close the time to treatment gap between NHB and NHW. There are several limitations to our study. First, DTI values are not exact and are capped because of limitations of the SEER database. However, overestimations or underestimations of DTI would be uniform across all groups and unlikely affect model results. Second, SEER does not include patient-level data on comorbidities, distance to referral center, healthcare utilization, provider recommendations, or prior screening that might impact DTI. Despite these limitations, the study has many strengths. We used a large population-level dataset with broad representation of racial/ethnic and SES groups to address an important question about cancer care and outcomes. Our study validates the results of prior small studies regarding the relationship between race/ethnicity and DTI but demonstrates these disparities on a national level.7Jones L.A. et al.Ann Epidemiol. 2017; 27: 731-738Crossref PubMed Scopus (19) Google Scholar Finally, our findings go beyond focusing on either race/ethnicity or SES to explore the interaction between these 2 factors on DTI, which is done infrequently but is important given the correlation between race/ethnicity and SES. In conclusion, multiple factors are associated with prolonged time to treatment for CRC, including a complex interaction between race/ethnicity and SES. Demographic groups in our study that were more likely to experience prolonged DTI, including NHB and NHAIAN, also have high CRC mortality. Differences in screening and treatment explain a large proportion of racial/ethnic CRC disparities in mortality.3Lansdorp-Vogelaar I. et al.Cancer Epidemiol Biomarkers Prev. 2012; 21: 728-736Crossref PubMed Scopus (101) Google Scholar As screening gaps narrow, CRC treatments improve, and as wait times for cancer care in the United States lengthen, minimizing differences in time to treatment becomes increasingly important to reduce disparities.2Edwards B.K. et al.Cancer. 2010; 116: 544-573Crossref PubMed Scopus (1381) Google Scholar Thus, our findings suggest that physicians must be more vigilant about timely care for minority and low-income patients diagnosed with CRC. Strategies include first developing an infrastructure to identify patients at high risk of care delays and patient-level barriers to receiving timely care. Then, implementing patient navigation to assist with overcoming barriers has shown promise in reducing disparities in cancer care.8Rodday A.M. et al.Cancer. 2015; 121: 4025-4034Crossref PubMed Scopus (39) Google Scholar Finally, action should be taken to address discriminatory practices or structural racism that contribute to differences in care recommendations or care receipt in underserved populations. Aileen Bui, MD (Conceptualization: Equal; Investigation: Equal; Methodology: Equal; Visualization: Lead; Writing – original draft: Lead; Writing – review & editing: Equal). Liu Yang, MD, MPH (Conceptualization: Equal; Formal analysis: Lead; Investigation: Equal; Methodology: Equal; Visualization: Equal; Writing – original draft: Supporting; Writing – review & editing: Equal). Anthony Myint, MD (Conceptualization: Supporting; Writing – review & editing: Supporting). Folasade Popoola May, MD, PhD, MPhil (Conceptualization: Equal; Formal analysis: Supporting; Investigation: Equal; Methodology: Equal; Supervision: Lead; Visualization: Equal; Writing – original draft: Supporting; Writing – review & editing: Equal). We performed a cross-sectional analysis using data from the National Cancer Institute (NCI) Surveillance, Epidemiology, and End Results (SEER) database. The SEER population is comparable to the US population with respect to income level (15.3% vs 15.1% below poverty level) and education (14.2% vs 13.0% with less than high school diploma), but has a higher proportion of foreign-born persons (17.9% vs 13.2%).1National Cancer InstituteSurveillance, epidemiology, and end results program.https://seer.cancer.gov/Google Scholar The SEER database also has higher representation of minorities and those with economic disadvantage than the overall population due to purposeful oversampling of epidemiologically significant population subgroups.1National Cancer InstituteSurveillance, epidemiology, and end results program.https://seer.cancer.gov/Google Scholar We used SEER 18, which includes data from 2000 to 2016 for 18 US regions and covers approximately 27.8% of the US population. SEER 18 offers broad geographic and demographic coverage of the SEER registries as well as granular data on participant race, ethnicity, and SES. Our study population included all individuals in the SEER program database age 20 to 79 with a histologically confirmed diagnosis of colon or rectal cancer between January 1, 2000 and December 31, 2016. Colon and rectal cancer diagnoses were based on the International Classification of Diseases for Oncology 3rd edition (ICD-O-3).2Fritz A, Percy C, Jack A, et al. International classification of diseases for oncology / editors, April Fritz ... [et al.]. In. 3rd ed ed. Geneva: World Health Organization; 2000.Google Scholar We excluded individuals with a history of other cancers, those who were missing data for time to treatment initiation, and those who were reported only through death certificate or autopsy as records may be incomplete or inaccurate (Supplementary Figure 1). The primary outcome was the diagnosis-to-treatment interval (DTI), defined as the number of months between CRC diagnosis and initiation of first CRC treatment (either surgery, radiation, or chemotherapy). SEER provides this variable in years and months and caps DTI at 24 months.1National Cancer InstituteSurveillance, epidemiology, and end results program.https://seer.cancer.gov/Google Scholar While there are no national guidelines or benchmarks for optimal DTI, prior studies examining the impact of DTI on CRC outcomes have defined prolonged DTI as greater than 30 days.3Lee Y.H. Kung P.T. Wang Y.H. et al.Effect of length of time from diagnosis to treatment on colorectal cancer survival: A population-based study.PLoS One. 2019; 14e0210465Crossref PubMed Scopus (32) Google Scholar, 4Zarcos-Pedrinaci I. Fernandez-Lopez A. Tellez T. et al.Factors that influence treatment delay in patients with colorectal cancer.Oncotarget. 2017; 8: 36728-36742Crossref PubMed Google Scholar, 5Bleicher R.J. Ruth K. Sigurdson E.R. et al.Time to Surgery and Breast Cancer Survival in the United States.JAMA Oncol. 2016; 2: 330-339Crossref PubMed Scopus (222) Google Scholar In addition, median DTI for all cancers in the United States is 31 days.6Bilimoria K.Y. Ko C.Y. Tomlinson J.S. et al.Wait times for cancer surgery in the United States: trends and predictors of delays.Ann Surg. 2011; 253: 779-785Crossref PubMed Scopus (171) Google Scholar Therefore, we created two DTI categories: 1) ≤ 1 month and 2) >1 month. In our analyses, DTI >1 month was considered prolonged DTI. Additional subject demographic and clinical characteristics included age at CRC diagnosis, year of diagnosis, sex, race, ethnicity, SES, urban or rural setting of residence, and stage at CRC diagnosis. For race and ethnicity, we created a single variable (race/ethnicity) with mutually exclusive categories: non-Hispanic white (NHW), non-Hispanic black (NHB), non-Hispanic Asian/Pacific Islander (NHAPI), non-Hispanic American Indian/Alaska Native (NHAIAN), and Hispanic. For SES, we used the NCI’s census tract-level SES index as a proxy for individual-level SES. The SES index is derived from seven SEER variables that measure different aspects of SES for a census tract: median household income, median house value, median rent, percent below 150% of poverty line, education index, percent working class, and percent unemployed.7Jones L.A. et al.Ann Epidemiol. 2017; 27: 731-738Crossref PubMed Scopus (19) Google Scholar,8Rodday A.M. et al.Cancer. 2015; 121: 4025-4034Crossref PubMed Scopus (39) Google Scholar After the SES scores are generated for each year, census tracts are categorized into tertiles (low, medium, and high SES) with equal populations in each tertile across the entire SEER catchment area. For stage at CRC diagnosis, we used the NCI SEER Historic Stage A variable to categorize tumors into one of four categories: unstaged, localized, regional, and distant. Localized disease was defined as disease confined to the colon/rectum or with intraluminal extension but no lymph node involvement at the time of diagnosis. Regional disease was defined as direct extension into nearby structures and/or involvement of ≤3 lymph nodes. If there was further contiguous extension or distant organ or lymph node metastases, the tumor was categorized as distant.9Young JL Jr RS, Ries LAG, Fritz AG, Hurlbut AA (eds). SEER Summary Staging Manual - 2000: Codes and Coding Instructions. In. NIH Pub. No. 01-4969 ed. Bethesda, MD: National Cancer Institute; 2001.Google Scholar We used SEER∗Stat version 8.3.6 to generate the study population and collect demographic and clinical characteristics. We calculated the rate of prolonged DTI by age, sex, race/ethnicity, SES, year of diagnosis, urban/rural setting of residence, and disease stage. We then used Chi-square tests to compare prolonged DTI rates by these subject characteristics. We used univariate and multivariable logistic regression to determine predictors of prolonged DTI, with particular attention to the impact of race/ethnicity and SES on DTI. We first examined crude associations between each covariate and prolonged DTI in univariate models and then included variables suggesting significant unadjusted association (P<.05) in the multivariable model. We assessed the interaction between SES and race/ethnicity by adding an interaction term to the multivariable main effects model with the intention to include the interaction term in the final model if interactions were significant at the P<.05 level. Statistical analyses were performed with SAS version 9.4 (SAS institute, Cary, NC). The study was deemed exempt by the UCLA Institutional Review Board at UCLA. NOTE. Adjusted mean time to treatment (months): Hispanics=0.62, NHAPI=0.57, NHAIAN=0.57, NHB=0.54, NHW=0.49 Results reported as N (%). DTI, diagnosis-to-treatment interval; SEER, Surveillance, Epidemiology, and End Results; NHW, non-Hispanic white; NHB, non-Hispanic black; NHAPI, non-Hispanic Asian/Pacific Islander; NHAIAN, non-Hispanic American Indian/Alaska Native.