Racial comparisons in treatment of rectal adenocarcinoma and survival in the military health system

Abstract

Background

Racial disparities in treatment and outcomes of rectal cancer have been attributed to patients’ differential access to care. We aimed to study treatment and outcomes of rectal cancer in the equal access Military Health System (MHS) to better understand potential racial disparities.

Methods

We accessed the MilCanEpi database to study a cohort of patients aged 18 and older who were diagnosed with rectal adenocarcinoma between 1998 and 2014. Receipt of guideline recommended treatment per tumor stage, cancer recurrence, and all-cause death were compared between non-Hispanic White and Black patients using multivariable regression models with associations expressed as odds (AORs) or hazard ratios (AHRs) and their 95% confidence intervals (CIs).

Results

The study included 171 Black and 845 White patients with rectal adenocarcinoma. Overall, there were no differences in receipt of guideline concordant treatment (AOR = 0.76, 95% CI = 0.45 to 1.29), recurrence (AHR = 1.34, 95% CI = 0.85 to 2.12), or survival (AHR = 1.08, 95% CI = 0.77 to 1.54) for Black patients compared with White patients. However, Black patients younger than 50 years of age at diagnosis (AOR = 0.34, 95% CI = 0.13 to 0.90) or with stage III or IV tumors (AOR = 0.28, 95% CI = 0.12 to 0.64) were less likely to receive guideline recommended treatment than White patients in stratified analysis.

Conclusions

In the equal access MHS, although there were no overall racial disparities in rectal cancer treatment or clinical outcomes between Black and White patients, disparities among those with early-onset or late-stage rectal cancers were noted. This suggests that factors other than access to care may play a role in the observed disparities and warrants further research.

Colorectal cancers are a common malignancy among US men and women, with most being adenocarcinoma histology (1). However, colon and rectal cancers differ in treatment, recurrence, and survival (2-7). In colon cancer, upfront surgery is recommended for locoregional tumors and adjuvant chemotherapy is recommended for advanced or high-risk tumors (2,8). For rectal cancers, neoadjuvant or adjuvant chemoradiation is recommended alongside surgery to improve outcomes in most tumors (2,3,8). Several population studies have compared outcomes in colon versus rectal cancers, finding that patients with rectal cancer generally have less favorable rates of recurrence and survival (5-7).

Because colon and rectal cancers have historically been studied together as colorectal cancer, racial disparities in rectal cancer treatment and survival are not well understood (9-17). Nevertheless, some evidence shows that Black patients are less likely than non-Hispanic White patients to receive standard treatment(s) of surgery, chemotherapy, or radiation for rectal cancers (10-13); and Black patients with rectal cancer have less favorable survival compared with non-Hispanic White patients (9,10,14-17). Recent evidence suggests racial disparities in rectal cancer may be more pronounced in patients with early-onset disease before age 50 or among those with locally advanced disease (10,18-20). Access to care may explain some, but not all, observed racial disparities in treatment and survival (11,13,19,21-24). Thus, studying treatment and survival of rectal cancer in health systems offering equal access to care may provide valuable information related to racial disparities in rectal cancers.

The US Military Health System (MHS) provides access to medical services to more than 9.6 million Department of Defense (DoD) beneficiaries each year, including active-duty service members, retirees, and their dependents (25). Beneficiaries are provided equal medical care regardless of race or other sociodemographic characteristics. Thus, the MHS provides an excellent setting for investigating whether there are any racial differences in treatment receipt and outcomes for patients diagnosed with rectal cancer in an equal access health system. This study aimed to compare Black and White patients in receipt of guideline-recommended treatment for rectal adenocarcinoma and in the outcomes of cancer recurrence and overall survival in the MHS population to better understand potential racial disparities in rectal cancer.

Methods

Data source

The Military Cancer Epidemiology (MilCanEpi) database was used for this study. It is described in detail elsewhere (26,27). Briefly, MilCanEpi contains information for patients who are diagnosed or treated for cancer in military treatment facilities in the MHS contained in the DoD cancer registry and linked administrative and medical encounter data from both military and civilian care. The MilCanEpi data were approved for use in research by the Uniformed Services University of the Health Sciences Institutional Review Board.

Study population

Eligible patients (n = 1155) included men and women aged 18 or older with a pathologically confirmed diagnosis of primary rectal adenocarcinoma (ICD-O-3 primary site code C209 and histology codes 814[x]3, 82103, 82203, 82613, 84803, and 84903) between January 1, 1998, and December 31, 2014, who had race recorded as Black or White in the MilCanEpi data. Hispanic ethnicity was also assessed in the database, and patients with Hispanic ethnicity (n = 87) were excluded because of a small number and potential effects on racial comparisons in treatment receipt or survival (10,13,15,28). Patients with unknown stage tumors (n = 52) were excluded due to the inability to define guideline-recommended treatment for those patients (29).

Study variables

Information on cancer diagnosis date, tumor stage, and tumor grade were obtained from MilCanEpi. Tumor stage was given as the American Joint Committee on Cancer (AJCC) summary stage I, II, III, or IV. Tumor grade was given as AJCC grade 1-4 (G1-G4) or unknown (Gx). Other patient information obtained from the MilCanEpi data included age at diagnosis, sex, marital status at diagnosis, and active-duty military status at diagnosis. Information on comorbid conditions in the Elixhauser index, adapted for persons with cancer (30,31), and gastrointestinal conditions including Crohn’s disease, colitis, inflammatory bowel disease, irritable bowel syndrome, diverticulitis, and high-risk diseases and conditions with symptoms similar to colorectal cancer were also captured (32).

Study outcomes

The primary study outcomes included treatment receipt, cancer recurrence, and overall survival. Treatment information was gathered from MilCanEpi and included encounters for surgery, radiation therapy, or chemotherapy. Treatment dates were consolidated using systematic procedures (27). Primary treatment was defined as any treatment type(s) initiated within 6 months of the diagnosis date (3,8). The time between cancer diagnosis and the first treatment (ie, time-to-treatment) was calculated in days, and “delay” in treatment initiation was considered to be time-to-treatment greater than 6 weeks (42 days) (33,34). Adjuvant treatment included any treatment initiated within 4 months after completion of the primary treatment (35-38). Recommendations that applied to the years of the data (ie, 1998-2014) were used to define “guideline-concordant treatment” on the basis of tumor stage as surgery for stage I, surgery and neoadjuvant or adjuvant chemotherapy or radiation for stages II and III, and chemotherapy or radiation for stage IV (31,39). Patients who received treatment other than the recommended therapy on the basis of the tumor stage were classified as having “non-concordant treatment” and were differentiated from patients with “no treatment.” Cancer recurrence was captured for patients with stage I-III tumors in the data by a diagnosis code for secondary malignant neoplasm occurring at least 60 days after the cancer diagnosis date for patients without primary treatment or 60 days after initiation of primary treatment for patients with primary treatment (40-42). Overall survival was determined for all patients in the study by indication of all-cause death and date of death in the data.

Statistical analysis

First, we examined the distribution of characteristics between non-Hispanic Black and White patients using χ² tests. Then we compared receipt of guideline recommended treatment(s) between racial groups using multinomial logistic regression models with outcome levels of “no treatment,” “non-concordant treatment,” and “guideline concordant treatment.” In the models, “no treatment” was the reference group for treatment, and White was the reference group for patient race. The multivariable model was adjusted for potential confounders and tumor characteristics. Measures of association were expressed as adjusted odds ratios (AORs) and their 95% confidence intervals (CIs). For patients with “non-concordant treatment,” we then evaluated whether treatment was non-concordant due to not receiving recommended surgery or chemoradiation and compared it between racial groups. Next, we compared patients in time-to-treatment more than 6 weeks overall. Analysis on time-to-treatment was not conducted in combination with “guideline concordant” and “non-concordant” treatment because of small sample sizes. Then we described frequency of recurrence (patients with stage I-III tumors) and survival (patients with stage I-IV tumors) and compared the risk of the outcomes between racial/ethnic groups using Cox proportional hazards regression models with White as the reference group for race. Multivariable Cox models were adjusted for potential confounders, tumor features, and treatment(s) received. Treatment was modeled as a time-dependent variable to account for any differences in time-to-treatment and reduce immortal time bias (43,44). Risk of the outcomes was expressed as adjusted hazard ratios (AHRs) and 95% confidence intervals. Statistical tests were 2-sided. For both the logistic and Cox regression models, results were considered statistically significant when the AOR or AHR, respectively, and 95% confidence intervals did not contain the null value of 1.00. We repeated the multivariable analyses with stratification by age at diagnosis (<50 years or ≥50 years), sex (men or women), and tumor stage (I/II or III/IV) to assess potential differences in the association between race and the outcomes among patient subgroups. Analysis was conducted using SAS 9.4 (SAS, Inc, Cary, NC).

Results

The study included 1016 patients with rectal adenocarcinoma. Patients had an average age at diagnosis of 56.4 (+/−13.6 SD) years, and 68.4% were men. In comparing patient characteristics between Black (n = 171) and White (n = 845) patients, overall, a higher proportion of Black patients were younger than 50 years of age at diagnosis compared with White patients (45.6% vs 27.7%) (Table 1). The distribution of patients within each geographic region and sponsor rank also varied by race (Table 1). There were no apparent differences in sex, marital status, active-duty status, or comorbidity. Regarding tumor features, there were no differences in distribution of tumor stage among the races, but a lower proportion of Black patients had grade 2 (55.6% vs 63.2%) tumors, and a higher proportion had unknown grade (19.9% vs 12.2%) tumors than White patients (Table 1).

Table 1.

Description of patients diagnosed with stage I-IV rectal adenocarcinoma in the US Military Health System, 1998-2014

Characteristic	Non-Hispanic Black (N = 171)	Non-Hispanic White (N = 845)	P
Age at diagnosis	No. (%)	No. (%)	<.001
18-39	28 (16.4)	88 (10.4)
40-49	50 (29.2)	146 (17.3)
50-59	50 (29.2)	231 (27.3)
60-69	27 (15.8)	219 (25.9)
70+	16 (9.4)	161 (19.1)
Sex			.15
Men	109 (63.7)	586 (69.3)
Women	62 (36.3)	259 (30.7)
Marital status at diagnosis			.77
Married	129 (75.4)	646 (76.4)
Not Marrieda	42 (24.6)	199 (23.6)
Active duty at diagnosis			.53
No	125 (73.1)	651 (77.0)
Yes	32 (18.7)	132 (15.6)
Unknown	14 (8.2)	62 (7.3)
TRICARE region			.001
North	62 (36.3)	234 (27.7)
South	70 (40.9)	285 (33.7)
West	33 (19.3)	291 (34.4)
Overseas	6 (3.5)	35 (4.1)
Sponsor rank			<.001
Enlisted (E1-E9)	135 (78.9)	525 (62.1)
Warrant/Officer (WO1-5, O1-10)	22 (12.9)	235 (27.8)
Unknown	14 (8.2)	85 (10.1)
Elixhauser comorbidity count			.69
0	91 (53.2)	472 (55.9)
1-2	58 (33.9)	250 (29.6)
3-4	15 (8.8)	79 (9.3)
5+	7 (4.1)	44 (5.2)
GI comorbidity b			.63
No	150 (87.7)	752 (89.0)
Yes	21 (12.3)	93 (11.0)
Tumor stage			.62
I	51 (29.8)	245 (29.0)
II	41 (24.0)	212 (25.1)
III	44 (25.7)	246 (29.1)
IVA/IVB	35 (20.5)	142 (16.8)
Tumor grade			.04
G1	21 (12.3)	118 (14.0)
G2	95 (55.6)	534 (63.2)
G3/4	21 (12.3)	90 (10.7)
Gx	34 (19.9)	103 (12.2)

^aIncludes single, legally separated, divorced, widowed, or unknown.

^bCrohn’s disease, colitis, inflammatory bowel disease, irritable bowel syndrome, diverticulitis, and high-risk diseases and conditions with symptoms similar to colorectal cancer; at least 1 inpatient or 3 outpatient records with a diagnosis code occurring less than 30 days before cancer diagnosis.

In considering treatment receipt on the basis of tumor stage, overall, 63.7% of Black patients and 67.6% of White patients received the recommended treatment(s), and 21.1% of Black patients and 19.6% of White patients received non-concordant treatment(s). There were no statistically significant racial differences in treatment receipt in logistic regression models (Table 2). Black patients were equally as likely to receive guideline-recommended treatment (AOR = 0.76, 95% CI = 0.45 to 1.29) or to receive non-concordant treatment (AOR = 1.24, 95% CI = 0.65 to 2.35) relative to no treatment as White patients. Among the 36 Black and 166 White patients with non-concordant treatment, 55.6% of Black and 49.4% of White did not receive recommended surgery, and 44.4% of Black and 50.6% of White did not receive recommended chemotherapy or radiation (χ²P = .50).

Table 2.

Receipt of stage-appropriate treatment by race/ethnicity for patients with stage I-IV rectal adenocarcinoma in the US Military Health System, 1998-2014

	Non-Hispanic Black			Non-Hispanic White
Treatment	N (%)	OR (95% CI)	AOR (95% CI)a	N (%)	OR (95% CI)
No primary treatment	26 (15.2)	—	—	108 (12.8)	—
Non-concordant treatmentb	36 (21.1)	0.90 (0.52 to 1.58)	1.24 (0.65 to 2.35)	166 (19.6)	1.00 (Referent)
Guideline concordant treatmentc	109 (63.7)	0.79 (0.49 to 1.28)	0.76 (0.45 to 1.29)	571 (67.6)	1.00 (Referent)

^aModel adjusted for age at diagnosis, sex, marital status, active-duty status, TRICARE region, sponsor rank, Elixhauser comorbidity, GI comorbidity, cancer diagnosis year, tumor stage at diagnosis, and tumor grade.

^bNon-concordant treatment is any treatment received other than the guideline recommended therapy for stage.

^cGuideline-concordant treatment is considered surgery for stage I, surgery and chemotherapy or radiation for stage II and III, and chemotherapy or radiation for stage IV according to treatment recommendations.

In analysis of treatment stratified by age at diagnosis, Black patients younger than 50 years of age were less likely to receive guideline-concordant treatment (AOR = 0.34, 95% CI = 0.13 to 0.90 vs White; Table 3). This association was not observed among patients aged 50 years or older at diagnosis (Table 3). The reason for non-concordant treatment (ie, no chemotherapy or no surgery) was similar between races among patients younger than 50 years (P = .61) and patients 50 years or older (P = .66) at diagnosis (data not shown because of small cell sizes and violation of CMS Cell Suppression Policy).

Table 3.

Receipt of stage-appropriate treatment for rectal adenocarcinoma by race-ethnicity in the US Military Health System, 1998-2014, stratified by age at diagnosis, sex, and tumor stage

Age at diagnosis	Age <50 years					Age ≥50 years
Race	Non-Hispanic Black			Non-Hispanic White		Non-Hispanic Black			Non-Hispanic White
Treatment	N (%)	OR (95% CI)	AOR (95% CI)a	N (%)	OR (95% CI)	N (%)	OR (95% CI)	AOR (95% CI) a	N (%)	OR (95% CI)
No primary treatment	12 (15.4)	—	—	18 (7.7)	1.00 (Referent)	14 (15.1)	—	—	90 (14.7)	1.00 (Referent)
Non-concordant treatmentb	12 (15.4)	0.53 (0.20 to 1.42)	0.53 (0.16 to 1.76)	34 (14.5)	1.00 (Referent)	24 (25.8)	1.17 (0.57 to 2.38)	1.70 (0.76 to 3.83)	132 (21.6)	1.00 (Referent)
Guideline-concordant treatmentc	54 (69.2)	0.45 (0.20 to 0.98)	0.34 (0.13 to 0.90)	182 (77.8)	1.00 (Referent)	55 (59.1)	0.91 (0.48 to 1.71)	0.97 (0.48 to 1.94)	389 (63.7)	1.00 (Referent)
Sex	Men					Women
Race	Non-Hispanic Black			Non-Hispanic White		Non-Hispanic Black			Non-Hispanic White
Treatment	N (%)	OR (95% CI)	AOR (95% CI) d	N (%)	OR (95% CI)	N (%)	OR (95% CI)	AOR (95% CI) d	N (%)	OR (95% CI)
No primary treatment	19 (17.4)	—	—	75 (12.8)	1.00 (Referent)	<11 (<17.7)	—	—	33 (12.7)	1.00 (Referent)
Non-concordant treatmentb	20 (18.3)	0.64 (0.32 to 1.28)	0.91 (0.41 to 2.01)	123 (21.0)	1.00 (Referent)	>12 (>19.4)	1.75 (0.65 to 4.76)	2.57 (0.76 to 8.73)	43 (16.6)	1.00 (Referent)
Guideline-concordant treatmentc	70 (64.2)	0.71 (0.41 to 1.25)	0.70 (0.38 to 1.32)	388 (66.2)	1.00 (Referent)	39 (62.9)	1.01 (0.41 to 2.44)	0.93 (0.33 to 2.61)	183 (70.7)	1.00 (Referent)
Stage at diagnosis	I and II					III and IV
Race	Non-Hispanic Black			Non-Hispanic White		Non-Hispanic Black			Non-Hispanic White
Treatment	N (%)	OR (95% CI)	AOR (95% CI) e	N (%)	OR (95% CI)	N (%)	OR (95% CI)	AOR (95% CI) e	N (%)	OR (95% CI)
No primary treatment	13 (14.1)	—	—	75 (16.4)	1.00 (Referent)	13 (16.5)	—	—	33 (8.5)	1.00 (Referent)
Non-concordant treatmentb	19 (20.7)	1.25 (0.58 to 2.69)	1.78 (0.72 to 4.44)	88 (19.3)	1.00 (Referent)	17 (21.5)	0.55 (0.24 to 1.27)	0.58 (0.22 to 1.54)	78 (20.1)	1.00 (Referent)
Guideline-concordant treatmentc	60 (65.2)	1.18 (0.61 to 2.26)	1.40 (0.68 to 2.91)	294 (64.3)	1.00 (Referent)	49 (62.0)	0.45 (0.22 to 0.91)	0.28 (0.12 to 0.64)	277 (71.4)	1.00 (Referent)

^aModel adjusted for patient sex, marital status, active-duty status, TRICARE region, sponsor rank, Elixhauser comorbidity, GI comorbidity, cancer diagnosis year, tumor stage at diagnosis, and tumor grade.

^bNon-concordant treatment is any treatment received other than the guideline-recommended therapy for stage.

^cGuideline-concordant treatment is considered surgery for stage I, surgery and chemotherapy or radiation for stage II and III, and chemotherapy or radiation for stage IV according to treatment recommendations.

^dModel adjusted for age at diagnosis, marital status, active-duty status, TRICARE region, sponsor rank, Elixhauser comorbidity, GI comorbidity, cancer diagnosis year, tumor stage at diagnosis, and tumor grade.

^eModel adjusted for age at diagnosis, patient sex, marital status, active-duty status, TRICARE region, sponsor rank, Elixhauser comorbidity, GI comorbidity, cancer diagnosis year, tumor stage, and tumor grade.

In analysis of treatment by tumor stage, Black patients were less likely to receive guideline-concordant treatment(s) among those with stage III or IV tumors (AOR = 0.28, 95% CI = 0.12 to 0.64 vs White), but not among those with stage I or II tumors (Table 3). The reason for non-concordant treatment was similar between races within strata by tumor stage (P_stageI/II = .57 and P_stageIII/IV = .10; data not shown). In strata by patient sex, there were no statistically significant racial differences in receipt of recommended therapy (Table 3) or in reason for non-concordant treatment (P_men = .13 and P_women = .76; data not shown).

In evaluation of treatment timing independent of guideline concordance, treatment was initiated more than 6 weeks after diagnosis for 21.4% of Black and 17.4% of White patients (Figure 1), and the likelihood of treatment more than 6 weeks after diagnosis was similar between racial groups (AOR = 1.52, 95% CI = 0.94 to 2.45, data not shown). However, for patients younger than 50 years of age at diagnosis, a higher frequency of Black patients (21.2%) had treatment initiation more than 6 weeks after diagnosis compared with White patients (9.3%, P = .009), whereas the frequency of treatment initiation more than 6 weeks was similar among patients 50 years of age or older (Figure 1). The frequency of treatment initiation more than 6 weeks after diagnosis was similar between races within tumor stage and sex strata (Figure 1).

Figure 1.

Percentage of patients with treatment initiated more than 6 weeks after diagnosis with rectal adenocarcinoma in the Military Health System, 1998-2014, overall and by age at diagnosis, sex, and tumor stage.

In the assessment of clinical outcomes, overall, 19.9% of Black patients and 13.4% of White patients with stage I-III disease experienced a recurrence, and 24.6% of Black and 26.9% of White patients died during the study period (Table 4). In multivariable regression models, there were no statistically significant racial differences in recurrence (AHR = 1.34, 95% CI = 0.85 to 2.12 for Black vs White, stage I-III) or survival (AHR = 1.09, 95% CI = 0.77 to 1.54 for Black vs White, all stages). In analysis stratified by patient age at diagnosis, there was a statistically higher risk of death for Black patients younger than 50 years of age compared with White in the univariable model (HR = 1.79, 95% CI = 1.08 to 2.98), which was attenuated and not statistically significant in the multivariable model (AHR = 1.46, 95% CI = 0.79 to 2.68) (Table 5). There were no statistically significant racial differences in recurrence or survival in the other strata by age at diagnosis, sex, or tumor stage (Table 5).

Table 4.

Cox regression adjusted hazard ratio (AHR) and 95% confidence intervals (CIs) for recurrence and survival of patients with stage I-IV rectal adenocarcinoma in the US Military Health System, 1998-2014

	Non-Hispanic Black				Non-Hispanic White
Outcome	Ntotal	Nevent (%)	HR (95% CI)	AHR (95% CI)a	Ntotal	Nevent (%)	HR (95% CI)
Recurrence b	136	27 (19.9)	1.50 (0.98 to 2.30)	1.34 (0.85 to 2.12)	703	94 (13.4)	1.00 (Referent)
All-cause death	171	42 (24.6)	0.97 (0.70 to 1.35)	1.09 (0.77 to 1.54)	845	227 (26.9)	1.00 (Referent)

^aModel adjusted for age at diagnosis, patient sex, marital status, active-duty status, TRICARE region, sponsor rank, Elixhauser comorbidity, GI comorbidity, cancer diagnosis year, tumor stage at diagnosis, tumor grade, and primary and adjuvant treatment(s) received.

^bRecurrence includes local or distant recurrence and applies to patients with stage I-III tumors at diagnosis only; recurrence identified in data by record with ICD-9 diagnosis code for secondary malignant neoplasm.

Table 5.

Cox regression adjusted hazard ratio (AHR) and 95% confidence intervals (CIs) for recurrence and survival of patients with stage I-IV rectal adenocarcinoma in the US Military Health System, 1998-2014, by age at diagnosis, sex, and tumor stage

Age at diagnosis	Age <50 years							Age >=50 years
Race	Non-Hispanic Black				Non-Hispanic White			Non-Hispanic Black				Non-Hispanic White
Outcome	Ntotal	Nevent (%)	HR (95% CI)	AHR (95% CI)a	Ntotal	Nevent (%)	HR (95% CI)	Ntotal	Nevent (%)	HR (95% CI)	AHR (95% CI)a	Ntotal	Nevent (%)	HR (95% CI)
Recurrence b	59	15 (25.4)	1.60 (0.87 to 2.96)	1.69 (0.85 to 3.36)	188	32 (17.0)	1.00 (Referent)	77	12 (15.6)	1.29 (0.69 to 2.39)	1.07 (0.54 to 2.14)	515	62 (12.0)	1.00 (Referent)
All-cause death	78	23 (29.5)	1.79 (1.08 to 2.98)	1.46 (0.79 to 2.68)	234	42 (17.9)	1.00 (Referent)	93	19 (20.4)	0.69 (0.43 to 1.10)	0.78 (0.48 to 1.27)	611	185 (30.3)	1.00 (Referent)
Sex	Men							Women
Race	Non-Hispanic Black				Non-Hispanic White			Non-Hispanic Black				Non-Hispanic White
Outcome	Ntotal	Nevent (%)	HR (95% CI)	AHR (95% CI) c	Ntotal	Nevent (%)	HR (95% CI)	Ntotal	Nevent (%)	HR (95% CI)	AHR (95% CI) c	Ntotal	Nevent (%)	HR (95% CI)
Recurrence b	84	17 (20.2)	1.64 (0.96 to 2.80)	1.37 (0.76 to 2.48)	479	63 (13.2)	1.00 (Referent)	52	<11 (<17.7)	1.29 (0.63 to 2.64)	1.38 (0.62 to 3.11)	224	31 (13.8)	1.00 (Referent)
All-cause death	109	31 (28.4)	1.17 (0.80 to 1.72)	1.12 (0.75 to 1.68)	586	167 (28.5)	1.00 (Referent)	62	11 (17.7)	0.71 (0.37 to 1.35)	1.23 (0.59 to 2.58)	259	60 (23.2)	1.00 (Referent)
Stage at diagnosis	I and II							III and IV
Race	Non-Hispanic Black				Non-Hispanic White			Non-Hispanic Black				Non-Hispanic White
Outcome	Ntotal	Nevent (%)	HR (95% CI)	AHR (95% CI) d	Ntotal	Nevent (%)	HR (95% CI)	Ntotal	Nevent (%)	HR (95% CI)	AHR (95% CI) d	Ntotal	Nevent (%)	HR (95% CI)
All-cause death	92	<11 (<12.0)	0.70 (0.36 to 1.36)	0.93 (0.45 to 1.90)	457	81 (17.7)	1.00 (Referent)	79	32 (40.5)	1.05 (0.72 to 1.55)	1.16 (0.77 to 1.76)	388	146 (37.6)	1.00 (Referent)

^aModel adjusted for patient sex, marital status, active-duty status, TRICARE region, sponsor rank, Elixhauser comorbidity, GI comorbidity, cancer diagnosis year, tumor stage at diagnosis, tumor grade, and primary and adjuvant treatment(s) received.

^bRecurrence includes local or distant recurrence and applies to patients with stage I-III tumors at diagnosis only; recurrence identified in data by record with ICD-9 diagnosis code for secondary malignant neoplasm.

^cModel adjusted for age at diagnosis, marital status, active-duty status, TRICARE region, sponsor rank, Elixhauser comorbidity, GI comorbidity, cancer diagnosis year, tumor stage at diagnosis, tumor grade, and primary and adjuvant treatment(s) received.

^dModel adjusted for age at diagnosis, patient sex, marital status, active-duty status, TRICARE region, sponsor rank, Elixhauser comorbidity, GI comorbidity, cancer diagnosis year, tumor grade, and primary and adjuvant treatment(s) received.

Discussion

In the Military Health System, which provides equal access to care, we found no overall differences in treatment receipt or timing, cancer recurrence, or survival between non-Hispanic Black and White patients with rectal adenocarcinoma. The overall similar likelihood of Black and White patients to receive guideline recommended treatment may contribute to the lack of racial disparities in clinical outcomes in the study population. Our main results on treatment and outcomes contrast studies in the general US population where access to care may differ by socioeconomic and demographic characteristics (12,13,15-17,24). For example, in a study of patients with rectal cancer in the Surveillance Epidemiology and End Results program, which contains patients from across the United States with varying access to care, authors found that Black patients were less likely to receive treatment(s) and had worse overall survival (13). Other studies of the National Cancer Database have reported a 44% higher risk of recurrence for Black compared with White patients with stage I-III disease receiving surgery (17), and between 12% and 26% higher risk of death for patients with all stages of rectal cancer (15,16). The observed disparities in these studies persisted even once controlling for sociodemographic characteristics and treatment received (13,15-17).

Although studies solely of rectal cancer are limited, previous investigations in equal access settings have demonstrated similar treatment, including surgery and adjuvant chemotherapy, and survival rates for patients of different racial/ethnic backgrounds with colon or colorectal cancer (45-50). Our study provides important evidence on possible disparities among patients with rectal adenocarcinoma in the MHS using comprehensive registry and claims data. The findings of no overall racial differences in the measured study outcomes further supports the notion that access to care may reduce racial health disparities in rectal cancer observed in the general US population. However, there were some observed racial differences in stratified analysis by age at diagnosis and tumor stage that need attention.

Black patients younger than 50 years of age at diagnosis had a lower likelihood to receive guideline recommended treatment, a higher frequency of treatment less than 6 weeks after diagnosis, and a non–statistically significant increased risk of all-cause death compared with White patients in this age group. As early-onset colon and rectal cancers gain attention and may be different from late-onset cancers in terms of risk factors and response to treatment (51-54), understanding potential disparities in treatment and outcomes is increasingly important. A study of the National Cancer Database found that Black patients younger than 50 years of age with rectal cancer were 27% less likely to receive guideline-recommended care than White patients (18). The authors reported that health insurance differences accounted for a good portion, but not all, of the disparity (18). Meanwhile, another study of the National Cancer Database reported racial disparities in overall survival for patients with early-onset colorectal cancer even among those with private insurance and similar income or education levels (19). This evidence along with the results of our study in the MHS suggests that factors other than access to care may be related to disparities in treatment and outcomes for patients with early-onset rectal cancer. Future efforts that aim to understand patient perceptions of disease, its risk for progression, patient–provider communication, social support, and other factors that may influence treatment receipt and timing and survival among those younger than 50 years of age who are diagnosed with rectal cancers are warranted.

Among patients with stage III and IV rectal cancers, we observed a lower likelihood of receiving recommended therapy for Black compared with White patients. Although the literature on racial disparities in treatment of stage IV rectal cancer is limited (55), several studies have reported a lower likelihood of Black patients with stage II and III rectal cancers to receive recommended treatment(s) even when controlling for sociodemographic and economic variables (10,20,22,56). The results of our study in the MHS further suggest factors other than access to care, which are likely multifactorial, in treatment disparities among patients with late-stage disease. Several studies in locally advanced rectal cancer also reported racial disparities in outcomes, pointing at differences in biologic and socioeconomic factors as a source of the disparities after adjusting for treatment(s) received (10,20,22). In our study, the racial difference in treatment among patients with stage III or IV tumors did not appear to translate to differences in overall survival. This observation may be related to other complex factors, such as age at diagnosis and competing comorbid conditions, which were adjusted for in the Cox models. As treatment of advanced and metastatic rectal cancers continue to evolve and overtreatment is becoming a topic of discussion (57-61), further investigation is needed to determine the contributors to racial disparities in treatment and outcomes for patients with advanced disease and to optimize treatment strategies to achieve desirable clinical outcomes.

We used comprehensive data from an equal access health system to study racial disparities in rectal cancer treatment and outcomes. However, our study has a few limitations. First, the data available at the time of study included patients diagnosed between 1998 and 2014. Thus, treatment included reflected those available during that time, and we cannot exclude possible racial differences in receipt of recently developed therapies, such as targeted therapy or immunotherapy. Nevertheless, our study provides important information about racial disparities in receipt of standard treatments of surgery, radiation, and chemotherapy, which have consistently been included in the treatment guidelines (3,62-67). Second, there were overall too few patients among treatment subgroups (eg, surgery with adjuvant chemotherapy) to evaluate multiple time-to-treatment intervals, and the patient sample sizes were too small to further evaluate likelihood of treatment delay in conjunction with receipt of guideline concordant treatment. Next, all-cause death was used as the survival outcome because information on cancer-specific death was not available in the data. However, we examined potential racial differences in cancer recurrence, finding none in our data. Nevertheless, we cannot rule out any potential racial differences in cancer-specific death. Last, the overall lower number of Black patients might lead to a relatively low study power in stratified analysis by age at diagnosis, sex, or tumor stage. However, we did find some racial differences in the outcomes in the stratified analyses.

In the equal access MHS, there were no overall racial disparities in rectal cancer treatment or clinical outcomes between Black and White patients. The lack of racial disparities in receiving guideline-recommended treatment receipt may contribute to the similar rates of recurrence and survival for Black and White patients in the study population overall. However, some potential disparities were observed among patients younger than 50 years of age at diagnosis and with late-stage tumors, which may be related to factors other than access to care and warrants further study.

Data availability

The data that support the findings of this study are not publicly available due to the restrictions in the access and use of the MilCanEpi data specified in the data-sharing agreements and regulatory approvals. The Department of Defense Central Cancer Registry (DoD CCR) data and data dictionary may be requested from the Joint Pathology Center and online at https://jpc.capmed.mil/. The Military Health System Data Repository (MDR) data and data dictionary may be requested from the Defense Health Agency and online at https://www.health.mil/.

Author contributions

Yvonne L. Eaglehouse, PhD, MPH (Conceptualization; Data curation; Investigation; Methodology; Writing—original draft; Writing—review & editing), Sarah Darmon, PhD (Data curation; Formal analysis; Methodology; Writing—review & editing), Michele M. Gage, MD (Investigation; Writing—review & editing), Craig D. Shriver, MD (Funding acquisition; Project administration; Resources; Software; Writing—review & editing), Kangmin Zhu, MD, PhD (Conceptualization; Funding acquisition; Investigation; Methodology; Project administration; Resources; Software; Supervision; Writing—original draft; Writing—review & editing).

Funding

This project was supported by the Murtha Cancer Center Research Program (MCCRP) of the Department of Surgery, Uniformed Services University of the Health Sciences (USUHS) under the auspices of the Henry M. Jackson Foundation for the Advancement of Military Medicine (HJF, Inc), grant numbers HHU0001-16-2-0014 and HU0001-18-2-0032 awarded to C.S.

Conflicts of interest

YE, SD, and KZ are/were employed by the Henry M. Jackson Foundation for the Advancement of Military Medicine, Inc, at the time work was performed. The authors have no other potential conflicts of interest to declare.

Disclaimer

The contents of this article are the sole responsibility of the authors and do not necessarily reflect the views, assertions, opinions, or policies of the Uniformed Services University of the Health Sciences, the Henry M. Jackson Foundation for the Advancement of Military Medicine, Inc, the Department of Defense, or the Departments of the Army, Navy, or Air Force. Mention of trade names, commercial products, or organizations does not imply endorsement by the US government.