Five-year survival of patients with late-stage prostate cancer: comparison of the Military Health System and the U.S. general population

Abstract

Background

While the 5-year survival rate for local and regional prostate cancer is nearly 100%, it decreases dramatically for advanced tumours. Accessibility to health care is an important factor for cancer prognosis. The U.S. Military Health System (MHS) provides universal health care to its beneficiaries, reducing financial barriers to medical care. However, whether the universal care translates into improved survival among patients with advanced prostate cancer in the MHS is unknown. In this study, we compared the MHS and the U.S. general population in survival of patients with advanced prostate cancer (stages III and IV).

Methods

The MHS patients (N = 5379) were identified from the Department of Defense’s (DoD) Automated Central Tumor Registry (ACTUR). Patients in the U.S. general population (N = 21,516) were identified from the Surveillance, Epidemiology, and End Results (SEER) programme. The two populations were matched on age, race, and diagnosis year.

Results

The ACTUR patients exhibited longer 5-year survival than the matched SEER patients (HR = 0.74, 95% CI = 0.67–0.83), after adjustment for the potential confounders. The improved survival was observed for ages 50 years or older, both White patients and Black patients, all tumour stages and grades. This was also demonstrated despite the receipt of surgery or radiation treatment.

Conclusions

MHS beneficiaries with advanced prostate cancer had longer survival than their counterparts in the U.S. general population.

Background

Prostate cancer is the second leading cause of cancer mortality in U.S. men [1]. While the 5-year relative survival rate is close to 100% for patients with early-stage tumours, it decreases to 30% for patients with distant stage of the disease [2]. In the U.S. general population, accessibility to health care, as reflected by the status of health insurance, affects survival among cancer patients [3–6]. Barriers in health care access influence cancer care from cancer screening, diagnosis, treatment, to post-diagnosis quality of care [3, 7–10] and therefore the cancer outcomes. As shown in the U.S. general population, prostate patients without health insurance or with Medicaid (a joint federal and state programme that helps with medical costs for some people with limited income and resources) had shorter survival than patients with other insurance coverages [11–13].

The U.S. Military Health System (MHS) provides universal health care to its beneficiaries including active-duty service members, retirees, and their family members [14]. As the beneficiaries receive free medical care or pay minimum out-of-pocket costs for care, barriers to medical care observed in the U.S. general population are minimised in the MHS. In order to assess whether universal health care provided by the MHS has translated into improvement in cancer survival, we previously compared patients with lung cancer [15], breast cancer [16], colon cancer [17], and brain cancer [18], from the MHS and the U.S. general population. These studies showed longer survival for the MHS beneficiaries, suggesting the survival benefit of MHS’s universal health care access to cancer patients. However, no research has been conducted for prostate cancer, which is the most common cancer among men. In this study, we compared overall survival of prostate cancer patients in the MHS with that in the general population. Particularly, we focused on advanced prostate cancer (stages III and IV) because of its poorer prognosis than early-stage tumours [19] and its higher likelihood to reflect prostate cancer-specific survival when overall survival is the study outcome. We hypothesised that the MHS beneficiaries with advanced prostate cancer, who benefit from the universal health care, have longer survival than their counterparts in the U.S. general population.

Methods

Data sources

In this study, we used the data from the Department of Defense’s (DoD) Automated Central Tumour Registry (ACTUR) and the National Cancer Institute’s (NCI) Surveillance, Epidemiology, and End Results (SEER) programme. ACTUR contains information on demographics, cancer diagnosis, tumour characteristics, cancer treatment, follow-up, vital status and other data from cancer patients diagnosed or treated at military treatment facilities (MTFs). Data items in ACTUR comply with the guidelines of the North American Association of Central Cancer Registries (NAACCR) [20]. SEER is a National Cancer Institute’s cancer registry programme that collects population-based cancer data in SEER-served cancer registries. The SEER data include demographics, cancer diagnosis and characteristics, first course of treatment, follow-up, vital status and other information. In this study, we used data from the SEER-18 registries (Atlanta, Connecticut, Detroit, Hawaii, Iowa, New Mexico, San Francisco-Oakland, Seattle-Puget Sound, Utah, Los Angeles, San Jose-Monterey, Rural Georgia, the Alaska Native, Greater California, Greater Georgia, Kentucky, Louisiana, and New Jersey), which represents 28% of the U.S. general population [21].

Study subjects

The MHS patients were identified from ACTUR and patients from the U.S. general population were identified from SEER. Men were eligible for the study if they were histologically diagnosed with advanced-stage prostate adenocarcinoma between January 1, 1987 and December 31, 2013 and aged 18 years or older at diagnosis. This time frame was used based on the availability of the ACTUR data at the time of data acquisition. Since adenocarcinoma comprises 95% of all malignant prostate tumours [22], we included only adenocarcinoma to preclude the potential effects of various histologic types on the study results. Prostate adenocarcinoma was defined with the cancer site codes (C61.9) and morphology code (8140) according to the International Classification of Diseases for Oncology, third edition (ICD-O-3) [23], or converted to ICD-O-3 for patients diagnosed in earlier periods. Patients with diagnosis from death certificate or autopsy only were excluded. Patients with multiple primary cancers were excluded to minimise possible effects of other cancers on survival. Because age, race, and diagnosis year are associated with cancer survival and differ between the two populations, we matched SEER patients to ACTUR patients on age (within 5 years), race (White, Black, Asian/Pacific Islander), and diagnosis year group (1987–1989, 1990–1994, 1995–1999, 2000–2004, 2005–2009, 2010–2013) with a matching ratio of 4:1.

Study variables

The study outcome was all-cause mortality. Cancer-specific mortality was not used because information on cause of death was not complete in the ACTUR data. Tumour stages (III and IV) were defined according to the American Joint Committee on Cancer (AJCC)’s criteria [24]. Tumour grade was defined as well-differentiated (grade I), moderately differentiated (grade II), poorly differentiated (grade III), non-differentiated (grade IV), or unknown. Prostate-specific antigen (PSA) values were grouped into the following categories: <4, 4–<10, 10–<20, and ≥20 [25, 26]. Because the information on PSA data became available in SEER and complete in ACTUR in 2010, only the data from 2010 and later were used for PSA.

Site-specific cancer-directed surgery codes were used to define “surgery received”, “no surgery” or “unknown or missing”. Radiation treatment was grouped into “radiation therapy received”, “radiation therapy not received”, or “unknown” according to SEER guidelines. Information on chemotherapy or hormone therapy receipt was unavailable in the SEER data for public use and therefore for the analysis. Demographic information included age, race, Hispanic ethnicity, and year of diagnosis. Hispanic ethnicity was grouped into “Non-Hispanic”, “Hispanic”, or “unknown”. Other demographic variables were used as matching variables and were described above.

This study was based on the non-identifiable ACTUR data approved by the institutional review board of Walter Reed National Military Medical Center. The SEER data included in this study are de-identified data for public use.

Statistical analysis

We first used the Chi-square test to compare the distributions of demographic and tumour characteristics between ACTUR and SEER patients. We then conducted survival analysis comparing the two patient populations. The study outcome was all-cause death in the five years following the diagnosis. If a patient died during the 5-year period, the follow-up time was calculated as the time from diagnosis to death. If death was not observed during the period, follow-up time was censored at the end of the fifth year. For patients who were diagnosed in more recent years and thus had <5 follow-up years after diagnosis, and if they were alive through the end of the study, survival time was censored at the study ending date, December 31, 2013. We compared overall 5-year survival between ACTUR and SEER using the Kaplan–Meier curve and log-rank test. Multivariable Cox proportional hazards model for matched data was then performed to estimate hazard ratios (HRs) and their 95% CIs for ACTUR compared to SEER with adjustment for variables other than the matching variables, which may potentially confound the results due to their associations with survival and differ between the two populations. These variables included Hispanic ethnicity, tumour stage, tumour grade, PSA, surgery, and radiation treatment. We also adjusted for age as a continuous variable to control for its residual confounding effects. We further conducted analysis stratified by age group, race, tumour stage, tumour grade, surgery, or radiation treatment. The purpose of the stratified analyses was to assess whether the differences between the two populations might vary by these variables. However, we did not conduct stratified analysis by PSA due to lack of information prior to 2010 and small numbers of patients diagnosed during 2010–2013. Instead, we performed stratified analysis by calendar year (<2010, no PSA, vs. 2010–2013 when PSA was available). To further evaluate the potential effects of PSA on results, for the 2010–2013 period, we performed the analysis with and without PSA adjustment. As PSA screening was recommended for men ages 50 or older, the analysis was confined to this age group. In all stratified analyses described above, a heterogeneity test was performed to test if stratum-specific HRs for a variable were statistically different from each other [27]. The proportional hazards assumption was evaluated by plotting the log–log survival curves. The assumption of proportional hazard was met.

All statistical analyses were conducted using the SAS software version 9.4.0 (SAS Institute, Inc.). All reported p values were two-sided with the significance level at p < 0.05.

Results

There were 5379 ACTUR patients and 21,516 matched SEER patients (Table 1). As shown in Table 1, the ACTUR and SEER patients had the same distributions on the matching variables, age (p = 1.00), race (p = 1.00) and year of diagnosis (p = 1.00). While Hispanic ethnicity constituted 7.0% of SEER patients and 4.2% of ACTUR patients, respectively, data were missing for 32.3% of ACTUR patients. Compared with SEER patients, ACTUR patients were more likely having stage III disease (58.9 vs. 55.1%) and less likely having stage IV disease (41.1 vs. 44.9%) (p < 0.001). In tumour grade, ACTUR patients were more likely to have grade I and unknown grade tumours, but less likely to have grades III and IV than SEER patients (p < 0.0001). A high percentage of patients had unknown PSA for both ACTUR (77.3%) and SEER (91.8%) due to the unavailability of information prior to 2010. The proportion of patients who received surgery was lower (60.7 vs. 64.8%, p < 0.001), but the proportion of radiation treatment was higher (26.6 vs. 23.7%, p < 0.001) for ACTUR than SEER.

Table 1.

Demographic and tumour characteristics of advanced-stage prostate cancer cases diagnosed during 1987–2013 in ACTUR and SEER registries.

	ACTUR (N = 5379)		SEER (N = 21,516)		p value
	N	%	N	%
Age group					1.00
<50	189	3.5	756	3.5
50–64	2570	47.8	10,280	47.8
≥65	2620	48.7	10,480	48.7
Race					1.00
White	4132	76.8	16,528	76.8
Black	1048	19.5	4192	19.5
Asian or Pacific Islander	199	3.7	796	3.7
Year of diagnosis					1.00
1987–1989	451	8.4	1804	8.4
1990–1994	1822	33.9	7288	33.9
1995–1999	1125	20.9	4500	20.9
2000–2004	850	15.8	3400	15.8
2005–2009	647	12.0	2588	12.0
2010–2013	484	9.0	1936	9.0
Hispanic ethnicity					<0.001
Non-Hispanic	3416	63.5	20,012	93.0
Hispanic	226	4.2	1504	7.0
Unknown	1737	32.3	0	0.0
Tumour stage					<0.001
Stage III	3167	58.9	11,866	55.1
Stage IV	2212	41.1	9650	44.9
Tumour grade					<0.001
Well differentiated, grade1	380	7.1	735	3.4
Moderately differentiated, grade 2	2177	40.5	9902	46.0
Poorly differentiated, grade 3	2242	41.7	9622	44.7
Undifferentiated, grade 4	62	1.2	213	1.0
Unknown	518	9.6	1044	4.9
PSA (ng/ml)a					<0.001
<4	160	3.0	108	0.5
4–<10	458	8.5	667	3.1
10–<20	210	3.9	350	1.6
≥20	391	7.3	650	3.0
Unknown/missing	4160	77.3	19,741	91.8
Surgery					<0.001
No	2037	37.9	7551	35.1
Yes	3265	60.7	13,935	64.8
Unknown	77	1.4	30	0.1
Radiation therapy					<0.001
No	3741	69.5	16,108	74.9
Yes	1430	26.6	5090	23.7
Unknown	208	3.9	318	1.5

PSA prostate-specific antigen.

^aData prior to 2010 were not available in SEER and sparse in ACTUR.

The median follow-up times for deceased patients were 26 and 23 months for ACTUR and SEER, respectively, and were 60 months for alive or censored patients in both populations. The Kaplan–Meier survival curve showed no significant difference in overall 5-year survival between the two populations (Fig. 1; p = 0.16). However, in multivariable Cox proportional hazards model, the ACTUR patients exhibited a 26% lower 5-year hazard of death than the SEER patients (HR = 0.74, 95% CI = 0.67–0.83) after adjustment for the potential confounders (Table 2). In the analysis further stratified by demographic characteristics (Table 2), the lower hazard of death for ACTUR was observed in those aged 50–64 (HR = 0.73, 95% CI = 0.60–0.88) and 65 or older (HR = 0.75, 95% CI = 0.66–0.85), but was not seen in younger patients under the age of 50 (HR = 1.46, 95% CI = 0.60–3.57). Black men from ACTUR had 32% lower hazard of death compared to their SEER counterparts (HR = 0.68, 95% CI: 0.54–0.85), while the corresponding percentage was 25% for White men (HR = 0.75, 95% CI = 0.66–0.84). However, survival was not significantly different between the two populations for Asian/Pacific Islanders (HR = 1.29, 95% CI = 0.76–2.21). In stratified analysis by calendar year, the HRs were 0.78 (95% CI = 0.71–0.86) for <2010 and 0.79 (95% CI = 0.51–1.22) for 2010–2013. For the latter period, the additional analyses evaluating the potential effects of PSA among patients ages 50 or older showed that the HRs were 0.78 (95% CI = 0.50–1.21) and 0.68 (95% CI = 0.45–1.04) with and without PSA adjustment, respectively (results not shown). When survival was compared by tumour stage, the lower hazard for ACTUR than SEER was seen for both stage III (HR = 0.84, 95% CI = 0.76–0.96) and stage IV (HR = 0.71, 95% CI = 0.65–0.77) tumours. The ACTUR patients tended to have a lower hazard than the SEER patients despite tumour grade while the difference between the two populations tended to be more obvious for patients with well-differentiated tumour (Grade I) (HR = 0.47, 95% CI = 0.34–65), undifferentiated tumour (Grade IV) (HR = 0.52, 95% CI = 0.36–0.77), and unknown grade (HR = 0.50, 95% CI = 0.42–0.60). When the survival was compared by surgery receipt, the survival advantage of ACTUR was observed regardless of whether patients received surgery or not. Similarly, the survival advantage of ACTUR patients was observed in patients who did not receive radiation (HR = 0.72, 95% CI = 0.66–0.78) as well as who received it (HR = 0.78, 95% CI = 0.69–0.88). Heterogeneity tests showed that the variation of the stratified HRs was statistically significant only for tumour grade (p < 0.001).

Fig. 1. Kaplan–Meier survival curve for ACTUR and SEER patients with late-stage prostate cancer diagnosed during 1987–2013.

The black panel represents ACTUR patient population and the grey panel represents the SEER patient population. The symbol of each panel is shown in the figure.

Table 2.

Five-year overall and stratified hazard ratios of all-cause mortality comparing ACTUR with SEER among advanced-stage prostate cancer patients diagnosed between 1987 and 2013.

Variables	Numbers		Adjusted HRa (95% CI)
	All cases (N = 26,895)	Deaths (N = 7731)
Overall
SEER	21,516	6231	1.00 (ref.)
ACTUR	5379	1500	0.74 (0.67–0.83)
By age
<50	945	190
SEER	756	160	1.00 (ref.)
ACTUR	189	30	1.46 (0.60–3.57)
50–64	12,850	2528
SEER	10,280	2049	1.00 (ref.)
ACTUR	2570	479	0.73 (0.60–0.88)
65 or older	13,100	5013
SEER	10,480	4022	1.00 (ref.)
ACTUR	2620	991	0.75 (0.66–0.85)
By race
White	20,660	5744
SEER	16,528	4593	1.00 (ref.)
ACTUR	4132	1151	0.75 (0.66–0.84)
Black	5240	1750
SEER	4192	1458	1.00 (ref.)
ACTUR	1048	292	0.68 (0.54–0.85)
Asian/Pacific Islander	995	237
SEER	796	180	1.00 (ref.)
ACTUR	199	57	1.29 (0.76–2.21)
By year
1987–2009	24,475		7420
SEER	19,580	5968	1.00 (ref.)
ACTUR	4895	1452	0.78 (0.71–0.86)
2010–2013	2420		311
SEER	1936	263	1.00 (ref.)
ACTUR	484	48	0.79 (0.51–1.22)
By tumour stage
Stage III	15,033	1567
SEER	11,866	1222	1.00 (ref.)
ACTUR	3167	345	0.84 (0.74–0.96)
Stage IV	11,862	6164
SEER	9650	5009	1.00 (ref.)
ACTUR	2212	1155	0.71 (0.65–0.77)
By tumour grade
Well differentiated	1115	201
SEER	735	150	1.00 (ref.)
ACTUR	380	51	0.47 (0.34–0.65)
Moderately differentiated	12,079	2119
SEER	9902	1713	1.00 (ref.)
ACTUR	2177	406	0.84 (0.74–0.95)
Poorly differentiated	11,864	4381
SEER	9622	3570	1.00 (ref.)
ACTUR	2242	811	0.78 (0.71–0.86)
Undifferentiated	275	167
SEER	213	133	1.00 (ref.)
ACTUR	62	34	0.52 (0.36–0.77)
Unknown	1562	863
SEER	1044	665	1.00 (ref.)
ACTUR	518	198	0.50 (0.42–0.60)
By surgery
Yes	17,200	2631
SEER	13,935	2179	1.00 (ref.)
ACTUR	3265	452	0.82 (0.73–0.92)
No	9588	5047
SEER	7551	4034	1.00 (ref.)
ACTUR	2037	1013	0.71 (0.65–0.77)
Unknown	107	53
SEER	30	18	1.00 (ref.)
ACTUR	77	35	0.45 (0.25–0.79)
By radiation
Yes	6520	1920
SEER	5090	1533	1.00 (ref.)
ACTUR	1430	387	0.78 (0.69–0.88)
No	19,849	5679
SEER	16,108	4615	1.00 (ref.)
ACTUR	3741	1064	0.72 (0.66–0.78)
Unknown	526	132
SEER	318	83	1.00 (ref.)
ACTUR	208	49	0.87 (0.61–1.25)

HR hazard ratio, CI confidence interval.

^aHRs were estimated from multivariable Cox proportional hazard model for matched data. HR was additionally adjusted for age (as continuous variable), Hispanic origin, tumour stage, tumour grade, PSA (when available), surgery and radiation. The stratified variable itself was not adjusted.

Discussion

This study found that patients with advanced prostate cancer in the MHS had improved survival compared to their counterparts in the U.S. general population, and the survival advantage was consistently observed in all subgroups defined by race (except Asian/Pacific Islander), tumour stage, tumour grade, surgery, and radiation treatment. The survival advantage of the MHS beneficiaries was observed for middle-aged (50–64 years old) and older (65 or older) men, but not for younger (aged 49 or younger) men.

Our findings suggest the benefits of the MHS’s universal health care for survival of patients with advanced stage prostate cancer. Our study further stresses the importance of health care access for survival of prostate cancer patients. In the studies that compared prostate cancer outcomes between patients with different insurance types in the U.S. general populations, it was reported that compared to those with Medicaid or who were uninsured, men with private insurance or Medicare had improved survival [3, 13, 28, 29], and they were less likely to present with advanced stage [12, 3, 13, 30, 31], and more likely to receive cancer-directed treatment [3, 12, 13, 30, 32] Our study provides additional evidence on the benefits of health insurance for survival of men with high-stage prostate cancer, showing that patients within the MHS had improved survival compared to those in the U.S. general population.

The improved survival in the MHS reported in this study is consistent with our previous studies on lung cancer [15], colon cancer [17], breast cancer [16] and glioma [18], which vary in diagnosis and treatment. The consistency suggests that the improved survival among the MHS beneficiaries than the general population may not be cancer-site specific; accessibility to timely care and factors at the health system level may contribute to the survival benefit. The MHS beneficiaries receive medical care free of charge or with minimal out-of-pocket cost, thus financial barriers to medical care are largely minimised. Health care accessibility influences the entire cancer care continuum, including adherence to cancer screening guidelines [7, 8], diagnosis and early detection [12, 30, 31], cancer treatment delivery, and quality of care received [33–35].Universal health care systems may reduce barriers in these stages of cancer care and thus improve the outcomes of advanced prostate cancer.

The improved survival for the MHS was observed for almost all the demographic subgroups except patients aged 49 years or younger and Asian patients. In regard to age, although the heterogeneity test showed that the overall differences in stratum-specific HRs were not statistically significant as a whole, a tendency that the difference existed only in older patients was observed. If this is true, it might be related to the utilisation of prostate cancer screening. PSA screening for prostate cancer had been recommended for average-risk men aged 50 or older until 2012 when the U.S. Preventive Service Task Force (USPSTF) recommended that the screening should be an informed decision between patients and their providers [36] (although the implementation of the informed decision might start prior to 2012). In the MHS, all men aged 50 or older are provided with an opportunity to receive annual PSA screening tests and digital rectal examinations without out-of-pocket costs [37, 38]. As a result, the MHS beneficiaries in this age group might be more likely to use PSA screening and therefore have improved survival. However, our findings of no significant differences between <2010 (no PSA used) and 2010–2013 (PSA used) as well as with versus without adjustment for PSA for 2010–2013 suggest that the potential effects of PSA screening might be limited, and other factors might pay a role.

In the stratified analysis by race, both White patients and Black patients in the MHS showed longer survival than their counterparts in the U.S. general population. While the race-specific HRs were not significantly different as a whole in the heterogeneity test, the 32% lower risk of death in the MHS than the general U.S. population among Black patients in contrast to 25% among White patients is noteworthy. In a large study of based on the National Cancer Database, Krimphove et al. reported that among variables of demographics, tumour characteristics and treatment variables related to access to care, access-related variables explained 84.7% of the excess risk of death in Black patients with advanced prostate cancer [11]. Survival outcome among Black patients and White patients was reported to be similar in a number of studies performed in equal access health care system [39–43]. Although our study was not aimed to address racial disparity in prostate cancer survival, the findings of our study suggest the significance of accessibility health care for reducing racial disparity in survival of advanced prostate cancer patients. The survival benefit of the MHS among Asian/Pacific Islander was not present, likely due to a small number of Asian/Pacific Inslander patients in ACTUR that limited the statistical power.

In stratified analysis by tumour grade, the HRs were lower for patients with well-differentiated tumours, indicating greater survival benefit for these patients than patients with more aggressive tumours. This is different from our previous studies on glioma [18], in which patients with aggressive glioblastoma from the MHS benefited more compared to those with less aggressive brain tumours. While the reasons for the different findings are unclear, we do not exclude the potential impacts of comorbidities on the study results. Compared to patients with high-grade cancer, patients with low-grade tumours are more likely to die of causes other than prostate cancer, given the same tumour stage. Therefore, the greater benefits of universal health care for well-differentiated tumours might result from a lower likelihood of dying of other causes in the ACTUR patients than their SEER counterparts. However, due to the lack of information on comorbidities in SEER data, we are unable to evaluate the effects of comorbidity on the study results.

The longer survival of MHS beneficiaries than SEER patients was observed with adjustment for tumour stage and grade, and across all tumour stages and grades in the stratified analyses. This suggests that post-diagnosis treatments and services, given tumour stage and grade, may contribute to the improved survival among MHS beneficiaries with advanced stage of the disease. However, the MHS patients had longer survival than the SEER patients regardless of surgery or radiation treatment, which implies that the receipt of surgery or radiation treatment might not be a main contributor to the survival difference between the two patient populations. Nevertheless, the effects of treatment on the survival difference could not be thoroughly evaluated due to lack of information on chemotherapy and hormone therapy in SEER and lack of detailed and complete information on treatment such as timeliness of treatment in both databases.

Our study has some limitations. In addition to the lack of detailed treatment information, there were no (SEER) or incomplete (ACTUR) data on comorbidity. Therefore, the possible effects of comorbidities on all-cause death could not be ruled out. Nevertheless, because all patients had advanced stage prostate cancer, which has a high case fatality, most patients might die of the disease and the effects of comorbidities on the results may be limited. The “healthy worker effect” should be considered when comparing active-duty military personnel to general public because active-duty members must maintain required fitness level to stay in the military. In this study, active-duty military patients constituted only 3.9% of all patients, thus the impact of “healthy worker effect” due to the active-duty component might be limited. However, military retirees may constitute a large proportion of the patients. We do not exclude the possible “healthy worker effect” from the retirees in the study since they might be healthier than the general population of similar age.

In conclusion, this study suggests that the MHS’s universal care has translated into improved survival of patients with advanced prostate cancer. Future studies are warranted to identify specific factors along the cancer care continuum, which contribute to the improved survival in the MHS.

Disclaimer

The contents of this publication are the sole responsibility of the author(s) and do not necessarily reflect the views, opinions, or policies of Uniformed Services University of the Health Sciences (USUHS), The Henry M. Jackson Foundation for the Advancement of Military Medicine, Inc., the Department of Defense (DoD), the Departments of the Army, Navy, or Air Force. Mention of trade names, commercial products, or organisations does not imply endorsement by the U.S. Government.

Author contributions

Each of the authors (JL, DN, JJ, GTC, CDS, and KZ) significantly contributed to the project conception that led to data acquisition and results interpretation. All authors contributed to manuscript drafting, revision, and final approval. The corresponding author (KZ) confirmed that he had full access to the data in the study and final responsibility for the decision to submit for publication.

Funding

This project was supported by John P. Murtha Cancer Center Research Program, Uniformed Services University of the Health Sciences under the auspices of the Henry M. Jackson Foundation for the Advancement of Military Medicine, Inc.

Data availability

The data sets generated during and/or analysed during the current study are not publicly available following DoD MHS regulations.

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

This study was based on the non-identifiable ACTUR data and SEER de-identified public use data. The study was approved by the institutional review board of Walter Reed National Military Medical Center. The study was performed in accordance with the Declaration of Helsinki.

Consent for publication

Not applicable.