Abstract
OBJECTIVE:
To evaluate the effects of race and insurance status on the use of brachytherapy for treatment of cervical cancer.
METHODS:
This is a retrospective cohort study of the National Cancer Database. We identified 25,223 patients diagnosed with stage IB2 through IVA cervical cancer who received radiation therapy during their primary treatment from 2004 to 2015. A univariate analysis was used to assess covariate association with brachytherapy. A multivariable regression model was used to evaluate the effect of race and insurance status on rates of brachytherapy treatment. The Cox proportional hazards model and the multiplicative hazard model were used to evaluate overall survival. P<.05 indicated a statistically significant difference for comparisons of primary and secondary outcomes.
RESULTS:
Non-Hispanic black patients received brachytherapy at a significantly lower rate than non-Hispanic white patients (odds ratio [OR] 0.93;95% CI 0.86–0.99; P=.036); Hispanic (OR 0.93; 95% CI 0.85–1.02; P=.115) and Asian (OR 1.13; 95% CI 0.99–1.29; P=.074) patients received brachytherapy at similar rates. Compared with patients with private insurance, those who were uninsured (OR 0.72; 95% CI 0.65–0.79; P<001), had Medicaid (OR 0.83; 95% CI 0.77–0.89; P<001) or Medicare insurance (OR 0.85; 95% CI 0.78–0.92; P<001) were less likely to receive brachytherapy. Brachytherapy was not found to be a mediator of race and insurance-related disparities in overall survival.
CONCLUSION:
Racial and insurance disparities exist for those who receive brachytherapy, with many patients not receiving the standard of care, but overall survival was not affected.
Cervical cancer is the third most prevalent cancer among women in the world and disproportionately affects racial and ethnic minorities in the United States.1–3 Non-Hispanic black women have an increased cervical cancer incidence and poorer survival compared with non-Hispanic white women.1 In fact, non-Hispanic black women are twice as likely to die from cervical cancer as white women.4 Similarly, multiple studies suggest that this disparity exists for women who are uninsured or have Medicaid insurance, compared with those with private insurance.5,6 However, there is a paucity of literature investigating whether these groups receive brachytherapy at different rates compared with their non-Hispanic white and privately insured counterparts, and whether those differences contribute to the observed disparity in cervical cancer mortality.
A National Cancer Database study by Robin et al7 suggested that fewer than half of patients in the United States receive the standard of care treatment with chemo-radiation and brachytherapy, despite known survival benefits. This statistic disproportionally represents minorities and those with suboptimal insurance. The American Brachytherapy Society recommends that all patients receiving radiation therapy for cervical cancer should receive brachytherapy as a component of their treatment.8 This is reiterated in the current National Comprehensive Cancer Network practice guidelines, particularly for patients with International Federation of Gynecology and Obstetrics (FIGO) stage IB2 through IVA disease.9 Numerous studies have shown that brachytherapy is strongly associated with higher rates of survival and, thus, is the standard of care.10,11
The aim of this study was to explore whether racial minorities or uninsured or underinsured patients receive brachytherapy at the same rates as non-Hispanic white and privately insured patients. Our secondary aim was to investigate whether differences observed in rates of brachytherapy contribute to disparities in cervical cancer mortality among racial minorities and among those with Medicaid, Medicare, or no insurance.
METHODS
This retrospective cohort study was granted exempt status by the institutional review board at Abington Hospital-Jefferson Health. This study was deemed exempt from full institutional review board review because it involved the study of existing, nonidentifiable data that were collected before the project began. Proposal ID 2015.3463 was accepted by the National Cancer Database. The National Cancer Database is a nationwide oncology outcomes database for approximately 1,500 Commission on Cancer accredited cancer programs in the United States. Data are collected and submitted using nationally standardized data item and coding definitions as defined in the Commission on Cancer’s Facility Oncology Registry Data Standards. Data reported from Commission on Cancer-approved hospitals are abstracted from patient charts by Certified Tumor Registrars who undergo training specific to the cancer registry. All data submitted to the National Cancer Database undergo integrity checks, and internal quality monitoring and validity reviews. Additionally, every few years surveyors from the Commission on Cancer evaluate each hospital’s data collection processes. We were awarded a National Cancer Database Participant User File for all patients diagnosed with cervical cancer between 2004 and 2015. This included 115,747 patients. Of these patients, 53,273 underwent radiation therapy as part of their primary treatment. Radiation therapy includes external beam with or without vaginal brachytherapy. Only those patients with FIGO stage IB2, II, III, and IVA disease were included in the analysis, but patients with stage IB2 disease who received surgery before radiation therapy were excluded. Finally, patients with incomplete data were also excluded. The final cohort included 25,223 patients who received radiation therapy during their primary treatment. Of note, an additional 3,012 patients were offered radiation as part of their primary treatment by their physician but did not receive it.
We employed a logistic regression model to evaluate the effects of covariates included in this study: age, race, insurance status, high school completion, comorbidities (Charlson-Deyo score), residence, year of diagnosis, median household income, facility location, and clinical stage. Race subgroups were categorized as non-Hispanic white, non-Hispanic black, Hispanic, Asian, and other race. Insurance status included those with private insurance, Medicaid, Medicare, other government insurance (eg, Department of Veterans Affairs, Indian Health Service, Public Health Service), and those who are uninsured.
A univariate association of brachytherapy receipt with each of the covariates previously described was evaluated using the Wilcoxon two-sample test for age and the Pearson’s χ2 for the remainder of the covariates. We used a multivariate logistic regression model to discern the effect of race, insurance, and other covariates on rates of brachytherapy treatment. Pearson’s χ2 test was used to test the association between insurance status and race and the given reason for no radiation treatment. Duration of radiation was dichotomized as 0–56 days (0–8 weeks) compared with more than 56 days (more than 8 weeks). Differences in the duration of radiation were also evaluated using Pearson’s χ2 test.
The Cox proportional hazards model and the multiplicative hazard model (an extension of the Cox model that allows incorporation of nonparametric time-varying coefficients for the predictors that violate the proportional hazards assumption) were used to evaluate the overall survival using separate models for cancer stage (IB2, II, III, and IVA).12 Data from only 2004 to 2014 were included in survival models, because outcomes for the 2015 diagnosis year were not available. Also, 286 participants in the smallest insurance group (other government) were excluded from the data set for survival analysis. Race, insurance group, Charlson-Deyo score, year of diagnosis, facility location, dichotomized age (younger than 65 years vs at least 65 years), brachytherapy, and interactions between brachytherapy and other covariates were considered as predictors of overall survival in each stage. Only significant predictors P<.05) were included in the final models. The survival models included the propensity score adjustment to decrease potential selection bias in retrospective analyses.13 Propensity scores were calculated as predicted probabilities of receipt of brachytherapy from the logistic regression model described previously. The propensity adjustment was implemented using the inverse probability of brachytherapy treatment weighting.14 The data were analyzed in R (The R Foundation for Statistical Computing at R-project.org) and SAS 9.4. P<.05 indicated a statistically significant difference for all comparisons of primary and secondary outcomes. However, P<.001 was used for the propensity analysis to allow for more informative conclusions given the large number of comparisons.
RESULTS
Of the 25,223 patients who received radiation therapy during primary treatment for cervical cancer, 16,025 (63%) self-reported as non-Hispanic white, 4,735 (19%) as non-Hispanic black, 2,846 (11%) as Hispanic, 1,178 (5%) as Asian, and 439 (2%) as another race. A total of 9,207 (37%) patients were reported as having private insurance, whereas 2,828 (11%) were uninsured, 5,890 (23%) had Medicaid, 7,012 (28%) had Medicare and 286 (1%) had other government insurance.
Univariate analysis by receipt of brachytherapy is shown in Table 1. Brachytherapy treatment was significantly associated with all covariates (P=.008 for residence and P<.001 for all others) except for income status (P=.076). In the multivariable logistic regression model, the interaction between insurance and race was not significant (P=.315) and was excluded. In the model with main effects only, all predictors were significant except for high school completion (P=.086) and income status (P=.297). The final model included race, insurance, Charlson-Deyo score, residence, year of diagnosis, age, treatment facility location, and FIGO clinical stage.
Table 1.
Brachytherapy by Sociodemographic and Clinical Characteristics Among Women With Stage IB2 to IVA Cervical Cancer in the National Cancer Database (2004–2015)
| Variable | Brachytherapy (n=16,007) | No Brachytherapy (n=9,216) | P |
|---|---|---|---|
| Race | |||
| Non-Hispanic white | 10,331 (64) | 5,694 (36) | <.001 |
| Hispanic | 1,712 (60) | 1,134 (40) | |
| Non-Hispanic black | 2,904 (61) | 1,831 (39) | |
| Asian | 760 (65) | 418 (35) | |
| Other | 300 (68) | 139 (32) | |
| Insurance | |||
| Private | 6,404 (70) | 2,803 (30) | <.001 |
| Uninsured | 1,701 (60) | 1,127 (40) | |
| Medicaid | 3,767 (64) | 2,123 (36) | |
| Medicare | 3,941 (56) | 3,071 (44) | |
| Other government | 194 (68) | 92 (32) | |
| High school completion (%) | |||
| Quartile 1 (less than 79) | 4,360 (61) | 2,791 (39) | <.001 |
| Quartile 2 (79.1–87) | 4,841 (64) | 2,778 (36) | |
| Quartile 3 (87.1–93) | 4,583 (65) | 2,416 (35) | |
| Quartile 4 (greater than 93) | 2,223 (64) | 1,231 (36) | |
| Median household income ($) | |||
| Quartile 1 (less than 38,000) | 4,405 (63) | 2,643 (38) | .076 |
| Quartile 2 (38,000–47,999) | 4,209 (64) | 2,359 (36) | |
| Quartile 3 (48,000–62,999) | 3,948 (63) | 2,317 (37) | |
| Quartile 4 (greater than 63,000) | 3,445 (64) | 1,897 (36) | |
| Charlson-Deyo score | |||
| 0 | 13,495 (64) | 7,483 (36) | <.001 |
| 1 | 1,979 (61) | 1,282 (39) | |
| 2 | 395 (55) | 318 (45) | |
| 3 or higher | 138 (51) | 133 (49) | |
| Residence | |||
| Metropolitan county | 13,125 (63) | 7,683 (37) | .008 |
| Urban county | 2,587 (66) | 1,355 (34) | |
| Rural county | 295 (62) | 178 (38) | |
| Treatment facility location | |||
| New England | 591 (64) | 328 (36) | <.001 |
| Middle Atlantic | 2,761 (68) | 1,328 (32) | |
| South Atlantic | 3,479 (64) | 1,918 (36) | |
| East North central | 2,847 (66) | 1,461 (34) | |
| East South central | 1,391 (61) | 890 (39) | |
| West North central | 1,187 (69) | 533 (31) | |
| West South central | 1,421 (56) | 1,127 (44) | |
| Mountain | 513 (59) | 356 (41) | |
| Pacific | 1,817 (59) | 1,275 (41) | |
| Year of diagnosis | |||
| 2004 | 1,033 (58) | 760 (42) | <.001 |
| 2005 | 1,145 (63) | 686 (37) | |
| 2006 | 1,112 (60) | 736 (40) | |
| 2007 | 1,141 (59) | 782 (41) | |
| 2008 | 1,220 (61) | 772 (39) | |
| 2009 | 1,286 (62) | 803 (38) | |
| 2010 | 1,363 (61) | 857 (39) | |
| 2011 | 1,560 (68) | 735 (32) | |
| 2012 | 1,522 (68) | 727 (32) | |
| 2013 | 1,582 (69) | 721 (31) | |
| 2014 | 1,537 (66) | 790 (34) | |
| 2015 | 1,506 (64) | 847 (36) | |
| FIGO clinical stage | |||
| IB2 | 1,603 (75) | 528 (25) | <.001 |
| II | 6,737 (71) | 2,819 (29) | |
| III | 7,064 (60) | 4,761 (40) | |
| IVA | 603 (35) | 1,108 (65) |
FIGO, International Federation of Gynecology and Obstetrics.
Data are n (row %) unless otherwise specified.
Table 2 highlights the results of the multivariate regression analysis. All reported odds ratios (ORs) are adjusted ORs. Hispanic patients (OR 0.93; 95% CI 0.85–1.02; P=.115) and Asian patients (OR 1.13; 95% CI 0.99–1.29; P=.074) received brachytherapy at similar rates to their non-Hispanic white counterparts. However, non-Hispanic black patients (OR 0.93; 95% CI 0.86–0.99; P=.036) received brachytherapy at significantly lower rates compared with non-Hispanic white patients. Compared with patients with private insurance, those who were uninsured (OR 0.72; 95% CI 0.65–0.79; P<.001), had Medicaid (OR 0.83; 95% CI 0.77–0.89; P<.001), or had Medicare (OR 0.85; 95% CI 0.78–0.92; P<.001) were significantly less likely to receive brachytherapy. Patients with Charlson-Deyo scores of 2 (OR 0.78; 95% CI 0.67–0.92; P=.003), or 3 or more (OR 0.60; 95% CI 0.46–0.77; P<.001) were less likely to receive brachytherapy compared with patients with a Charlson-Deyo score of 0. Patients who lived in metropolitan counties (the most populated, 250,000 to more than 1 million residents) received significantly less brachytherapy than those in urban counties (2,500 to 250,000 residents) (OR 1.13; 95% CI 1.05–1.22; P=.002). However, patients in rural counties (the least populated, fewer than 2,500 residents) received brachytherapy at similar rates to patients in metropolitan counties (OR 1.01; 95% CI 0.83–1.24; P=.886). There was a trend towards increased use of brachytherapy in patients who were diagnosed starting in 2010 as compared with patients diagnosed before 2010. Compared with patients with stage IB2 disease, those with stage III (OR 0.90; 95% CI 0.48–0.60; P<.001) or IVA disease (OR 0.20; 95% CI 0.17–0.23; P<.001) received significantly less brachytherapy.
Table 2.
Multivariate Regression Analysis Evaluating Predictors of Brachytherapy
| Variable | OR | 95% CI | P |
|---|---|---|---|
| Age | 0.08 | 0.98–0.98 | <.001 |
| Race (vs non-Hispanic white) | |||
| Hispanic | 0.93 | 0.85–1.02 | .115 |
| Non-Hispanic black | 0.93 | 0.86–0.99 | .036 |
| Asian | 1.13 | 0.99–1.29 | .074 |
| Other | 1.12 | 0.91–1.39 | .284 |
| Insurance (vs private) | |||
| Uninsured | 0.72 | 0.65–0.79 | <.001 |
| Medicaid | 0.83 | 0.77–0.89 | <.001 |
| Medicare | 0.85 | 0.78–0.92 | <.001 |
| Other government | 1.01 | 0.77–1.31 | .940 |
| Charlson/Deyo score (vs 0) | |||
| 1 | 0.94 | 0.87–1.02 | .156 |
| 2 | 0.78 | 0.67–0.92 | .003 |
| 3 or higher | 0.60 | 0.46–0.77 | <.001 |
| Residence (vs metropolitan county) | |||
| Urban county | 1.13 | 1.05–1.22 | .002 |
| Rural county | 1.01 | 0.83–1.24 | .886 |
| Year of diagnosis (vs 2004) | |||
| 2005 | 1.20 | 1.05–1.38 | .010 |
| 2006 | 1.09 | 0.95–1.25 | .213 |
| 2007 | 1.05 | 0.92–1.20 | .469 |
| 2008 | 1.17 | 1.02–1.34 | .023 |
| 2009 | 1.18 | 1.03–1.35 | .017 |
| 2010 | 1.20 | 1.05–1.37 | .006 |
| 2011 | 1.63 | 1.42–1.89 | <.001 |
| 2012 | 1.60 | 1.40–1.83 | <.001 |
| 2013 | 1.69 | 1.48–1.93 | <.001 |
| 2014 | 1.47 | 1.29–1.68 | <.001 |
| 2015 | 1.34 | 1.18–1.53 | <.001 |
| Treatment facility location (vs Pacific) | |||
| New England | 1.42 | 1.21–1.67 | <.001 |
| Middle Atlantic | 1.57 | 1.41–1.73 | <.001 |
| South Atlantic | 1.33 | 1.20–1.47 | <.001 |
| East North central | 1.40 | 1.26–1.56 | <.001 |
| East South central | 1.09 | 0.96–1.23 | .171 |
| West North central | 1.58 | 1.38–1.81 | <.001 |
| West South central | 0.89 | 0.79–0.99 | .035 |
| Mountain | 1.03 | 0.88–1.21 | .067 |
| FIGO clinical stage (vs IB2) | |||
| II | 0.90 | 0.82–1.00 | .050 |
| III | 0.53 | 0.48–0.60 | <.001 |
| IVA | 0.20 | 0.17–0.23 | <.001 |
OR, odds ratio; FIGO, International Federation of Gynecology and Obstetrics.
High school completion and median household income were not included in this multivariate analysis.
Univariate analysis illustrated that non-Hispanic black patients and Hispanic patients were less likely to complete radiation therapy within the standard 56 days compared with non-Hispanic white patients (P=.007) (Table 3). Similarly, patients with Medicaid insurance were less likely to complete radiation therapy within 56 days compared with privately insured patients (P<.001).
Table 3.
Duration of Radiation Therapy by Race and Insurance Status
| 56 d or Less | Longer than 56 d | P | |
|---|---|---|---|
| Race | |||
| Non-Hispanic white | 8,751 (58) | 6,460 (42) | .007 |
| Hispanic | 1,448 (57) | 1,114 (43) | |
| Non-Hispanic black | 2,469 (55) | 2,029 (45) | |
| Asian | 649 (59) | 448 (41) | |
| Other | 251 (60) | 165 (40) | |
| Total | 13,568 (57) | 10,216 (43) | |
| Insurance | |||
| Private | 1,502 (57) | 1,132 (43) | <.001 |
| Uninsured | 5,054 (58) | 3,664 (42) | |
| Medicaid | 2,904 (53) | 2,595 (47) | |
| Medicare | 3,945 (59) | 2,713 (41) | |
| Other government | 163 (59) | 112 (41) | |
| Total | 13,568 (57) | 10,216 (43) |
Data are n (row %) unless otherwise specified.
Patients who did not receive radiation as part of their primary cervical cancer treatment were categorized into one of two groups: “not planned or contraindicated” or “recommended but not received” (Table 4). This analysis suggested that non-Hispanic black and Asian patients, compared with non-Hispanic white (P=.047), as well as patients who are uninsured or covered by Medicaid or Medicare insurance, compared with private insurance (P<.001), were more likely to fall into the “recommended but not received” group.
Table 4.
Reason Patient Did Not Receive Radiation Therapy for Treatment of Cervical Cancer, by Race and Insurance Status
| Not Planned or Contraindicated* | Recommended But Not Received† | P | |
|---|---|---|---|
| Race | |||
| Non-Hispanic white | 1,572 (84) | 309 (16) | .047 |
| Hispanic | 298 (84) | 57 (16) | |
| Non-Hispanic black | 466 (80) | 120 (20) | |
| Asian | 105 (76) | 33 (24) | |
| Other | 42 (81) | 10 (19) | |
| Total | 2,483 (83) | 529 (17) | |
| Insurance | |||
| Private | 860 (89) | 105 (11) | <.001 |
| Uninsured | 299 (78) | 86 (22) | |
| Medicaid | 438 (80) | 107 (20) | |
| Medicare | 864 (79) | 229 (21) | |
| Other government | 22 (92) | 2 (8) | |
| Total | 2,483 (83) | 529 (17) |
Data are n (row %) unless otherwise specified.
Includes those who did not receive radiation because it was not a planned part of the first course of treatment owing to patient risk factors, or patient died before the therapy.
Includes patients to whom radiation was recommended but refused by the patient or family member or was not administered for an unknown reason.
The results of the final multiplicative hazard models with propensity score adjustment for OS in each stage are presented in Table 5. For every stage, there was a significant interaction between brachytherapy and age group (P=.015 for stage IB2 and P<.001 for stages II–IV); brachytherapy, age, and their interaction all violated the proportional hazard assumptions in each stage model. Therefore, the brachytherapy effect was evaluated separately in patients younger than 65 years old and in patients who were 65 years and older. Brachytherapy significantly improved overall survival in patients younger than 65 years old and in patients 65 years or older across all stages (Table 5 and Fig. 1, Appendix 1 [Appendix 1, available online at http://links.lww.com/AOG/B476].
Table 5.
Final Multiplicative Hazard Models With Propensity Score Adjustment for Overall Survival in International Federation of Gynecology and Obstetrics Stage IB2 Through IVA Disease
| FIGO IB2 | FIGO II | FIGO III | FIGO IVA | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Variable | HR | 99% CI | P | HR | 99% CI | P | HR | 99% CI | P | HR | 99% CI | P |
| Race (vs non-Hispanic white) | ||||||||||||
| Hispanic | 0.641 | 0.36–1.14 | .002 | 0.630 | 0.50–0.79 | <.001 | 0.676 | 0.57–0.80 | <.001 | 0.682 | 0.44–1.05 | .010 |
| Non-Hispanic black | 1.189 | 0.84–1.69 | .060 | 1.006 | 0.87–1.17 | .892 | 0.981 | 0.88–1.10 | .605 | 1.062 | 0.81–1.39 | .548 |
| Asian | 0.618 | 0.25–1.52 | .044 | 0.661 | 0.48–0.91 | <.001 | 0.774 | 0.61–0.98 | <.001 | 0.750 | 0.40–1.41 | .117 |
| Other | 0.676 | 0.19–2.43 | .244 | 0.869 | 0.52–1.46 | .319 | 0.725 | 0.48–1.09 | .011 | 0.988 | 0.34–2.84 | .967 |
| Insurance (vs private) | ||||||||||||
| Uninsured | 1.165 | 0.69–1.97 | .240 | 1.022 | 0.82–1.28 | .710 | 1.073 | 0.92–1.25 | .144 | 0.985 | 0.69–1.41 | .911 |
| Medicaid | 1.510 | 1.01–2.26 | <.001 | 1.210 | 1.02–1.43 | <.001 | 1.152 | 1.02–1.30 | <.001 | 1.041 | 0.78–1.38 | .678 |
| Medicare | 1.645 | 0.98–2.56 | <.001 | 1.461 | 1.21–1.76 | <.001 | 1.217 | 1.05–1.41 | <.001 | 1.171 | 0.84–1.64 | .199 |
| Charlson-Deyo score (vs 0) | ||||||||||||
| 1 | 1.405 | 0.94–2.11 | .003 | 1.374 | 1.17–1.61 | <.001 | 1.299 | 1.15–1.47 | <.001 | 1.111 | 0.83–1.48 | .299 |
| 2 | 1.458 | 0.66–3.24 | .098 | 1.595 | 1.20–2.12 | <.001 | 1.533 | 1.22–1.93 | <.001 | 1.196 | 0.72–0.98 | .326 |
| 3 or higher | 6.028 | (2.82–12.89 | <.001 | 1.897 | 1.14–3.16 | <.001 | 1.699 | 1.16–2.50 | <.001 | 1.523 | 0.62–3.73 | .288 |
| Year of diagnosis (vs 2004) | ||||||||||||
| 2005 | 0.755 | 0.39–1.48 | .109 | 1.024 | 0.82–1.28 | .714 | 0.920 | 0.75–1.12 | .217 | 0.976 | 0.60–1.60 | .900 |
| 2006 | 1.023 | 0.54–1.96 | .892 | 0.882 | 0.70–1.11 | .055 | 0.843 | 0.69–1.03 | .011 | 1.072 | 0.66–1.75 | .718 |
| 2007 | 1.177 | 0.61–2.27 | .368 | 0.921 | 0.73–1.17 | .210 | 0.841 | 0.69–1.02 | .008 | 1.034 | 0.64–1.67 | .853 |
| 2008 | 0.828 | 0.43–1.61 | .279 | 0.945 | 0.74–1.20 | .403 | 0.785 | 0.64–0.96 | <.001 | 0.785 | 0.49–1.27 | .155 |
| 2009 | 0.899 | 0.45–1.81 | .565 | 0.862 | 0.68–1.10 | .025 | 0.779 | 0.64–0.95 | <.001 | 0.844 | 0.54–1.33 | .281 |
| 2010 | 0.675 | 0.35–1.29 | .021 | 0.762 | 0.59–0.99 | <.001 | 0.752 | 0.62–0.92 | <.001 | 0.754 | 0.48–1.18 | .068 |
| 2011 | 0.630 | 0.32–1.25 | .012 | 0.779 | 0.60–1.02 | .001 | 0.722 | 0.59–0.88 | <.001 | 0.799 | 0.50–1.27 | .153 |
| 2012 | 0.646 | 0.32–1.29 | .020 | 0.807 | 0.61–1.06 | .006 | 0.737 | 0.60–0.91 | <.001 | 0.804 | 0.50–1.29 | .174 |
| 2013 | 0.715 | 0.36–1.44 | .078 | 0.830 | 0.62–1.10 | .021 | 0.760 | 0.62–0.94 | <.001 | 0.756 | 0.46–1.24 | .088 |
| 2014 | 0.870 | 0.41–1.85 | .477 | 0.724 | 0.51–1.02 | .001 | 0.659 | 0.52–0.83 | <.001 | 0.912 | 0.56–1.48 | .555 |
| Treatment facility region (vs Pacific) | ||||||||||||
| New England | — | — | — | 0.941 | 0.66–1.34 | .546 | 0.977 | 0.75–1.28 | .788 | 0.802 | 0.46–1.39 | .199 |
| Middle Atlantic | — | — | — | 0.903 | 0.71–1.14 | .117 | 1.034 | 0.87–1.23 | .535 | 0.680 | 0.44–1.05 | .004 |
| South Atlantic | — | — | — | 1.034 | 0.83–1.29 | .593 | 1.064 | 0.90–1.26 | .238 | 0.865 | 0.58–1.30 | .269 |
| East North central | — | — | — | 0.968 | 0.77–1.22 | .604 | 1.090 | 0.92–1.30 | .112 | 0.824 | 0.54–1.25 | .140 |
| East South central | — | — | — | 0.945 | 0.72–1.23 | .439 | 1.149 | 0.94–1.40 | .029 | 1.052 | 0.66–1.68 | .752 |
| West North central | — | — | — | 1.038 | 0.78–1.38 | .634 | 1.096 | 0.89–1.36 | .166 | 0.727 | 0.44–1.22 | .074 |
| West South central | — | — | — | 0.969 | 0.74–1.26 | .663 | 0.959 | 0.79–1.17 | .502 | 0.859 | 0.54–1.37 | .327 |
| Mountain | — | — | — | 0.993 | 0.69–1.43 | .944 | 1.030 | 0.79–1.35 | .737 | 0.981 | 0.55–1.75 | .930 |
| Age (y) (older than 65 vs younger than 65) | — | — | <.001 | — | — | <.001 | — | — | <.001 | — | — | .003 |
| Brachytherapy (yes vs no) | ||||||||||||
| Age younger than 65 y | — | — | .001 | — | — | <.001 | — | — | <.001 | — | — | <.001 |
| Age 65 y or older | — | — | .010 | — | — | <.001 | — | — | <.001 | — | — | <.001 |
HR, hazard ratio; FIGO, International Federation of Gynecology and Obstetrics.
Fig. 1.
Overall survival stratified by brachytherapy and categorized age. International Federation of Gynecology and Obstetrics (FIGO) IB2 (A); FIGO II (B); FIGO III (C); FIGO IVA (D).
The results reported in Table 5 imply that the differences in survival associated with race and insurance were significant with adjustment for brachytherapy and other significant predictors of overall survival. Across all stages, Hispanic and Asian patients had a lower hazard of death compared with non-Hispanic white patients. Asian patients had a lower hazard of death compared with non-Hispanic white patients for FIGO IB2, II, and III disease. Non-Hispanic black patients had similar hazard ratios to non-Hispanic white patients across all stages. Additionally, patients with Medicaid and Medicare had lower survival probabilities than those with a private insurance in stage IB2, II, and III disease. Hazard ratios were similar for stage IV disease.
Figure 2 presents Kaplan-Meier plots by race and insurance group with brachytherapy as strata for each clinical stage. Appendixes 2 and 3, available online at http://links.lww.com/AOG/B476, specifies the number of patients at risk in 12 month intervals. Differences in survival by race and insurance group were similar regardless of brachytherapy receipt for all stages. In general, Asian and Hispanic patients had the best survival probabilities, whereas non-Hispanic white and non-Hispanic black patients had the worst survival probabilities across all stages. In addition, patients with private insurance or those who were uninsured had the best survival probabilities, whereas patients with Medicare or Medicaid insurance had the worst probabilities. This remained true across all stages.
Fig. 2.
Overall survival by race (A) and insurance group (B) with brachytherapy as strata for each International Federation of Gynecology and Obstetrics (FIGO) clinical stage.
DISCUSSION
Racial and ethnic disparities in cervical cancer diagnosis, treatment, and survival have been well documented.1,4 Similarly, multiple studies suggest that these disparities also exist for women who are uninsured or are covered by Medicaid insurance.5,6,15 Nevertheless, there are few studies that specifically investigate disparities in brachytherapy for treatment of cervical cancer.15,16
Current literature remains conflicted with regard to racial disparities and treatment. Some suggest that minorities receive less treatment, whereas others have found no correlation.1,17–19 Markt et al20 found that non-Hispanic black women were more likely to receive radiation therapy than surgery, but that they receive brachytherapy at lower rates than their non-Hispanic white counterparts for late stage disease. Another study by Mundt et al18 illustrated that African American patients receive less brachytherapy than Caucasian patients and that this disparity can be attributed to technical and health problems amongst that patient population. However, in multivariate analysis, race was not found to be significant in terms of disease-free survival or cause specific survival.18
Review of National Cancer Database data in this study showed that non-Hispanic black patients, and those who are uninsured or covered by Medicaid or Medicare receive brachytherapy at lower rates than their non-Hispanic white and privately insured counterparts, respectively. This remains true when adjusting for stage and socioeconomic factors. Though there are few published studies on this subject, our data are consistent with the existing literature.7,15,21 In addition, we found that despite these differences, brachytherapy was not a mediator of race- or insurance-related disparities in overall survival because overall survival for all race and insurance groups remained the same with and without brachytherapy across all cancer stages.
Our study also showed that non-Hispanic black and Hispanic patients in the National Cancer Database appear less likely to complete their radiation therapy (external beam radiation therapy +brachytherapy) within the current standard of care of 56 days.22 Medicaid patients were also less likely to complete their radiation therapy in 56 days compared with those with private insurance. Multiple studies have demonstrated poorer local control and survival in patients with prolonged treatment.23–25 One retrospective cohort study by Cohen et al found that the most common cause of protracted radiation therapy was psychosocial stressors common in lower socioeconomic populations. However, in that study, insurance and race did not significantly affect the timely completion of therapy.26
There were a large number of patients in the database who qualified for radiation therapy as part of their primary treatment but did not receive it. We found that these patients were more likely to have Medicaid, Medicare, or no insurance. We were limited in our ability to interpret these data any further owing to the information provided in the database, but one can speculate that factors such as patient preference, transportation issues, unwillingness for providers to accept suboptimal insurance, and other social and financial barriers may have contributed. Another National Cancer Database study by Robin et al7 suggested that fewer than half of patients in the United States receive standard of care treatment for cervical cancer with chemo-radiation and brachytherapy, despite known survival benefits.
The idea that insurance and race affects cancer treatment has been well published in the field of gynecologic cancer. Goff et al27 demonstrated that payer status (private insurance compared with Medic-aid) affected a patient’s chance of undergoing optimal surgical management for ovarian cancer. In an analysis of 11 different cancer types, Harlan et al28 noted significantly lower adherence to treatment guidelines for black patients with Medicaid compared with black patients with Medicare or private insurance.
Differences in overall survival by race and insurance group were estimated to be similar regardless of brachytherapy status for all stages. This finding suggests that, despite observed differences in rates of brachytherapy, the differences seen in overall survival are likely due to other variables. However, we found that non-Hispanic black and non-Hispanic white patients have similar overall survival, whereas Hispanic patients have a better overall survival when controlling for insurance, Charlson-Deyo score, year of diagnosis, facility location, age, and brachytherapy. This contradicts existing literature, which suggests that non-Hispanic black patients have worse cancer survival than non-Hispanic white patients.1–5,29–31 It is important to consider that there may be other factors, such as variations in tumor biology between racial groups, which could be contributing to these results.
Additionally, our study found that Hispanic women receive brachytherapy at similar rates to non-Hispanic white women, however survival probabilities for Hispanic women were better. This is an interesting paradox because Hispanic women tend to have a similar socioeconomic status as non-Hispanic black women, yet Hispanic women have a survival advantage. This observation has been well documented in the literature.32,33 This could be attributed to differences in social support or other cultural influences that cannot be accounted for in our study.
Limitations to this study were those inherent to using a large database. Though we performed a logistic regression analysis to account for multiple variables, we could not correct for inaccurate coding, which is a concern with all similar databases. We attempted to account for factors that may have confounded our results regarding race and insurance disparities, but certain factors such as income and education were based on county data, not individual data. There was also no way to infer patient preference as it related to treatment selection. Given that this was such a large data set, we believe these intangible variables were stable across all groups.
Despite these limitations, we demonstrated that rates of brachytherapy for primary treatment of cervical cancer are lower among non-Hispanic black patients and patients with Medicare, Medicaid, and uninsured payer status. Therefore, these patients were less likely to receive the standard of care. However, despite the identified disparities, these did not have an effect on overall cancer survival.
Supplementary Material
Financial Disclosure
Inna Chervoneva and Misung Yi report having money paid to their institution from NIH/NCI. Mitchell I. Edelson has received payment by Blumberg & Wolk, LLC and Post & Schell, P.C. to serve as an expert witness for legal cases. His wife is an employee of Merck and receives a salary. Mark S. Shahin has received payment for serving on the speaker’s bureau for Tesaro, Merck, Astra Zeneca, and Clovis Oncology. The other authors did not report any potential conflicts of interest.
Footnotes
Presented as a poster at the Society of Gynecologic Oncology’s Winter Meeting, January 16–19, 2019, Lake Tahoe, California.
Each author has confirmed compliance with the journal’s requirements for authorship.
REFERENCES
- 1.Chatterjee S, Gupta D, Caputo TA, Holcomb K. Disparities in gynecological malignancies. Front Oncol 2016;6:36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Shavers VL, Brown ML. Racial and ethnic disparities in the receipt of cancer treatment. J Natl Cancer Inst 2002;94:334–57. [DOI] [PubMed] [Google Scholar]
- 3.U.S. Cancer Statistics Working Group. U.S. cancer statistics: data visualizations tool, based on November 2017 submission data (1999–2015). Available at: www.cdc.gov/cancer/dataviz. Retrieved May 1, 2019.
- 4.American Cancer Society. Cancer facts and figures for African Americans 2016–2018. Atlanta (GA): American Cancer Society; 2016. [Google Scholar]
- 5.Bradley CJ, Given CW, Roberts C. Health care disparities and cervical cancer. Am J Public Health 2004;94:2098–103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Ward E, Halpern M, Schrag N, Cokkinides V, DeSantis C, Bandi P, et al. Association of insurance with cancer care utilization and outcomes. CA Cancer J Clin 2008;58:9–31. [DOI] [PubMed] [Google Scholar]
- 7.Robin TP, Amini A, Schefter TE, Behbakht K, Fisher CM. Disparities in standard of care treatment and associated survival decrement in patients with locally advanced cervical cancer. Gynecol Oncol 2016;143:319–25. [DOI] [PubMed] [Google Scholar]
- 8.Viswanathan AN, Erikson B, Gaffney DK, Bosch W. Comparison and consensus guidelines for delineation of clinical target volume for CT- and MR-based brachytherapy in locally advanced cervical cancer. Int J Radiat Oncol Bio Phys 2014;90:320–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.National Comprehensive Cancer Network. Cervical cancer (version 3.2019). Available at: https://www.nccn.org/professionals/physician_gls/PDF/cervical.pdf. Retrieved January 3, 2019.
- 10.Logsdon MD, Eifel PJ. Figo IIIB squamous cell carcinoma of the cervix: an analysis of prognostic factors emphasizing the balance between external beam and intracavitary radiation therapy. Int J Radiat Oncol Biol Phys 1999;43:763–75. [DOI] [PubMed] [Google Scholar]
- 11.Holschneider CH, Petereit DG, Chu C, Hsu IC, Ioffe YJ, Klopp AH, et al. Brachytherapy: a critical component of primary radiation therapy for cervical cancer: from the Society of Gynecologic Oncology (SGO) and the American Brachytherapy Society (ABS). Gynecol Oncol 2019;152:540–7. [DOI] [PubMed] [Google Scholar]
- 12.Martinussen T, Scheike TH. Dynamic regression models for survival data. New York (NY): Springer Science & Business Media; 2007. [Google Scholar]
- 13.Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983;70:41–55. [Google Scholar]
- 14.Austin PC. An introduction to propensity-score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011;46:399–424. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Churilla T, Egleston B, Dong Y, Shaikh T, Murphy C, Mantia-Smaldone G, et al. Disparities in the management and outcome of cervical cancer in the United States according to health insurance status. Gynecol Oncol 2016;141:516–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Sharma C, Deutsch I, Horowitz DP, Hershman DL, Lewin SN, Lu YS, et al. Patterns of care and treatment outcomes for elderly women with cervical cancer. Cancer 2012;118:3618–26. [DOI] [PubMed] [Google Scholar]
- 17.Howell EA, Chen YT, Concato J. Differences in cervical cancer mortality among black and white women. Obstet Gynecol 1999;94:509–15. [DOI] [PubMed] [Google Scholar]
- 18.Mundt AJ, Connell PP, Campbell T, Hwang JH, Rotmensch J, Waggoner S. Race and clinical outcome in patients with carcinoma of the uterine cervix treated with radiation therapy. Gynecol Oncol 1998;71:151–8. [DOI] [PubMed] [Google Scholar]
- 19.Del Carmen MG, Montz FJ, Robert KM, Bristow RE, Bovicelli A, Cornelison T, et al. Ethnic differences in patterns of care of stage IA1 and IA2 cervical cancer: a SEER database study. Gynecol Oncol 1999;75:113–17. [DOI] [PubMed] [Google Scholar]
- 20.Markt SC, Tang T, Cronin AM, Katz IT, Howitt BE, Horowitz NS, et al. Insurance status and cancer treatment mediate the association between race/ethnicity and cervical cancer survival. PLoS One 2018;13:e0193047. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Parikh-Patel A, Morris C, Kizer KW. Disparities in quality of care: the role of health insurance and population demographics. Medicine 2017;96:e9125. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Nag S, Erickson B, Thomadsen B, Orton C, Demanes JD, Petereit D, et al. The American Brachytherapy Society recommendations for high-dose brachytherapy for carcinoma of the cervix. Int J Radiat Oncol Biol Phys 2000;48:201–11. [DOI] [PubMed] [Google Scholar]
- 23.Grinsky T, Rey A, Roche B, Haie C, Gerbaulet A, Randrianarivello H, et al. Overall treatment time in advanced cervical carcinomas: a critical parameter in treatment outcome. Int J Radiat Oncol Biol Phys 1993;27:1051–6. [DOI] [PubMed] [Google Scholar]
- 24.Perez CA, Grigsby PW, Castro-Vita H, Lockett MA. Carcinoma of the uterine cervix. I. Impact of prolongation of overall treatment time and timing of brachytherapy on outcome of radiation. Int J Radiat Oncol Biol Phys 1995;32:1275–88. [DOI] [PubMed] [Google Scholar]
- 25.Petereit DG, Sarkaria JN, Chappell R, Fowler JF, Hartmann TJ, Kinsella TJ, et al. The adverse effect of treatment prolongation in cervical carcinoma. Int J Radiat Oncol Biol Phys 1995;32: 1301–7. [DOI] [PubMed] [Google Scholar]
- 26.Cohen J, Harper A, Nichols EM, Rao GG, Mohindra P, Roque DM. Barriers to timely completion of radiation therapy in patients with cervical cancer in an urban tertiary care center. Cureus 2017;9:e1681. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Goff BA, Matthews BJ, Larson EH, Andrilla CHA, Wynn M, Lishner DM, et al. Predictors of comprehensive surgical treatment in patients with ovarian cancer. Cancer 2007;109:2031–2042. [DOI] [PubMed] [Google Scholar]
- 28.Harlan LC, Greene AL, Clegg LX, Mooney M, Stevens JL, Brown ML. Insurance status and the use of guideline therapy in the treatment of selected cancers. J Clin Oncol 2005;23:9079–88. [DOI] [PubMed] [Google Scholar]
- 29.Siegel RL, Miller KD, Jemal A. Cancer statistics for Hispanics/Latinos, 2015. CA Cancer J Clin 2015;65:5–29. [DOI] [PubMed] [Google Scholar]
- 30.DeSantis C, Naishadham D, Jemal A. Cancer statistics for African Americans. CA Cancer J Clin 2013 2013;63:151–66. [DOI] [PubMed] [Google Scholar]
- 31.Harper S, Lynch J. Methods for measuring cancer disparities: using data relevant to healthy people 2010 cancer-related objectives. NCI cancer surveillance monograph series, number 6. NIH Publication No. 05-5777. Bethesda (MD): National Cancer Institute; 2005. [Google Scholar]
- 32.Coker AL, Desimone CP, Eggleston KS, White AL, Williams M. Ethnic disparities in cervical cancer survival among Texas women. J Womens Health (Larchmt) 2009;18:1577–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Liao Y, Cooper RS, Cao G, Durazo-Arvizu R, Kaufman JS, Luke A, et al. Mortality patterns among adult Hispanics: findings from the NHIS, 1986 to 1990. Am J Public Health 1998; 88:227–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.