In the United States, cervical cancer will be diagnosed in an estimated 12,990 women in 2016 and will cause an estimated 4,100 deaths.1 In cervical cancer, racial and ethnic disparities in mortality have persisted for many decades and socioeconomic disparities have increased in recent years.2,3 The Institute of Medicine's report on disparities noted that after cancer is diagnosed, minorities and those of lower socioeconomic status are less likely to receive high-quality care, thereby contributing to disparities in outcomes.4 Although measuring quality of care in health care is challenging, the National Comprehensive Cancer Network treatment guidelines provide a framework to do so.5 Adherence to guideline-based care has been shown to be associated with improved survival in ovarian, colorectal, and head and neck squamous cell carcinoma.6–9 However, in cervical cancer, the specific effect of receiving guideline-based care has not been studied.
It is often hypothesized that minorities and women of low socioeconomic status are treated in low-volume hospitals, where they are less likely to receive guideline-based care. However, little data exist about racial–ethnic disparities in receipt of guideline-based care and their relationship to hospital volume. In this context, the primary objective of this study was to investigate racial–ethnic disparities in guideline-based care in locally advanced cervical cancer and their relationship to hospital volume. The secondary objective of this study was to quantify the independent effect of guideline-based care on overall survival.
MATERIALS AND METHODS
Using the National Cancer Database, we performed a retrospective cohort study of women diagnosed between 2004 and 2012 with locally advanced squamous or adenocarcinoma of the cervix undergoing definitive primary radiation therapy. The National Cancer Database (referred to as database) is a joint program of the American Cancer Society and the Commission on Cancer of the American College of Surgeons. This database uses a hospital-based registry and the Commission on Cancer requires approved programs to abstract and follow all malignant tumors diagnosed, treated, or both at the hospital.10 Currently, 95.6% of the cervical cancer cases in the United States are reported to the database.11 Reported data reported are deidentified to ensure confidentiality; therefore, this study is exempt from obtaining informed consent by the study participants and approval by the University of Michigan institutional review board.
Cases were selected using the diagnostic code C53.9 and restricted to locally advanced squamous or adenocarcinoma of the cervix defined as International Federation of Gynecologists and Obstetricians stages IB2 to stage IVA. We only included patients undergoing primary radiation therapy as definitive. Staging reported in the database is clinical; therefore, to exclude patients with more advanced disease, the following group of patients was excluded: metastatic disease to the lung, bones, brain, or liver; those receiving multiagent chemotherapy greater than 4 weeks before radiation therapy, because this scenario would either represent metastatic disease or neoadjuvant therapy before definitive chemoradiation; those receiving radiation beyond the pelvis and paraaortic regions (eg, lung, abdominal targets); those treated with palliative intent; and cases in which details of therapies administered were not available (ie, date of chemotherapy initiation). Lastly, in an attempt to account for potential underreporting of chemotherapy or brachytherapy by some hospitals, we excluded hospitals that had a lower than 20% rate of administering chemotherapy or brachytherapy based on previous studies investigating adherence to guidelines.12
Race and ethnicity were used to construct single mutually exclusive race–ethnicity variables (non-Hispanic white, non-Hispanic black, Hispanic, or other or unknown). Tumor stage was determined using the revised 2009 staging criteria of the International Federation of Gynecologists and Obstetricians.13 Insurance status was categorized as uninsured or Medicaid, private, Medicare, or unknown. Age was categorized in deciles starting at age 40 years (younger than 40, 41–50, 51–60, 61–70, 71–80, and older than 80 years). Tumor histology was categorized as squamous or adenocarcinoma. Tumor grade was categorized as grade 1 or 2, grade 3 or undifferentiated, or unknown. Median household income from zip code of residence was derived from the 2012 American Community Survey data and was categorized into quartiles to be used as a proxy for socioeconomic status.
To assess the prevalence of comorbid disease in our cohort, we used the Charlson Comorbidity Index. Patients were assigned a score of 0, 1, or 2 or greater.14 The mean hospital volume was calculated by dividing the total number of cases reported by the hospital by the time period the hospital reported the cases to the database. The mean hospital volume was then ranked into quartiles—hospitals in the lowest quartile were labeled “low volume,” the second and third quartiles were labeled “intermediate volume,” and those in the fourth quartile were labeled “high volume.” Distance traveled to the hospital (as reported by the database) was categorized into four categories by dividing the entire range of values into quartiles. The resulting categories were first quartile (less than 4.5 miles), second quartile (4.6–10 miles), third quartile (10.1–85 miles), and fourth quartile (greater than 85.1 miles).
The National Comprehensive Cancer Network guidelines recommend external beam radiation therapy with concurrent platinum-based chemotherapy and vaginal brachytherapy for locally advanced cervical cancer. These guidelines have been consistent without any significant change since 2004.15 Patients receiving all three components of recommended therapy or those receiving external beam radiation and concurrent chemotherapy followed by completion hysterectomy were labeled as the “guideline-adherent” group. All other patients were categorized as the “guideline-discordant” group.
We compared the covariate distributions across outcome groups using χ2 tests and independent two-sample t tests. Linear changes in study variables over the timeframe of the study were assessed using a nonparametric trend test (Stata, nptrend).16 Factors associated with receipt of guideline-based care with prior knowledge based on previous research or clinical plausibility were included in multivariable analyses. A random-effects multivariable logistic regression model, with hospital as a random intercept, was used to assess factors associated with guideline-based care. Factors considered in the model included patient medical and sociodemographic factors (age, race, income, Charlson Comorbidity Index, and insurance status), tumor factors (stage, histology, and grade of disease), and treatment factors (hospital volume, distance from place of residence to the hospital, and year of diagnosis). Using the odds ratios derived from logistic regression analysis, we calculated the risk-adjusted probability of receiving guideline-based care. This was obtained by postestimation computation of predicted or expected values from the fitted logistic regression model using methodology previously described.17,18
Propensity score matching was performed to create a cohort in which participants who did and did not receive guideline-based care were balanced on covariates that might confound the effect of treatment approach on survival. We fit logistic regression models to estimate the probability of receiving guideline-based care. We then matched each woman who received guideline-based care with a woman who did not based on a similar propensity using a caliper of 0.001 to obtain a propensity-matched cohort in a 1:1 ratio.
After verifying the assumption of proportionality, a Cox proportional hazards model was fitted to evaluate the effects of patient, tumor, and treatment factors on overall survival. Survival analysis was performed for the entire cohort as well as for the propensity-matched cohort. Adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) were generated. Risk-adjusted survival was calculated based on the fitted models to estimate the independent effect of the covariates studied. Data management and analyses were performed using STATA 14. All tests were two-tailed, and α was set at 0.05.
The National Cancer Database registry data collectors abstract the information regarding chemotherapy administration regardless of whether all the treatment was given in the same hospital or in more than one facility. Nevertheless, to account for the possibility that some centers routinely refer their patients for treatment to other hospitals, we conducted sensitivity analysis by repeating all procedures in a subset of patients who received all of their treatment in one hospital only. In addition, we repeated our analysis by including previously excluded hospitals with a less than 20% rate of chemotherapy or brachytherapy.
The final cohort consisted of 16,195 patients who met inclusion criteria. A flowchart to outline the cohort development is provided in Appendix 1, available online at http://links.lww.com/AOG/A908. The demographic, clinicopathologic, and treatment characteristics of the cohort are listed in Table 1. The median age was 52 years (range 18–90 years). The majority of the patients in this cohort were non-Hispanic white (61%), were underinsured (Medicaid or uninsured, 40%), had squamous cell carcinoma (91%), and had stage III disease (43%). Overall, 57% of the patients (n=9,136) received guideline-based care and were labeled as “guideline-adherent” and 43% of patients (n=7,059) did not receive guideline-based care and were labeled as “guideline-discordant.” This cohort included patients in three different categories: 1) those receiving external beam radiation only, 10% (n=1,668); 2) those receiving external beam radiation and brachytherapy, completion hysterectomy, or both but no concurrent chemotherapy, 11% (n=1,750); and 3) those receiving external beam radiation and concurrent chemotherapy but no brachytherapy or completion hysterectomy, 22% (n=3,641).
Figure 1 shows the rate of adherence to guideline-based care by race–ethnicity. The rate of guideline-based care was 58.4% (95% CI 57.4–59.4%) for non-Hispanic white, 53% (95% CI 51.4–54.9%) for non-Hispanic black, and 51.5% (95% CI 49.4–53.7%) for Hispanic women (P<.001). Over the study period, the rate of guideline-based care improved from 49.5% (95% CI 47.1–51.9%) to 59.1% (95% CI 56.9–61.2%) (P trend<.001). Figure 2 shows the temporal trend in the rate of adherence to guideline-based care. Despite a decrease in the disparity between races over time, the rates of receipt of guideline-based care remained disparate in 2012. A larger proportion of non-Hispanic white patients received guideline-based care in 2012 (59%) compared with non-Hispanic black (54.6%) or Hispanic patients (53.7%) (P<.001).
Table 2 highlights factors independently associated with receipt of guideline-based care. The elderly, non-Hispanic black patients, Hispanic patients, those who were uninsured or Medicaid-insured, those with two or more comorbidities or a higher stage, and those receiving care at a low-volume hospital were less likely to receive guideline-based care. The risk-adjusted predicted probability of each patient receiving guideline-based care was calculated based on the logistic regression model and the mean rates stratified by hospital volume were plotted (Fig. 3). Compared with low-volume hospitals, the increase in adherence to guideline-based care in high-volume hospitals was 48–63% for non-Hispanic white, 47–53% for non-Hispanic black, and 41–54% for Hispanic women. As a result, there was a much larger gap in guideline-based care at high-volume centers. The results were similar when the estimates were dichotomized between the lower income (less than $38,000/year) group and those with higher income (greater than $38,000/year).
A total of 3,578 patients in the guideline-adherent cohort were matched 1:1 with the guideline-discordant cohort based on the propensity score matching. The characteristics of the propensity-matched cohort of patients stratified by adherence to guideline-based care are available in Appendix 2, available online at http://links.lww.com/AOG/A909. The propensity-matched groups were well balanced in patient, tumor, and treatment characteristics—all with comparison P values >.05.
Results of the Cox proportional hazard model are presented in Table 3. Increasing age, comorbidities, stage, and grade were associated with worse survival. Hispanic patients (adjusted HR 0.68, 95% CI 0.62–0.74); those with private insurance (adjusted HR 0.89, 95% CI 0.84–0.94), year of diagnosis 2010–2012 (adjusted HR 0.86, 95% CI 0.81–0.91), income $63,000 or greater (adjusted HR 0.86, 95% CI 0.79–0.82); and those receiving guideline-based care (adjusted HR 0.65, 95% CI 0.62–0.68) had lower odds of mortality. After adjustment for the receipt of guideline-based care, hospital volume was not independently associated with survival. These results were validated in the propensity-matched cohort. Based on the Cox proportional hazard regression model, risk-adjusted survival based on race (stratified by stage) and guideline-based care status was calculated and is presented as a figure in Appendix 3, available online at http://links.lww.com/AOG/A910.
Sensitivity analysis that included hospitals with a less than 20% rate of chemotherapy and brachytherapy administration as well as analysis of patients getting their entire treatment in one center showed similar results.
Adherence to guideline-based care has been shown by multiple investigators to significantly improve survival for multiple malignancies.6,7,19 Unsurprisingly, our results highlight that when appropriate guideline-based care was provided, survival improved regardless of race–ethnicity, stage, and hospital volume. The Centers for Disease Control and Prevention data show that minority women have the highest incidence as well as the highest death rate of cervical cancer.20–22 Although the reasons for this disparity are multifactorial, our data support the notion that not receiving guideline-based care might play a significant role.
Hospital volume has been shown to be associated with an improvement in survival in patients with locally advanced cervical cancer receiving radiation.23 However, when we adjusted for guideline-based care in our study, hospital volume was no longer a significant factor influencing survival. These results are consistent with a previous study using the National Cancer Database, which concluded that guideline-based care, rather than hospital volume, has a significant effect on survival.24 Although our analyses reveal that as the hospital volume increases, the rate of guideline-based care increases; the increase in guideline adherence was significantly greater for non-Hispanic white women compared with non-Hispanic black and Hispanic women. These findings—that racial–ethnic minorities are less likely to receive guideline-based care—are consistent with data from other oncologic and nononcologic diseases.4,25 Therefore, based on these data, it is reasonable to conclude that the most important reason for disparate care is not the lack of access to high-volume centers, but rather the lack of guideline-based care.26
Clearly identifying the etiologies of health care disparities remains an elusive task. Data from individual institutions, however, may provide some insight. In one such series of patients with cervical cancer, omission of brachytherapy and treatment interruptions was likely the result of acute toxicity in white women, whereas comorbidities and technical factors (poor geometry, inability to place tandem) and patient refusal were more common in black women.27 The higher frequency of technical factors leading to brachytherapy omission was attributed to the higher incidence of locally advanced disease, which can make applicator placement more challenging.28,29 Another factor preventing brachytherapy administration could be medical comorbidities. Although we included the Charlson Comorbidity Index as a covariate, it might not be specific enough to capture and distinguish between comorbidities that influence the receipt of chemotherapy, anesthesia required for Smit sleeve placement, or brachytherapy administration. Although it is certainly possible that technical factors and comorbidities contribute to reduced brachytherapy utilization, factors such as implicit biases of treatment teams could have contributed as well.30–32
Our study has several limitations. First, data on how treatment decisions were made, with respect to the patient's perception of how therapy would affect quality of life, are not available through the National Cancer Database. Clinical decisions often incorporate several factors, including patient preference, physician judgment, and resource utilization—none of which are reported in the database. Second, it is possible that some patients received chemotherapy, brachytherapy, or completion hysterectomy at a different hospital than the one administering radiation therapy. Although this concern is valid, we excluded facilities with a less than 20% yearly rate of administration to account for this possibility. In addition, we performed two sensitivity analyses: 1) restricting analysis to those receiving treatment in only one facility and 2) including facilities with a less than 20% rate. The results were similar.
The results of this large study demonstrate that even in the highest volume hospitals, substantial racial disparities persist in the treatment of locally advanced cervical cancer. This disparity in care is not attributable to income, access to care, disease stage, or histology; in fact, it is worse in the largest and most sophisticated medical facilities and has the very real capacity to reduce long-term survival in non-Hispanic black and Hispanic women with locally advanced cervical cancer. We suggest that further research should seek to evaluate the specific reasons for guideline discordance in this clinical setting in an effort to identify behaviors and biases on the part of health care providers as well as gaps in the education and support of patients that may be actionable and amenable to improvement.
1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2015. CA Cancer J Clin 2015;65:5–29.
2. Sheppard CS, El-Zein M, Ramanakumar AV, Ferenczy A, Franco EL. Assessment of mediators of racial disparities in cervical cancer survival in the United States. Int J Cancer 2016;138:2622–30.
3. Simard EP, Fedewa S, Ma J, Siegel R, Jemal A. Widening socioeconomic disparities in cervical cancer mortality among women in 26 states, 1993–2007. Cancer 2012;118:5110–6.
4. Committee on Understanding and Eliminating Racial and Ethnic Disparities in Health Care, Board on Health Sciences Policy, Institute of Medicine. Unequal treatment: confronting racial and ethnic disparities in health care (with CD). Washington, DC: National Academies Press; 2009.
5. DeMartino JK. Measuring quality in oncology: challenges and opportunities. J Natl Compr Canc Netw 2013;11:1482–91.
6. Bristow RE, Chang J, Ziogas A, Campos B, Chavez LR, Anton-Culver H. Sociodemographic disparities in advanced ovarian cancer survival and adherence to treatment guidelines. Obstet Gynecol 2015;125:833–42.
7. Boland GM, Chang GJ, Haynes AB, Chiang YJ, Chagpar R, Xing Y, et al. Association between adherence to National Comprehensive Cancer Network treatment guidelines and improved survival in patients with colon cancer. Cancer 2013;119:1593–601.
8. Sharma A, Schwartz SM, Méndez E. Hospital volume is associated with survival but not multimodality therapy in Medicare patients with advanced head and neck cancer. Cancer 2013;119:1845–52.
9. Bristow RE, Powell MA, Al-Hammadi N, Chen L, Miller JP, Roland PY, et al. Disparities in ovarian cancer care quality and survival according to race and socioeconomic status. J Natl Cancer Inst 2013;105:823–32.
11. Bilimoria KY, Stewart AK, Winchester DP, Ko CY. The National Cancer Data Base: a powerful initiative to improve cancer care in the United States. Ann Surg Oncol 2008;15:683–90.
12. Chagpar R, Xing Y, Chiang Y-J, Feig BW, Chang GJ, You YN, et al. Adherence to stage-specific treatment guidelines for patients with colon cancer. J Clin Oncol 2012;30:972–9.
13. Pecorelli S, Odicino F. Cervical cancer staging. Cancer J 2003;9:390–4.
14. Rauh-Hain JA, Pepin KJ, Meyer LA, Clemmer JT, Lu KH, Rice LW, et al. Management for elderly women with advanced-stage, high-grade endometrial cancer. Obstet Gynecol 2015;126:1198–206.
15. Teng N, Abu-Rustum NR, Bahador A, Bookman RA, Bristow RE, Campos S, et al. Cervical cancer guidelines. Clinical practice guidelines in oncology. J Natl Compr Canc Netw 2004;2:612–30.
16. Scher AI, Petterson B, Blair E, Ellenberg JH, Haan E, Reddihough DS, et al. The risk of mortality or cerebral palsy in twins: a collaborative population-based study. Pediatr Res 2002;52:671–81.
17. Onukwugha E, Bergtold J, Jain R. A primer on marginal effects—part II: health services research applications. Pharmacoeconomics 2014;33:97–103.
18. Williams R. Using the margins command to estimate and interpret adjusted predictions and marginal effects. Stata J 2012;12:308–31.
19. Sharma P, Ashouri K, Zargar-Shoshtari K, Luchey AM, Spiess PE. Racial and economic disparities in the treatment of penile squamous cell carcinoma: results from the National Cancer Database. Urol Oncol 2016;34:122.e9–15.
23. Lin JF, Berger JL, Krivak TC, Beriwal S, Chan JK, Sukumvanich P, et al. Impact of facility volume on therapy and survival for locally advanced cervical cancer. Gynecol Oncol 2014;132:416–22.
24. Wright JD, Huang Y, Ananth CV, Tergas AI, Duffy C, Deutsch I, et al. Influence of treatment center and hospital volume on survival for locally advanced cervical cancer. Gynecol Oncol 2015;139:506–12.
25. Virnig BA, Baxter NN, Habermann EB, Feldman RD, Bradley CJ. A matter of race: early- versus late-stage cancer diagnosis. Health Aff (Milwood) 2009;28:160–8.
27. 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.
28. 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.
29. Howell EA, Chen YT, Concato J. Differences in cervical cancer mortality among black and white women. Obstet Gynecol 1999;94:509–15.
30. Penner LA, Dovidio JF, Gonzalez R, Albrecht TL, Chapman R, Foster T, et al. The effects of oncologist implicit racial bias in racially discordant oncology interactions. J Clin Oncol 2016;34:2874–80.
31. Blair IV, Steiner JF, Fairclough DL, Hanratty R, Price DW, Hirsh HK, et al. Clinicians' implicit ethnic/racial bias and perceptions of care among black and Latino patients. Ann Fam Med 2013;11:43–52.
32. Haider AH, Sexton J, Sriram N, Cooper LA, Efron DT, Swoboda S, et al. Association of unconscious race and social class bias with vignette-based clinical assessments by medical students. JAMA 2011;306:942–51.
Supplemental Digital Content
© 2017 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.