Underuse of Radiation in Younger Women With Breast Cancer
Underuse of Radiation in Younger Women With Breast Cancer
Our study of women aged less than 65 years who underwent BCS estimated that the compliance rate of RT was 86%, which was higher than the rate reported among Medicare beneficiaries. Although the overall compliance rate among our study cohort was high, it is puzzling why nearly 15% of these BCS patients did not receive RT given that all patients in our sample had employer-sponsored health insurance. What is even more puzzling is that women in the youngest age group (aged 20–50 years) were statistically significantly less likely to receive RT than their older counterparts, although these younger women had the most to gain from the reduction in the risk of recurrence associated with RT. This study provides important insights to understanding these counterintuitive patterns.
Anecdotal stories from patients indicated that the need to arrange for childcare can be challenging for women undergoing RT because the entire course of treatment typically involves spending 30 to 60 minutes at a radiation facility 5 days a week for 3 to 7 weeks. Our study concluded that for women aged 20 to 50 years, those in families with at least one child aged less than 7 years were less likely to receive RT after BCS than patients in families with no children aged less than 7 years. Although this finding is not surprising because children aged less than 7 years are highly dependent on their parents, to our knowledge this is the first study in the medical literature to formally explore the association between family structure and receipt of guideline-concordant cancer therapy. Findings from our study suggest that child-care supplements provided by employers may have additional health benefits: it reduces the disincentive for employees or their spouses to seek care because of the concern of child-care need. To what extent this fringe benefit influences a young breast cancer patient's decision to undergo RT is a topic worth exploring in future research.
The receipt of RT after BCS represents one aspect of quality cancer care; therefore, a higher compliance rate is likely to be observed among patients whose other aspects of care are of high quality. We included several clinical variables to examine the association between RT and other breast cancer quality measures endorsed by the National Quality Forum. Indeed, positive associations were observed between RT and the other three quality measures (chemotherapy, staging imaging, and axillary surgery) in the full sample and for each age group. This finding suggests that devoting efforts to improving overall quality of breast cancer care could also improve RT compliance in addition to interventions targeting RT as an isolated event.
We found that patients whose BCS was performed by a provider not within the same Census division as their residence were less likely to receive RT. Using patients' travelling to a different Census division for BCS as a proxy for having to travel a longer distance to receive RT, our study concluded that distance may create an access barrier for the receipt of RT. This observation was consistent with findings in the literature. We also found wide variation in RT compliance across Census divisions, with the lowest rate observed in the Pacific region. Consistent with this finding, prior studies indicate that among patients with DCIS, RT compliance is lower in Los Angeles than in other SEER registries. Similarly, a national Medicare study of older women with invasive breast cancer found low RT compliance and substantial racial disparity among women residing in California. Potential causative factors underlying geographic variation remain unclear, although it should be noted that some of the leading academic physicians advocating for decreased use of RT in treatment of DCIS practice in the Los Angeles area, suggesting that this academic work reflects and/or influences regional practice patterns.
Institutional and patient-specific factors also matter. Patients enrolled in HMOs or capitated PPOs were less likely to receive RT than those in other plan types. Although this finding is alarming, our data do not contain sufficient information to allow further assessment of whether specific features in insurance benefit designs, such as less generous coverage or more limited choices of providers, have created access barrier for RT. With the exception of the youngest age group, women who were the primary holders of an insurance policy were more likely to receive RT. Policy implications of this finding were less clear because we were not able to determine to what extent the finding may reflect the completeness of claims for those who were primary holders vs those who were not. Lastly, patient comorbidities were not associated with RT among the younger age groups. This was probably because 94.7% and 90.3% of women in the groups aged 20 to 50 years and 51 to 55 years, respectively, had zero comorbidity.
Our study has several limitations. First, we were not able to determine whether there may be racial/ethnic disparities in the receipt of RT because race and ethnicity was not collected in the MarketScan database. The observed association between young children and RT would likely be attenuated if minorities tended to be diagnosed at younger age (and thus were more likely to have young children) and also were either more likely to be diagnosed at late stage or less likely to receive guideline-concordant care. Second, the MarketScan database is not linked to cancer registries; therefore, we did not have pathologic information, such as cancer stage, grade, and hormone receptor status that may affect the receipt of RT. Also, we had to apply a claims-based algorithm to identify the incident cases of breast cancer. Our algorithm was modified from a previously validated algorithm by using only breast cancer surgery and diagnosis as the initial step to ascertain our study cohort without including all other "qualifiers" (including RT) in the Nattinger et al. algorithm. However, because our study focused on RT compliance, false positives of incident cases should be less problematic once a BCS was identified. Despite our efforts to exclude potential prevalent cases, a proportion of our cohort could still be prevalent cases (ie, BCS performed to either rule out or treat an in-breast tumor recurrence), which can lead to underestimation of the rate of RT. Third, relying on the enrollment data as the sole criterion to identify children would miss information on children who were not covered under the health insurance plans in the MarketScan database, causing misclassification of the family structure variable. However, results from perturbation analyses suggested that such misclassification should not have a large impact on our estimates. Lastly, we relied on a proxy measure to determine whether patients may need to travel a longer distance for their RT because we did not have information on the location of all radiation facilities accessible to patients in our study cohort.
A wealth of literature has characterized population-based treatment patterns and outcomes among older women with breast cancer, yet, to date, little work has evaluated younger women. Our finding that a young child in the home is a barrier to completion of appropriate breast cancer therapy underscores the unique challenges confronted by younger (aged 20–50 years) cancer patients. Additional work is needed to understand the impact of family structure on other aspects of cancer care and to develop robust interventions tailored to the unique needs of younger cancer patients.
Discussion
Our study of women aged less than 65 years who underwent BCS estimated that the compliance rate of RT was 86%, which was higher than the rate reported among Medicare beneficiaries. Although the overall compliance rate among our study cohort was high, it is puzzling why nearly 15% of these BCS patients did not receive RT given that all patients in our sample had employer-sponsored health insurance. What is even more puzzling is that women in the youngest age group (aged 20–50 years) were statistically significantly less likely to receive RT than their older counterparts, although these younger women had the most to gain from the reduction in the risk of recurrence associated with RT. This study provides important insights to understanding these counterintuitive patterns.
Anecdotal stories from patients indicated that the need to arrange for childcare can be challenging for women undergoing RT because the entire course of treatment typically involves spending 30 to 60 minutes at a radiation facility 5 days a week for 3 to 7 weeks. Our study concluded that for women aged 20 to 50 years, those in families with at least one child aged less than 7 years were less likely to receive RT after BCS than patients in families with no children aged less than 7 years. Although this finding is not surprising because children aged less than 7 years are highly dependent on their parents, to our knowledge this is the first study in the medical literature to formally explore the association between family structure and receipt of guideline-concordant cancer therapy. Findings from our study suggest that child-care supplements provided by employers may have additional health benefits: it reduces the disincentive for employees or their spouses to seek care because of the concern of child-care need. To what extent this fringe benefit influences a young breast cancer patient's decision to undergo RT is a topic worth exploring in future research.
The receipt of RT after BCS represents one aspect of quality cancer care; therefore, a higher compliance rate is likely to be observed among patients whose other aspects of care are of high quality. We included several clinical variables to examine the association between RT and other breast cancer quality measures endorsed by the National Quality Forum. Indeed, positive associations were observed between RT and the other three quality measures (chemotherapy, staging imaging, and axillary surgery) in the full sample and for each age group. This finding suggests that devoting efforts to improving overall quality of breast cancer care could also improve RT compliance in addition to interventions targeting RT as an isolated event.
We found that patients whose BCS was performed by a provider not within the same Census division as their residence were less likely to receive RT. Using patients' travelling to a different Census division for BCS as a proxy for having to travel a longer distance to receive RT, our study concluded that distance may create an access barrier for the receipt of RT. This observation was consistent with findings in the literature. We also found wide variation in RT compliance across Census divisions, with the lowest rate observed in the Pacific region. Consistent with this finding, prior studies indicate that among patients with DCIS, RT compliance is lower in Los Angeles than in other SEER registries. Similarly, a national Medicare study of older women with invasive breast cancer found low RT compliance and substantial racial disparity among women residing in California. Potential causative factors underlying geographic variation remain unclear, although it should be noted that some of the leading academic physicians advocating for decreased use of RT in treatment of DCIS practice in the Los Angeles area, suggesting that this academic work reflects and/or influences regional practice patterns.
Institutional and patient-specific factors also matter. Patients enrolled in HMOs or capitated PPOs were less likely to receive RT than those in other plan types. Although this finding is alarming, our data do not contain sufficient information to allow further assessment of whether specific features in insurance benefit designs, such as less generous coverage or more limited choices of providers, have created access barrier for RT. With the exception of the youngest age group, women who were the primary holders of an insurance policy were more likely to receive RT. Policy implications of this finding were less clear because we were not able to determine to what extent the finding may reflect the completeness of claims for those who were primary holders vs those who were not. Lastly, patient comorbidities were not associated with RT among the younger age groups. This was probably because 94.7% and 90.3% of women in the groups aged 20 to 50 years and 51 to 55 years, respectively, had zero comorbidity.
Our study has several limitations. First, we were not able to determine whether there may be racial/ethnic disparities in the receipt of RT because race and ethnicity was not collected in the MarketScan database. The observed association between young children and RT would likely be attenuated if minorities tended to be diagnosed at younger age (and thus were more likely to have young children) and also were either more likely to be diagnosed at late stage or less likely to receive guideline-concordant care. Second, the MarketScan database is not linked to cancer registries; therefore, we did not have pathologic information, such as cancer stage, grade, and hormone receptor status that may affect the receipt of RT. Also, we had to apply a claims-based algorithm to identify the incident cases of breast cancer. Our algorithm was modified from a previously validated algorithm by using only breast cancer surgery and diagnosis as the initial step to ascertain our study cohort without including all other "qualifiers" (including RT) in the Nattinger et al. algorithm. However, because our study focused on RT compliance, false positives of incident cases should be less problematic once a BCS was identified. Despite our efforts to exclude potential prevalent cases, a proportion of our cohort could still be prevalent cases (ie, BCS performed to either rule out or treat an in-breast tumor recurrence), which can lead to underestimation of the rate of RT. Third, relying on the enrollment data as the sole criterion to identify children would miss information on children who were not covered under the health insurance plans in the MarketScan database, causing misclassification of the family structure variable. However, results from perturbation analyses suggested that such misclassification should not have a large impact on our estimates. Lastly, we relied on a proxy measure to determine whether patients may need to travel a longer distance for their RT because we did not have information on the location of all radiation facilities accessible to patients in our study cohort.
A wealth of literature has characterized population-based treatment patterns and outcomes among older women with breast cancer, yet, to date, little work has evaluated younger women. Our finding that a young child in the home is a barrier to completion of appropriate breast cancer therapy underscores the unique challenges confronted by younger (aged 20–50 years) cancer patients. Additional work is needed to understand the impact of family structure on other aspects of cancer care and to develop robust interventions tailored to the unique needs of younger cancer patients.
