Factors Contributing to Underuse of Radiation Among Younger Women With Breast Cancer

Abstract

Background

Evidence-based literature has confirmed the effectiveness of radiation therapy (RT) after breast-conserving surgery (BCS), especially for young women. However, women with young children may be less likely to be compliant. This study explores factors associated with noncompliance of RT among insured young patients.

Method

Using the MarketScan Database, we identified the study cohort as women aged 20 to 64 years who had a BCS between January 1, 2004, and December 31, 2009, and had continuous enrollment 12 months before and after the date of BCS. Patients who had any radiation claims within a year of BCS were considered compliant. Adjusted odds of compliance were estimated from logistic regressions for the full sample and age-stratified subgroups. Sensitivity analyses were performed to evaluate the robustness of study findings. All statistical tests were two-sided.

Results

Eighteen thousand one hundred twenty of 21 008 (86.25%) nonmetastatic BCS patients received RT. Among patients aged 20 to 64 years, those with children aged 7 to 12 years, those with children aged 13 to 17 years, and those with no children or children aged 18 years or older were more likely to receive RT than patients with at least one child aged less than 7 years (7–12 years: odds ratio (OR) = 1.32, 95% confidence interval (CI) = 1.05 to 1.66, P = .02; 13–17 years: OR = 1.41, 95% CI = 1.13 to 1.75, P = .002; no children or ≥18 years: OR = 1.38, 95% CI = 1.13 to 1.68, P = .001). Stratified analyses showed that the above association was primarily driven by women in the youngest age group (aged 20–50). Other important factors included breast cancer quality of care measures, enrollment in health maintenance organizations or capitated preferred provider organizations, travelled to a Census division outside their residence for BCS, and whether patients were primary holders of the insurance policy.

Conclusions

Competing demands from child care can constitute a barrier to complete guideline-concordant breast cancer therapy. Younger patients may be confronted by unique challenges that warrant more attention in future research.

Clinical benefits of radiation therapy (RT) after breast-conserving surgery (BCS) for women with early-stage breast cancer are well documented in the literature (1,2). Many professional associations now consider RT after BCS to be an important quality of care indicator for breast cancer (3). Several population-based studies have investigated the compliance rates of RT and the associated factors. The majority of these studies focused on an elderly population and used either Medicare claims or the Surveillance, Epidemiology and End Results (SEER)–Medicare data (4–14). However, recent estimates reported that approximately 60% of breast cancer patients diagnosed between 2005 and 2009 were aged less than 65 years (15), yet the pattern of RT after BCS among young women remained underexplored.

Understanding the pattern of RT after BCS among young patients and factors associated with noncompliance has important clinical and economic implications. Today it is well accepted that breast cancers diagnosed at younger ages differ from those diagnosed at older ages; younger women tend to have more aggressive tumors that are associated with higher risks of local and distant recurrence (16). A meta-analysis reported that RT after BCS increased survival and reduced local recurrence, especially among young patients (1,2). Thus, young breast cancer patients who underwent BCS are expected to benefit more from RT. From an economic perspective, many women aged less than 65 years are active participants in the workforce; therefore, their absence from the labor market because of morbidity associated with cancer symptoms and treatment or premature death from cancer will lead to substantial productivity loss. Indeed, the estimated costs of productivity loss because of breast cancer were $10.9 billion in 2010 (17).

The few studies that included young breast cancer patients used cancer registry data, such as SEER or the National Cancer Database (18–21), to determine the pattern of RT after BCS. Although the rate of RT may be underestimated in these studies because validation studies have raised the concern of underascertainment for RT identified from registry data (22,23), an interesting pattern was observed in the two studies that focused exclusively on young breast cancer patients (19,21). Both reported a lower rate of RT after BCS among patients at a younger age. One study found that the rate of RT was 81% for women aged 50 to 55 years and was 75% for those aged 35 years or younger (19); the other reported the rate to be 69%, 73%, and 80% for women in the age ranges of 35 years or younger, 36 to 40 years, and 61 to 64, respectively (21).

The purpose of our study was twofold. First, we used claims and enrollment data from employer-sponsored insurance plans to explore factors associated with the use of RT after BCS among young patients with breast cancer. Second, we tested a hypothesis that the lower rate of RT observed among younger women may be associated with competing demand of their time in providing child care. Specifically, we investigated whether younger women with young children were less likely to undergo RT after BCS compared to women at similar ages without young children.

Methods

Data Sources

We used a large, nationwide, employment-based MarketScan database (24) that contains information on medical and outpatient prescription drug claims for employees and their spouses and dependents; the data represent claims from approximately 45 large employers and capture claims records from more than 100 payers. The University of Chicago’s Institutional Review Board exempted this study for approval because all observations were deidentified.

Ascertainment of Study Cohort

We applied a previously published algorithm for claims data to identify incident cases of breast cancer (14). We first identified our study cohort as women aged 20 to 64 years who had at least one International Classification of Diseases, 9th Edition (ICD-9) diagnosis code indicating invasive breast cancer (ICD 9 CM 174.XX) and a procedure code of breast cancer surgery between January 1, 2004, and December 31, 2009. BCS was identified by common procedural terminology 4 and ICD-9 procedure codes (6). The earliest claims date indicating BCS was chosen as the index date. Patients who had mastectomy or both BCS and mastectomy claims concurrently were excluded. We then limited the study cohort to those who had continuous enrollment in health insurance 12 months before and after the index date to ensure that information obtained from claims in this 24-month window was complete. We excluded patients who had a breast cancer diagnosis more than 12 months before the index date, a prior history of breast cancer, RT before BCS, and mastectomy within 12 months of BCS. Lastly, we excluded patients with missing information on geographic region of residence or comorbidities and those with claims indicating distant metastasis. The final sample was 21 008 young women (Figure 1).

Figure 1.

Flow chart for data selection step. BCS = breast conserving surgery; ICD-9 = International Classification of Diseases, 9th Edition.

Identification of Radiation Therapy and Other Clinical Variables

We identified RT using ICD-9 procedure, common procedural terminology, and revenue center codes based on published algorithms (14,25–27). We defined patients who were compliant as those who had any RT claims within 12 months of BCS. Other treatment-related variables included chemotherapy, use of staging imaging, axillary surgery, pre-BCS mammography screening, and inpatient vs outpatient BCS. Codes used to identify these variables are provided in Supplementary Table 1 (available online).

Other clinical variables included axillary lymph node involvement and whether the patient had a ductal-carcinoma-in-situ (DCIS) diagnosis within 12 months of the index date (25–28). Comorbidity was constructed using Klabunde’s modified algorithm of Charlson comorbidity score (29).

Patients and Area Characteristics

We obtained patient age, geographic regions of residence, and whether the patient was the primary holder of insurance directly from the MarketScan database. We derived several patient characteristics variables using information provided in the data. First, we took advantage of the age information for the patients’ dependents in the enrollment file to infer patients’ family structure based on whether the family had a child of young age. Specifically, we classified families into four categories: families with at least one child aged less than 7 years, with no children aged less than 7 years but at least one child aged 7 to 12 years, with no children aged less than 13 years but at least one child aged 13 to 17 years, and those with no children or only children aged greater than 17 years. Our choice of cutpoints corresponds to the typical age ranges in nursery, elementary, middle/high school, and postsecondary education in the US education system (30). These categories reflect the child-care need, with the greatest need in families with at least one child aged less than 7 years. Second, we regrouped the insurance plan type variable into two categories based on whether patients were more likely to be tightly managed under the payment policy of their insurance: Health maintenance organization (HMO) or preferred provider organization (PPO) with capitation vs all other plan types. Third, we constructed a proxy variable to indicate whether patients may need to travel a long distance to receive RT by evaluating whether the geographic area (in terms of Census division) of the patient differed from that of her BCS provider.

We obtained variables of area characteristics by linking patients’ county of residence to the 2004 to 2009 Area Resource File (31) and measured the socioeconomic status (in quartiles) as the percentage of families with median household income below poverty level and percentage of adults with college degree. For health-care resources related to radiation therapy, we reorganized counties into health services area based on the modified health service area definition by the National Cancer Institute (32). Using health service area as the geographic unit that formed the patient’s local market for RT, we included two supply-side variables in quartiles: density of hospitals with radiation therapy and density of radiation oncologists.

Statistical Analysis

We used SAS version 9.2 (SAS Institute Inc. Cary, NC) for data management and STATA version 11.0 (StataCorp LP, College Station, TX) for statistical analyses. We used Pearson χ² test to compare each covariable for women with RT within 12 months of BCS and those without and ran logistic regressions to examine factors associated with the receipt of RT. We applied three methods for model diagnosis, including ‘linktest’ in STATA, the Hosmer–Lemeshow goodness-of-fit test, and the Bayesian information criterion. The model specification that passed all three tests was chosen as the final model. In addition, we used perturbation analyses to assess potential impact of collinearity and misclassification (33,34). Next, we conducted stratified analyses by patients’ age (aged 20–50, 51–55, 56–60, and 61–64 years) to determine whether certain covariables (eg, age of children) were only relevant to a specific age group. We also performed sensitivity analyses to evaluate the robustness of study findings across various subgroups. All statistical tests were two-sided.

Results

Eighteen thousand one hundred twenty (86.25%) of the 21 008 patients received RT within a year of BCS. Table 1 shows that the majority of the patients (89.86%) had no comorbidity. More than half of the patients (59.51%) were the primary holders of the insurance policy. Among the study cohort, 4.25% had at least one child aged less than 7 years, and the rate of RT was 80.61% for this group. Geographic variations in RT compliance were also observed, ranging from less than 80% in the Pacific Census division to close to 90% in the West-North-Central Census division.

Table 1.

Analysis of univariate χ² test and multivariable logistic regression (full study sample)*

		Univariate analysis			Logistic regression
Variables	No. of case patients	%	Percentage who received RT	P	Odds ratio (95%CI)	P
Total No. of case patients	21008		86.25
Age, y				<.001		<.001†
20 – 50	7270	34.61	84.26		1.00 (referent)
51 – 55	4756	22.64	86.75		1.22 (1.08 to 1.37)	.001
56 – 60	5604	26.68	87.90		1.38 (1.22 to 1.55)	<.001
61 – 64	3378	16.08	87.09		1.28 (1.11 to 1.47)	<.001
Comorbidities				.10		.02†
0	18878	89.86	86.23		1.00 (referent)
1	1813	8.63	87.09		0.95 (0.82 to 1.10)	.51
≥2	317	1.51	82.65		0.64 (0.47 to 0.87)	.005
Employee status				<.001
Primary holder of insurance policy	12501	59.51	87.11		1.00 (referent)
others	8507	40.49	85.00		1.20 (1.10 to 1.31)	<.001
Family status				<.001		.01†
With at least one child < age 7 years†	892	4.25	80.61		1.00 (referent)
With all children > age 6 and at least one child between ages 7 and 12 years	1584	7.54	84.34		1.32 (1.05 to 1.66)	.02
With all children > age 12 and at least one child between ages 13 and 17 years	2016	9.60	85.07		1.41 (1.13 to 1.75)	.002
With no children or all children > age 18 years	16516	78.62	86.89		1.38 (1.13 to 1.68)	.001
HMO or PPO with capitation				<.001
No	17447	83.19	87.29		1.00 (referent)
Yes	3531	16.81	81.14		0.70 (0.63 to 0.77)	<.001
Census divisions				<.001		<.001†
Pacific (AK, CA, HI, OR, WA)	2895	13.78	79.52		1.00 (referent)
New England (CT, MA, NH, ME, RI, VT)	691	3.29	88.57		1.89 (1.45 to 2.48)	<.001
Middle Atlantic (NJ, NY, PA)	1168	5.56	86.99		1.75 (1.42 to 2.14)	<.001
East North Central (IL, IN, MI, OH, WI)	4647	22.12	87.76		1.65 (1.41 to 1.93)	<.001
West North Central (IA, KS, MN, MO, NE, ND, SD)	1122	5.34	89.57		1.89 (1.51 to 2.38)	<.001
South Atlantic (DE, DC, FL, GA, MD, NC, SC, VA)	4791	22.81	87.27		1.60 (1.39 to 1.83)	<.001
East South Central (AL, KY, MS, TN)	1726	8.22	84.99		1.30 (1.09 to 1.56)	.003
West South Central (AR, LA, OK, TX)	2890	13.76	88.03		1.49 (1.27 to 1.74)	<.001
Mountain (AZ, CO,ID, MT, NV, NM, UT, WY)	1078	5.13	84.88		1.26 (1.03 to 1.54)	.02
Metropolitan area				.52
No	3514	16.73	85.91		1.00 (referent)
Yes	17494	83.27	86.32		1.13 (0.99 to 1.29)	.06
Axillary LN Dissection				<.001
No	5633	26.81	76.97		1.00 (referent)
Yes	15375	73.19	89.65		2.30 (2.11 to 2.51)	<.001
Axillary LN involvement				<.001
No	17813	84.79	85.41		1.00 (referent)
Yes	3195	15.21	90.95		1.14 (0.99 to 1.31)	.08
DCIS diagnosis within 1 year of surgery				<.001
No	13056	62.15	88.36		1.00 (referent)
Yes	7952	37.85	82.80		0.80 (0.73 to 0.87)	<.001
Surgery type
Inpatient BCS	743	3.54	89.64	.006	1.00 (referent)
Outpatient BCS	20265	96.46	86.13		0.73 (0.57 to 0.94)	.02
Chemotherapy				<.001
No	11810	56.22	82.35		1.00 (referent)
Yes	9198	43.78	91.26		1.83 (1.65 to 2.03)	<.001
Imaging				<.001
No	10705	50.96	83.36		1.00 (referent)
Yes	10303	49.04	89.26		1.33 (1.22 to 1.45)	<.001
Mammogram screening				<.001
No	3463	16.48	81.72		1.00 (referent)
Yes	17545	83.52	87.15		1.57 (1.42 to 1.74)	<.001
Density of hospitals with radiation therapy within a health services area				<.001		.76†
Q4	5295	25.20	86.82		1.00 (referent)
Q3	5146	24.50	87.68		1.03 (0.91 to 1.17)	.63
Q2	5453	25.96	85.00		0.97 (0.86 to 1.10)	.64
Q1	5114	24.34	85.57		1.02 (0.90 to 1.17)	.71
Density of radiation oncologist within a health services area				.001		.54†
Q4	5581	26.57	87.67		1.00 (referent)
Q3	4926	23.45	85.28		0.92 (0.81 to 1.04)	.16
Q2	5264	25.06	85.43		0.97 (0.85 to 1.11)	.67
Q1	5237	24.93	86.48		0.96 (0.84 to 1.10)	.53
Percentage of Families income under poverty level				.005		.012†
Q4	5770	27.47	86.26		1.00 (referent)
Q3	4950	23.56	84.83		0.88 (0.78 to 0.98)	.03
Q2	5243	24.96	87.09		1.05 (0.93 to 1.19)	.42
Q1	5045	24.01	86.78		0.92 (0.80 to 1.06)	.26
Percentage of neighborhood with college education				.50		.047†
Q4	5366	25.54	86.81		1.00 (referent)
Q3	5251	25.00	86.31		0.93 (0.82 to 1.06)	.29
Q2	5257	25.02	86.06		0.84 (0.74 to 0.95)	.008
Q1	5134	24.44	85.82		0.84 (0.72 to 0.98)	.03
Surgery year				.49		.07†
2004	2588	12.32	85.16		1.00 (referent)
2005	3000	14.28	86.27		1.01 (0.86 to 1.18)	.93
2006	3020	14.38	86.69		1.07 (0.91 to 1.26)	.39
2007	4271	20.33	86.02		1.13 (0.98 to 1.32)	.095
2008	4124	19.63	86.28		1.15 (1.00 to 1.34)	.06
2009	4005	19.06	86.84		1.21 (1.04 to 1.41)	.01
Travel across Census division to receive BCS				.001
No	20045	95.42	86.43		1.00 (referent)
Yes	963	4.58	82.55		0.72 (0.60 to 0.86)	<.001

* All statistical tests were two-sided. BCS = breast conserving surgery; CI = confidence interval; DCIS = ductal carcinoma in situ. HMO = health maintenance organization; LN= lymph node; PPO = preferred provider organization; RT = radiation therapy.

† P value from likelihood ratio test.

Results from the logistic regression (Table 1) showed that compared with patients aged 50 years of age or younger, the odds ratio (OR) of receiving radiation therapy was 1.22 (95% confidence interval (CI) = 1.08 to 1.37), 1.38 (95% CI = 1.22 to 1.55), and 1.28 (95% CI = 1.11 to 1.47) for those in the age ranges 51 to 55 years, 56 to 60 years, and 61 to 64 years, respectively. Having at least one child aged less than 7 years resulted in statistically significantly lower odds of radiation therapy than having no children or older children. Among young patients (aged 20–64 years), compared with those with at least one young child (aged <7 years), patients with children aged 7 to 12 years, patients with children aged 13 to 17 years, and patient with no children or children aged 18 years or older were more likely to receive RT (children 7–12 years: OR = 1.32, 95% CI = 1.05 to 1.66, P = .02; children 13–17 years: OR = 1.41, 95% CI = 1.13 to 1.75, P = .002; no children or children ≥18 years: OR = 1.38, 95% CI = 1.13 to 1.68, P = .001). Perturbation analyses (Supplementary Table 2, available online) showed that although women’s age group and the family structure variable were correlated, the correlation was not high enough to cause incorrect inferences. Statistically significantly lower odds of RT were observed among patients who were enrolled in a HMO or a capitation PPO (OR = 0.70; 95% CI = 0.63 to 0.77), patients who travelled across Census division to receive BCS (OR = 0.72; 95% CI = 0.60 to 0.86), or patients who received BCS in an outpatient setting (OR = 0.73; 95% CI = 0.57 to 0.94). In addition, patients who were the primary holder of insurance policy were more likely to have RT (OR = 1.20; 95% CI = 1.10 to 1.31). Those living in areas with lower percentages of the population with college education were statistically significantly less likely to receive RT. In addition, DCIS was found to be statistically significantly negatively associated with the receipt of RT. Patients with chemotherapy, imaging, and mammogram screening were more likely to receive RT after BCS.

Analyses stratified by age group (Table 2) showed that the association between young children and lower likelihood of RT was statistically significant only in the youngest age group of women (aged 20–50 years). This suggested that the association observed in the full sample was primarily driven by women in the youngest age group. Other associations observed above largely persisted across all age groups. In addition, sensitivity analyses (Figure 2) for women in the youngest age group showed that our hypothesized relationship was supported regardless of patients’ DCIS status or whether they were primary holders of an insurance policy.

Table 2.

Comparison of multivariable logistic regression models among four age groups*

Variables	Aged 20-50 y	Aged 51-55 y	Aged 56-60 y	Aged 61-64 y
	Odds ratio (95%CI)	Odds ratio (95%CI)	Odds ratio (95%CI)	Odds ratio (95%CI)
Comorbidities	.87†	.64†	.04†	.06†
0	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
1	0.97 (0.71 to 1.32)	1.06 (0.76 to 1.48)	0.84 (0.65 to 1.09)	1.01 (0.73 to 1.41)
≥2	1.24 (0.47 to 3.24)	0.71 (0.32 to 1.54)	0.55 (0.35 to 0.88)	0.48 (0.28 to 0.82)
Employee status
Primary holder of insurance policy	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
Others	1.10 (0.96 to 1.27)	1.34 (1.12 to 1.60)	1.22 (1.03 to 1.45)	1.29 (1.04 to 1.61)
Family status	.003†	.10†	.61†	.26†
With at least one child < age 7 years	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
With all children > age 6 years and at least one child between ages 7 and 12 years	1.40 (1.10 to 1.78)	1.08 (0.45 to 2.55)	2.00 (0.44 to 9.14)	0.84 (0.72 to 9.79)
With all children > age 12 years and at least one child between ages 13 and 17 years	1.57 (1.23 to 2.01)	1.11 (0.51 to 2.40)	2.54 (0.68 to 9.50)	5.44 (0.47 to 63.06)
With no children or all children > age 18 years	1.40 (1.14 to 1.74)	1.47 (0.70 to 3.09)	1.99 (0.60 to 6.58)	2.30 (0.30 to 17.83)
HMO or PPO with capitation
No	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
Yes	0.72 (0.61 to 0.85)	0.73 (0.59 to 0.91)	0.67 (0.54 to 0.83)	0.56 (0.42 to 0.74)
Census Divisions	<.001†	<.001†	.04†	.005†
Pacific (AK, CA, HI, OR, WA)	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
New England (CT, MA, NH, ME, RI, VT)	1.55 (1.02 to 2.35)	2.44 (1.37 to 4.34)	2.01 (1.11 to 3.61)	2.23 (1.07 to 4.64)
Middle Atlantic (NJ, NY, PA)	1.49 (1.09 to 2.04)	2.44 (1.53 to 3.89)	1.49 (0.95 to 2.33)	2.16 (1.26 to 3.71)
East North Central (IL, IN, MI, OH, WI)	1.42 (1.09 to 1.85)	1.93 (1.39 to 2.68)	1.68 (1.21 to 2.33)	1.83 (1.25 to 2.68)
West North Central (IA, KS, MN, MO, NE, ND, SD)	1.62 (1.14 to 2.32)	2.67 (1.59 to 4.47)	2.04 (1.25 to 3.34)	1.62 (0.92 to 2.86)
South Atlantic (DE, DC, FL, GA, MD, NC, SC, VA)	1.64 (1.30 to 2.06)	1.93 (1.44 to 2.58)	1.39 (1.05 to 1.84)	1.48 (1.06 to 2.06)
East South Central (AL, KY, MS, TN)	1.01 (0.76 to 1.35)	1.39 (0.96 to 2.02)	1.39 (0.96 to 2.01)	2.06 (1.25-3.39)
West South Central (AR, LA, OK, TX)	1.09 (0.84 to 1.42)	1.69 (1.19 to 2.38)	1.66 (1.19 to 2.31)	2.44 (1.58 to 3.77)
Mountain (AZ, CO,ID, MT, NV, NM, UT, WY)	1.09 (0.77 to 1.53)	1.33 (0.87 to 2.03)	1.37 (0.90 to 2.10)	1.40 (0.87 to 2.25)
Metropolitan area
No	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
Yes	1.28 (1.03 to 1.59)	1.20 (0.90 to 1.58)	1.05 (0.81 to 1.37)	0.93 (0.66 to 1.30)
Axillary LN dissection
No	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
Yes	2.17 (1.88 to 2.50)	2.77 (2.30 to 3.34)	2.12 (1.76 to 2.55)	2.34 (1.87 to 2.94)
Axillary LN involvement
No	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
Yes	1.26 (1.01 to 1.59)	1.28 (0.94 to 1.75)	1.02 (0.77 to 1.36)	0.92 (0.63 to 1.36)
DCIS diagnosis within 1 year of surgery
No	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
Yes	0.80 (0.69 to 0.92)	0.85 (0.70 to 1.03)	0.73 (0.61 to 0.87)	0.82 (0.65 to 1.03)
Surgery type
Inpatient BCS	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
Outpatient BCS	0.81 (0.53 to 1.24)	0.58 (0.33 to 1.03)	0.77 (0.47 to 1.27)	0.61 (0.33 to 1.11)
Chemotherapy
No	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
Yes	2.18 (1.86 to 2.55)	1.62 (1.31 to 2.02)	1.63 (1.31-2.02)	1.62 (1.22 to 2.15)
Imaging
No	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
Yes	1.26 (1.09 to 1.45)	1.33 (1.11 to 1.61)	1.43 (1.20 to 1.71)	1.37 (1.09 to 1.71)
Mammogram Screening
No	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
Yes	1.41 (1.20 to 1.65)	1.80 (1.44 to 2.25)	1.71 (1.37 to 2.14)	1.56 (1.18 to 2.05)
Density of hospitals with radiation therapy within a health services area	.01†	.18†	.44†	.56†
Q4	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
Q3	1.18 (0.96 to 1.46)	0.86 (0.66 to 1.12)	1.08 (0.83 to 1.40)	0.87 (0.64 to 1.19)
Q2	0.85 (0.70 to 1.05)	1.06 (0.81 to 1.39)	1.04 (0.80 to 1.34)	1.10 (0.79 to 1.52)
Q1	1.09 (0.88 to 1.35)	1.18 (0.89 to 1.57)	0.87 (0.67 to 1.13)	0.96 (0.69 to 1.33)
Density of radiation oncologist within a health services area	.82†	.12†	.24†	.08†
Q4	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
Q3	0.98 (0.80 to 1.20)	0.93 (0.72 to 1.22)	0.95 (0.73 to 1.22)	0.68 (0.50 to 0.94)
Q2	0.95 (0.77 to 1.18)	0.93 (0.71 to 1.23)	1.21 (0.92 to 1.58)	0.77 (0.55 to 1.07)
Q1	0.90 (0.72 to 1.13)	1.26 (0.94 to 1.70)	1.03 (0.79 to 1.34)	0.69 (0.49 to 0.97)
Percentage of families income under poverty level	.01†	.56†	.33†	.81†
Q4	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
Q3	0.75 (0.62 to 0.90)	0.88 (0.68 to 1.13)	1.04 (0.82 to 1.32)	0.96 (0.71 to 1.29)
Q2	0.96 (0.79 to 1.17)	1.02 (0.78 to 1.33)	1.22 (0.95 to 1.57)	1.09 (0.80 to 1.49)
Q1	0.84 (0.67 to 1.05)	0.88 (0.64 to 1.19)	1.00 (0.76 to 1.32)	1.11 (0.78 to 1.58)
Percentage of neighborhood with college education	.09†	.32†	.21†	.28†
Q4	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
Q3	0.93 (0.76 to 1.13)	1.14 (0.86 to 1.49)	0.77 (0.60 to 0.99)	1.04 (0.75 to 1.45)
Q2	0.78 (0.63 to 0.97)	0.90 (0.68 to 1.19)	0.91 (0.70 to 1.19)	0.80 (0.57 to 1.10)
Q1	0.77 (0.60 to 0.99)	0.91 (0.65 to 1.27)	0.83 (0.61 to 1.12)	0.98 (0.66 to 1.46)
Surgery year	.31†	.12†	.24†	.41†
2004	1.00 (referent)	1.00	1.00 (referent)	1.00 (referent)
2005	1.02 (0.79 to 1.31)	1.26 (0.89 to 1.77)	0.75 (0.54 to 1.04)	1.17 (0.78 to 1.76)
2006	1.05 (0.81 to 1.34)	1.27 (0.90 to 1.78)	0.90 (0.64 to 1.27)	1.25 (0.83 to 1.88)
2007	1.22 (0.96 to 1.55)	1.34 (0.98 to 1.84)	0.94 (0.68 to 1.28)	1.07 (0.74 to 1.55)
2008	1.18 (0.93 to 1.51)	1.19 (0.87 to 1.63)	0.98 (0.71 to 1.34)	1.45 (0.99 to 2.11)
2009	1.23 (0.97 to 1.57)	1.59 (1.15 to 2.19)	1.08 (0.7 to 1.49)	1.11 (0.77 to 1.59)
Travel across Census division to receive BCS
No	1.00 (referent)	1.00 (referent)	1.00 (referent)	1.00 (referent)
Yes	0.63 (0.47 to 0.83)	0.63 (0.44 to 0.92)	0.85 (0.58 to 1.25)	1.04 (0.60 to 1.79)

* All statistical tests were two-sided. BCS = breast conserving surgery; CI = confidence interval; DCIS = ductal carcinoma in situ. HMO = health maintenance organization; LN = lymph node; PPO = preferred provider organization; RT = radiation therapy;

† P value from likelihood ratio test.

Figure 2.

Adjusted odds ratios in sensitivity analyses for family structure variables, by subgroups. Error bars represent 95% confidence intervals. All statistical tests were two-sided. DCIS = ductal carcinoma in situ.

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 (14). 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 (35). 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 (36–38). 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 (39,40). 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 (14). 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 (41), 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 (14) was modified from a previously validated algorithm (25) 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 (25). 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.

Funding

This work was support by the Agency for Healthcare Research and Quality (R01 HS018535 to YCTS); the University of Chicago Cancer Research Foundation Women’s Board; and the Cancer Prevention & Research Institute of Texas (RP101207 to BDS).

The authors would like to thank Nicole Babuskow for editing the paper. The work was completed while I.W. Pan was an employee at the University of Chicago.