Currently Viewing:
The American Journal of Managed Care June 2014
Comparison Between Guideline-Preferred and Nonpreferred First-Line HIV Antiretroviral Therapy
Stephen S. Johnston, MA; Timothy Juday, PhD; Amanda M. Farr, MPH; Bong-Chul Chu, PhD; and Tony Hebden, PhD
The Value of Specialty Pharmaceuticals - A Systematic Review
Martin Zalesak, MD, PhD; Joyce S. Greenbaum, BA; Joshua T. Cohen, PhD; Fotios Kokkotos, PhD; Adam Lustig, MS; Peter J. Neumann, ScD; Daryl Pritchard, PhD; Jeffrey Stewart, BA; and Robert W. Dubois, MD
Currently Reading
Health Insurance and Breast-Conserving Surgery With Radiation Treatment
Askal Ayalew Ali, MA; Hong Xiao, PhD; and Gebre-Egziabher Kiros, PhD
Impact of a Patient Incentive Program on Receipt of Preventive Care
Ateev Mehrotra, MD; Ruopeng An, PhD; Deepak N. Patel, MBBS; and Roland Sturm, PhD
Novel Predictive Models for Metabolic Syndrome Risk: A "Big Data" Analytic Approach
Gregory B. Steinberg, MB, BCh; Bruce W. Church, PhD; Carol J. McCall, FSA, MAAA; Adam B. Scott, MBA; and Brian P. Kalis, MBA
Primary Care Diabetes Bundle Management: 3-Year Outcomes for Microvascular and Macrovascular Events
Frederick J. Bloom Jr, MD; Xiaowei Yan, PhD; Walter F. Stewart, PhD; Thomas R. Graf, MD; Tammy Anderer, PhD; Duane E. Davis, MD; Steven B. Pierdon, MD; James Pitcavage, MS; and Glenn D. Steele Jr, MD
Association of Electronic Health Records With Cost Savings in a National Sample
Abby Swanson Kazley, PhD; Annie N. Simpson, PhD; Kit N. Simpson, DPH; and Ron Teufel, MD
Learning About 30-Day Readmissions From Patients With Repeated Hospitalizations
Jeanne T. Black, PhD, MBA
Removing a Constraint on Hospital Utilization: A Natural Experiment in Maryland
Noah S. Kalman, MD; Bradley G. Hammill, MS; Robert B. Murray, MA, MBA; and Kevin A. Schulman, MD
Using Clinically Nuanced Cost Sharing to Enhance Consumer Access to Specialty Medications
Jason Buxbaum, MHSA; Jonas de Souza, MD; and A. Mark Fendrick, MD
Real-World Impact of Comparative Effectiveness Research Findings on Clinical Practice
Teresa B. Gibson, PhD; Emily D. Ehrlich, MPH; Jennifer Graff, PharmD; Robert Dubois, MD; Amanda M. Farr, MPH; Michael Chernew, PhD; and A. Mark Fendrick, MD
A Systematic Review of Value-Based Insurance Design in Chronic Diseases
Karen L. Tang, MD; Lianne Barnieh, PhD; Bikaramjit Mann, MD; Fiona Clement, PhD; David J.T. Campbell, MD, MSc; Brenda R. Hemmelgarn, MD, PhD; Marcello Tonelli, MD, SM; Diane Lorenzetti, MLS; and Braden J. Manns, MD, MSc

Health Insurance and Breast-Conserving Surgery With Radiation Treatment

Askal Ayalew Ali, MA; Hong Xiao, PhD; and Gebre-Egziabher Kiros, PhD
Type of health insurance plays a significant role in the likelihood of receiving the recommended treatment among women diagnosed with early-stage breast cancer.
receive BCS with radiation, pij= P( yif = 1), is modeled using where x T/ij  is the vector of observed individual and census tract covariates, β is the vector of regression coefficients to be estimated and uj is the census tract-level random effect. The random effect uj represents unobserved census-tract factors that affect treatment choice and is shared by all women residing in census tract j. The distributional assumptions made about uj were uj ~ N(0, ou2). Odds ratios (ORs) and 95% CIs for each predictor variable considered were estimated overall and for each health insurance type. Log-likelihood ratio test was employed to assess goodness of fit of the estimated models. All the statistical analyses in this study were performed using STATA 11 and all statistical tests were 2-sided.


Descriptive Findings

The descriptive statistics of selected characteristics of women used in this study is presented in Table 1. The mean age at diagnosis was 66 years for the study period 1997 to 2002. Most women were 60 years and above (68.18%), married (57.65%), and non-Hispanic/white (86.07%). The distribution of women by health insurance type was 31.54% privately insured, 1.74% Medicaid, 50.12% Medicare, 14.07% other, and 2.54% uninsured. The primary payers for more than 81% of women in this study were either Medicare or private insurers. The majority of women (58.62%) received BCS. The remaining women had a mastectomy (38.61%) or received no surgery (2.77%). Of those who received BCS, only 47.45% received RT following BCS. Overall, 27.81% women received BCS with RT. In total, 65% of tumors were below 2 cm, 27.24% were 2 cm or larger, and 7.8% had an unknown size.

Table 2 provides the distribution of surgical treatment by type of insurance, race/ethnicity, marital status, age, tumor size, and year of diagnoses. The first 3 columns add up to 100%. The percentages presented in the last column were calculated among women who received BCS treatment. χ² analysis showed significant associations between each of the predictor variables listed in Table 2 and surgical treatment choice. Medicaid-insured and uninsured women were less likely to have surgical treatment than women with either private health insurance or Medicare or “other” types of government insurance (P <.01). The majority of women with private (61.67%), Medicare (57.22%), and other (60.24%) insurance received BCS. The percentage of women who received BCS among Medicaid insured and uninsured women was less than 50% and almost identical (47.18% and 47.13%). Furthermore, there were significant variations by health insurance in the type of surgical treatment received. The percentages of women who received RT after BCS were similar for those insured by Medicaid and those uninsured (41.30% and 40.94%, respectively) and lower compared with women insured privately (45.93%), through Medicare (47.87%), and through other government programs (51.05%). It is important to note that the results presented in Table 2 show that in addition to substantial variations in women’s surgical treatment within a specific type of health insurance, there were also considerable differences between types of health insurance coverage. More specifically, the bivariate analysis reveals that women who were either Medicaid beneficiaries or uninsured had a significant disadvantage in receiving BCS as well as the recommended RT following BCS than women insured by the other 3 health insurance types.

Concerning racial/ethnic differences, non-Hispanic black women were more likely to have no surgery than non-Hispanic white women (P <.01). Similarly, separated, single, and divorced women were less likely to have surgery than married women (P <.01). Married women (49.94%) received RT after BCS at a higher rate than separated (39.18%), single (41.21%), divorced (43.81%), or widowed (46.44%) women. Older women (70+ years) were less likely to receive surgery than women in other age categories. Specifically, patients older than 70 years (45.74%) or 40 to 49 years (43.23%) were less likely to receive RT after BCS relative to women 50 to 69 years (48.94%) and 60 to 69 years (51.44%). Patients with a tumor smaller than 2 cm (50.79%) were more likely to get RT after BCS than women with a tumor 2 cm or larger (39.82%). The percentage of women with no surgery decreased from 5.32% in 1997 to 2.94% in 2002. Despite an upward trend in the use of BCS between 1997 and 2002 to treat early-stage breast cancer among women, the use of RT following BCS did not follow a similar trend.

Results From Multilevel Logistic Regression

Multilevel multivariate logistic regression was used to investigate the impact of health insurance and to identify factors related to receiving the recommended treatment for women diagnosed with early-stage cancer. Agesquared was included in the models to capture the nonlinearity of age in affecting use of BCS with radiation. The percentage of the study population with income below the poverty level and the percentage of those who were 25 years and older and had a high school diploma were skewed, thus natural logarithm transformations of these variables were used in the regression models. The regression analysis was restricted to non-Hispanic whites, non-Hispanic blacks, and Hispanics due to the small sample size for the other race categories. Women who were separated were also excluded from the analysis due to the small sample size.

Predictors of BCS With RT: Non-Stratified Analysis

Table 3 presents the likelihood of receiving BCS with radiation, using a 2-level logistic regression model. The results show that health insurance type had a significant impact on which patients received BCS with radiation even after controlling for sociodemographic factors. The odds of receiving RT after BCS decreased by 27% (OR = 0.73; 95% CI, 0.60-0.88) among women without health insurance relative to women with private health insurance. On the other hand, Medicare-insured women were more likely (OR = 1.10; 95% CI, 1.02-1.18) to receive the recommended treatment than women insured privately. Moreover, women insured by “other” insurance programs (other than Medicaid or Medicare, including Tricare, Military, Veterans Affairs, and Indian/Public Health Services) were more likely (OR = 1.10; 95% CI, 1.01-1.19) to receive the recommended treatment than women who were insured privately. However, women insured by Medicaid were not significantly different from women insured by private health insurance (OR = 0.80; 95% CI, 0.64, 1.01) in their likelihood of having the recommended treatment.

The impact of race/ethnicity was not significant in who received the recommended treatment. The effect of marital status was significant, however. Single, divorced, or widowed women were significantly less likely to receive BCS with RT compared with married women. Age also had a significant effect on receiving BCS with RT. As a woman gets older, she is more likely to receive the recommended treatment; however, the increase was not linear, as it began to decline starting at 70 years (Table 2) and by the estimated coefficient of age-squared. As expected, tumor size also had a significant impact on the odds of receiving RT after BCS. Patients with a tumor 2 cm or larger had significantly lower odds of having RT after BCS (OR = 0.45; 95% CI, 0.43-0.49). In addition, year of diagnosis was significantly associated with use of BCS with RT. Women diagnosed in 1997 were significantly less likely to receive the recommended treatment compared with women diagnosed in 2002.

Moving on the to the impacts of census tract variables, women who resided in census tracts with higher percentages of individuals with at least a high school education were found to be significantly associated with higher odds of receiving RT after BCS. This finding is consistent with those reported in the literature: that women with low education levels are at a disadvantage with respect to breast cancer treatment. The other census tract-level predictor variable considered in this analysis— percentage of people living below poverty—was not significant. The estimate of the random effects variation at the census tract-level was highly significant. This indicates the appropriateness of the multilevel modeling approach in our analysis instead of using the commonly used multivariate logistic regression, which does not take into account the correlation among women who resided in the same neighborhood, as they may be influenced by observed or unobserved common environmental factors.

Predictors of BCS With RT Stratified by Health Insurance Type

Our findings show that receiving the recommended BCS followed by RT among women diagnosed with earlystage breast cancer varied substantially by type of health insurance. Of note is the fact that 1 of the objectives of this study was to investigate whether racial/ethnic disparities exist in the receipt of RT after BCS among patients insured by the same type of health insurance. To meet this objective, before we ran the multilevel regression models that included other predictors and controls, we explored the bivariate relationships between race/ethnicity and the receipt of RT after BCS by controlling for the type of health insurance. The Figure displays the percentage of women who received RT following BCS by race/ ethnicity for each health insurance type. A χ² test was also conducted to measure associations between race/ ethnicity and the receipt of RT after BCS for each health insurance type.

Race/ethnicity was unrelated to the receipt of RT following BCS among women who were insured privately, through Medicare, had “other” insurance, or were uninsured. However, among Medicaid-insured women, there was significant association (P <.05) between race/ethnicity and the receipt of RT after BCS. Among Medicaid-insured women, non-Hispanic white women (36%) and Hispanic women (39%) were less likely to receive RT following BCS compared with non-Hispanic black women (53%). It is also noteworthy to indicate that uninsured non-Hispanic white women fared better (44%) than Medicaid-insured non-Hispanic white women (36%) in receiving RT after BCS.

We were further interested in whether the significant racial/ ethnic differences in using the recommended treatment among Medicaid- insured women observed in the Figure would hold true when the effects of other factors were controlled. Therefore, in an attempt to better understand what other factors may also contribute to such differences, and to identify barriers to the receipt of the recommended treatment, we analyzed the data stratified by health insurance type. To achieve this, 5 separate insurance-specific multilevel logistic regressions were fitted. The large sample size we used gave us the opportunity to detect associations by fitting regression models stratified by health insurance type without being worried that the power of the test may diminish. The results are presented in Table 4.

Copyright AJMC 2006-2020 Clinical Care Targeted Communications Group, LLC. All Rights Reserved.
Welcome the the new and improved, the premier managed market network. Tell us about yourself so that we can serve you better.
Sign Up