real-world evidence study of factors associated with ......real-world evidence study of factors...

Real-World Evidence Study of Factors Associated With Breast Cancer Conserving Surgery for Females Diagnosed With Early Stage Breast Cancer

Judy George, PhD,1 Joseph Tkacz, MS,1 M. Christopher Roebuck, PhD, MBA,2 Fredy Reyes, PhD,1 Yull E. Arriaga, MD, 1 Gretchen P. Jackson, MD, PhD 1,3 Irene Dankwa-Mullan, MD, MPH 1.1 IBM Watson Health, Cambridge, MA, USA 2 RXECONOMICS LLC, Hunt Valley, MD, USA 3 Vanderbilt University Medical Center, Nashville TN

National and Regional Time Trends in BCS Logistic Model of BCS versus Mastectomy Results (n=57,299)Problem• Breast Conserving Surgery (BCS) is now considered an acceptable

alternative to mastectomy for patients with non-metastatic invasivebreast cancer (BCa).

• Limited research examining factors influencing BCS reveal importantpredictors such as insurance, sociodemographic characteristics, andavailability of health care services

Research Objective• Examine factors associated with BCS in a large, contemporary cohort of

commercially insured females with breast cancer.

This retrospective observational study of the IBM® MarketScan®Commercial and Medicare Supplemental Database: 1/1/2012 –3/31/2018.

This study showed increased BCS with older age, access to and use of genetic services, living in communities with a higher density ofphysicians specialized in medical genetics & nuclear medicine physicians, and decreased BCS use with certain co-morbid conditions andthe regions of the South, Midwest, or West. This real-world evidence study contributes to informing clinical practice on the significantpatient- and community-level factors influencing BCS in a non-metastatic invasive breast cancer cohort.

Higher BCS use was significantly associated with(Figure 3):• Older age (OR: 1.96-2.44, p<.0001)• Genetic testing (OR: 1.25, p<.0001)• More recent year of diagnosis (OR: 1.08-1.72,

p<.0001-p=.02)• Neoadjuvant radiation therapy (OR: 4.20, p<.0001)• Higher density of physicians in specialties of medical

genetics (OR: 6.50, p<.01; ME=.41) or nuclearmedicine (OR: 2.24, p=.04; ME=.17)

Lower BCS use was associated with:• Residing in the South, Midwest, or West (OR: 0.71-

.89, p<.0001-p=.01)• Comorbid dementia (OR: 0.58, p=.01) and comorbid

congestive heart failure (OR: 0.79, p=.01)• Higher density of hospitals with medical/surgical

services (OR: 0.26, p<0.01; ME=-.29) or plasticsurgeons (OR: 0.66, p<.0001; ME=-.09)

BACKGROUND

METHODS

57,299 met inclusion criteria of which, 67.1% (38,474)had BCS, ranging from 62.9% in 2012 to 73.3% in 2017(Figure 2). While BCS increased overall, differences wereobserved on most factors considered between patientswho had BCS and those who did not (Table 1). However,of 15 comorbidities measured, only two (diabetes[p<.001] and chronic obstructive pulmonary disease[p=.04]) varied significantly.

CONCLUSIONSThis cross-sectional study relied upon privately-insured commercial claims dataand results may not be generalizable. We inferred community-level characteristicsbased on county-level data. Additionally, clinical data sources (e.g., biomarker,hormone receptor status) may further explain BCS vs. mastectomy selection.

LIMITATIONS

RESULTS

A

Figure 2. National and Regional Trends in BCS. Average of BCS proportionsacross US census regions (A) Proportion of BCS (2012-2017) in each USCensus region per year (B)..

Figure 3. Odds Ratios (ORs) with 95% confidence intervals (A, B) and Marginal Effects (MEs) (C) of factors influencing BCS. Factors arecategorized according to patient-level and Community(ZIP3)-level (+) data. # represents number per 10,000 residents. Significancelevels are designated (*= p<.05; **= p<.001). ZIP3-level p-values were based on clustered standard errors.

Comparison of Patients Receiving BCS or Mastectomy n=57,299

B A C

Patients 60 years and older were at least

more likely to receive BCS relative to those younger than 50

Patients with BRCA 1 or 2 genetic testing were

more likely to receive BCS

Patients in communities with a higher density of medical geneticists and nuclear medicine physicians were more likely to receive BCS

Figure 1. Cohort selection process

Table 1. Characteristics of study population. Results from Kruskal-WallisEquality of Populations Tests.

B

Patient-Level Factors

Breast Conserving

Surgery(N=38,474)

Mastectomy (N=18,825) p-value*

Age (in Years)Under 50 18.1% 35.0% p<.001 50-59 34.0% 32.6% 0.01 60-69 30.1% 20.9% p<.001 70-79 12.4% 7.6% p<.001 80+ 5.3% 3.9% 0.004RegionNortheast 21.5% 16.4% p<.001Midwest 24.9% 22.3% p<.001South 37.7% 45.7% p<.001West 15.9% 15.6% 0.59Plan TypePPO, POS, or Comprehensive 74.3% 73.0% 0.01HMO or Exclusive Provider Organization (EPO) 12.3% 11.6% 0.20Consumer Directed Health Plan (CDHP) 7.7% 8.8% 0.04High Deductible Health Plan (HDHP) 4.5% 5.3% 0.13BRCA 1/2 Testing StatusHad a Genetic Test 87.2% 84.9% p<.001In Situ Diagnosis Status

Also Had In Situ on Index Date 18.4% 18.1% 0.55Also Had In Situ pre-Index Date 24.6% 28.4% p<.001Neoadjuvant TherapyChemotherapy Prior to Surgery 4.1% 10.5% p<.001Radiation Therapy Prior to Surgery 1.7% 0.4% 0.01