Autologous Breast Reconstruction Versus Implant-Based Reconstruction

How Do Long-Term Costs and Health Care Use Compare?

Valerie Lemaine, M.D., M.P.H.; Stephanie R. Schilz, B.A.; Holly K. Van Houten, B.A.; Lin Zhu, M.D.; Elizabeth B. Habermann, Ph.D.; Judy C. Boughey, M.D.


Plast Reconstr Surg. 2020;145(2):303-311. 

In This Article

Patients and Methods

Data Source

We conducted a retrospective analysis of data from OptumLabs Data Warehouse, which includes deidentified claims data for privately insured and Medicare Advantage enrollees in a large, private, U.S. health plan.[5] The database contains longitudinal health information on enrollees, representing a diverse mixture of ages, ethnicities, and geographic regions across the United States.[6] The health plan provides comprehensive full insurance coverage for physician, hospital, and prescription drug services.

Study Population

We identified all women (aged 18 to 64 years) with a diagnosis of breast cancer who underwent immediate breast reconstruction between January of 2004 and December of 2014. (See Table, Supplemental Digital Content 1, which shows the CPT codes used for cohort creation for mastectomy and for reconstruction, The date of the mastectomy and immediate breast reconstruction hospitalization was defined as the index date. Patients were required to have at least 12 months of continuous medical coverage before the index surgery date (baseline period) and 2 years after the discharge date.

Independent Variables

Baseline characteristics included age, race (i.e., white, black, Hispanic, Asian, unknown), geographic region (i.e., Midwest, Northeast, South, West, other), Charlson-Deyo comorbidity index (modified), year of reconstruction, type of reconstruction (i.e., autologous or implant), hospital length of stay, and laterality. Age was stratified into four categories (i.e., 18 to 34 years, 35 to 44 years, 45 to 54 years, and 55 to 64 years). The Charlson-Deyo index was modified using only non–breast cancer baseline comorbidities captured by International Classification of Diseases, Ninth Revision, codes in any position on claims occurring within 12 months before the index reconstruction date.[7] For analysis, the index was stratified into four categories (i.e., 0, 1, 2, and ≥3).


The primary outcomes were 2-year use and cost of care. Use measures included rates of subsequent hospitalizations, emergency room visits, and office visits during the 2 years after discharge of reconstruction hospitalization. Rates were presented per 100 women. OptumLabs data contain total paid amounts, which is the sum of out-of-pocket (patient) and health plan (insurer) spending associated with all medical claims (i.e., ambulatory visits, emergency room visits, inpatient hospitalizations, and other services). Costs were reported for the 2-year follow-up period from reconstruction hospitalization discharge and include professional services. All costs were adjusted to 2015 U.S. dollars using the Consumer Price Index.[8] Reconstruction costs included all professional services from admission to discharge date of reconstruction hospitalization. The postreconstruction costs refer to all costs incurred over the subsequent 2 years after reconstruction hospitalization including inpatient, outpatient, and clinical services costs (including imaging) and any subsequent reconstruction costs. Of note, office visits during the global period were captured by CPT codes but have zero cost.

Statistical Analysis

Patient characteristics (i.e., age, race, region, modified Charlson-Deyo comorbidity, year of reconstruction, type of immediate reconstruction, length of stay, and laterality) were described using mean (standard deviation) or count (percentage) as appropriate. Furthermore, we examined differences in patient characteristics among patients with autologous breast reconstruction versus implant-based breast reconstruction using chi-square and t tests to assess univariate differences. Unadjusted use rates were compared between the autologous and implant-based breast reconstruction patients using t tests. A generalized linear model was used to adjust for patient characteristics and calculate predicted costs. Statistical differences in incremental costs were determined using 95 percent confidence intervals. A value of p < 0.05 was considered statistically significant. All statistical analyses were performed using SAS software version 9.4 (SAS Institute, Inc., Cary, N.C.).