Author

DANMENG HUANG

Date of Award

12-2019

Degree Name

Doctor of Philosophy (PhD)

Advisor(s)

STEPHEN H. LINDER

Second Advisor

SUJA S. RAJAN

Third Advisor

DAVID R. LAIRSON

Abstract

Current guidelines for screening mammography recommend different start ages and intervals for women with average cancer risk. Longer intervals between mammograms allow more time for a tumor to grow, to become symptomatic and clinically detectable, and more likely to be at advanced stage. Existing literature on the associations between mammography screening frequency and risk of more advanced breast cancer have mixed results. Studies have showed that advanced breast cancer costs more to treat, but real-world cost estimates following different mammography screening frequencies are unavailable. To fill the gaps, this dissertation aimed to quantify the clinical and financial impact following annual versus biennial screening. Commercial claims database provides rich real-world information on diagnoses, medical resources used and the associated costs. To supplement claims database with breast cancer stage information, this study first developed and validated an algorithm with classification and regression tree method using SEER-Medicare data. The performance was measured with sensitivity, specificity, positive predictive value, negative predictive value and area under the receiver operating characteristic curve. The algorithm was then applied to The

MarketScan® Commercial Claims and Encounters Database to estimate breast cancer stage at diagnosis. Incident breast cancer cases identified in the MarketScan database were categorized as annual, biennial and non-screeners based on their pre-diagnosis screenings. Partial proportional odds model was used to estimate the odds ratio of having more advanced breast cancer. Stratified analysis by age was also conducted. For the three screening groups, total healthcare costs, insurer costs, and out-of-pocket costs adjusted by generalized linear model with gamma distribution and log-link function were reported. The staging algorithm had improved performance than others, especially in the prediction of non-invasive cases, early stage cancers and metastases. Generally, regular screening protects women from more advanced breast cancer. Compared to biennial screening, annual screening was associated with a reduced risk of later stage invasive cancers for both women in their 40s and older. For health insurance payers, there were cost savings in healthcare costs with regular screening, especially for annual screening. Cost reduction was more obvious among women aged 40-49. Compared to insurer’s costs, out-of-pocket costs borne by patients were minimal. Although this study showed both clinical and financial benefits in annual versus biennial screening, the optimal screening frequency should be an individual decision weighing both the benefits and harms.

Share

COinS