MINING PATIENT CLAIM DATA TO DETECT DRUG SIGNALS THAT CAN REDUCE THE RISK OF BARRETT'S ESOPHAGUS PATIENT DEVELOPING ESOPHAGEAL ADENOCARCINOMA
Author(s)
Wu Y, Cohen T, DeSantis S, Iyengar S, Hoot N, Soysal E, Tao C, Jiang M, Xu H
The University of Texas Health Science Center at Houston, Houston, TX, USA
OBJECTIVES: Esophageal Adenocarcinoma (EA) is a rapidly progressing fatal disease. Reported five-year survival rates are less than 20%. This research aims to identify any medication that can reduce the risk of EA development in BE patients using a large scale claim data. METHODS: We developed a BE patient cohort from the Blue Cross Blue Shield claim data in Texas (BCBSTX). Case subjects were selected among the BE patients who developed EA at least six months after the diagnosis of BE. This group was matched to the control subjects who didn’t develop EA by incidence density sampling. We focused on three promising drug categories (statins, aspirins and PPI drugs).We used a regression model to examine the variables to find the drugs that could potentially reduce the risk of BE patient developing EA. RESULTS: The case subjects were composed of 115 patients. We used the incidence density sampling algorithm to match the cases to the controls at a ratio of 1:6. There were 305 patients with exposure to statins, 547 patients with exposure to PPIs and 252 patients with exposure to aspirins. The results showed that the statins (incidence density ratio, 0.56; 95% CI, 0.36–0.89) and aspirins (incidence density ratio, 0.76; 95% CI, 0.47–1.2) were more likely to reduce the risk of developing EA from BE. However, the signal can not be replicated for PPIs. CONCLUSIONS: BE is a frequent health problem with an average estimated prevalence 2-3%. Any factor delaying or blocking this progress will have a great impact on controlling the development of EA. The preliminary results revealed two possible medications, statins and aspirins, which may reduce the risk of esophageal adenocarcinoma in Barrett’s Esophagus patients. Due to the limitations, we need to conduct further study to examine the signals by exploring more types of data, such as the patient record data from clinical practice.
Conference/Value in Health Info
2016-05, ISPOR 2016, Washington DC, USA
Value in Health, Vol. 19, No. 3 (May 2016)
Code
PCN46
Topic
Epidemiology & Public Health
Disease
Oncology