Academic Commons

Theses Doctoral

The Causal Effect of Early Surgical Intervention on Biliary Atresia Patients Clinical Outcome Using Economic and Education Models

Hoshino, Eri

Purpose and Background: The aim of this paper is to investigate the causal effect of the timing of Kasai procedure, a surgical intervention, on Biliary Atresia (BA) patient outcomes using quasi-experimental techniques developed in economics and education. BA is a rare pediatric disease and is the most common cause of liver disease-related death in children. The causes of BA are not well established and are probably multifactorial. The Kasai procedure is usually performed at the early stage of disease, within the first few months of a patient’s life. Although performance of the Kasai procedure before 60 days of age suggests improved prognosis, no available research to date has investigated the causal effect of the timing of the Kasai procedure on patient outcomes. Methods: This study uses data drawn from the Japan Biliary Atresia Registry. The dataset contains a total of 2743 patients; 1743 girls, and 1000 boys registered since 1989 to 2012. Subjects were eligible for study inclusion provided that they were born between 1989 and 2012 and underwent the Kasai procedure. In addition, eligible subjects had to have had an observable primary predictor variable and outcome. The primary predictor variable was the age at Kasai procedure and the outcome was 1-year native liver survival without jaundice. Other variables included in the analyses were sociodemographics (i.e. gender, birth order, birth weight, gestational age, parental age), clinical characteristics (i.e. types of obstruction, associated anomalies, type of choleretic agents), and hospital characteristics defined as hospital caseload. Bivariate analysis, multivariate logit/probit regressions, stratified multivariate logit regression, hospital fixed effects, and instrumental variable (IV) approaches were conducted. Findings: The IV approach was used to examine whether the timing of the Kasai procedure was endogenous in the model. Although using the length of stay at the last hospital prior to Kasai procedure as a proxy for accessibility to the nearest high-level hospital had its strengths as the instrument as confirmed by the first-stage diagnostics, the Wu-Hausman test indicated that the estimates of the model with and without IV were consistent. The results from the multivariate covariate-adjusted logit regression suggested that for patients with obstruction types I and I with cyst, the timing of Kasai procedure was not associated with 1-year native liver survival. However, in patients with type III obstruction (the majority of BA patients [86%]), when comparing to the timing of Kasai procedure at 61 to 75 days, the timing of Kasai procedure at less than 30 days was suggestively associated with the primary outcome of 1-year native liver survival without jaundice (log-odds ratio: 0.43 [t=1.79]). This was 1.5 times as likely to achieve 1-year native liver survival without jaundice while holding other variables constant. The timing of Kasai procedure after 76 days had a statistically negative association (log-odds ratio: -0.36 [p<0.05]) with the outcome. Moreover, the Kasai procedure at 106 to 121 days of age decreased the log odds of 1-year native liver survival without jaundice by 0.83, which equated to an increase in likelihood of achieving the primary outcome of 0.43 times. The Kasai procedure after 121 days of age also decreased the log odds of the outcome by 1.17 that was 0.3 times as likely to achieve 1-year native liver survival without jaundice. These differences are quite significant; therefore, the effort to increase the patients who could have the Kasai procedure at less than 30 days and to prevent the Kasai procedure after 76 days should be made to improve 1-year native liver survival without jaundice for the type III patients. As previously suggested, cholangitis and the use of corticosteroid have a negative association with the outcome after adjusting for covariates. Ursodeoxycholic acid indicates a positive association with the outcome, as does the hospital caseload, even when both are adjusted for covariates.

Files

  • thumnail for Hoshino_columbia_0054D_12984.pdf Hoshino_columbia_0054D_12984.pdf binary/octet-stream 4.85 MB Download File

More About This Work

Academic Units
Economics and Education
Thesis Advisors
Levin, Henry M.
Degree
Ph.D., Columbia University
Published Here
October 15, 2015
Academic Commons provides global access to research and scholarship produced at Columbia University, Barnard College, Teachers College, Union Theological Seminary and Jewish Theological Seminary. Academic Commons is managed by the Columbia University Libraries.