« Back to Results

Current Topics in Health Economics and Policy

Paper Session

Tuesday, Jan. 5, 2021 12:15 PM - 2:15 PM (EST)

Hosted By: Chinese Economic Association in North America
  • Chair: Shin-Yi Chou, Lehigh University

Characteristics of Statewide Prescription Drug Monitoring Programs and Associated Potentially Inappropriate Opioid Prescribing: A Machine Learning Application

Hsien-Chang Lin
,
Indiana University
Zhi Wang
,
Indiana University
Yi-Han Hu
,
Indiana University
Anne Buu
,
University of Texas-Houston
Kosali Simon
,
Indiana University

Abstract

Unnecessary opioid prescription has become a major public health concern in the U.S. Statewide prescription drug monitoring programs (PDMPs) have been implemented to improve safe prescribing practice and reduce unnecessary opioid prescribing. Yet, no previous studies have comprehensively evaluated the effectiveness of PDMPs in reducing opioid-related potentially inappropriate prescribing (PIP) practices. The objective of this study is to apply machine learning methods to comprehensively evaluate PDMP effectiveness by examining how different PDMP characteristics (including program administration arrangements, access requirements, and reporting schemes) across states are associated with opioid-related PIP for non-cancer chronic pain (NCCP) treatment. This study analyzed over 1.8 million NCCP-related health insurance claims with opioid prescriptions from a major insurer during January 2007 to July 2016. Opioid-related PIP practices (including dosage strength, days supply, and benzodiazepine-opioid co-prescription) were examined to assess the importance of different PDMP characteristics using five regression and tree-based models. Opioid-related PIP practices are prevalent in NCCP treatment and therefore improving the effectiveness of PDMPs is indeed critical to reduce unnecessary opioid prescriptions. Our results indicated that penalized regressions and tree-based models, compared to logistic regression models, provided slightly better but limited improvement in model performance when predicting opioid-related PIP practices. Our study also identified the most important PDMP characteristics that are associated with PDMP effectiveness. With the knowledge of the most essential PDMP characteristics, the study findings could imply the redesign of PDMPs to reduce opioid-related potentially inappropriate prescribing practices, and may in turn, improve opioid-related health outcomes.

Information Campaign and Pain Relief Prescription

Shin-Yi Chou
,
Lehigh University
Pei-Chuan Ho
,
Lehigh University

Abstract

This study investigates how physicians responded to a series of information regarding pain reliefs and pain management in terms of prescribing behaviors since late 1990 in the United States. In 1996, Purdue Pharma L.P. introduced Oxycotin to the U.S. market and marketed it for non-cancer pain treatment. The marketing targeted physicians who prescribe fewer pain reliefs as well as primary care physicians. In 2000, Veterans Health Administration (VHA) had established a protocol of pain treatment, while the Joint Commission introduced a series of pain management guidelines in 2001. Physicians may respond to the series of information regarding pain management by changing prescribing behaviors. We investigated whether physicians’ prescribing behaviors varied across the level of pain associated with different medical conditions. We analyzed the National Ambulatory Medical Care Survey (NAMCS) data in 1995 to 2015 and found that, in office-based visits, physicians did not respond to the information immediately. Since early 2000’s, physicians started to prescribe more pain reliefs for medical conditions associated with mild pain rather than medical conditions associated with severe pain. The behavioral change was highly persistent: this increased pain relief use for mild pain remained salient even after the opioid abuse already caught the attention of the public.

Global Movement Restrictions in a Labor-Leisure Pandemic World

Erkmen G. Aslim
,
Grand Valley State University
Manuel Hoffmann
,
Stanford University
Ruixin Jia
,
Texas A&M University
Murat C. Mungan
,
George Mason University

Abstract

Almost all nations of the world are affected by the current mitigation measures that aim to control the spread of Covid-19. We study the implications of lifting those measures on health, economic behavior and market outcomes across the world by partnering with a university having students dispersed across the globe. We hypothesize the effect of these interventions to be heterogeneous since individuals may have different preferences for social connectivity. We incorporate social preferences during work and leisure time in the canonical labor-leisure model. The model predicts smaller reductions in productivity and labor market expectations for individuals with higher social dependence on leisure activities, as opposed to individuals who are socially more connected in their work lives. Moreover, individuals with higher absolute social dependence are likely to experience larger reductions in overall well-being. We test the predictions by identifying the social preference types in each domain via administrative, survey, and experimental variation from our partner university. In our surveys, we elicit responses related to time spent on leisure and labor activities with and without social connectivity. Moreover, a randomized intervention of leisure and labor activities allows us to cross-validate these types by assessing how much individuals value social connectivity in each domain. We interact the types with the natural policy variation from Covid-19 to estimate the heterogenous causal effect of the mitigation measures. Overall, our findings will inform policy decisions related to social, economic, and psychological disruptions, with the objective of normalizing social and living conditions amid the pandemic.

Is Home Eviction Associated with Child Abuse and Neglect? Empirical Evidence from a National Database of Eviction Records and Child Protective Services Data

Shichao Tang
,
Centers for Disease Control and Prevention
Daniel A. Bowena
,
Centers for Disease Control and Prevention
Laura Chadwick
,
U.S. Department of Health and Human Services
Emily Madden
,
U.S. Department of Health and Human Services
Robin Ghertner
,
U.S. Department of Health and Human Services

Abstract

Home eviction can increase the likelihood of homelessness and exacerbate the negative effects of poverty on families. Eviction is especially detrimental for low-income families who may not have a safety net (e.g., no other places to live, no income resources to support basic living expenses) after eviction. Home eviction has been linked to a higher risk of negative mental health outcomes for both parents and their children. Parental stress as a result of eviction may further increase the risk of their children experiencing maltreatment. This study aimed to understand the relationship between home eviction and child abuse and neglect (CAN). Data from 2000-2016 National Child Abuse and Neglect Data System (NCANDS) and Adoption and Foster Care Analysis and Reporting System (AFCARS) were used. We focused our analyses on all children’s foster care entries and substantiated child maltreatment cases reported to child protective services (CPS). The home eviction data are from the Eviction Lab at Princeton University, which collected formal eviction county-court records from 48 states and the District of Columbia from 2000-2016. We used two-way fixed-effect models to examine the association between home eviction and CAN. Preliminary results show that a 1 percent increase in the home eviction filing rate is associated with a 0.46 percent (p = 0.041) increase in the rate of children’s foster care entries. In addition, a 1 percent increase in the home eviction rate is associated with a 1.14 percent (p = 0.008) increase in the rate of children’s foster care entries. No significant association was found between the home eviction rate or the home eviction filing rate and substantiated child maltreatment rate. Providing stable and affordable housing may reduce foster care entries and prevent CAN. Approaches that strengthen family’s economic supports (e.g., assisted housing mobility), that might effectively reduce parental stress and CAN.
Discussant(s)
Shin-Yi Chou
,
Lehigh University
Zhi Wang
,
Indiana University
Wei Fu
,
University of Pennsylvania
Chia-Lun Liu
,
University of Pennsylvania
JEL Classifications
  • I1 - Health
  • K4 - Legal Procedure, the Legal System, and Illegal Behavior