Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

Assessing the potential effects of unmeasured confounders in a single and a meta-analysis of opioid use disorder studies

Shi, Xiaojun (2022) Assessing the potential effects of unmeasured confounders in a single and a meta-analysis of opioid use disorder studies. Master's Thesis, University of Pittsburgh. (Unpublished)

This is the latest version of this item.

Download (323kB) | Preview


Incidence of opioid abuse disorder (OUD) and overdose death have reached epidemic proportions. In the intervention of OUD, it is necessary for healthcare providers to understand the relationship between subjects’ characteristics and usage of medication for OUD. As the largest financing source in the US for OUD treatments, Medicaid provides a wealth of data for researchers to examine the exposure-outcome relationship, yet a major concern is that unmeasured confounders could explain away the observed relationship.
The goal of this thesis is to apply a sensitivity analysis measure to examine the unmeasured confounding effect that could influence the association between the effect of a Medicaid enrollee’s characteristics and the OUD outcomes, where the effects included individual characteristics of interest include race/ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, others) and eligibility group (pregnant women, youth, adults with disability, nondisabled adults), and the outcomes included receiving medication for OUD and OUD medication continuity for 180 days. The associations have been previously studied among Medicaid enrollees form a US individual state and from meta-analysis with multiple US states. Based on the published estimated associations, the study investigated about how strong the unmeasured confounding must be in order to completely explain away the effect of an enrollee’s characteristics on the OUD outcomes. Results showed that the reported 95% confidence intervals of the estimated effects changes so much as to include the null hypothesis of no effects when the effects of unmeasured confounders, in a risk ratio scale, are ranging from 1.03 to 1.51. For meta-analysis, the minimum confounding strength required to explain away the confounded effects in one out of ten states with true effects exceeding the scientific importance threshold is ranging from 2.24 to 2.80. These results showed that the associations between enrollees’ characteristics and the OUD outcomes are weak and not robust to unmeasured confounders.
Public health significance: E-value is a useful tool used to perform sensitivity analysis of the impact of unmeasured confounding for observational study. Results of sensitivity analysis should be included for researchers to correctly assess the robustness of exposure-outcome relationships leading to a design strategy or a medical decision.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
Shi, Xiaojunxis32@pitt.eduxis32
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairChang, Chung-Chou H.changj@pitt.educhangj
Committee MemberTang, Lulutang@pitt.edulutang
Committee MemberTalisa, Victor Brodzikvit13@pitt.eduvit13
Date: 4 January 2022
Date Type: Publication
Defense Date: 8 December 2021
Approval Date: 4 January 2022
Submission Date: 17 December 2021
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Number of Pages: 36
Institution: University of Pittsburgh
Schools and Programs: School of Public Health > Biostatistics
Degree: MS - Master of Science
Thesis Type: Master's Thesis
Refereed: Yes
Uncontrolled Keywords: E-value, sensitivity analysis, observational study, opioid abuse disorder, causal treatment effect
Date Deposited: 04 Jan 2022 15:39
Last Modified: 04 Jan 2022 15:39

Available Versions of this Item

  • Assessing the potential effects of unmeasured confounders in a single and a meta-analysis of opioid use disorder studies. (deposited 04 Jan 2022 15:39) [Currently Displayed]


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item