A Social Network Analysis of Provider Networks Leading
to Opioid Initiation Among US Adults Volume 4 - Issue 2
Divyan Chopra1, Chenghui Li?1, Jacob T Painter?1, Jonathan P Bona?2, Intawat Nookaew?2, and Bradley C Martin?1*
1Division of Pharmaceutical Evaluation and Policy, University of Arkansas for Medical Sciences, USA
2Department of Biomedical Informatics, University of Arkansas for Medical Sciences, USA
Received:May 08, 2021; Published: May 25, 2021
Corresponding author: Bradley C Martin, PharmD, PhD, Professor, Division of Pharmaceutical Evaluation and Policy, University of
Arkansas for Medical Sciences College of Pharmacy, 4301 W. Markham, slot 522 Little Rock, AR 72205, 501.603.1992 (office), USA
Background: About 38% of U.S. adults use prescription opioids. Opioid-related overdoses have risen in the last decade. Studies
suggest that duration and type of first opioid prescribed play a significant role towards risky and chronic opioid use.
Objective: The primary objective was to identify key players involved with initiation of patients on opioids. This study also
identified patient characteristics most likely to receive care from influential opioid providers as opioid initiation interventions could
be directed to such physicians.
Design/Participants: A cohort of patients with incident opioid use were identified using Arkansas All-Payer Claims Database
from 2013-2018. A social network comprising provider as nodes and referrals as edges was constructed based on healthcare
utilization 180 and 90 days preceding incident opioid use for potentially non-acute and potentially acute pain patients respectively.
Main Measures: Network centrality measures such as indegree (referrals received), eigenvector (neighbor centrality),
betweenness (involvement) and closeness (reach) were estimated. Outcomes included influential providers determined by network
centrality measures, initial opioid providers, and patients of influential providers. Covariates included provider demographics,
patient demographics and clinical characteristics. Generalized linear mixed effect models were used for statistical analyses.
Key Results: There were 150,676 incident opioid users who visited 12,629 healthcare providers. Primary care providers
(PCPs) were found to have higher centralities except for betweenness. Initial opioid prescribers showed higher network centralities.
More complex patients such as those with more pain conditions and higher levels of comorbidity were more likely to seek opioid
providers at the periphery of the network (lower indegree and eigen centrality).
Conclusions: This study highlights the characteristics of networks leading to opioid use. Findings that PCPs have higher influence
and higher influence of initial opioid prescribers can guide opioid initiation guidelines as targeted interventions. Moreover, efforts
should be undertaken to direct patients requiring judicious opioid prescribing to influential providers.
Keywords: Opioids; physicians; social network analysis; centrality.