Empirical Evidence on Telehealth and Its Clinical Implications

Abstract

Chapter 1: The first study evaluates the impact of state Medicaid telehealth parity policies on opioid use disorder (OUD) related admissions to treatment facilities. Using Treatment Episode Data Set–Admissions (TEDS-A) data from 2010 to 2019 linked with state telehealth policy information from the American Telemedicine Association, I estimate the effects of telehealth parity adoption using a generalized difference-in-differences framework with two-way fixed effects. To strengthen causal interpretation, the analysis incorporates robustness checks addressing potential confounders such as Medicaid expansion, section 1115 substance use disorder waivers, interstate licensure, and OUD dispensing laws, in addition to falsification tests. The findings indicate that Medicaid telehealth parity is associated with a non-statistically significant reduction in OUD-related admissions. In the fully adjusted model, parity adoption is associated with a modest statistically insignificant reduction relative to the pre-period mean. Stratified analyses show that these statistically insignificant reductions are found across all treatment settings; suggesting that while telehealth does play an integral role in the OUD care continuum, it may not be associated with downstream clinical effects on admissions Chapter 2: The second study examines whether telehealth utilization declined when states ended their COVID-19 emergency orders and, in some cases, allowed temporary private insurance reimbursement parity requirements to expire. This study uses Household Pulse Survey data and a staggered-adoption framework utilizing timing variation of states’ emergency-order ending. Due to the variation created in the withdrawal of a major telehealth incentive, this setting offers an opportunity to test whether telehealth use remained dependent on parity requirements once telehealth had already become integrated in care delivery. Using Callaway-Sant’Anna difference-in-differences models and event-study analyses, I find no statistically significant overall decline in telehealth utilization following the end of state emergency orders and the loss of private parity in the pooled analysis. Telehealth use fell from pandemic peaks, but exhibited broadly similar post-period patterns, suggesting continuity rather than sharp contraction. Stratified analyses were likewise null. These findings suggest that once telehealth has been incorporated into clinical workflows, patient expectations, and organizational infrastructure, the withdrawal of a single telehealth policy component may not be sufficient to reverse utilization in the short term. More broadly, this study indicates that telehealth persistence in the post-emergency period may reflect institutional adaptation, sunk implementation costs, and lagged insurer responses, rather than simple dependence on payment parity requirements alone. Chapter 3: The third study evaluates telehealth use and relational continuity of care among outpatient neurology patients at Tulane University. Using outpatient scheduling data from March 2020 through April 2021, this study analyzes encounter-level information on appointment disposition, and patient characteristics to assess which patients used telehealth and how appointment dispositions varied across groups. Logistic regression models, using cluster bootstrapped standard errors, are used to estimate factors associated with ever using telehealth and with continuity-related outcomes such as no-shows and appointment completion. The findings show that telehealth users in this specialty setting were disproportionately White, female, and commercially insured. Medicare patients were significantly less likely than commercially insured patients to use telehealth. At the same time, no-show odds varied sharply across demographic and insurance categories: Black patients had substantially higher odds of no-shows than White patients, and Medicaid and self-pay patients also had higher odds of missed visits relative to commercially insured patients. Telehealth use rose rapidly early in the pandemic, increasing from negligible levels to a peak of roughly 45% of encounters before declining to a lower but still higher than baseline level. Taken together, these findings suggest that telehealth became an important component of outpatient neurology care, but that its adoption and continuity-related benefits were unevenly distributed across patient groups. In this sense, telehealth may improve flexibility and help preserve access for some patients while still leaving underlying inequities in continuity of care unresolved.