Health Care Visit Modality Among Older Adults: Telehealth and Subsequent Emergency Department Use

Abstract

Introduction:

Telehealth may address deficiencies in care provision for older adults, but potential downsides when used in place of in-person care are unclear. The use of subsequent care may be informative in a comprehensive evaluation of telehealth.

Methods:

We used administrative claims data from Optum’s de-identified Clinformatics^® Data Mart Database to explore telehealth use in outpatient care and subsequent emergency department (ED) utilization before and during the COVID-19 pandemic.

Results:

We found evidence that telehealth use is associated with a higher likelihood of subsequent ED visits within six months relative to in-person care, but estimated differences vary across years and by age group. In 2021, estimated differences are statistically significant and increasing by age group.

Conclusion:

Telehealth may be associated with more frequent subsequent use of ED care compared with in-person care; however, more work is needed to understand healthcare quality differentials by visit modality and whether the results are causal.

Introduction

Telehealth utilization increased dramatically during the COVID-19 pandemic, creating the potential to extend access to care while reducing disease spread. ¹ Synchronous audio/video telehealth has transformed outpatient care delivery in the United States. ² A significant demographic group possibly gaining from increased telehealth access in outpatient care is older patients, both for physical access and disease prevention reasons. Existing literature has documented the rise of telehealth for aging patients as well as future benefits.^3,4 However, differences in subsequent emergency care between synchronous telehealth and in-person services are understudied. While barriers to telehealth uptake in older populations have been documented, how patient outcomes are correlated with telehealth services used in place of in-person care is less understood.^5,6

In this study, we explore differences in the likelihood of a patient emergency department (ED) visit within six months of an office/outpatient evaluation and management (E/M) service claim by visit modality and age cohort. We evaluate the association between increased telehealth utilization and subsequent ED visits, discussing the implications of telehealth use in older patients.

Methods

Data and study design

We used medical claims, diagnosis, and member enrollment data from Optum’s de-identified Clinformatics^® Data Mart Database, a comprehensive commercial claims database with coverage in all 50 U.S. states, with representativeness discussed in the existing literature. ⁷ Our sample includes encounter-level telehealth and in-person office and other outpatient E/M service claims (Current Procedural Terminology^® [CPT^®] codes 99201–99215) between January 2018 and March 2022 for patients enrolled in commercial and Medicare Advantage insurance plans. De-identified claims data are longitudinal, such that individuals can be linked across time using anonymized identifiers. Synchronous audio/video telehealth visits were identified by either CPT^® modifier -95 or place of service code 02 or 10 and otherwise were considered in-person.

This study was reviewed by the Human Research Protection Program at Indiana University and was determined to not require Institutional Review Board (IRB) review because the project does not involve human subjects as defined in 45 CFR 45.102(f).

Variables of interest

The outcome of interest is whether a patient with an office/outpatient E/M service visit experiences any ED visit within six months of the encounter date. An individual ED visit is identified through any patient-linked encounter with revenue codes 0450 through 0459. The primary grouping variable is patient age, where we group patients into four distinct cohorts: below 65, 65 to 74, 75 to 84, and 85 years and older. The key independent variable is an indicator variable reflecting encounters using telehealth for the visit modality, with in-person encounters as the comparison group.

Additional covariates in the study design include patient age, race, gender, COVID-19 diagnosis, and Charlson Comorbidity Index (CCI). Clinician state is included to account for potential geographic variation in telehealth utilization. Visit CPT^® code levels are included to account for variations in severity across E/M service encounters.

Statistical analysis

Using logistic regression, we estimate the average marginal effects associated with telehealth relative to in-person modality on the likelihood of patient ED visit within six months using an encounter-level model. We report average monthly totals of office/outpatient E/M services from January 2018 to March 2022. To observe a full six-month outcome window, we limit our analysis to January 2018 through September 2021. For reported results, estimates are summarized into annual averages.

To complement the primary findings, we report trends in patient six-month ED visit rates by age cohort. Rates are reported in counts of ED visits per 1,000 encounters for each month between January 2018 and September 2021. All analyses in this study were executed using R Statistical Software (v.4.1.1) with tidyverse (v.2.0.0), fixest (v.0.11.0), and marginaleffects (v.0.9.0) packages.^8–11

Results

Figure 1 displays the mean count of monthly office/outpatient E/M service encounters by visit modality and year. The typical monthly number of office/outpatient E/M encounters ranges between five to six million individual visits between 2018 and 2022. Monthly telehealth use rises from less than 1% of all visits before 2020 to 7–12% of all office/outpatient E/M services in 2020 and onward.

FIG. 1.

Average monthly outpatient or other office visit E/M service claims, 2018–2022 (by visit modality). This figure displays a bar plot of the mean count of monthly office/outpatient E/M service encounters by visit modality and year, constructed using encounters from Optum’s De-identified Clinformatics^® Data Mart Database from January 2018 to March 2022. Overall average monthly total encounters in each year are reflected by the sum of mean monthly telehealth (abbreviated TH, light gray) and in-person (abbreviated F2F, dark gray) encounters. F2F, in-person; E/M, evaluation and management; TH, telehealth.

Figure 2 displays the average marginal effects of telehealth utilization on the likelihood of a subsequent patient ED visit within six months of the initial encounter by age cohort and aggregated by year from 2018 to 2021. A subsequent ED visit is statistically significantly more likely to occur after a telehealth visit than after an in-person visit among patients below the age of 65 years each year. Patients between 65 and 74 years show a positive, significant association in each year after 2018, and patients above 75 years do not reflect a statistically significant association at the annual level until 2021. We note that 2021 reflects an increase in the likelihood of ED visits with telehealth utilization within six months across all age cohorts in the sample.

FIG. 2.

Average marginal effects of telehealth visit on the probability of ed visit six months from visit (by age group). This figure displays the results of logistic regression analysis of telehealth utilization relative to in-person modality on the likelihood of a subsequent patient ED visit within six months of initial encounter. Results are shown in the form of mean additional ED visits per 1,000 encounters by age cohort and aggregated across months to the annual level from 2018 to 2021. Data uses telehealth and in-person encounters from Optum’s De-identified Clinformatics^® Data Mart Database from January 2018 to September 2021 to observe a full six-month window. Results include 95% confidence intervals for aggregated estimates and asterisks denote where annual averages are statistically significant at the 95% level. Individual bar plots reflect each age group cohort (below 65, 65 to 74, 75 to 84, and 85 and up) with the darkest to lightest shade of gray, respectively. ED, emergency department.

In the typical month in 2020 and 2021, we estimate the magnitude of the association between telehealth use and subsequent ED visit to range between 10 to 20 additional ED visits per 1,000 encounters for patients below 75 years of age. For patients 75 years old and above, we estimate the average magnitude to be approximately 30 additional ED visits per 1,000 encounters in 2021. The average monthly estimated association size in the pre-pandemic period is often much larger across age groups, but confidence intervals frequently overlap with zero, reflecting less frequent and more restricted telehealth use during this time period.

Figure 3 shows trends in rates of six-month ED visits by age cohort and month, displayed as the residual or additional ED visit likelihood per 1,000 encounters with telehealth versus in-person encounters.

FIG. 3.

Six month residual ED visit rate by office/outpatient E/M service visit (TH vs. F2F encounters by age group). This figure displays line graph trends in incidence rates of six-month ED visits by age cohort and month, displayed as the residual or additional ED visit likelihood per 1,000 encounters with telehealth versus in-person encounters. Graph is constructed using the incidence of subsequent ED visits following telehealth and in-person encounters from Optum’s De-identified Clinformatics^® Data Mart Database from January 2018 to September 2021 to observe the full six-month window. Individual line plots reflect each age group cohort (below 65, 65 to 74, 75 to 84, and 85 and up) with the darkest to lightest shade of gray, respectively. F2F, in-person; ED, emergency department; E/M, evaluation and management; TH, telehealth.

Discussion

Telehealth use expanded during the COVID-19 pandemic, but no research to date compares the nature of subsequent health care use following telehealth and in-person visits, marking a missed opportunity to investigate comprehensively the tradeoffs associated with visit modality. We find a positive association between telehealth use and the likelihood of subsequent ED visits among patients using office/outpatient E/M services before and throughout the COVID-19 pandemic. In 2021, we show the association between telehealth use and the likelihood of an ED visit increases by age.

Although additional ED visits indicate undesired outcomes subsequent to an E/M visit, our study does not capture the potential convenience of telehealth, such as the peace of mind of avoided contagion risks. To identify the proper role of telehealth for aging patients, a causal study design and comprehensive modality comparison are needed.

Although this finding is novel to the telehealth literature, available measures in claims data may not perfectly capture the care goals met through telehealth use, especially for aging patients. Beyond the COVID-19 pandemic, more must be done to evaluate the impacts of telehealth use on healthcare quality in all patients.

Limitations

There are limitations to this study. The study data do not provide insight into how the visit modality was decided. Unobserved factors correlated with subsequent ED visits may have played roles in determining modality. Evolving billing guidance and reimbursement reflect the potential for clinician confusion when billing for telehealth services, requiring caution when identifying telehealth through billed claims. The data may not be generalizable outside the universe of Optum’s de-identified Clinformatics^® Data Mart Database. Our data use agreement limits using socioeconomic status variables or geographic variables in our analysis for this study. Finally, beyond the constructed Charlson Comorbidity Indices, the impact of other care received is not captured.

Conclusions

Synchronous audio/video telehealth utilization in office/outpatient E/M service visits rose dramatically during the COVID-19 pandemic. We investigate differences in the likelihood of patient ED visits within six months of telehealth and in-person services by age cohort. Relative to in-person services, telehealth encounters are associated, on average, with higher likelihood of subsequent ED visits. However, the causal impact of telehealth use on the quality of care for aging patients remains unclear.

Data Sharing Statement

The health care claims data used in this article were available through a license negotiated between Indiana University and Optum. Unfortunately, we are not permitted to share the data further. The data access is described at https://www.optum.com/content/dam/optum/resources/productSheets/Clinformatics_for_Data_Mart.pdf.