Comparison of Visit Rates Before vs After Telehealth Expansion Among Patients With Mental Health Diagnoses Treated at Federally Qualified Health Centers

This cohort study assesses visit rates before and after telehealth expansion to assess whether telehealth availability at federally qualified health centers is associated with visit rates for patients with mental health diagnoses.

This supplementary material has been provided by the authors to give readers additional information about their work. eAppendix 1. FQHCs in study sample compared to FQHCs excluded from the study sample   Further, we excluded from the telehealth definition instances where encounter type included: "COVID" or "triage." All encounter types, encounter type descriptions, and charge codes were reviewed by two research team members, our analytic vendor, and by the C3 ACO leadership team. This included manually reviewing and classifying over 5,500 encounter type descriptions. eAppendix 3. Defining high versus low telehealth FQHCs To categorize FQHCs are "high" versus "low" telehealth FQHCs, for each FQHC, we examined the percentage of all FQHC visits that were delivered via telehealth during our post-period (i.e. April 2020 -March 2021). For qualitative validation, we compared these estimates to two external sources of FQHClevel telehealth availability, which weren't specific to mental health: (1) biweekly survey data, reported by every FQHC to HRSA during the COVID-19 pandemic, that reported the percentage of all medical visits delivered via telehealth and (2) a C3 ACO-administrated telehealth assessment, that qualitatively categorized each FQHC's telehealth adoption level. In instances where our study population estimates did not qualitatively align with the other data sources (i.e., in instances where EHR data likely underreported telehealth use, as rates were very low), we considered these EHR classifications invalid and excluded those FQHCs from our sample.
Among FQHCs with valid EHR telehealth data, we examined the distribution of "percentage telehealth visits among all visits" from April 2020 -March 2021. Percentages ranged from 11% to 88%. The 50% threshold was where the two groups naturally diverged: within the study population, all FQHCs had either >53% or <34% of visits delivered via telehealth (i.e. no FQHCs delivered 34-53% of visits via telehealth -it was always higher or lower than this range). In the "high telehealth" group, the percentage ranged from 54-88% of visits (mean=68.0%). In the "low telehealth" group, the percentage ranged from 11-34% of visits (mean=25.7%).

eAppendix 4. Identifying video vs audio-only telehealth visits
Identifying whether telehealth visits are audio only versus video is challenging, as data are often missing in EHRs and the validity of these EHR and claims measures is not well understood. This has recently been described in the literature (Hailu et al, 2022).
Nevertheless, we were able to estimate the distribution of telehealth visits that were audio only vs telehealth. We did this using "encounter type" codes in the EHRs and charge data from FQHC billing systems. For all patient encounters that were otherwise flagged as telehealth, we further classified telehealth visits as "video" vs "audio-only" if any of the following were true: Audio-only visits phon* (e.g., "Phone Preventive Care 30" ; "Phone Visit 20" ; "Psych Phone 20 F/U Visit")

99441-99443
Within our study population, of all FQHC encounters (i.e. visits) that were otherwise flagged as telehealth, 55% were classified as either video or audio-only. The remaining 45% of telehealth visits could not be classified. Of the 55% that were classified, 36% were classified as video and 64% were classified as audioonly. We report these statistics within our manuscript, but do so with the caveat that these classifications are subject to numerous limitations.