The outcome variable, outpatient telehealth service utilization, was defined based on published literature.7 (link),21 (link),22 (link) The telehealth visits were billable and were defined by outpatient Current Procedural Terminology (CPT)/Healthcare Common Procedural Coding System (HCPCS) procedural codes. The outpatient telehealth visits included virtual check-in encounter codes (G2010, G2012), telephone evaluation and management visit codes (99441, 99443), FQHC distant site telehealth code (G2025), and if the reason for visit was recorded as a telehealth visit.
Individual patient-level factors may be associated with telehealth service utilization and the following baseline patient characteristics were captured at the time of each telehealth visit utilization. Patient race and ethnicity was classified into four categories: (1) Hispanic, (2) Non-Hispanic Black, (3) Non-Hispanic White, and (4) Other race. Other race category predominantly included other race (95%) followed by Asian, American Indian, Native Hawaiian or Pacific Islander, and Vietnamese. Patient age (in years) and gender (male, female) were included. Clinical factors included the presence of a mental disorder (yes/no), and the Charlson comorbidity index (CCI). Mental health disorders included depression, anxiety (ICD-10 codes: F32.3, F33.3, F43.21, F32.9, F41.9, F32.9, F33.9, F43.21, F32.9), bipolar disorder, schizophrenia, or other psychotic disorders (F20.89, F22, F32.3, F33.3, F06.2, F06.0, F30.10, F31.10, F31.30, F31.60, F31.9, F39). The CCI is a longstanding assessment tool of a patient's unique clinical situation that has been found to predict long-term mortality in different clinical populations with excellent reliability, concurrent validity, sensitivity, and predictive validity.23 (link)The neighborhood-level characteristics were assessed at zip code and census track levels by linking patient residence zip codes and addresses with zip code- and census track-level neighborhood factors, respectively. These factors included residence in a low-income area or a HPSA measured at the zip code level. Patient residence ZIP codes were linked to the CMS database of HPSA ZIP codes to identify patients residing in low-income areas or HPSAs.24 These are regions with a lack of primary care providers based on need for care.17 (link) Census track level factors included income defined as percent of population in the census track below 100% of the Federal Poverty Level. The percent population below the federally designated poverty level was classified along the three categories: (1) less than 20%, (2) 20% to 30%, and (3) more than 30%. The other census track level neighborhood-level characteristics that are indicators of socioeconomic status included the percentage of the population who commute more than 30 minutes to a health facility (an indicator of geographic access), percentage of the population who had at least a college education, the percentage who rented a living place, the vacancy rate (i.e., the percentages of all available units in a rental property like an apartment complex that is unoccupied, with higher rates being indicative of lower income in the area), and the percentage of households without internet access. The study was reviewed and approved by the University of Tennessee Health Science Center Institutional Review Board.
Individual patient-level factors may be associated with telehealth service utilization and the following baseline patient characteristics were captured at the time of each telehealth visit utilization. Patient race and ethnicity was classified into four categories: (1) Hispanic, (2) Non-Hispanic Black, (3) Non-Hispanic White, and (4) Other race. Other race category predominantly included other race (95%) followed by Asian, American Indian, Native Hawaiian or Pacific Islander, and Vietnamese. Patient age (in years) and gender (male, female) were included. Clinical factors included the presence of a mental disorder (yes/no), and the Charlson comorbidity index (CCI). Mental health disorders included depression, anxiety (ICD-10 codes: F32.3, F33.3, F43.21, F32.9, F41.9, F32.9, F33.9, F43.21, F32.9), bipolar disorder, schizophrenia, or other psychotic disorders (F20.89, F22, F32.3, F33.3, F06.2, F06.0, F30.10, F31.10, F31.30, F31.60, F31.9, F39). The CCI is a longstanding assessment tool of a patient's unique clinical situation that has been found to predict long-term mortality in different clinical populations with excellent reliability, concurrent validity, sensitivity, and predictive validity.23 (link)The neighborhood-level characteristics were assessed at zip code and census track levels by linking patient residence zip codes and addresses with zip code- and census track-level neighborhood factors, respectively. These factors included residence in a low-income area or a HPSA measured at the zip code level. Patient residence ZIP codes were linked to the CMS database of HPSA ZIP codes to identify patients residing in low-income areas or HPSAs.24 These are regions with a lack of primary care providers based on need for care.17 (link) Census track level factors included income defined as percent of population in the census track below 100% of the Federal Poverty Level. The percent population below the federally designated poverty level was classified along the three categories: (1) less than 20%, (2) 20% to 30%, and (3) more than 30%. The other census track level neighborhood-level characteristics that are indicators of socioeconomic status included the percentage of the population who commute more than 30 minutes to a health facility (an indicator of geographic access), percentage of the population who had at least a college education, the percentage who rented a living place, the vacancy rate (i.e., the percentages of all available units in a rental property like an apartment complex that is unoccupied, with higher rates being indicative of lower income in the area), and the percentage of households without internet access. The study was reviewed and approved by the University of Tennessee Health Science Center Institutional Review Board.
