Telehealth utilization in U.S. Medicare beneficiaries aged 65 years and older during the COVID-19 pandemic

BMC Public Health

Table 2 Informative variables for predicting telehealth coverage and Internet access from Random Forest analyses

	Coverage of telehealth				Access to the Internet
Variable	Est	SE	P value	OR^§	Est	SE	P value	OR^§	Sig^‡
Age	0.10	0.02	0.000	1.65	0.96	0.10	0.000	2.74	***+++
Race/ethnicity	1.14	0.27	0.000	1.92	2.05	0.44	0.000	3.85	***+++
Metro residence	2.00	0.24	0.000	2.00	-0.04	0.03	0.878	1.70	***
Income	0.10	0.05	0.026	0.49	3.83	0.49	0.000	0.14	*+++
Medicare-Medicaid dual eligibility	0.50	0.53	0.172	0.50	9.78	1.16	0.000	0.13	+++
Positive COVID-19 test	0.25	0.10	0.005	1.31	0.03	0.14	0.411	0.82	**
Able to pay rent/mortgage	0.47	0.21	0.011	1.39	0.24	0.26	0.172	2.13	*
Able to get food	0.99	0.30	0.001	1.58	0.83	0.24	0.000	1.18	**+++
Feel lonely or sad	0.38	0.13	0.002	1.05	0.14	0.07	0.018	1.06	**+
Weak immune system due to health cond	-0.06	0.01	1.000	1.34	0.01	0.01	0.039	0.98	+
Angina pectoris/CHD	-0.01	0.02	0.724	0.98	0.13	0.02	0.000	1.16	+++
Other heart conditions, eg valve/rhythm	0.01	0.01	0.026	0.96	0.04	0.01	0.000	1.02	*+++
High cholesterol	0.12	0.01	0.000	1.04	-0.02	0.01	0.999	0.95	***
Depression	0.05	0.01	0.000	1.08	-0.05	0.01	1.000	0.88	***
Any arthritis	0.06	0.01	0.000	0.83	-0.04	0.01	1.000	0.82	***
Any heart condition	0.01	0.01	0.141	0.89	0.05	0.01	0.000	0.90	+++
Ever used e-cigarette	0.15	0.07	0.016	0.92	0.15	0.06	0.008	1.32	*++
Smoke e-cigarette now	0.61	0.09	0.000	0.72	0.04	0.03	0.152	1.60	***

Est and SE indicate estimation and standard error for Random Forest variable importance (VIMP)
^§OR indicates the survey-weighted odds ratio indicating the direction of effects: if the value is larger than one, the first category of the variable in Table 1 is more likely with a positive outcome than the second category. For example, the odds ratio of age is 1.65, indicating that the 65 to 74 age group was more likely with telehealth coverage than the over-74 age group.
^‡Sig indicates significant level according to P values of VIMP: when the outcome is telehealth coverage, * for \(p\le\) 0.05, ** for \(p\le\) 0.01, *** for \(p\le\) 0.001; when the outcome is Internet access, + for \(p\le\) 0.05, ++ for \(p\le\) 0.01, and +++ for \(p\le\) 0.001

ISSN: 1471-2458