Introduction

The COVID-19 pandemic in the United States precipitated a dramatic increase in telehealth use, defined as telecommunication and remote technology to provide health services (Koonin et al., 2020). Studies have indicated that during the COVID-19 pandemic, telehealth use increased between 154% (Koonin et al., 2020) to 4081% nationwide (Whaley et al., 2020), compared to pre-pandemic. The restriction of mobility and limitations in transportation options combined with concerns of getting infected at health facilities (Gupta & Kraschnewski, 2021) motivated telehealth use (Malliaras et al., 2021). In response, the United States government enacted policies to allow for greater adoption of telemedicine, such as regulatory waivers to facilitate telehealth reimbursement through the Medicaid program (“Medicaid Telehealth Policies,” 2020).

Several studies have highlighted the potential for telehealth to help reduce healthcare costs, engage patients, and offer convenience for service recipients (Chunara et al., 2021). However, existing inequities in access to health services have been exacerbated during the COVID-19 pandemic and have been documented in telehealth services (Sachs et al., 2021). Specifically, the in-person health access inequities between Black and Latinx Americans have been identified in telehealth use (Chunara et al., 2021).

Among older adults (60 +), telehealth uptake has generally shown to be lower than those under 60 years old (Choi et al., 2022), despite a high level of interest (Li et al., 2023). Factors contributing to this lower uptake include functional limitations (Kim & Ang, 2023), efficacy of telehealth (Turcotte et al., 2023), and a reluctance to utilize video-telehealth (Li et al., 2023). In a systematic review, telehealth was found to be as effective for older adults as usual care in terms of feasibility, chronic disease management, and patient satisfaction (Sahin et al., 2021).

There is a gap in understanding factors influencing telehealth use among older adults (Pang et al., 2022; Sheng et al., 2023). However, to achieve health equity in the digital health sphere, the reasons for failure to access or barriers to telehealth services must be understood (Crawford & Serhal, 2020). This study aims to address this gap by exploring the usage patterns and determinants of telehealth services, as well as to identify factors associated with telehealth use, among older New Yorkers during the COVID-19 pandemic.

Methods

We conducted a telephone survey using a sample frame of 864,410 landline telephone numbers of community-dwelling older adults aged 70 and older in New York City from December 2020 to March 2021. We selected a stratified random sample of 64,850 telephone numbers. The five boroughs and 18 zip codes with the highest case counts of COVID-19 available in October 2020 made up the six strata for random probability selection.

Calls were made by interviewers who used a computer-assisted telephone interview (CATI) software. To increase likelihood of participation, interviewers made calls during various times of the day and all days of the week. Furthermore, interviewers made at least three attempts to contact a household member at each telephone number. Any adult household member age 70 + was eligible to participate in the interview, and if there were two or more eligible household members, the participant with the most recent birthday was selected to interview. All interviewers asked participants to verbally consent to participation in the interview. The survey included up to 73 questions and took about 30 min to complete.

Interviews were offered in English and Spanish, and using a simultaneous third-party translator, in Mandarin, Cantonese, and Haitian Creole. Respondents received a $10 gift card or a check in the mail if they completed the survey. Further details on survey methods were previously described (Greenleaf et al., 2022).

Measures

Sociodemographic and health information was collected, including age, sex (male or female), borough of residence (Brooklyn, Bronx, Manhattan, Queens, or Staten Island), marital status (married or living together, divorced, widowed, separated or never married), household annual income (< $25,000, ≥ $25,000 to \(\le\)$50,000, > $50,000 to \(\le\) $100,000, or \(\ge\)$100,000), country of birth (US born or born in another country), educational attainment (less than high school, high school diploma/GED, some college/university, or college/university degree), employment status (currently working, not currently working), internet availability (cellular data plan, broadband, satellite, dial-up, no internet), and race and ethnicity (See variables in Table 1). The exposure of interest in this study was race and ethnicity (White, Black, Hispanic/Latinx, Another Race). Respondents were asked first if they identified as Hispanic or Latinx. Respondents who identified as Hispanic/Latinx were then excluded from the following race categories: White, Black, and those of another race. Those of another race included individuals who did not identify as Hispanic, White, or Black. This category included individuals who identified as multi-racial or Asian; the latter a group that made up a very small proportion of our sample. Any difference between race is due largely to societal differences given race represents no biological or cultural differences between groups (Adkins-Jackson et al., 2022). Health status was self-reported on a scale of (excellent, very good, good, fair, and poor). In this analysis, income, internet, and health status were recoded as dichotomous variables. Income was categorized as ($0-$50,000, $50,000 +) and health status was categorized as (“Fair or Poor Health”, “Excellent, Very Good, or Good” health status). Internet was categorized as (“No Internet,” “Has Internet”). Anyone who said yes to the following question was considered a telehealth user: “In the past three months, have you had an appointment with a doctor, nurse, or other health professional by video or by phone?”.

Table 1 Socio-demographic characteristics of study participants (N = 676)
Full size table

Statistical Analysis

We conducted bivariate and multivariable logistic regression with odds ratios (OR) and 95% confidence intervals in SAS version 9.4. Descriptive analyses were weighted to account for survey design, eligibility, nonresponse rates, and post-stratification. We did not weight the logistic regression.

We ran a crude logistic regression model for our main exposure of interest (race and ethnicity) and our outcome of telehealth use to calculate crude odds ratios, 95% confidence intervals, and p-values. We also conducted crude and adjusted logistic regression models individually for each potential confounder (chosen a priori); i.e. age, sex, income, health status and internet access. All significance tests were assessed using a 95% confidence limit.

Effect measure modification was interpreted as an interaction between exposure and outcome that differs between population groups (Lopez et al., 2019). We examined effect measure modification of health status and income in the relationship between race and telehealth usage.

Results

The characteristics of the 676 study participants are summarized in Table 1. Sixty-two percent of participants were between the ages of 70–79 and 38% were 80 years or older. Forty-five percent self-identified as Non-Hispanic White, 21% as Non-Hispanic Black/African American, 20% as Hispanic/Latinx, and 14% as another race. Twenty-nine percent lived in Queens, 28% lived in Brooklyn, 22% lived in Manhattan, 14% lived in Bronx, and 7% lived in Staten Island. Sixty-three percent of participants were female and 37% were males. Thirty-five percent of participants had an income of less than $25 k per year, 29% between $25 k-$50 k per year, 21% between $50 k-$100 k per year, and 15% more than $100 k per year. Fifty-two percent of participants were born in another country and 48% were born in the United States. Fifty-six percent of participants did not complete a telehealth provider appointment by phone or internet while 44% had such an appointment in the three months prior to the survey. Overall, 20% of participants did not have access to internet at home.

In the adjusted multivariable model, we examined the association of socio-demographic characteristics and use of telehealth (Table 2). After adjustment for various sociodemographic factors, use of telehealth was significantly associated with race and ethnicity: Black (adjusted odds ratio: 2.15, CI: 1.33–3.44, P-Value: 0.001), Latinx (OR: 2.27, CI: 1.19–4.27, P-Value: < 0.001), and those of another race (OR: 3.45, CI: 1.67–7.08, P-Value: < 0.001) all had higher odds of telehealth use compared to White participants.

Table 2 Bi-variate and Multivariable logistic regression model assessing relationship between sociodemographic characteristics and telehealth use
Full size table

Individuals with “Fair or Poor” health status had 2.11 times the odds of telehealth usage compared to individuals with “Excellent, Very Good, or Good” health status (CI: 1.48–3.01, P-Value: < 0.001). In addition, individuals with internet access had 1.74 times the odds of telehealth usage compared to individuals who did not have internet (CI: 1.15–2.61, P-Value: < 0.001).

Finally, when examining effect measure modification of health status and income in the relationship between race and telehealth usage, we did not observe statistically significant interaction for the association between health status and telehealth usage (P-Value: 0.92) and income and telehealth usage (P-Value: 0.26) at a significance level of 0.05.

Discussion

While much of the COVID-19 literature has focused on the disproportionate morbidity and mortality in older adults, few studies have examined health-seeking behaviors in this population. In this study, we found that older New York City residents 70 years or older who were Black, Latinx, and those of another race were more likely to have telehealth visits by internet or phone compared to White participants, even when adjusting for health status. Poor health status and internet access was also associated with telehealth use.

Studies have shown that older adults face greater barriers to telehealth use compared to their younger counterparts. For example, in one study, adults 65 years of age or older have decreased telehealth uptake compared to younger adults due to factors such as limited access to internet, technology literacy, and low socio-economic status (Zhai, 2020). In New York City, 42% of adults over the age of 65 in 2017 lacked access to broadband internet, compared to 23% of adults ages 18–64 (Kalicki et al., 2021). In our study, 22% of participants 70 year or older and 34% of those 80 years or older did not have access to the internet (Greenleaf et al., 2022). Furthermore, other factors beyond internet access have been noted to create challenges in use of telehealth by older adults. These include lack of a caregiver to assist with technology, cognitive decline, impairments in vision, dexterity, and refined motor skills (Kalicki et al., 2021). Nonetheless, increasing internet access and fluency could increase telehealth use. Several programs in New York City have worked to bridge the digital divide through educational programs to support older adults in accessing internet and computer technology, which has shown to support overall engagement of older adults within the community (Weil et al., 2021; Finkelstein et al., 2023).

The higher usage of telehealth by Black and Latinx participants in our study may be due to the disproportionate impact of COVID-19 on such communities (Don Bambino Geno Tai et al., 2021). Beyond the impact of COVID-19, communities of color have been noted to be at higher risk for chronic health conditions (Selvin et al., 2014). This may have motivated increased telehealth use among this population, consistent with our finding of higher reported telehealth use among those reporting poorer health status. Future telehealth research should better capture multilevel and multidimensional structural racism, helping better explain differences in use by race and ethnicity (Adkins-Jackson et al., 2022).

The study had several strengths. This included the use of random-number dial sampling of persons 70 years or older residing at home in New York City. We also conducted the survey using multiple languages which allowed for inclusion of a diverse population. Finally, we analyzed our data in a variety of ways including exploratory, interaction, and confounding analyses. The study also has some limitations. In this study, telehealth use was defined as contact with a provider either through internet or phone contact, rather than distinguishing a narrower definition of telehealth, such as telephone-only visits versus video-only visits, as done in other studies (Chunara et al., 2021; Kalicki et al., 2021; Rivera et al., 2021; Smith & Bhardwaj, 2020). Therefore, it is important to note that variations in how researchers define telehealth may affect the findings from the various studies. The racial and ethnic groups are representative of New York City, except that Asian participants were under-represented; an additional limitation. Among New York City residents over 65, 44% identify as Non-Hispanic White, 22% identify as Non-Hispanic Black/African American, and 21% identify as Hispanic/Latinx (New York City Department of Health & Mental Hygiene, 2019). Asian individuals proved to be difficult to reach through landline survey and made up only 3% of our sample compared to the 12% in New York City (New York City Department of Health & Mental Hygiene, 2019). The other demographic characteristics we examined reflected the population over 65 in New York City.

Conclusions

The study findings demonstrate that telehealth use differs by race and ethnicity, health status, and internet availability among older adults in New York City. These determinants of telehealth use provides important insights for increasing telehealth access among older adults. Future research should explore reasons telehealth was used rather than in-person care, accessibility of telehealth for older adults and satisfaction with telehealth.