Data were retrieved from CHS using the Clalit Research Data sharing platform powered by MDClone [72] for the Sharon-Shomron District, which includes populations from representative subregions and according to SES, ethnicity (Arab, Jewish-orthodox, and Jewish-general), and area of residence (central or remote according to geographic information system location data, and urban or rural). Our sample consisted of health records of all insured patients in the Sharon-Shomron District who met the inclusion criteria (“users of telehealth services”) and those who did not meet such criteria (“nonusers”).

Data included five characteristics: (1) sociodemographic characteristics (age, sex, demographic sector, area level SES, residency, and periphery types), (2) number and types of chronic health conditions, overall morbidity burden score (Charlson Comorbidity Index), (3) telehealth services usage (number of telehealth service visits per period for each service type), (4) regular (in-person) family physician services usage (number of visits per period), and (5) telehealth and regular family physician consultation outcomes: related medical diagnosis and care received post-visit (emergency room or hospitalization within 7 days).

The databases integrate electronic health records, insurance claims, demographic information, and geographic data. The database captures all health care encounters, including traditional in-person visits and telehealth consultations, providing comprehensive longitudinal health information for all enrolled members.

The target population comprised all CHS members aged 25 years and older residing in the Sharon-Shomron District. We established a baseline population of adults who maintained continuous membership throughout the entire study observation period.

We designed a 3-period temporal analysis framework to capture telehealth usage patterns in relation to the COVID-19 pandemic:

This framework allows examination of telehealth adoption trajectories, identification of pandemic-specific effects, and assessment of sustained behavioral changes in health care seeking patterns.

For each time period, we calculated (1) overall telehealth usage rates by demographic subgroups, (2) service-specific usage patterns (telephone vs video vs TYTO services), (3) after-hours consultation frequency and timing, and (4) longitudinal changes in usage patterns across the 3 periods.

This study represents an analysis of telehealth usage patterns across 3 distinct time periods rather than longitudinal tracking of individual changes over time. Our design examines usage rates within each period among the eligible population during that specific timeframe, not repeated measures of identical individuals across periods. Each time period (pre-COVID-19, COVID-19, and post-COVID-19) includes all CHS members who met the inclusion criteria during that respective timeframe, regardless of their membership status in other periods. This cross-sectional approach to each time period allows for examination of population-level telehealth adoption patterns and demographic disparities within the context of the COVID-19 pandemic timeline, but does not permit analysis of individual-level behavioral changes or temporal interaction effects that would require following the same cohort longitudinally. The temporal framework serves to understand how telehealth usage evolved across different population segments during distinct phases of the pandemic rather than to track changes in disparity magnitude among identical individuals over time.

In telehealth usage, we focused on the data obtained for “telemedicine users,” that is, those with at least 2 telemedicine visits per year, in contrast to “nonusers”—those with less than 2 telemedicine visits annually. We also examined a binary indicator defined as ≥1 telehealth consultation during each study period for broader usage analysis.

By “use of telehealth services,” we refer to 3 types of services representing different levels of involvement of patients and health care providers mentioned below.

The telehealth service categories are listed below:

In after-hours usage, telehealth consultations occurring outside standard clinic operating hours (evenings, nights, weekends, and holidays), calculated as a percentage of total telehealth visits. Exploring the use of telehealth consultations during after-work hours is crucial for understanding its role in enhancing health care accessibility and equity. Specifically, it is important to examine the usage trends, for example, how the demand for after-work hours telehealth services compares to traditional primary care physician visits during the same timeframe. The examination of the impact on health care outcomes is also of high importance, for example, exploring whether after-work-hours telehealth usage reduces strain on emergency departments or facilitates timely medical interventions.

Age is categorized into 5 groups (25‐35, 35‐50, 50‐60, 60‐70, ≥70 years) based on the age at study initiation, and sex has a binary classification (male and female) based on administrative records. The demographic sector has a four-category classification reflecting Israel’s diverse population:

The SES has a 3-level classification (low, medium, and high) based on Israel Central Bureau of Statistics neighborhood-level socioeconomic indices, incorporating income, education, employment, and housing quality indicators.

Residency type has a six-category classification based on community characteristics:

The periphery is classified into a five-level geographic classification based on distance from major urban centers and service availability:

We used descriptive statistics to describe the sociodemographic and health-related characteristics of the target population, types of telehealth services used, and their frequency and consultation outcomes.

We calculated frequencies and percentages for categorical variables and means with SDs for continuous variables. Telehealth usage rates were computed for each demographic subgroup across all 3 time periods, with 95% CIs calculated using exact binomial methods.

We assessed changes in telehealth usage across the three study periods using:

We have conducted univariate analyses to examine the relationships between sociodemographic and health-related characteristics and the various types of telehealth service use during each study period.

We examined bivariate associations between each sociodemographic factor and telehealth usage using chi-square tests of independence for categorical variables and t tests for continuous variables. Statistical significance was set at P<.05 for all analyses.

We constructed multivariable logistic regression models to assess independent associations between sociodemographic factors and telehealth usage while controlling for potential confounders. The models included all sociodemographic variables and the Charlson Comorbidity Index as predictors.

A multivariable logistic regression model was constructed to assess the independent association of each sociodemographic and health-related factor with the likelihood of telehealth usage. Bootstrapping (1000 resamples) was applied to ensure robust estimates. The adjusted bootstrapped odds ratios (ORs) and their 95% CIs were reported to quantify the strength and direction of associations. The model controlled for confounding by including all predictors simultaneously. Statistical significance was set at P<.05. Model fit was assessed using the Hosmer-Lemeshow goodness-of-fit test. Multicollinearity was examined using the variance inflation factor to ensure no strong correlations between independent variables.

The model specification is classified into three categories:

For robust estimation, we applied bootstrapping with 1000 resamples to obtain robust SEs and 95% CIs, accounting for potential nonnormal distribution of residuals and heteroscedasticity.

The model diagnostics are classified into three categories: (1) Hosmer-Lemeshow goodness-of-fit test to assess model calibration, (2) variance inflation factor analysis to detect multicollinearity (variance inflation factor>10 considered problematic), and (3) residual analysis to identify influential observations.

We conducted specialized analyses of after-hours telehealth usage, including (1) calculation of after-hours usage percentages by demographic subgroups and time periods, (2) chi-square tests to assess differences in after-hours usage patterns, and (3) logistic regression models predicting after-hours versus regular-hours telehealth use.

We assessed patterns of missing data across all variables. Given the administrative nature of the data source, missing data rates were minimal (<1% for most variables). Complete case analysis was performed for the primary analyses, with sensitivity analyses conducted using multiple imputation for variables with >1% missing data.

All analyses were conducted using R version 4.0 or later (R Core Team). Statistical code and analysis protocols are available upon request to ensure reproducibility.

Telehealth usage demonstrated a marked increase during the study period. In the pre-COVID-19 period (March 2019-February 2020), 4% (20,146/499,607) of members used at least 1 telehealth service. This proportion more than doubled to 8.9% (44,678/499,607) during the COVID-19 period (March 2020-February 2021) and further increased slightly to 9.4% (46,868/499,607) in the post-COVID-19 period (March 2021-February 2022).

Telephone services constituted the primary mode of telehealth delivery across all periods, with usage increasing from 4% (20,146/499,607) pre-COVID-19 to 8.7% (43,494/499,607) during COVID-19, and 9.1% (45,225/499,607) post-COVID-19. TYTO online services, while less frequently used, exhibited proportionally greater growth, with usage rates increasing from 0.04% (178/499,607) pre-COVID-19 to 0.4% (1787/499,607) during COVID-19 and 0.5% (2363/499,607) post-COVID-19.

The telehealth service usage in the population is summarized in Table 3.

Significant disparities in telehealth usage were observed across all sociodemographic variables (P<.01 for all comparisons), with these differences persisting across all study periods.

The distribution of telehealth service usage by different sociodemographic factors of the population during each study period is summarized in Table 4.

Women consistently demonstrated higher telehealth adoption rates than men across all periods. These rates are listed as follows:

This gender disparity was statistically significant (P<.01) and persisted throughout all study periods.

A pronounced inverse relationship was observed between age and telehealth usage. The youngest age group (25‐35 years) exhibited the highest usage rates (pre-COVID-19: 8059/102,533, 7.9%; COVID-19: 16,749/102,533, 16.3%; post-COVID-19: 18,333/102,533, 17.9%), while the oldest age group (≥70 years) demonstrated the lowest rates (pre-COVID-19: 1871/72,280, 2.6%; COVID-19: 4137/72,280, 5.7%; post-COVID-19: 4129/72,280, 5.7%; P<.01).

The youngest adults (25-35 years) consistently used telehealth at nearly triple the rate of older adults (70+ years). This pattern persisted across all time periods, with the difference remaining statistically significant (P<.01). Notably, while both groups showed substantial increases in adoption from pre- to post-COVID-19 periods, the absolute gap in usage (between those groups) widened from 5.3 percentage points to 12.2 percentage points.

Substantial disparities in telehealth usage were observed across ethnocultural sectors. The Religious Jewish sector exhibited the highest usage rates (pre-COVID-19: 443/7630, 5.8%; COVID-19: 966/7630, 12.7%; post-COVID-19: 904/7630, 11.8%), while the Arab and Bedouin sector demonstrated markedly lower rates than the general Jewish population (pre-COVID-19: 438/76,895, 0.6%; COVID-19: 1214/76,895, 1.6%; post-COVID-19: 1125/76,895, 1.5%; P<.01).

The nearly 10-fold difference between the highest-adopting sector (Religious Jewish) and the lowest-adopting sector (Arab and Bedouin) represents one of the most pronounced disparities in the study. Each demographic sector category proportion differed significantly from the others at the 0.05 level during each study period, suggesting deep-rooted cultural, linguistic, technological, or structural barriers to telehealth adoption in certain communities.

The data revealed one of the most profound disparities in telehealth adoption across socioeconomic lines: a strong positive association was observed between SES and telehealth usage. Members with high SES used telehealth services at nearly 4 times the rate of those with low SES (pre-COVID-19: 8256/172,011, 4.8% vs 587/56,154, 1.0%; COVID-19: 18,151/172,011, 10.6% vs 1404/56,154, 2.5%; post-COVID-19: 19,064/172,011, 11.1% vs 1328/56,154, 2.4%; P<.01).

This nearly 5-fold difference between high and low SES groups represents a significant digital divide with important health equity implications. While both groups saw increases in adoption during the COVID-19 pandemic, the proportional gap remained relatively unchanged across the periods, suggesting that the pandemic did not meaningfully close the socioeconomic telehealth divide.

Telehealth usage varied significantly by residence type and peripherality. Members residing in large towns demonstrated the highest usage rates (pre-COVID-19: 6761/124,157, 5.4%; COVID-19: 14,524/124,157, 11.7%; post-COVID-19: 14,845/124,157, 12%), while those in non-Jewish settlements exhibited the lowest rates (pre-COVID-19: 371/73,969, 0.5%; COVID-19: 1072/73,969, 1.4%; post-COVID-19: 993/73,969, 1.3%; P<.01).

The data show more than a 10-fold difference in telehealth adoption between large towns and non-Jewish settlements. There were no significant differences in usage between patients residing in a Kibbutz (a collective community in Israel, traditionally based on agriculture, where members share ownership, resources, and responsibilities), a Moshav (small agricultural community with personally owned household farms), or small towns, suggesting that the urban-rural divide may be less pronounced than cultural or community-specific factors.

Notably, a U-shaped relationship was observed with peripherality, with both very peripheral and very central locations demonstrating higher usage rates (pre-COVID-19: 466/10,259, 4.5% to 9698/186,121, 5.2%; COVID-19: 980/10,259, 9.6% to 21,173/186,121, 11.4%; post-COVID-19: 1115/10,259, 10.9% to 21,986/186,121, 11.8%) compared to medium peripheral locations (pre-COVID-19: 3500/186,121, 3%; COVID-19: 7576/118,320, 6.4%; post-COVID-19: 1115/10,259, 6.9%; P<.01) (Figure 1).

There were no significant differences in the percentage of telehealth services usage between patients who reside in locations of peripheral and central types, as well as between patients who reside in locations of peripheral and very peripheral types.

This U-shaped distribution challenges conventional assumptions about telehealth adoption. Both very peripheral and very central locations showed higher adoption rates than medium peripheral areas. This might reflect the different motivations driving telehealth adoption: in very peripheral areas, telehealth may overcome physical distance barriers, while in very central areas, it may cater to tech-savvy populations seeking convenience.

We used a multivariable logistic regression model to identify sociodemographic and health-related predictors of telehealth usage across 3 periods: pre-COVID-19, COVID-19, and post-COVID-19. Telehealth use, defined as at least 1 consultation during the study period, served as the primary binary outcome. Bootstrapping (1000 resamples) was applied for robust estimates, with adjusted odds ratios (ORs) and 95% CIs reported (Table 5). All predictors were included simultaneously to control for confounding, and statistical significance was set at P<.05.

The results reveal notable disparities in telehealth usage across demographic and geographic groups. Controlling for all included covariates, men were 20% less likely to use telehealth services compared to women (OR 0.802, 95% CI 0.790‐0.814).

Age also played a significant role: individuals aged 35‐50 years were 49% less likely to use telehealth than those aged 25‐35 years (OR 0.513, 95% CI 0.504‐0.523), and those aged 50+ years were 71% less likely (OR 0.290, 95% CI 0.283‐0.298), suggesting that younger adults, more tech-savvy and with fewer barriers, are the primary adopters, beyond all other characteristics.

Even after controlling for other personal and community-related factors, ethnic and cultural sectors still exhibited stark contrasts. Individuals from the Arab, Bedouin, and Druze sectors were 82% less likely to use telehealth services (OR 0.183, 95% CI 0.150‐0.284), indicating substantial cultural or systemic barriers. Conversely, members of the Ultra-Orthodox Jewish sector were 21% more likely to use telehealth than the General Jewish population (OR 1.215, 95% CI 1.150‐1.284), possibly due to tailored community solutions or preferences for remote care.

Geography further influenced usage, beyond ethnicity: residents of peripheral (OR 0.824, 95% CI 0.795-0.854), medium peripheral (OR 0.806, 95% CI 0.789-0.824), and even central areas (OR0.816, 95% CI 0.802-0.830) were all significantly less likely to use telehealth compared to those living in very central regions, where access to infrastructure and services is likely better.

A higher Charlson Comorbidity Index is associated with a 2.3% increase in the likelihood of using telehealth services per unit increase in the index (OR 1.023, 95% CI 1.018‐1.028). This highlights the role of telehealth in managing patients with chronic or complex conditions.

People in medium and high SES (OR 1.314, 95% CI 1.254-1.378 and OR 1.353, 95% CI 1.289-1.421, respectively) were significantly more likely to use telehealth services, as compared to people with low SES.

Our analysis reveals significant shifts in after-hours telehealth usage patterns across different time periods and demographic groups, highlighting important trends in health care accessibility.

A notable shift occurred in the temporal distribution of telehealth consultations across the study periods: the percentage of telehealth consultations occurring during after-work hours has steadily decreased from the pre-COVID-19 era through the postpandemic period. Before COVID-19, after-hours telehealth consultations represented 65% (324,744/499,607) of total telehealth usage, indicating that telehealth primarily served as an after-hours alternative to traditional care. During the pandemic, this figure dropped to approximately 55% and further declined to 49% in the post-COVID-19 period.

Notable changes are listed as follows:

This decline suggests a fundamental transformation in telehealth’s role within the health care ecosystem, evolving from a predominantly after-hours convenience to an integrated service available throughout the day.

Our data demonstrates significant expansion in telehealth usage, with usage more than doubling from pre-COVID-19 (4.06%) to post-COVID-19 (9.38%) periods. This dramatic increase demonstrates how the pandemic functioned as a catalyst for telehealth adoption, with the sustained high usage in the post-COVID-19 period suggesting a fundamental shift in health care delivery preferences and behaviors rather than just a temporary response to pandemic restrictions.

Similar patterns were observed by Baum et al [79] and Patel et al [73], who noted dramatic increases followed by plateauing. The predominance of telephone services highlights the importance of familiar, low-technology solutions, aligning with Rodriguez et al [36] findings.

Gender disparities favoring women align with Fischer et al [80] findings across multiple health care systems, potentially reflecting health care-seeking behaviors and caregiving responsibilities [36].

The pronounced age-related digital divide, with the youngest cohorts using telehealth at triple the rate of the oldest groups, reflects well-documented patterns. Haimi et al [81] reported older adults’ concerns about telemedicine quality, while Lam et al [37] found that older adults faced multiple barriers. Roberts and Mehrotra [82] documented that 26.3% of Medicare beneficiaries lacked digital access, particularly affecting older adults and communities of color. This age-related disparity is particularly concerning, given that older adults typically have higher health care needs and could potentially benefit most from the convenience of telehealth services [24,37,83].

The nearly 4-fold socioeconomic difference aligns with Eberly et al [3] research, suggesting possible ongoing access barriers and income-related disparities. Cultural variations, with Religious Jewish communities showing the highest usage and Arab and Bedouin communities showing dramatically lower rates, reflect patterns observed by Rodriguez et al [36] among minority populations and Yoon et al [84] regarding linguistic minorities.

The geographic variations in telehealth usage present an intriguing pattern. The higher rates in large towns compared to non-Jewish settlements align with expected urban-rural divides in technology access. However, the U-shaped relationship with peripherality, where both very peripheral and very central locations demonstrated higher usage than medium peripheral locations, may indicate a more complex dynamic.

This pattern may reflect several things: targeted telehealth promotion in very peripheral areas to address physical access barriers, greater technology adoption and comfort in very central areas, different health care resource allocation patterns across geographic regions, and potential variations in telehealth service design and implementation across localities.

Similar geographic complexities have been documented by Chu et al, who identified that telemedicine adoption increased in rural and remote areas during the COVID-19 pandemic, but its use increased in urban and less rural populations [85]. Drake et al further demonstrated that telehealth adoption in rural communities varied significantly based on the presence of targeted implementation support and provider training [86].

In very peripheral areas, telehealth may address physical access barriers, consistent with findings by Hirko et al that rural patients valued telehealth primarily for reducing travel burden [44]. Conversely, in very central areas (socially advantaged neighborhoods), adoption may be driven by convenience and technological readiness, as suggested by Weiner et al in their analysis of urban telehealth adoption patterns [87].

The lower adoption in medium peripheral areas and non-Jewish settlements may reflect infrastructure limitations or community-specific barriers. It may also reflect the intersection between geography, SES, and culture.

Saeed et al [88] identified reliable internet connectivity as a critical prerequisite for telehealth equity, while Patel et al [73] noted that broadband access varies substantially across communities even within the same geographical region.

The findings found in this study have significant implications for health care policy and service delivery. The overall decline in after-hours telehealth usage suggests successful integration of telehealth into standard care hours, potentially improving health care system efficiency. However, the persistent socioeconomic disparities highlight the ongoing need for targeted interventions to improve health care equity.

The continued higher reliance on after-hours services among lower SES groups underscores the need for targeted interventions to ensure equitable health care access.

In the post-COVID-19 period, the 50‐ to 60-year age group maintained the highest after-hours usage (52%), while the youngest adults (25-35 years) showed the lowest reliance (48%). These patterns suggest that younger populations adapted more readily to regular-hours telehealth services during and after the pandemic, while older adults maintained a slightly higher preference for after-hours consultations, possibly due to greater healthcare needs or established usage patterns.

As health care systems continue to evolve postpandemic, these findings emphasize the importance of maintaining flexible telehealth scheduling options, particularly for vulnerable populations who rely more heavily on after-hours services. Additionally, the observed cultural and religious variations suggest that telehealth implementation strategies should be tailored to meet the specific needs of diverse communities.

The significant shift in after-hours telehealth usage patterns represents a fundamental transformation in telehealth’s function within the health care ecosystem. Prepandemic, telehealth primarily served as an alternative access point outside regular hours, aligning with Mehrotra’s characterization of telehealth as a convenience-oriented service, emphasizing that before the pandemic, telemedicine was mostly used by patients in remote and rural areas of Australia, Canada, and the United States to videoconference with specialists [89].

During and after the pandemic, it evolved into an integrated component of routine health care delivery throughout the day. This transformation reflects what Wosik et al described as the “mainstreaming” of telehealth—its evolution from a niche service to a core health care delivery channel [90]. The normalization of telehealth during regular hours suggests what Dorsey and Topol (2020) called a “virtualist” approach to care, where digital interactions become standard rather than exceptional [1].

The implications of this shift extend beyond mere scheduling flexibility. As Keesara et al argued, the integration of telehealth into routine care represents a catalyst for broader digital transformation in health care delivery models [91]. This mainstreaming may facilitate what Bokolo termed “hybrid care models”—integrated approaches that strategically blend in-person and virtual care based on clinical appropriateness rather than merely emergency conditions [92].

Our findings show that, in general, older individuals, males, and residents of peripheral areas are less likely to use telehealth, highlighting a need for targeted outreach and education to improve adoption in these groups. Significant underutilization by Arab, Bedouin, or Druze populations suggests the need for culturally sensitive strategies to improve telehealth accessibility and trust within these communities. Lower usage rates in peripheral and medium peripheral areas may reflect geographic inequities. Investment in digital infrastructure and incentives for telehealth adoption in these regions could help bridge the gap. Telehealth appears to be an essential tool for patients with chronic conditions, as indicated by its association with the Charlson Comorbidity Index. Expanding remote monitoring capabilities could enhance care delivery for these populations.

As our findings showed, all observed differences remained significant after controlling for all sociodemographic determinants. These findings underscore the importance of an intersectional lens in understanding telehealth usage. Patterns of use are shaped not only by individual characteristics, such as age, gender, or health status, but by their intersections with cultural, geographic, and socioeconomic contexts.

For example, the markedly lower usage among older Arab, Bedouin, or Druze individuals that appears even after controlling for their tendency to reside in peripheral areas and their relatively younger age reflects how multiple, overlapping forms of marginalization (ethnicity, age, and geography) compound barriers to access. This finding supports research showing that digital health equity requires attention to multiple, overlapping social factors [29,93]. Effective strategies must address combined barriers rather than single issues.

Addressing disparities in telehealth adoption thus demands tailored, multidimensional strategies that go beyond single-axis solutions, ensuring culturally sensitive outreach, digital infrastructure development, and inclusive design of services.

Our findings suggest that comprehensive cultural adaptation is essential. Interventions should include Arabic-language telehealth platforms with culturally appropriate interfaces, community health worker programs leveraging trusted local leaders, and gender-specific telehealth services addressing cultural preferences for same-gender providers. Partnership with existing community organizations and religious institutions for outreach, combined with family-centered telehealth models that align with cultural health care decision-making patterns, may address the multiple barriers identified.

Despite high usage, this community demonstrated unique patterns requiring specialized approaches. Expanding “kosher” telehealth devices that meet religious requirements, developing after-hours services accommodating religious observance schedules, and creating rabbinic endorsement programs for telehealth technologies could further optimize access. Privacy-enhanced platforms addressing community concerns about digital exposure should be prioritized.

The triple-rate difference between the youngest and oldest adults requires age-specific approaches that consider cultural context. For older Arab and Bedouin adults, combining cultural mediators with simplified technology training may address both digital literacy and cultural barriers. For younger adults in peripheral areas, leveraging mobile-first platforms and peer education models could optimize adoption. Family-supported telehealth models may be particularly effective for older adults across all communities.

The U-shaped peripherality relationship suggests different intervention needs across geographic contexts. Medium peripheral areas showing the lowest adoption require targeted infrastructure investment combined with community-based digital literacy programs. Low-SES urban areas may benefit from device lending programs with multilingual technical support, while high-SES peripheral areas could use advanced telehealth services leveraging existing technological comfort.