The Relationship Between Broadband Speeds,Device Type,Demographic Characteristics,and Care-Seeking Via Telehealth

Abstract

Background:

This study sought to examine the complex relationship between individual and environmental characteristics, broadband access, device type (computer or smartphone), and telehealth utilization as it relates to the digital divide.

Methods:

We analyzed a combination of electronic health record and publicly available zip code-level data for 2,770 men seeking treatment on a large, nationally available, direct-to-consumer telehealth platform. Using logistic regression, we determined the likelihood of accessing the platform through a smartphone (vs. a computer) based on key features of the environment, including broadband access and income, and demographic characteristics, including age and race.

Results:

We found that living in areas with higher rates of broadband adoption significantly decreased the likelihood of accessing virtual care using a smartphone (odds ratio [OR] = 0.17, p < 0.001). Compared with the 18–29 age category, the odds of accessing virtual care using a smartphone decreased for men between the age categories of 40–59 (OR = 0.63, p < 0.01) and over 60 (OR = 0.29, p < 0.001) years. Belonging to historically marginalized communities of color (Black, Hispanic, and Native American) almost doubled the odds of using a smartphone to access the platform (OR = 1.8, p < 0.001). Broadband availability and median area income were not significantly associated with mobile use.

Conclusions:

Telehealth platform design and policy solutions intended to expand access to virtual care should be flexible enough to accommodate the sometimes competing needs of patients who are at the greatest risk of being left behind.

Introduction

The COVID-19 pandemic spotlighted the numerous benefits of telehealth, from reducing unnecessary exposure to infection to its potential to increase access to care by removing time and distance barriers. However, the digital divide is an area of continuing concern for policymakers and practitioners seeking to ensure that historic health inequities are not reproduced or exacerbated in a post-COVID-19 landscape, where virtual services are playing an increasingly larger role in care delivery.^1
–3

While the pandemic catalyzed swift positive changes at the regulatory level to make telehealth more broadly accessible,⁴ patient-level access barriers remain an unwieldy issue: the divide itself refers to a multifactorial set of disparities that prevent individual telehealth utilization, including access to high-speed internet services, equipment ownership, and digital literacy.^1,5 These elements of the digital divide correlate with pre-existing disparities associated with access to in-person care, threatening to contribute to health inequity despite telehealth's inherent advantages.

Discourse on the digital divide often centers on access to broadband as the most critical component of telehealth utilization disparities.^6

–9 Broadband is defined by the Federal Communications Commission (FCC) as internet connection with download speed of 25 megabits per second (Mbps) and upload speed of 3 Mbps.¹⁰ This can include both wired or fixed internet (DSL or fiber-optic cable) or wireless (satellite, LTE, or 5G cellular).⁹

Broadband access can be understood as an umbrella concept encompassing two distinct hurdles.¹¹ To access broadband, it must be available, meaning that residential high-speed broadband service must be offered by internet service providers and can be purchased in a given location. Many studies on broadband access focus solely on challenges related to availability.¹

In rural America, where proximity and provider shortages already undermine access to in-person health care services, the availability of broadband services is more limited than in areas with denser populations.¹² However, even in locations where broadband is available, it is not necessarily adopted, meaning that individuals might not subscribe to in-home, high-speed internet services or mobile data plans despite the availability of technical infrastructure.¹³

Studies identifying variation in broadband access in urban areas are generally highlighting disparities in adoption: while service providers might operate in those locations, households that have historically experienced broader socioeconomic disadvantages are also less likely to purchase in-home high-speed internet for themselves.¹³ One survey found that 80% of White adults report having a broadband connection at home, compared with 71% of Black and 65% of Hispanic adults.¹⁴ Results from the same study also revealed that cost is the main determining factor of whether or not households subscribe to, or adopt, broadband services.

Another component of the digital divide is access to the necessary devices. Research indicates that disparities exist in device ownership: 80% of White adults report owning a desktop or laptop, compared with only 69% of Black adults and 67% of Hispanic adults. Smartphone ownership, on the other hand, is more equitable across racial and ethnic categories.¹⁵ This phenomenon might be driven by income disparities: in a recent survey, one quarter (26%) of smartphone-only respondents reported annual household incomes of $30,000 or less, which is twice the rate of the U.S. population.¹⁶

The second most common reason for not subscribing to in-home broadband was that ‘smartphones do everything they need.’ This suggests that connecting to high-speed internet using a cellular data plan (or local Wi-Fi hotspots) is a possible workaround for those who cannot afford in-home high-speed internet.

Indeed, some researchers hypothesize that despite evidence to the contrary, broadband access is not a barrier to telehealth utilization. They speculate that most patients have access to internet that—while not high speed—is fast enough, and smartphones do provide a workaround to lack of wired connection.¹⁷ If this is the case, it might be true only under certain conditions: while asynchronous (store-and-forward) telehealth does not require copious amounts of data, other modalities such as video visits require heavy data usage that might be unaffordable on cellular plans.

Nonetheless, policymakers looking to bridge the digital divide are beginning to propose smartphone-oriented solutions that would make care available to an entire population who may not have previously considered telehealth as a viable platform to receive medical care.¹⁸ While this is a sensible approach given certain factors, a hyperfocus on smartphone-driven solutions studies might widen another gap: research shows that older adults who are already less likely to own any kind of device that can be used to access telehealth have significantly lower rates of smartphone adoption compared with computers.¹⁹

One recent review²⁰ examines multiple studies that apply various theoretical frameworks to explain why patients do or do not adopt telehealth. The authors conclude that patient adoption of telehealth is a complex phenomenon that likely reflects a combination of motivation, social pressure and support, and facilitative conditions (environmental characteristics). What many studies fail to consider is that patient-level telehealth adoption is a stepwise process that can hinge on the adoption of other technologies—broadband and devices—while also being subject to other individual and contextual factors that are associated with disparities.

Little is known about the intersection between environmental characteristics, broadband adoption, device adoption, and telehealth utilization. The purpose of this observational study is to shed some preliminary light on what role the additional layers of adoption might play in facilitating telehealth utilization by examining what fixed or contextual characteristics are associated with using a smartphone to access telehealth services.

Specifically, we aim to better understand (1) the extent to which access to wired broadband was a necessary component of telehealth utilization; (2) whether smartphones are a workaround to lack of wired broadband access in seeking virtual care; and (3) what role age, race, and income might play. We draw from data from a national sample of 2,551 men seeking treatment on a direct-to-consumer (DTC), telehealth cash-pay platform.

A better understanding of how the digital divide is or is not reflected in the characteristics of treatment seekers could inform policymakers of key gaps and trends while considering a broad set of solutions to increase equitable access to telehealth.

Materials and Methods

DATA SOURCES

Platform EHR data

We started by extracting electronic health record data for 2,847 men located across the United States seeking treatment for erectile dysfunction (ED), a condition commonly treated on DTC platforms,^21,22 between January 17 and 20, 2020. As part of an initial evaluation, men in the sample engaged in an asynchronous online visit. This involved filling out an adaptive, structured intake form that collects relevant health history, symptom, and demographic information.

In that visit, they were asked to provide their zip code of residence, age, and race (selection options were Asian/Pacific Islander, Black or African American, Hispanic or Latino, I prefer not to answer, Other, Native American, and White, and instructions were to select all that applied). If a patient selected Hispanic or Latino alone or with another race, they were categorized as Hispanic or Latino. If men selected more than one race that did not include Hispanic or Latino, they were categorized as “Other.” These three self-reported variables were extracted from the telehealth platform's EHRs for the purpose of this analysis. A small number of patients (77) selected “I prefer not to answer” and were excluded from analysis; the final number of men included in the sample was 2,770.

The platform itself relies on integrated device specifications to automatically direct someone to a mobile-friendly or desktop version of the online visit for ease of use. This distinction is logged, and we included this as a variable in the dataset. Patients who used iPhone or Android devices were categorized as smartphone users. A small number of patients used an iPad tablet; because of its portability and cellular capabilities, we opted to put these patients in the smartphone category. All other patients were either desktop or laptop users.

If a patient used multiple devices to complete the intake form, we associated them with the first device used; this only occurred in a handful of cases.

Publicly available data

Patient-level relevant EHR fields were then integrated with publicly available data on broadband availability and adoption, income, and geographic categorization by converting patient zip codes into zip code tabulation areas (ZCTAs) and then matching them to the ZCTAs provided in each respective dataset.

Zip codes were converted to ZCTAs using the crosswalk provided by the Housing and Urban Development Office of Policy Development and Research.²³ ZCTA-level variables on broadband access were linked from the National Neighborhood Data Archive (NaNDA).²⁴ Median income was linked from the 2015 American Community Survey estimates.

STATISTICAL ANALYSIS

We dichotomized the device-type variable into computer and mobile variables and employed logistic regression to test the relationship between the technology used to access the platform and socioeconomic characteristics. These variables included a patient's self-reported race, age category, whether they lived in a designated urban or rural area, and ZCTA-level median income. We also included two variables measuring broadband availability and adoption at the ZCTA level from the NaNDA dataset.

We defined broadband availability as whether there was at least one residential, high-speed internet service provider in a ZCTA and adoption as the estimated percent of households within the ZCTA that had in-home high-speed internet. Due to small subgroup numbers, we collapsed the race variable into two groups: one in which non-Hispanic Black men, Hispanic or Latino men, and Native American men were grouped together and the other in which non-Hispanic White, Asian American, and men who selected “Other” were grouped.

We chose this categorization because of copious research that indicates that Black, Hispanic, and Native Americans have poorer health care access and outcomes than White or Asian Americans, both before and during the pandemic.^25
–27

The vast majority (83%) of zip codes included in the sample contained only one observation, but there were instances where men were clustered in the same zip code (these clusters were small, with most of them containing only two individuals and the largest containing seven). To account for any bias introduced by within-zip code correlation, we calculated robust standard errors. We also tested for multicollinearity by examining variance inflation factors for variables like income, broadband adoption, and broadband availability as they are theoretically related, and any correlation among them could bias model estimates.

We ran additional models that included interaction terms to look for possible moderating effects: for example, the relationship between race and device usage might vary across different age categories or different income levels. We assessed whether these models increased fit by examining residual deviance and AIC values, and by comparing likelihood ratio chi-square tests. None of the models with interaction term(s) demonstrated better fit or were significantly different from the initial set of independent variables used to predict device usage; therefore, we included results from the most parsimonious model. This study was approved by the WCG IRB.

Results

The majority of treatment-seeking men whose records were used in this study were non-Hispanic White, lived in urban settings, and used mobile phones to access the platform (Table 1). A plurality of men were between the ages of 45 and 59 years (36.3%). Geographically, almost every single state was represented (47 states plus Washington, D.C.), but most were located in the southern region of the United States (40%). Approximately three-fourths (76.1%) of men used a mobile device to access the platform.

Table 1.

Sample Demographic breakdown

	% (N)
Race/ethnicity
Asian/Pacific Islander	4.4 (122)
Non-Hispanic Black or African American	13.2 (366)
Hispanic or Latino	9.3 (258)
Native American or American Indian	0.9 (24)
Non-Hispanic White	71.1 (1970)
Other/multiple	1.1 (30)
Geography
Rural	13.9 (385)
Midwest	18.9 (525)
Northeast	18.6 (514)
South	40 (1107)
West	22.5 (624)
Age group, years
18–29	11.7 (325)
30–44	25.0 (692)
45–59	36.3 (1006)
60+	27.0 (747)

We then broke down device type and ZCTA-level broadband access and income by race (Table 2). We found that non-Hispanic Black men seeking treatment lived, on average, in ZCTAs with a lower median income ($66,743 compared to $84,700 and $97,928 for non-Hispanic White and Asian men, respectively). Non-Hispanic White men used mobile devices at lower rates compared with others, and non-Hispanic Black, Hispanic or Latino, and Native American men were the most likely to access the platform using a smartphone.

Table 2.

Device Usage and Local Technological and Demographic Context by Race/Ethnicity and Geography

RACE/ETHNICITY	AGE 60+ YEARS (%)	USED A MOBILE DEVICE TO ACCESS THE PLATFORM (%)	RESIDE IN A ZCTA WITH NO RESIDENTIAL HIGH-SPEED INTERNET SERVICE PROVIDERS (%)	MEDIAN ZCTA INCOME	HOUSEHOLDS IN ZCTA THAT ADOPTED (PURCHASED) IN-HOME BROADBAND (%)
Asian/PI	41.9	74.6	50.8	$97928.79	28.3
Non-Hispanic Black/AA	50.1	84.4	64.2	$66743.77	23.7
Hispanic or Latino	42.6	88.0	55.4	$73413.19	23.3
Native American/AI	43.3	95.8	50.0	$73351.64	21.6
Non-Hispanic White	50.7	72.8	57.6	$84700.28	26.7
Other/Multiple	44.1	76.7	46.7	$87014.13	28.9

AA, African American; AI, American Indian; ZCTA, zip code tabulation area.

Results of the logit model (Table 3) showed that living in ZCTAs with higher rates of broadband adoption significantly decreased the likelihood of accessing virtual care using a smartphone (odds ratio [OR] = 0.17, p < 0.001). Compared with the 18–29 age category, the odds of accessing virtual care using a smartphone decreased for men between the age categories of 40–59 (OR = 0.63, p < 0.01) and over 60 (OR = 0.29, p < 0.001) years. Belonging to historically marginalized communities of color (Black, Hispanic, and Native American) almost doubled the odds of using a smartphone to access the platform (OR = 1.8, p < 0.001). Broadband availability and ZCTA-level income were not significantly associated with mobile use.

Table 3.

Results of Logistic Regression Predicting Smartphone Usage

	OR	SE	Z	PR (>\|Z\|)
Intercept	9.25	0.25	8.16	0.00^***
% Households with broadband at the ZCTA level	0.17	0.85	−2.04	0.03^*
No HS BB service providers operating in the ZCTA	1.01	0.10	0.12	0.90
Age group, years
18–29	Referent
30–44	1.0	0.20	0.02	0.99
45–59	0.63	0.19	−2.51	0.01^*
60+	0.29	0.19	−6.63	0.00^***
Race/ethnicity
Non-Hispanic White, Asian, or Other	Referent
Non-Hispanic Black/AA, Hispanic, and Native American/AI	1.8	0.13	−4.57	0.00^***
Median income (ZCTA level)	1.0	0.00	1.10	0.27
ZCTA classified as urban	1.02	0.16	0.11	0.91

***

<0.001; ^** <0.01; and ^*< 0.05.

HS, high-speed; BB, broadband; OR, odds ratio; PR, p-value; SE, standard error.

Table 4 describes the variance inflation factors for each predictor. Variance inflation factor coefficients are all below 10, indicating that multicollinearity was not present.²⁸

Table 4.

Variance Inflation Factors

VARIABLE	VIF	DEGREES OF FREEDOM	GENERALIZED VIF^1/(2⋅DF)
% Households with broadband (ZCTA level)	1.78	1	1.33
Age	1.02	3	1.01
Non-Hispanic Black, Hispanic, and Native American	1.06	1	1.03
Median income (ZCTA level)	1.68	1	1.30
ZCTA classified as urban	1.15	1	1.07

VIF, variance inflation factor.

Discussion

Results showed that living in areas with higher rates of broadband adoption and belonging to a younger age group was associated with accessing virtual care using a smartphone. Belonging to historically marginalized communities of color (Black, Hispanic, and Native American) almost doubled the odds of using a smartphone to access the platform. Broadband availability and median area income were not significantly associated with mobile use.

Findings align with research on smartphone ownership across racial and age categories and corroborate studies that found that Black and Hispanic patients were more likely to use e-health portals with smartphones.²⁹ Our results also support other studies that conclude that due to disparities in computer ownership, modality-neutral virtual services that support both desktop and mobile use might facilitate access to virtual services.³⁰

This study has several limitations. First, telehealth utilization was defined as participating in an asynchronous onboarding process; while this provides insight into the extent to which low-bandwidth telehealth is accessible under certain conditions, it is unclear whether the need for a video call for further evaluation might have led to bandwidth-related access barriers. Second, several of the predictor variables are measured on the ZCTA level and might not accurately reflect individual attributes. Last, results from this study might not be generalizable due to the sampling frame: findings were derived from data from a single DTC telehealth company and were further limited to those seeking treatment for ED, which also meant the sample excluded females.

It is difficult to speculate on whether broadening the population to include women might have altered results. Rates of smartphone ownership are similar across genders, but most population-based studies have identified notable differences in in-person treatment-seeking behaviors across biological sex categories.^31
–33

However, research on gender differences in telehealth utilization is less clear,^34,35 and one study found that women are more likely to have a health-related app on their phone.³⁶ It is also unclear whether broadening the population to include those seeking treatment for other conditions might have produced different results. ED is a condition that confers stigma^37,38 and privacy benefits associated with the portability of smartphones might have driven their adoption for this purpose.

It is unclear if there would be parallels with other stigmatized conditions, such as mental health, which are increasingly being addressed through telehealth. In addition, the sample was created from a group of telehealth users, revealing no concrete insight into those who would benefit from telehealth, but are entirely unable to access it. More research is needed to better understand how smartphone usage, environmental characteristics, and demographics interact in a broader set of circumstances involving telehealth utilization.

Limitations notwithstanding, this analysis provides unique insight; unlike previous studies, the sample is nationally distributed, not confounded by variation in insurance coverage, and not tethered to a specific payor or health system. Several of our findings have implications for policymakers.

First, the majority of men seeking treatment through telehealth live in ZCTAs where there are no wired high-speed internet providers, suggesting that low-bandwidth modalities are accessible in areas where wired high-speed internet is not available and, when appropriate, asynchronous telehealth might be a useful tool for patients without access to high-speed wired internet. Asynchronous care requires less bandwidth than video visits, meaning it can be used with lower-speed wired internet or less cellular data usage; prior research indicates that low-income patients are concerned about mobile health (mHealth) and the cost-related impact on their data plans.³⁹ Patients who reside in states where the definition of telemedicine excludes asynchronous care and/or requires an in-person visit before allowing asynchronous care might benefit from prudent reconsideration of these policies.

Second, high rates of smartphone usage for platform access, and the inverse relationship between broadband adoption and smartphone usage, suggest that infrastructural investment in cellular data expansion might be considered as a parallel path alongside wired broadband expansion. To ensure access, policymakers must also consider subsidies for data plans for low-income patients who would benefit from data-heavy telecare.

Third, in communities where broadband adoption is low, high-speed Wi-Fi hotspots might help patients to access telehealth. Some communities have already done so; for example, in response to the pandemic, Washington State created drive-in Wi-Fi hotspots.⁴⁰ These offer the added benefit of allowing patients to talk with providers through video in the privacy of their vehicles (something that is a challenge in places where free high-speed Wi-Fi is often found, such as public libraries). However, any boon for privacy through Wi-Fi hotspots could easily be undermined by hardware security or gaps in regulating smartphone-based data exchange; broader privacy concerns should be considered along with the shift to smartphone-based health care.⁴¹

Fourth, we also found that patients of color who have historically experienced the greatest access disparities are significantly more likely to use smartphones to access telehealth. Somewhat counterintuitively, this relationship existed after controlling for income, although this might be a result of ZCTA-level data obscuring true observational-level incomes. Conversely, older patients were more likely to use computers even after controlling for race. This corroborates prior research on device adoption: for example, a recent survey of patients residing in Appalachia found high rates of smartphone ownership and cellular and Wi-Fi access, but increasing age was a strong predictor of lack of access to telehealth.⁴²

Policy solutions designed to expand access to telehealth should accommodate the sometimes competing needs of patients who are at the greatest risk of being left behind.

Conclusions

The results of this study underscore the need for additional research, but may still offer practical insights to those responsible for designing and delivering telecare. Menendez et al. emphasize that it is pivotal that telehealth applications prioritize ease of use and maximize compatibility with existing electronic devices, while ensuring that data security and privacy standards are met.⁴³

Similarly, a review of 24 studies on telehealth adoption concludes that easy-to-use applications are of utmost importance. Our findings suggest that the definition of “easy to use” will likely vary depending on patient-level characteristics. As such, it is critical that telehealth platforms continue to invest in and offer technical flexibility. For example, while mHealth apps offer advanced communication and computing features and are especially attractive given the high and more equitable rates of mobile usage,^44,45 older adults who rely on computers might not benefit from them as readily.

More research should be conducted to consider how platform design can facilitate access to those who, unlike the patients in this study, have yet to be able to access any telehealth at all. They are likely to be the most at risk and would incur the greatest benefits from advances in e-health technology.

Footnotes

Disclosure Statement

No competing financial interests exist.

Funding Information

No funding was received for this article.

References

Beaunoyer

, Dupéré

, Guitton

. COVID-19 and digital inequalities: Reciprocal impacts and mitigation strategies. Comput Hum Behav, 2020; 111:106424.

Khilnani

, Schulz

, Robinson

. The COVID-19 pandemic: New concerns and connections between EHealth and Digital Inequalities. J Inf Commun Ethics Soc, 2020; 18:393–403.

Maese

, Seminara

, Shah

, et al. Perspective: What a difference a disaster makes: The telehealth revolution in the age of COVID-19 pandemic. Am J Med Qual, 2020; 35:429–431.

Mehrotra

, Ray

, Brockmeyer

, et al. Rapidly converting to “Virtual Practices”: outpatient care in the era of Covid-19. NEJM Catal Innov Care Deliv, 2020; e1–e5.

Ramsetty

, Adams

. Impact of the digital divide in the age of COVID-19. J Am Med Inform Assoc, 2020; 27:1147–1148.

Bauerly

, McCord

, Hulkower

, et al. Broadband access as a public health issue: The role of law in expanding broadband access and connecting underserved communities for better health outcomes. J Law Med Ethics, 2019; 47(2_suppl):39–42.

Lai

, Widmar

. Revisiting the digital divide in the COVID-19 era. Appl Econ Perspect Policy, 2021; 43:458–464.

Perzynski

, Roach

, Shick

, et al. Patient portals and broadband internet inequality. J Am Med Inform Assoc, 2017; 24:927–932.

Rachfal

, Gilroy

. Broadband internet access and the digital divide: Federal Assistance Programs. Congr Res Serv Rep RL30719. Available at: https://fas.org/sgp/crs/misc/RL30719.pdf/ (last accessed December 19, 2021 ).

10.

Kruger

LG.

Defining broadband: Minimum threshold speeds and broadband policy. Congr Res Serv Rep R45039. Available at: https://sgp.fas.org/crs/misc/R45039.pdf/ (last accessed December 10, 2021 ).

11.

Kolko

. A New Measure of US Residential Broadband Availability. Telecommun Policy, 2010; 34:132–143.

12.

Ali

. The politics of good enough: Rural broadband and policy failure in the United States. Int J Commun, 2020; 14:23.

13.

Reddick

, Enriquez

, Harris

, et al. Determinants of broadband access and affordability: An analysis of a community survey on the Digital Divide. Cities, 2020; 106:102904.

14.

Atske

, Perrin

. Pew Research

Center;

2021. Available at: www.pewresearch.org/fact-tank/2021/07/16/home-broadband-adoptioncomputer-ownership-vary-by-race-ethnicity-in-the-u-s/ (last accessed January 13, 2022 ).

15.

NW. 1615L. St, Washington S 800 and Inquiries D20036U-419-4300|M-857-8562|F-419-4372|M. Demographics of internet and home broadband usage in the United States. n.d. Available at: https://www.pewresearch.org/internet/fact-sheet/internet-broadband/ (last accessed March 23, 2022 ).

16.

Horrigan

Measuring the gap: What's the right approach to exploring why some americans do not subscribe to broadband? National Digital Inclusion Alliance; 2020. Available at: https://www.digitalinclusion.org/measuring-the-gap/

17.

Struminger

, Arora

. Leveraging telehealth to improve health care access in rural America: It takes more than bandwidth. Ann Intern Med, 2019; 171:376–377.

18.

Park

, Erikson

, Han

, et al. Are state telehealth policies associated with the use of telehealth services among underserved populations?. Health Aff (Millwood), 2018; 37:2060–2068.

19.

Choi

, DiNitto

, Lee

, et al. Internet and health information technology use and psychological distress among older adults with self-reported vision impairment: Case-control study. J Med Internet Res, 2020; 22:e17294.

20.

Harst

, Lantzsch

, Scheibe

. Theories predicting end-user acceptance of telemedicine use: Systematic review. J Med Internet Res, 2019; 21:e13117.

21.

Jain

, Lu

, Mehrotra

. Prescriptions on demand: The growth of direct-to-consumer telemedicine companies. JAMA, 2019; 322:925–926.

22.

Jain

, Mehrotra

. Comparison of direct-to-consumer telemedicine visits with primary care visits. JAMA Netw Open, 2020; 3:e2028392–e2028392.

23.

Wilson

, Din

. Understanding and enhancing the US Department of Housing and Urban Development's ZIP Code Crosswalk Files. Cityscape, 2018; 20:277–294.

24.

, Gomez-Lopez

, Khan

, Clarke

, Chenoweth

. National Neighborhood Data Archive (NaNDA): Broadband Internet Access by Census Tract, United States, 2014-2018. Inter-University Consortium for Political Social Research Distributor, 2020.

25.

Orgera

, Artiga

. Disparities in health and health care: Five key questions and answers. Henry J Kaiser Family Foundation, 2018.

26.

Qeadan

, VanSant-Webb

, Tingey

, et al. Racial disparities in COVID-19 outcomes exist despite comparable elixhauser comorbidity indices between Blacks, Hispanics, Native Americans, and Whites. Sci Rep, 2021; 11:1–11.

27.

Williams

, Rucker

. Understanding and addressing racial disparities in health care. Health Care Financ Rev, 2000; 21:75–90.

28.

Thompson

, Kim

, Aloe

, et al. Extracting the variance inflation factor and other multicollinearity diagnostics from typical regression results. Basic Appl Soc Psychol, 2017; 39:81–90.

29.

Chang

, Blondon

, Lyles

, et al. Racial/ethnic variation in devices used to access patient portals. Am J Manag Care, 2018; 24:e1–e8.

30.

Graetz

, Gordon

, Fung

, et al. The digital divide and patient portals. Med Care, 2016; 54:772–779.

31.

Keene

, Li

. Age and gender differences in health service utilization. J Public Health, 2005; 27:74–79.

32.

Manuel

. Racial/ethnic and gender disparities in health care use and access. Health Serv Res, 2018; 53:1407–1429.

33.

, Borders

. Gender, health, and physician visits among adults in the United States. Am J Public Health, 2003; 93:1076–1079.

34.

Stevens

, Mechanic

, Markson

, et al. Telehealth use by age and race at a Single Academic Medical Center during the COVID-19 pandemic: Retrospective cohort study. J Med Internet Res, 2021; 23:e23905.

35.

Zhang

, Shi

, Han

, et al. Disparities in telehealth utilization during the COVID-19 pandemic: Findings from a Nationally Representative Survey in the United States. J Telemed Telecare, 2021; [Epub ahead of print]; DOI:1357633x211051677.

36.

Escoffery

. Gender similarities and differences for E-Health behaviors among US adults. Telemed Ehealth, 2018; 24:335–343.

37.

Chiang

, Yafi

, Dorsey

, et al. The dangers of sexual enhancement supplements and counterfeit drugs to “Treat” Erectile Dysfunction. Transl Androl Urol, 2017; 6:12–19.

38.

Perelman

, Shabsigh

, Seftel

, et al. Attitudes of men with erectile dysfunction: A cross-national survey. J Sex Med, 2005; 2:397–406.

39.

Liu

, Astudillo

, Velez

, et al. Use of mobile health applications in low-income populations: A prospective study of facilitators and barriers. Circ Cardiovasc Qual Outcomes, 2020; 13:e007031.

40.

Anonymous.Washington State Drive-In WiFi Hotspots Location Finder. n.d. Available at https://www.commerce.wa.gov/building-infrastructure/washington-state-drive-in-wifi-hotspots-location-finder/ (last accessed March 23, 2022 ).

41.

Guadarrama

. Mind the gap: Addressing gaps in HIPAA coverage in the mobile health apps industry. Houst Law Rev, 2018; 55:3876.

42.

Madabhushi

, McLouth

, King

, et al. Age and medicare insurance are barriers to telemedicine access—A quality improvement project. Am Surg, 2022. [Epub ahead of print].

43.

Menendez

, Moverman

, Puzzitiello

, et al. The Telehealth Paradox in the neediest patients. J Natl Med Assoc, 2021; 113:351–352.

44.

Bidargaddi

, Almirall

, Murphy

, et al. To prompt or not to prompt? A microrandomized trial of time-varying push notifications to increase proximal engagement with a Mobile Health App. JMIR Mhealth Uhealth, 2018; 6:e10123.

45.

Vaghefi

, Tulu

. The continued use of mobile health apps: Insights from a longitudinal study. JMIR Mhealth Uhealth, 2019; 7:e12983.