Abstract
Video telehealth visits (VTV) have emerged as a critical tool for oncology care delivery, with potential to address longstanding access disparities. We examined the association between broadband internet availability, individual digital literacy factors, and VTV utilization among patients with cancer. In a retrospective cohort of 13,897 patients across a multi-site practice, VTV utilization was significantly lower in areas with ≤1 internet service provider (ISP) offering download speeds ≥25 Mbps (p = 0.0009). Validation in a regional cohort (n = 6665) confirmed lower VTV utilization in low-broadband areas. Among 1134 surveyed patients, higher digital literacy was the strongest predictor of VTV use (OR 2.5; p < 0.001), even where broadband was limited. This study demonstrates that while both broadband availability and digital literacy independently influence VTV utilization, individual digital skills can partially offset structural limitations, underscoring the need for concurrent investment in broadband infrastructure and targeted digital literacy initiatives to advance access to care.
Subject terms: Cancer, Health care, Medical research
Introduction
Telehealth has become a vital tool in healthcare with the COVID-19 pandemic accelerating its adoption across all medical fields, including oncology1–5. Telehealth has the potential to overcome long-known disparities in accessing cancer care for many socially, geographically, and economically disadvantaged populations6. However, despite the promised benefits, disparities in telehealth utilization persist, largely due to variations in broadband internet availability, digital literacy, and socioeconomic factors1,3,5–12.
Broadband availability has been identified as an important determinant of telehealth utilization, particularly among rural populations8,13–16. The Federal Communications Commission (FCC) Broadband Data Collection and National Broadband Map (NBM)17 provides detailed insights into broadband infrastructure and availability across the United States (US), offering an opportunity to evaluate the association of broadband parameters—such as the number of internet service providers (ISPs) and available download speeds—with video visit adoption among patients. However, no studies have previously assessed the potential association of specific broadband parameters with real-world video telehealth visit utilization in cancer patients.
Additionally, recent studies have highlighted that while broadband availability is an important prerequisite for telehealth engagement18, it is insufficient to overcome access disparities, underscoring the need for a more comprehensive understanding of the interplay between digital infrastructure, social determinants of health (SDoH), and individual digital literacy8,13,14. Within oncology care, where continuity and timely access to care are essential, understanding the multifactorial determinants of telehealth utilization is particularly critical. To address these challenges, institutions have developed novel assessment tools to provide insights into the non-infrastructural barriers to digital health engagement14,19,20. For this purpose, we previously developed the Digital Equity Screening Tool (DEST)20. DEST is a 5-item questionnaire used to assess patient digital access and literacy, developed through a community-engagement approach alongside patient stakeholders as a tool to capture patients’ experiences with accessing and interacting with technology for healthcare delivery in a real-world setting.
This study aims to define the relationship between broadband internet availability, demographic (e.g. rurality, sex), SDoH factors, and patient experience accessing and interacting with technology with video telehealth visit utilization among cancer patients. Utilizing the FCC NBM, the primary objective was to identify key broadband parameters associated with video visit use. Additionally, through validation within a defined regional population and assessment of individual experience with digital access and literacy via the DEST questionnaire, a key secondary objective was to evaluate the interplay between digital infrastructure and patient-specific digital determinants of telehealth utilization.
Results
Identifying high-impact broadband internet availability parameters
From July 2020–October 2021, 13,897 unique patients received longitudinal cancer care at our institution. As outlined in Table 1, broadband internet availability parameters that were found to have a statistically significant association with lower video visit utilization included total number of ISPs available (specifically, 0–1 unique ISPs available; parameter estimate = −0.031, p = 0.0009) and maximum download speed available (specifically, <25 Mbps, parameter estimate = −0.873, p = 0.0148).
Table 1.
Multivariable association of selected demographics and broadband internet availability with video visits
| Maximum likelihood parameter estimate | 95% confidence interval | p value | |
|---|---|---|---|
| Age <65 | 0.0176 | −0.008; 0.0433 | 0.1778 |
| Age 65+ | ref | ||
| Female sex | 0.0171 | −0.0001; 0.0343 | 0.051 |
| Male sex | ref | ||
| Race: non-white | 0.006 | −0.0245; 0.0365 | 0.6983 |
| Race: white | ref | ||
| No interpreter services required | 0.0866 | −0.0122; 0.1853 | 0.0858 |
| Interpreter services required | ref | ||
| Non-rural residence | 0.0454 | 0.0264; 0.0644 | <0.0001 |
| Rural residence | ref | ||
| 0–1 unique ISP available | −0.0305 | −0.0484; −0.0126 | 0.0009 |
| 2 or more unique ISPs available | ref | ||
| 25a Mbps download speed (DS) not available | −0.8733 | −1.5757; −0.171 | 0.0148 |
| 25 Mbps DS available | ref | ||
| 100b Mbps DS not available | −0.0257 | −0.0697; 0.0183 | 0.252 |
| 100 Mbps DS available | ref |
Bold values signify those that have achieved statistical significance (as defined by p value <0.05).
RUCA Rural-Urban Commuting Area, FCC Federal Communications Commission, Mbps megabits per second, DS download speed, Ref reference.
aThis speed parameter defines the FCC “Unserved” category, which refers to any location without access to internet which provides the federally required minimum download speed of at least 25 Mbps.
bThis speed parameter defines the FCC “Underserved” category, which refers to any location without access to internet which provides minimum download speed of at least 100 Mbps download speed.
EHR SDoH data—specifically elements related to transportation, financial hardship, and educational level—were also evaluated for potential association with video telehealth visit utilization. The overall rate of patient completion of SDoH questionnaires was not high enough across the cohorts (51.3%; n = 7129) to be incorporated into a formal multivariable analysis. Among SDoH respondents, the vast majority (>98%; n > 6989) reported no concerns for medical transportation, regardless of video visit utilization (data not shown). However, as highlighted in Table 2, patients with higher educational levels were found to be significantly more likely to utilize video visits (32% completed college degree or beyond versus 27% completed high school with some college versus 23% up to high school, p < 0.0001). Additionally, video visit utilization was not associated with increasing levels of financial hardship (p = 0.37).
Table 2.
Financial strain and educational level among cancer patient, shown by proportion of video visit utilization
| Financial resource strain: “How Hard is it for you right now to pay for the very basics like food, housing, medical care, and heating?” | |||
|---|---|---|---|
| Answer: | Patients (n = 7127) | Video visit | Video visit utilization [p = 0.37]a |
| “Not Hard At All” | 4466 | 1275 | 28.5% |
| “Not Very Hard” | 1597 | 470 | 29.4% |
| “Somewhat Hard,” “Hard,” or “Very Hard” | 1064 | 326 | 30.6% |
| Educational Level: “Describe your total years of educational experience.” | |||
| Answer: | Patients (n = 7003) | Video visit | Video visit utilization [p < 0.0001]a |
| Up to high school only | 1430 | 330 | 23.1% |
| Completed high school with some college | 1538 | 418 | 27.2% |
| Completed college degree or beyond | 4035 | 1303 | 32.3% |
aChi-square test was used to compare responses for those subjects who utilized video visits versus those who did not utilize video visits.
Validation in a regional population
Among the cohort of patients in phase 1, 6665 (48%) were identified as residing in the 3-state Upper Midwest region. The residential address was classified as “rural” or “highly rural” by RUCA code for 3988 (59.8%) patients in this subgroup. Within the overall subgroup, 1218 (18.2%) experienced at least 1 video visit from July 2020 to October 2021. A summary of video visit utilization according to total ISP availability and maximum download speed reported per ISP was generated for further analysis (Table 3).
Table 3.
Video visit utilization according to ISP availability and download speed thresholds
| Internet service providers (ISP) available | ISP maximum download speed reported | p value | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| >100 Mbps | >50 Mbps | >25 Mbps | ||||||||
| Patients | At least 1 video visit | % | Patients | At least 1 video visit | % | Patients | At least 1 video visit | % | ||
| 0–1 | 2976 | 534 | 17.9% | 1939 | 341 | 17.6% | 678 | 103 | 15.2% | p = 0.23 |
| 2+ | 3679 | 682 | 18.5% | 4716 | 875 | 18.6% | 5977 | 1113 | 18.6% | p = 0.99 |
| p value | p = 0.53 | p = 0.35 | p = 0.029 | |||||||
Bold values signify those that have achieved statistical significance (as defined by p value <0.05).
Among patients residing at addresses with two or more ISPs available, video visit utilization occurred at a similar rate regardless of maximum download speed reported (18.5% versus 18.6% versus 18.6% for >100 Mbps, >50 Mbps, and >25 Mbps, respectively, with all groups being mutually exclusive). However, for patients with ≤1 ISP available, video visit utilization decreased progressively according to maximum download speed available. In particular, at the lowest maximum download speed threshold of 25 Mbps, video visit utilization was 15.2% among patients with ≤1 ISP available versus 18.6% for those with 2+ ISPs available (p = 0.029). Namely, video visit utilization was found to be significantly lowest in areas where only ≤1 ISP was available to provide download speeds of >25 Mbps. Therefore, this combination was used to define “low” broadband availability for purposes of further analysis and patient selection for phase 3 of the study (Supplementary Table 2).
Assessing the impact of individual digital access and literacy factors
The overall response rate among the 2000 patients mailed a DEST questionnaire was 56.7% (n = 1134), including 535 with low and 599 with high broadband internet availability and 568 who did and 566 who did not utilize video telehealth visits. Demographics and overall results from DEST questionnaire respondents are reported in Table 4. ADI distribution remained well-balanced across respondents from all four groups.
Table 4.
Overall comparison of demographics and results from DEST questionnaire respondents
| A: Low broadband, No video use (n = 255) | B: Low broadband, Yes video use (n = 280) | C: High broadband, No video use (n = 311) | D: High broadband, Yes video use (n = 288) | Total (N = 1134) | p value | |
|---|---|---|---|---|---|---|
| Age at appointment | <0.001a | |||||
| n | 255 | 280 | 311 | 288 | 1134 | |
| Mean (SD) | 67.4 (11.32) | 62.7 (12.62) | 68.1 (12.27) | 62.5 (13.27) | 65.2 (12.67) | |
| Median | 69.0 | 64.0 | 69.0 | 65.0 | 67.0 | |
| Median (IQR) | 69.0 (62.0, 75.0) | 64.0 (55.0, 71.0) | 69.0 (62.0, 77.0) | 65.0 (56.0, 71.0) | 67.0 (58.0, 74.0) | |
| Range | 28.0, 89.0 | 24.0, 90.0 | 24.0, 93.0 | 21.0, 92.0 | 21.0, 93.0 | |
| Gender, n (%) | 0.742b | |||||
| F | 137 (53.7%) | 150 (53.6%) | 179 (57.6%) | 157 (54.5%) | 623 (54.9%) | |
| M | 118 (46.3%) | 130 (46.4%) | 132 (42.4%) | 131 (45.5%) | 511 (45.1%) | |
| Married, n (%) | 0.021b | |||||
| No | 55 (21.7%) | 58 (20.7%) | 93 (30.0%) | 81 (28.3%) | 287 (25.4%) | |
| Yes | 198 (78.3%) | 222 (79.3%) | 217 (70.0%) | 205 (71.7%) | 842 (74.6%) | |
| Missing | 2 | 0 | 1 | 2 | 5 | |
| Race, n (%) | 0.161b | |||||
| Non-white | 1 (0.4%) | 7 (2.5%) | 9 (2.9%) | 8 (2.8%) | 25 (2.2%) | |
| White | 254 (99.6%) | 272 (97.5%) | 300 (97.1%) | 280 (97.2%) | 1106 (97.8%) | |
| Missing | 0 | 1 | 2 | 0 | 3 | |
| Ethnicity, n (%) | 0.783b | |||||
| Hispanic | 2 (0.8%) | 3 (1.1%) | 4 (1.3%) | 5 (1.7%) | 14 (1.2%) | |
| Non-Hispanic | 252 (99.2%) | 274 (98.9%) | 305 (98.7%) | 282 (98.3%) | 1113 (98.8%) | |
| Missing | 1 | 3 | 2 | 1 | 7 | |
| Rural designation (RUCA), n (%) | <0.001b | |||||
| Non-rural | 87 (34.1%) | 83 (29.6%) | 147 (47.3%) | 136 (47.2%) | 453 (39.9%) | |
| Rural | 168 (65.9%) | 197 (70.4%) | 164 (52.7%) | 152 (52.8%) | 681 (60.1%) | |
| Area deprivation index (ADI) quintile, n (%) | 0.998b | |||||
| ADI 1 | 58 (22.7%) | 66 (23.6%) | 77 (24.8%) | 63 (21.9%) | 264 (23.3%) | |
| ADI 2 | 58 (22.7%) | 62 (22.1%) | 66 (21.2%) | 61 (21.2%) | 247 (21.8%) | |
| ADI 3 | 49 (19.2%) | 51 (18.2%) | 59 (19.0%) | 64 (22.2%) | 223 (19.7%) | |
| ADI 4 | 49 (19.2%) | 54 (19.3%) | 59 (19.0%) | 51 (17.7%) | 213 (18.8%) | |
| ADI 5 | 41 (16.1%) | 47 (16.8%) | 50 (16.1%) | 49 (17.0%) | 187 (16.5%) | |
| ADI quintile categories, n (%) | 0.887b | |||||
| ADI 1 or 2 | 116 (45.5%) | 128 (45.7%) | 143 (46.0%) | 124 (43.1%) | 511 (45.1%) | |
| ADI 3 + | 139 (54.5%) | 152 (54.3%) | 168 (54.0%) | 164 (56.9%) | 623 (54.9%) | |
| Total visits | <0.001a | |||||
| n | 255 | 280 | 311 | 288 | 1134 | |
| Mean (SD) | 7.3 (5.90) | 9.9 (7.74) | 7.9 (5.88) | 11.2 (8.33) | 9.1 (7.21) | |
| Median | 5.0 | 7.0 | 6.0 | 8.0 | 7.0 | |
| Median (IQR) | 5.0 (4.0, 9.0) | 7.0 (4.0, 12.5) | 6.0 (4.0, 9.0) | 8.0 (5.0, 14.0) | 7.0 (4.0, 11.0) | |
| Range | 3.0, 46.0 | 3.0, 43.0 | 3.0, 33.0 | 3.0, 42.0 | 3.0, 46.0 | |
| Video visits | <0.001a | |||||
| n | 255 | 280 | 311 | 288 | 1134 | |
| Mean (SD) | 0.0 (0.00) | 1.9 (1.78) | 0.0 (0.00) | 1.9 (1.86) | 1.0 (1.60) | |
| Median | 0.0 | 1.0 | 0.0 | 1.0 | 1.0 | |
| Median (IQR) | 0.0 (0.0, 0.0) | 1.0 (1.0, 2.0) | 0.0 (0.0, 0.0) | 1.0 (1.0, 2.0) | 1.0 (0.0, 1.0) | |
| Range | 0.0, 0.0 | 1.0, 17.0 | 0.0, 0.0 | 1.0, 20.0 | 0.0, 20.0 | |
| Virtual visits (video or phone) | <0.001a | |||||
| n | 255 | 280 | 311 | 288 | 1134 | |
| Mean (SD) | 0.2 (0.64) | 2.2 (1.97) | 0.4 (0.97) | 2.3 (2.03) | 1.3 (1.81) | |
| Median | 0.0 | 2.0 | 0.0 | 2.0 | 1.0 | |
| Median (IQR) | 0.0 (0.0, 0.0) | 2.0 (1.0, 2.0) | 0.0 (0.0, 0.0) | 2.0 (1.0, 3.0) | 1.0 (0.0, 2.0) | |
| Range | 0.0, 4.0 | 1.0, 17.0 | 0.0, 9.0 | 1.0, 21.0 | 0.0, 21.0 | |
| Face to face visits | 0.010a | |||||
| n | 255 | 280 | 311 | 288 | 1134 | |
| Mean (SD) | 7.0 (5.85) | 7.7 (7.57) | 7.5 (5.72) | 8.9 (8.04) | 7.8 (6.90) | |
| Median | 5.0 | 5.0 | 5.0 | 6.0 | 5.0 | |
| Median (IQR) | 5.0 (4.0, 8.0) | 5.0 (2.0, 10.0) | 5.0 (4.0, 9.0) | 6.0 (3.0, 12.0) | 5.0 (3.0, 10.0) | |
| Range | 0.0, 45.0 | 0.0, 41.0 | 1.0, 33.0 | 0.0, 41.0 | 0.0, 45.0 | |
| Site of oncology care, n (%) | <0.001b | |||||
| MCHS NW WI | 61 (23.9%) | 37 (13.2%) | 9 (2.9%) | 7 (2.4%) | 114 (10.1%) | |
| MCHS SW WI | 21 (8.2%) | 25 (8.9%) | 6 (1.9%) | 14 (4.9%) | 66 (5.8%) | |
| MCHS SE MN | 6 (2.4%) | 2 (0.7%) | 22 (7.1%) | 18 (6.3%) | 48 (4.2%) | |
| MCHS SW MN | 14 (5.5%) | 21 (7.5%) | 34 (10.9%) | 35 (12.2%) | 104 (9.2%) | |
| ROCHESTER | 153 (60.0%) | 195 (69.6%) | 240 (77.2%) | 214 (74.3%) | 802 (70.7%) | |
| DEST summary score, n (%) | <0.001b | |||||
| 0 | 9 (3.5%) | 0 (0.0%) | 7 (2.3%) | 5 (1.7%) | 21 (1.9%) | |
| 1 | 6 (2.4%) | 3 (1.1%) | 10 (3.3%) | 2 (0.7%) | 21 (1.9%) | |
| 2 | 30 (11.8%) | 9 (3.3%) | 27 (8.9%) | 9 (3.1%) | 75 (6.7%) | |
| 3 | 28 (11.0%) | 19 (6.9%) | 32 (10.6%) | 16 (5.6%) | 95 (8.5%) | |
| 4 | 78 (30.7%) | 63 (22.9%) | 68 (22.4%) | 66 (23.1%) | 275 (24.6%) | |
| 5 | 103 (40.6%) | 181 (65.8%) | 159 (52.5%) | 188 (65.7%) | 631 (56.4%) | |
| Missing | 1 | 5 | 8 | 2 | 16 | |
| DEST score, n (%) | <0.001b | |||||
| Score 0–4 | 151 (59.4%) | 94 (34.2%) | 144 (47.5%) | 98 (34.3%) | 487 (43.6%) | |
| Score 5 | 103 (40.6%) | 181 (65.8%) | 159 (52.5%) | 188 (65.7%) | 631 (56.4%) | |
| Missing | 1 | 5 | 8 | 2 | 16 | |
| DEST score, n (%) | <0.001b | |||||
| Score 0–2 | 45 (17.7%) | 12 (4.4%) | 44 (14.5%) | 16 (5.6%) | 117 (10.5%) | |
| Score 3–5 | 209 (82.3%) | 263 (95.6%) | 259 (85.5%) | 270 (94.4%) | 1001 (89.5%) | |
| Missing | 1 | 5 | 8 | 2 | 16 | |
| DEST Question 1. Devices use or have access to regularly, n (%) | 0.059b | |||||
| No | 10 (3.9%) | 2 (0.7%) | 10 (3.3%) | 5 (1.7%) | 27 (2.4%) | |
| Yes | 245 (96.1%) | 278 (99.3%) | 297 (96.7%) | 282 (98.3%) | 1102 (97.6%) | |
| Missing | 0 | 0 | 4 | 1 | 5 | |
| DEST Question 1.1: Smart Phone, n (%) | ||||||
| Yes | 215 (84.3%) | 263 (93.9%) | 255 (82.0%) | 263 (91.3%) | 996 (87.8%) | |
| No | 40 (15.7%) | 17 (6.1%) | 56 (18.0%) | 25 (8.7%) | 138 (12.2%) | |
| DEST Question 1.2: Tablet, n (%) | ||||||
| Yes | 97 (38.0%) | 131 (46.8%) | 114 (36.7%) | 152 (52.8%) | 494 (43.6%) | |
| No | 158 (62.0%) | 149 (53.2%) | 197 (63.3%) | 136 (47.2%) | 640 (56.4%) | |
| DEST Question 1.3: Laptop, n (%) | ||||||
| Yes | 122 (47.8%) | 170 (60.7%) | 148 (47.6%) | 183 (63.5%) | 623 (54.9%) | |
| No | 133 (52.2%) | 110 (39.3%) | 163 (52.4%) | 105 (36.5%) | 511 (45.1%) | |
| DEST Question 1.4: Desktop, n (%) | ||||||
| Yes | 76 (29.8%) | 104 (37.1%) | 115 (37.0%) | 107 (37.2%) | 402 (35.4%) | |
| No | 179 (70.2%) | 176 (62.9%) | 196 (63.0%) | 181 (62.8%) | 732 (64.6%) | |
| DEST Question 1.5: Internet Kiosk, n (%) | ||||||
| Yes | 1 (0.4%) | 0 (0.0%) | 7 (2.3%) | 1 (0.3%) | 9 (0.8%) | |
| No | 254 (99.6%) | 280 (100.0%) | 304 (97.7%) | 287 (99.7%) | 1125 (99.2%) | |
| DEST Question 1.6: Gaming Console, n (%) | ||||||
| Yes | 3 (1.2%) | 12 (4.3%) | 5 (1.6%) | 13 (4.5%) | 33 (2.9%) | |
| No | 252 (98.8%) | 268 (95.7%) | 306 (98.4%) | 275 (95.5%) | 1101 (97.1%) | |
| DEST Question 2. Type of internet connection use at home or in public places, n (%) | 0.010b | |||||
| No | 18 (7.1%) | 6 (2.2%) | 19 (6.2%) | 8 (2.8%) | 51 (4.5%) | |
| Yes | 237 (92.9%) | 273 (97.8%) | 287 (93.8%) | 278 (97.2%) | 1075 (95.5%) | |
| Missing | 0 | 1 | 5 | 2 | 8 | |
| DEST Question 3. Comfortable managing your health care online, n (%) | <0.001b | |||||
| Not at all comfortable | 43 (16.9%) | 12 (4.3%) | 44 (14.3%) | 16 (5.6%) | 115 (10.2%) | |
| A little comfortable | 30 (11.8%) | 19 (6.9%) | 31 (10.1%) | 16 (5.6%) | 96 (8.5%) | |
| Somewhat comfortable | 77 (30.3%) | 63 (22.8%) | 71 (23.1%) | 68 (23.6%) | 279 (24.8%) | |
| Very comfortable | 104 (40.9%) | 182 (65.9%) | 162 (52.6%) | 188 (65.3%) | 636 (56.5%) | |
| Missing | 1 | 4 | 3 | 0 | 8 | |
| DEST Question 4. Need help to access health care online, n (%) | 0.010b | |||||
| All of the time | 34 (18.6%) | 32 (15.8%) | 32 (15.4%) | 20 (9.8%) | 118 (14.8%) | |
| Some of the time | 49 (26.8%) | 46 (22.8%) | 45 (21.6%) | 35 (17.2%) | 175 (22.0%) | |
| Need help, but do not have it | 7 (3.8%) | 2 (1.0%) | 9 (4.3%) | 4 (2.0%) | 22 (2.8%) | |
| Do not require any help | 93 (50.8%) | 122 (60.4%) | 122 (58.7%) | 145 (71.1%) | 482 (60.5%) | |
| Missing | 72 | 78 | 103 | 84 | 337 | |
| DEST Question 5.1 It is not in my native language, n | 0 | 0 | 2 | 0 | 2 | |
| DEST Question 5.2 The information is hard to understand, n | 32 | 22 | 29 | 19 | 102 | |
| DEST Question 5.3 I do not know how to find what I need, n | 68 | 45 | 63 | 28 | 204 | |
| DEST Question 5.4 There is too much information, n | 23 | 14 | 20 | 15 | 72 | |
| DEST Question 5.5 I do not have any difficulty, n | 135 | 192 | 191 | 209 | 727 | |
| DEST Question 5.6 Other, n | 46 | 32 | 59 | 33 | 170 |
aANOVA F-test p value.
bChi-square p value.
Two-group comparisons are reported separately in Supplementary Table 1A–C, and multivariable analysis to identify the strongest independent predictors of video visit utilization across respondents in areas with low and high broadband availability, respectively, is provided in Table 5. Younger age and increased visit volume were both significantly associated with video telehealth visit utilization, regardless of broadband internet availability. Additionally, a DEST score of at least 5 was significantly associated with increased video visit utilization overall (Table 4). Upon multivariable analysis (Table 5), video visit utilization was associated with a DEST score of 5 (odds ratio = 2.54; 95% CI: 1.73–3.77; p < 0.001), younger age (OR = 0.98; 95% CI: 0.96–0.99; p = 0.003), and increased visit volume (OR = 1.06; 95% CI: 1.03–1.09; p < 0.001) for cancer patients residing in areas with low, but not high, broadband availability.
Table 5.
Multivariable analysis of demographic and digital access parameters as potential predictors of video visit utilization
| A. Among patients with low broadband availability (0–1 ISPs, >25 Mbps) | ||
|---|---|---|
| Variable | Odds ratio (95% confidence interval) | p value |
| DEST score 5 | 2.544 (1.728–3.767) | <0.001 |
| Age | 0.975 (0.959–0.991) | 0.003 |
| Male sex | 1.15 (0.789–1.68) | 0.468 |
| Married | 0.841 (0.526–1.34) | 0.468 |
| Race (white) | 0.094 (0.003–0.74) | 0.06 |
| Ethnicity (non-Hispanic) | 4.067 (0.315–107.842) | 0.303 |
| Rural residence | 1.468 (0.979–2.21) | 0.064 |
| Total visits | 1.058 (1.028–1.092) | <0.001 |
| Site of care (academic) | 1.356 (0.919–2.004) | 0.125 |
| B. Among patients with high broadband availability (3+ ISPs, >100 Mbps) | ||
| Variable | Odds ratio (95% confidence interval) | p value |
| DEST score 5 | 1.401 (0.967–2.032) | 0.075 |
| Age | 0.967 (0.953–0.981) | <0.001 |
| Male sex | 1.242 (0.867–1.781) | 0.238 |
| Married | 0.952 (0.642–1.41) | 0.807 |
| Race (white) | 1.426 (0.484–4.393) | 0.523 |
| Ethnicity (non-Hispanic) | 0.628 (0.144–2.597) | 0.516 |
| Rural residence | 1.043 (0.73–1.489) | 0.817 |
| Total visits | 1.071 (1.044–1.1) | <0.001 |
| Site of care (academic) | 0.786 (0.518–1.188) | 0.254 |
Bold values signify those that have achieved statistical significance (as defined by p value <0.05).
Responses to DEST question #3 (“How comfortable are you using technology to manage your health care online?”) were highly associated with video telehealth visit utilization regardless of broadband internet availability (Supplementary Tables A1 and 1B), while responses to DEST question #1 (“Which of the following devices do you use or have access to regularly (at least once per month)?”) were only modestly associated with video visit utilization overall. However, though not statistically significant, patients utilizing video telehealth visits did report consistently higher rates of access to all three primary device types listed in subquestions 1.1–1.6 (smart phone, tablet, laptop) as compared with patients not using video telehealth visits, regardless of broadband availability (Table 4).
Patients with lower digital literacy, as defined by answering less than “very comfortable” on question #3, were additionally asked to answer questions about digital assistance (question #4) and potential language barriers (question #5). In assessing digital assistance needs, those reporting that they “do not require any help” with accessing their health care online were significantly more likely to use video telehealth visits (p = 0.010; Table 4), regardless of broadband internet availability. In assessing potential language barriers, barriers due to native language were not identified amongst this cohort (Question 5.1), while patients using video telehealth visits tended to report “no difficulty” (Question 5.5) more often than those not using video telehealth visits.
Discussion
This study provides novel insights into the impact of broadband internet availability, individual digital access, and digital literacy on video visit utilization among patients with cancer, including a subgroup residing in the Upper Midwest. To our knowledge, this study is the first to use the FCC National Broadband Map to define specific broadband internet availability parameters associated with video telehealth visit utilization among cancer patients. These findings have significant implications for both healthcare delivery and policy, particularly in the context of ongoing efforts to expand broadband access through federal and state initiatives.
Since the passage of the Infrastructure Investment and Jobs Act (IIJA) in 2021, which includes substantial financial resources allocated to states to improve broadband access, the definition of “unserved” or “underserved” areas have become instrumental in guiding state lawmakers to prioritize high-impact funding11,21,22. However, these definitions have also evolved over time to reflect increasing technological requirements for various applications, and states have adopted different criteria for federal funding allocation. For example, in 2015, the FCC updated the definition of “broadband internet” as services meeting a minimum threshold of 25 Mbps download speed and 3 Mbps upload speed; consequentially, residences without either of these parameters available, according to the FCC Area Broadband Map at the time this study was conducted, were considered “unserved”23. However, many states’ formal legislative definitions of “broadband programming requirements” vary widely, even among neighboring states. Among the three Upper Midwest states included in the present study, Minnesota has adopted the FCC definition of “unserved” based on these federally defined speed parameters (MN Statute 116J.394), while Iowa uses only the term “underserved,” defining this as “areas with no broadband providers” (IA Code 8B.1). Wisconsin, meanwhile, designates “unserved” as “areas lacking at least one fixed wireless or wireline provider offering actual upload and download speeds of at least 20% of those defined by the FCC” (WI Statute 196.504)11.
This study demonstrates that real-world video visit utilization is significantly impacted by a combination of FCC broadband internet availability parameters (namely, number of available ISPs and sufficient download speed) rather than any individual speed-based metric alone. Specifically, our data show that video telehealth visit utilization is significantly lower in areas where only one or fewer ISPs offer download speeds above 25 Mbps (Table 2). This finding aligns with prior research suggesting that competition among ISPs enhances broadband quality and affordability, leading to improved access24. This finding also has implications for policy-making and infrastructure development in the United States, as it provides a real-world standard by which broadband internet availability may improve access to telehealth services for rural populations.
Prior studies have shown that differences in telehealth access can have a major impact on telehealth utilization25,26. However, our study adds to a growing body of evidence that broadband availability alone does not guarantee equitable telehealth access and utilization1,3,6,13. A particularly important observation from our analysis of DEST questionnaire respondents is that comfort level with digital devices was a stronger predictor of video visit utilization than was access to the devices. Furthermore, high DEST scores were strongly predictive of video visit utilization among patients residing in areas with low broadband availability. This suggests that cancer patients can leverage high individual digital literacy to overcome structural barriers such as low broadband internet availability. Additionally, our study identified several additional factors that significantly impacted video visit utilization among cancer patients. SDoH surveys revealed that educational level was more strongly associated with video visit utilization than financial hardship. Multivariable analysis demonstrated that younger age and higher visit volumes were independent predictors of increased video visit utilization, regardless of broadband availability.
Recently, Tilhou et al. have shown that while high-speed internet availability may improve receipt of care through telehealth utilization, it is insufficient to close utilization and access gaps in the context of primary care13. Our findings in the setting of cancer care additionally reinforce that while broadband availability is an important factor associated with telehealth utilization, improving broadband infrastructure alone likely will be inadequate to ensure equitable access to virtual healthcare for many patients.
There were several strengths of this study, including large cohorts, longitudinal follow-up, regional validation in a population with significant rural representation, use of a novel tool (DEST) developed with input from community stakeholders, and over 1100 questionnaire respondents with equal ADI distribution to mitigate potential impact due to socioeconomic differences. The study also has limitations, including retrospective design, single-institutional experience, subgroup focus on the Upper Midwest (and thus lower representation of racial and ethnic minority patients as compared to the US population overall), and lack of assessment for alternative forms of connectivity to facilitate video visit utilization beyond fixed broadband internet (for example, satellite and cellular-enabled technologies, which have grown significantly in recent years). Additionally, this study was conducted from 2020–2022, and with the continued evolution of digital literacy, connectivity, and access across the US population since then, these results may not fully reflect the current state. Insurance coverage was also not included in our model, which is an important factor that may influence telehealth use.
Since the onset of the COVID-19 pandemic, studies have shown that telehealth remains a fixture of medical and surgical oncology care for many patients and practices across the United States27–29. This study contributes to a growing body of evidence suggesting that a deeper and more nuanced understanding of the complex interplay between broadband infrastructure, SDoH, and personal digital literacy factors will be critical to optimize telehealth engagement and access to care for cancer patients. In response to these findings, future efforts should aim to incorporate digital access and literacy screening into routine cancer care. Tools such as the novel DEST and traditional SDoH questionnaires can help identify patients facing digital barriers and inform tailored interventions. In addition to broadband expansion efforts under the IIJA, multistakeholder collaborations—including healthcare systems, policymakers, and technology developers—are essential to ensuring that telehealth services are accessible to all patients.
Methods
This study was performed in accordance with recognized ethical guidelines, including the Declaration of Helsinki and the U.S. Common Rule, and was approved by the Mayo Clinic institutional review board (IRB) (approvals 22-013109, 22-007797). All patients participating in the study signed informed consent for research authorization for use in data analysis. The study was conducted in three phases. Phases 1–2 were conducted retrospectively and thus deemed exempt from additional informed consent per the institutional ethics committee, as per IRB #22-013109. For study phase 3, additional informed consent was obtained for each participant, according to IRB #22-007797.
Study phase 1: identifying high-impact broadband internet availability parameters
The first phase of the study aimed to determine which broadband internet availability parameters (as reported by the FCC NBM)17 were associated with video telehealth visit utilization by cancer patients at our institution.
Patient selection
We drew upon a previously defined cohort of patients1 who experienced at least one established/return visit within the Mayo Clinic Cancer Practice (MCCP) between July–August 2020. From this cohort, we identified those receiving longitudinal care, defined as three or more visits over the subsequent 16 months (July 2020–October 2021). This frequency was chosen deliberately to select for patients being seen at least every <6 months, on average, over this timeframe. This longitudinal care cohort was developed to represent patients being actively managed at our institution and excluded those being seen for consultation only.
Data collection
The FCC is an independent U.S. government agency regulating radio, television, wire, satellite, and cable communications. The FCC provides detailed information about internet services available at specific locations across the country, as reported by ISPs through the FCC NBM17. We identified information on fixed broadband internet availability for patients by linking their home addresses, recorded in the electronic health records (EHR), with 2020 FCC Broadband Map data. Specifically, we calculated each patient’s geographic latitude and longitude, mapped these coordinates to their corresponding Federal Information Processing Standard (FIPS) codes, and then extracted broadband internet availability parameters (per Supplementary Table 2) to the associated neighborhood block level using the FCC Application Programming Interfaces (APIs). As such, all references to broadband activity in this study pertain to the individual-level.
Key demographic data elements previously shown to impact video visit utilization were collected from the EHR, including patient sex, rural-urban commuting area (RUCA) code, and need for a medical interpreter. We defined rural residence type as RUCA 4-14 and nonrural as RUCA 1-3, as previously described1,30. Additionally, patient-reported SDoH related to transportation, financial hardship, and educational level were abstracted from the EHR.
Analysis
The association of video visit utilization with selected broadband internet availability parameters (Supplementary Table 2) and key demographic elements was assessed using multivariable logistic regression.
Study phase 2: validation in a regional population
The second phase of the study aimed to validate phase 1 findings within a defined geographical region.
Patient selection
Of the original phase 1 cohort, a subgroup was identified as those cancer patients residing in a state within the “Upper Midwest” that is home to a Mayo Clinic site, including the tertiary referral campus in Rochester, Minnesota or one of the Mayo Clinic Health System community-based clinics in Minnesota, Iowa, and Wisconsin. Residential addresses of patients in this subgroup were classified according to RUCA-coded designations30, as previously described for phase 1.
Study phase 3: assessing the impact of individual digital access and literacy factors on telehealth utilization
The third and final phase of the study aimed to determine whether individual digital literacy and access factors were associated with the utilization of video telehealth visits among the Upper Midwest MCCP patient cohort.
Patient selection
Cancer patients from the Upper Midwest cohort (phase 2) were further separated into two categories of broadband availability: low availability (0–1 ISP available to provide >25 Mbps download speed) and high availability (3 + ISP available to provide >100 Mbps download speed). These categories were divided into video visit user and non-user subgroups. This design is summarized in Supplementary Table 3.
To account for potential differences in digital access and literacy due to socioeconomic status, patients in these four categories (A, B, C, D) were equally stratified according to Area Deprivation Index (ADI) quintiles31. ADI is a highly-validated, publicly-available neighborhood-level (exposome) measure that compiles several key elements to generate standardized scores of “disadvantage” to neighborhoods across the United States. The most commonly used social exposome measure within NIH-funded research, ADI levels are known to correlate with poor health outcomes. One hundred patients within each ADI quintile (500 patients per category) were then randomly selected to receive a DEST questionnaire via standard mail. Deceased patients were excluded.
DEST questionnaire and scoring
DEST20 is a 5-item questionnaire used to assess patient digital access and literacy (Supplementary Fig. 1). Total score for DEST can range from 0–5, with three of the five items requiring a response to generate a score. For item 1 (device access), any response except “None of the above” was scored as 1, while “none of the above” was scored as 0. For item 2 (access to the internet), any response except “No access at all” was scored as 1, while “no access at all” was scored as 0. For item 3 (digital literacy), “Very comfortable” was scored as 3, “Somewhat comfortable” was scored as 2, “A little comfortable” was scored as 1, and “Not at all comfortable” was scored as 0. Respondents were not required to answer items 4 and 5 if “very comfortable” digital literacy was indicated by item 3.
DEST comparative and multivariable analyses
Individual survey scores and patient characteristics were summarized and compared across groups using analysis of variance and chi-square tests. Two-group comparisons were also performed using two-sample t-tests and chi-square tests. Of note, with increasing digital literacy trends across the general population, and through our experiences developing the DEST tool in other community-engaged settings, we hypothesized that DEST scores of 0–1 would be rare. Therefore, two additional variables were created for comparative analysis of DEST scores: 0–4 versus 5 (indicating the highest level of digital literacy), and 0–2 versus 3–5 (indicating moderate digital literacy). The ADI variable, values 1–5 and 1–2 versus 3+, is also summarized. Multivariable logistic regression with select demographic and digital access parameters, including DEST scores, was conducted to assess the association of these parameters with video visit utilization. In all cases, p values < 0.05 were considered statistically significant.
Supplementary information
Acknowledgements
Funding/Support: Noaber Foundation Digital Health Award (J.C.P., P.S., T.C.H., C.A.P.); Wohlers Family Foundation Grant (T.C.H., J.C.P.); Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery, through the Kern Scholar Program (J.C.P.). Role of the Funder/Sponsor: the funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Author contributions
J.C.P. and T.C.H. had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design: J.C.P. and P.S. Drafting of the manuscript: J.C.P., P.S., C.A.P., and T.C.H. Statistical analysis: M.H., R.D., P.A.D., R.J., and B.J.B. Obtained funding: J.C.P., P.S., C.A.P., and T.C.H. Acquisition, analysis, or interpretation of data and critical review of the manuscript for important intellectual content: J.C.P., P.S., M.H., R.D., T.A.B., J.P.M., C.C.K., H.A., P.A.D., R.J., J.T., N.K., L.C.B., J.C.T., B.J.B., C.A.P., and T.C.H.
Data availability
The datasets generated and/or analyzed during the current study are not publicly available due to patient privacy and confidentiality considerations (for example, geographic residences of participants), but are available from the corresponding author on reasonable request.
Code availability
Code developed for this study is not publicly available but may be shared upon reasonable request at the discretion of the study authors.
Competing interests
B.J.B. discloses consulting for Boehringer-Ingelheim on unrelated health economics and outcomes research projects. T.C.H. discloses grant funding to Mayo Clinic from Takeda Oncology and Puma Biotechnology, unrelated to this project. All other authors declare no financial or non-financial competing interests.
Footnotes
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
The online version contains supplementary material available at 10.1038/s41746-026-02397-9.
References
- 1.Pritchett, J. C. et al. Patient- and provider-level factors associated with telehealth utilization across a multisite, multiregional cancer practice from 2019 to 2021. JCO Oncol. Pract.19, 750–758 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Zon, R. T. et al. Telehealth in oncology: ASCO standards and practice recommendations. JCO Oncol. Pract.17, 546–564 (2021). [DOI] [PubMed] [Google Scholar]
- 3.Pritchett, J. C., Patt, D., Thanarajasingam, G., Schuster, A. & Snyder, C. Patient-Reported Outcomes, Digital Health, and the Quest to Improve Health Equity. Am. Soc. Clin. Oncol. Educ. Book43, e390678 (2023). [DOI] [PubMed]
- 4.Doshi, S. D. et al. Perspectives on telemedicine visits reported by patients with cancer. JAMA Netw. Open7, e2445363 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Boucher, A. A. et al. Adult hematology/oncology patient perspectives on telemedicine highlight areas of focus for future hybrid care models. Telemed. J. E Health29, 708–716 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Haemmerle, R., Paludo, J., Haddad, T. C. & Pritchett, J. C. The growing role of digital health tools in the care of patients with cancer: current use, future opportunities, and barriers to effective implementation. Curr. Oncol. Rep.26, 593–600 (2024). [DOI] [PubMed] [Google Scholar]
- 7.Wood, B. R. et al. Advancing digital health equity: a policy paper of the Infectious Diseases Society of America and the HIV Medicine Association. Clin. Infect. Dis.72, 913–919 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Morris, B. B., Rossi, B. & Fuemmeler, B. The role of digital health technology in rural cancer care delivery: a systematic review. J. Rural Health38, 493–511 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Pandit, A. A. et al. Association between broadband capacity and telehealth utilization among Medicare Fee-for-service beneficiaries during the COVID-19 pandemic. J. Telemed. Telecare31, 41–48 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Kelpin, S. et al. Increasing digital equity to promote online smoking cessation program engagement among rural adults: a randomized controlled pilot trial. Commun. Med. 4, 194 (2024). [DOI] [PMC free article] [PubMed]
- 11.Pew Research Center. Mobile Fact Sheet. Tech Adoption Trends. https://www.pewresearch.org/internet/fact-sheet/mobile/ (2024).
- 12.Haynes, N. et al. “Can you see my screen?” Addressing racial and ethnic disparities in telehealth. Curr. Cardiovasc. Risk Rep.15, 23 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Tilhou, A. S., Jain, A. & DeLeire, T. Telehealth expansion, internet speed, and primary care access before and during COVID-19. JAMA Netw. Open7, e2347686 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Sharma, P. & Patten, C. A. A need for digitally inclusive health care service in the United States: recommendations for clinicians and health care systems. Perm. J.26, 149–153 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Richardson, S., Lawrence, K., Schoenthaler, A. M. & Mann, D. A framework for digital health equity. NPJ Digital Med.5, 119 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Adedinsewo, D. et al. Health disparities, clinical trials, and the digital divide. Mayo Clin. Proc.98, 1875–1887 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.FCC. Federal Communications Commission. FCC National Broadband Map. https://broadbandmap.fcc.gov/home (2025).
- 18.Acharya, M., Shoults, C. C., Hayes, C. J. & Brown, C. C. Association between broadband capacity and social vulnerability factors in the United States: a county-level spatial analysis. Popul. Health Manag.25, 798–806 (2022). [DOI] [PubMed] [Google Scholar]
- 19.Nelson, L. A., Pennings, J. S., Sommer, E. C., Popescu, F. & Barkin, S. L. A 3-item measure of digital health care literacy: development and validation study. JMIR Form. Res.6, e36043 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Sharma, P. et al. Development and initial cognitive testing of the Digital Equity Screening Tool (DEST): community participatory approach to assessing digital inequality. J. Clin. Transl. Sci.6, e117 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.GFOA. Government Finance Officers Association. Infrastructure Investment and Jobs Act (IIJA) Implementation Resources. https://www.gfoa.org/the-infrastructure-investment-and-jobs-act-iija-was (2021).
- 22.Landrieu, M. President Joe Biden Building a Better America. Build.gov. A Guidebook to the Bipartisan Infrastructure Law for State, Local, Tribal, and Territorial Governments, and Other Partners. https://gfoaorg.cdn.prismic.io/gfoaorg/0727aa5a-308f-4ef0-addf-140fd43acfb5_BUILDING-A-BETTER-AMERICA-V2.pdf (2022).
- 23.Cooper, T. Broadbandnow. Broadbandnow Research. How the government defines broadband: analysis and history. https://broadbandnow.com/report/fcc-broadband-definition (2018).
- 24.Bauerly, B. C., McCord, R. F., Hulkower, R. & Pepin, D. 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. Ethics47, 39–42 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Ko, J. S. et al. Disparities in telehealth access, not willingness to use services, likely explain rural telehealth disparities. J. Rural Health39, 617–624 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Shih, K. K. et al. Telehealth preferences among patients with advanced cancer in the post COVID-19 vaccine era. J. Pain. Symptom Manag.67, 525–534.e1 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Parameswaran, V. et al. Telemedicine trends in ambulatory surgical oncology: a five-year analysis of visit volume and utilization at a high-volume academic center. Ann. Surg. Oncol.32, 6829–6837 (2025). [DOI] [PubMed] [Google Scholar]
- 28.Parameswaran, V. et al. Impact of telemedicine on access to care in surgical oncology clinics: distance and time to visit analysis for new patients at a multispecialty cancer center. J. Clin. Oncol.43, e23345–e23345 (2025). [Google Scholar]
- 29.Parameswaran, V. et al. Comparing trends in telehealth utilization by medical oncology practices at Stanford Medicine and Mayo Clinic from 2019–2023. J. Clin. Oncol.43, e13890–e13890 (2025). [Google Scholar]
- 30.GSA. General Services Administration. data.gov Data Catalog. Rural-Urban Commuting Area Codes. https://catalog.data.gov/dataset/rural-urban-commuting-area-codes (2024).
- 31.Powell, W., Sheehy, A. & Kind, A. J. H. Scientifically validated social exposome tool available for policies advancing health equity. https://www.healthaffairs.org/content/forefront/area-deprivation-index-most-scientifically-validated-social-exposome-tool-available. Health Affairs Forefront. (Accountable Care for Population Health) 10.1377/forefront.20230714.676093 (2023).
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets generated and/or analyzed during the current study are not publicly available due to patient privacy and confidentiality considerations (for example, geographic residences of participants), but are available from the corresponding author on reasonable request.
Code developed for this study is not publicly available but may be shared upon reasonable request at the discretion of the study authors.