Background
People living with HIV (PLWH) in the United States are aging and experiencing a greater psychosocial and comorbidity burden compared to the general population [
1‐
5]. To manage HIV and other comorbidities, older PLWH rely upon routine HIV primary and specialty healthcare services. Many of them receive care in Ryan White HIV/AIDS Programs, publicly funded comprehensive systems of HIV primary medical care, and essential support services including medication coverage for low-income patients [
6]. In mid-March 2020, as COVID-19 began to surge in New York State and the “New York Pause” went into effect closing most face-to-face non-essential services [
7], in-person outpatient HIV and specialty healthcare was substantially reduced or closed, and telehealth options rapidly expanded in an effort to maintain care while enabling necessary COVID-19 safety protocols.
Telehealth encompasses a spectrum of activities to deliver care remotely without direct physical contact including through synchronous video and telephone visits, asynchronous messaging, as well as remote monitoring [
8]. Prior to COVID-19, studies of telehealth found it to be a promising means of care delivery [
9‐
12] particularly in rural areas with few HIV providers and long travel distances to receive care [
13]. Yet, broad adoption was sparse due to the reimbursement landscape and interpretation of ‘retention in care’ [
11,
14] despite calls for its growth by some healthcare providers [
15]. Telehealth expansion necessitated by the advent of COVID-19 has increased interest among many providers, administrators, researchers, and payers as a format for safe and effective HIV care delivery [
14,
16,
17]. However, accounts of real-world implementation suggest challenges with uneven digital access [
18], video technology [
19], technology literacy, and making meaningful interpersonal connections [
20]. Concerns have been captured in HIV and other populations requiring primary and specialty chronic care management [
21,
22]. The unique stigma experiences of PLWH [
23], pre-existing health and socioeconomic disparities and complex chronic disease management raise questions around the equity and effectiveness of telehealth as a medium of care for older PLWH. As calls to expand telehealth intensify, it is beneficial to assess early telehealth experiences among older PLWH during COVID-19.
Conceptual framework
We sought to explore telehealth as an equitable means of delivering care to older PLWH drawing upon the constructs of access, use, and quality, which are identified in the literature as critical components to building digital equity [
8,
24,
25] Access is defined as the potential to use healthcare services based on its availability including resource (e.g., device) availability on delivery and recipient sides. ‘Use’ connotes the actual delivery of a telehealth visit, a function of availability plus uptake or ‘demand.’ ‘Demand’ is largely framed in the literature as patients’ abilities with technology. Quality is the measure of an effective visit outcome to improve lives [
8] or produce ‘good effects’ [
25].
To consider questions related to telehealth equity, we developed and conducted a survey inclusive of open-ended questions with PLWH > 50 years receiving HIV care either in a large urban academic hospital-based medical center (AMC) in New York City, or a rural/semi-urban federally qualified health center (FQHC) in the Mid-Hudson Valley of New York State. Each program receives Ryan White funding to provide care to their majority publicly insured patients using a team-based, comprehensive medical and supportive services model. We also aimed to explore common and divergent experiences related to access, use and quality of telehealth across participants. To our knowledge there are no telehealth studies about older PLWH during the COVID-19 period using a multi-dimensional digital equity framework. Nor do we know of studies seeking to discern similarities and variations across a diverse, chronically ill population.
Methods
Program settings
Located in an urban academic hospital, the AMC HIV program is both inpatient and outpatient and provides HIV and primary care along with co-located mental health services and some specialty services. Many older PLWH who receive care at the AMC HIV program access specialty and inpatient care within the same facility. The program serves approximately 2400 patients, 42% of whom are > 50 years. In contrast, the FQHC’s HIV program is spread across several clinical locations in rural and peri-urban communities. Programs co-locate some specialties (mental health, dental, OB/GYN) but others require patients to seek care in the wider system. In two of the FQHC clinics of focus in this study, 240 HIV positive patients are cared for with 60% > 50 years. The AMC and FQHC are similar in that overall, 90% of their patients are virally suppressed. Both programs serve roughly the same percentage of Black patients (~ 40%) but the AMC serves a larger Latinx population (55% AMC compared to 21% FQHC) and a smaller white patient population (10% AMC compared to 39% FQHC).
In mid-March 2020, the HIV programs at the AMC and the FQHC largely transitioned to a telehealth care delivery model, though the AMC continued to see some patients in-person. Many community- and hospital-based specialty services used by PLWH at both sites converted to telehealth or temporarily closed except in emergency circumstances. Starting in July, the AMC and FQHC HIV programs began a fuller re-opening, with increased in-person care over the remainder of 2020.
Sampling & recruitment
We randomly selected and made a list of 40% of each site’s census of HIV positive patients > 50 years of age. A member of the clinical or research team introduced the survey to patients on this list at a regularly scheduled health visit or by telephone. If interested, the patient was consented, and a research staff member administered the survey. At the AMC, out of 400 eligible participants on the list, approximately 170 were contacted, and of those 50 enrolled. At the AMC, out of 58 eligible participants, all were contacted and of those 30 enrolled. Common reasons for non-enrollment were: being unreachable by phone, disinterest, lack of time, and privacy concerns. Participants received a $25 ‘thank you’ for the time they devoted to the study. Approval for the study was granted by Columbia University Medical Center Institutional Review Board, IRB registration #00008612.
Survey design & administration
The survey was developed by the study team based on clinical and content area expertise with some questions drawn from the Community Health Advisory & Information Network survey, a longitudinal study of representative samples of PLWH in New York City and the Tri-County region that began in 1996 [
26] (see Additional File
1 for survey). The survey was administered by one of four trained study staff using Qualtrics, in English (3 staff conducted
N = 63) or Spanish (1 staff conducted
N = 17), in-person (
N = 27) or via telephone (
N = 53), between July and November 2020. Surveys were completed on a laptop or tablet by the study team member and, including the consent process, took between 30 and 60 minutes. Surveys were audio recorded for transcription and coding. English recordings were transcribed by three of the research staff directly into an a priori coding spreadsheet comprised of categories derived from the question areas in the survey (see questions and analysis for more details). Spanish language recordings were translated and transcribed by two of the study team members or a professional service, and content from the transcription was extracted and added to the coding spreadsheet.
Closed and open-ended questions
Participants responded to a total of 44 open- and closed-ended questions (some with sub-questions applied with skip logic as applicable). We collected information in the following areas: 1) Physical and mental health history and current status; 2) COVID-19 infection history; 3) Overall management of HIV and other diseases during COVID-19; 4) Social networks, physical/social distancing and sources of support during COVID-19; and 5) Sociodemographic information and COVID-19 impact on resources. Questions were asked using a multi-select, Likert scale, and write-ins of specific answers (e.g., ‘how old are you today?’). Open-ended questions followed some closed questions to capture descriptions, meaning, and contexts of the standardized responses. Interviewers were encouraged to pick up on and further probe responses to clarify answers and elicit further detail. For example, after learning that a telehealth video or phone visit had occurred, participants were asked in an open manner what it was like to set up the visit, the quality of the interaction, the outcome of the visit, and any additional concerns. After asking about the quality of a telehealth visit using the Likert scale better/same/worse, interviewers would probe, ‘what made the visit better/same/worse?’ and followed up with additional questions based on the responses. The interview ended by asking participants to share ‘anything else important about [their] physical or mental health at the present time.’ Since surveys were recorded, unexpected explanations of close-ended questions to contextualize responses were also spontaneously captured and incorporated into the qualitative analysis.
Analysis
For the quantitative analysis, responses were exported from Qualtrics to Microsoft Excel for preliminary descriptive summaries. R version 4.0.3 was used to conduct analyses. Closed ended questions were dichotomized and their responses aggregated to provide sufficient cell counts. For example, participant’s ease of technological use was determined by asking them to select challenges (if any) with technology. The list of available options included, “I don’t have technology for telehealth”, “I don’t know how to work the technology”, “I didn’t feel comfortable speaking with my provider by video/phone” or “I don’t have any challenges with technology”. To achieve sufficient cell counts, a participant that reported a challenge was categorized as
Yes for difficulty using technology. The two outcomes of interest were use of telephone or video telehealth for HIV or specialty visits, and the reported quality of those visits (better/same/worse) compared to traditional in-person visits. We dichotomized the quality measure by combining better/same due to insufficient cell count and the study team’s specific interest in examining negative health experiences due to the telehealth format. Pearson’s chi-square test for independence was conducted to analyze all bivariate relationships of interest between specific variables with the outcomes. Variables of interest were selected a priori and derived from factors known in the literature to influence telehealth including physical components like age [
27], degree of illness [
28], being from a minority group, level of education, and socioeconomic status [
11,
25]. Fisher’s Exact Test was used when expected cell counts were less than five.
P-values < 0.05 were used to identify statistically significant associations.
The qualitative component used a framework method [
29] in which a priori coding categories mapped to the five main areas of the survey. For example, for descriptions of ‘Overall Management of HIV and Other Diseases during COVID-19,’ coding categories (divided by HIV and other services) consisted of changes to visits, visit experiences, and telehealth comfort. In this process, text that supported these predicted codes also led to additional emergent codes (e.g., telehealth comfort vs technology comfort), iteratively expanding upon and updating the codebook to include relevant concepts. Reliability was assessed by the supervising coder who reviewed a subset of audio recordings/transcripts alongside the coded extractions. Analytic memos produced and discussed by the study team on use and quality of care revealed overall data patterns and outliers of experiences, which were supported by illustrative quotations.
Quantitative and qualitative analyses of the survey data were conducted in parallel using a convergent design in which independently arrived at findings pertaining to telehealth use and quality were then integrated [
30]. Triangulating the two types of data helped in the development of more nuanced findings through the identification of compatibility, complementarity or divergence of findings and subsequently, explaining data patterns. For example, as we will describe, by bringing the data together we developed an explanation for how and why specific characteristic of PLWH > 50 years (e.g., level of formal education) might constrain telehealth use and opinions about quality.
Discussion
Overall, almost three-fourths of study participants had at least one telehealth visit by video or telephone for an HIV or specialty visit, and half had an HIV visit between the March 2020 New York State Pause and the time of the survey. Roughly 40% of these visits occurred by telephone rather than video. Based solely on measuring use, participants overall appeared to be relatively connected to care by either telephone or video during the period, which is consistent with other findings of HIV care in the US context during COVID-19 [
19,
31] Further, a mental health or substance use diagnosis was a significant predictor of a telehealth visit for a non-HIV visit, a finding that may indicate that some participants with these diagnoses in our sample maintained their routine therapy.
Combining quantitative results with qualitative themes provided a deepened and more complex picture of telehealth challenges, including the types of challenges and how they negatively affected telehealth equity. Many of the participants lacked the necessary resources to engage in telehealth, particularly the video delivery format. Receiving care at the FQHC was a significant predictor of whether a visit occurred. We attributed this to the fact that use of video telehealth was harder at the AMC where access and use of the MyChart App proved onerous, compounded by the need for high end communication devices (e.g., smart phones or laptops), consistent functionality and the know-how to navigate the platform. The FQHC’s one-click log-in and use required some video stability but seemed to be operational with less sophisticated device and participant technology savvy. Telehealth support, a feature advocated in the literature [
9], facilitated access and use at both locations, but it could not entirely overcome the significant resource barriers. Moreover, even with support some technology know-how was needed by the participant, as suggested by the individual who received help on a landline while installing the technology on a portable device.
The demanding nature of the App shows how, under certain circumstances, the delivery of digital care is not only potentially underpinned by existing health disparities but those disparities can widen in the digital sphere [
8,
20,
32]. This widening is shaped by barriers in access to and use of devices, connectivity technology, and Apps. Findings further describe how the unpredictable nature of telehealth components exacerbate disparities when existing resources were inconsistently operational, graphically illustrated by video technology that worked in one moment but failed in the next. Participants in this context then needed to engage in attempts - with support from their HIV programs - to both secure the necessary resources (i.e., have/find/borrow a phone or internet terminal) and then mobilize them towards being functionally sufficient for telehealth (i.e., clear out memory, seek out a public WiFi point).
Time and again, participants demonstrated adapting to new telehealth ‘interaction chains’ [
33] to: secure multiple kinds of resources (each with the built-in possibility of failing); mobilize them towards a successful visit (their successful alignment also uncertain); and apply know-how to address routine and emergent complications. Given these conditions, it is not hard to infer how some participants had missed or substandard visits, as much as ones that occurred and felt of reasonable quality. As telehealth expands the potential for access, our findings suggest that it can concomitantly create – or intensify – the makings of a fragile environment for some patients as well as additional work for them to acquire and mobilize necessary resources for routine care.
Findings also point to distinguishing participants’ degree of comfort with technology from comfort with telehealth, both important features of ‘telehealth demand.’ Participants across urban and rural settings put specific language (‘lack of savvy,’ ‘illiterate’) to lacking the skills to navigating technology, particularly as challenges arose. The need for better support, training and coaching to use technology have been widely identified to increase the effective use of telehealth [
8,
19,
33]. The analysis further suggests the value of designing coaching to assist patients with real-time point of use problem-solving. Wootton et al [
34] similarly suggest using ongoing text messaging to accompany a patient’s visit to address challenges arising during the entirety of a visit. Further, our findings point out the separate and important influence of telehealth discomfort due to privacy concerns. This issue has been identified in telehealth services involving PLWH due to HIV disclosure fears [
35], as well as in other ‘remote’ health programs such as mobile clinics where, by moving beyond clinic walls, patients raise concerns about inadvertent disclosure of HIV and other health-related information [
36].
Overall, of those who completed a visit, nearly 70% felt it was worse compared to their usual in-person visits, a departure from several studies prior to [
13,
37] and during [
31,
32] the pandemic that identified patient perceptions of telehealth as equal or better to in-person care. This may be due to our framing telehealth in comparison to in-person care; in so doing and in explaining what was worse, several insights about the features of telehealth emerged. The first and overarching point was the influence of technology not only on access and use, but also on the quality of the visit itself. As others have also noted [
31], technology problems, such as when a video visit defaults to telephone, are frustrating and can undermine the visit’s quality. Second, participants also described in detail how telehealth fostered a different kind of patient-provider interaction. Telehealth visits were focused on verbal communication, whereas in-person visits had sensory and physical elements patients appreciated. This finding supports this idea of in-person care as ‘co-present,’ defined as a place to deal with “patients’ complex and often existential problems” through a body-to-body … intimacy.” In contrast, telehealth is a “therapeutic alliance between health professionals and patients … in a more diffuse social relationship” [
38]. Significantly, in-person ‘co-present’ interactions unfolded organically and triggered patients’ memories, enabling them to bring up different and sometimes unanticipated issues they wished to discuss. The telehealth environment was not as conducive to recalling different issues of concern because it created what has been called a “specific and concentrating” effect [
39]. Paradoxically, it may have hindered providers’ memories or ability to retain details about the patient as well.
In addition to non-remembering, the telehealth effect seems to have extended to self-censoring for some participants, particularly in the therapeutic telehealth session. Patient-side communication behaviors should be factored in when considering which specialties are amenable to telehealth, as the encounter quality will be a function of the provider’s assessment and what the patient is willing to share. Finally, specific outcomes associated with a good visit (receiving referrals and follow up care and coordination) also felt lacking in telehealth care, some of significant concern.
Participants that described the telehealth visit in positive terms (approximately 30%) indicated that it covered the same content as an in-person visit. They tended to have a long-term established relationship with their provider, used telehealth to address a non-urgent medical need, and had the type of visit that primarily relied on non-emotive verbal communication (e.g., informational exchange of lab results). Ease of access by not having to travel (driving, public transportation) to and wait in the clinic contributed to these patients’ positive views of telehealth, a finding consistent with other studies on telehealth that identified its perceived benefits [
37] and with suggestions to apply a typology of factors to determine when telehealth is (and is not) an appropriate delivery medium [
40,
41].
Through this lens where quality is shaped by access and use barriers, creating a different kind of relationship between patient and provider, and feelings that follow-up outcomes are worse, we may further come to understand how the ‘worse’ rating was given significantly more frequently by those with lower levels of formal education. Education may be a proxy for socio-economic status and participants in this subgroup likely had more constrained telehealth resources and mobilizing power. However, in this subgroup, even participants who had the necessary resources and attended a telehealth visit seemed to find it challenging to establish a close relationship with their providers in the technological landscape. This finding calls for further investigation, but one hypothesis is that those with less formal education may be particularly challenged by the complex demands afforded by telehealth to combine cognitive navigation tasks, abstract medical information, and the performance of meaningful communication. Challenges also arose among all seven Spanish speakers who rated telehealth a significantly worse experience. These individuals, some with discordant language relationships with providers, described their usual in-person care as a back and forth dance with providers, a form of interaction less available in technology-based time and space. The Spanish-speaking participant whose physical therapy referral was ordered for the wrong leg suggests that without the visual in-person cues and ability to take time to communicate, it may have been easier for providers to mislabel the complaint side of the body.
In sum, findings suggest that a more nuanced examination of telehealth consequences is needed before we manage “to build a telemedicine model in HIV care that empowers patients … rooted in trust, patient-provider connection, and effective communication” [
18]. Fostering a positive connection with technology itself may have something to do with how it can be better integrated in a clinical visit [
37], as well as finding methods to develop substantial real-time flexibility given the many moving parts we have observed. That said, major structural challenges of access and use must also be overcome while recognizing them for their nuance to truly develop equitable, high quality care. We need to understand how technology can be harnessed towards facilitating access and use in an unstable and variable landscape, while supporting better connection-making.
Limitations
Our data have some limitations. First, the small sample size limited some analyses. We were unable to disaggregate and analyze patterns of access or use between video and telephone users, nor fully explore use and quality by the diversity of specialty services. We believe this is needed to better understand the structural dynamics informing use and how different specialists might build relationships using telehealth. Further, telehealth offerings are occurring against an evolving COVID-19 landscape where telehealth itself is evolving. Our findings will need to be reviewed as telehealth matures, but our study is a benchmark of a particular moment of the pandemic and associated telehealth expansion and can serve as a comparative snapshot as well an evolving analytic around access, use and quality and their interactions. In addition, given our team’s capacities, we did not review the participants’ medical charts. A follow up would benefit from examining associations between healthcare use (telehealth and in-person), perceptions of quality, and documented disease processes and outcomes using chart review. This work is also limited by a lower representation of some participant groups including older transgender participants living with HIV. Further, we only conducted Spanish language surveys at the AMC, limiting our understanding of Spanish-speaking patients in other geographies, and we know from the findings that non-English language groups need attention. Since many interviews were conducted over the telephone, the study design also limits our ability to fully represent the experiences of those with limited telephone access, and comfort to participate in an interview in this format, and by extension may have even higher barriers to telehealth engagement.
Conclusions
Through an equity lens, our study revealed how the rapid introduction of telehealth during the COVID-19 pandemic appears to have disadvantaged some PLWH, particularly individuals constrained by technology resources, difficulty with technology’s unpredictability, more limited know-how, and greater discomfort relating to providers by telehealth. Telehealth expansion and sustaining it should thus be considered carefully using an equity framework to avoid intensifying pre-existing or add emergent disparities in care. Further, equity framework components should be conceptualized not as individual elements but as related and dynamic. Access, use, and quality are interactive, and technology crosscuts them in important ways. The relationship between language and technology also needs to be better understood. Telehealth is a unique and specific practice with signature qualities related to digital possibilities as well as its current limitations. As next steps, a deeper and more nuanced examination of telehealth should be undertaken in diverse primary and specialty care settings to meaningfully broaden the care landscape for older PLWH in patient-centered ways, while not exacerbating hurdles to high quality care. Future broader analyses of all persons with HIV or comparisons by age group might be useful to help determine which telehealth barriers are more pronounced for older age groups. From the practice perspective, clinical members of the study team are currently involved in expanding the delivery of virtual care for older individuals living with HIV through a program that includes home technology assessments, tablet devices to those who need them, and coaching to enhance comfort and competency with technology and telehealth. These interventions recognize the important role telehealth now plays in an evolving world of diverse care delivery modalities, but in keeping with the study findings also recognize that further research and novel programming are needed that ensure its promise while avoiding negative consequences of innovation.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.