Facilitators and Barriers to the Adoption of Telemedicine During the First Year of COVID-19: Systematic Review

Introduction

Rationale

The virulent nature of COVID-19 forced social distancing and a decrease of in-person visits to clinics around the world. Telemedicine presented health care providers with solutions that enabled a social-distancing window into the clinical environment and a continuation of the doctor-patient relationship.

Telemedicine is defined by the World Health Organization as healing from a distance through information communications technologies by all health care professionals for the “exchange of valid information for diagnosis, treatment and prevention of disease and injuries, research and evaluation” [1]. Telemedicine is not a perfect means of patient care; however, it offers great advantages to overcome geographical barriers to improve health outcomes [1]. Validated and peer-reviewed international statistics are elusive on adoption figures, but a recent question-and-answer session indicates overall low adoption of telemedicine internationally [2]. In the United States, prior to the pandemic, telemedicine had only been adopted by 8% of providers [3]. Providers have recognized wide acceptance of telemedicine by patients; however, prior to the desperate circumstances of COVID-19, they had not been willing to adopt telemedicine on a wide scale [4]. The largest challenges to the adoption of telemedicine were identified as technically challenged staff, resistance to change, cost, reimbursement, and education level of the patient [5]. Telemedicine saves patients time, consultation fees, and travel expenses [6]. However, telemedicine requires users at both ends to possess certain levels of technological skills such as those required to enable video teleconferencing [7]. Fortunately, some countries enacted legislation to expand the adoption of telemedicine. For example, in the United States, telemedicine was not easily reimbursed by federal programs until the Coronavirus Aid, Relief, and Economic Security (CARES) Act legislation [8], which greatly increased reimbursement mechanisms for the telemedicine modality. This change in reimbursement structure should not be ignored, and it most likely provided a significant catalyst to the adoption of telemedicine.

A large number of articles were published in the first 12 months of the pandemic (February 2020 to February 2021) on the rapid implementation efforts of telemedicine to enable clinics and hospitals to continue to see patients and care for their needs [9,10]. However, providers acknowledge some of the shortfalls inherent to this modality, such as lack of technical infrastructure, cost, lack of technical staff, computer literacy of both staff and patients, and a negative impact on the patient-to-provider relationship [4,11-13]. A systematic review performed in 2020 on telemedicine and COVID-19 evaluated 44 articles along four service lines and identified 10 themes of efficiency [14]. However, the authors did not evaluate facilitators and barriers to adoption or health outcomes. Another systematic review [5] was performed in 2016 on the barriers to the adoption of telemedicine worldwide, which evaluated 30 articles across all service lines in all countries; however, it also did not evaluate facilitators or health outcomes.

Although analyses have been published that highlight the advantages to the adoption of telemedicine, with an 8% adoption rate in the United States, the conclusions of these previous studies may not be as robust as possible. The circumstances presented by the pandemic have encouraged wider adoption of this modality of care. Therefore, with proper systematic review techniques, reviewer observations this far into the pandemic will undoubtedly be more robust and widely applicable to medicine.

Objectives

The purpose of this systematic review was to evaluate the facilitators and barriers to the adoption of telemedicine worldwide, including an analysis of health outcomes and patient satisfaction. A brief comparison of the results of this review with those of reviews performed prior to COVID-19 was further performed to identify changes in these factors in light of the pandemic.

Methods

Protocol and Registration

The Kruse protocol for writing a systematic review was followed, and the findings are reported in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) guidelines [15,16]. This systematic review was registered in PROSPERO on August 2, 2021 (ID CRD42021235933).

Eligibility Criteria

The search parameters were established to find articles published in 2020 and 2021 concerning telemedicine in all aspects of care and for all ages of patients, published in peer-reviewed journals, using any method of study (mixed method, quantitative, and qualitative). Other systematic reviews were excluded because we wanted to compare our results to these previous reviews without confounding the findings. The Johns Hopkins Nursing Evidence-Based Practice Rating Scale (JHNEBP) was used to assess the quality of all articles analyzed [17]. Any studies below level IV C were discarded due to poor quality.

Information Sources

Four research databases were searched: PubMed (MEDLINE), CINAHL (excluding MEDLINE), Web of Science, and Science Direct. We also performed a journal-specific search of the Journal of Medical Internet Research.

Search Strategy

Google Scholar was used to determine the general trends of publication on this topic previously and to collect key terms from published articles. These key terms were entered into the US Library of Medicine’s Medical Subject Headings (MeSH) to create an exhaustive search string using Boolean terms. The actual search string used was: (telemedicine OR telehealth OR “mobile health” OR mhealth OR ehealth) AND (COVID-19 OR coronavirus). The same search string was used in all databases. Similar filters were used in each database (not all filters are the same between databases).

Study Selection Process

Once the search string was entered into each database, we filtered the results and screened abstracts for applicability. Although filters for the four research databases differ, we generally filtered for the date range (2020-2021), scholarly journals (no theses or opinions), and “full text” to ensure that we would have access to the entire article. Articles were rejected for a variety of reasons: protocol (no results to analyze); opinion (no data); reviews; did not use telemedicine; or did not contribute to our objective statement of identifying facilitators, barriers, or effects on patient satisfaction. The κ statistic was calculated to identify the level of agreement between reviewers [18].

Data Collection Process

An Excel spreadsheet was used as a data-extraction tool to collect data for reporting and analysis. This spreadsheet was standardized according to the Kruse protocol [15]. We held three consensus meetings to screen abstracts, analyze articles, and discuss possible themes. After the second consensus meeting, we performed a narrative analysis to identify themes in the articles analyzed [19]. Because there were only two authors on this project, both authors analyzed all articles (n=46).

Data Items

In accordance with the Kruse protocol, PRISMA standard, and JHNEBP, the following fields were collected: database source; date of publication; journal; authors; study title; PICOS (participants, intervention, results, outcomes, study design); sample size; bias within study; effect size; country of origin; statistics used; quality metrics from the JHNEBP scale; and reviewer observations as they relate specifically to the objective statement in areas of patient satisfaction, and facilitators and barriers to adoption [15,17,20]. All data items were independently collected and discussed in subsequent consensus meetings.

Risk of Bias Within and Across Studies

The JHNEBP rating scale was used for assessment of bias within and across studies. Observations of bias and methodological weaknesses were noted [17]. The JHNEBP ratings also provided insight into bias because poor-quality results can limit the external validity of the experiment.

Summary Measures

Because we included mixed methods and qualitative studies, we were unable to standardize summary measures as would be performed in a meta-analysis.

Additional Analyses

We performed a narrative, or thematic, analysis of the observations to convert them into themes (common threads between articles) [19]. We calculated the frequency of occurrence of both themes and individual observations and report these in a series of affinity matrices (tables). This technique was used to identify the statistical probability for identifying each theme, which does not identify a level of importance but rather identifies a frequency of mention of these themes in the literature during the period of observation.

Results

Study Selection

The database search and study selection process are illustrated in Figure 1. The κ statistic was 0.95, indicating almost perfect agreement between reviewers [18,21]. Several studies made it through all filters, but were still eliminated because they were protocols (no results), opinions, out of the date range, or other systematic reviews.

‎View this figure

Study Characteristics

Reviewers collected study characteristics identified by the PRISMA standard such as PICOS (see Table 1). Of the 46 six studies analyzed over the 15-month period, 2 (4%) involved adolescents, 6 (13%) involved adults >60 years, and 38 (83%) involved adults >18 years as participants. Most participants were current or former patients who agreed to participate in studies. More than half the interventions were mobile health (mHealth), telephone/televideo, or eHealth (26/46, 56%). The rest were interventions involving telemonitoring, patient portals, telecoaching, web chat, and social media, which could be cross-platform. In these 46 studies, 18 resulted in a positive outcome over a control group (23%), 12 of which involved medically measured outcomes (21%) as opposed to clinical and administrative outcomes. Only 9 of the 46 (20%) studies resulted in no statistically significant difference between the intervention and control groups, which means that positive results could be obtained through telemedicine commensurate with those obtained using traditional means of care. Four articles analyzed were published in 2021 [22-25], with the remaining 42 articles published in 2020 [26-67]. Further explanation of the results and medical outcomes can be found the Additional Analysis subsection.

Risk of Bias Within and Across Studies

Table 2 summarizes the quality indicators assessed for each article with the JHNEBP tool. The strength of evidence most frequently observed was level III followed by level I and level II. Nearly half of the articles reported strong-evidence studies that included both a control group and randomization; the next most common study type was nonexperimental (no control group) or qualitative, with the least frequent type being quasiexperimental (included a control group but no randomization). The quality of evidence most frequently observed was A (high quality), followed by B (good quality). The most common combination of strength and quality was III B, followed closely by I A, which speaks to both the strength and quality of evidence evaluated by this review. The III B combination highlights the number of qualitative studies with smaller samples or selection bias.

Many studies used geographically localized samples, which may limit the external validity of the results. Some studies focused only on one gender or race, speaking to the convenience sample or volunteer-basis of their design. Asking for volunteers in a technology-oriented experiment invites bias because the self-selection allows for those who are already technology-oriented or comfortable with technology to participate. This group as the intervention can skew the results because those already comfortable with technology will not experience the frustration experienced by those who are not comfortable with technology. This selection bias also limits the external validity of the results. A comprehensive list of bias, country of origin, sample size, strength, and quality of evidence identified for each study can be found in Multimedia Appendix 1.

Thematic Analysis Based on Results of Individual Studies

During the analysis phase of the systematic review process, the reviewers recorded observations to identify instances of patient satisfaction, as well as both facilitators and barriers to the adoption of telemedicine. A thematic analysis was then performed to make sense of the observations [19]. Multiple instances of the same observation become a theme. A translation of observations to themes is provided in Multimedia Appendix 2. The summary of analysis is provided in Table 3, which lists the themes/observations from reviewers that correspond with the objective statement and sorts articles from the most recent to the oldest.

Patient satisfaction was reported as “strong satisfaction” or “satisfaction” in 9 (20%) and 1 (2%) of the 46 studies, respectively, and 36 studies did not report any measure of patient satisfaction. No studies reported a decline in patient satisfaction as a result of using telemedicine as the intervention.

Twenty-five facilitator themes and seven individual observations were identified in the literature by the two reviewers. Only two studies did not identify facilitators. Facilitator themes are listed in Table 4.

The most commonly identified themes were increased self-management, acceptance of the technology from the pandemic, adherence to treatment protocols, access, and social support. For the 46 articles, these themes represent 38% of all 132 occurrences. Other themes included convenience of telemedicine and perceived ease of use, decreased cost, opportunity for education, technical literacy, availability of technology, an increase in patient-provider communication, faster initiation of treatment, increased connectedness, perceived usefulness, and past experience with technology. Health literacy and improved health behaviors were identified less frequently, and increased office efficiencies, medical concerns adequately addressed, enabled social interaction, increased quality of life, improved standard of care, increased flexibility, and increased weight loss were the least frequent themes identified. The following seven individual observations accounted for 5% of the total observations: decreased anxiety, increased technical literacy, televideo enabled reading of body language, fewer miles driven to appointment, long-term use may not be required to develop good habits, decreased emergency room visits, and no significant difference in cost of care.

Twelve themes and five individual observations were identified as barriers from the literature by the reviewers; 11 studies did not identify barriers (11%). Table 5 lists the themes and individual observations.

The most commonly listed barriers were technical literacy, technology needs further development, availability of technology, and patient preference, accounting for 55% of the total 86 occurrences. Cost, connectivity, and confidentiality/security were also identified, as well as health literacy, limits of reimbursement for telemedicine, and lack of personal desire to get better with less frequent occurrences (2 each). The remaining five observations made up a total of 6% of the total occurrences: decrease in quality of life after intervention, discomfort for wearing monitors, workflow issues for providers, lack of data infrastructure, and interoperability.

Additional Analyses

Distribution of Publications by Country

Eighteen of the 46 studies (39%) were performed in North America, 11 (24%) were performed in Europe, 7 (15%) were performed in Asia, 5 (11%) were performed in Australia, 3 (7%) were performed in South America, and 2 (4%) were performed in multiple countries and continents.

Comparisons to a Control Group

Table 6 summarizes the themes and observations recorded for results as compared to the control group identified by the two reviewers. There is some overlap between this set of observations and medical outcomes; the latter represent clinical observations only, whereas the former are both clinical and administrative in nature. Ten themes and four individual observations were identified by the reviewers for a total of 66 occurrences in the literature. Eleven studies were nonexperimental in nature, which had no control group.

Eighteen of the studies demonstrated either a clinical or administrative improvement compared to the control group, whereas nine reported no statistically significant results from the control group. Both of these themes demonstrate the efficacy of the telemedicine modality. The remainder of the list in Table 6 demonstrates the specific improvements that occurred (multiple improvements occurred in multiple articles), including improved patient satisfaction, improved behaviors, improved compliance/adherence to treatment protocol, improved self-management of condition or disease, decreased fat mass, improved emotional support, improved companionship, and improved health literacy. The remainder were individual observations that combined accounted for 5% of the total observations: improved informational support, decreased cost, and increased quality of life. Only one article did not report a result as compared to the control group because it was a posttrial analysis and it did not address the control group.

Medical Outcomes Commensurate With an Intervention

Table 7 summarizes the medical outcomes observed. Seven themes and nine individual observations were recorded commensurate with the adoption of telemedicine for a total of 30 occurrences. Twenty-eight studies did not report clinical outcomes.

Twelve studies reported 12 statistically significant improvements in clinical outcomes and three reported no statistically significant difference between modalities of care. Both of these themes demonstrated the efficacy of the telemedicine modality. The most commonly observed theme for medical outcomes was decreased body weight, followed by decreased fat mass, improved self-management, increase in controlled asthma, and increased quality of life. The following individual observations contributed to 30% of the total observations: reduction in blood pressure, reduction in mean tacrolimus trough coefficient of variation, improved annual rate of estimated glomerular filtration rate (eGFR) decline, decrease in medication use, decrease incidence of heart failure, decreased mortality, improved mental health inventory, decreased intracranial hemorrhage, and telemedicine identified other areas for intervention.

Interactions Between Observations

Interventions of mHealth resulted in seven occurrences of a result (clinical and administrative outcomes) and six occurrences of an improvement in at least one clinical outcome. The interventions with telephone or televideo resulted in four instances of improved patient satisfaction and a decrease in eGFR and weight loss. The interventions of eHealth resulted in very few instances of either clinical or administrative outcomes other than improved compliance and health behaviors.

Discussion

Principal Findings

Telemedicine is examined in countries worldwide, and it is clear that the COVID-19 pandemic caused a rapid adoption of this modality of medicine to ensure the viability of practices. A key issue for discussion is the differences in findings between this systematic review and another recent similar review [14]. This systematic review identified key facilitators and barriers, and further analyzed health outcomes. The other similar review identified themes of effectiveness but failed to meet the expectations for a systematic review in terms of medical outcomes [68]. Common themes between the two reviews were: rapid telemedicine expansion, education, improved access, convenience, and patient satisfaction.

Summary of Evidence

This systematic review exercised a set Boolean search string in four common research databases to analyze 46 articles originating from five continents for themes of facilitators, barriers, and medical outcomes. Nearly 50% of the articles demonstrated the strongest evidence and nearly 60% demonstrated the highest quality of evidence. Various forms of telemedicine were examined: eHealth, mHealth, audio only, telemonitoring, telecoaching, telerehab, robotics or artificial intelligence, and televideo. Twenty-five facilitator themes and individual observations, 12 results themes and observations, and 7 medical outcome themes and observations were recorded and analyzed. Forty-one percent of barrier themes recorded either an improvement or no statistically significant improvement in results compared to the control group. Forty percent of the observations recorded an improvement in at least one medical outcome.

Health care administrators can focus on the findings demonstrating that implementation of telemedicine will increase self-management [27,30,35,37,38,40,41,45,46,49,57,58], adherence [32,34,35,37,38,40,43,46,58,67], access [24,27-30, 41,52,53,60], and social support [49,50,54,55,62-64,66]. Telemedicine is shown to be an effective modality of treatment [28,32,35,38,40,41,43,46,48-50,53-55,58-60,64] at a decreased cost [31,46,52,56,59,60]. Patients perceive the modality to be convenient and easy to use [26,34,36,43-45,47], and its implementation increases patient satisfaction [24,25,43, 47,49,50,63].

Health policymakers should focus on several barriers to increase the adoption of telemedicine. Because technical literacy, availability of technology, and connectivity are listed as the most often cited barriers, public programs should be offered to assist those with these difficulties. Technical literacy is often associated along age or socioeconomic lines, and researchers acknowledge the dearth of research in the area of how to overcome this obstacle [69]. However, community centers that provide access to computers, classes on computers, and a dedicated broadband connection can all contribute to solutions to these barriers.

A key similarity between the 2020 systematic review [14] and this review is the rapid expansion of telemedicine. Eleven articles analyzed in this review used a phrase similar to “the pandemic created an acceptance of telemedicine technology” [23,25,31,33,41,43,47,51,52,61,62]. A systematic review published in 2018 cited cost as the chief barrier to adoption, whereas this review only found cost as a barrier in 8% of all observations [5]. The COVID-19 pandemic forced acceptance of the technology and enabled providers to not focus so intently on the cost of its implementation.

Limitations

This systematic review selected 46 articles for analysis from four commonly available research databases. A larger group for analysis could have yielded richer results. This review also only utilized two researchers to analyze the data; additional researchers could have identified additional themes. Selection bias was controlled through independent analysis of all articles by both reviewers followed by consensus meetings. Publication bias is the largest limitation because we were unable to query and analyze unpublished articles.

Conclusion

The COVID-19 pandemic caused huge problems to deliver medicine traditionally. However, these problems created an environment that limited face-to-face medical encounters and fostered legislation to reimburse the telemedicine modality for broad and rapid adoption of telemedicine to expand the access of care beyond the physical walls of the clinic. Physicians should feel confident that the telemedicine modality will be reimbursed and will have very little effect on patient satisfaction. Health care administrators who have not already adopted telemedicine should feel confident in the technology; however, they should ensure that sufficient confidentiality and security measures are in place. Policymakers should enact legislation to remove or mitigate barriers such as availability of technology, technical literacy, and connectivity, as these are commonly referred to in the literature.