The Arksey and O’Malley scoping review method was used to achieve the aims of this study [9]. This methodology involved an initial literature review, followed by an expert focus group, and finally targeted literature searches based on the focus group discussion (Figure 1) [9]. A scoping review method was chosen due to the volume of literature [10]. Reporting of the methods and results was performed in accordance with the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) Checklist [11].

To identify literature reporting the results of economic evaluations in telehealth, initial literature searches were conducted using terms relating to telehealth, telemedicine, and economics. Searches were conducted using the PubMed, EMBASE, and Scopus databases. Search results were restricted to publications available in English, and no restriction was placed on the country of origin or health system model described. The scope of the review was restricted to short- to mid-term (6 months to 3 years) term and cost implications reported from the perspective of the health system payer (ie, health service or hospital). After the completion of database searches, duplicates were removed, titles and abstracts were screened by 2 authors (MT and CS), and a full-text review of articles was conducted by 3 authors (MT, LC, and CS). Included studies were categorized as cost-minimization analysis (CMA), cost-effectiveness analysis (CEA), and cost-utility analysis (CUA) studies [6,12,13]. Further descriptions of these categories are provided in Table 1.

Data were extracted from the included studies by two investigators, and any disagreements in data extraction were discussed until a consensus was reached. The results were synthesized, and descriptions of systems in which telehealth reduces costs to the health system were identified and reported. All articles were quality assessed using the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) [15]. CHEERS is a 24-point quality assessment tool for published economic evaluations. It includes domains to assess the reporting of relevant economic principles such as currency, discounting, time horizon, effectiveness measures, choice of outcome, assumptions, and model choice. The CMA studies were assessed according to 20 of the 24 points in the checklist; choice of health outcome, measurement of effectiveness, measurement and valuation of preference-based outcomes, and incremental costs and outcomes were not applicable. All costs were converted to 2019 US $ to allow for ease of comparison. The synthesis of the findings was narrative.

Ethics approval to undertake the focus group was received from The University of Queensland’s Human Research Ethics Committee (Approval # 2018002428).

The framework proposed by Arksey and O’Malley [9] identifies that reviews can be enhanced by including a consultation process. To this end, investigators conducted an expert focus group to identify the domains where telehealth could reduce health system costs.

National experts in telehealth and economics were sent an invitation to attend the focus group on “How can telehealth reduce health system costs?”. Experts were identified and recruited through existing relationships between the research team and the Australasian Telehealth Society who assisted with recommending individuals with both health economics and telehealth expertise. A total of 16 experts were invited to participate and 9 agreed to represent 7 different organizations. Participation was via focus group (n=7) or via email and telephone with the investigators (n=2).

Attendees were given the option to attend in person or via videoconference. Before attendance, invitees were emailed a discussion paper. Invitees were asked if they agreed or disagreed with the points in the discussion paper and if any additional factors should be considered. Some invitees who were unable to attend the meeting provided feedback via email or phone. Open-ended questions on the potential for cost savings arising from telehealth were used to facilitate discussion in the focus group. The discussion was recorded and notes were recorded. Topics suggested by the expert panel were synthesized and categorized under distinct topic headings (domains). These domains were used to direct further literature searches. By including the focus group consultative exercise where participants were given potential domains, it was possible to elicit more ideas on the subject matter, resulting in richer data [9].

Subsequent to the focus groups, further literature searches were conducted to locate evidence on domains identified by the expert panel. Literature searches were conducted using both broad search terms and domain-specific terms to identify supporting evidence. Searches were conducted in PubMed, EMBASE, Scopus, and gray literature. Additionally, members of the expert panel contributed relevant evidence items (designated as hand searches). No restrictions on publication date were set during any search process. Searches were conducted from May 2018 to January 2019. Articles were omitted when the telehealth modality referenced telephone only. Search terms included, but were not limited to, telehealth, telemedicine, store-and-forward and other telehealth nomenclature, and domain-specific search terms such as secondary care or productivity. No restrictions were placed on the country of origin or the health system described in the analyses.

The level of evidence was described using the National Health and Medical Research Council guidelines [16], where, for example, I is a systematic review of randomized controlled trials (RCTs), II is an RCT, III is a comparative study with or without controls, and IV is a case series. The results of the subsequent articles identified through the targeted searches were summarized.

Searches and screening identified 17 cost-minimization studies that reported their results from the perspective of the health system. Of these 17, 9 studies (53%) reported telehealth to be cost saving compared with conventional care (Figure 2), 6 studies (35%) reported telehealth to be cost saving after a workload threshold was achieved, and 2 studies (12%) reported telehealth to be more expensive than conventional care. The overall quality of reporting was sound, with an average score of 15 out of 20 (Table 2).

The most common situation where telehealth reduced health system costs reported in these studies was when it offsets patient or clinician travel funded or subsidized by the health system [17,19,21,25,26,28,30,31]. Hence, savings are most likely to be realized in the public health system, as it is unusual that other service models cover patient transport costs with the exception of the Department of Veterans Affairs. It is more likely that savings will be realized when patient travel is substituted with telehealth versus when clinician travel is substituted due to the volume, that is, saved transport cost for one clinician versus saved transport costs for many patients. One of the reviewed studies found telehealth to be less expensive than subsidized patient travel but more expensive than outreach clinics (where clinicians travel to outlying areas to provide consults) [25]. Prevention of emergency transfers was another way in which telehealth could contribute to reduced travel costs [24].

Other scenarios where potential savings can be realized include remote monitoring of high-risk pregnancies, reducing the need for in-hospital monitoring [18]. Remote monitoring in lieu of in-home visits also realized savings due to saved staff travel costs, the reduced salary of home-visit nurses, and the reduced number of hospitalizations that resulted from continual monitoring [22]. Teletriage, such as when a nurse screens a patient to determine if tertiary care is necessary, was also shown to save costs by reducing the number of hospital visits [30,31]. The two services that found telehealth to be more expensive than conventional care were due to the additional salary of a patient-end clinician (eg, general practitioner) attending a specialist consultation via videoconference [32,33].

In a number of CMA studies, telehealth was found to be cost saving as compared with standard care models only after a certain caseload was exceeded, known as a threshold or break-even point [19,21,24-26,28]. The break-even point was when the initial investment (typically in equipment setup and staff training) was offset by realized savings later on. Many studies calculated a break-even point; however, only one study reported the payback period [17]. The payback period ranged from near immediate to 9 years after implementation [34]. The payback period could be reduced if the activity was higher and the service had the capacity for increased activity [23,31]. Cost savings for telehealth have been slow to materialize for many early adopters due to the time lag between capital investments and the broader adoption of telehealth [35]. The marginal costs of a teleconsultation were found to be less than the marginal costs of conventional consultations, indicating that telehealth is likely to benefit from economies of scale [21].

A total of 8 cost-effectiveness studies from the past 5 years that reported costs from the perspective of the health system were identified [36-43]. Of the 8 studies, 4 (50%) were in quartile 2 of the cost-effectiveness plane (Figures 2 and 3), meaning they demonstrated cost savings and an increased or equivalent clinical effectiveness for telehealth compared with conventional in-person care. These are also known as dominant strategies and are recommended for implementation. Two of these studies reported results from a remote monitoring intervention for heart failure [36,42]. The remaining studies (4/8, 50%) were in quartile 1 of the cost-effectiveness plane, meaning increased clinical effectiveness for increased costs, requiring a value judgment to be made.

Moreover, 3 of the 4 cost-effectiveness studies in quartile 1 scored highly on quality as assessed using the CHEERS checklist [15], indicating comprehensive reporting of study results.

The telehealth services that were shown to reduce direct health system costs and be equally or more effective than their comparators were in smoking cessation [43], cardiovascular remote monitoring [36,42], and physiotherapy telerehabilitation after orthopedic surgery [41] (Table 3). Telehealth interventions for cardiovascular remote monitoring [36,42] and postdischarge monitoring for neonates [44] have been shown to reduce hospital admissions, readmissions, and emergency department (ED) presentations, which have the potential to reduce overall costs to the health system [36,42,44]. In these studies, hospital events were used as a measure of the effectiveness of the telehealth intervention. The premise being that preventing hospital events would reduce costs for the health system if it was translated into a dollar value.

CEA studies use measures of effectiveness that reflect expected outcomes from the intervention. Many of these effects represent positive health gains and potentially medium- to long-term cost savings for the health system, such as avoided treatment of smoking-related diseases. However, the studies in which these effects have been demonstrated have not valued these gains in terms of cost.

A total of 25 cost-utility studies from the past 5 years that reported costs from the perspective of the health system and changes in health-related quality of life (HRQoL) were identified [37,40,45-65]. Of these 25 studies, 8 (32%) studies were in quartile 2 of the health economics plane (Figures 2 and 4), as they demonstrated costs savings and increased or equivalent changes in effect as measured by HRQoL [45,47,49,50,57,60,65,66].

Telehealth interventions that save money and increase quality-adjusted life years (QALYs) should be considered for implementation. As shown in Figure 5, studies in the lower right quadrant of the plane (Q2) represent telehealth interventions that, when compared with usual care, represent a decrease in health care costs and a gain in QALYs. Each of these studies describes an intervention that, if implemented, could potentially save money for the health system.

The remaining studies (17.25, 68%) represent telehealth interventions that, when compared with usual care, represent an increase in health care costs and a gain in QALYs (quartile 1; Figure 4). Therefore, these interventions would only be implemented by decision makers if they were willing to pay for the relative increase in HRQoL.

All the cost-utility studies scored highly on quality as assessed using the CHEERS checklist [15], indicating the comprehensive reporting of study results.

An incidental finding of the review was that most telehealth interventions examined provided a small marginal improvement in the number of QALYs [37,40,45,46,48-68]. This small marginal change is likely due to the sensitivity of the instruments used to measure the quality of life and their ability to respond to the changes resulting from changing the service delivery model. Increases in QALYs for telehealth interventions compared with usual care were between 0.0006 and 0.12, irrespective of the clinical area or telehealth intervention type [37,40,45,46,48-68]. This is below what is considered clinically meaningful. Studies conducted from a health system perspective reported cost savings between US $32 and US $3523. All studies were located in the dominant quartile, demonstrating that the studies that reduced costs for the health system also demonstrated either equal or increased QALYs (Figure 5; Table 4).

The expert panel focus group ran for 1 hour, and 7 experts attended from 5 institutions. Two further experts from 2 different organizations were unable to attend and provided input via email or telephone instead. They covered telehealth expertise in academic research, health service provision, and economics. Seven experts who were contacted did not respond. The videoconference session was recorded so that the investigators could review the comments during the writing of this report.

At the conclusion of the expert panel, the telehealth domains where health system costs could potentially be reduced were finalized. These domains are listed in Table 5.

The first domain identified by the expert focus group was productivity gains, which is when the productivity of a system is increased, cost savings can be realized due to the increased capacity of the system. In the case of telehealth, it is possible that a greater volume of patients can be managed with similar resources, which increases the productivity of the health system and reduces the marginal cost per patient. Evidence of increasing physician productivity with the use of telehealth exists in various other forms, including case control, RCT, CMA, and case series. Telehealth has the potential to increase health system productivity through a number of mechanisms, including reducing travel, reducing consultation time, and substituting for in-person service modalities.

Secondary care involves services provided by specialist or tertiary centers and includes hospital admissions, specialist outpatient visits, and ED presentations. Avoidance of secondary care in favor of other methods of care has the potential to reduce health system costs [89]. The evidence was from RCTs, case-control prospective and observational studies, and reviews of health services. Most of the studies, except the CMA conducted by Pare et al [22], reported economic findings as a secondary result. This resulted in moderate quality studies examining changes in secondary care usage related to telehealth, but a level of extrapolation from the results was needed to interpret cost savings. The use of telehealth to reduce secondary care may be realized through a number of scenarios, including remote monitoring, hospital avoidance, and triage.

Direct-to-consumer telehealth services are often funded by consumer payments. User pays funding models can potentially reduce costs, particularly if they substitute for government-funded or government-subsidized health services. Hence, savings to the health system are based on the assumption that individuals who access direct-to-consumer services would have otherwise accessed an equivalent health system service if the telehealth option was unavailable to them. In Australia, very few direct-to-consumer services are eligible for reimbursement from the Medicare Benefits Scheme. Consumers accessing these services are required to pay the full service fee as an out-of-pocket cost. This leads to cost savings for the government when consumers access these services instead of Medicare-funded services. Users may be willing to pay a higher out-of-pocket cost for these services because they offer elements that they value, for instance, either convenience or timely specialist access. However, due to the convenience of telehealth services, there is a risk that direct-to-consumer services may not reduce health system usage, but rather may increase overall health service utilization and costs to the health system [102]. That is, individuals who would not normally use health services may begin accessing telehealth services because of the convenience provided by the direct-to-consumer model. When these individuals are referred to a health system service provider or given a Pharmaceutical Benefits Scheme reimbursed prescription, they increase service utilization and costs to the overall health system.

There is limited academic literature that reports on savings to the health system from direct-to-consumer telehealth. Examples of commercial direct-to-consumer telehealth models include Qoctor (previously called DrSicknote), which is an Australian web-based general practitioner service. Qoctor provides a range of general practitioner services through their website, including medical certificates, prescriptions, and specialist referral letters. In January 2019, Qoctor reported $1,040,566 in saved costs to Medicare to date by diverting appropriate patients from standard general practitioner clinics [103]. A further example is iDoc24, where consumers capture images of skin lesions (eg, a mole), transmit the image to a dermatologist, and receive a diagnosis within 24 hours [104].

Deferring treatment to less qualified and, therefore, less expensive staff could potentially result in reduced health system costs. This can be facilitated by telementoring. Arguably, the most well-known telementoring program is the Project Extension for Community Healthcare Outcomes (Project ECHO), a model that was started in the United States but has been adopted and practiced internationally [105]. Using this model, primary care staff are upskilled using videoconference sessions with specialists in the form of weekly telementoring sessions. During these sessions, primary care providers can present cases and receive specialist advice on diagnosis, management, and treatment for their patients.

Although telementoring was identified by the focus group as a way to potentially reduce costs, limited evidence was found to support this view. A pre- and poststudy found that telementoring in the context of geriatric mental health resulted in a small reduction in per-patient cost when medication, specialist outpatient visits, hospitalization, and ED visits were quantified from insurance claim data [106]. Although overall costs were reduced, a subanalysis did reveal increased costs resulting from an increase in antipsychotic medication prescriptions. The time horizon for reported cost reduction was 6 months. The cost of setting up and running the telementoring was not considered in this analysis.

Project ECHO, a program that provides telementoring for primary care physicians, for the management of hepatitis C, was found to be cost-effective with an incremental cost-effectiveness ratio (ICER) of US $10,351 per QALY [107]. This would indicate that telementoring for hepatitis C will increase costs to the health system but has the potential to increase population-related quality of life.

From another perspective, savings to the health system as a result of telementoring may be realized in the retention of staff and patients at remote medical practices [88].

Telehealth was shown to reduce costs to the health system in the short to medium term in 53% of CMA, 50% of CEA, and 32% of CUA studies reviewed. The predominant reason for reduced costs was when the health system funded travel and either patient or clinician travel was reduced or avoided. In the remaining studies (not reviewed in detail here), telehealth increased costs but was also shown to improve care [37-40,42-44,46,48,51,52,54-56,58,59,61-64]. For example, evidence indicates that remote patient monitoring is currently a poor cost minimizer; however, it is very effective for improving overall health and reducing morbidity and hospitalization.

The models of the care and the contexts in which telehealth is used are heterogeneous. The question as to whether telehealth decreases the cost of health care delivery is complex, as are all economic evaluations in telehealth [7]. There are many compounding factors to consider, for example, the modality (real-time videoconferencing, remote patient monitoring, and store-and-forward) of telehealth used or the way telehealth consultations are remunerated. In Australia, Medicare reimburses provider telehealth consultations at 150% of an equivalent in-person consultation [108]. Furthermore, reimbursement for telehealth in Australia under Medicare and activity-based funding have both provider and patient-end payments, automatically making telehealth more expensive than conventional care. Although in the United States, state-by-state reimbursement under Medicaid means a telehealth consultation may be reimbursed at a lower or equal rate to an equivalent in-person consultation.

By improving accessibility, telehealth may also increase the cost of providing health care as populations served by traditional models of care who had limited or no access to care can now access services. Further, by improving the convenience of access, there is potential for excess use [109,110]. Furthermore, the potential for telehealth to become adjunctive to traditional in-person care (rather than substitutive) also increases the potential for increasing use and associated costs.

Telehealth is not implemented at scale in some jurisdictions (eg, Australia) [79,111], which may impact the ability to reduce costs. The marginal cost or average cost of a telehealth consultation was found to be less than an equivalent in-person consultation in a number of studies [19,32]. However, costs of running the telehealth service remained higher than the equivalent in-person service until a threshold number of consultations was reached. A threshold number of examinations required is necessary to counteract the implementation costs of telehealth, including technology and project costs. Telehealth technology costs are decreasing. Video conference equipment costs are reducing with a move from room-based systems to consumer videoconferencing platforms. A similar cost decrease of remote monitoring equipment and resultant cost decrease in remote monitoring costs per patient per year has also been reported [7]. Future studies may find a reduction in both the threshold number of examinations and the payback period resulting from a decrease in technology costs.

Many telehealth services do not result in short- to medium-term savings. For this reason, health services considering implementing telehealth should be motivated by the benefits of telehealth other than cost reduction and recognize that implementing telehealth will require recurrent investment costs.

Asynchronous consultations are likely to reduce consultation time compared with equivalent in-person consultations, thereby improving clinician productivity. There is potential to increase the cost to the health system under fee-for-service models; hence, reimbursement for consultations where the patient is not present needs to account for increased productivity. One such example is a recent application from the Australasian College of Dermatologists for Medicare funding for store-and-forward dermatology, which sought a lower reimbursement than in-person consultations [112].

This study has identified a number of opportunities for future research. Many of the reviewed economic analyses of telehealth are older than 10 years. Contemporary economic analysis of telehealth is needed to consider changes in technology costs. Our study has also identified a gap in economic analyses related to telementoring. Furthermore, the use of telehealth to triage referrals in dermatology, wound care, otolaryngology, and ophthalmology can be effective in reducing unnecessary specialist outpatient appointments; however, there are no published cost analyses. Additionally, the use of remote patient monitoring has been shown to reduce costs and increase costs in different contexts. Many remote monitoring studies did not report overall cost savings (eg, the cost of implementing and running the service). Analyses of remote monitoring services need to include overall costs rather than direct health costs alone to determine if remote patient monitoring reduces system costs.

Positive reporting bias is likely to result in an increased number of studies reporting cost-effectiveness and cost-minimization of telehealth. Savings on travel resulting from telehealth are more beneficial to the health system in countries such as Australia. The large geographical areas and a public health system that funds travel favor such contexts. For this reason, the findings of cost savings are rarely generalizable. Additionally, it is unlikely that the studies examined were statistically powered for economic findings, as it is routine practice to power for the primary clinical outcome, meaning economic outcomes may not be precise. The aim of this study was to identify short- to medium-term cost savings; therefore, a further limitation of this study is that studies examining the long-term cost impact of telehealth interventions were not included.

This study aimed to determine whether telehealth has the potential to improve the sustainability of health systems by reducing costs. Telehealth was shown to reduce costs to the health system in the short to medium term in 53%, 50%, and 32% of the cost-minimization, cost-effectiveness, and cost-utility studies reviewed, respectively. The predominant reason for reduced costs was when the health system–funded travel (patient or clinician) was reduced or avoided. In the remaining studies, telehealth increased costs, albeit with improved care.

The expert focus group identified 4 areas of potential savings from telehealth: productivity gains, reductions in secondary care use, emerging alternate funding models for care provision, and savings resulting from telementoring effects. Telehealth can increase clinician productivity when it is used to convert travel time to clinical time. In terms of consultation time, there are unlikely to be productivity gains when substituting an in-person consultation with a video consultation. The use of asynchronous consultations as a substitute for in-person consultations is more likely to increase productivity by reducing consultation time. However, under activity-based funding mechanisms, there is a likelihood that these productivity gains could result in cost increases.

Mitigation of secondary care through remote patient monitoring, teletriage, and hospital avoidance has the potential to reduce costs to the provider. However, there is currently a lack of economic evidence to support this. Similarly, telementoring has scant economic evaluations to demonstrate cost savings.

Alternate funding models from telehealth systems have the potential to save the health system money in situations where the consumers pay fully out of pocket to receive services, thereby mitigating the cost to the health system. The convenience of telehealth may influence consumers to pay out-of-pocket fees. This may be considered cost-shifting as opposed to cost saving.

The available evidence has indicated that telehealth does not always reduce the cost of care from the perspective of the health system in the short to medium term. Health services considering implementing telehealth should be motivated by benefits other than cost reduction.