The Agency for Healthcare Research and Quality (AHRQ), through its Evidence-Based Practice Centers (EPCs), sponsors the development of evidence reports and technology assessments to assist public- and private-sector organizations in their efforts to improve the quality of health care in the United States. This report was requested and funded by the Centers for Medicare & Medicaid Services. The reports and assessments provide organizations with comprehensive, science-based information on common, costly medical conditions and new health care technologies. The EPCs systematically review the relevant scientific literature on topics assigned to them by AHRQ and conduct additional analyses when appropriate prior to developing their reports and assessments.
To bring the broadest range of experts into the development of evidence reports and health technology assessments, AHRQ encourages the EPCs to form partnerships and enter into collaborations with other medical and research organizations. The EPCs work with these partner organizations to ensure that the evidence reports and technology assessments they produce will become building blocks for health care quality improvement projects throughout the Nation. The reports undergo peer review prior to their release.
AHRQ expects that the EPC evidence reports and technology assessments will inform individual health plans, providers, and purchasers as well as the health care system as a whole by providing important information to help improve health care quality.
We welcome comments on this evidence report. They may be sent by mail to the Task Order Officer named below at: Agency for Healthcare Research and Quality, 540 Gaither Road, Rockville, MD 20850, or by e-mail to epc@ahrq.gov.
Carolyn M. Clancy, M.D.
Director
Agency for Healthcare Research and Quality
Jean Slutsky, P.A., M.S.P.H.
Director, Center for Outcomes and Evidence
Agency for Healthcare Research and Quality
Mark B. McClellan, M.D., Ph.D.
Administrator
Centers for Medicare & Medicaid Services
Kenneth S. Fink, M.D., M.G.A., M.P.H.
Director, EPC Program
Agency for Healthcare Research and Quality
Martin Erlichman, M.S.
EPC Program Task Order Officer
Agency for Healthcare Research and Quality
Valuable assistance for this project was provided by external peer reviewers, and the research team would like to acknowledge their helpful contributions to the project. A listing of those individuals appears in Appendix D. We also acknowledge the efforts of the Evidence-based Practice Center support staff, specifically Andrew Hamilton, MLS, MS. Our Task Order Officer from the Agency for Healthcare Research and Quality, Martin Erlichman, MS, also provided invaluable assistance and guidance in the preparation of this report.
Context. Telemedicine services are increasingly utilized by patients, clinicians, and institutions. Although private and Federal insurers are covering some telemedicine services, the rationale for these coverage decisions is not always evidence-based.
Objectives. The goal of this report was to assess the peer-reviewed literature for telemedicine services that substitute for face-to-face medical diagnosis and treatment that may apply to the Medicare population. We focused on three distinct areas: store-and-forward, home-based, and office/hospital-based services. We also sought to identify what progress had been made in expanding the evidence base since the publication of our initial report in 2001 (AHRQ Publication No. 01-E012.)
Data Sources. Ovid MEDLINE®, reference lists of included studies, and non-indexed materials recommended by telemedicine experts.
Study Selection. Included studies had to be relevant to at least one of the three study areas, address at least one key question, and contain reported results. We excluded articles that did not study the Medicare population (e.g., children and pregnant adults) or used a service that does not require face-to-face encounters (e.g., radiology or pathology diagnosis).
Data Extraction. Our literature searches initially identified 4,083 citations. Using a dual-review process, 597 of these were judged to be potentially relevant to our study at the title/abstract level. Following a full-text review, 97 studies were identified that met our inclusion criteria and were subsequently included in the report's evidence tables.
Data Synthesis. Store-and-forward services have been studied in many specialties, the most prominent being dermatology, wound care, and ophthalmology. The evidence for their efficacy is mixed, and in most areas, there are not corresponding studies on outcomes or improved access to care.
Several limited studies showed the benefits of home-based telemedicine interventions in chronic diseases. These interventions appear to enhance communication with health care providers and provide closer monitoring of general health, but the studies of these techniques were conducted in settings that required additional resources and dedicated staff.
Studies of office/hospital-based telemedicine suggest that telemedicine is most effective for verbal interactions, e.g., videoconferencing for diagnosis and treatment in specialties like neurology and psychiatry.
Conclusions. There are still significant gaps in the evidence base between where telemedicine is used and where its use is supported by high-quality evidence. Further well-designed and targeted research that provides high-quality data will provide a strong contribution to understanding how best to deploy technological resources in health care.
Telemedicine is the use of telecommunications technology for medical diagnostic, monitoring, and therapeutic purposes where distance and/or time separates the patient and health care provider. Both federal and private health insurers are now covering some telemedicine services, despite the fact that the benefits and costs of providing many of these services have not been well studied.
This report assesses the peer-reviewed literature of specific telemedicine study areas, with a focus on those that substitute for face-to-face medical diagnosis and treatment of the Medicare population. Thus, this report targets face-to-face clinical specialties (as opposed to radiology and pathology) and the Medicare population. It does not evaluate studies examining populations that usually are not covered by Medicare, such as children and pregnant women. The report identifies health care services that could be provided using telemedicine and describes existing programs in three categories of telemedicine: store-and-forward, home-based and office/hospital-based services.
This evidence report provides an update on the state of telemedicine, following the 2001 publication of Telemedicine for the Medicare Population.1 It identifies whether there has been significant progress in the number and types of telemedicine studies being conducted. More specifically, we searched for well-designed studies that evaluated telemedicine services in three technological categories described below.
In store-and-forward telemedicine, clinical data are collected, stored, and then forwarded to be interpreted later. A store-and-forward system eliminates the need for the patient and the clinician to be available at either the same time or place. The following key questions were addressed for store-and-forward telemedicine services in Medicare-eligible patient populations.
Does store-and-forward telemedicine result in comparable diagnostic decisions and recommendations for clinical management?
Does store-and-forward telemedicine result in comparable health outcomes?
Does the availability of store-and-forward telemedicine services improve access to care?
Home-based telemedicine services enable physicians and other health care providers to monitor physiologic measurements, test results, images, and sounds, usually collected in a patient's residence or a care facility. The use of home-based telemedicine services in Medicare-eligible patient populations was examined, asking the following questions relative to conventional care using face-to-face encounters.
Does home-based telemedicine result in comparable diagnostic decisions and recommendations for management?
Does the use of home-based telemedicine result in comparable health outcomes?
Does the use of home-based telemedicine improve access to care?
Office/hospital-based telemedicine services are real-time clinician-patient interactions that substitute for face-to-face encounters between a patient and a physician or other health care provider. The use of office/hospital-based telemedicine was evaluated relative to face-to-face encounters on the basis of the following questions.
Does office/hospital-based telemedicine result in comparable diagnosis and appropriateness of recommendations for management?
Does office/hospital-based telemedicine result in comparable health outcomes?
Does the availability of office/hospital-based telemedicine improve access to care?
The Oregon Evidence-based Practice Center (EPC) team that developed this report sought to identify peer-reviewed literature in the three study areas. We searched the peer-reviewed literature using the Ovid MEDLINE® electronic bibliographic database. We also searched through telemedicine reports and compilations, including their reference lists.
The inclusion criteria for the evidence report update were that the study was relevant to at least one of the three study areas, addressed at least one key question in the analytic framework for that study area, and contained reported results. Exclusion criteria for the evidence report update were that the study did not address a key question, lacked reported results, had a study population that was not relevant to the Medicare population (i.e., children and pregnant adults) or substituted for a service that did not historically require face-to-face encounters (e.g., diagnostic services in radiology or pathology).
We categorized the included articles by the key question(s) they addressed. The included studies for each study area and key question were critically appraised to determine the strengths and limitations of the most important studies following a detailed rationale for the appraisal of study characteristics related to quality.
For each study area, we constructed an evidence table summarizing the strength of the evidence for each key question. We then constructed summary tables for clinical specialties or domains. For studies of diagnosis and management decisions, we explicitly noted whether studies assessed concordance (without a diagnostic gold standard) or accuracy (with a gold standard) of the telemedicine system when compared to conventional care.
Following review of the abstracts of all studies retrieved in the literature search, a total of 597 citations were determined potentially to have evidence for the efficacy of one of the three study areas. The full text of these 597 articles was reviewed. After exclusion criteria were applied, there were 97 articles included in evidence tables. Of these, 35 articles assessed store-and-forward telemedicine services, 27 articles evaluated home-based services, and 38 articles assessed office/hospital-based services. Some studies assessed more than one telemedicine study area.
Individual studies were assessed for evidence based on criteria applicable to the study question. Studies were too heterogeneous to undertake any quantitative aggregate analyses such as meta-analysis.
Store-and-forward telemedicine. Similar to our original evidence report, the studies we found of store-and-forward telemedicine only assessed diagnosis or management decisions and not outcomes. As we also found in the original report, some aspects of the telemedicine systems used in home and office-hospital settings made use of store-and-forward techniques, but in the context of larger and/or interactive interventions.
1. Does store-and-forward telemedicine result in comparable diagnostic decisions and recommendations for clinical management?
Similar to the original report, the largest number of studies came from the specialty of dermatology. The most commonly assessed aspect of teledermatology was interobserver concordance. The range of concordance varied widely, from 41 percent to 87 percent for complete agreement to 51 percent to 96 percent for disease-category agreement. All of these studies were limited by the lack of measurement of concordance among more than one face-to-face examiner. Concordance studies assessing management decisions typically looked at decision to biopsy. While one study found complete agreement, others found lesser concordance. The studies of diagnostic accuracy typically compared the telemedicine diagnosis to some sort of gold standard, often a biopsy of a pigmented lesion. In these studies, telemedicine generally was nearly as good as face-to-face in correctness of diagnosis. The second most frequently studied clinical area was wound care. These studies demonstrated that some characteristics of skin wounds and ulcerations could be assessed effectively using store-and-forward telemedicine. However, most of these studies had small numbers of patients and very small numbers of clinicians, raising statistical power issues. Five studies provide data on store-and-forward applications in ophthalmology. Four of these studies show that a high accuracy of diagnosing diabetic retinopathy could be obtained. Other specialties studied included gynecology and gastroenterology.
2. Does store-and-forward telemedicine result in comparable health outcomes?
Similar to the previous report, there were no studies that assessed health outcomes using store-and-forward telemedicine interventions. This is problematic for assessing the overall benefit of store-and-forward telemedicine, since the outcomes from its use for diagnosis and management decisions are unclear.
3. Does the availability of store-and-forward telemedicine services improve access to care?
When store-and-forward telemedicine systems have been evaluated as a method for performing specialty consultations of patients followed by general practitioners or primary care clinicians, the systems have had only a small impact on reducing the need for subsequent face-to-face clinical evaluations by specialists. While these systems can aid in the triage of patients referred for consultation, they have not been shown conclusively either to improve access to care or to have a negative influence on access to care. Five studies reported evidence on the effect of store-and-forward techniques upon access to care. The methodologic quality of these studies generally was low.
Home-based telemedicine. Home-based telemedicine is most commonly used for management of chronic diseases or specific conditions, such as heart disease, diabetes mellitus, and rehabilitation. Some studies show telemedicine applied in this setting can be efficacious, although many are limited by small sample sizes, inadequate length of follow-up, and inconclusive results.
1. Does home-based telemedicine result in comparable diagnostic decisions and recommendations for management?
Two studies assessed diagnostic capabilities in the home in the areas of congestive heart failure assessment and pulmonary function monitoring. These studies found various levels of agreement and disagreement depending on the specific observation.
2. Does home-based telemedicine result in comparable health outcomes?
A variety of published studies have assessed chronic diseases afflicting patients in the Medicare population, such as congestive heart failure, diabetes mellitus, coronary artery disease, and hypertension. Unfortunately, the studies are very heterogeneous, and their limitations prevent broad conclusions. Interventions showing multi-faceted interventions demonstrate more benefit than single interventions, such as monitoring of blood sugar or blood pressure. However, in most studies, it is not possible to assess whether improved outcomes are due to the increased level of care provided by dedicated clinical staff versus the technology intervention.
3. Does the use of home-based telemedicine improve access to care?
No studies were identified that examined the effect of home-based telemedicine services on access to care.
Office/hospital-based telemedicine. A variety of studies were found that assessed office/hospital-based telemedicine. The largest number of studies assessed diagnosis and management decisions, and these studies examined a broad range of medical specialties. There were, however, very few studies of high methodologic quality.
1. Does office/hospital-based telemedicine services result in comparable diagnosis and appropriateness of recommendations for management?
For diagnosis and management decisions, the most frequently studied specialty was ophthalmology. As with store-and-forward studies, some aspects of ophthalmologic evaluation were amenable to interactive telemedicine, while others were not. Other frequently studied specialties included neurology and psychiatry, which demonstrated that some diagnostic assessments can be successfully administered via telemedicine.
2. Does office/hospital-based telemedicine result in comparable health outcomes?
Studies of outcomes also showed that for most of the clinical specialties assessed, outcomes with telemedicine interventions are comparable to those using conventional clinical evaluations. However, most of these studies are limited by small sample sizes and/or other problems. None of these studies attempted to measure their statistical power to avoid type-2 (beta) errors.
3. Does the availability of office/hospital-based telemedicine improve access to care?
Studies having relatively weak methodologies suggest that office/hospital-based telemedicine can improve access to care for patients in rural locations in medical applications in which patient evaluations can be performed using standard teleconferencing equipment.
This update on evidence about the efficacy of telemedicine for the Medicare population covered published peer-reviewed literature for the five years between 2000 and 2004. Similar to the findings of our original report a half-decade ago, there are still serious gaps in the evidence base for telemedicine. While this situation is hardly unique to telemedicine, having a solid evidence base is essential given that there is increased advocacy for health care payers, especially Medicare, to provide coverage for its use.
The best evidence for the effectiveness of telemedicine is in medical specialties for which verbal interactions are a key component of the patient assessment, such as psychiatry and neurology. Various psychiatric and neurological assessments can be administered effectively via interactive videoconferencing. Likewise, treatments administered in these specialties via telemedicine appear to achieve comparability with face-to-face care.
Our systematic review also identified several studies, a few of them of high methodologic quality, showing benefits of home-based telemedicine interventions in chronic diseases. These systems appear to enhance communication with health care providers and provide closer monitoring of general health, but the studies of these techniques were conducted in settings that required additional resources and dedicated staff. With ongoing improvements in telecommunications technology, particularly broadband connections to the home, further research, including larger clinical trials, will likely be informative.
The specialty with the largest number of studies is dermatology, and most studies of teledermatology have evaluated store-and-forward techniques. The body of evidence summarized in this report is consistent with the findings of the earlier report. There continues to be highly variable rates of interobserver and intraobserver agreement in diagnoses. This issue can only be resolved by high-quality studies that compare not only the concordance of telemedicine versus face-to-face diagnosis, but also the concordance of face-to-face versus face-to-face diagnosis in the same situation.
Of course, rates of concordance in a vacuum, i.e., without a clinical context of how the patient fared, are also limited from an evidence standpoint. What we ultimately need to know is the patient outcome. In other words, did the teledermatology encounter at least provide comparable care for the patient? A corollary question that must be answered is whether teledermatology resulted in harm from any missed diagnoses or other consequences of the telemedicine encounter. These questions can only be answered in studies of clinical outcomes, none of which were identified in this report. In general, advocacy for an expanded role for teledermatology will require further studies that examine rates of missed diagnoses, incorrect treatments, and when the technology is insufficient to avoid in-person encounters.
Other widely studied domains include ophthalmology and wound care. Teleophthalmology appears to result in high rates of diagnostic concordance and accuracy, but only for some eye conditions. It appears to be particularly efficacious in assessment of diabetic retinopathy. Studies of wound care show potential, but are limited by small sample sizes, use of only one assessor, and lacking comparison to other in-person examiners when assessing concordance.
There may be situations when the use of telemedicine is warranted even if the evidence is lacking. For example, there may be situations when care would be otherwise impossible to deliver except via telemedicine. This could include remote rural areas or other locations where medical care is not available locally and the patient is for whatever reason unable to travel to a setting where it can be obtained. However, even in these instances it is important to understand the efficacy of telemedicine so that any clinical shortcomings can be anticipated.
The present evidence base provides guidance on the clinical areas in which future research is most likely to be useful. It now is clear that continued small or methodologically weak studies are unlikely to add to the evidence base for telemedicine. In teledermatology, larger and more comprehensive analyses that assess key patient outcomes are needed. Likewise, there is a need for similar studies of clinical outcomes using office/hospital-based telemedicine in fields such as psychiatry and neurology. Well-designed randomized controlled trials will likely provide valuable information on the potential of these clinical applications. Longitudinal observational studies and demonstration projects also will be useful. Studies of home-based telemedicine should carefully address the independent contributions of technology and human resources in the complex delivery models for patients with chronic diseases.
The goal of this report is to present an overview of the scientific evidence on diagnosis and management decisions, clinical outcomes, and access to services through the use of three categories of telemedicine services: store-and-forward, home-based, and office/hospital-based. The report is intended to provide an update on the evidence for the efficacy of telemedicine services in the Medicare population. Consequently, the scope of this report is limited to telemedicine programs and clinical settings that have been used for, or are likely to be applied to, Medicare beneficiaries.
This report provides an update on the state of telemedicine, following the 2001 publication of Telemedicine for the Medicare Population (AHRQ Publication No. 01-E012).1 Our initial report found that while telemedicine was in widespread use, the evidence of efficacy for those uses was lacking. But even more problematic was the quality of evaluative studies. Many studies were performed using poor methodologies and small sample sizes. Indeed, our major conclusion was not that telemedicine was not efficacious, but rather that the quality of the studies evaluating it prevented one from making that determination at all. Another goal of the current report is to determine whether there has been progress in the number and quality of telemedicine studies that have been conducted since then.
We note that this report only presents a view of telemedicine from the standpoint of the peer-reviewed medical literature. This does not represent the sum of all experience with telemedicine. Indeed, telemedicine continues to be widely used. According to the 2004 TRC Report on Telemedicine Activity (Telemedicine Research Center and Telemedicine Information Exchange),2 48,194 non-radiology teleconsultations took place in 2003 in 46 states. Mental health, cardiology, dermatology, orthopedics, and neurology represent some of the clinical specialties most actively utilizing telemedicine services. This report also identified a number of barriers to telemedicine services. The most significant of these barriers are the difficulty associated with integrating telemedicine services into health care delivery, lack of long-term funding, and lack of reimbursement for the provision of telemedicine services.2
Despite the peer-reviewed literature representing only a subset of all telemedicine experience, it is important to analyze telemedicine using an evidence-based approach, especially when the goal is to inform decisions about coverage. As such, this report undertakes a systematic if limited review of the efficacy for the telemedicine services and usages described above.
Telemedicine is the use of telecommunications technology for medical diagnostic, monitoring, and therapeutic purposes when distance and/or time separates the participants. Some descriptions use the broader term telehealth to indicate care beyond that provided in medical encounters (e.g., health education, health-related Web sites, etc.). Other descriptions use narrower terms focused on medical specialties, such as teledermatology or teleradiology. A telemedicine encounter is the event where clinical services are provided using telemedicine. The narrower term teleconsultation is used when a traditional specialist medical consultation is performed using telemedicine.
This report examines telemedicine services in three areas: store-and-forward, home-based, and office/hospital-based services. Each of these three services are evaluated for their efficacy in three functions: diagnosis and management decisions, clinical outcomes, and access to care. Because the decision to use telemedicine is only predicated on it performing comparably to face-to-face care, studies are assessed from the standpoint of telemedicine to perform comparable to, but not necessarily better than, conventional care.
The terminology used in this update varies from the terminology used in the 2001 report. The term self-monitoring/testing has been replaced with home-based, and the term clinician-interactive has been replaced with office/hospital-based. The updated terms more accurately reflect the services that they describe and match the terminology used in subsequent publication of the findings of the 2001 evidence report in the peer-reviewed literature.3, 4
Store-and-forward telemedicine services collect medical data, store them, and then forward them to be interpreted later. Store-and-forward systems provide the ability to capture and store digital still or moving images of patients, as well as audio and text data. A store-and-forward system eliminates the need to have the patient and the specialist available at the same time. Store-and-forward is therefore an asynchronous, non-interactive form of telemedicine. It is usually employed as a clinical consultation (as opposed to an office or hospital visit).
Home-based telemedicine services enable physicians and others to monitor physiologic measurements, test results, images, and sounds, usually collected in a patient's residence or a nursing facility. Post-acute-care hospital patients, patients with chronic illnesses, and patients with conditions that limit their mobility often require close monitoring and follow-up. These patients also may be taking medications that require testing and/or titration of dosage.
Telemedicine systems use a variety of strategies to accomplish this monitoring. For example, several technologies allow patients to upload monitoring data directly to a health care system or to enter it into a home computer, whereby it can be transferred to a provider. Others make use of high-bandwidth phone or cable television infrastructure to apply two-way interactive video, audio, and medical diagnostic instrumentation. The close monitoring afforded by these approaches may allow better health care through early detection of problems or more precise dosing of medications and biologic agents, potentially reducing costs.
Some common forms of home-based telemedicine services are blood pressure measurement and blood glucose measurement performed by a diabetic patient and used by a clinician to evaluate the patient's glycemic control and to recommend changes in management. Other medical conditions for which home-based telemedicine services have been developed include asthma (in which spirometry is measured), congestive heart failure (weights, symptoms, blood pressure), cardiac arrhythmias (electrocardiography), anticoagulation therapy (prothrombin time), and post-acute hospital care. Monitoring may facilitate preventive measures to be taken before problems get so severe that hospitalization becomes necessary. Telemetry devices could also provide a more cost-effective method of care, by reducing medical visits for conditions that are not severe. Home-based telemedicine systems also may enhance patient-provider communication.
Office/hospital-based telemedicine services are real-time, clinician-patient interactions that conventionally would require face-to-face encounters between a patient and a health professional. Examples of office/hospital-based services that might be delivered by telemedicine include office visits, hospital visits, consultations, and home visits, as well as a variety of specialized examinations and procedures.
Telemedicine is commonly used to make diagnosis or management decisions, often by a specialist located remotely from the patient. Because many diagnostic decisions in medicine are not made definitively, it is often adequate to demonstrate that telemedicine results in concordant as opposed to accurate decisions. This is particularly true in specialties like dermatology, where diagnoses are made by visualizing skin lesions and not confirmed with definitive testing before treatment is begun. Making accurate decisions usually requires more definitive testing, such as biopsies, which are not routinely done for many conditions, especially those that are not life-threatening.
For this reason, we distinguish in this report between concordance and accuracy in making diagnoses. We note other dimensions of concordance, such as the measure used (usually either percent agreement or Cohen's kappa measure for categorical data and correlation for continuous data), comparison to the same diagnostician (intraobserver) versus a different one (interobserver), and whether concordance of telemedicine versus face-to-face diagnosis is compared to face-to-face versus face-to-face diagnosis. Studies of the highest quality must include the latter as a reference to the comparison of telemedicine. In our original report, few studies made this important comparison. Accuracy is usually measured by comparison with some gold standard using sensitivity and specificity.
The clinical outcomes assessed in this study are limited to measures of clinical care, such as health status, improvement in clinical parameters (e.g., blood glucose or blood pressure), and recovery from disease. We do not focus on utilization or economic outcomes.
This review has a number of limitations of scope. Our gathering of evidence is limited to the peer-reviewed literature. Only studies assessing populations relevant to Medicare (non-pregnant adults) are assessed. Studies focusing on non-clinical care (teleradiology and telepathology) are excluded, as are those focusing on economics as well as patient or provider satisfaction. We also do not look explicitly at telemedicine from the context of changes in the health care system or implementation of wider information technologies unless such aspects were specifically assessed in peer-reviewed studies.
Several other reports have analyzed the quality of telemedicine studies, some of which are systematic reviews. These include:
Bashshur R, Shannon G, Sapci H. Telemedicine evaluation. 2004. In: Proceedings of the Second International Symposium on the Future Directions for Telemedicine; 2004 May 22–24; Ann Arbor, MI.
Brantley D, Laney-Cummings K, and Spivack R. Innovation, Demand and Investment in Telehealth. 2004, Department of Commerce: Washington, DC. http://www.technology.gov/reports/TechPolicy/Telehealth/2004Report.pdf.
Currell R, et al., Telemedicine versus face-to-face patient care: effects on professional practice and health care outcomes, in The Cochrane Database of Systematic Reviews. 2000, Update Software: Oxford, UK. CD002098.
Grigsby J, Brega A, DeVore P. Health services research and evidence of telemedicine. In: Proceedings of the Second International Symposium on the Future Directions for Telemedicine; 2004 May 22–24; Ann Arbor, MI.
Hailey D, Ohinmaa A, Roine R. Study quality and evidence of benefit in recent assessments of telemedicine. J Telemed Telecare 2004 10(6):318–24.
Heinzelmann P et al. A review of clinical outcomes in telemedicine/telehealth. In: Proceedings of the Second International Symposium on the Future Directions for Telemedicine; 2004 May 22–24; Ann Arbor, MI.
Hersh W, Wallace J, Patterson P, et al. Telemedicine for the Medicare Population. Evidence Report/Technology Assessment No. 24. Rockville, MD: U.S. Department of Health and Human Services, Agency for Health Care Quality and Research. 2001 Jul. AHRQ Publication No. 01-E102.
Jennett P, Scott R, Hailey D, et al. Socio-Economic Impact of Telehealth: Evidence Now for Health Care in the Future: Volume One: State of the Science Report. 2003, Health Telematics Unit, University of Calgary: Calgary, AB.
Evidence reports aim to define the limits of the evidence, clarifying when assertions about the value of the intervention are based on strong evidence from clinical studies. The quality of the evidence on effectiveness is a key component, but not the only component, of decision-making on coverage decisions. Both national and local Medicare coverage determinations are also based on whether a service has been determined to be “reasonable and necessary” based on “descriptive information, and scientific and clinical evidence.”5
The Medicare Payment Advisory Commission has recognized the value that telemedicine services can add to patient care, particularly following the passage of the Medicare, Medicaid, and SCHIP Beneficiary Act of 2000 (BIPA.)6 However, the potential for overuse of telemedicine services and the need for demonstrated efficacy of telemedicine services prior to making coverage decisions remains an overriding concern.
To determine the key questions and guide the review of the literature in the evaluation of telemedicine, we developed an analytic framework, as shown in Figure 1
We then made explicit the key questions for each of the three study areas. For studies of diagnosis and management decisions and of clinical outcomes, the key questions were assessed from the standpoint of determining whether the telemedicine system provided comparable care, since telemedicine can be deemed efficacious when the quality of clinical care provided is as good as, but not necessarily better than, in-person care.
Store and forward
Does store-and-forward telemedicine result in comparable diagnostic decisions and recommendations for clinical management?
Does store-and-forward telemedicine result in comparable health outcomes?
Does the availability of store-and-forward telemedicine services improve access to care?
Home-based
Does home-based telemedicine result in comparable diagnostic decisions and recommendations for management?
Does the use of home-based telemedicine result in comparable health outcomes?
Does the use of home-based telemedicine improve access to care?
Office/hospital-based
Does office/hospital-based telemedicine result in comparable diagnosis and appropriateness of recommendations for management?
Does office/hospital-based telemedicine result in comparable health outcomes?
Does the availability of office/hospital-based telemedicine improve access to care?
We searched the literature for information about ongoing telemedicine programs, activities, and services. This search focused on obtaining English-language journal articles and reports pertaining to the three study areas. We identified programs from the following:
Electronic bibliographic database. The search strategy was similar to that used in the prior report. It was designed to find any publications about telemedicine and was used to search the MEDLINE® database using Ovid, version 19.2.0. The key MeSH terms associated with the search include telemedicine, remote consultation, telecommunications and delivery of health care. Appendix A * details the complete search string. The initial search, which identified telemedicine articles published between January 2000 and June 2004, resulted in 3,848 citations. An update of the same strategy in November 2004 yielded 224 additional citations. Other databases, such as EMBASE® and CINAHL®, were not searched, since their coverage tends to duplicate MEDLINE® for telemedicine.
Reference lists. Reference lists of previously published telemedicine systematic reviews were searched, resulting in the inclusion of 11 studies to those identified by our search strategy. Of these 11 studies, six were subsequently included following application of the inclusion criteria. Reference lists of included studies also were searched.
Note: Articles may fit into more than one key question category.
| Code # | Explanation |
|---|---|
| Inclusion | |
| 1 | KEY QUESTION AND DATA |
| Addresses a key question on one of the service areas and contains data (results) | |
| Exclusion | |
| 2 | NO KEY QUESTION |
| Does not address a key question | |
| 3 | KEY QUESTION, NO DATA |
| Addresses a key question, but does not contain data | |
| 4 | WRONG POPULATION |
| Addresses key question and contains data, but population of study is outside our scope | |
| 5 | GOOD BACKGROUND MATERIAL and/or REVIEW |
| However, does not meet inclusion criteria | |
| 6 | OTHER |
| Indicate reason | |
| Reviewers | Percent agreement | Kappa value |
|---|---|---|
| WRH, SMS | 89.5 | .56 |
| WRH, TLD | 91.9 | .59 |
| DHH, SMS | 85.8 | .40 |
| DHH, TLD | 91.2 | .42 |
All studies rated as relevant on the basis of review of titles and abstracts were retrieved, photocopied, and distributed to one of the investigators (WRH or DHH). Studies judged to have evidence about a key question were then abstracted. For each key question, data from each study were abstracted using electronic abstraction forms (Appendix B 1), and entered into an evidence table. Evidence tables are presented in full in Appendix C.* A second investigator reviewed all studies included in evidence tables to verify the evidence table content. The study quality ratings of all included studies were assigned at this time (see further details in the next section). A complete list of excluded studies appears in Appendix D.*
| Study Class | Characteristic |
|---|---|
| I | Properly designed random controlled trials |
| II | Random controlled trials that contain design flaws preventing specification of Class I |
| Properly designed trials with control groups not randomized | |
| Multi-center of population-based longitudinal (cohort) study | |
| Case control studies | |
| III | Descriptive studies (uncontrolled case series) |
| Clinical experience | |
| Expert opinion | |
| Case reports | |
| Study Class | Characteristic |
|---|---|
| I | Case series of consecutive patients from relevant population of individuals who would use telemedicine; using an objective gold standard with blinded interpretation of results; with inter-observer analysis |
| II | Case series of patients from relevant population of individuals who would use telemedicine; using an objective gold standard |
| III | Case series not from relevant population or not using appropriate methodology for diagnostic test evaluation |
In appraising studies addressing access to care, we adapted criteria described by the Institute of Medicine (IOM)10 as applied to the use of telemedicine. The model of access to care incorporated three types of indicators: barriers (structural, financial, and personal); utilization; and outcomes (mortality, well-being, or functionality). The IOM has recommended that studies of access to care measure both utilization and outcomes, and our criteria included both measures. Studies that examined only outcomes of care were assigned to the Outcomes category rather than to the Access category.
The definition of access that we used had originally been proposed in a report by the IOM published in 19937 and had been widely disseminated prior to the period of time covered by the studies reviewed for the current report. Other models of access to care have been described,11, 12 but these models include elements of staff deployment and scheduling strategies that have rarely been addressed in studies of telemedicine. Thus, we found the IOM model to be best suited to the published literature in this domain.
| Study Class | Characteristic |
|---|---|
| I | Appropriately comparable comparison group not exposed to telemedicine services; valid measures of utilization and outcomes |
| II | Appropriately comparable comparison group not exposed to telemedicine services; valid measure of utilization |
| III | Comparison group absent or not comparable in some respects; valid measure of utilization; outcomes may also be measured |
| A | Strong improvement or clearly comparable |
| B | Weak improvement or probably comparable |
| C | Conflicting evidence for improvement or comparability |
| D | Negative effect (evidence that technology does not provide comparability or improvement) |
Because of the larger evidence base for studies of diagnosis/management and clinical outcomes, we excluded Class III studies from the analysis in these categories. Class III studies were included in the evidence tables for access to care. We included tallies of Class III studies in all summary tables that show the number of studies and their class and effect for each specialty.
Results of the evidence report update are presented in full in the evidence tables (Appendix C 2). The investigator for each key question constructed separate evidence tables for each of the three study areas. In general, the evidence tables include author/date, key research question(s), study design/level, population, sample/selection, measures, results, quality rating, and limitations.
For the study areas with more than two studies, we constructed a summary table of specialties or domains and the strength of the evidence for each key question and type of telemedicine. The efficacy for telemedicine can therefore be gleaned from the number of studies that have a “positive” direction of effect, i.e., are rated A or B. For those procedures or services that have evidence, the summary tables show which analytic framework links are supported by evidence. We also interpret our synthesis and discuss the limitations of our approach to this evaluation.
Thirteen peer reviewers were selected based on their expertise in the field of telemedicine and their availability to review the draft report. Refer to Appendix E 3 for the list of peer reviewers. The draft report was submitted to the peer reviewers along with a peer review form. The review form was developed by the research team and was based on one used in our original study and those used by other research teams at the Oregon Evidence-based Practice Center. The peer reviewers had three weeks to respond. The comments from the peer reviewers were received and distributed to the investigators for their consideration and response. A spreadsheet was prepared that contained the comments of the peer reviewers and our response to them. The peer review comments aided in creating this improved, comprehensive final document.
| UPDATE | ORIGINAL REPORT | |||||||
|---|---|---|---|---|---|---|---|---|
| Modality | Diagnosis and management | Access | Outcomes | Total | Diagnosis and management | Access | Outcomes | Total |
| Store and forward | 30 | 5 | 0 | 35 | 22 | 2 | 0 | 24 |
| Home-based | 2 | 0 | 25 | 27 | 4 | 2 | 19 | 25 |
| Office/hospital-based | 20 | 9 | 9 | 38 | 33 | 7 | 6 | 46 |
| Total | 52 | 14 | 34 | 100 | 59 | 11 | 25 | 95 |
| Summary by key question | Total | I-A | I-B | I-C | II-B | II-C | II-D | III-B | III-C |
|---|---|---|---|---|---|---|---|---|---|
| Diagnosis and management store-and-forward | 30 | 2 | 3 | 2 | 21 | 2 | 0 | ||
| Diagnosis and management home-based | 2 | 0 | 0 | 0 | 0 | 2 | 0 | ||
| Diagnosis and management office/hospital-based | 20 | 1 | 3 | 1 | 13 | 2 | 0 | ||
| Diagnosis and management total | 52 | 3 | 6 | 3 | 34 | 6 | 0 | ||
| Outcomes store-and-forward | 0 | 0 | 0 | 0 | 0 | 0 | 0 | ||
| Outcomes home-based | 25 | 6 | 2 | 0 | 17 | 0 | 0 | ||
| Outcomes office/hospital-based | 9 | 2 | 0 | 0 | 6 | 1 | 0 | ||
| Outcomes total | 34 | 8 | 2 | 0 | 23 | 1 | 0 | ||
| Access store-and-forward | 5 | 0 | 1 | 0 | 1 | 0 | 0 | 2 | 1 |
| Access home-based | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Access office/hospital-based | 9 | 0 | 1 | 1 | 3 | 1 | 0 | 3 | 1 |
| Access total | 14 | 0 | 2 | 1 | 4 | 1 | 0 | 5 | 2 |
| All total | 100 | 11 | 10 | 4 | 61 | 8 | 0 | 5 | 2 |
Another problem with many of these studies concerns a statistical issue. As noted earlier, the goal of most telemedicine studies is only to show that telemedicine is “comparable” to in-person care, not necessarily “better,” especially when it can provide that care with decreased cost, increased convenience, and/or access to care when none was previously available. One way to determine whether two approaches are comparable is to show there is no statistically significant difference between them. However, lack of statistical significance can also occur even when there is a difference, but the study lacks adequate statistical power to detect it. For this reason, studies with small sample sizes should compute, in addition to an alpha value (the well-known p value), the value of beta. This value estimates the probability that a difference between two comparison groups truly exists when the study results fail to show a difference (also known as a type 2 error). Virtually none of the studies we reviewed assessed beta error, and as such, the “statistical comparability” may exist because there was inadequate statistical power to show otherwise.
In assessing diagnostic and management decision studies, a Class I study had to include not only a comparison of the telemedicine and in-person decisions but also one of the following:
In the case of concordance studies, a comparison of a “baseline” concordance between two or more face-to-face examiners,
In the case of accuracy studies, have measurement against a suitable “gold standard” with measures such as sensitivity and specificity.
Therefore, when there was just a comparison of telemedicine and in-person concordance, the study was rated as Class II. Studies were also rated as Class II when the diagnostic assessment did not include a definitive gold standard consisting of an objective test (e.g., biopsy) or a commonly accepted clinical judgment (e.g., visual findings on gastrointestinal endoscopy or of diabetic retinopathy). Class III studies were excluded from our analysis of diagnosis and management decisions.
For the strength of evidence, a grade of A or B was given when the study set out to demonstrate comparability and did so. Class II studies were not graded higher than B, since studies with this level of methodology do not have the quality of evidence to provide convincingly strong results.
For outcomes studies, a Class I study had to be a randomized controlled trial (RCT). RCTs with clear and obvious flaws were rated as Class II, as were cohort, pre-post, and observational studies. For the strength of evidence, a grade of A or B was given when the study set out to demonstrate comparability and did so, or when the study set out to show superiority of telemedicine and did so. Similar to diagnostic and management efficacy studies, Class II studies of outcomes were also not graded higher than B.
| UPDATE | Total | I-A | I-B | I-C | II-B | II-C | II-D | III-B | III-C |
| Dermatology | 13 | 2 | 9 | 2 | |||||
| Gastroenterology | 2 | 2 | |||||||
| Gynecology | 3 | 2 | 1 | ||||||
| Ophthalmology | 5 | 2 | 1 | 2 | |||||
| Plastic Surgery | 7 | 7 | |||||||
| Total | 30 | 2 | 3 | 2 | 21 | 2 | 0 | ||
| ORIGINAL | Total | I-A | I-B | I-C | II-B | II-C | II-D | III-B | III-C |
| Ambulatory Care | 1 | 1 | |||||||
| Dentistry | 1 | 1 | |||||||
| Dermatology | 9 | 2 | 5 | 2 | |||||
| Neonatology | 1 | 1 | |||||||
| Ophthalmology | 5 | 2 | 1 | 1 | 1 | ||||
| Otolaryngology | 2 | 1 | 1 | ||||||
| Wound Care | 1 | 1 | |||||||
| Total | 20 | 0 | 2 | 0 | 9 | 3 | 1 | 5 | |
The most commonly assessed aspect of teledermatology was interobserver concordance.13–19 The range of concordance varied widely, from 41 percent to 87 percent for complete agreement to 51 percent to 96 percent for disease-category agreement. Unfortunately, all of these studies were limited by the lack of measurement of concordance among more than one face-to-face examiner. In other words, none of the studies compared face-to-face versus telemedicine agreement with face-to-face versus face-to-face agreement. As such, none of the studies could be rated as Class I. In our previous report, two studies did assess concordance of face-to-face examiners.20, 21 Concordance studies assessing management decisions typically looked at decision to biopsy. While one study found complete agreement,22 others found lesser concordance.13, 23–25
The studies of diagnostic accuracy typically compared the telemedicine diagnosis to some sort of gold standard, often a biopsy of a pigmented lesion.13, 23, 26, 27 Most of these studies did not calculate statistical significance, but some did show a trend towards less accuracy for telemedicine.
Store-and-forward applications of teledermatology have generally used commonly available digital cameras and varying techniques for storing and transmitting the digital photographs. Teledermoscopy is a technique by which a low-power lens is used to generate a magnified image of a discrete skin lesion. This methodology was tested in two studies of store-and-forward techniques and found to be comparable to face-to-face diagnosis of pigmented lesions.26, 27
The second most frequently studied clinical area was wound care. Seven studies, all Class II, demonstrated that some characteristics of skin wounds and ulcerations could be assessed effectively using store-and-forward telemedicine. However, most of these studies had small numbers of patients and very small numbers of clinicians, raising the statistical power issues described above.
Five studies provide data on store-and-forward applications in ophthalmology. Four of these studies show that a high accuracy of diagnosing diabetic retinopathy (DR) could be obtained.28–31 One of them found, however, that concordance was lower for severity of DR and specific abnormalities.31
Other specialties studied include gynecology and gastroenterology. The gynecology studies assessing colposcopy were hindered by the limitations of that procedure even when done in person.32, 33
| UPDATE | Total | I-B | II-B | III-B | III-C |
| Dermatology | 2 | 1 | 1 | ||
| Ophthalmology | 1 | 1 | |||
| General Surgery | 1 | 1 | |||
| Multiple Specialties | 1 | 1 | |||
| Total | 5 | 1 | 1 | 2 | 1 |
| ORIGINAL | Total | I-B | II-B | III-B | III-C |
| Dermatology | 2 | 1 | 1 | ||
The studies of access provide information about how telemedicine systems have been deployed in real-world situations and thereby provide an estimate of the actual clinical impact of the systems. All of the studies measured utilization of traditional (non-telemedicine) clinical services following the telemedicine intervention, and all reported the proportion of patients for whom the telemedicine service was the only care received in the index clinical episode. However, two of the studies34, 35 collected no data to assess whether the care provided by the telemedicine service was adequate.
All of the studies that included data on access to care examined the deployment of store-and-forward telemedicine systems for screening patients referred for medical or surgical specialty services following referral by clinicians in primary care or general practice settings. One study36 used only text information submitted by electronic mail, while the other four studies all were based on the collection of digital photographs, usually to supplement conventional clinical information submitted in a text format. The effect on utilization of specialty services was generally modest. In the two studies of teledermatology, more than 80 percent of patients were recommended to have subsequent face-to-face evaluations by dermatologists.37, 38 In a randomized trial of all specialty consultation requests in a rural Finnish community, the electronic mail-based store and forward system had no effect on the proportion of patients who received follow-up care in the local community rather than at the regional centers providing specialty services.36 The two other studies34, 35 used photography-based screening systems by which the majority of patients were recommended not to have specialty follow-up. However, these studies were of relatively low quality and did not collect any follow-up information on the screened patients.
The only study of access to care that was given a Class I rating evaluated a store-and-forward technique for screening primary care patients referred for dermatologic consultation in a Veterans Affairs medical center.38 That study was a randomized trial that included a measure of the time to completion of the consultation. Dermatologists evaluating the patients randomized to teledermatology could determine the time interval to a face-to-face dermatology appointment, while patients randomized to the conventional care group had only a routine appointment scheduled. Patients randomized to the teledermatology group had significantly shorter time intervals until the face-to-face appointment. We judged this to be an unfair comparison, because the study design itself favored improved access to care for the teledermatology group. Since it is likely that the hospital in which the study was conducted had a fixed number of appointment slots for the dermatology clinic, scheduling sooner appointments for patients in the teledermatology group would tend to reduce the pool of available appointments and cause the appointments available to the patients randomized to the conventional care group to be, on average, further distant in time.
| UPDATE | Total | I-A | I-B | I-C | II-B | II-C | II-D | III-B |
| Congestive Heart Failure | 6 | 1 | 1 | 4 | ||||
| Chronic Disease | 3 | 1 | 2 | |||||
| Coronary Artery Disease | 2 | 1 | 1 | |||||
| Diabetes Mellitus | 5 | 1 | 1 | 3 | ||||
| Hypertension | 3 | 1 | 2 | |||||
| Lung Transplantation | 1 | 1 | ||||||
| Multiple Sclerosis | 1 | 1 | ||||||
| Spinal Cord Injury | 1 | 1 | ||||||
| Obesity | 1 | 1 | ||||||
| Psychiatry | 1 | 1 | ||||||
| Pulmonary Medicine | 1 | 1 | ||||||
| Total | 25 | 6 | 2 | 0 | 17 | 0 | 0 | |
| ORIGINAL | Total | I-A | I-B | I-C | II-B | II-C | II-D | III-B |
| AIDS | 2 | 1 | 1 | |||||
| Alzheimer’s | 1 | 1 | ||||||
| Cardiology | 1 | 1 | ||||||
| Chronic Disease | 3 | 1 | 1 | 1 | ||||
| Diabetes Mellitus | 10 | 1 | 1 | 8 | ||||
| Hypertension | 2 | 1 | 1 | |||||
| Neonatology | 1 | 1 | ||||||
| Pulmonary | 1 | 1 | ||||||
| Total | 21 | 3 | 6 | 0 | 10 | 0 | 0 | 2 |
A common characteristic of the studies of home-based telemedicine was that the intervention included dedicated staff (usually nursing staff) that monitored the data recorded in the home and developed clinical management plans. Some of the studies were randomized controlled trials that compared such systems (technology and dedicated staff) to conventional care (such as visiting nurse services). These studies found improved outcomes with the telemedicine-based interventions, but the design of the studies made it difficult to discern the benefit of the dedicated program staff from the telemedicine intervention.
While a small number of the studies were well-designed RCTs,39–43 the rest were limited by either small sample sizes and/or control groups of dubious value. In addition, while all of the studies assessed achieved at least comparable benefits in clinical outcomes, and thus obtained an effect rating of A or B, the value of such comparability (e.g., same but not better blood sugar control or weight loss) was not clear.
Three studies of chronic disease in the elderly showed benefit of the dedicated programs in both patient functional status and reduced emergency department visits and hospital admissions.40–44 Some interventions tailored for specific diseases were found to be effective in congestive heart failure (CHF), hypertension, and pulmonary disease.39, 42–45 Other home-based interventions, such as blood sugar measurements intended to improve management of diabetes mellitus, were not found consistently superior to usual care.46–50 Interventions in other domains, such as obesity51 and lung transplantation,52 also failed to show benefit over usual care.
While two studies were identified in the original report that examined the effect of home-based telemedicine systems on access to care, no studies were identified in the 2000-2004 period. Home-based systems have nearly always been used to enhance the care of patients who already receive conventional clinical services, either through clinic visits or via home care agencies. The primary rationale for home-based telemedicine is to improve data collection and/or communication rather than to supplant conventional care (such as clinic or home visits). Thus, the lack of studies examining conventional measures of access to care is not surprising. Expanded definitions of access to care, such as “patient-centered access”11 are applicable to home-based systems and may provide suitable models for future research on the deployment of home-based systems.
| UPDATE | Total | I-A | I-B | I-C | II-B | II-C | II-D | III-B | III-C |
| Cardiology | 2 | 2 | |||||||
| Dermatology | 1 | 1 | |||||||
| Gastroenterology | 1 | 1 | |||||||
| Neurology | 4 | 4 | |||||||
| Ophthalmology | 5 | 1 | 1 | 1 | 2 | ||||
| Otolaryngology | 2 | 1 | 1 | ||||||
| Psychiatry | 3 | 1 | 2 | ||||||
| Rheumatology | 1 | 1 | |||||||
| Vascular Surgery | 1 | 1 | |||||||
| Total | 20 | 1 | 3 | 1 | 13 | 2 | 0 | 0 | 0 |
| ORIGINAL | Total | I-A | I-B | I-C | II-B | II-C | II-D | III-B | III-C |
| Cardiology | 5 | 1 | 3 | 1 | |||||
| Dentistry | 1 | 1 | |||||||
| Dermatology | 7 | 2 | 5 | ||||||
| Emergency Medicine | 3 | 3 | |||||||
| Neurology | 2 | 1 | 1 | ||||||
| Ophthalmology | 2 | 1 | 1 | ||||||
| Otolaryngology | 2 | 1 | 1 | ||||||
| Psychiatry | 7 | 2 | 5 | ||||||
| Pulmonary | 1 | 1 | |||||||
| Rheumatology | 1 | 1 | |||||||
| Trauma | 1 | 1 | |||||||
| Urology | 1 | 1 | |||||||
| Total | 33 | 0 | 4 | 0 | 15 | 3 | 1 | 2 | 8 |
| UPDATE | Total | I-A | I-B | I-C | II-B | II-C | II-D | III-B |
| Critical Care | 1 | 1 | ||||||
| Neurology | 1 | 1 | ||||||
| Orthopedics | 2 | 2 | ||||||
| Otolaryngology | 1 | 1 | ||||||
| Psychiatry | 3 | 1 | 2 | |||||
| Wound Care | 1 | 1 | ||||||
| Total | 9 | 2 | 0 | 0 | 6 | 1 | 0 | |
| ORIGINAL | Total | I-A | I-B | I-C | II-B | II-C | II-D | III-B |
| Dermatology | 1 | 1 | ||||||
| Emergency Medicine | 1 | 1 | ||||||
| Intensive Care | 1 | 1 | ||||||
| Neonatology | 1 | 1 | ||||||
| Neurosurgery | 2 | 2 | ||||||
| Total | 6 | 1 | 0 | 0 | 3 | 0 | 0 | 2 |
| UPDATE | Total | I-B | I-C | II-B | II-C | III-B |
| Neurology | 2 | 1 | 1 | |||
| Orthopedics/Rheumatology | 3 | 1 | 1 | 1 | ||
| Ophthalmology | 1 | 1 | ||||
| Multiple specialties | 1 | 1 | ||||
| Psychiatry | 1 | 1 | ||||
| Hematology | 1 | 1 | ||||
| Total | 9 | 1 | 1 | 3 | 1 | 3 |
| ORIGINAL | Total | I-B | I-C | II-B | II-C | III-B |
| Neurosurgery | 3 | 2 | 1 | |||
| Cardiology | 2 | 2 | ||||
| Multiple specialties | 2 | 2 | ||||
| Total | 7 | 0 | 0 | 2 | 0 | 5 |
For diagnosis and management decisions, the most commonly studied specialty was ophthalmology. As with store-and-forward studies, some aspects of ophthalmologic evaluation were amenable to interactive telemedicine, while others were not. One Class I study showed rates of disagreement in eye injuries under 10 percent,53 while another found disagreement was consistently higher with telemedicine than when comparing two in-person evaluations.54
Other frequently studied specialties included neurology and psychiatry. Although the studies were rated Class II, two studies showed neurological diagnosis was highly concordant55, 56 and two studies showed that the NIH Stroke Scale could be reliably administered via telemedicine.57, 58 A few Class II studies demonstrated concordance on a variety of psychiatric scales.59–61 Studies in other specialties, such as dermatology,62 rheumatology,63 and vascular surgery64 demonstrated that some diagnostic assessments can be successfully administered interactively via telemedicine.
Studies of clinical outcomes also showed that for most of the clinical specialties assessed, outcomes between conventional and telemedicine interventions were comparable. However, most of these studies were limited by small sample sizes (with the caveats concerning statistical significance described earlier), lack of randomization, and assessment of less than the full range of clinical outcomes. None of these studies attempted to measure statistical power to avoid beta error. Class I RCTs showing comparable outcomes were done in otolaryngology65 and psychiatry.66
The studies of access have examined the use of office-based telemedicine in both suburban and rural settings and have examined both specialist evaluations and follow-up continuity care. In limited studies of patients with sickle cell anemia67 and patients with chronic psychiatric disorders,68 office-based telemedicine appeared to be adequate for the ongoing routine care of patients in rural areas, with few problems reported. For new evaluations by specialists of patients referred by general practitioners, the use of office-based telemedicine led to a significantly greater rate of diagnostic test utilization than face-to-face consultations for neurology patients69 but not for patients needing other types of specialty care.70, 71 Two studies compared office-based telemedicine to telephone consultations between a referring physician and a specialist.63, 72 Both these studies had weak designs but had results suggesting that the telemedicine system provided faster access to definitive care.
A total of 28 studies identified in the literature search were excluded from consideration for evidence tables because they were conducted in populations not eligible for Medicare services based on their demographic characteristics. All of these 28 studies were considered to have evidence potentially applicable to the key questions on the basis of the initial title and abstract review. The populations examined in these studies included children, pregnant women, incarcerated prisoners, and active duty military personnel. These studies were subjected to further review, and twelve were deemed to have evidence suitable for inclusion in evidence tables if we had not applied the population exclusion. Seven of the studies examined home-based telemedicine, three examined store-and-forward techniques, and two examined office-based telemedicine applications. The studies' findings were consistent with the findings of the studies included in the evidence tables. We concluded that excluding studies conducted on non-Medicare populations had not biased our overall conclusions.
This update on evidence about the efficacy of telemedicine for the Medicare population covered published peer-reviewed literature for the five years between 2000 and 2004. Similar to the findings of our original report a half-decade ago, there are still serious gaps in the evidence base for telemedicine. While this situation is hardly unique to telemedicine, having a solid evidence base is essential given that there is increased advocacy for health care payers, especially Medicare, to provide coverage for its use. In the discussion that follows, we will review telemedicine by specialty in the case of store-and-forward and office/hospital-based telemedicine and by disease domain in the case of home-based telemedicine.
The best evidence for the effectiveness of telemedicine is in medical specialties for which verbal interactions are a key component of the patient assessment, such as psychiatry and neurology. Various psychiatric and neurological assessments can be administered effectively via interactive videoconferencing. Likewise, treatments administered in these specialties via telemedicine appear to achieve comparability with face-to-face care. It can probably be concluded that medical care administered via interactive videoconferencing can achieve results that are comparable to their in-person counterparts.
Our systematic review also identified several studies, a few of them of high methodologic quality, showing benefits of home-based telemedicine interventions in chronic diseases. These systems appear to enhance communication with health care providers and provide closer monitoring of general health, but the studies of these techniques were conducted in settings that required additional resources and dedicated staff. Deployment of home monitoring technology in the absence of these integrated systems is unlikely to be beneficial. Systems designed to facilitate specific aspects of care, such as blood sugar and blood pressure measurements, provide less clear benefit. With ongoing improvements in telecommunications technology, particularly broadband connections to the home, further research, including larger clinical trials, will likely be informative.
The specialty with the largest number of studies is dermatology, and most studies of teledermatology have evaluated store-and-forward techniques. The body of evidence summarized in this report is consistent with the findings of the earlier report. There continues to be highly variable rates of interobserver and intraobserver agreement in diagnoses. This issue can only be resolved by high-quality studies that compare not only the concordance of telemedicine versus face-to-face diagnosis, but also the concordance of face-to-face versus face-to-face diagnosis in the same situation. It should be noted that the teleophthalmology field has done this in most of their diagnostic concordance studies.
The published studies of teledermatology have other flaws as well. For example, most of them included only a small number of teledermatologists. Over half of the studies we identified used three or fewer teledermatologists. In addition, most of the studies deployed teledermatology only in a laboratory type of setting. The few studies of real-world use of teledermatology found that most patients required subsequent face-to-face clinical encounters. Thus, it appears that the expense and time commitment of teledermatology systems have not yet demonstrated the potential for improving access to care.
Of course, rates of concordance in a vacuum, i.e., without a clinical context of how the patient fared, are also limited from an evidence standpoint. What we ultimately need to know is the patient outcome. In other words, did the teledermatology encounter at least provide comparable care for the patient? A corollary question that must be answered is whether teledermatology resulted in harm from any missed diagnoses or other aspects of the telemedicine situation. These questions can only be answered in studies of clinical outcomes, none of which were identified in this report. One study has been completed (personal communication, J Whited), but at the time of this writing has not yet been submitted for publication. In general, advocacy for an expanded role for teledermatology will require further studies that examine rates of missed diagnoses, incorrect treatments, and when the technology is insufficient to avoid in-person encounters.
Most published studies of teledermatology have examined store-and-forward techniques, with relatively few studies of real-time office-based techniques. Despite its current widespread use, additional evidence is required to conclude that store-and-forward teledermatology can be routinely substituted for face-to-face encounters in the evaluation of new referrals to dermatologists. Dermatologic practice also involves follow-up visits of patients who have previously received a comprehensive dermatologic evaluation. Store-and-forward teledermatology may be better suited to such follow-up visits, but there have been no published reports of experience with this type of visit. Store-and-forward techniques also may be a useful adjunct in dermatologic consultations for settings in which patients are located a great distance from the consultant (such as isolated rural settings). The published studies suggest that a fraction of such patients may successfully avoid face-to-face visits to complete the dermatologic evaluation.
Teleophthalmology has been widely studied, and this field has produced commercial systems for retinal photographs that are becoming widely used as a clinical tool to augment face-to-face evaluations of patients at risk for diabetic retinopathy.73 The quality of studies in this domain is slightly higher than in dermatology, although the results are equivocal. Essentially, teleophthalmology results in high rates of diagnostic concordance and accuracy for only some eye conditions. It appears to be most efficacious for the assessment of diabetic retinopathy. However, there are a number of diagnoses for which it fares less well, and it is often unusable altogether when certain patient characteristics are present, such as cataracts and other lens abnormalities. The value of a technology that is only useful for some conditions must be assessed in the larger picture of clinical outcomes and ultimately the economics of investing in equipment that is not always useful.
Also widely studied is the use of telemedicine for wound care. The key observation from studies assessing telemedicine for this purpose is that all of the studies have small sample sizes, use only one assessor, and do not compare in-person examiners when assessing concordance. These studies present a trend of comparability, but serious questions remain about their statistical power and reproducibility.
The situation of gynecology, in particular telecolposcopy, is also instructive. Studies show that the accuracy of diagnosis by telemedicine is comparable to face-to-face assessment. However, the accuracy of neither of these approaches is terribly efficacious, in the range of 50–60 percent.
An often-touted benefit of telemedicine is the provision of care to rural areas, where specialists are less prevalent and individuals in need of them must travel great distances to see them. Studies of rural populations have tended to be of poorer methodological quality than studies of urban and suburban populations. The limited evidence available supports the use of office/hospital-based telemedicine for providing continuity care of stable patients by specialists. The technological platform for such systems is relatively uncomplicated and can be based on widely available teleconferencing equipment.
In general, the role of telemedicine most likely to demonstrate value could be as an adjunct to care that is centered around the in-person visit. As noted above, it could, for example, serve as a means to triage skin lesions, injuries, and other problems that arise where appropriate specialty care is not available. In most instances, clinical care will likely still require in-person diagnosis and management. Likewise, telemedicine may also play a role in managing the growing number of elderly and other infirmed individuals with chronic diseases. Its value may not be as a substitute for in-person care as much as an adjunct to it.
Experience with telemedicine has similarities to other attempts to apply computer technology to clinical environments. Computer-based expert systems went through a long period of experimental evaluation and limited deployment, and systems designed to enhance (or even replace) clinician judgment were found to be best suited to narrow clinical domains.74–76 Nevertheless, the use of computer-based decision support has steadily increased, as it has been more appropriately integrated into the clinical care process, serving more as an assistant than a replacement of clinical judgment and expertise.77 Well-designed and definitive clinical trials of this type of decision support have enhanced the adoption of the technology.78, 79
As noted in Chapter 1, this review is limited in scope to the peer-reviewed literature of telemedicine. While this spectrum of data does not cover all or even most of the experience with telemedicine, it does provide the most objective, evidence-based assessment of this technology. The presence of a small number of well-designed studies with positive outcomes that our analysis identified shows that it is possible to demonstrate efficacy of telemedicine.
Of course, there may be situations when the use of telemedicine is warranted even if the evidence is lacking. For example, there may be situations when care would be otherwise impossible to deliver except via telemedicine. This could include remote rural areas or other locations where medical care is not available locally and the patient is for whatever reason unable to travel to a setting where it can be obtained. However, even in these instances it is important to understand the efficacy of telemedicine so any clinical shortcomings can be anticipated. We are reassured that no studies show telemedicine to cause any significant harm.
We also acknowledge that the efficacy of telemedicine is not immune to other forces in health care, such as the structure of the delivery system. There are instances when reimbursement or other incentives are not amenable to innovations, technical or otherwise. In particular, fee-for-service health care will likely provide incentive for modalities of care that are reimbursed, not necessarily those that provide the best quality care. Therefore it will be important for ongoing research to take into account the health care setting when evaluating the efficacy of telemedicine.
The present evidence base provides guidance on the clinical areas in which future research is most likely to be useful. It now is clear that continued small or methodologically weak studies are unlikely to add to the evidence base for telemedicine. In teledermatology, larger and more comprehensive analyses that assess key patient outcomes are needed. Likewise, there is a need for similar studies of clinical outcomes using office/hospital-based telemedicine in fields such as psychiatry and neurology. Well-designed RCTs will likely provide valuable information on the potential of these clinical applications. Longitudinal observational studies and demonstration projects also will be useful. Studies of home-based telemedicine should carefully address the independent contributions of technology and human resources in the complex delivery models for patients with chronic diseases.
We recognize the limitations of advice that telemedicine be studied with more and larger RCTs. Not only are such trials expensive, but they are difficult to carry out. They also have a long lead-time from their planning and inception to completion and analysis of results. Another challenge with RCTs in this area is that telemedicine is not a single technology or intervention. It is a tool that is used to deliver different aspects of clinical care for diverse diseases. Due to the time and expense of RCTs, other means to assess telemedicine interventions objectively should also be explored. Given the growing use of electronic health records, selective data could be extracted on patients with telemedicine interventions to assess them longitudinally. Such studies will be most feasible in large integrated delivery networks with advanced electronic health record systems.
This report has found that the evidence base for telemedicine is incomplete yet improving. Further well-designed and targeted research that provides high-quality data will provide a strong contribution to understanding how best to deploy technological resources in health care.
The promise of telemedicine is not matched by the strength of its evidence base. The technology to administer telemedicine is prevalent and, in some locations, ubiquitous. Telemedicine is widely used, with increasing numbers of health care payers reimbursing for its use. However, outside of a small number of clinical specialties, the evidence base for the efficacy of telemedicine is weak. Areas where telemedicine is most promising include home health and specialties where care can be delivered via interactive videoconferencing, such as psychiatry and neurology. There is mixed evidence for the efficacy of telemedicine in dermatology and ophthalmology. Further research must address the limited evidence base so that the optimal use of telemedicine can be ascertained.
Database: Ovid MEDLINE(R) Version: re l9.2.0
exp TELEMEDICINE
telemedicine.mp.
telehealth.mp.
remote consultation$.mp.
1 or 2 or 3 or 4
exp Home Care Services
Home Nursing
6 or 7
exp Therapy, Computer-Assisted
exp COMPUTERS
exp Computer Communication Networks
exp Medical Informatics
exp TELECOMMUNICATIONS
exp Monitoring, Physiologic
monitor$.mp.
blood glucose self-monitoring
Self-Examination
self exam$.mp.
self monitor$.mp.
self test$.mp.
14 or 15 or 16 or 17 or 18 or 19 or 20
tele$.mp.
(remote or offsite or distance).mp.
Rural Population
Rural Health Services
HOSPITALS, RURAL
rural.mp.
22 or 23 or 24 or 25 or 26 or 27
21 and 28
9 or 10 or 11 or 12 or 13 or 29
8 and 30
31 not 5
limit 32 to english language
32 not 33
limit 34 to abstracts
33 or 35
5 or 36
limit 37 to yr=2000-2004
exp Computer Communication Networks
Patient Participation
exp Consumer Satisfaction
“Delivery of Health Care”
exp Home Care Services
exp Home Nursing
house calls/ or house call$.mp. or housecall$.mp.
40 or 41 or 42 or 43 or 44 or 45
39 and 46
limit 47 to english language
limit 48 to yr=2000-2004
38 or 49

Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]Individuals
| Reviewer | Affiliation |
|---|---|
| Jeanette C. Hartshorn, PhD, RN, FAAN | UTMB Telehealth Center |
| University of Texas | |
| Galveston, TX | |
| Michael A. Hillman, MD, MBA | Marshfield Clinic |
| Marshfield, WI | |
| Penny Jennett, PhD | Health Telematics Unit |
| University of Calgary | |
| Calgary AB | |
| Bonnie J. Wakefield, PhD, RN | Center for Research in the Implementation of Innovative Strategies in Practice (CRISP) |
| VA Medical Center | |
| Iowa City, IA | |
Organizations
| Organization/Agency | Reviewer | Affiliation |
|---|---|---|
| American Academy of Dermatology | Hon Pak, MD | Department of Dermatology |
| Brooke Army Medical Center | ||
| Fort Sam Houston, TX | ||
| American Telemedicine Association | Nina Antoniotti, PhD, MBA, RN | Marshfield Clinic Telehealth Network |
| Marshfield, WI | ||
| Association of Telehealth Service Providers | Josie Henderson | Telemedicine Research Center |
| Portland, OR | ||
| National Association for Home Care & Hospice | Theresa Forster | Vice President for Research NAHC |
Federal Agencies
| Agency for Healthcare Research and Quality | J. Michael Fitzmaurice, Ph.D., FACMI | Senior Science Advisor for Information Technology |
| 540 Gaither Road, Suite 3026 | ||
| Rockville, Maryland | ||
| Agency for Healthcare Research and Quality | Martin Erlichman, MS | Task Order Officer |
| 540 Gaither Road | ||
| Rockville, MD | ||
| Centers for Medicare & Medicaid Services | Shamiram R. Feinglass, MD, MPH | Medical Officer |
| Medicare Coverage & Analysis Group | ||
| 7500 Security Boulevard | ||
| Mailstop C1-09-06 | ||
| Baltimore, MD | ||
| Indian Health Service | Mark Carroll, MD | Flagstaff, AZ |
| U.S. Department of Health and Human Services | Dena Puskin, ScD | Office for the Advancement of Telehealth |
| Rockville, MD | ||
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]
Free Full text in PMC]Appendices cited in this report are provided electronically at http://www.ahrq.gov/downloads/pub/evidence/pdf/telemedup/telemedup.pdf.
Appendices cited in this report are provided electronically at http://www.ahrq.gov/downloads/pub/evidence/pdf/telemedup/telemedup.pdf.
Appendices cited in this report are provided electronically at http://www.ahrq.gov/downloads/pub/evidence/pdf/telemedup/telemedup.pdf.
Appendices cited in this report are provided electronically at http://www.ahrq.gov/downloads/pub/evidence/pdf/telemedup/telemedup.pdf.
Evidence tables cited in this report are provided electronically at http://www.ahrq.gov/downloads/pub/evidence/pdf/telemedup/telemedup.pdf in Appendix C.
Evidence tables cited in this report are provided electronically at http://www.ahrq.gov/downloads/pub/evidence/pdf/telemedup/telemedup.pdf in Appendix C.
Evidence tables cited in this report are provided electronically at http://www.ahrq.gov/downloads/pub/evidence/pdf/telemedup/telemedup.pdf in Appendix C.
Evidence tables cited in this report are provided electronically at http://www.ahrq.gov/downloads/pub/evidence/pdf/telemedup/telemedup.pdf in Appendix C.