Bill Grigsby, Ph.D. Telemedicine Research Center, Portland, OR. Bill.Grigsby@atsp.org
Ace Allen, M.D., Editor.

Keeping tabs on telemedicine programs is like keeping track of a busload of campers at an amusement park. They seem quite manageable as they get off the bus. Then they’re off, only to reappear occasionally, then off again in a new direction. At the end of the day they return to the bus, some full of wild stories, others matter-of-fact, and others tired to the point of silence.

This is our 4^th year of tracking down the unruly statistics of telemedicine activity. The first survey (Vol. 1, #4; 1993) was simple, when an afternoon on the telephone yielded the stats on the 12 active programs in the U.S. and Canada. With few exceptions there was only one technology: interactive video with room units or rollabout units. There was little teleradiology, no telepathology, no tele-home health care, and few store-and-forward (S&F) consultations. There was no low-bandwidth consultative telemedicine (i.e., via ISDN or regular phone lines), and no desktop telemedicine. There was very little telemedicine outside North America and Norway.

In the ensuing years we have made it our mission to fairly profile the face of the industry through in-depth surveys, to present the primary data to our readers, and to let the numbers speak for themselves. Some examples of important trends identified from past surveys:

Most of the early programs went through a 2-year "plateau" phase, when telemedicine activity remained stagnant. This was followed by a significant increase in activity presumably due to improvements in program management and organization.

Many of the most active programs were for care of prisoners, probably because the cost-benefit justification is so clear, and because most of these are managed care contracts rather than fee-for-service teleconsultations (which continue to be unpredictably reimbursed).

Most interactive video (ITV) programs operated at 384 Kbps, which was sufficient for the great majority of teleconsultations and tele-education.

Most programs used about 50% of their aggregate telemedicine network time for consultations. The remainder was for medical education, administration, and business rental.

This year’s survey, a cooperative effort of Telemedicine Today and the Association of Telemedicine Service Providers (ATSP), was more ambitious than in years past. While it was most directed to collecting information on ITV and S&F-mediated patient-clinician consultations, the survey also addressed teleradiology, pathology, and tele-home health care. The survey was directed only to programs in the U.S. and in Norway. A review of Canadian telemedicine was published in the last issue of Telemedicine Today.

We are very grateful to our survey respondents for their often time-consuming commitment to accuracy and completeness. They have contributed in an important way to the understanding and advance of telemedicine.

Identifying the Programs

A master list of programs was identified from the extensive files of the Telemedicine Information Exchange (TIE; http://tie.telemed.org/), the ATSP, and the Telemedicine Today databases. In late May, 1997 a query letter was sent to 334 personnel at active or potentially active programs in the U.S. In most cases these were the identified telemedicine coordinators or administrators; in cases where the proper contact person couldn’t be identified mailings were sent to more than one potential respondent. Targets included University, military, VA, prison and private hospital programs, as well as public and private clinics and tele-home health agencies.

Programs that saw 50 or more consultations in 1996 were asked to complete a 3-page survey; an overview of these responses is found in Table _____ . For programs seeing fewer than 50 consultations, the number (both radiology and non-radiology) done in 1996 and the first four months of 1997 was solicited (Table _____). These less active programs were not asked to complete the more in-depth 3-page questionnaire, although many chose to do so.

Non-responders to the initial mailed survey were sent a postcard reminder in mid-June. Known active programs that still had not responded were sent a second survey in late June. Persistent nonresponders were then contacted by FAX, telephone, and email. In the end, we received information from 87 programs that were active in 1996 and/or 1997.

What Was Measured?

The scope of telemedicine has increased markedly since our last survey. In the past year, there has arisen a much broader variety of interactions between patients and clinicians, and between clinicians. This matter of changing study conditions causes problems when designing and analyzing a survey, since serial comparisons are maximally reliable only when the study conditions remain stable. However, we felt that a survey that tried to distinguish, for example, between patient-physician telecardiology and physician-to-physician "curbside consult" telecardiology would be too cumbersome. In the past, we tried to make this distinction, often unsuccessfully. Thus, there may be a tendency to "lump" types of interactions that hitherto have been separated out. This makes some comparisons between our findings this year and in previous years a bit more difficult. We caution the reader to realize that there is an aspect of comparing ‘apples and oranges’ when contrasting 1996 data to 1995 and before. In general, we have tried to err on the side of conservatism in our reckoning and in the conclusions we’ve drawn.

It is becoming increasingly difficult to distinguish between "hub" and "spoke" sites. Many "spoke" sites that present cases to a tertiary "hub" also, in turn, provide telemedicine services to even smaller clinics – becoming effectively hubs themselves. In a few cases we received survey responses from "spoke" sites whose numbers had already been accrued into the responses of a larger "hub" site. In these cases, we did not list the spoke site as an independent program.

Tele-home health interactions were logged as number of patients followed/year, rather than as number of consultations. We did this because tele-home nursing interactions can be extremely frequent (twice or more daily) and we felt it might be misleading to pool the data from these very specialized applications with the more general patient-clinician data. Also, some programs track only the number of patients followed, rather than the number of interactions.

Caveats

A few known active programs did not respond to this survey, although we believe that the numbers they would have contributed would have not significantly altered our findings. We believe that this survey provides a very fair overview of the state of clinical, consultative telemedicine in the U.S. today. Although teleradiology, telepathology, and tele-home health care coverage were not as comprehensive, the results are included here and provide useful information.

The figures tabulated here have not been independently verified by Telemedicine Today nor by the ATSP. In several cases, obvious discrepancies in survey responses were discussed with the respondent and corrected as necessary. Any questions about individual program responses should be directed to the contact person identified in the appropriate Table.

The Findings

(U.S. only – patient-clinician interactions, excluding teleradiology)

Programs

Eighty-seven programs were identified as doing patient-clinician interactions in 1996 and/or the first four months of 1997. In 1996 there were 80 active programs in 38 states and Washington, D.C. Of these 80 programs, 72 (including 3 programs that did only tele-home health nursing) did interactive video (ITV) mediated consultations, some of which may have incorporated store-and-forward (S&F) technology as well. The other eight programs did only S&F interactions (See Table___).

Among the 80 programs tracked in 1996 there were 1,032 sites supporting telemedicine interactions; about 92% of these were "spoke" sites, the remainder were "hubs" – recognizing that the distinction between "hub" and "spoke" is not always absolute.

ITV vs. S&F

In 1996, there were a total of 21,274 patient-clinician interactions, excluding teleradiology and tele-home health (which are treated separately below). There were 69 programs that did a total of 19,380 ITV-mediated consultations, or consultations involving both ITV and S&F. A total of 1,894 consultations were S&F only. Thus, 91.1% of interactions involved ITV technology, and 8.9% involved S&F only.

Breakdown of 1996 telemedicine interactions:

ITV only: 17,395 (see below for details)

S&F only*: 1,894 (63% cardiology; 15.5% dermatology; 12.2% neurology; 6.6% pathology; 2.7% misc.)

Both ITV and S&F: 1,985 (26.8% ophthalmology; 21.2% dermatology; 13.4% cardiology; 11.0% pediatrics; 7.7% pulmonary; 6.5% orthopedics; 13.4% misc.)

*One program, Worldcare, reported 7,347 S&F "clinical drug trials." Because we were not sure that these fell within the usual understanding of a clinical interaction, we elected to delete them from our tallies.

ITV Activity

The 19,380 non-teleradiology ITV teleconsultations in the U.S. in 1996 compares to the 6,138 we documented in 1995 – a tripling of activity in a year. At the same time, the number of active ITV-mediated programs (excluding programs doing only tele-home health care) has increased from 38 in 1995 to 69 in 1996. The number of sites (hubs+spokes) has more than doubled from 442 to 993. (See Figures ___ and ___)

In the first four months of 1997 we documented 12,122 ITV consultations among 79 programs (including 5 that had not been active in 1996). Straight-line extrapolation suggests there will be over 36,000 consultations in 1997, approximately double the 1996 figure.

The median U.S. ITV-mediated program doing 50 or more consults in 1996 (n=47; programs doing only tele-home health excluded – see below) had 8 sites (range: 2-300). The average number of sites/program was 17.0, skewed upwards by Viewsend with its 300 sites. If that outlier is excluded from the calculation, the average drops to 10.8 sites/program. The average program seeing 50 or more consults in 1996 saw (18,766 consults)/(498 sites) = 37.7 consults/site/year, or about 3/site/month (Viewsend excluded from calculation). This is a considerable improvement since 1995, when the average consults/site/year for programs seeing 50 or more consults was (5,824 consults)/(352 sites) = 16.5 consults/site/year. This metric is important, since it can serve as a weak proxy for program efficiency. Given all else equal (a big given!), a program that sees more patients/site/year is likely making better use of its fixed resources than one that is less active.

Figure ___.

The Most Active ITV-Mediated Programs – And The Least

(Tables ___ and ___)

The 10 most active programs saw 10,661 ITV-mediated consultations, or over half the total. Indeed, the top 5 programs saw 6,758 – over one third of all consultations, and more than were seen by all 38 active programs in 1995. These top programs have clearly addressed the problems of low utilization that have bedeviled programs in the past. Most have developed regularly scheduled (in addition to "prn") clinics in a contractual, managed care setting. Four (UTMB, OH State, Arapahoe, Tm Systems Corp) are prison programs; their 4,188 consults account for almost 22% of the total in 1996. On the other hand, there were 22 programs doing fewer than 50 ITV-mediated consults/year in 1996 (tele-home health programs excluded). These accounted for only 614 consults, or about 3% of the total. While most of these less active programs are relative newcomers, many have been operational for at least three years.

Table _______

ITV-Mediated U.S. Telemedicine Programs (non-teleradiology) with <50 Consultations in 1996

(arranged by activity level):

S&F Mediated Programs (Table ___)

Aside from teleradiology, store-and-forward technologies have not been much utilized for telemedicine consultations. This is beginning to change, especially for cardiology (see Vol. 4, No. 3), dermatology, and most recently ophthalmology (see Vol. 4, No. 5) where S&F and ITV techniques are combined. In 1996 we documented 1,894 non-teleradiology S&F consultations. Of these, 1,359 were done by programs that did only S&F; the remaining 535 were done by programs whose major activity was ITV-mediated.

Table____. Programs Doing Only S&F Consultations in 1996 (excluding teleradiology)

Most Common Specialties

(Table _____)

As in years past, mental health consultations remain by far the most common type, probably because they are less technology-intensive, don’t demand many (or any) peripheral devices, and can be done effectively over low bandwidths (128 Kbps). They accounted for 21.3% of consultations in 1996. However, a ‘new kid on the block’ arose for second place: emergency/triage, accounting for 15.9% (2,574) of consultations. Hitherto there has been extremely little documented use of telemedicine for emergencies. While the very high numbers accrued by the DigitalCare System of Tulsa (1,392 emergency consultations) account for over half of this, there were 10 other programs doing emergency consultations, suggesting that this isn’t just a one-program phenomenon. The third most common specialty was cardiology (12.4% of the total), followed by dermatology (11.1%). Number five was a surprise: surgery consultations (8.3% of the total, of which 74.4% were orthopedic; the balance were plastic, hand, and urology). While surgery itself is not being done much via telemedicine, and most of that is ‘telementoring’ (see vol. 5, no. 3), there is a great deal of pre- and post-surgical evaluation, primarily in a prison setting. These top 5 specialties account for nearly 70% of all activity.

Table ____

External support

About 60% of these programs (data not shown) were identified as being initially supported by substantial -- typically greater than 50% of start-up funds -- federal (40.7% of programs) or state (19.8%) monies. Interestingly, of the seven programs begun in 1997, only one is federally funded. Two others are supported by state funds. Of the 10 most active U.S. programs in 1996, only one (APPAL-LINK) received significant federal support. This suggests that federal support does not assure an active program (See "External vs. Internal Funding: Does It Matter Who Pays the Bills?" Vol. 5, no. 1). This may be because federal grants have tended to support projects that provide service to rural areas that may be by their nature low-volume. However, many of the most active programs (DigitalCare, UTMB-G, U of KS, TX Tech) receive no federal support yet cater almost exclusively to rural areas.

Tele-Home Health Care

(Table ___ )

Ten respondents reported doing tele-home health care in 1996-7. Three did only tele-home health care; for the seven others tele-home health was incorporated into their other telemedicine services. One program (Tele-Home Health, Sacramento, CA) is reported on in this issue. A report on another (H.E.L.P. Innovations, Lawrence, KS) was done in Vol. 3, No. 3 and will be updated this fall (Vol. 5, #6).

Table____. Interactive Video-Mediated Tele-Home Health Programs, U.S., 1996-7

Arranged by activity level

*Did tele-home health only **POTS = Plain Old Phone System

This survey was not specifically designed to capture home telemonitoring services, and very likely missed some important tele-home health programs. It is interesting to note that for the majority of these programs simple, inexpensive phone lines appear to suffice.

Teleradiology (Tables___ and ___)

This survey was not designed specifically to quantify teleradiology activity. Nevertheless, 15 programs reported doing teleradiology. (Fig: Teleradiology Reads: 1996) Of these, 6 programs identified teleradiology as their primary activity. In all, 13,653 teleradiology cases were read, over 98% in store-and-forward format and as plain films or CT/MRI scans. Telemedicine Today’s last complete overview of teleradiology activity was done a year ago (Vol. 4, #6, 1996). We estimated then that among 13 programs in the U.S. about 115,000 interpretations would be done via teleradiology in 1996. The great majority were diagnostic rather than preliminary (interim) reads. Quantifying preliminary reads, typically done from the radiologist’s home after hours, is extremely difficult because the technology is becoming ubiquitous. In 1996, the average program interpreted about 9,000 studies/year (median: 3,960). Among the 6 U.S. programs we now report on whose primary activity was to interpret plain films/CT/MRI scans, the average yearly studies was 2,053 (median: 1,325). The U.S. programs reported here probably represent well under 10% of total teleradiology activity in 1996.

Table ___

Teleradiology Activity in 1996 Among 15 programs: ITV vs. S&F vs. Both

*nearly all of these are plain films and CT/MRI scans

Table ____

6 U.S. Programs Doing Teleradiology as their Primary Activity in 1996 (ordered by activity level)

*S&F readings are nearly always plain films/CT/MRI

Barriers to Sustainability

(Table___)

Seventy-two U.S. programs active in 1996 responded to a query about barriers to program sustainability. The standouts: reimbursement and the cost of telecommunications. As noted in another article in this issue (Mintzer), average monthly cost for point-to-point digital services may range from $157 to nearly $4,000, depending on distance and bandwidth. High installation costs bring the overall cost of digital telecommunications even higher.

Table ____

The 9 Most Important Barriers to Program Sustainability

(as identified by 72 U.S. programs active in 1996)

* one apiece: Contract renewal; tech support; technol availability; network accommodation; lack of standards; provider relations; health care integration

Bandwidth Issues (Table___)

We received responses from 86 programs in the U.S. about the primary bandwidth used. These include S&F and teleradiology systems. Because many programs are now using different bandwidths for different applications, it is becoming more difficult to come up with a discrete answer to a question such as, "What is the most common bandwidth used for interactive video telemedicine consultations?" The most commonly used bandwidths continue to be º-T1 (384 Kbps) and T1 (1.54 Mbps). Among the top 10 most active ITV-mediated programs, four transmitted primarily at 768 Kbps, two at 384 Kbps, two at 128 Kbps, and one each at 1.54 Mbps and at 56 Kbps. There is definitely an increase in the use of lower bandwidths, particularly 128 Kbps and POTS. This is probably because of the increasing availability of ISDN, and because many telepsychiatry programs are finding that the lower bandwidths are adequate, and are relatively inexpensive.

Table_. Bandwidth Used Most Frequently by S&F and ITV Telemedicine Programs in the U.S.1996-7

*Plain Old Telephone System; includes tele-home health programs

Norway

The Department of Telemedicine at the University Hospital of Troms¯ in northern Norway has been at the forefront of telemedicine since 1989. Their activity has been tracked by Telemedicine Today since 1993. Correspondent Steinar Pedersen, M.D. provides the following statistics for 1996:

* Most common transmission bandwidth: 384 Kbps

* Total consultations: 9,000

* Total non-radiology consultations: 1,000

* Breakdown of non-radiology consultations:

*      dermatology: 500

*       ENT: 250

*       mental health: 200

*       pathology: 50

* Most important barriers to program sustainability: lack of standards; lack of electronic medical records.

Trends and Conclusions

* Location: Until 1995 ITV-mediated telemedicine was demonstrably Midwest-centric. There were very few programs on the east or west coasts. This has changed considerably, perhaps because telemedicine is evolving from a specific solution for improving medical access in rural areas to a more general solution that applies to metropolitan areas as well. However, in terms of activity levels, the center of gravity remains in the Midwest. Of the 21,274 non-teleradiology interactions done in 1996, 12,533 (about 60%) took place between the Appalachians and the Rockies, and outside the South.

* Program number & activity: The number of programs in the U.S. continues to (approximately) double yearly, and program activity has tripled since 1995. This continues the trend of yearly doubling that we have documented since our First Annual Survey in 1993. (Figures __ and __ ) There is no evidence to suggest this trend will slow down.

* Utilization: Utilization per site is increasing, having approximately doubled over the past year. This may be due to more efficient clinic scheduling and better clinician acceptance, although there may also be some artifact to this statistic. Here’s why: for these surveys, the number of sites in a program was tallied whether the site had been active the entire year (thus able to accrue patients for the year) or had been activated only on the last week of the year (thus able to accrue only a few, or no, patients). Clearly, the average activity/site will be much less if a significant portion of the sites came on line late in the year.

* Prisons: As in previous years, prison telemedicine accounts for a significant percent of overall activity. Prison telemedicine has especially compelling cost-benefit considerations because of the averted costs – and risks – of patient transfer.

* Military: We were disappointed not to receive responses from more military telemedicine programs. Only AKAMAI responded this year. In past surveys military programs have reported extremely low utilization.

* Specialties. While telemental health continues to be the most common application, accounting for over 20% of all consults in 1996, and cardiology is second, emergency medicine has emerged for the first time as a numerically meaningful application. (Table ___)

* Bandwidth. 384 Kbps transmission for ITV interactions still reigns as the acceptable compromise for enabling adequate resolution and motion handling while minimizing bandwidth cost. Increasingly, however, programs are turning to lower bandwidths (128 Kbps ISDN) for applications such as telepsychiatry.

* Modality. For non-teleradiology applications, only 9% were S&F only. The great majority of interactions included interactivity. There may be a trend toward more S&F applications as software improves and as teledermatology, pathology, and ophthalmology become more accepted within their specialties. Cardiology stood out as a "S&F only" application, while ophthalmology and dermatology seemed especially suited to a combination of S&F and ITV.

Telemedicine deployment is continuing unabated, on the same trajectory it has followed since 1993. Individual programs are becoming much more adept at using their telemedicine infrastructure, with increasing consultations/site/year and a broadening base of specialties and technologies. There appears to be decreasing dependence on external (federal) funding, which may bode well for improved self-sufficiency. The most important perceived barriers to program sustainability are reimbursement and telecommunications costs. These barriers are dropping rapidly.

Finally, the concept of a "telemedicine program" is evolving. While all early deployments emerged within the structure of large health care organizations (typically university-based hospitals, multispecialty clinics, or the military), there are now several independent, for-profit "free agents" providing telemedicine services. Worldcare is the best example of this. This trend started with teleradiology service providers (see Vol. 4, No. 6 for a review of 13 such providers), and has extended to other clinical services.

All rights reserved.
Telemedicine Today Magazine