Telemedicine Spending by Medicare: A Snapshot from 2012

Introduction

Medicare first began paying for a limited number of medical services delivered via telemedicine in 1997. Coverage was later expanded in the Medicare, Medicaid, and the State Children's Health Insurance Program (SCHIP, or just CHIP) Benefits Improvement and Protection Act of 2000. Medicare's telemedicine coverage has stayed structurally unchanged since that time, with only the addition of a few new eligible services and a small expansion of eligible sites adding to its scope and overall obligations. ^1,2

In 2000, the U.S. Office of Management and Budget estimated that covering telemedicine services for rural Medicare beneficiaries would cost approximately $150 million over 5 years. ³ The actual expenditures for the first 5 years of Medicare telemedicine coverage totaled $3 million, and although subsequent years showed expenditures growing quickly on a percentage basis, they remained very low relative to overall Medicare expenditures. The most recent year for which a published report on Medicare telemedicine is available is 2009. In that year, Medicare covered some 38,000 clinical encounters, but the available report does not detail expenditures. ¹ Proceedings of the 2012 Institute of Medicine workshop, “The Role of Telehealth in an Evolving Health Care Environment,” report expenditures at “about $6 Million annually” without further elaboration. ³ Very little information on the distribution of that utilization, either geographically or by medical specialty or service, has been published to date.

As interest in all forms of telehealth continues to expand, questions about the cost of services and current distribution are increasingly germane. Medicare provides practically identical reimbursement for telemedicine across all 50 states, but variability in state adoption rates, if it exists, and in the overall national rate of adoption has been largely unstudied. Is telemedicine adoption growing rapidly, as its proponents claim? Are some states adopting telemedicine more rapidly than others, and if so, why? Is rurality a primary driver of adoption, because Medicare reimbursement for telemedicine is limited to rural areas? Are some specialties more inclined than others to use telemedicine to provide services? Answers to these questions may be of interest to many, including policy makers and others outside the community of telemedicine researchers and providers. It seems reasonable, then, to explore what information is available to address these questions from the perspective of the oldest and largest payer in telemedicine.

This article analyzes data available in Medicare public use files that contain a summary of all Medicare outpatient claims for calendar year 2012. Using publicly available sources, the claims are mapped to their states of origin and categorized by billing code and medical specialty to see how the use of telemedicine services under Medicare is distributed both geographically and by specialty and service.

Materials and Methods

The publicly available Medicare Physician/Supplier Procedure Summary (PSPS) data file was used for this analysis. The Centers for Medicare & Medicaid Services (CMS) Web site (www.cms.hhs.gov) describes this data file as follows ⁴ :

This file is a 100% summary of all Part B Carrier and [Durable Medical Equipment Regional Carrier] DMERC Claims processed through the Common Working File and stored in the National Claims History Repository. The file is arrayed by carrier, pricing locality, Healthcare Common Procedure Coding (HCPC), modifier 1, modifier 2, specialty, type of service and place of service. The summarized fields are total submitted services and charges, total allowed services and charges, total denied services and charges, and total payment amounts.

The PSPS data are broken into subfiles by HCPC codes to limit the size of the individual files. We analyzed the files containing specific codes indicating telemedicine services covered by Medicare in all states. We ignored “store-and-forward” telemedicine services, which are reimbursed by Medicare only in Alaska and Hawaii. The list of Medicare telemedicine codes, as printed in the 2012 Medicare Telemedicine Fact Sheet, is shown in Table 1.

Telemedicine services do not have their own unique HCPC billing codes. Rather, services provided via telemedicine are billed using the standard HCPC codes and definitions for the clinical services provided, with the addition of the modifier “GT.” This modifier indicates that the service was provided “Via interactive audio and video telecommunications system.” ⁵ Billed charges for a HCPC code that is on the list of eligible telemedicine codes and include the “GT” modifier indicate telemedicine services. The PSPS files contain two modifier fields. We extracted all claims for telemedicine-eligible codes that included the “GT” modifier in either modifier field. Teleradiology claims, which are fully covered by Medicare and may be quite significant in number, are unfortunately not captured by these data and methods.

Claims in the PSPS files are associated with a carrier, the contracted fiscal intermediary that handles payment for the Medicare claim. Although companies may contract to function as Medicare carriers in multiple states, each carrier code refers exclusively to a carrier's line of business within a particular state. We used the carrier codes to link each claim to the state in which the service was billed. By definition, this corresponds to the state in which the patient was located when the services occurred. ⁶ There is no complementary method for determining the provider's physical location at the time of service using the PSPS files.

These methods allowed us to extract from the PSPS files a comprehensive list of all Medicare claims submitted and either paid or denied for telemedicine services (except for teleradiology services, as noted earlier), categorized by state, provider specialty, and HCPC or Current Procedural Terminology (CPT) code/service type.

Data on Medicare enrollment are available from several organizations, although all are sourced from CMS. We collected enrollment data from the Kaiser Family Foundation Web site ⁷ to make use of the more convenient format available on this site. Medicare sponsors both the traditional Medicare fee-for-service national plan and many Medicare Advantage plans. Only expenditures incurred through reimbursement of services delivered to beneficiaries in the traditional plan are included in the PSPS files. For this reason, the calculation of expenditures per enrollee was made using only beneficiaries enrolled in Medicare's traditional plan.

A variable indicating expenditures per traditional Medicare enrollee was calculated as a way of measuring the relative prevalence of telemedicine use within the Medicare population in a given state. This was done by dividing the amount spent by Medicare on telemedicine services within the state by the total number of enrolled Medicare beneficiaries in the state. Ranks were assigned based on this calculation of relative spending per beneficiary on telemedicine services. States with higher levels of Medicare telemedicine spending per enrollee were ranked higher.

Data analysis was performed using Microsoft^® (Redmond, WA) Access^® and Excel^®.

Results

Summaries of the Medicare telemedicine services and spending for calendar year 2012 are provided in Tables 2 and 3, which summarize the same data, but Table 2 categorizes spending by state, whereas Table 3 categorizes all services by CPT code, ignoring states. Rank variables reflect the relative telemedicine spending per beneficiary in Table 2 and the top 20 billing codes by total payment in Table 3.

Table 4 lists expenditures by provider specialty according to Medicare's provider categories. Because originating site facility fees can be assumed to be billed in conjunction with a corresponding professional fee, claims for the originating site facility fee, HCPC code Q3014, were excluded from Table 4 to give a more accurate indication of the relative utilization among specialties. The top 20 specialties were ranked according to the number of allowed codes rather than total spending, to better reflect relative volume for specialists who bill at different rates.

An examination of Tables 3 and 4 yields several notable findings. First, mental health services and mental health service providers dominate the expenditures. If originating site facility fees are ignored, 69.8% of Medicare telemedicine professional fees are related to mental health. The true proportion may be higher because mental health services can be billed in several ways, some of which are indistinguishable from outpatient medical services.

Second, relative penetration of telemedicine services varies widely among states and is not simply a function of the rurality of the state. The northern Great Plains region contains three of the top five telemedicine-using states (South Dakota, North Dakota, and Minnesota), and these states rank 7th, 10th, and 25th, respectively, in the percentage of the population that lives in a rural area. The other two states in the top five for telemedicine use, Oklahoma and Missouri, rank 16th and 20th, respectively, in rurality. The top four states in terms of rural population percentage (Maine, Vermont, West Virginia, and Mississippi) all ranked less than 20th in terms of Medicare telemedicine penetration. Delaware, Rhode Island, and Hawaii are missing from Table 2 because they did not record any Medicare telemedicine billing in 2012.

The total nationwide cost of including telemedicine in the traditional Medicare benefit package for 2012 was about $0.09 per enrolled beneficiary per year, or about three-quarters of a penny per member per month. As can be seen in Table 2, the statewide costs vary from a hundredth of a penny up to nearly $0.07 per member per month (or $0.82 annually).

Discussion

Medicare has been a key payer for telemedicine services since it began covering telemedicine in the late 1990s. Despite this, the current findings show that telemedicine-related costs remain a relatively miniscule part of overall Medicare expenditures. The distribution of these expenditures across states, services, and specialties suggests that implementation has not been uniform and that factors other than simply rurality or need have driven adoption. It is likely that multiple factors, including policies of other potential payers and business concerns internal to healthcare organizations, also contribute significantly to implementation decisions.

The calculation of expenditures per Medicare enrollee by state gives a measure of the prevalence of telemedicine use across the Medicare population in each state in terms that will be familiar to those who understand risk-based contracts for healthcare. The “per member per month” metric translates the absolute expenditures for telemedicine services into terms that are scaled for the eligible population within a state. It is hoped that these data may inform and enrich discussions of the relative value of access in rural areas and embolden policy makers to expand coverage into other underserved areas as well.

This report must be considered preliminary in some respects because the numbers reported here are almost certainly an underestimate of the true costs of telemedicine to the Medicare program. Because telemedicine services can be paid even if they are not claimed as telemedicine services (simply by omitting the “GT” modifier), it is possible that a significant percentage of actual telemedicine is not being counted in Medicare's claims database. On the other hand, some reports (J. Stensland, personal communication) (the Institute of Medicine workshop ³ ) suggest that a significant percentage of Medicare payments for telemedicine in 2009 did not meet the conditions of payment, leading the recorded payments to overestimate the true costs of telemedicine. Only rigorous review could determine the true error rate and real costs. Until then, assuming error rates that are not significantly different than other types of services within Medicare is not unwarranted.

Finally, these data make clear that telemedicine has been only minimally explored as a viable option for healthcare service delivery to date, at least within the Medicare population. Despite its promise for expanding access in highly effective and efficient ways and the existence of a large and growing research literature to support this claim, telemedicine has been used in what could be justifiably called only a “pilot” basis for the past 20 years. The challenge for the next 20 years may be to give it a real test.