The ACA 6 Years In: Spreading Physicians Thin

Since the passing of the ACA, the United States healthcare system has seen a net influx of approximately 20 million newly insured individuals under the often-criticized law. Although opinions on the ACA are many and often politically motivated, the fact remains that the system is now burdened with individuals seeking care that was previously unavailable to them. Working under the assumption that the upcoming election will not radically alter coverage availability for the newly insured, one study by the Association of American Medical Colleges estimates the demand for physicians to increase by two percent due to ACA insurees alone (17 percent for the system overall) by 2025. The same study estimates the shortfall of PCPs to range from 12,500 to 31,100 by the same year -- which considering physician training and education can take a decade, this is today’s problem.

Addressing this issue is no small task and there is no single answer. A variety of potential solutions exist ranging from increased primary care capabilities of nurse practitioners, extending office hours for providers (although this adds additional encumbrance to them), forgiving medical school debt, and improving government funding for additional training and residency programs post-medical school. Telemedicine, or (in simplistic terms) the practice of exchanging medical information electronically between two sites, is yet another option growing in popularity for managing routine and sometimes urgent care situations. A selection of disparate market forecasts (methodological differences aside) estimate a global market worth anywhere from $34B to $35B to $36.3B -- all within less than a decade. Needless to say, there is indeed authentic interest in developing telemedicine technology from public, private, and patient levels.

Measuring Potential: The Telemedicine Needs Assessment

Telemedicine is not a new care delivery model and has been around since the early 1960’s space program for managing astronaut health. However, with the proliferation of telecommunications technology globally it has become a viable delivery option in an overtaxed healthcare system. Although government funding is on the table -- $16M recently awarded by the Health Resources and Services Administration (HRSA) for a variety of programs -- a substantial number of legal barriers exist.

Without getting into serious detail about all the restrictions and legal wins/losses (for those interested, check out examples of the recent legal situations in Texas and Arkansas), suffice it to say that implementing a telemedicine program isn’t as easy as simply snapping one’s fingers and making it happen. Misalignment at the state and federal level make this process arduous in many cases. Furthermore, not every potential area that could benefit from a telemedicine program has adequate systems in place for one.

This is where the needs assessment process comes in. A telemedicine needs assessment serves several purposes for those interested in exploring the viability of implementation as well as obtaining funding from a variety of public and private entities. Identifying service and accessibility gaps that could readily be addressed through telemedicine systems (especially if some semblance of those systems are already in place) becomes an appealing argument in a competitive funding pool. Conversely, the needs assessment may determine that telemedicine implementation isn’t viable for reasons including costliness, lack of significant need, lack of patient/provider acceptance, and so on. It is therefore arguable that the needs assessment is paramount to the process of telemedicine implementation where hundreds of thousands (if not millions) of dollars are at stake.

While there is no definitive framework for the telemedicine needs assessment, this is probably ideal. As with any business problem, the variables that come into play to address the viability of telemedicine implementation vary by the unique mixture of internal, organizational constraints and external market forces. In other words, it’s inherently difficult to structure a needs assessment framework that is useful across the board.

The Telemedicine Impact Score (TIS)

Interest in telemedicine is growing rapidly and the supporting research is struggling to keep up the same pace. Expected X launched the Telemedicine Impact Score (TIS) study to address this gap. The goal of the TIS study is not to be the be-all-end-all solution to developing a solid telemedicine needs assessment. Instead, we set out to create a method for assessing telemedicine implementation on a broad scale to tackle this “inherently difficult” task. We believe that the TIS study should be thought of as an input to the needs assessment -- not the needs assessment itself.

Continuing in that vein, our philosophy is not to offer cookie-cutter solutions. We treat the TIS study as just one way of using public data to answer a business problem -- in this case telemedicine implementation. Different organizations will have different considerations and/or opinions on the inputs to our scoring algorithm. We welcome those conversations not only to improve our clients’ positions but our own understanding of the telemedicine landscape.

Data and Methodology

The healthcare industry is flush with publically available data collected and compiled by government, private, and academic agencies alike. This means that a good portion of the data management process has already been completed by experts in the field. The difficulty is in determining what measures are suitable for scoring individual geographies by telemedicine impact.

We looked at the problem from two angles:

  • What is the level of opportunity for telemedicine implementation within a given geography?
  • What is the level of need for telemedicine implementation within a given geography?

When thinking about the problem this way you not only consider how telemedicine could address a particular population in need (say, for instance, 65+ Type-2 Diabetes patients), but what geographies are well suited to help the most patients with that condition via a telemedicine solution -- and with the fewest restrictions. Based on our evaluation, the answer may surprise you.

County Health Rankings and Roadmaps (CHR&R): The CHR&R study compiles data from a variety of sources to create a county-level, in-state ranking for the quality of healthcare compared to other counties. Rather than use the study’s aggregate score, we used two variables:

  • Percent “Poor or Fair Health”: Self-reported health status from the CDC’s Behavioral Risk Factor Surveillance System (BRFSS) -- a randomly-selected, annual survey of adults 18+.
    • Reason for inclusion: Self-reported health status has been shown to correlate well with mortality.
    • Limitations: Data is self-reported and scaled on a Likert “excellent” to “poor” scale that cannot be medically validated.
  • Diabetes prevalence: Estimates from the CDC's Behavioral Risk Factor Surveillance System (BRFSS) for the percentage of adults aged 20 and above with diagnosed diabetes.
    • Reason for inclusion: Diabetes is not only one of the costliest conditions in the country (estimated at costing ~$250B annually) but also one where telemedicine treatment can play a significant role in reducing cost/burden on the healthcare system.
    • Limitations: Data is self-reported “yes/no” response to question "Has a doctor ever told you that you have diabetes?" Cannot be medically validated.

Community Health Status Indicators (CHSI): The CHSI study collects key indicators for health outcomes and a selection of factors determined to impact population health for all 3,144 counties in the United States. The CHSI is intended to identify gaps and disparities between counties to promote population health improvement.

  • Deaths per 100,000 (Heart Disease): Proportion of coronary heart disease-related deaths (identified by ICD-10 codes) by average population within county for all available years’ data.
    • Reason for inclusion: Heart disease is the leading cause of death in the United States but remote management (especially following a hospitalization event) is being studied as a means to prevent readmission.
    • Limitations: Counts were aggregated across age group and race thereby eliminating the ability to stratify by these demographics.
  • Deaths per 100,000 (Cancer): Proportion of cancer deaths (identified by ICD-10 codes) by average population within county for all available years’ data.
    • Reason for inclusion: Cancer continues to be a leading cause of death and medical expenditure in the United States but several oncology-related treatments are available via telemedicine systems which can help reduce cost and patient burden.
    • Limitations: Counts were aggregated across age group and race thereby eliminating the ability to stratify by these demographics.

Health Professional Shortage Area (HPSA): Generally speaking, HPSAs are geographies, subpopulations, or medical facilities with a substandard ratio of providers to patients (typically less than 1:3,500, respectively). While HPSA designations exist for physicians, dentists, and mental health professionals, we chose to work specifically with physician HPSAs (pediatrics, ob/gyn, general internal medicine, and family practice physicians) that are geographically defined. Other designation types can be added in future iterations of the TIS study.

  • HPSA Score: A variety of measures are considered during the HPSA designation process. The HPSA Score metric is an aggregate measure that we felt sufficiently covered important aspects. First, four individual metrics are scored and assigned point values: patient/provider ratio, percent population below poverty level, infant health index (combines low birth weight and mortality), and distance to nearest medical facility. These are then summed to create the HPSA Score from 0 to 26 with higher values indicating a greater need.
    • Reason for inclusion: HPSA designation is key to Medicare reimbursement for most telemedicine services and provides a simple measure of need within a geography.
    • Limitations: Both Medicaid and private insurers use different standards for evaluating need/access that may not incorporate HPSA.

Hospital Compare: The Hospital Compare is a component of CMS’s Hospital Quality Initiative and rates Medicare-certified hospitals and systems on several quality measures. Data is either collected directly by CMS or via claims data for the two measures used in our scoring algorithm.

  • Rate of readmission after discharge from hospital (30-day): The rate of readmission hospital-wide following a discharge.
    • Reason for inclusion: The assumption is that a high rate of readmission signals an unmet need in recovery care -- a gap telemedicine could fill for certain conditions.
    • Limitations: The measure does not specify the reason/condition for readmission so not every readmission may be telemedicine-ready (however, disaggregated measures do exist and may be incorporated in future TIS studies).
  • Able to receive lab results electronically and Able to track patients’ lab results, tests, and referrals electronically between visits: Both measures are used to serve as a proxy for a hospital’s ability to effectively implement a telemedicine solution.
    • Reason for inclusion: A hospital that is already effectively managing an EMR system is assumed to have technical expertise available to extend that into the telemedicine space. Additionally, many telemedicine solutions require EMRs as a prerequisite -- especially for store-and-forward applications.
    • Limitations: The measure does not indicate if a hospital has already implemented a telemedicine program. We assume EMR implementation and telemedicine implementation may share some infrastructure components.

National Broadband Map: Database designed to help measure, promote, and develop broadband access across the country. Originally built by the National Telecommunications and Information Administration (NTIA) and Federal Communications Commission (FCC).

  • Reason for inclusion: Simply put, neither a real-time video conference nor a remote monitoring telemedicine solution can be initiated without adequate access to broadband Internet speeds. Store-and-forward is not necessarily impacted by slower speeds.
  • Limitations: Database is no longer maintained and data is from 2014. However, broadband availability is nearly ubiquitous in most US counties as of 2016.

Our next challenge after selecting the data for our scoring algorithm was to actually construct the algorithm itself. This was no trivial task and could easily be composed differently from the method we implemented:

  • Each measure is a county-level measure. We examined each in respect to a state-level and national-level aggregation but we recommend state-level since most telemedicine restrictions involve in-state legislature (i.e. multi-state physician licensing, reimbursement parity, store-and-forward laws, etc.).

  • Measures were normalized by dividing each by the maximum value of that variable (with the maximum determined by the level of analysis -- maximum within state counties or maximum within all counties nationwide). For example, a normalized score of “1” at the state level means that a county has the highest score in the state. Any county scoring less than the top county is scored as a proportion of the maximum value, or: 

countymaxequation.gif
  • Each “raw score” was multiplied by 10.
  • Each of these was then summated to create a “raw TIS” which was multiplied by a factor of 100 to create the final TIS on a scale from 0 - 100.

However, implementation of a telemedicine solution needs to take practical matters into account as well. For instance, if a county’s TIS is extremely high compared to its neighboring counties, it usually wouldn’t make financial sense to implement telemedicine in a single county. Therefore, we added a “smoothing function” to our scoring algorithm so that adjacent counties essentially “share” some of their scores with each other. We won’t get into the technical details here, but suffice it to say the process uses the exponential decay function below:

Results and Findings

The interactive data visualization below can be used to inspect each county's TIS using the state or national-level calculation (due to heavy traffic, you may need to refresh the page if "Not Found" displays below):

Traditionally, telemedicine solutions have been marketed as a means to address the needs of populations living in rural areas where access to a physician is prohibitive. Unfortunately, due to often-conflicting state and federal regulations many telemedicine services were limited to populations in these areas when it comes to reimbursement. In urban settings that do have telemedicine offerings there were often only a limited number of available services that don’t match their rural counterparts.

Yet these laws are starting to change and telemedicine is reaching a greater population with reimbursement parity becoming more common than not and rural vs. urban being removed from the equation. According to our study, this is coming at a great time.

Out of the 3,144 counties in the United States, approximately ⅓ exist as part of a MSAs (Metropolitan Statistical Areas). Among that subset of counties we found that 45 states in the country had their highest scoring county in an MSA*. So while the prevailing notion of the recent past may have been that telemedicine was a “rural-only” solution, our study suggests the contrary.

Thinking about this deeper from the perspective of the approach we took to create the TIS, it makes sense. While rural communities may indeed have a greater need for telemedicine, the opportunity may not be there. Yes, access to care may be an issue (particularly those counties in HPSAs), but if the population is in good health on average then a costly implementation may only impact a handful of individuals -- and even some of them may not be willing to accept telemedicine as an alternative to face-to-face physician care!** The opportunity within a given rural geography may also be decreased due to factors like the lack of a solid broadband infrastructure (again, mostly for real-time video conferencing). One would need to address that before even beginning to consider implementation.

Continuing on, for the sake of brevity we chose to examine the two states, Illinois and California, a little further.

Illinois

When considering national-level TISs, Illinois is home to the county with the highest score: Cook County. Cook also scores highest at the state-level and its five contiguous counties (Lake, McHenry, Kane, DuPage, Will) score among the highest in the state. Thinking about this from our study’s perspective of weighing need and opportunity we see that Cook County not only has a high prevalence of chronic conditions and hospital readmissions (the need) but also a substantial number of health facilities we deemed “telemedicine-ready” (the opportunity).

Now, to be fair, being home to one of the largest metropolitan areas in the country, Chicago, Cook County already has many telemedicine implementations in use. The story is the same in other high-scoring counties within MSAs. Although county population wasn’t explicitly used in calculating the TIS, it naturally has an impact from both the need and opportunity perspective.

If we look further at the distribution of TISs within Illinois we see a disparity between Cook County and the rest of the state. The histogram shown uses the raw TISs adjusted by the smoothing function described earlier. We chose these because the final, normalized score creates an arbitrary uniform distribution that can act more like a within-state ranking.

While Cook County occupies a place at the top of the state in terms of TIS, the remaining counties are distributed far lower than Cook County. The state’s median TIS at 52.7 is less than half of Cook County’s 111 adjusted raw score. We suspect this is because Illinois’s population density is heavily distributed to the Chicagoland area. The TIS tends to score urban areas higher simply because the greater number of residents usually indicates both a greater need and opportunity.

California

California is presented in contrast to Illinois. The two states differ in a few ways with respect to their embrace of telemedicine. For example, California’s telemedicine legal restrictions are fewer in comparison to Illinois. California has required private payer reimbursement for telemedicine services since 2011. Illinois has several bills pending legislation, but still lags behind many states in terms of private payer reimbursement.

From another standpoint, California’s geography differs substantially from Illinois. Not only is California’s geographic area nearly three times that of Illinois, but its population distribution is not concentrated heavily in a single region of the state. From the perspective of our TIS calculation, this should create a much flatter distribution of scores across the state and a greater number of counties that score higher when “boosted” by our smoothing function. This appears to be the case.

Conclusion and Considerations

As we stated earlier, the TIS is just one way to measure the impact of telemedicine implementation. We invite our clients and prospective clients to weigh-in with their opinions on how to strengthen our calculation. A few adjustments for future iterations may include:

  • Apply an additional weight to each input variable to vary their impact in the final scoring. We suspect this will be most useful when performed at the state level.
  • Greater incorporation of legal measures to determine opportunity. Ultimately, legal restrictions (or lack thereof) will help drive telemedicine adoption. Again, performed at the state level.
  • A potentially larger issue in telemedicine adoption is the patient’s willingness to accept and utilize it. A multi-million-dollar telemedicine system with only a handful of users will fail to achieve sufficient ROI. In a future iteration of the TIS study, we envision incorporating a patient-level component to gauge telemedicine awareness, consideration, and acceptance within areas with high TISs. This can act as an additional weighting factor or input variable itself.

Questions/Comments: john@expectedx.com


*MSA’s defined by Bureau of Economic Analysis (BEA). Includes District of Columbia which is unique.

**Although recent research seems to indicate that patients are becoming more accepting of telemedicine, we did not incorporate this element  in our TIS. An additional primary research element may be incorporated in a future iteration.