Musculoskeletal Telemedicine Trends Preceding the COVID-19 Pandemic and Potential Implications of Rapid Telemedicine Expansion

Introduction Telemedicine was rapidly deployed at the onset of the COVID-19 pandemic. Little has been published on telemedicine in musculoskeletal care prior to the COVID-19 pandemic. This study is aimed at characterizing trends in telemedicine for musculoskeletal care preceding the COVID-19 pandemic. Methods This retrospective study used insurance claims from the Truven MarketScan database. Musculoskeletal-specific outpatient visits from 2014 to 2018 were identified using the musculoskeletal major diagnostic category ICD-10 codes. Telemedicine visits were categorized using CPT codes and Healthcare Common Procedure Coding Systems. We described annual trends in telemedicine in the overall dataset and by diagnosis grouping. Multivariable logistic regression modeling estimated the association between patient-specific and telemedicine visit variables and telemedicine utilization. Results There were 36,672 musculoskeletal-specific telemedicine visits identified (0.020% of all musculoskeletal visits). Overall, telemedicine utilization increased over the study period (0% in 2014 to 0.05% in 2018). Orthopedic surgeons had fewer telemedicine visits than primary care providers (OR 0.57, 95% CI 0.55-0.59). The proportion of unique patients utilizing telemedicine in 2018 was higher in the south (OR 2.28, 95% CI 2.19-2.38) and west (OR 5.58, 95% CI 5.36-5.81) compared to the northeast. Those with increased comorbidities and lower incomes and living in rural areas had lower rates of telemedicine utilization. Conclusions From 2014 to 2018, there was an increase in telemedicine utilization for musculoskeletal visits, in part due to insurance reimbursement and telemedicine regulation. Despite this increase, the rates of telemedicine utilization are still lowest in some of the groups that could derive the most benefit from these services. Establishing this baseline is important for assessing how the roll-out of telemedicine during the pandemic impacted how/which patients and providers are utilizing telemedicine today.


Introduction
Telemedicine in musculoskeletal care offers great promise. Individual hospitals have published data showing increased telemedicine utilization since the start of the pandemic [1,2], but this has not yet been shown on a national scale. Unfortunately, there has been limited research in the United States documenting prepandemic baseline telemedicine rates or telemedicine trends in musculoskeletal care.
Widespread implementation of telemedicine in musculoskeletal care may lead to an improvement in patient care and health outcomes. Prior analyses of telemedicine in musculoskeletal settings have shown that patient safety is not compromised with its use [3][4][5], and patients are comparably satisfied with the remote interactions [6][7][8]. In addition, these visits are cost-effective for both patients and health systems [8][9][10]. However, despite these advantages, patients that have lower median household income and/or live in rural areas-those who have the most to theoretically gain from telemedicine-often have been the least able to access it [11].
Restrictive reimbursement rates have limited the expansion of telemedicine services in the past [12]. Medicare, in particular, has consistently been one of the most restrictive in telemedicine coverage, and many commercial payers have also implemented similar restrictions [13]. This may in part be due to a desire to minimize costs, as individuals have higher healthcare utilization when telemedicine is available [14]. With increasing public and private reimbursement since the onset of the COVID-19 pandemic, telemedicine has expanded across all specialties, including musculoskeletal care. Many musculoskeletal and orthopedic clinics began to offer telemedicine for the first time following the onset of the pandemic [2,15], with 83% of academic orthopedic surgery clinics offering telemedicine services as a direct result of the COVID-19 pandemic [15]. The institutions most likely to offer telemedicine were located in the northeast and south regions of the United States, regions that were "hot spots" of COVID-19 at the time of the article's publication [15]. Without baseline data on differences in telemedicine utilization, it cannot be concluded if this correlation is truly a result of infection rates, as these regional differences could have existed prepandemic.
This study is aimed at (1) documenting the prevalence of musculoskeletal telemedicine utilization within the United States, (2) analyzing trends in telemedicine visits and patients over the 4-year study period, and (3) tracking these trends by visit subtype. These prepandemic rates of telemedicine utilization in musculoskeletal care establish a baseline necessary for contextualizing telemedicine trends following the onset of the COVID-19 pandemic. This contribution allows for better analysis of the deployment of telemedicine in musculoskeletal care.

Materials and Methods
Approval of this study was obtained from the Institutional Review Board.
2.1. Study Design, Database, and Sample. This retrospective study used patient-level private insurance claims from >100 payers from the Truven MarketScan database to identify telemedicine visits that occurred between 2014 and 2018 (n = 846,461,609 visits; copyright © 2017 Truven Health Analytics Inc.; dataset access was limited to Hospital for Special Surgery employees). To extract musculoskeletal-specific outpatient visits (n = 190,299,246 visits), the cohort was filtered using ICD-10 codes that are associated with the musculoskeletal major diagnostic category (MDC). Telemedicine visits were defined using current procedural terminology codes 99441-99444; Healthcare Common Procedure Coding System codes G0406-G0408, G0459, G0508-G0509, G0425-G0427, Q3014, and T1014; or any code with either a procedure modifier of GT, GQ, or 95 or a location of service listed as "telehealth" [16].

Study
Variables. The primary outcome of this analysis was the utilization of telemedicine as the modality of a patient visit. The study variables considered included both patientspecific variables and telemedicine visit characteristics.
Telemedicine visit characteristics included the following: diagnosis, encounter type, provider type, copayment, and year of visit (2014-2018). Using ICD-10 codes, the diagnoses were classified into 5 major groups (hip/knee pain, low back or neck pain, musculoskeletal aftercare, and other). The encounter type was determined using the MarketScan variable "SVCSCAT," which identifies the detailed service type (

Statistical Analysis.
In this paper, we analyzed annual trends in telemedicine for adults > 18 who had a visit classified as "MDC code 8 (musculoskeletal)." This analysis was performed both for the overall dataset and by diagnosis group. Cochran-Armitage trend tests assessed linear trends.
The multivariable logistic regression model estimated the association between both patient-specific and telemedicine visit study variables and telemedicine utilization. Odds ratios (OR) and 95% confidence intervals (CI) were reported; analyses were performed using SAS statistical software v9.4.

Results
From 2014 to 2018, a total of 190,299,246 adult outpatient visits classified with MDC code 8 (musculoskeletal) were observed in our study. Of these, 36,672 were telemedicine visits (0.020%). The number of unique telemedicine patients 2 International Journal of Telemedicine and Applications ranged from 1837 (0.018%) to 15,338 (0.270%) over the years of the study. The majority of people who used telemedicine had just 1 visit (n = 28,831). Of the remaining patients, 2885 had 2 telemedicine visits, and 1231 had over 2 visits. Telemedicine utilization increased over the study period from nearly 0% to 0.05% of visits and 0.02% to 0.27% of patients, with the proportion of telemedicine patients increasing more quickly than telemedicine visits ( Figure 1(a)). An inflection point is observed in 2016, with a significant increase in the number of patients utilizing telemedicine in the subsequent years. Similar patterns were seen for low back or neck pain (0% to 0.06% of visits; 0.02% to 0.27% of patients) (Figure 1(b)) and hip/knee pain (0% of visits to 0.05% of visits; 0.01% to 0.17% of patients) (Figure 1(c)). Each year, there was an increase in telemedicine visits and patients for these groups. Musculoskeletal aftercare was unique because while there was an overall increase in visits and patients (0% to 0.02% of visits; 0.01% to 0.1% of patients), there was a decrease from 2017 to 2018 (Figure 1(d)).
In the 2014-2018 period of study, the proportion of unique telemedicine visits and patients increased across types of encounters ( Figure 2). The proportions were the highest and rose the steadiest for nonspecialized physicians (0.02% of visits to 0.47% of visits; 0.03% to 0.64% of patients) (Figure 2(a)). Telemedicine utilization rates were lower and more variable in their growth among specialized physician (0% of visits to 0.02% of visits; 0.01% to 0.04% of patients) (Figure 2(b)) and nonphysician "professional" (0% of visits to 0.01% of visits; 0% to 0.07% of patients) (Figure 2(c)) visits/patients. Upon stratifying telemedicine utilization by the type of provider ( Figure 3), growth in the telemedicine proportions was observed from 2014 to 2018. Of the types of providers assessed, telemedicine made up the greatest proportion of primary care visits/patients (0.02% to 0.37% of visits; 0.03% to 0.57% of patients) (Figure 3(a)) and the least proportion of orthopedic surgeon visits/patients (0% to 0.01% of visits; 0% to 0.02% of patients) (Figure 3(c)). Telemedicine utilization rates steadily increased as a proportion of total visits/patients for primary care providers (Figure 3(a)), urgent care providers (0.01% to 0.12% of visits; 0.01% to 0.14% of patients) (Figure 3(b)), and "other" providers (0% to 0.02% of visits; 0.01% to 0.10% of patients) (Figure 3 3 International Journal of Telemedicine and Applications followed by the south region (0% to 0.05% of visits; 0.02% to 0.22% of patients) (Figure 4(c)), with the northeast (0% to 0.02% of visits; 0.01% to 0.09% of patients) (Figure 4(a)) and north central (0% to 0.02% of visits; 0.02% to 0.10% of patients) (Figure 4(b)) regions demonstrating comparably low utilization.
Telemedicine utilization varied significantly across a range of telemedicine visit characteristics (Table 1). It was more likely to be used for musculoskeletal aftercare (OR 1.47, CI 1.34-1.61) than for low back/neck pain and less likely to be used for hip/knee pain (OR 0.75, CI 0.72-0.77) or other visits (OR 0.85, CI 0.83-0.87). Visits were less likely to be office visits with a specialized physician (OR 0.15, CI 0.14-0.15), visits with nonphysician "professional" (OR 0.03, CI 0.03-0.03), or other (OR 0.01, CI 0.01-0.01) than office visits with a nonspecialized physician. Similarly, visits were less likely to be with urgent care (OR 0.57, CI 0.55-0.59), with an orthopedic surgeon (OR 0.36, CI 0.34-0.39), or other (OR 0.51, CI 0.49-0.53) than with primary care. Telemedicine visits were significantly less likely to have a copayment (OR 0.27, CI 0.26-0.28). Finally, telemedicine visit data showed that there was an increasing likelihood of having a telemedicine visit from 2014 through 2018 (OR 18.3, CI 17.5-19.2).
Significant differences in utilization were also seen for several patient characteristics (Table 1). Females were more likely than males to have had a telemedicine visit (OR 0.86, CI 0.85-0.88). There was a decreasing likelihood of having a telemedicine visit for individuals < 55 years of age (OR 0.69, CI 0.67-0.71), increased Charlson comorbidity burden (OR 0.86, CI 0.84-0.88) to >2 (OR 0.68, CI 0.66-0.70), and an active opioid prescription (OR 0.90, CI 0.88-0.93). Patients in urban areas were more likely to have had a telemedicine visit than those in rural areas (OR 2.92, CI 2.79-3.05). In terms of location of residence, those that lived in the west region were the most likely to have had a telemedicine visit (OR 5.58, CI 5.36-5.81), followed by those from the south region (OR 2.28, CI 2.19-2.38), the north central region (OR 1.35, CI 1.28-1.42), and the northeast region (reference). Finally, patients with a median household income of <$45,000 were less likely (OR 0.74, CI 0.70-0.78) than those with a median household income of $45,000-$60,000 to have a telemedicine visit, and those with a median household income >$60,000 were more likely (OR 2.57, CI 2.49-2.65)

Discussion
Generally, there were very low rates of telemedicine utilization for musculoskeletal visits in the years preceding the COVID-19 pandemic. From 2014 to 2018, 0.020% of visits classified as musculoskeletal were conducted via telemedicine with 0.098% of patients having at least one visit via telemedicine. A stepwise increase was observed year-over-year,   International Journal of Telemedicine and Applications indicating an upward trend in the usage of telemedicine. The upward trends were similar across musculoskeletal visit subtype, with the exception of musculoskeletal aftercare which saw a slight decrease in unique telemedicine visits and patients in 2018. There was a rapid increase in the proportion of unique telemedicine patients, but the number of telemedicine visits per patient overall did not vary dramatically from year to year and remained around 1.13 (range 1.11-1.16). This finding indicates that telemedicine was becoming widely used and that increased overall telemedicine visits was not due to the same telemedicine patients having more frequent visits. This is in contrast to the literature as providers prefer using telemedicine for follow-up visits [18], a class of visit that typically outnumbers those such as consults or preoperative planning. As evidenced by the low prepandemic rates of telemedicine, musculoskeletal care was virtually naïve to telemedicine. This is an important finding as it means the COVID-19 pandemic required the musculoskeletal field to drastically rollout and adapt to telemedicine. Immature telemedicine programs were likely deployed, with equity a minimal focus in the era of crisis. With this low rate of prepandemic utilization established, characterizing the baseline of what types of visits were telemedicine and which patients participated in them is important for tracking how these trends evolved across the pandemic to the present day. These analyses are crucial for refining telemedicine programs and ensuring equity.
There were notable differences observed in telemedicine utilization across visit characteristics. Musculoskeletal aftercare yielded a higher proportion of telemedicine visits than low back/neck pain, a trend that will likely be found to be reversed during the COVID-19 pandemic due to a pause on elective musculoskeletal surgeries and a predictive increase in low back/neck pain as a result of more sedentary activity throughout the pandemic. Additionally, orthopedic surgeons previously utilized telemedicine at lower rates than primary care providers. Figures 2 and 3 together illustrate that the upward trends in telemedicine utilization were not as consistent for musculoskeletal visits as they were in primary care settings.
Interestingly, the regions with the highest proportion of unique patients utilizing telemedicine were the south (0.22% in 2018) and the west (0.85% in 2018), while the proportion in the northeast was only 0.09% in 2018. This data provides interesting prepandemic context, as institutions in the "hot spot" northeast and south regions of the United States were more likely to offer telemedicine at the onset of the pandemic [15]. While comparable claims data from 2019 to 2020 would provide additional insight into how the rate of telemedicine utilization has changed, the increase in telemedicine in these regions may have been the result of COVID-19 rather than simply a reflection of previous trends.
Unfortunately, at this time, telemedicine has not been adopted equally across the entire patient population. Here, female patients and older patients were more likely to have had a telemedicine visit. Conversely, patients that have more comorbidities, have lower incomes, and live in rural areas are less likely to have had a telemedicine visit. These groups already suffer from health disparities-disparities that translate into prepandemic telemedicine utilization, as well.
The observed disparities could be due to a variety of factors. For instance, patients with lower incomes may have more limited access to necessary telemedicine technology and may therefore be less likely to have had a telemedicine visit. This disparity is especially damaging as individuals with lower income could theoretically benefit the most from having flexibility in the location in which they attend their appointments. Previous studies have also shown that individuals with lower socioeconomic status are more likely to have a failed telemedicine video visit [19,20]. Flexibility in the modality of telemedicine visit may help address this disparity, as telephone visits have been shown to be comparably successful across income groups [19]. Additionally, those living in urban areas have increased access to musculoskeletal care nearby; therefore, one would expect rural areas to have higher rates of telemedicine utilization. However, this is not the case in our data, as telemedicine utilization rates are higher for those from urban areas. Urban clinics may have more resources to develop and implement telemedicine programs, resulting in a roll-out 7 International Journal of Telemedicine and Applications effect that favors their earlier adoption of telemedicine practices. It is clear that significant disparities existed in telemedicine preceding COVID-19, many of which may have been exacerbated by the pandemic. Prepandemic telemedicine usage must be known to accurately assess and address the impact of COVID-19 on these disparities.
With regard to assessment of the trends of annual telemedicine utilization, a sharp uptick is observed in 2016. Several policy factors, germane to the period leading into 2016, might be implicated in the observed uptick. First, health plans began to expand the coverage of telemedicine services. The Center for Medicare & Medicaid Services (CMS) reported that, from 2014 to 2016, there was a 48.3% increase in the number of plans that covered telemedicine services [21]. Additionally, states began to increasingly regulate private insurers to cover telemedicine services [22]. This expansion of the coverage of telemedicine services by health plans likely resulted in the subsequent increase in telemedicine visits. Another potential policy driver of telemedicine involves the advent of Accountable Care Organizations (ACOs). Many ACOs sought to save costs through bundled payment reimbursement schemes, charging a single price up front for all associated costs of a procedure. In an effort to improve cost-savings, it is possible that ACOs turned to telemedicine as a cheaper modality of patient care in musculoskeletal settings.

Limitations.
This study is limited in its use of commercial claims data as it omits many older and lower-income patients that are enrolled in Medicare and Medicaid, respectively. Additionally, the lack of information on telemedicine modality may limit this research, as the trends in telephone visits may differ from those of video visits; further stratification of telemedicine modality may provide additional insight into disparities in telemedicine.

Conclusions
This study demonstrates that there were overall low rates of telemedicine, particularly in musculoskeletal care, in the prepandemic period, though utilization trended upward from 2014 to 2018. Those with increased comorbidities, lower incomes, and living in rural areas had lower rates of telemedicine utilization in the prepandemic period. This information provides much needed information on previous trends in telemedicine, and this baseline can be used for comparison as research is conducted on telemedicine in the era of COVID-19.

Data Availability
This study utilized the Truven MarketScan database, a commercial dataset produced by Truven Health Analytics (copyright © 2017 Truven Health Analytics Inc.).

Conflicts of Interest
Dr. Memtsoudis is the owner of SGM Consulting, LLC and co-owner of Centauros Healthcare Analytics and Consulting, LLC. He is a partner in Parvizi Surgical Innovations, LLC and investor in HATH. He has a US patent application for a Multicatheter Infusion System (US-2017-0361063). He is a one-time consultant for Teikoku Pharma Inc. All other authors have no potential conflicts of interest to disclose.