Factors Driving Rapid Adoption of Telemedicine in an Academic Orthopedic Surgery Department

Abstract

Introduction:

With the COVID-19 epidemic ever-expanding, nonemergent access to health care resources has been reduced to decrease the exposure for patients and health care providers. Alternatives to in-office outpatient medical evaluations are necessary. We aimed to analyze how quickly orthopedic surgery providers at a large academic institution adopted telemedicine, and identify any factors that were associated with earlier or “faster” telemedicine adoption.

Methods:

We analyzed the telemedicine activity of 39 providers within the Department of Orthopedic Surgery between March 16, 2020, and May 30, 2020, and constructed logistic regression models to identify characteristics with significant association to earlier or faster telemedicine adoption.

Results:

No significant predictors of percentage of visits conducted via telemedicine were found. However, increased experience and practice at multiple locations was associated with slower telemedicine adoption time, while Professor level academic rank was associated with a faster time to achieving 10% of pre-COVID visit volumes via telemedicine. Higher pre-COVID visit volumes were also significantly associated with faster telemedicine adoption. Demographic factors, including, age, gender, practice locations, academic degrees, pediatric specialty, and use of physician assistants/nurse practitioners, were not found to have significant associations with telemedicine use.

Conclusions:

These results indicate that telemedicine has an important role to play within academic orthopedic surgery practices, with a wide and diverse range of orthopedic surgery providers choosing to utilize it during the COVID-19 pandemic. Given the rapid expansion and urgency driving the adoption of telemedicine, these results illustrate the importance of considering provider-side characteristics in ensuring that providers are well equipped to utilize telemedicine.

Introduction

With the COVID-19 epidemic ever-expanding, nonemergent access to health care resources has been reduced to decrease exposure for patients and their health care providers.¹ Traditionally, outpatient visits were performed in a face-to-face environment at a physician's office. However, as social distancing, lockdown, and shelter-in-place become common terms, alternatives to in-office outpatient medical evaluations, such as telemedicine, are necessary and need to be evaluated.

Telemedicine in orthopedics, or “tele-orthopedics,” was initially used sparingly by providers for brief follow-up visits or other appointments with limited physical examination necessary. More recently, it has evolved to include postoperative follow-up, physical examinations, and even teleconsultations directly to patients, mainly as a result of the COVID-19 pandemic.² Despite the rapid growth and emergent needs for teleorthopedic care and technologies, many orthopedic providers still show resistance.³ Previous studies have cited provider characteristics and subspecialties, such as sex, years in practice, academic rank, multiple degrees, and specialization, may influence likelihood of individual providers' telemedicine adoption.^3,4

Although the COVID-19 pandemic has vastly expanded telemedicine usage overall, no studies have yet examined the rate at which orthopedic surgery providers have accepted telemedicine during this time, and what, if anything, contributed to their adoption.¹ We aimed to analyze how quickly orthopedic surgery providers at a large academic institution adopted telemedicine amidst the COVID-19 pandemic, and identify any factors that were associated with earlier or “faster” telemedicine adoption.

Materials and Methods

This study was approved by the Johns Hopkins Institutional Review Board, study number IRB00250908.

Data Collection

Data on the telemedicine usage of 39 providers within the Department of Orthopedic Surgery was obtained from Tableau dashboards created by the Office of Johns Hopkins Physicians Decision Support Analytics. These data included the monthly average number of in-person visits per provider before March 16, 2020 (recorded as the start of the COVID-19 pandemic), the total number of telemedicine and in-person visits per provider between March 16, 2020, and May 30, 2020 (May 30 was recorded as the end of the most severe government restrictions on in-person visits), the date of each providers' first telemedicine visits, and each providers' academic ranks, specialties, degrees, genders, years in practice, ages, locations practiced at, primary practice locations, and their individual use of physician assistants (PAs) and nurse practitioners (NPs). The date on which 10% of each provider's total visits were telemedicine, defined as a percentage of the average pre-COVID monthly visit volume. Average pre-COVID monthly visit volume was based on data from January 1, 2019, to December 31, 2019. Of note, all providers had equal access to resources from the Office of Telemedicine, including training and IT support. These resources were rolled out from a system level and provided to all providers to facilitate telemedicine use. In this study, telemedicine included both audio and video visits and was used for all types of visits, including new patient visits, follow-ups, and postoperative visits. Visits were conducted via the Epic Electronic Medical Record, Zoom, or Doximity.

Data Processing and Exploration

Providers with no telemedicine usage were excluded. For the purpose of this study, those individual providers with 1 or 0 telemedicine visits in the period between March 16, 2020, and May 30, 2020, were considered to be nonusers. The raw data were used to calculate several derived variables: (1) the percentage of visits conducted by telemedicine between March 16, 2020, and May 30, 2020, (2) the time to adoption of telemedicine in days, defined as the time between March 16, 2020, and date of first telemedicine visit, and (3) the time in days from March 16, 2020 until 10% of a provider's average monthly visit volume before March 16, 2020 was achieved via telemedicine visits.

The raw data and derived variables were then converted to binary or categorical types. Binary variables were created for (1) practicing at multiple locations, (2) using a PA or NP, (3) holding multiple degrees, (4) having a primarily academic-based versus community-based practice, and (5) practicing pediatric orthopedics. Categorical variables were created by dividing the following variables into quartiles 1, 2, 3, and 4: (1) age, (2) years in practice, (3) average monthly visit volume before March 16, 2020, (4) percent telemedicine visits between March 16, 2020 and May 30, 2020, (5) adoption time for telemedicine after March 16, 2020, and (6) time till 10% of a provider's average monthly visit volume before March 16, 2020, was achieved via telemedicine visits. Finally, these variables were used as independent and dependent variables in the logistic regression models. Table 1 summarizes the variables and their types.

Table 1.

Summary of Variables Used in Logistic Regression Analysis

NAME	DESCRIPTION	TYPE	USAGE
Academic rank	Assistant, Associate, Professor	Categorical	Independent
Gender	Male, female	Binary	Independent
Mult Locs	Practicing at multiple locations	Binary	Independent
PA	Use of PA or NP	Binary	Independent
Mult degrees	Holding multiple degrees	Binary	Independent
Academic based	Primary practice at JHOC or Bayview	Binary	Independent
Peds	Pediatric specialist or not	Binary	Independent
Age quartile	Provider age, quartile (Q1 is youngest providers)	Categorical	Independent
Experience quartile	Provider experience quartile (Q1 is providers in practice for the least time)	Categorical	Independent
PreCovidVolume	Provider pre-COVID volume quartile (Q1 is providers whose pre-COVID visit volumes were the lowest)	Categorical
10p Quartile	Quartile for time till 10% of visits were telemedicine (Q1 is providers who reached this threshold most quickly)	Categorical	Dependent
Adoption quartile	Quartile for time till telemedicine was first used (Q1 is providers who reached this threshold most quickly)	Categorical	Dependent
Percent telemed quartile	Quartile for percentage of telemedicine (Q1 is providers with the lowest percentages of practice being telemedicine)	Categorical	Dependent

PA, physician assistant; NP, nurse practitioner.

Testing for Predictor's of Telemedicine Utilization

The statsmodel Python module⁵ was used to conduct logistic regression for six dependent variables: bottom three quartiles versus top quartile and bottom two quartiles versus top two quartiles for each of the following dependent variables: quartile for time till 10% of pre-COVID monthly volume was achieved in telemedicine, quartile for adoption time, and quartile for percent of telemedicine visits. For the time to 10% of pre-COVID volume and adoption time quartiles, quartile 1 is the top quartile with the fastest adoption of telemedicine. For percent telemedicine quartile, quartile 1 refers to the top quartile with the highest percentage of telemedicine use. The academic rank and gender variables were one-hot encoded as there is no numerical relationship between the categories (female and assistant professor were the reference categories). In contrast, the quartile variables were maintained in a categorical format since the quartiles have an ordered numerical relationship. A constant intercept was added before model fitting. The basin hopping optimization method was utilized. Each model was fit 25 times, and the coefficient set with the highest log-likelihood was selected for analysis. The code and data needed to run this code are available in Supplementary Files S1 and S2.

Results

Data Processing and Exploration

Of 39 individuals for whom data were available, 6 were excluded as nonusers of telemedicine during the period between March 16, 2020, and May 30, 2020. The remaining 33 individuals conducted between 20% and 100% of their visits via telemedicine during this period (Fig. 1). Table 2 offers summary statistics for some variables of interest for these providers.

Fig. 1.

Distribution of percentage of telemedicine visits for 33 providers who used telemedicine between March 16, 2020, and May 30, 2020.

Table 2.

Summary Statistics for the 33 Providers Who Used Telemedicine and Were Studied in Logistic Regression Analysis

VARIABLE	SUMMARY
Academic rank	Assistant Professor: n = 15
	Associate Professor: n = 10
	Professor: n = 8
Gender	Male: n = 28
	Female: n = 5
No. of locations practiced at	Mean: 2.12 locations
	SD: 0.93 locations
Work with PAs and/or NPs	Yes: n = 15
	No: n = 18
No. of degrees	Mean: 1.18 degrees
	SD: 0.39 degrees
Academic based	Yes: n = 18
	No: n = 15
Pediatric	Yes: n = 4
	No: n = 29
Age	Mean: 47.70 years
	SD = 9.96 years
Years in practice	Mean: 21.82 years
	SD: 10.35 years
Monthly visit average (before March 16, 2020)	Mean: 156.21 visits
	SD: 60.39 visits
Time till 10% of visits were telemedicine	Mean: 36.58 days
	SD: 17.66 days
Time till telemedicine was first used	Mean: 12.58 days
	SD: 9.14 days
Percentage of telemedicine visits	Mean: 59.42%
	SD: 24.65%

SD, standard deviation.

Interestingly, the dependent variables of adoption time and time to achieving 10% of pre-COVID monthly volume in telemedicine visits were positively correlated (Pearson correlation coefficient = 0.757, p = 3.408e-07). However, adoption time had no correlation with percentage of telemedicine visits during the study period (Pearsons correlation coefficient = −0.0012, p = 0.995). Moreover, average monthly pre-COVID visit volumes had no correlation with the percentage of telemedicine visits during the study period (Pearson correlation coefficient = −0.0167, p = 0.9265). This indicates that the relationship between adoption time, pre-COVID visit volumes, and percentage of telemedicine use is likely complex and influenced by other provider-side variables.

A preliminary qualitative exploration of the variables showed that providers of all ages, experience levels, and pediatric versus adult practice were distributed across the four quartiles of percentage telemedicine visits (Fig. 2). Notably, in the same time-period before the COVID-19 pandemic (March 16, 209 to May 30, 2019), these 39 providers saw 38,181 patients with all being in-person visits and no telemedicine use.

Fig. 2.

Relationship between percentage of telemedicine visits and pre-COVID volume quartile (a), age quartile (b), experience quartile (c), and pediatric versus adult provider (d). In (d), 1 corresponds to pediatric providers and 0 to adult providers. In (a–c), 1–4 correspond to the quartile.

Testing for Predictors of Telemedicine Utilization

No significant predictor variables were found for being in the bottom three quartiles versus top quartile and bottom two quartiles versus top two quartiles for percent telemedicine. For adoption time bottom three quartiles versus top quartile, experience quartile was a significant predictor of being in the top quartile, with a coefficient of −5.625 (p = 0.044, 95% confidence interval [CI] −11.093 to −0.157) and practice at multiple locations was a significant predictor of being in the top quartile with a coefficient of −3.7673 (p = 0.026, 95% CI −7.088 to −0.447). The negative coefficients indicate that the odds of being in the top quartile decreased with experience or practice at multiple locations. The top quartile consists of the 25% of providers who adopted telemedicine most quickly. For adoption time bottom two quartiles versus top two quartiles, pre-COVID volume quartile was a significant predictor of being in the top two quartiles, with a coefficient of 2.6628 (p = 0.022, 95% CI 0.385 to 4.941). The top two quartiles consist of the 50% of providers with lowest adoption times (“faster” adopters). For time to 10% telemedicine visits bottom three quartiles versus top quartile, no significant predictor variables were found. For time to 10% telemedicine visits bottom two quartiles versus top two quartiles, an academic rank of Professor was a significant predictor of being in the top two quartiles, with a coefficient of 4.4597 (p = 0.041, 95% CI 0.185 to 8.735). The top two quartiles consist of the 50% of providers who took the least time to reach 10% of their pre-COVID volume in telemedicine (“faster” adopters). The statsmodel output is available in Supplementary Files S3–S8.

Discussion

While the significant predictors reported here are minimal, there are some interesting patterns. It appears that higher pre-COVID volumes were associated with earlier or “faster” telemedicine adoption times, while practice at multiple locations was associated with slower telemedicine adoption times. In addition, while higher experience levels were associated with slower adoption times, Professor-level academic rank was associated with a faster time to reaching 10% of pre-COVID volume in telemedicine. Interestingly, those providers with Professor level academic rank are also more experienced (exclusively in experience quartiles 3 and 4) suggesting that while more experienced providers were slower in adopting telemedicine, they appeared to reach 10% of their pre-COVID volume more rapidly. Moreover, 50% of those in the top two experience quartiles were not Professors, indicating that the impact of experience level may be mediated by academic rank.

Importantly, gender, age, academic degrees, academic-based practice, specialty and use of PA/NPs were not significant predictors of telemedicine adoption or utilization. This suggests that a diverse range of orthopedic surgery providers are willing and able to use telemedicine effectively and that differences in adoption and utilization are more likely driven by extrinsic factors such as number of patients—those with higher pre-COVID volumes perhaps had more incentive to utilize telemedicine quickly to maintain their clinical practice.

These results fit within an emerging body of literature suggesting that telehealth has an important role to play in patient care, especially during times of public health crisis. Recent literature has suggested that the COVID-19 pandemic will result in a large backlog of elective surgical cases in orthopedic surgery.⁷ Adoption of telemedicine ensures continuity of care and may aid in prioritizing patients for surgery whose care is deferred. Moreover, a recent study of telemedicine in orthopedic surgery practice found high satisfaction with telemedicine appointments from both the patient and provider sides, with 93% of patients indicating that they would participate in telemedicine again and surgeons reporting that 78.4% of the time telemedicine successfully replaced an in-person visit.⁶ Most studies to date utilizing orthopedic practices have focused on patient-side characteristics that predict telemedicine utilization. Another recent study of fracture clinic patients found that older patients were more likely to choose telehealth appointments,⁸ while a study of patients in an orthopedic trauma clinic found that there was no difference in the percentage of no-show appointments between patients using telemedicine versus traditional appointments.¹

In the department studied here, between March 16, 2019, and May 30, 2019, (the same time-frame as studied here, but before the COVID-19 pandemic), the patients seen had an average age of 49 years; 55.28% were female and 44.71% were male; 63.46% were white, 24.44% were black, 5.35% were Asian, and 7.91% had their listed race as “Other.” Of them, 61.82% had the MyChart tool activated, 26.88% had activation pending, and 10.27% had inactivated or declined the tool (this is how telemedicine appointments were accessed from the patient side). Based on the high access to the MyChart tool, and the high overall adoption of telemedicine during the 2020 study period, it is likely that these demographics are representative of the patients who used telemedicine during the COVID-19 pandemic. However, further study is needed to ascertain this, and demographic disparities in patient-side telemedicine adoption will be an important factor to understand and address to ensure equitable access.

Despite the benefits of telemedicine, prior studies suggest that before the COVID-19 pandemic, rates of telemedicine utilization were low and that most clinicians had little knowledge of telemedicine³; this is in contrast to our findings here that 33 of 39 providers within the department used telemedicine to a great extent. Much of this adoption may have been driven by necessity due to restrictions placed on in-person appointments during the pandemic, but this adoption still illustrates that telemedicine is a viable option for delivering orthopedic patient care.

Given the rapid expansion of telemedicine and the tremendous urgency driving its adoption, it is also important to consider the provider-side characteristics that predict telemedicine utilization. A recent study of ophthalmology providers found that increased experience (practicing >36 years) was significantly associated with telemedicine adoption, as was clinical assistant, clinical associate, instructor or associate professor academic rank, and female gender.⁴ These results are markedly different from the results of this study, illustrating that there are likely significant differences between medical specialties, and that further investigation in this area is needed to ensure that all providers are equipped and supported in the effort to utilize telemedicine. Another factor to consider is different providers' desires to establish telemedicine practices—different providers may have different interest in this modality depending on how the pandemic has impacted their practice volumes and the types of patients they see. To this end, future efforts should attempt to integrate a broader range of providers across multiple departments, and to simultaneously analyze both patient-side and provider-side characteristics.

Authors' Contributions

All authors contributed to the conception of the work or the acquisition and interpretation of the data, drafting the piece or critically revising it, and the final version of the article.

Disclosure Statement

Author D.S. is a consultant for Alcon. Author U.S. reports personal fees and other from Tigon Medical, personal fees from Conventus, personal fees from Fx Shoulder USA, personal fees from Orthofix, grants from Depuy/Synthes, grants from Arthrex, grants from Wright, grants from Smith & Nephew, personal fees from Heron, personal fees from Pacira, grants from ASES, grants from OMEGA, outside the submitted work; In addition, Dr. Srikumaran has a patent Conventus pending, a patent Fx Shoulder USA pending, and a patent Tigon Medical issued. All other authors declare that (s)he has no relevant or material financial interests that relate to the research described in this article.

Funding Information

We acknowledge support for the statistical analysis from the National Center for Research Resources and the National Center for Advancing Translational Sciences (NCATS) of the National Institutes of Health through Grant Number 1UL1TR001079. Author A.V.A. acknowledges support from the NIH Medical Scientist Training Program 1T32GM136577 and the Helen and Joseph Pardoll Scholarship for MSTP Students.

Supplementary Material

Supplementary File S1

Supplementary File S2

Supplementary File S3

Supplementary File S4

Supplementary File S5

Supplementary File S6

Supplementary File S7

Supplementary File S8