Editor’s Spotlight/Take 5: Do Orthopaedic Virtual Clinic Visits Demonstrate Cost and Time Efficiencies Compared With In-person Visits?

Telemedicine is one of the new hot topics in healthcare. When I graduated from medical school (in 1991, for those keeping count), five studies addressed telemedicine, according to PubMed. By the time I entered practice in 1998, that number was close to 800. Last year, over 8000 studies were published on the topic.

As a patient, I like it. I had a course of treatment last year and it was entirely online—I never actually met my clinician in person! And many other patients think similarly [1, 2, 5]. Aside from concerns over COVID-19, telemedicine offers convenience, less time spent in transit, and no parking hassles, among other perks. From the clinical side, using telemedicine seems like a winner at first glance: an improved patient experience, better (or at least equivalent) care, and reduced costs. It’s not surprising that hospitals and clinicians are all increasing their use of telemedicine—why not do better for cheaper?

But as always, the details matter—in particular, how one measures costs. In this month’s Editor’s Spotlight, we’re featuring the work of Livingston et al. [4] from the Henry Ford Health System in Detroit, MI, who performed an in-depth evaluation of the costs of patient visits using telemedicine compared to in-person visits. Contrary to my expectations, there was no financial benefit to telemedicine. The importance of this paper’s work lies in its use of time-derived activity-based costing (TDABC), which helps us see how costs truly are generated. Here come those details I mentioned…

Traditional ways to estimate costs include cost-to-charge or basing estimates on resource value unit (RVU) payments. In the cost-to-charge method, hospital costs are allocated on a patient-by-patient basis, and each patient's bill contains charges, which are then converted to costs using an ad hoc ratio of cost-to-charge (RCC). For example, if a patient had USD 300 in pharmacy charges, and the hospital's RCC for pharmacy was 0.5, then cost would be allocated as USD 150. In the RVU method, each item consumed in a department is assigned a value that reflects its relative expense as a function of departmental baseline costs. The RVUs of items used by patients in each department are added together, then multiplied by the department's cost per RVU. In both methods, there’s an astonishing level of guesswork and assumption. Essentially, you figure out what you were paid, you figure out what you spent, and create a ratio.

A more accurate method is activity-based costing (ABC), which attempts to open the black box between resources and products: What do you need to do to get from raw materials to finished product? Once costs of the activities have been identified, the cost of each activity is attributed to each product to the extent that the product uses the activity.

To use a medical example: Cardiac surgery is a high-skill, high-visibility, resource-intense service line that traditionally results in generous hospital reimbursements. Therefore, lots of hospitals pursue cardiac surgery programs. If you simply went with an RVU-based approach, this would be an obvious choice. But with ABC, you’d look at how much each segment of treatment actually costs in terms of personnel and equipment. For example, if you have a cardiac surgery program, you need a cardiac care unit. You need perfusionists. You need all sorts of special gadgets. And you might find that cardiac surgery breaks even, or actually loses money.

But activity-based costing isn’t easy. To build an ABC model for a department that performs one activity, you would ask employees to estimate the percentage of time they spend or expect to spend on that particular activity, find the average, and assign the department’s resource expenses according to that average percentage. As Kaplan and Anderson [3] pointed out almost 20 year ago, “This approach works well in a limited setting, typically a single department, plant, or location. Difficulties arise, however, when you try to roll this approach out on a large scale for use on an ongoing basis. In one large bank’s brokerage operation, the ABC data-gathering process required 70,000 employees at more than 100 facilities to submit monthly reports of their time allocation. The company employed 14 people full-time just to manage the data collection, processing, and reporting” [4]. Clearly, this is not practical.

TDABC, by contrast, requires less data. Two parameters are required: the unit cost of supplying capacity (such as the cost per minute of an employee’s time) and the typical time required to perform a transaction or an activity. Because you’re not asking employees to recall how they spent their last month, it’s more accurate and realistic: Your data are bottom-up, based on actual processes and resources. TDABC allows for a quick assessment of what resources are being used and a logical transition to improve the process. The problem with TDABC is that it requires an in-depth assessment at every level of care. When that’s done, as Livingston and colleagues [4] found, the costs are often quite different from the estimates garnered by a traditional top-down approach.

Now that we understand how to estimate healthcare costs more accurately, let’s take a look at whether telemedicine truly saves money. Join me in the Take 5 interview that follows with Dr. Day, who has thought a lot about these issues and how to apply them in clinical practice.

Charles S. Day MD, MBA

Take 5 Interview with Charles S. Day MD, MBA, senior author of “Do Orthopaedic Virtual Clinic Visits Demonstrate Cost and Time Efficiencies Compared With In-person Visits?”

Paul A. Manner MD: Much of the time spent on a patient visit involves documentation and entering data in the medical record. To what degree might other solutions, such as the use of scribes or voice recognition software (which presumably make documentation more efficient and less time-consuming), make a difference here? Based on your discoveries, what are you doing to try to improve this side of things in your practice?

Charles S. Day MD, MBA: Even without the use of telemedicine, physician groups or practices could theoretically decrease costs by shifting documentation to scribes or voice recognition software. Scribes make about 1/20 the salary of a surgeon, and while voice recognition software may require a hefty installation fee and an annual subscription, this is still much less than what a surgeon makes. In our study [4], documentation for virtual visits was only completed by the attending physician, whereas documentation for in-person visits was primarily completed by physician assistants (PAs) and residents who have salaries that are about 1/5 and 1/9, respectively, of the surgeon. When comparing differences in salaries and the costs of other tools to help offset the time allocated by physicians, finding a more cost-effective alternative for clinic visit documentation would decrease costs for both visit modalities. And as our study shows, greater cost savings would likely be generated from implementing these changes for virtual visits because documentation was completed solely by the physician.

Our ability to unveil this cost-saving workflow improvement would not have been possible using the RCC methodology. Our data showed that there was a negligible difference in “costs” between virtual and in-person for every level of service using RCC, due to the fact that RCC cost is based solely on the charges of that level of service. On the other hand, TDABC demonstrated that the cost of virtual visits is 10% greater than in-person visits. In addition, while RCC is not sensitive enough to measure the cost differential between virtual and in-person visits, it overestimates the actual direct variable labor costs of both visit modalities by an average of 83% higher than TDABC.

Dr. Manner: During the lockdowns for COVID-19, a lot of companies instituted “work from home” with virtual visits on the theory that this made workers more productive. When the lockdowns ended, many of those companies stopped virtual work because they felt it made workers less productive. How do you think productivity changed with the use of virtual visits?

Dr. Day: From our study, we found that the average in-person visit was 17 minutes and the average virtual visit was 13 minutes. I think productivity could have increased for physicians because of the reduction in virtual visit time relative to in-person visits, which would have allowed physicians to squeeze in more patients on a given day without altering a physician's work hours. However, this time efficiency didn’t translate to a cost efficiency because while the overall visit was shorter, the attending spent a longer time in a virtual visit compared to an in-person visit, and because of the added expense of attending time, the decrease in time per visit didn’t translate to decreased costs. Nevertheless, if a visit takes less time virtually, there may be space to include additional visits at the end of the day, which might result in increased productivity.

Dr. Manner: Returning to the theme of “work from home,” telecommuting is nice for college-educated workers, but it may have a negative impact on individuals with lower educational levels or those who work in vocations rather than professions. It is thus likely to have substantially disparate distributional impacts. To what degree did you observe this? That is, were patients with lower educational levels or those who were less computer savvy more adversely affected? How might one create an economic model to assess this?

Dr. Day: We didn't directly examine whether patient education level would be impacted positively or adversely with respect to utilization of virtual visits. However, we did take steps to ensure and equalize our patients’ “computer savviness” so that they felt comfortable engaging with providers on our platform. Before each visit, a staff member called patients to confirm that they could maneuver the virtual environment with confidence. As for an economic model, I might suggest doing away with that extra step involving a staff member and seeing how many patients would choose virtual visits relative to their educational level, and whether educational level affected their comfort level while using the virtual platform during their visits.

Dr. Manner: You looked at three surgeons with the highest volume in their groups. How might your findings apply more broadly to physicians with lower clinical volumes in your practice or in other practices?

Dr. Day: Physicians with lower virtual visit volumes are likely to be less efficient in managing a virtual visit compared to their in-person visits. For those providers who want to host virtual visits in a more‐selective fashion, perhaps they could start with routine postoperative visits and/or test result follow-ups as a way to streamline those experiences and minimize the use of virtual visits for new patients, second opinions, or complicated follow-ups.

Dr. Manner: As a patient, I love virtual visits. I avoid traffic, paying for parking, sitting in waiting rooms, and so forth. But not all patients feel like I do. How might we calculate whether there’s a cost saving for patients?

Dr. Day: In a separate study of ours [6], survey results on patient perceptions of virtual care showed that 76.9% of patients preferred virtual visits. When it comes to calculating costs and savings for patients, you can certainly calculate costs based on the time saved from travel, parking, waiting, and other logistical aspects of care. That being said, for many patients, cost savings isn't a primary reason that drives whether they choose to see a provider virtually or in-person. In another study currently in review, we found that patients feel better about their experience in-person, yet others feel like they can be more honest during a virtual visit.