Comparing Outcomes of Musculoskeletal Radiographs from In-Person and Telemedicine Primary Care Cohorts, April 2019–June 2021

Discussion

The literature has repeatedly called for studies evaluating actual outcomes of televisits.^6⇓–8 Yet in 2022, the authors of a systematic review of orthopedic televisits still concluded that “further studies are needed to determine whether telemedicine consultations from the patient’s own home are truly reliable and achievable.”¹⁸ To our knowledge, this study is the first to explore outcomes of radiographs ordered at televisits.

The results indicate that primary care physicians may overutilize musculoskeletal radiographs at televisits. A possible explanation for these results is that, during the height of the pandemic, patients with minor symptoms avoided in-person visits, whereas patients with more severe conditions sought in-person care. The idea of patient self-selection accords with a survey of primary care patients who delayed care for orthopedic concerns that arose during the pandemic.¹⁹ A study of Medicare patients found that up to 45% of them avoided medical facilities during the pandemic for fear of contracting the virus.²⁰ Delaying care may have inflated radiographic abnormalities for in-person visits that occurred after the height of the pandemic. Ongoing surveillance is needed to see if this trend persists.

Alternatively, physicians may have overutilized radiographs at televisits to compensate for the lack of a tactile physical examination. Unfortunately, the literature on imaging outcomes from virtual visits is sparse. A study looking specifically at patients presenting for low back pain found that less imaging was ordered for patients presenting virtually compared with in-person.²¹ However, national guidelines recommend against routine imaging for low back pain,²² and the study did not explicitly look at imaging results. A study of virtual and in-person encounters for abdominal complaints in the ER found that, compared with the in-person evaluation, clinicians performing the virtual evaluation ordered less imaging.²³ Records from the ER televisits documented greater diagnostic uncertainty, and the authors hypothesized that concerns about cost and imaging utilization may have led the virtual clinicians to order less imaging. Clinical uncertainty may also explain the results of our study. Clinicians may have ordered imaging at televisits to compensate for the lack of a reliable physical examination. A virtual musculoskeletal examination has been shown to provide good accuracy for elements like range of motion, but special tests are more difficult to perform.¹² Multiple studies document the challenges that physicians and patients face when engaging in a virtual musculoskeletal examination, including the potential need for caregiver assistance, appropriateness of the patient’s environment, poor visualization, low inter-rater reliability, and technical issues such as video lag.^{24⇓⇓–27} Clinicians should not underestimate the harms of an incomplete or inaccurate physical examination,²⁸ though a growing familiarity with the virtual musculoskeletal examination¹¹ might impact future ordering patterns for radiographs. Forthcoming studies could evaluate which components of the in-person visit (eg, history and/or physical examination) best correlate with the need for musculoskeletal imaging.

In an age mindful of cost and resource utilization, we should also consider the economic impact of primary care televisits for musculoskeletal conditions. Prior studies exploring health systems and societal costs of orthopedic televisits did not specifically consider the cost of imaging; they either ignored imaging or assumed costs were similar for televisits.^13,29 Our study’s findings challenge this assumption. Telemedicine does not routinely reduce the cost of health care, largely due to the costs associated with technology upkeep and maintenance.³⁰ It has the potential to reduce health care costs if it mitigates expensive procedures and specialist referral,³⁰ but our study indicates that televisits may be less than ideal for musculoskeletal conditions. Unnecessary radiography would increase health care costs (in 2022 CMS reimbursed $34.26 for CPT 73030, multi-view shoulder radiograph)³¹ and further mitigate any savings to the health system gained by the televisit. Furthermore, telemedicine’s societal benefits are largely attributable to the decreased costs of travel and missed work.³² If telemedicine utilization is associated with unnecessary imaging orders, however, the need for travel to an imaging facility and time off of work would moderate the savings to society.

This study found no significant association between social determinants of health or comorbidities and telemedicine utilization, which contrasts with much of the existing literature.^14⇓–16 Local factors and the population studied likely account for this. We did see trends toward significance with age and ethnicity.

Our study has several limitations. As telemedicine becomes a regular medium for ambulatory visits, care must be taken to contextualize these results, which were obtained during the height of the COVID-19 pandemic. A major limitation of this study is that imaging outcomes were not compared with a patient’s documented symptoms in the medical record. Radiographic abnormalities may therefore be incidental findings. However, any error introduced from this would likely be nondifferential, because incidental findings would be present in all groups. Another limitation of the study involves the per-order analysis approach. A per-patient approach could have been used, but it would be more likely to introduce selection bias, because only 1 radiograph per patient could be chosen. Qualitative review of the results found that, of the patients who obtained multiple radiographs of a similar anatomic site, the proportion of concordant abnormal results were similar before and after the pandemic’s onset. Lastly, as a retrospective cohort study, we could not control the size of the cohorts. The number of radiographs from televisits were relatively small compared with those ordered at in-person visits. Nevertheless, retrospective data such as this is valuable, because there is so little evidence of telemedicine’s actual outcomes. In addition, the unpredictability of patients’ concerns makes a prospective primary care cohort difficult to arrange, and randomizing musculoskeletal concerns to telemedicine or in-person visits would be difficult to do in the primary care setting.

In summary, radiographs ordered at in-person primary care visits after the pandemic’s onset had higher odds of being abnormal compared with televisits. These findings indicate that prudence should be applied to ordering musculoskeletal radiographs in telemedicine encounters.