Photo Credit: Yaroslav Olieinikov

It is recommended that physicians develop mechanisms and alerts that allow them to know if a test previously ordered has not been completed.

Prior to COVID-19, telehealth made up less than 1% of medical visits, according to Maelys J. Amat, MD, MBA. At the beginning of the pandemic, telehealth became pervasive before dropping off to lower levels, she says, with 37% of adults initiating at least one telephone or video visit in 2022.

Little is known, however, about whether the probability of completing recommended tests and specialty referrals is linked with visit modality. “Our research focuses on diagnostic errors as these present a huge safety concern and cost the system billions of dollars,” Dr. Amat says. “One major source of diagnostic errors occurs when a test or referral is ordered but not completed, a concept that we refer to as ‘diagnostic loop closure.’ While investigating this issue, our team became interested in the impact of telehealth on diagnostic errors.”

Dr. Amat and colleagues set out to discover the percentage of tests and referrals that were ordered during a telehealth visit versus when an in-person visit was completed. “Our aim was to study this across a variety of test types to see how test characteristics may impact completion,” Dr. Amat explains. “We reviewed 4,133 test orders and referrals and assessed whether they were completed within an acceptable timeframe.” The study team focused on the prevalence of diagnostic loop closure for colonoscopy referrals (screening and diagnostic), dermatology referrals for suspicious skin lesions, and cardiac stress tests. Their findings were published in JAMA Network Open.

Tests Ordered During Telehealth Less Likely to Be Completed

The study team found that tests and referrals ordered during a telehealth visit were less likely to be completed compared with those ordered during in-person visits or outside of visits. “This was especially pronounced for colonoscopy orders,” Dr. Amat notes. “We also observed that the rate of test completion was suboptimal across all visit types. Finally, we noted that adjusting for several patient and healthcare professional characteristics did not impact the results, suggesting that this effect is inherent to our system and not impacted by individual factors.”

The take-home message, she adds, is that the overall odds of loop closure for all tests ordered during telehealth visits was 0.55 in the adjusted analysis, meaning that there is a 45% decrease in the odds of test completion when a test is ordered during a telehealth visit (Table).

“One way for primary care physicians to incorporate these findings into their practice is to develop mechanisms and alerts that will allow them to know if a test previously ordered has not been completed,” Dr. Amat says. “Ideally, these would be integrated into our medical records to remove the burden from primary care physicians. However, if these systems are not yet in place, we encourage clinicians to develop their own workflows to avoid diagnostic errors. Additionally, clinics should consider implementing a virtual checkout system, similar to what is done during in-person visits, to help patients better understand recommended next steps. Finally, our prior work has highlighted the importance of patient education and comprehension as an important contributor to test and referral completion.”

Dr. Amat and colleagues agree that future research should focus on further understanding why this disparity exists and on interventions to help improve loop closure rates for telehealth visits. “We strongly support the use of telehealth in outpatient medicine and hope that workflows can be strengthened so that patients can receive the same level of care during a telehealth visit as they do during an in-person visit,” Dr. Amat says.