IMPORTANCE: Low-value care is a persistent problem with direct and cascading harms. Telemedicine is now commonly used and may reduce low-value testing by introducing barriers to completing tests at a given visit or expand opportunities for low-value testing by contributing to higher visit volumes. OBJECTIVE: To quantify the association between telemedicine adoption and low-value testing among fee-for-service Medicare beneficiaries. DESIGN, SETTING, AND PARTICIPANTS: In this cohort study using 100% fee-for-service Medicare claims data, US health systems were divided into quartiles based on 2020 telemedicine adoption. Beneficiary-level linear regression in difference-in-differences (DiD) analyses was used to compare beneficiaries who were continuously enrolled from 2019 through 2022 and were attributed before telemedicine adoption (2019) to high telemedicine-adopting (top quartile) vs low telemedicine-adopting (bottom quartile) health systems on low-value test and visit outcomes in 2022 vs 2019. Data were analyzed from October 2023 to December 2024. EXPOSURE: Health system telemedicine adoption. MAIN OUTCOMES AND MEASURES: Receipt of, and spending on, 20 low-value screening, preoperative, chronic condition management, and acute diagnostic tests, as well as total visits (in person and virtual). RESULTS: The sample included 1 382 033 beneficiaries who were attributed to high-telemedicine systems (mean [SD] age, 71.6 [10.5] years
58.8% female) and 999 051 beneficiaries who were attributed to low-telemedicine systems (mean [SD] age, 71.8 [10.0] years
57.0% female). From 2019 to 2022, those in high-telemedicine systems had a small differential rise in visits (DiD visits per beneficiary, 0.12
95% CI, 0.03 to 0.21) and differential decreases in use of 7 of 20 low-value tests: cervical cancer screening (DiD, -0.45 percentage points [pp]
95% CI, -0.72 to -0.17 pp), screening electrocardiograms (DiD, -1.30 pp
95% CI, -1.96 to -0.65 pp), screening metabolic panels (DiD, -1.84 pp
95% CI, -2.87 to -0.80 pp), preoperative complete blood cell counts (DiD, -0.64 pp
95% CI, -1.06 to -0.22 pp), preoperative metabolic panels (DiD, -1.35 pp
-1.91 to -0.80 pp), total or free T3 (triiodothyronine) level testing for hypothyroidism (DiD, -0.90 pp
95% CI, -1.38 to -0.41 pp), and imaging for uncomplicated low back pain (DiD, -1.66 pp
95% CI, -2.35 to -0.98 pp). There were no statistically significant differences in other tests. Those in high-telemedicine systems saw statistically significant differential decreases in spending on visits per beneficiary (-7.87
95% CI, -6.85 to -.88) and on 2 of 20 low-value tests, but no differences in low-value spending overall. CONCLUSIONS AND RELEVANCE: In this cohort study, telemedicine adoption was associated with modestly lower use of 7 of 20 examined low-value tests (most point-of-care) and no changes in use of other low-value tests, despite a small rise in total visits that might offer more testing opportunities. Results suggest possible benefits of telemedicine and mitigate concerns about telemedicine contributing to increased spending.