Background: The COVID-19 (COVID) pandemic has been associated with care seeking and delivery system changes. Before COVID the management of low back pain (LBP) was variable and a common source of low-value care. The purpose of this retrospective cohort study was to examine how the management of LBP changed during the COVID pandemic in commercially insured (CI) and Medicare Advantage (MA) cohorts. Methods: A US national sample of LBP episodes with a duration of less than 91 days experienced during 2019-2021 was analyzed. Independent variables included whether an individual had CI or MA coverage, and the timing of LBP onset. Secondary independent variables included individuals home address State. Dependent measures included the percent of individuals initially contacting eighteen types of health care provider (HCP) and receiving twenty-two types of health care services, and total episode cost. Early and late COVID measures were compared with a pre-COVID baseline to examine COVID related change. The impact of the stringency of State level COVID public policy response was evaluated. Results: The study included 222,043 CI and 466,125 MA complete episodes of LBP. During the pre-COVID period the MA cohort was older (MA median 72 vs. CI 45), with higher percent female (61% vs. 52%), and from zip codes with a higher Area Deprivation Index (median 52 vs. CI 38). MA and CI cohort attributes remained nearly identical in the early and late COVID periods. Initial contact with licensed acupuncturists (LAc risk ratio (RR) 0.66) and physical therapists (PT RR 0.82) in the CI cohort, and with PTs (RR 0.78), urgent care (RR 0.86), and emergency medicine (RR 0.87) in the MA cohort experienced the largest decreases during the early COVID period. The largest increase in the CI cohort was to PCPs (RR 1.08), and in the MA cohort to PCPs (RR 1.11) and nurse practitioners (RR 1.09). During the late COVID period the largest decreases in the CI cohort were to neurologists (RR 0.84), PTs (RR 0.86), and physical medicine and rehabilitation physicians (RR PMR 0.87) and in the MA cohort to rheumatologists (RR 0.81), PMR (RR 0.89) and pain management physicians (RR 0.89). The largest increases during the late COVID period in the CI cohort were to radiologists (RR 1.22), hospitals (RR 1.07) and orthopedic surgeons (OS RR 1.04) and in the MA cohort to LAcs (RR 1.32), radiologists (RR 1.11) and hospitals (RR 1.10). Compared to the pre-COVID period during the early COVID period the percent of episodes including most health care services was unchanged or reduced. In the late COVID period in both the CI and MA cohorts the percent of episodes with imaging studies increased, MRI (RR CI 1.15, MA 1.21), CT (RR 1.16, 1.16), and plain film radiology (RR 1.06, 1.06). The stringency of State COVID public policy responses was not associated with significant variability in either the type of HCP initially contacted, or services received for LBP. Conclusions: In both CI and MA LBP cohorts COVID was associated with changes in the types of HCP initially contacted and subsequent services provided. Guideline concordant first-line service use declined during COVID, and the rate of diagnostic imaging was higher in the late COVID period than the pre-COVID period.