Korean Medicine Linked to Lower Surgery, Opioid Use in LDH
Key Takeaways
- Among matched patients with newly diagnosed lumbar disc herniation in South Korea, early Korean medicine use was associated with lower subsequent lumbar surgery than conventional medicine alone.
- Early Korean medicine use was also associated with lower composite opioid prescribing during follow-up (adjusted HR 0.891), but the absolute difference was small: 2,957 vs 3,103 events over 4 years, a 0.40 percentage-point difference (NNT 250).
- The association with lower lumbar surgery was stronger at 1 year than at 4 years.
- For non-tramadol opioids alone, results were mixed: the survival curves were not significantly different, and although the fully adjusted Cox model favored Korean medicine, the authors said that estimate should be interpreted cautiously because proportional-hazards assumptions were violated and there were relatively few events.
Using South Korea’s Health Insurance Review and Assessment Service (HIRA) claims data within the National Health Insurance (NHI) system, the Hyun et al nationwide HIRA cohort study of Korean medicine use in lumbar disc herniation identified patients with incident lumbar disc herniation (LDH) by a primary ICD-10 M51.1 diagnosis in 2015. The entry date was the first 2015 diagnosis, and the index date was 60 days later. Korean medicine (KM) exposure required at least 3 outpatient visits to KM institutions during that window and more KM than conventional medicine (CM) visits, whereas CM required at least 3 CM visits and no KM visits. After excluding prior disc disorder diagnoses, red-flag conditions, prior lumbar surgery, and prior opioid exposure before index date for the opioid analysis, 1:1 nearest-neighbor propensity score matching (PSM) with a caliper of 0.1 using sex, age group, insurance type, and Charlson Comorbidity Index (CCI) yielded matched cohorts of 121,720 for surgery and 72,684 for opioid analyses. Healthcare utilization quartiles served as a severity proxy, and Kaplan-Meier plus Cox models tracked outcomes for up to 4 years.
Lumbar surgery showed the clearest association. Because proportional hazards were violated for surgery, the primary surgery estimates were time-stratified: the 1-year hazard ratio was 0.715 (95% CI 0.665-0.769) and the 4-year hazard ratio was 0.801 (95% CI 0.762-0.842). Over 4 years, lumbar surgery occurred in 3,362/60,860 (5.52%) of the KM group versus 3,669/60,860 (6.03%) of the CM group, an absolute difference of 0.51 percentage points (NNT 196). Composite opioid prescribing was also lower with KM, with a 4-year hazard ratio of 0.891 (95% CI 0.844-0.940), but the absolute difference was small: 2,957 versus 3,103 events over 4 years, a 0.40 percentage-point difference (NNT 250); qualifying prescribing meant at least 14 days of tramadol or at least 7 days of a non-tramadol opioid. For non-tramadol opioids alone, results were mixed: the survival curves were not significantly different, and although the fully adjusted Cox model favored KM, the authors said that estimate should be interpreted cautiously because proportional-hazards assumptions were violated and there were relatively few events; a 1-year exposure-window sensitivity analysis was materially unchanged.
As an observational, claims-based comparison, the study cannot establish causation. Claims lacked direct severity measures such as imaging findings, neurologic deficits, and bowel or bladder dysfunction, and healthcare utilization was only an imperfect proxy that may also reflect health-seeking behavior. The dominant-modality design excluded patients receiving substantial combined KM-CM care, and non-reimbursed services, treatment crossover after the 60-day window, and long-term opioid use were not captured. For North American readers, these findings come from Korea’s national insurance system, where KM is integrated and reimbursed, so transferability to U.S. or other North American settings may be limited.
The authors concluded that early KM use after incident LDH was associated with lower hazards of later lumbar surgery and composite opioid prescribing over 4 years in this national cohort. They emphasized that richer clinical data and prospective studies are still needed to test whether those associations persist after fuller adjustment for baseline severity.
Clinician Questions
How was early Korean medicine exposure defined in newly diagnosed lumbar disc herniation?
The KM group included patients with at least 3 outpatient visits to KM institutions within 60 days of the first ICD-10 M51.1 diagnosis and with more KM than CM visits during that window, while the comparison group had at least 3 CM visits and no KM visits. This was a dominant-modality comparison rather than an analysis of individual KM treatments.
What opioid use counted as an outcome in this lumbar disc herniation cohort?
Qualifying opioid prescribing meant at least 14 days of tramadol or at least 7 days of a non-tramadol opioid during follow-up. The authors also evaluated non-tramadol opioids separately.
Which patients do these lumbar disc herniation findings apply to most directly?
These findings apply most directly to patients in South Korea with an incident primary ICD-10 M51.1 lumbar disc herniation diagnosis whose early care during the first 60 days was predominantly KM or exclusively CM. Patients receiving substantial combined KM-CM care were excluded, so the results do not directly describe blended-care patterns.
What limits causal interpretation of lower surgery and opioid use with Korean medicine in lumbar disc herniation?
Causal interpretation is limited because this was a claims-based observational comparison rather than a randomized trial. Direct severity measures such as imaging findings, neurologic deficits, and bowel or bladder dysfunction were unavailable, healthcare utilization was only an imperfect proxy for severity, and health-seeking behavior or treatment crossover could still influence the associations.