Wenjun Zhang, Xin Zhang, Lihua Huang, Shengdi Lu, Lei Yu
Under the assumptions of the target trial emulation, manual therapy was associated with greater improvement in disability than dry needling, consistent with the index trial. The unadjusted analysis overstated the association, underlining the value of confounding adjustment. Residual confounding from unmeasured factors cannot be excluded.
OBJECTIVE: To estimate the effect of dry needling versus manual therapy on disability in mechanical neck pain by emulating a target trial using electronic health record data from three tertiary hospitals in China.
DESIGN: This target trial emulation used a new-user, active-comparator design with alignment at time zero. Patients were new users of dry needling plus exercise (n = 1995) or manual therapy plus exercise (n = 3898). The primary outcome was the Neck Disability Index (NDI) at 3 months. Effects were estimated by inverse probability of treatment weighting (intention-to-treat-analogue) and clone-censor-weighting (per-protocol).
RESULTS: Among 5893 patients, dry needling initiators were more severe at baseline (maximum standardised mean difference, 0.72), consistent with confounding by indication. The unadjusted 3-month NDI difference (dry needling minus manual therapy) was 10.1 points (95% CI, 9.4-10.7). After weighting, the difference was 5.7 points (95% CI, 5.0-6.5), exceeding the minimal clinically important difference of 5.5. The per-protocol difference was 5.7 (95% CI, 4.9-6.4).
CONCLUSION: Under the assumptions of the target trial emulation, manual therapy was associated with greater improvement in disability than dry needling, consistent with the index trial. The unadjusted analysis overstated the association, underlining the value of confounding adjustment. Residual confounding from unmeasured factors cannot be excluded.