Expectancy effects in an open-label randomized trial of manual therapy for the care of cervicogenic headache

Introduction: Expected treatment outcomes and the influence of the provider-patient interaction on expectancy are important in randomized trials, particularly in the absence of complete blinding. In a secondary analysis, we identify the relationship of participant outcomes with measures of patient-expected treatment success and variables that could influence participant expectation.

Methods: 80 participants with chronic headache were randomized to 4 groups: 2 levels of treatment dose and 2 levels of therapy. Patients received 8 treatments (1/wk) or 16 treatments (2/wk) from a chiropractor. The study intervention was spinal manipulation and the control treatment was light massage. Regression analysis was used to determine the partial correlation between outcome and an expectancy variable after controlling for intervention, dose, and baseline severity. Dependent variables were pain, number of headaches, and appointments attended. Independent expectancy variables were optimal dosage and confidence in manipulation, massage, participant's treatment, and other treatment; and perceived DC enthusiasm and comfort treating the participant.

Result Highlights: Expectancy had little influence on appointment attendance. A greater product of baseline pain and expected optimal number of treatments (interaction effect) was associated with 12-week pain (Rinteraction = 0.47, P < .001). A greater product of baseline total headaches and perceived DC enthusiasm was associated with 24-week total headaches (Rinteraction = 0.42, P = .001).

Conclusions: Expectancy at baseline had a modest correlation with outcomes, whereas the doctor-patient interaction was poorly associated with outcomes except for headaches at 24 weeks. Expectancy and provider-participant interaction must be addressed to avoid bias in unblinded trials.

This research was supported by a grant from NCCAM/NIH (R21 AT002324).