BackgroundWe aimed to compare patients’ and physicians’ safety reporting using data from large acupuncture trials (44,818 patients) and to determine associations between patient characteristics and reporting of adverse reactions.MethodsSix pragmatic randomized trials with an additional non-randomized study arm that included those patients who refused randomization were evaluated. Patients received acupuncture treatment for osteoarthritis of the hip or knee, chronic neck pain, chronic low back pain, chronic headache, dysmenorrhea, or allergic rhinitis or asthma. Safety outcomes were evaluated by questionnaires from both the physicians and the patients. To determine level of agreement between physicians and patients on the prevalence of adverse reactions, Cohen’s kappa was used. With multilevel models associations between patient characteristics and reporting of adverse reactions were assessed.ResultsPatients reported on average three times more adverse reactions than the study physicians: for bleeding/haematoma, 6.7% of patients (n = 2458) vs. 0.6% of physicians (n = 255) and for pain, 1.7% of patients (n = 636) vs. 0.5% of physicians (n = 207). We found only minor agreements between patients and physicians (maximum Cohen’s kappa: 0.50, 95% confidence interval [0.49;0.51] for depressive mood). Being a female and participation in the randomization were associated with higher odds of reporting an adverse reaction.ConclusionsIn our study, patients’ and physicians’ reports on adverse reactions of acupuncture differed substantially, possibly due to differences in patients’ and physicians’ questionnaires and definitions. For the assessment of safety, we strongly support the inclusion of patients’ and physicians’ reports while ensuring standardization of data collection and definitions.