The Effect of Upper Cervical Mobilization/Manipulation on Temporomandibular Joint Pain, Maximal Mouth Opening, and Pressure Pain Thresholds: A Systematic Review and Meta-Analysis.

Only 8 trials, most with small samples. Every outcome was downgraded for imprecision because the pooled samples were under 400, and the confidence intervals were wide enough that the true effect could sit above or below the authors' threshold for clinical importance. None of the trials blinded participants or the people delivering treatment, so expectation could have inflated the patient-reported results. Five of the 8 had problems with how participants were allocated, and only 1 of the 8 was at low risk of bias overall. The pooled pain result against sham came from just 2 trials and was heterogeneous, so that summary estimate is unstable. Publication bias could not be assessed reliably: several comparisons had too few trials for the funnel plot test, and the plots were asymmetric for the pressure pain outcomes. In some trials the cervical technique was given together with exercise, education or craniocervical flexor training, so for the comparisons against no treatment the effect of the cervical treatment alone cannot be separated from the rest. The trials varied in how long people had had jaw pain, from over a month to several years, and 3 trials included women only, so the findings may not fit every patient. The authors could not rule out a placebo response at the longest follow-up, and adverse events may have been missed.

The supplied text does not include a funding or conflict-of-interest statement, so it is not possible to say who paid for this work or what the authors declared.

Do not read the significant pain and mouth-opening results as proof that upper cervical manual therapy works. Both pooled estimates fell below the thresholds the authors set for a clinically important change (2.0 points for pain, 2.5 mm for mouth opening), the pain result rests on two heterogeneous trials, and at each trial's earliest follow-up point the pain difference was no longer significant (P=.26).