Article type
Year
Abstract
Background: Splint therapy is the most frequently used treatment for temporomandibular disorders (TMDs). Previous reviews have used the standardized mean difference (SMD) to present the summary effect of splint therapy. The SMD is limited by its assumption of similar variability in patient populations across studies, and on knowledge users being comfortable interpreting results presented in SD units.
Objective: To systematically review all RCTs that compared splint therapy to minimal/no treatment in patients with TMDs, and to generate intuitively understandable summary estimates.
Methods: We searched MEDLINE, EMBASE and CENTRAL from inception of each database to August 2011. Eligible studies randomly assigned patients with TMDs to splint therapy or minimal/no treatment. We calculated the SMD to pool data for reporting pain across trials, and converted this measure into a risk difference (RD), an odds ratio (OR), and to natural units on the most familiar instrument, the 100 mm visual analogue scale (VAS). To calculate RD, we used 0.6 for the control group risk of continued pain and performed sensitivity analyses using different control group risks.
Results: Eleven studies proved eligible. Moderate quality evidence suggests that splint therapy reduced pain in the TMJ area: SMD (95% confidence interval [CI]) = c0.93 (−1.33 to −0.53); RD (95% CI) reduction in continued pain= −0.35 (−0.21 to −0.46); OR (95% CI) of reducing pain = 4.64 (8.98, 2.40); and mean difference (95% CI) of reduced pain on the VAS= −11.5 mm (−16.5 to −6.6 mm).
Conclusions: The SMD, OR and RD all suggest a large effect of splint therapy in reducing pain, whereas re-scaling to natural units suggests a modest effect. The explanation for the discrepancy between the estimates of magnitude remains uncertain, and leaves doubt about the appropriate interpretation.
Objective: To systematically review all RCTs that compared splint therapy to minimal/no treatment in patients with TMDs, and to generate intuitively understandable summary estimates.
Methods: We searched MEDLINE, EMBASE and CENTRAL from inception of each database to August 2011. Eligible studies randomly assigned patients with TMDs to splint therapy or minimal/no treatment. We calculated the SMD to pool data for reporting pain across trials, and converted this measure into a risk difference (RD), an odds ratio (OR), and to natural units on the most familiar instrument, the 100 mm visual analogue scale (VAS). To calculate RD, we used 0.6 for the control group risk of continued pain and performed sensitivity analyses using different control group risks.
Results: Eleven studies proved eligible. Moderate quality evidence suggests that splint therapy reduced pain in the TMJ area: SMD (95% confidence interval [CI]) = c0.93 (−1.33 to −0.53); RD (95% CI) reduction in continued pain= −0.35 (−0.21 to −0.46); OR (95% CI) of reducing pain = 4.64 (8.98, 2.40); and mean difference (95% CI) of reduced pain on the VAS= −11.5 mm (−16.5 to −6.6 mm).
Conclusions: The SMD, OR and RD all suggest a large effect of splint therapy in reducing pain, whereas re-scaling to natural units suggests a modest effect. The explanation for the discrepancy between the estimates of magnitude remains uncertain, and leaves doubt about the appropriate interpretation.