**3. Data analysis**

In addition to the complete reading of each of the included articles and its quality evaluation, the following procedures were followed:

Mantel-Haenzel test of homogeneity was used to evaluate the null hypothesis that the included studies were homogeneous. Statistical calculations were carried out following the Mantel-Haenzel Q heterogeneity statistical formula. This test was compared with Chi square distribution with n-1 degrees of freedom and a confidence level of 95%, where n was the number of studies included in the meta analysis, which in this case was n=4. If under these circumstances the calculated QMH value was above the tabulated value of X2, the

Effect of 1000 or More ppm Relative

to 440 to 550 ppm Fluoride Toothpaste – A Systematic Review 195

Fig. 1. DMF-S and dmf-s in children under 14 years old: Metaview

Fig. 2. DMF-T and dmf-t in children under 14 years old: Metaview

homogeneity hypothesis was not rejected and no heterogeneity between the included studies was considered.

The estimated effect was measured in terms of average dental caries in each of the studied groups according to the dmf and DMF index. The final analysis evaluated the mean difference between the groups by applying the DerSimonian-Laird method for random effects.

A metaview graphics illustrated the estimated values of the individual studies and the combined estimated effect.

Once the first meta-analytic approach was completed after combining the results from the independent studies, a sensibility analysis was done to establish the solidity of the combined estimated effect after one or more studies with extreme high or low values were withdrawn from the meta analysis.
