Abstract
No meta-analyses or systematic reviews have been conducted to evaluate numerous potential biasing factors contributing to the controversial results on congenitally missing teeth (CMT). We aimed to perform a rather comprehensive meta-analysis and systematic review on this subject. A thorough search was performed during September 2012 until April 2013 to find the available literature regarding CMT prevalence. Besides qualitatively discussing the literature, the meta-sample homogeneity, publication bias, and the effects of sample type, sample size, minimum and maximum ages of included subjects, gender imbalances, and scientific credit of the publishing journals on the reported CMT prevalence were statistically analyzed using Q-test, Egger regression, Spearman coefficient, Kruskal-Wallis, Welch t test (α=0.05), and Mann-Whitney U test (α=0.016, α=0.007). A total of 111 reports were collected. Metadata were heterogeneous (P=0.000). There was not a significant publication bias (Egger Regression P=0.073). Prevalence rates differed in different types of populations (Kruskal-Wallis P=0.001). Studies on orthodontic patients might report slightly (about 1%) higher prevalence (P=0.009, corrected α=0.016). Non-orthodontic dental patients showed a significant 2% decline [P=0.007 (Mann-Whitney U)]. Enrolling more males in researches might significantly reduce the observed prevalence (Spearman ρ=-0.407, P=0.001). Studies with higher minimums of subjects' age showed always slightly less CMT prevalence. This reached about -1.6% around the ages 10 to 13 and was significant for ages 10 to 12 (Welch t test P<0.05). There seems to be no limit over the maximum age (Welch t test P>0.2). Studies' sample sizes were correlated negatively with CMT prevalence (ρ=-0.250, P=0.009). It was not verified whether higher CMT rates have better chances of being published (ρ=0.132, P=0.177). CMT definition should be unified. Samples should be sex-balanced. Enrolling both orthodontic and dental patients in similar proportions might be preferable over sampling from each of those groups. Sampling from children over 12 years seems advantageous. Two or more observers should examine larger samples to reduce the false negative error tied with such samples.