Systematic review and meta-analysis studies combine the results of studies performed on a same topic and after the analysis present the final conclusion of the baseline original studies. The importance of these studies is in resolving the need to perform large scale and country wide studies. A systematic review study, collects, reviews, assesses and analyzes the published valid studies regarding a given topic. On the other hand, meta-analysis is a method in which the results of different studies are combined to obtain a more accurate estimation rather than traditional reviews. In other word, the aim of meta-analysis is to estimate the amount of real effect more powerful than an individual study (18).
Search time
In this study, all published studies from 1995 to 2018 regarding dental caries which were performed in all Iran provinces are collected and DMFT status of 12 years old Iranian population are presented through systematic review and meta-analysis. Regarding the cause of selecting 1995 to 2018 to collect articles, it is necessary to say that dental services were integrated to primary health care (PHC) in Iran in 1995 in order to increase access of low socioeconomic groups to the services. This integration led to equipment of dental healthcare centers, allocating dentists and dental care providers to them, performing educational, preventive and therapeutic programs and covering some services by public insurances.
Search strategy
In order to retrieve the most number of articles, a very extensive search strategy was selected. So that only “DMFT “keyword was searched in the internal (Persian) databases and “DMFT” and “IRAN” keywords were searched in the external databases. Though this study assesses DMFT in 12 years old population, but “12 years old” keyword was neglected when searching databases. This is because, however, some studies have not mentioned it in the title or abstract, but have reported some results regarding DMFT in 12 years old.
All epidemiological studies including cross-sectional, case- control, cohort and clinical trial were eligible to be entered to the study. The specific question of current study was: What is DMFT index in 12 years old Iranian population totally and separately for two sexes?
Persian databases of Scientific Information Databases (http://sid.ir/), and Magiran (http://www.magiran.com/) and external databases of Google Scholar (https://scholar.google.nl/), PubMed (http://www.ncbi.nlm.nih.gov/pubmed), Embase (https://www.embase.com), Scopus (www.scopus.com), and Web of Science (http://www.isiknowledge.com) were searched systematically.
Inclusion and exclusion criteria
The inclusion criteria were existence of DMFT word and 12 years old age group in the articles, studies with random sampling, response rate higher than 50% and attrition rate lower than 50 %. The exclusion criteria were caries studies in children covered by supportive organizations, diabetics, mental diseases, exceptional children, thalassemia, factory workers, asthma, kindergarten, blind, military children, hemophilia, kidney disease and other diseases and problems. Also, letter to editor articles, grey literature and studies before 1995 were excluded from the study. The final extracted studies from the databases were entered to the EndNote software and duplicated were deleted.
Assessment quality of articles
The quality of articles was assessed using STROBE checklist. This checklist includes 22 questions which assesses all sections of an article including title, abstract, introduction, methods, results and discussion. This instrument help authors in correct writing observational studies and help editors and reviewers to correct assesses articles to be published in the journals and lastly help readers to critically assesses published articles (19).
As a simple method to score this checklist, we consider 0-2 scores for each one of its questions. The lowest obtainable score for each article to be entered to the data extraction process and meta-analysis was considered 16. The articles in Tables 1 & 2 have obtained this lowest acceptable score.
Predict standard deviation
Articles in which only have stated sample mean not standard deviation (SD), it was estimated using linear regression. So that, at the first, mean and SD in studies which these two measures are specified were entered in 2 separate columns in the SPSS and linear regression test was performed. In the results, if R Square was more than 0.7, so we can use this model to predict SD in the studies which have not calculated it. In this way, by multiplying B amount obtained from linear regression test in the mean of articles which SD is not specified, we can estimate SD amount. In this study, R Square amount was obtained 0.83 which is more than 0.7, so by multiplying B coefficient amount (which is 0.788) in the reported mean DMFT of each study, SD amount was estimated. Also, articles in which have reported DMFT/dmft for boys and girls but have not reported pooled mean and SD, they were estimated using the statistical formula. The studies in which have reported confidence interval, they were converted to SD using the statistical formula.
Continuous vertical line in forest plot indicates “no effect” or neutral state which its amount for articles have reported mean is equal to zero. Interrupted vertical line is corresponding on diamond vertical line which is an indicative of index under study (20).
Heterogeneity and publication bias
Heterogeneity was predicted using I2 index. Because of predicting internal and external heterogeneity between studies, Random Effect Model Used to draw forest plot.
Funnel plot and Egger’s test used to indicate publication bias. Funnel plot indicates the effect measure of each study. Symmetric funnel plot indicates no publication bias. Eggers’ test was used to interpret funnel plot statistically.
If the study had no publication bias, so its Eggers’ test is insignificant and its confidence interval encompass 0 digit. Also, funnel plot turns into an inverted funnel. In forest plot, whatsoever study size become larger, its precision increases and its mean index would place in the higher place of funnel plot. This case is opposite in the smaller studies. Finally, the data were analyzed using Stata 13.1, Stata Corp, College Station, TX, USA.