28 Chapter 2 complication. It contains the estimated overall pooled proportion of a complication and the corresponding 95%-CI. If the total case rate of the specific complication was above 50, the pooled proportion and CI per birth defect was showed IF; (1) the complication was reported in at least three studies for the specific birth defect OR (2) if the total number of patients all studies on an birth defect combined was ≥ 100 patients; AND (3) there was at least one event of the complication present within all studies on a specific birth defect. Anomalies that did not met these criteria were reported under the forest plot as residual. These residuals were still included in the calculation of overall pooled proportions. For all studies with multiple arms, data of both trial-arms were combined or, if only one arm matched the inclusion criteria, the appropriate arm was used. Further extracted parameters were: author, country of conduct, year of publication, journal, study design, duration of follow-up, duration of study, number of participants, type of birth defect Data extraction Rayyan was used by two independent authors (LES, DN) for title and abstract screening. Rayyan is an online based software facilitating blind screening for reviewers. Disagreements were resolved by discussion. If consensus was not reached a specialist (JD, GM) was consulted. Then, full text of the remaining articles was assessed for inclusion eligibility (LES, DN). Authors were contacted if full texts were not available online. Of the included articles the reference list was cross checked to find any additional articles. Validity and Eligibility assessment All included articles were assessed for the methodological quality and risk of bias using the Newcastle Ottawa quality assessment scale. [32] For randomized controlled trials this was done using the Jadad scoring system.[33] The assessment was done by LES and DN separately. Data synthesis For each study, a weighted average of the logit proportions was determined by the use of the generic inverse variance method. The logit proportions were back transformed to the summary estimate and 95% CIs were obtained in a summary proportion representing the pooled proportion of the IH. Heterogeneity was assessed using the I2 and χ2 statistics. Analysis was performed using R-studio version 3.6.1 (package “meta” (Schwarzer, 2007) and “metaprop” (Viachtbauer, 2010)). The random-effects model was used for interpretation. Heterogeneity was deemed significant if the pooled data’s p value was <0.05 or χ2 statistics were ≥75. Heterogeneity was interpreted as small (I2≤0.25), medium (I2=0.25 – 0.50) or strong (I2≥0.50), according to Higgins.[34]
RkJQdWJsaXNoZXIy MTk4NDMw