Association Between Advanced Paternal Age and Congenital Heart Defects: A Systematic Review and Meta-analysis

A Systematic Review and Meta-analysis

F. Joinau-Zoulovits; N. Bertille; J.F. Cohen; B. Khoshnood


Hum Reprod. 2020;35(9):2113 

In This Article

Materials and Methods

We conducted our systematic review and meta-analysis following guidance from the Centre for Reviews and Dissemination (Tacconelli, 2010). We registered our protocol with PROSPERO (CRD42019135061). The results of the study are reported using Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (Moher et al., 2009).

Search Strategy

We searched MEDLINE (via PubMed) and EMBASE for relevant studies indexed between 1960 and 30 May 2019 (last update), unrestrictive of language or sample size. We used a combination of Medical Subject Headings (MeSH) terms and free text words such as 'paternal age', 'paternal factors', 'father's age', 'parental age', 'heart', 'cardiac', 'cardiovascular', 'abnormalities, congenital', 'birth defects', 'congenital malformations' and 'congenital abnormalities' (Supplementary Material). We also searched the references cited in each article to identify any relevant studies not included in our initial search.

Study Selection

Study selection, quality assessment and data extraction were done by two independent reviewers (F.J.-Z. and N.B.), and disagreements were discussed by the two reviewers and by seeking the opinion of the third reviewer (B.K.) if necessary. The Rayyan software was used for the study selection process (Ouzzani et al., 2016).

Eligibility Criteria and Data Extraction

We included observational studies aiming at assessing the association between paternal age and CHD. The included population could be live births, fetal deaths and terminations of pregnancy for fetal anomaly (TOPFA). To be included, studies had to provide either odds ratios (OR) with their 95% confidence interval (CI) or sufficient information to recalculate crude ORs with 95% CIs per paternal age category. Studies were excluded if they had no comparative group, if they were duplicate publications, reviews or case reports. First, screening was performed based on the titles and abstracts. Then, a second selection was based on full-text articles.

For each included study, we extracted the following information: first author, publication year, country, study design, type of CHD (isolated, chromosomal or syndromic), type of birth (live births, fetal deaths, TOPFA), period of diagnosis of the CHD (at birth, during infancy/childhood), sample size (number of cases and controls), covariates used for multivariable adjustment (if any), risk estimates and 95% confidence intervals.

Most studies that have found an association between advanced paternal age and CHD have shown it at the age of 35 years (Olshan et al., 1994; Materna-Kiryluk et al., 2009). Therefore, we defined advanced paternal age as an age ≥35 years; ORs (or raw data needed to recalculate crude ORs) were extracted accordingly.

Assessment of Methodological Quality

Quality assessment was performed using the Newcastle–Ottawa Assessment Scale (NOS) (Wells et al., 2013). We adapted the tool to our research question (Supplementary Table SI). The NOS assigns a maximum of 12 points to studies of the highest quality according to three risks of bias dimensions: (i) selection of study participants, (ii) comparability of groups and confounding and (iii) outcome or classification bias. Studies were classified as follows: low risk of bias if NOS score ≥9, medium risk of bias if NOS score 5–8 and high risk of bias if NOS score ≤4. To be classified as low risk of bias, studies had to fulfill the following criteria: (i) isolated CHD excluding transient ductus arteriosus, (ii) CHD diagnosed by independent experts, (iii) consecutive or obviously representative series of cases, (iv) community controls, (v) same population between included cases and eligible but not included cases, (vi) adjustment for maternal age, (vii) structured interview for paternal age, (viii) same method of ascertainment of paternal age for cases and controls.

Statistical Analysis

The primary measures of association were crude ORs and 95% CIs for the association between advanced paternal age and risk of CHD. Given the expected heterogeneity in the design and methods of analysis across studies (such as differences in the definition of exposure categories, in study populations (e.g. population-based or not), and in methods for adjustment for potential confounding factors), meta-analysis was performed using DerSimonian and Laird procedures for random-effects models (DerSimonian and Laird, 1986). Forest plots were used to display the results from individual studies, as well as summary estimates obtained through meta-analysis. Heterogeneity was visualized through forest plots and statistically assessed by calculating the I 2 statistic, with corresponding χ 2 tests (Higgins and Thompson, 2002; Higgins et al., 2003). Publication bias was assessed through the visual inspection of funnel plots and by Egger's test (Egger et al., 1997). We calculated e-values to estimate the potential effect of unmeasured confounding in observational studies (Haneuse et al., 2019).

Analyses were stratified based on the type of population included (live births only vs. live births, fetal deaths and TOPFA), type of CHD (isolated CHD only vs. isolated, chromosomal and syndromic), study design (population-based vs clinical-setting) and risk of bias (low risk vs. medium and high risk). We also performed a one-stage random-effects dose–response meta-analysis of the association between paternal age and risk of CHD (Crippa et al., 2019). To summarize the dose–response association, we used centered dose levels to take into account that the references of exposure are different and not zero across studies. To fit the potential non-linear trend, we performed a restricted cubic spline model. We evaluated the nonlinearity by testing that the coefficient of the second spline was different from zero (Orsini et al., 2012). All statistical analyses were performed using Stata V.12.0 (Stata Corp, College Station, TX, USA).