Study selection, data collection and data items
All studies (cohort studies and case-control studies) reporting data on the birthweight discordance affected by GHD in twin pregnancies were included. This review included all women in both DC and MC twin pregnancies complicated with gestational hypertension, chronic hypertension, preeclampsia and eclampsia. We excluded pregnancies with intrauterine death of one or both twins.
The outcome was the intertwin birthweight discordance, which was calculated utilizing the following equation: birthweight discordance (%) = (larger twin’s birthweight − smaller twin’s birthweight)/larger twin’s birthweight) × 100. We performed the review according to one of the most common cut‐offs of birthweight discordance (≥15%).4
Furthermore, according to the classification of GHD, we planned to perform sub-group analyses reporting cases with GH, PE, eclampsia and CH separately. According to the criteria outlined by ACOG,32,33 concretely, GH was defined as systolic blood pressure (BP) ≥140 mmHg or ≥ 90 diastolic mmHg after 20 weeks of gestation, PE met the same criteria as GH with either proteinuria or end-organ dysfunction (liver dysfunction, renal dysfunction, central nervous system disturbances, thrombocytopenia, or pulmonary edema). Eclampsia was defined as a new onset of seizures in women with PE, who showed a convulsive episode or other sign of altered consciousness. CH was diagnosed as a pre-existing hypertensive disease or BP ≥140 mmHg systolic or ≥ 90 mmHg diastolic preceding 20 weeks gestation.
In order to ascertain whether the association between GHD and intertwin birthweight discordance differs according to chorionicity, stratification by twin chorionicity (DC and MC) was performed. The assessment of chorionicity was clinically based on the ultrasound evaluation, including the number of gestational sacs, placenta and embryos at early pregnancy, τ-sign or λ-sign, the fetal sex,34 and after delivery it was confirmed by examining the placenta.
Two authors (YW, HYZ) screened all studies independently in view of the inclusion and exclusion standards. Where there were inconsistencies, consensus was reached by discussion with another reviewer (FZ). With full text copies of included studies obtained, the same two reviewers extracted relevant data in regard to study characteristics and pregnancy outcomes independently. In case over one study was published based on the identical cohort with uniform endpoints, the report demonstrating the most comprehensive information was included.
Quality assessment was performed with the Newcastle-Ottawa Scale (NOS) adapted for cohort and case-control studies. Each study is evaluated from three broad aspects: selection of patients, comparability of study groups and ascertainment outcome of interest. The scores of the NOS range from 0 to 9. The included studies were interpreted to be of high risk of bias (for scores ≤ 4), medium risk of bias (for scores 5-7), or low risk of bias (for scores ≥ 7).35 Articles with low risk of bias were included in the analysis.