Twin pregnancy has increased risks of preterm labor, spontaneous preterm birth, premature rupture of membranes, neonatal and perinatal morbidity and mortality [5, 10, 21, 22]. The occurrence of any potentially life-threatening conditions, maternal near miss or maternal death was twice as high or more, in twin pregnancies; they had complications in 15.3% while singleton pregnancies had only in 6.8%. Results were reasonably consistent across geographical regions. These outcomes were the object of study in at least another two articles with data from World Health Organization detailing the relationship between twin pregnancy and severe maternal morbidity [6, 15]. No explanations were found to variable rates of adverse maternal outcomes in twin pregnancies in different countries with similar income, however it may relate to differences in the quality of available care and local complication patterns [6, 15].
The reported preterm birth rates among twins are very similar to that found in other studies, ranging from 31% [6, 22] to 44% , but some reporting up to 63% . Early preterm births are less frequent than late (34–36 weeks), as Vogel et al. reported in the WHO Global Survey, with 11.9% of preterm birth below 34 completed weeks . Higher early preterm rates are important, as they are associated with higher neonatal morbidity and perinatal death rates, mainly due to respiratory complications [6, 23, 25, 26].
Low birth weight is also more frequent among twin pregnancies. A previous study found that this risk was 8.3 times higher than in singletons, with a mean birth weight of 2300 g , higher than that observed in our study (5 times higher). This risk is associated with the increase in Apgar score at 5th minute < 7 and death during the first year of life [22,23,24, 27]. Adequacy of weight for gestational age better assesses the size of the fetus for a given gestational age (compared to birth weight alone). This is particularly useful in populations where preterm birth rates are high. A fetus that is small for gestational age is more likely to experience perinatal morbidity and mortality and adverse effects in adult life . Few studies have evaluated this outcome among twin deliveries, but associations between twin pregnancies and higher rates of small-for-gestational-age have been reported [28, 29]. For these estimations, we used the curves of Fenton et al.  because we believed that it was more appropriate to be used when the prevalence of preterm birth is very high, as is the case among twin pregnancies in this population. However, due to the number of cases to have such estimates, it was not feasible to have such assessment performed using different nomograms for comparison.
The risk for low 5th minute Apgar score was three times higher for twin pregnancy (either for the first or second twin) than for singletons. Additionally, it was 1.3 times higher for the second when both twins were compared. This significantly lower Apgar score for the second twin is always taken into consideration in discussions about the best mode of delivery for twin pregnancies and the time interval between first and second twin, although not justifying an indication for a systematic Cesarean section for twin pregnancies [6, 30,31,32]. The higher rates of admission to a neonatal intensive care unit we found have also been reported by previous studies on the topic [6, 31].
Prevalence of fetal death of one of the twins varies from 0,5-6,8% with the worst result for monochorionic pregnancy presenting a high prevalence for this condition (50–70%) and risk for the surviving fetus including the fetal death of this co-twin, neurological morbidity and iatrogenic preterm delivery [33, 34]. In the current study, we have not data on chorionicity, however fetal death (death after 28 weeks) occurred over 1.5 times (3.6%) for the first twin and almost 3 times (5.7%) for the second twin when compared to singletons (2.0%).
Perinatal death has been described as up to four times higher in twin pregnancies than in singletons, mainly due to preterm birth, fetal growth restriction, low Apgar scores and extremely low birth weight [5, 6, 23, 25]. In our study, it was found to be 2.5 times higher for the first twin and 3.5 for the second one. This difference between both twins has already been described . In the current study, we also observed a higher risk for fetal and early neonatal death, supporting previous findings from other studies [6, 31].
Cesarean section, including that performed electively, was the most common mode of delivery in twin pregnancy in the present study. The debate on the best mode of delivery is extensive, especially considering higher adverse outcomes for the second twin, and that neither labor nor vaginal delivery is associated with worse perinatal outcomes, since the first twin is in cephalic presentation. There is currently no indication for a policy of planned cesarean delivery, although still some controversies frequently arise among professionals [11, 12, 30, 35, 36].
In the current analysis higher rates of maternal complications are directly related to twin pregnancy (15.9% in twins with APO, 14.1% in twins with no APO and 9.8% in singletons with APO and 5.3% in singletons with no APO). This reinforces some recent studies identifying twin pregnancy as a risk factor for the occurrence of severe maternal morbidity. In a WHO Global Survey analysis, Vogel et al. reported a 1.85 higher risk of occurrence of a severe maternal outcome (maternal death, admission to an intensive care unit, blood transfusion or hysterectomy) between twin pregnancies compared to singletons . Using the new WHO diagnostic criteria for severe maternal conditions, another recent study from our group using the same database identified that twin pregnancies increased twofold the risk of occurrence of PLTC, threefold the risk of MNM and fourfold of occurrence of MD compared with singleton pregnancies . These differences reinforce that twin pregnancy is associated with worse outcomes for both newborns and women. Whether this justifies the need for a more specialized care for women with a twin pregnancy, not only aiming at a good perinatal outcome but also for the maternal outcome, is not completely understood and deserves more specific studies [5, 6, 12, 15, 37].
In the multivariate analysis, both twin pregnancy and maternal complications (PLTC, MNM and MD) still appear as factors independently associated with acute or any adverse perinatal outcome. As already argued, a twin pregnancy is associated with a number of perinatal complications either acute or chronic. However, the relationship between adverse perinatal outcomes and severe maternal conditions reinforces that when the woman develops an adverse condition, the fetus suffers direct consequences (growth restriction and stillbirth) or indirectly by the need of interrupting pregnancy before term, with all the consequences of being preterm. In view of these results and knowing that twin pregnancy is not a modified condition but maternal complications are preventable conditions through improvement of the quality of obstetric care theattention to the pregnant woman is able to modify the perinatal outcomes associated with a twin pregnancy.
There were a few limitations to our study. We had no data on the chorionicity and pregnancies archived by ART/FIV for twin pregnancies – what could be associated with perinatal outcomes. In addition, there is no information at all on ethnicity and BMI of the women, what could also be associated with both twin pregnancy and perinatal outcomes. This was a big international multicenter study that for collecting information on all deliveries during a period of time should use a very short questionnaire to facilitate data collection. Considering twin pregnancy was not the main objective of the study, these variables were not included. Despite all quality control procedures some inconsistency could occur unnoticed from data collected from each individual women with a paper form until the feeding of the electronic database, either by the reviewer or by the system. The reported patterns probably relate more to Low- and Middle-Income Countries settings as a result of the included countries, and therefore the generalization of results for High-Income countries may be limited. In addition, this WHOMCS was mainly performed in secondary and tertiary facilities functioning as referral hospitals with a probable over-representation of maternal complications and maternal/perinatal deaths and considering that the results are based on the facility, countries with low health-facility coverage will be underrepresented, in those facilities the results will probably be lower mainly to severe maternal morbidity. These data might not be representative of maternal outcomes and coverage of essential interventions in smaller facilities or in the community, with variations among countries. In addition, the data collection included only women and newborns up to seven days postpartum or abortion with that cases progressed to maternal and neonatal complications beyond this period may be lost.
On the other hand, we could also highlight some strengths of the current study. The WHOMCS is a large, multi-country database and based on information collected in a standardized way; outcome data on more than 8000 twins were captured, and results obtained identify worse perinatal outcomes for twins, especially for the second and the association between severe maternal morbidity and twin pregnancy. These findings allow the understanding that twin pregnancy is not only associated with obstetric complications but also maternal death. In the clinical practice, these results could assist in the implementation of protocols for identification of risk conditions and maternal and perinatal care.