Maternal, Perinatal and Neonatal Outcomes of Triplet Pregnancies According to Chorionicity: A Systematic Review of the Literature and Meta-Analysis

Triplet pregnancies are rare events that affect approximately 93 in 100,000 deliveries in the world, especially due to the increased use of assisted reproductive techniques and older maternal age. Triplet pregnancies are associated with a higher risk of fetal and maternal morbidity and mortality compared to twins and singletons. Chorionicity has been proposed as a major determinant of perinatal and maternal outcomes in triplet pregnancies, although further evidence is needed to clarify the extent and real influence of this factor. Thus, the aim of this study was to conduct a systematic review of the literature and a meta-analysis of the maternal and perinatal outcomes of triplet pregnancies, evaluating how chorionicity may influence these results. A total of 46 studies with 43,653 triplet pregnancies and 128,145 live births were included. Among the main results of our study, we found a broad spectrum of fetal and maternal complications, especially in the group of monochorionic and dichorionic pregnancies. Risk of admission to NICU, respiratory distress, sepsis, necrotizing enterocolitis, perinatal and intrauterine mortality were all found to be higher in non-TCTA pregnancies than in TCTA pregnancies. To date, our meta-analysis includes the largest population sample and number of studies conducted in this field, evaluating a wide variety of outcome measures. The heterogeneity and retrospective design of the studies included in our research represent the main limitations of this review. More evidence is needed to fully assess outcome measures that could not be studied in this review due to scarcity of publications or insufficient sample size.


Introduction
Having triplets may be overwhelming news for many parents. It is indeed a diagnosis that is usually followed by multiple uncertainties and concerns that obstetric providers are expected to assess [1]. Given the limited data in the current published literature, physicians face a complex challenge when counselling patients on potential risks related to triplet

Eligibility Criteria
To be considered eligible for inclusion, studies had to be written in English and published within the last 15 years (January 2005 to December 2020). Exclusion criteria were as follows: studies using animal models, studies not including abstracts, posters without a full article, studies not differentiating triplet pregnancies from other non-triplet multiple pregnancies, clinical trials or interventional studies regarding fetal reduction or laser therapy and their results as their main objective, systematic reviews, meta-analyses, case series, expert opinions or consensus, secondary analyses, guidelines, studies that excluded TCTA pregnancies, studies that only focused on a specific outcome, and studies only including patients with severe morbidity such as patients who were admitted to the NICU (to avoid a potential bias on the severity of the results).

Selection of Studies
All articles were analyzed independently by two authors (MBC and IAM), and if the title and abstract did not provide useful information for our review or were considered irrelevant for our main purpose, the article was discarded from our analysis. All articles were evaluated according to the inclusion and exclusion criteria previously mentioned. Inconsistencies were assessed by a third author (MRM), discussed by the reviewers, and a consensus was reached between the three investigators.

Data Collection
Data were collected using a standardized format. The data extracted from each article were the authors, year of publication, patient recruitment years, design and study type (single or multicentre, population or observational retrospective study), the country in which the study was conducted, the total number of pregnancies and live births included, the use of ART, and chorionicity, which divided pregnancies into TCTA and both dichorionic and monochorionic, encompassing the latter in the non-TCTA category.
Data regarding maternal outcome included maternal age, parity, body mass index, as well as maternal obstetric morbidities such as diabetes, hypertensive disorders (including gestational hypertension, preeclampsia, and eclampsia), restricted intrauterine growth or small for gestational age, twin to twin transfusion syndrome, congenital malformations, preterm labor, premature rupture of membranes, cervical cerclage, use of corticosteroids, antenatal bleeding and postpartum hemorrhage.
Data on perinatal outcome included mode of delivery (vaginal vs cesarean section), gestational age at the time of delivery, considering preterm birth (<37 weeks), very preterm birth (<32 weeks), and extremely preterm birth (<28 weeks); birth weight, considering low birth weight (<1500 g) and very low birth weight (<1000 g); APGAR at 1 min and 5 min; and sex.
The following data on neonatal outcome was also included: neonatal intensive care unit (NICU) admission and complications, such as respiratory distress syndrome (RDS), hyaline membrane disease, need for mechanical ventilation, use of surfactant, bronchopulmonary dysplasia (BPD), neonatal enterocolitis (NEC), neonatal sepsis, severe intraventricular hemorrhage (IVH type III and IV), retinopathy of prematurity (ROP) and composite neonatal morbidity.
Likewise, we explored mortality outcomes, such as miscarriage (under 22 weeks), intrauterine fetal mortality (IUFM; over 22 weeks), neonatal death (during the first 28 days after birth), and perinatal death, defined as the sum of both intrauterine and neonatal mortality.
First, a descriptive analysis of all the articles included in the systematic review was carried out. In addition, we analyzed which of the articles compared perinatal and maternal outcomes based on chorionicity, and included them in a meta-analysis for each of the available outcome measures. Finally, graphic representations of the results were obtained.

Risk of Bias Assessment and Statistical Analysis
Quality assessment of the included studies was performed using the Newcastle-Ottawa Scale (NOS) for case-control studies. According to this scale, each study is evaluated from three perspectives: the selection of the study groups (the representativeness of the exposed and unexposed cohort, their exposure, and the fact that the variable of interest was not present from the start of the study), the comparability of the groups, and the precision in the measurement of the variable of interest. According to NOS, a study can be awarded a maximum of one star for each item in the selection and outcome categories of interest and a maximum of two stars for comparability.
Different meta-analyses of proportions have been performed to estimate the combined incidence of our main variables of interest (maternal, perinatal, neonatal, and mortality outcomes) in TCTA and non-TCTA triplet pregnancies. Meta-analyses were only performed when at least 3 studies could be included. The heterogeneity of the studies was evaluated using the I 2 statistic, considering a value >50% as indicative of high heterogeneity. Given the high variability found in many of the different studies, random effects models were chosen to estimate the pooled odds ratios. Under the assumption of random effects, in the weighting of the studies, both their own variability (intrastudy variability) and the variability among studies (interstudy variability) are considered. When random effects models are assumed, results are usually more conservative than those obtained under the assumption of fixed effects, obtaining wider confidence intervals for the combined odds ratios.
Potential publication bias was assessed using Egger's regression asymmetry test for small study effects. Given the low number of studies included for each of the meta-analyses, the application of other formal tests for the evaluation of publication bias, such as the funnel plot, was not considered convenient.

Results
The initial search revealed 1637 citations potentially eligible for inclusion, of which 860 met our inclusion criteria, and 701 were excluded based on their title or abstract. A total of 159 full manuscripts were reviewed, and 46 were ultimately included in the systematic review. Among the 46 studies, 12 were included in the meta-analysis, which studied the differences in maternal, perinatal, and neonatal outcomes when comparing TCTA with non-TCTA pregnancies. (Figure 1). Table 1 shows the characteristics of the studies included in the systematic review. Most of them were published in North America (37%) and Europe (28.33%) and carried out in a single centre (54.4%). The predominant study design was retrospective observational (36; 78.3%), followed by 6 (13%) population studies and only 4 (8.7%) prospective observational studies. Sixteen      Tables 2-6 show the data collected from the studies included in the systematic review.    A total of 46 studies with 43,653 triplet pregnancies and 128,145 live births were included. The number of outcome measures collected in the systematic review were: 6 gestational and maternal characteristics, 11 obstetric complications, 13 perinatal outcomes, 11 neonatal outcomes, and 6 mortality outcomes.

Maternal and Gestational Characteristics
Regarding the method of conception, 2962 (68.06%) pregnancies were obtained by ART. According to chorionicity, 64.1% were TCTA pregnancies, 24.88% were dichorionic triamniotic pregnancies, and 4.25% were MCTA pregnancies. It should be noted that three of the studies included in the review excluded monochorionic pregnancies [24][25][26]. The mean maternal age was 31.6 years ( Table 2). Table 3 describes the obstetric complications observed in the studies included in the systematic review. Preterm labor occurred in 26.39% of the pregnancies, and premature rupture of membranes in 11.52%. A total of 41.43% of the patients underwent cervical cerclage, although the heterogeneity was significant, varying from 3 to 63% in the 11 articles that included this variable. There were 4884.3 (34.92%) cases of intrauterine growth restriction or small for gestational age. TTTS complicated 2.79% of all the pregnancies included in the review. Two of the articles from the review specifically excluded pregnancies complicated by TTTS; thus, this percentage could be slightly underestimated [27,28]. Fetal malformations were diagnosed in 6.54% of pregnancies, although this was also an exclusion criterion for six of the included studies [25,[28][29][30][31][32].

Mortality
Miscarriages (<22 weeks) and intrauterine fetal mortality (IUFM) were close to 5%. On the other hand, the number of neonatal deaths was 44,089 (4.42% of live newborns for whom this variable was reported, which was 34.41% of live newborns). Four studies reported maternal mortality, which was one maternal death out of 721 pregnant women (0.14%) ( Table 6).

Meta-Analysis Based on Chorionicity
A total of 13 articles were included in the meta-analysis, with 2188 pregnant women, 5790 fetuses, and 5441 live newborns. Data were collected on seven maternal and perinatal characteristics, eight neonatal, and three mortality outcome measures. Among all the variables studied, those that accounted for the largest number of studies and incorporated most data were perinatal mortality (ten studies with 5583 fetuses), followed by intrauterine mortality (nine studies with 5367 fetuses) and ART (nine studies with 1807 pregnant women). The variable for which the least data are available is very preterm birth (<32 weeks) (three articles with 225 pregnancies). Table 7 describes the analysis of the maternal and perinatal characteristics. Chorionicity had no statistically significant differential impact on the mode of delivery or on prematurity. In the very preterm (<32 weeks) and very low birth weight infants, a tendency to lower risk was observed in TCTA pregnancies compared to non-TCTA pregnancies. Additionally, TCTA pregnancies were three times more likely to come from ART.
Regarding neonatal outcomes (Table 8), it is interesting to note that newborns from TCTA pregnancies had a significantly lower risk of being admitted to the NICU (OR 0.57; 95% CI 0.44-0.72) as well as of developing respiratory distress (OR 0.46; 95% CI 0.22). −0.97), sepsis (OR 0.57; 95% CI 0.37-0.89) and necrotizing enterocolitis (OR 0.32; 95% CI 0.15-0.69). However, no statistically significant differences were observed for very low birth weight (<1500 g), bronchopulmonary dysplasia, retinopathy of prematurity, or severe intraventricular hemorrhage stages III-IV.  Finally, in terms of mortality (Table 9), pregnancies without any monochorionic component (TCTA) showed a lower risk of both intrauterine (OR 0.29; 95% CI 0.14-0.62) and perinatal mortality (OR 0.32; 95% CI 0.20). −0.53).   Tables 7-9. As shown in these figures, due to the scarcity of articles that reported data on intrauterine growth restriction, APGAR < 7 at 5 min, and use of surfactant, the meta-analysis for these outcomes was not carried out.   Tables 7-9. As shown in these figures, due to the scarcity articles that reported data on intrauterine growth restriction, APGAR <7 at 5 min, and u of surfactant, the meta-analysis for these outcomes was not carried out.

Main Findings
Both our systematic review and meta-analysis collected data on a wide range of outcome measures from a large population sample. This systematic review brings together a large population sample from a large number of studies (46 studies with a total of 43,653 triplet pregnancies and 128,145 live births), collecting information on 47 variables [14,16,21,. Similarly, this meta-analysis included many articles with a large total population sample (12 studies, 2188 pregnant women, 5790 fetuses, and 5441

Main Findings
Both our systematic review and meta-analysis collected data on a wide range of outcome measures from a large population sample. This systematic review brings together a large population sample from a large number of studies (46 studies with a total of 43,653 triplet pregnancies and 128,145 live births), collecting information on 47 variables [14,16,21,. Similarly, this meta-analysis included many articles with a large total population sample (12 studies, 2188 pregnant women, 5790 fetuses, and 5441 live newborns), collecting up to 18 variables [14,16,21,24,26,33,36,41,44,56,57,63]. The vast majority of the studies have been published in the last 5 years (41.3%), which highlights the growing interest in triple pregnancies and their maternal and perinatal outcomes.
To date, the only meta-analysis published on perinatal morbidity and mortality in triplet pregnancies is the review conducted by Curado et al. [18], which concluded that chorionicity is a determining factor of perinatal morbidity in triplet pregnancies. Among its main findings, it is worth noting that non-TCTA triplet pregnancies show increased neonatal morbidity and higher intrauterine and perinatal mortality than TCTA pregnancies. Our main results are consistent with these findings and contribute to these statements, since the sample and the number of studies on which they are based have increased, in addition to enriching the existing literature with a greater number of variables.
Regarding our main results from the systematic review, it is striking that many articles lack information about baseline maternal characteristics, such as complications directly related to multiple pregnancies. Miscarriages in the first trimester are also underreported in comparison to the rest of the mortality analysis. Finally, the scarcity of data regarding maternal morbidity and mortality is surprising.
Regarding our meta-analysis, the main results are that the risk of admission to NICU or presenting neonatal complications such as respiratory distress, sepsis, or necrotizing enterocolitis is greater in non-TCTA pregnancies than in TCTA pregnancies. Perinatal and intrauterine mortality are also more frequent in non-TCTA triplet pregnancies.

Comparison with Previous Literature
In the meta-analysis by Curado et al. [18], a smaller number of studies were analyzed with a smaller population sample (9 studies, 1373 pregnancies, 4119 fetuses, and 3669 live births) and a smaller number of variables. Curado et al. analyzed intrauterine, perinatal, and neonatal mortality and gestational age at birth as main outcomes. The secondary outcomes were composite neurological, respiratory and infectious morbidity. However, they did not carry out an extensive systematic review of the literature on maternal and obstetric characteristics or perinatal and neonatal complications separately.

Socioeconomic and Geographic Determinants
The socioeconomic background has an undeniable role as a key determinant of maternal health and obstetric outcomes [67,68]. It has an impact on patients' access to health resources, ART, tertiary care centres, and NICU availability. Ref. [69], In our study, most articles included had been published in Western countries (35 of the 46 publications of the systematic review), with only one study in Latin America, and no representation from the African continent. Furthermore, only one study was from an upper-middle-income country, and two were from a lower-middle-income country. This is even more apparent among the studies that were finally included in the meta-analysis, thus influencing the results of our study. It is therefore vital for physicians to counsel their patients according to their social, cultural, and economic background, taking into account the importance of such differences on morbimortality outcomes in triplet pregnancies. Moreover, considering that 68% of all triplet pregnancies in our systematic review were a result of ART, special attention must be paid to the selection of patients and strategies implemented to minimize the number of triplet pregnancies and their complications [70][71][72][73][74][75][76][77][78][79]. When analyzing the impact of ART on chorionicity in the meta-analysis, TCTA pregnancies were found to be up to three times more frequent, which aligns with the results of Fennessy et al., who found TCTA pregnancies to be four times more likely after ART [44].

Maternal and Perinatal Characteristics
In the systematic review, the pooled proportion of the mode of delivery was 88% for cesarean section. Prematurity (before 37 weeks) occurred in 92.34% of all triplet pregnancies, with 40% being highly preterm (under 32 weeks). This is directly related to the fact that newborns from triplet pregnancies have high rates of preterm labor and premature rupture of membranes, even though in many cases, preterm termination of pregnancy is scheduled [59]. In either case, it is not surprising that corticosteroids for antenatal lung maturity are frequently administered. Prematurity is also related to low birth weight (more than a third showed low birth weight and 11.47% presented very low birth weight) and with lower average APGAR scores, although most recover at minute 5 as a result of the advanced neonatal and resuscitation care and resources available today.
In our meta-analysis, regarding the mode of delivery, there were no significant differences based on chorionicity, as expected, given that the current guidelines recommend delivery by scheduled cesarean section regardless of chorionicity [30]. Nevertheless, a trend towards greater prematurity and lower birth weight was observed in mono-or dichorionic triplet pregnancies compared to trichorionic pregnancies. These findings align with what has been known to occur in twin pregnancies, although our analysis did not reach statistical significance.

Neonatal Outcomes
In the systematic review, 78% of newborns were admitted to the NICU, the majority due to respiratory morbidity. Surfactant was administered to one-fifth (22.7%) of the newborns, and one-third (34.49%) required mechanical ventilation. As mentioned previously, these findings are consistent with the significant prematurity of newborns from triplet pregnancies.
In our meta-analysis, admission to the NICU in newborns of TCTA pregnancies was almost 50% compared to those of non-TCTA pregnancies (OR 0.57; 95% CI 0.44-0.72), which could be related to the trend, which did not achieve statistical significance, of lower risk for prematurity and low birth weight, hence with the lower risk of other neonatal complications. It was observed that newborns of TCTA pregnancies have less than half the risk of developing respiratory distress, which is the main complication of these newborns [37,41,48]. However, neither the administration of surfactant nor bronchopulmonary dysplasia were influenced by the chorionicity of triplet pregnancy, possibly due to their low prevalence and the few articles that reported these variables.
Other neonatal complications resulting from the increased prematurity in triplet pregnancies, are sepsis (5.48%), severe intraventricular hemorrhage (4.6%), retinopathy of the premature (4.12%) and necrotizing enterocolitis (1.67%). [14,24,28,29]. The risk of developing sepsis and necrotizing enterocolitis was higher in newborns from non-TCTA triplet pregnancies, while there were no differences for severe intraventricular hemorrhage or retinopathy. These last two events are rare events, and it is possible that greater statistical power is required to reach statistical significance. Regarding retinopathy, perhaps the inclusion of the article by Kawaguchi et al. [14] that found no significant differences in this disease despite a large study sample prevented the global trend from being statistically significant.

Mortality
Regarding the results of the systematic review, it is striking that miscarriages before 22 weeks were only reported for 2.88% of the fetuses included, and in fact, 32 studies excluded them from their analysis, so the interpretation of this data should be cautious. Intrauterine mortality was also excluded from 16 articles, although it was collected in 26.17% of the pregnancies included, and neonatal mortality was only collected in 34.41% of the live newborns included in the review. Even so, the sample on which these variables were reported was between 40,000 and 44,000 fetuses, so we consider that the data obtained can be considered representative. Intrauterine fetal mortality was found to affect 5% of all fetuses in triplet pregnancies, and neonatal mortality in 4.42% of newborns.
According to the results of our meta-analysis, intrauterine mortality is decreased in fetuses from TCTA pregnancies compared to those from non-TCTA pregnancies [14,16,18,21,24,33,41]. The pathophysiology of a shared placenta may be the underlying explanation of our findings, given the risk of developing TTTS, restricted intrauterine growth, anemia-polycythemia sequence, and cardiovascular compromise in monoamniotic fetuses [14,16,80].
As for neonatal mortality, we found no statistically significant differences when comparing TCTA pregnancies with non-TCTA pregnancies. One possible explanation for this finding would be that, despite greater neonatal morbidity in terms of the previously described complications, access to NICUs and high-complexity centers makes it possible to reduce neonatal mortality without differentiating newborns based on their chorionicity. However, there seems to be a certain temporal trend in the results of neonatal mortality in the articles that report it, showing higher neonatal mortality in older studies. This could be explained by the progress made in fetal surgery and interventionism in the last decade.
The risk of perinatal mortality was also shown to be significantly increased in fetuses and newborns from non-TCTA pregnancies, probably at the expense of intrauterine mortality.
Finally, there is a notable lack of data on maternal mortality in triplet pregnancies. Only four studies reported maternal mortality as a study outcome, which was an isolated case of maternal death out of a total of 721 pregnant women included in these studies (0.14%) [35,39,53,55]. Similarly, only one study reported a composite variable on maternal morbidity [34], which was present in 32 of the 35 patients on which this data was reported. It is surprising how little thought is given to maternal mortality and morbidity in studies on triplet pregnancies, especially when increased maternal mortality has been described in multiple pregnancies [81]. In fact, multiple pregnancies are risk factors for serious maternal complications, such as hypertensive disorders of pregnancy [17] or postpartum hemorrhages [82]. Maternal mortality, although less and less frequent in our environment, is always a tragic event, and in the case of multiple pregnancies, we cannot forget the scope of its repercussions. The loss of a mother in a triplet pregnancy results in three newborns losing a parent, with the social, economic, and emotional consequences that this might entail, thus tripling its impact [83].

Strengths
To our knowledge, this is the largest systematic review and meta-analysis conducted to date on maternal, fetal, and neonatal morbidity and mortality of triplet pregnancies and their relationship to chorionicity. One of the main strengths of this study is the gathering of such a large population sample from an event as rare as triplet pregnancies.
As mentioned above, the main difference with respect to the meta-analysis published by Curado et al. [18] is that they focused on variables related to fetal and neonatal mortality, including fewer composite secondary variables, and fewer studies (and consequently, smaller population sample and data) for analysis. Our meta-analysis, on the one hand, provides robust and comprehensive evidence to the common results with which it aligns, since it includes more data from more studies. On the other hand, this study adds richness and diversity to the currently published data since it does not focus exclusively on neonatal or mortality outcomes but also analyzes maternal, gestational, and obstetric outcomes.
Finally, the authors believe that conducting a systematic review prior to performing the meta-analysis is also one of the strengths of this article. It provides the clinician with a highly applicable global vision for the risks and complications that may occur in these high-risk pregnancies. It also facilitates the interpretation of the results of the meta-analysis on the impact that chorionicity has on morbidity and mortality. As it has been suggested, the subgroup analysis according to chorionicity is of special relevance, both for patient counsel as well as for obstetric clinical management.

Limitations
The main limitations of our meta-analysis are related to the retrospective observational designs of the studies included, their relatively small population samples, and their high heterogeneity. Finally, the publication bias analyzed using Egger's test in the variables included in the meta-analysis could be present in at least four of them.
Additionally, the review includes articles from 2005 to 2020, an extended period of time during which the management of triplet pregnancies and their newborns may have shifted substantially.
Another limitation is the lack of information not reported by numerous studies about variables that the authors considered relevant when evaluating morbidity and mortality in triplet pregnancies. Few articles describe complications intrinsic to multiple pregnancies, such as TTTS or postpartum hemorrhage. The current meta-analysis has evaluated the influence of chorionicity on fetal, maternal, and perinatal outcomes. However, other factors such as the mode of conception (different ARTs and spontaneous conception) have not been studied in detail in our meta-analysis. Addressing how fetal, perinatal, and maternal morbidity and mortality in triplet pregnancies might differ when comparing the different modes of conception is an interesting question for further research.
Finally, we consider that maternal mortality is a relevant outcome that should not be left in the background when addressing the morbidity and mortality of triplet pregnancies. Indeed, triplet pregnancies have been found to be, not only high-risk events for the newborns but also for the mothers themselves.
Finally, the authors are aware that recent studies that might have been published after finishing our systematic review, such as Badreldin et al. [84] and Mitsiakos et al. [85], were not included in this article and should be taken into account in future research.

Conclusions
The results of the systematic review described a high incidence of maternal, fetal, and obstetric morbidity, largely related to the prematurity of triplet pregnancies. In our meta-analysis, we observed a significant increase in maternal, fetal, and neonatal morbidity and mortality in non-TCTA triplet pregnancies compared to TCTA. Further research is needed to assess some of the outcomes that either could not be included in our study due to the scarcity of publications reporting them, or in which statistical significance was not reached, probably due to an insufficient sample size.