Factors and outcomes for placental anomalies: An umbrella review of systematic reviews and meta-analyses.

We conducted an umbrella review using standardised methodology and reported our findings according to the PRISMA and MOOSE guidelines [ 12 , 13 ]. The protocol was registered in PROSPERO (CRD42022300160). We searched PubMed, Embase, Web of Science, and the Cochrane Library from inception to 8 February 2023 to identify systematic reviews and meta-analyses of studies that analysed the association between factors and outcomes and placental anomalies, including PP, PAS, and VP. We conceptualised the search strategy around the following keywords: ‘morbidly adherent placenta,’ ‘abnormal placentation,’ ‘placenta previa,’ ‘placenta accreta spectrum disorders,’ or ‘vasa previa,’ combined with ‘systematic review’ or ‘meta-analysis’ (File S1 in the Online Supplementary Document ). We did not impose any language limitations when choosing the appropriate studies. We also manually searched the reference lists of relevant reviews and performed forward and backward citation chaining. Two researchers (DF and LL) independently screened and evaluated the full texts of potentially eligible articles. Disagreements were resolved through discussion. We included all meta-analyses of observational studies that investigated the causes and consequences of PP, PAS, and VP, irrespective of their publication date. We excluded systematic reviews without meta-analyses, animal studies, genetic studies, conference abstracts, letters, and editorials (File S2 in the Online Supplementary Document ). If we found two similar articles, we included the most recent in the analysis, as it likely comprises more studies and/or is of the highest quality per the AMSTAR 2 tool [ 14 ]. Two researchers (DF and JR) extracted the following data independently: first author, journal and year of publication, factor(s) and outcome(s) of interest (PP, PAS, or VP), and number of studies analysed. Where possible, they extracted other data such as estimated value, 95% confidence intervals (CIs), P -values, the number of participants in each groups, analytical data model, heterogeneity, and small-study effects. Disagreements in the extraction process were resolved through consensus. Two investigators (DF and YM) assessed the methodological quality of each included meta-analysis using the AMSTAR 2 tool, a reliable and valid instrument facilitates the quality assessment of meta-analysis and systematic reviews [ 14 – 16 ] (File S3 in the Online Supplementary Document ). AMSTAR 2 categorises the quality of a meta-analysis on a scale from critically low to high, based on 16 predefined items [ 17 ]. Each item has three responses – yes, partial yes, and no; items 2, 4, 7, 9, 11, 13, and 15 are key to the evaluation. Overall, the included articles are ranked as high, moderate, low, or critically low. We noted which associations met the following criteria to determine the strength of the epidemiologic evidence (i.e. the confidence in the effect estimate): Precision of the estimate (i.e. P -value <0.001, a threshold associated with significantly less false-positive results, and over 1000 cases with the disease); consistency of results ( I 2 0.10); and no evidence of small-study effects ( P -value >0.10) [ 18 – 20 ]. We ranked the strength of epidemiologic evidence as high (when all the above criteria were satisfied), moderate (if a maximum of one criterion was not satisfied and a P -value <0.001), or weak ( P -value <0.05). If the P -value was not reported, we calculated the 95% CI of the pooled effect estimate using a standard method. We displayed the derived random-effects estimates using forest plots to show the relationships between various factors or outcomes and PP, PAS, or VP. We computed and presented the random-effects estimate whenever a fixed-effects model was originally used. In practice, these different measures (rate ratio (RR), odds ratio (OR), and hazard ratio (HR)) of effect yield similar estimates, since PP, PAS, or VP is a rare occurrence. We converted these various measurements to ORs using a standardised methodology [ 21 ]. We used the Shapiro-Wilk test was used to test for normality of continuous variables and presented data with skewed distribution using medians and interquartile ranges (IQRs). We considered P -value <0.05 statistically significant, except for heterogeneity and small-study effects. All statistical analyses were conducted in Stata software, version 12.0 (StataCorp, College Station, Texas, USA).

We retrieved 689 studies, of which 34 met the eligibility criteria ( Figure 1 ). They were published between 2003 and 2023. Twenty-seven articles focussed on PP [ 9 , 10 , 15 – 17 , 22 – 43 ], eight on PAS [ 10 , 11 , 25 , 44 – 48 ], and only two on VP [ 9 , 49 ]. Two articles [ 10 , 25 ] focussed on PP and PAS and one [ 9 ] on PP and VP. We found 55 factors (27 for PP, 22 for PAS, and 6 for VP) and 16 outcomes (12 for PP, and 4 for VP) in the 34 included studies. No meta-analysis evaluated the outcomes for PAS. There was a median of six primary studies per evidence synthesis (IQR = 3–12) with a median number of 1772 cases (IQR = 359–9532) and 278 459 subjects (IQR = 48 209–1 055 206) ( Table 1 ). Flowchart of literature search. Characteristics, quantitative synthesis, and assessment of the included meta-analyses AMA – advanced maternal age, APGAR – appearance, pulse, grimace, activity, and respiration, ART – assisted reproductive techniques, CHB – chronic hepatitis B, CS – caesarean section, eSET – elective single embryo transfer, ET – embryo transfer, HDP – hypertensive disorders of pregnancy, NS – not significant, OR – odds ratio, PAS – placenta accreta spectrum disorders, PP – placenta previa, SES – socioeconomic status, SGA – small for gestational age, SISP – short interval between prior caesarean section and subsequent pregnancy (<23 mo), STPP – second trimester placenta previa, TNF – tumour necrosis factor, UAE – uterine artery embolization, VCI – velamentous cord insertion, VP – vasa previa, wk – weeks. Eleven studies were of ‘high’ (32.35%), twenty-one of ‘low’ (61.77%), and two of ‘critically low’ quality (6.45%). The most frequent flaw was the absence of a registered protocol (item 2: 23 meta-analyses (67.65%)) and inadequacy of the literature search (item 4: 2 meta-analyses (5.88%)) (File S3 in the Online Supplementary Document ). Seven factors (maternal cocaine use (for PP), uterine leiomyoma (for PP), prior abortion (spontaneous) (PP), threatened miscarriage (PP), maternal obesity (PP), maternal smoking (PAS), male foetus (PAS)) had high epidemiological evidence. Twelve factors (advanced maternal age (>45 years) (PP), advanced maternal age (>35 years) (PP), assisted reproductive techniques (singleton pregnancy) (PP), prior abortion (induced) (PP), prior uterine artery embolization (PAS), placenta previa (PAS), assisted reproductive techniques (frozen embryo transfer) (PAS), prior caesarean section (PAS), prior uterine surgery (PAS), multiparity (PAS), multiple gestations (PAS), low socioeconomic status (PAS)) and six outcomes (preterm delivery (<32 weeks) (PP), blood transfusion in caesarean section (PP), preterm delivery (<34 weeks) (PP), preterm delivery (<28 weeks) (PP), neonatal death (PP), neonatal intensive care unit (PP)) had moderate epidemiological evidence. Twenty-two factors (13 for PP, 4 for PAS, and 5 for VP) and eight outcomes (4 for PP, and 4 for VP) showed significant association, but with weak credibility. Other fourteen factors (5 for PP, 8 for PAS, and 1 for VP) and two outcomes (for PP) showed no statistically significant estimates ( Table 1 , Figure 2 , and Figure 3 ; File S4 in the Online Supplementary Document ). The results of factors and outcomes for placenta previa. AMA – advanced maternal age, ART – assisted reproductive techniques, CHB – chronic hepatitis B, CI – confidence interval, CS – caesarean section, eSET – elective single embryo transfer, ET – embryo transfer, HDP – hypertensive disorders of pregnancy, IBD – inflammatory bowel disease, IMH – isolated maternal hypothyroxinaemia, IUGR – intrauterine growth restriction, NICU – neonatal intensive care unit, NS – not significant, PP – placenta previa, SGA – small for gestational age, TNF – tumour necrosis factor, UAE – uterine artery embolization. The results of factors for placenta accrete spectrum disorders and the factors and outcomes for vasa previa. AMA – advanced maternal age, ART – assisted reproductive techniques, CI – confidence interval, CILTTU – cord insertion in the lower third of the uterus at first trimester ultrasound, CS – caesarean section, ET – embryo transfer, HDP – hypertensive disorders of pregnancy, NS – not significant, PAS – placenta accreta spectrum disorders, PP – placenta previa, SES – socioeconomic status, SGA – small for gestational age, SISP – short interval between prior caesarean section and subsequent pregnancy (<23 months), STPP – second trimester placenta previa, UAE – uterine artery embolization, VCI – velamentous cord insertion, VP – vasa previa.

In this umbrella review, we identified seven factors demonstrating high strength of epidemiologic evidence, as well as twelve factors and six outcomes demonstrating moderate strength of epidemiologic evidence. Meanwhile, the estimate of effect’s degree of confidence was weaker another 22 factors and 8 outcomes. The methodological quality across meta-analyses differed slightly. A previous umbrella study has attempted to explore the risk factors for PP [ 7 ] and found seven high and two weak risk factors. We evaluated a further sixteen factors and found two which had two high epidemiological evidence (threatened miscarriage and maternal obesity). We also twelve outcomes – six moderate, four weak, and two of insignificant epidemiologic evidence. Compared to previous study [ 7 ], we not only explored risk factors, but also reported outcomes for PP, addressing the existing gap in the literature. A previously published umbrella review evaluated and found seven environmental risk factors for PAS [ 8 ]; in our study, we found another 15 – one of high, five of moderate, three of weak, and six of statistically insignificant epidemiological evidence. The factors found in the previous umbrella review and our study differ slightly. This includes four factors for PP (hypertensive disorders of pregnancy; assisted reproductive techniques (singleton pregnancy); assisted reproductive techniques (twin pregnancy); and endometriosis) and one for PAS (hypertensive disorders of pregnancy). This is mainly caused by the different included articles. For example – regarding hypertension in pregnancy in PAS, we selected the latest study by Li et al. [ 44 ] rather than the one by Wang et al. [ 50 ] study. For endometriosis in PP, we chose the study by Matsuzaki et al. [ 24 ], not the one by Zullo et al. [ 51 ]. We determined by detailed evaluation and comparison that the two studies by Li et al. [ 44 ] and Matsuzaki et al. [ 24 ] are more recent and have higher quality (per the AMSTAR 2 tool), and are also more in line with our inclusion criteria. VP is a rare but life-threatening obstetric disease, with an incidence from 0.46 to 0.60 for every 1000 deliveries, according to two systematic reviews [ 5 , 49 ]. Each effect size is relatively large for the six factors and four outcomes in VP. However, the epidemiologic evidence for this estimation was weak, mainly because the sample size of the included studies was too small. Therefore, more research is needed for this rare obstetrical condition. While three other systematic reviews and meta-analyses assessed VP [ 5 , 52 , 53 ], their topics were not related to and could not be evaluated in this umbrella review. The previous review noted antepartum haemorrhage, postpartum haemorrhage, and septicaemia (among others) as adverse pregnancy outcomes for placental anomalies [ 8 ]. However, we did not find several of these outcomes in our review. Therefore, more systematic reviews and meta-analyses are needed to evaluate the effect of placental anomalies on other adverse pregnancy outcomes in the future. The AMSTAR 2 tool assisted us in evaluating the methodological quality of the included studies. The most common flaw and reason for downgrading the quality assessment was the lack of protocol (n = 23). The most recent registration among the included studies took place in 2016. While preregistration of study protocols was an uncommon practice until recently, especially in the field of obstetrics, we found that implementing it could improve the quality of published meta-analysis. To our knowledge, this is the first umbrella review to provide a broad overview of the scope and validity of the reported associations of various factors and pregnancy outcomes with placental anomalies, including PP, PAS, and VP. However, some limitations should be considered. First, our results exclusively rely on meta-analyses of observational studies, and are thus subject to the same limitations – including over/under-reporting, recall bias, and reverse causation. Second, because most of included meta-analysis were of low quality according to the AMSTAR 2 assessment, the results should be interpreted with caution. Third, we found an uneven covariate mix across primary studies, which could have affected our findings so that it might be difficult to assess both the effect size or its direction. Fourth, the results may not necessarily correspond well with the clinical studies. Finally, due to the lack of raw data, we are not able to conduct further analyses.

In this review, we provide a comprehensive overview and critical evaluation of the contributing factors and outcomes of placental anomalies. Across 55 factors and 16 outcomes, seven (five for PP and two for PAS) and 12 factors (four for PP, and eight for PAS) and six outcomes (for PP) showed high/moderate epidemiologic evidence for placental anomalies. The results can be used to reassure women or refer them to to pre-conception counselling clinics or antenatal clinics. Clinicians should consider and communicate these factors and outcomes when counselling their patients. Regarding future research, more broadly implementing the reporting criteria and registering observational studies that test hypotheses could help strengthen the evidence. Likewise, new meta-analyses are needed to obtain, evaluate, and validate the novel and strongest evidence.