Prevalence of placenta praevia by world region: a systematic review and meta-analysis

Authors


Corresponding Author Jenny A. Cresswell, London School of Hygiene & Tropical Medicine, Keppel St, London WC1E 7HT, UK. E-mail: jenny.cresswell@lshtm.ac.uk

Abstract

Objectives

(i) To estimate the prevalence burden of placenta praevia in each world region, and (ii) to investigate potential sources of heterogeneity.

Methods

Systematic review of the literature and random-effects meta-analysis. Potential sources of heterogeneity were investigated using meta-regression.

Results

The overall prevalence of placenta praevia was 5.2 per 1000 pregnancies (95% CI: 4.5–5.9). However, there was evidence of regional variation (P = 0.0001); prevalence was highest among Asian studies (12.2 per 1000 pregnancies; 95% CI: 9.5–15.2) and lower among studies from Europe (3.6 per 1000 pregnancies; 95% CI: 2.8–4.6), North America (2.9 per 1000 pregnancies; 95% CI: 2.3–3.5) and Sub-Saharan Africa (2.7 per 1000 pregnancies; 95% CI: 0.3–11.0). The prevalence of major placenta praevia was 4.3 per 1000 pregnancies (95% CI: 3.3–5.4).

Conclusion

The prevalence of placenta praevia is low at around 5 per 1000 pregnancies. There is some evidence suggestive of regional variation in its prevalence, but it is not possible to determine from existing data whether this is due to true ethnic differences or other unknown factor(s).

Abstract

Objectif

(i) Estimer la prévalence du placenta praevia dans chaque région du monde, (ii) investiguer les sources potentielles d'hétérogénéité.

Méthodes

Revue systématique de la littérature et méta-analyse à effets aléatoires. Les sources potentielles d'hétérogénéité ont été étudiées à l'aide d'une méta-régression.

Résultats

La prévalence globale de placenta praevia était de 5,2 pour 1000 grossesses (IC95%: 4,5 à 5,9). Cependant il y a des preuves de variations régionales (P = 0,0001), la prévalence était la plus élevée dans les études asiatiques (12,2 pour 1000 grossesses; IC95%: 9,5 à 15,2) et plus faible dans les études en provenance d'Europe (3,6 pour 1000 les grossesses IC95%: 2,8 à 4,6), d'Amérique du nord (2,9 pour 1000 grossesses; IC95%: 2,3 à 3,5) et d'Afrique subsaharienne (2,7 pour 1000 grossesses; IC95%: 0,3 à 11,0). La prévalence du placenta praevia majeur était de 4,3 pour 1000 grossesses (IC95%: 3,3 à 5,4).

Conclusion

La prévalence du placenta praevia est faible, environ 5 pour 1000 grossesses. Il existe des preuves suggérant des variations régionales dans la prévalence, mais il n'est pas possible de déterminer à partir des données existantes si cela est dû à de véritables différences ethniques ou à d'autres facteurs inconnus.

Abstract

Objetivo

(i) Calcular la prevalencia de la carga de placenta previa en cada región del mundo; (ii) investigar las fuentes potenciales de heterogeneidad.

Métodos

Revisión sistemática de la literatura y meta-análisis de efectos aleatorios. Las fuentes potenciales de heterogeneidad se investigaron utilizando una meta-regresión .

Resultados

La prevalencia total de placenta previa era de 5.2 por 1000 embarazos (IC 95%: 4.5–5.9). Sin embargo había evidencia de variación regional (P = 0.0001); la prevalencia era mayor en estudios asiáticos (12.2 por cada 1000 embarazos; IC 95%: 9.5–15.2) y menor en estudios de Europa (3.6 por cada 1000 embarazos; IC 95%: 2.8–4.6), de América del Norte (2.9 por cada 1000 embarazos; IC 95%: 2.3–3.5) y África subsahariana (2.7 por cada 1000 embarazos; IC 95%: 0.3–11.0). La mayor prevalencia de placenta previa era de 4.3 por cada 1000 embarazos (IC 95%: 3.3–5.4).

Conclusión

La prevalencia de placenta previa es baja con cerca de 5 por cada 1000 embarazos. Hay alguna evidencia que sugiere una variación regional de la prevalencia, pero no es posible determinar, utilizando los datos existentes, si se debe realmente a diferencias étnicas o a otros factores desconocidos.

Introduction

Placenta praevia is a potentially severe obstetric complication where the placenta lies within the lower segment of the uterus, presenting an obstruction to the cervix and thus to delivery. Risk factors for placenta praevia include those that increase the likelihood of uterine scar tissue (including higher parity, prior caesarean delivery or prior abortion) or multiple gestations (Ananth et al. 1997; Faiz & Ananth 2003; Gurol-Urganci et al. 2011).

A previous systematic review by Faiz and Ananth 2003 of studies published between 1966 and 2000 calculated the prevalence of placenta praevia to be 4.0 per 1000 pregnancies; both the search strategy and evidence synthesis of this review emphasised North American literature. Our study, conducted as part of a larger project aiming to quantify the global burden of maternal haemorrhage (Calvert et al. 2012), updates these results and puts particular emphasis on capturing and including studies originating outside of the USA and high-income countries.

Placenta praevia can result in life-threatening maternal complications such as haemorrhage and shock and in adverse infant outcomes such as prematurity, stillbirth and neonatal death (Crane et al. 1999, 2000; Bahar et al. 2009). National hospital surveillance data from the USA demonstrate a case-fatality rate of 17.3 deaths per 100 000 White women with placenta praevia and 40.7 deaths per 100 000 among Black women (Tucker et al. 2007). The case-fatality rate in low-income settings is likely to be considerably higher due to reduced antenatal screening and lack of quality emergency obstetric care. The primary objective of this paper was to estimate the prevalence of placenta praevia in each world region; the secondary objective was to investigate potential sources of heterogeneity.

Methods

Data sources & search strategy

This review was part of a larger study investigating the prevalence of maternal haemorrhage and the causes of haemorrhage (Calvert et al. 2012). The databases EMBASE, Medline, Popline, CAB Abstracts, African Index Medicus, Eastern Mediterranean Region Index Medicus, Latin American and Caribbean Center on Health Sciences Information, and Western Pacific Region Index Medicus were searched using thesaurus and free-text terms to identify literature published from 1 January 1990 onwards. The initial search strategy identified studies published up until 2009; the search was later updated in April 2012.

The search strategy included thesaurus and free-text terms relating to ‘haemorrhage’, ‘placenta praevia’ and ‘bleeding’. A particular effort was made to identify literature from low-income settings by searching the WHO regional databases. The reference lists of eligible studies were hand searched to identify further publications. The full search strategy is available upon request.

Inclusion criteria

Studies were eligible for inclusion if they reported the number of cases of placenta praevia, along with a suitable denominator for the total number of deliveries or births in the population, and the median year of the study was 1990 onwards. To capture population-representative data, facility-based studies were excluded if local or national (if local data unavailable) skilled birth attendance was <95%. Studies that reported the total number of caesarean deliveries as the denominator were excluded.

Case reports and publications that did not report original data were excluded; all other study types were eligible. Studies with a sample size of fewer than 30 deliveries were excluded. No language restrictions were imposed. Where cases were potentially reported in multiple publications due to overlap in the study site and dates, the publication covering the longest time period was selected.

Statistical analysis

Placenta praevia exists in different grades, according to the precise location of the placenta and the degree of overlap with the cervical os Oppenheimer and Farine 2009. Due to inconsistencies in the severity of placenta praevia across the different studies, the results are presented in two groups: (1) studies representing all cases of placenta praevia (marginal, partial or complete coverage of the os) and (2) studies representing major placenta praevia (partial and complete coverage of the os) only.

Meta-analysis was conducted using the DerSimonian–Laird random-effects model (Viechtbauer 2010). Proportions were transformed prior to meta-analysis using the Freeman–Tukey double arcsine transformation for variance stabilisation (Freeman & Tukey 1950; Rucker et al. 2009). Pooled proportions were subsequently back-transformed to the original scale (Miller 1978).

Meta-regression was used to investigate potential sources of heterogeneity. Geographic region, World Bank income, study setting, time at which diagnosis was confirmed and denominator definition were hypothesised a priori to potentially contribute towards between-study heterogeneity. Analyses were carried out using R 2.12.2 and Stata 12.0.

Results

Our initial search identified 13 205 potentially relevant studies (Figure 1). A total of 58 articles met the review inclusion criteria; 10 of these were duplicate publications reporting on the same location and time period, leaving 48 unique studies included in our review (41 of which report on all cases of placenta praevia and 22 of which report the number of major praevia cases). Studies are described in Appendix S1.

Figure 1.

Search strategy flow diagram.

Studies were identified from Asia (n = 9), Australasia (n = 3), Europe (n = 15), Latin America (n = 3), North Africa/Middle East (n = 6), North America (n = 10) and Sub-Saharan Africa (n = 2). Most studies originated from high-income settings (n = 36).

Overall, the pooled prevalence of all cases of placenta praevia was 5.2 cases per 1000 pregnancies (95% CI: 4.5–5.9) (Table 1). Prevalence was highest among Asian studies (12.2 per 1000) and lower among studies from Europe (3.6 per 1000), North America (2.9 per 1000) and Sub-Saharan Africa (2.7 per 1000). The pooled prevalence of major placenta praevia was 4.3 cases per 1000 pregnancies (95% CI: 3.3–5.4).

Table 1. Random-effects meta-analysis of pooled prevalence of placenta praevia, stratified by covariates
CharacteristicNo. of Studies kPooled prevalence per 1000 pregnanciesVariation due to Study Heterogeneity (I2)
n Prevalence(95% CI)
All cases of placenta praevia (marginal, partial or complete)
GBD RegionAsia (Hendricks et al. 1999; Yamada et al. 2005; Fujii et al. 2010; Huang et al. 2011; Jang et al. 2011; Matsuda et al. 2011)6378 85612.2(9.5, 15.2)97.7%
Australasia (Olive et al. 2006; Lain et al. 2008; McCormack et al. 2008)350 5829.5(4.8, 15.7)95.0%
Europe (Taipale et al. 1998; D'Souza 2000; Vettraino et al. 2001; Grgic et al. 2004; Love et al. 2004; Romundstad et al. 2006; Tuzovic 2006; Papinniemi et al. 2007; Tata et al. 2007; Milosevic et al. 2009; Vazquez Rodriguez et al. 2010; Daskalakis et al. 2011; Rosenberg et al. 2011)131 461 9283.6(2.8, 4.6)98.1%
Latin America & the Caribbean Cabrera (Hernández et al. 1999; Faneite et al. 2001; Rivas et al. 2001)340 0585.1(2.5, 8.7)94.4%
North Africa/Middle East (Ismail 2001; Bhat et al. 2004; Bahar et al. 2009; Celik Acioglu et al. 2010; Davood et al. 2010; Alshami et al. 2011)699 5446.4(5.6, 7.3)57.8%
North America (Ananth et al. 2001; Francois et al. 2003; Koroukian 2004; Shen et al. 2005; Predanic et al. 2007; Yang et al. 2009; Aliyu et al. 2011a,b; Eichelberger et al. 2011)819 688 4262.9(2.3, 3.5)99.5%
 Sub-Saharan Africa (Prual et al. 2000; Buambo-Bamanga et al. 2004)240 5602.7(0.3, 11.0)99.2%
World Bank ClassificationHigh income (Taipale et al. 1998; Hendricks et al. 1999; D'Souza 2000; Ananth et al. 2001; Vettraino et al. 2001; Francois et al. 2003; Bhat et al. 2004; Koroukian 2004; Love et al. 2004; Shen et al. 2005; Yamada et al. 2005; Olive et al. 2006; Tuzovic 2006; Romundstad et al. 2006; Papinniemi et al. 2007; Predanic et al. 2007; Tata et al. 2007; McCormack et al. 2008; Bahar et al. 2009; Yang et al. 2009; Bragg et al. 2010; Fujii et al. 2010; Vazquez Rodriguez et al. 2010; Aliyu et al. 2011a,b; Alshami et al. 2011; Daskalakis et al. 2011; Eichelberger et al. 2011; Jang et al. 2011; Matsuda et al. 2011; Rosenberg et al. 2011)3021 559 5725.2(4.4, 6.0)99.6%
Upper-middle income (Cabrera Hernández et al. 1999; Faneite et al. 2001; Ismail 2001; Rivas et al. 2001; Grgic et al. 2004; Milosevic et al. 2009; Celik Acioglu et al. 2010; Davood et al. 2010; Huang et al. 2011)9149 8225.8(3.5, 8.8)97.8%
Low and Lower-middle income (Prual et al. 2000; Buambo-Bamanga et al. 2004)240 5602.7(0.3, 11.0)99.2%
Language of PublicationEnglish (Taipale et al. 1998; Hendricks et al. 1999; D'Souza 2000; Ananth et al. 2001; Ismail 2001; Francois et al. 2003; Bhat et al. 2004; Koroukian 2004; Love et al. 2004; Shen et al. 2005; Yamada et al. 2005; Olive et al. 2006; Romundstad et al. 2006; Tuzovic 2006; Tata et al. 2007; Papinniemi et al. 2007; Predanic et al. 2007; McCormack et al. 2008; Bahar et al. 2009; Yang et al. 2009; Bragg et al. 2010; Davood et al. 2010; Fujii et al. 2010; Aliyu et al. 2011a,b; Alshami et al. 2011; Daskalakis et al. 2011; Eichelberger et al. 2011; Huang et al. 2011; Jang et al. 2011; Matsuda et al. 2011; Rosenberg et al. 2011)3121 590 3225.5(4.7, 6.3)99.6%
 Non-English (Cabrera Hernández et al. 1999; Prual et al. 2000; Faneite et al. 2001; Rivas et al. 2001; Vettraino et al. 2001; Buambo-Bamanga et al. 2004; Grgic et al. 2004; Milosevic et al. 2009; Celik Acioglu et al. 2010; Vazquez Rodriguez et al. 2010)10159 6224.2(2.7, 6.2)96.5%
Study SettingFacility-based (Taipale et al. 1998; Cabrera Hernández et al. 1999; Hendricks et al. 1999; D'Souza 2000; Faneite et al. 2001; Ismail 2001; Rivas et al. 2001; Vettraino et al. 2001; Francois et al. 2003; Bhat et al. 2004; Buambo-Bamanga et al. 2004; Grgic et al. 2004; Love et al. 2004; Yamada et al. 2005; Olive et al. 2006; Tuzovic 2006; Papinniemi et al. 2007; Predanic et al. 2007; McCormack et al. 2008; Bahar et al. 2009; Milosevic et al. 2009; Celik Acioglu et al. 2010; Davood et al. 2010; Vazquez Rodriguez et al. 2010; Alshami et al. 2011; Daskalakis et al. 2011; Eichelberger et al. 2011; Jang et al. 2011)28547 8935.2(4.3, 6.2)95.9%
Population-based (Prual et al. 2000; Ananth et al. 2001; Koroukian 2004; Shen et al. 2005; Romundstad et al. 2006; Tata et al. 2007; Yang et al. 2009; Bragg et al. 2010; Fujii et al. 2010; Aliyu et al. 2011a,b; Huang et al. 2011; Matsuda et al. 2011; Rosenberg et al. 2011)1321 202 0605.0(3.9, 6.3)99.8%
Confirmation of Placenta Praevia DiagnosisConfirmed at delivery (Taipale et al. 1998; Cabrera Hernández et al. 1999; Hendricks et al. 1999; D'Souza 2000; Ananth et al. 2001; Rivas et al. 2001; Francois et al. 2003; Buambo-Bamanga et al. 2004; Olive et al. 2006; Romundstad et al. 2006; Papinniemi et al. 2007; Bahar et al. 2009)131 617 9404.4(3.2, 5.7)98.9%
Present at last ultrasound prior to delivery (Faneite et al. 2001; Ismail 2001; Bhat et al. 2004; Love et al. 2004; Yamada et al. 2005; Tuzovic 2006; Predanic et al. 2007; Daskalakis et al. 2011; Eichelberger et al. 2011; Jang et al. 2011; Matsuda et al. 2011; Rosenberg et al. 2011)11626 8846.1(3.7, 9.1)99.4%
 Not reported (Prual et al. 2000; Vettraino et al. 2001; Grgic et al. 2004; Koroukian 2004; Shen et al. 2005; Tata et al. 2007; McCormack et al. 2008; Milosevic et al. 2009; Yang et al. 2009; Bragg et al. 2010; Celik Acioglu et al. 2010; Davood et al. 2010; Fujii et al. 2010; Vazquez Rodriguez et al. 2010; Aliyu et al. 2011a,b; Alshami et al. 2011; Huang et al. 2011)1719 505 1305.2(4.3, 6.2)99.5%
Definition of denominatorAll deliveries (Cabrera Hernández et al. 1999; Prual et al. 2000; Faneite et al. 2001; Ismail 2001; Rivas et al. 2001; Vettraino et al. 2001; Francois et al. 2003; Bhat et al. 2004; Buambo-Bamanga et al. 2004; Grgic et al. 2004; Love et al. 2004; Shen et al. 2005; Yamada et al. 2005; Olive et al. 2006; Tuzovic 2006; Tata et al. 2007; Milosevic et al. 2009; Bragg et al. 2010; Celik Acioglu et al. 2010; Davood et al. 2010; Rosenberg et al. 2011)211 749 6484.6(3.6, 5.7)98.5%
All Live Births (Huang et al. 2011)121 23417.5(15.8, 19.4)
All singleton deliveries (Taipale et al. 1998; Hendricks et al. 1999; D'Souza 2000; Romundstad et al. 2006; Papinniemi et al. 2007; Predanic et al. 2007; McCormack et al. 2008; Bahar et al. 2009; Fujii et al. 2010; Vazquez Rodriguez et al. 2010; Aliyu et al. 2011a,b; Alshami et al. 2011; Daskalakis et al. 2011; Eichelberger et al. 2011; Jang et al. 2011; Matsuda et al. 2011)162 607 9805.8(4.0, 8.0)99.7%
All Singleton Live Births (Ananth et al. 2001; Koroukian 2004; Yang et al. 2009)317 371 0923.5(2.4, 4.8)99.6%
Overall4121 749 9545.2(4.5, 5.9)99.5% 
Major placenta praevia (partial or complete)
GBD RegionAsia (Hendricks et al. 1999; Hung et al. 2007; Sumigama et al. 2007; Hasegawa et al. 2009; Jang et al. 2011)5164 5598.0(6.0, 10.3)95.6%
Australasia0
Europe (Taipale et al. 1998; D'Souza 2000; Becker et al. 2001; Grgic et al. 2004; Love et al. 2004; Guariglia et al. 2006; Daskalakis et al. 2011)7127 4853.2(1.9, 4.8)94.9%
Latin America & the Caribbean Cabrera (Hernández et al. 1999; Faneite et al. 2001; Rivas et al. 2001)340 0583.3(1.3, 6.1)93.8%
North Africa/Middle East (Bhat et al. 2004; Bahar et al. 2009)251 4944.1(3.6, 4.7)0.0%
North America (Crane et al. 2000; Cleary-Goldman et al. 2005; Predanic et al. 2007; Eichelberger et al. 2011)4166 4583.0(1.5, 5.0)97.1%
 Sub-Saharan Africa (Buambo-Bamanga et al. 2004)120 2345.3(4.4, 6.4)
World bank classificationHigh income (Taipale et al. 1998; Hendricks et al. 1999; Crane et al. 2000; D'Souza 2000; Becker et al. 2001; Bhat et al. 2004; Love et al. 2004; Cleary-Goldman et al. 2005; Guariglia et al. 2006; Predanic et al. 2007; Sumigama et al. 2007; Bahar et al. 2009; Hasegawa et al. 2009; Daskalakis et al. 2011; Eichelberger et al. 2011; Jang et al. 2011)16463 2844.3(3.3, 5.4)96.3%
Upper-middle income (Cabrera Hernández et al. 1999; Faneite et al. 2001; Rivas et al. 2001; Grgic et al. 2004; Hung et al. 2007)586 7704.1(1.1, 8.7)98.6%
 Low and Lower-middle income (Buambo-Bamanga et al. 2004)120 2345.3(4.4, 6.4)
Language of PublicationEnglish (Taipale et al. 1998; Hendricks et al. 1999; Crane et al. 2000; D'Souza 2000; Becker et al. 2001; Bhat et al. 2004; Love et al. 2004; Cleary-Goldman et al. 2005; Hung et al. 2007; Predanic et al. 2007; Sumigama et al. 2007; Bahar et al. 2009; Hasegawa et al. 2009; Daskalakis et al. 2011; Eichelberger et al. 2011; Jang et al. 2011)16473 5604.5(3.3, 6.0)97.7%
Non-English (Cabrera Hernández et al. 1999; Faneite et al. 2001; Rivas et al. 2001; Buambo-Bamanga et al. 2004; Grgic et al. 2004; Guariglia et al. 2006)696 7283.6(2.2, 5.4)94.2%
Study SettingFacility-based (Taipale et al. 1998; Cabrera Hernández et al. 1999; Hendricks et al. 1999; D'Souza 2000; Becker et al. 2001; Faneite et al. 2001; Rivas et al. 2001; Bhat et al. 2004; Buambo-Bamanga et al. 2004; Grgic et al. 2004; Love et al. 2004; Cleary-Goldman et al. 2005; Guariglia et al. 2006; Hung et al. 2007; Predanic et al. 2007; Sumigama et al. 2007; Bahar et al. 2009; Hasegawa et al. 2009; Daskalakis et al. 2011; Eichelberger et al. 2011; Jang et al. 2011)21476 2924.3(3.3, 5.5)97.1%
 Population-based (Crane et al. 2000)193 9963.2(2.9, 3.6)
Confirmation of Placenta Praevia DiagnosisConfirmed at delivery (Taipale et al. 1998; Cabrera Hernández et al. 1999; Hendricks et al. 1999; Crane et al. 2000; D'Souza 2000; Becker et al. 2001; Rivas et al. 2001; Buambo-Bamanga et al. 2004; Cleary-Goldman et al. 2005; Guariglia et al. 2006; Bahar et al. 2009)13366 6884.5(3.2, 6.0)97.4%
Present at last ultrasound prior to delivery (Faneite et al. 2001; Bhat et al. 2004; Love et al. 2004; Predanic et al. 2007; Sumigama et al. 2007; Hasegawa et al. 2009; Daskalakis et al. 2011; Eichelberger et al. 2011; Jang et al. 2011)8194 5904.4(2.6, 6.4)97.2%
 Not reported (Grgic et al. 2004; Hung et al. 2007)19 0101.3(0.7, 2.2)
Definition of denominatorAll deliveries (Cabrera Hernández et al. 1999; Crane et al. 2000; Becker et al. 2001; Faneite et al. 2001; Rivas et al. 2001; Bhat et al. 2004; Buambo-Bamanga et al. 2004; Grgic et al. 2004; Love et al. 2004; Guariglia et al. 2006; Sumigama et al. 2007; Hasegawa et al. 2009)12322 6014.1(3.0, 5.3)95.8%
All live births0
All singleton deliveries (Taipale et al. 1998; Hendricks et al. 1999; D'Souza 2000; Cleary-Goldman et al. 2005; Hung et al. 2007; Predanic et al. 2007; Bahar et al. 2009; Daskalakis et al. 2011; Eichelberger et al. 2011; Jang et al. 2011)10247 6874.6(2.8, 6.7)98.1%
 All Singleton Live Births0
Overall22570 2884.3(3.3, 5.4)97.2% 

A very high level (>97%) of heterogeneity was observed for all and major cases, which remained even after stratification by study characteristic (Table 1). In the meta-regression (Table 2), the only study characteristic with a significant effect on prevalence of all cases of placenta praevia was geographic region (P = 0.0001). Although region could explain around half of the between-study variance, there was still a very high level of residual heterogeneity after accounting for region (98.5%).

Table 2. Meta-regression
Study characteristicOR[95% confidence interval]Test for covariate with Knapp-Hartung modificationProportion of between-study variance explained (Adjusted R2)Residual variation due to study heterogeneity (Residual I2)
All cases of placenta praevia (marginal, partial or complete)
GBD RegionAsia1.00 F = 6.96; P = 0.000150.4%98.5%
Australasia0.78[0.36–1.68]
Europe0.29[0.17–0.49]
Latin America & the Caribbean0.42[0.19–0.90]
North Africa/Middle East0.55[0.30–1.03]
North America0.24[0.13–0.42]
Sub-Saharan Africa0.19[0.08–0.46]
World Bank ClassificationHigh income1.00 F = 1.32; P = 0.2786  
Upper-middle income1.13[0.64–1.99]
Low and Lower-middle income0.44[0.14–1.33]
Language of PublicationEnglish1.00 F = 0.89; P = 0.3522  
Non-English0.77[0.45–1.34]
Study SettingFacility-based1.00 F = 0.49; P = 0.4889  
Population-based0.84[0.51–1.39]
Confirmation of Placenta Praevia DiagnosisConfirmed at delivery1.00 F = 0.56; P = 0.5754  
Present at last ultrasound prior to delivery1.38[0.74–2.56]
Not reported1.11[0.63–1.93]
Definition of DenominatorAll deliveries1.00 F = 1.70; P = 0.1846  
All live births4.22[0.96–18.49]
All singleton deliveries1.28[0.78–2.08]
All singleton live births0.82[0.34–2.00]
Major placenta praevia (partial or complete)
GBD RegionAsia1.00 F = 2.64; P = 0.053330.9%95.3%
Europe0.39[0.21–0.75]
Latin America & the Caribbean0.40[0.18–0.90]
North Africa/Middle East0.53[0.21–1.31]
North America0.36[0.17–0.75]
Sub-Saharan Africa0.67[0.21–2.18]
World Bank ClassificationHigh income1.00 F = 0.16; P = 0.8548  
Upper-middle income0.89[0.44–1.81]
Low and Lower-middle income1.31[0.33–5.19]
Language of PublicationEnglish1.00 F = 0.48; P = 0.4959  
Non-English0.81[0.43–1.53]
Confirmation of Placenta Praevia DiagnosisConfirmed at delivery1.00 F = 1.43; P = 0.2630  
Present at last ultrasound prior to delivery0.96[0.54–1.69]
Not reported0.32[0.08–1.31]
Definition of DenominatorAll deliveries1.00 F = 0.10; P = 0.7557  
All singleton deliveries1.09[0.62–1.92]

Discussion

Our results show that the prevalence of placenta praevia was around 5.2 cases per 1000 pregnancies (95% CI: 4.5–5.9); the prevalence of major placenta praevia was 4.3 cases per 1000 pregnancies (95% CI: 3.3–5.4). Heterogeneity was very high, even after stratification on study characteristics.

Geographic region was the only study characteristic with a significant influence on prevalence. Prevalence was highest among Asian studies (12.2 per 1000) and lower among studies from Europe (3.6 per 1000), North America (2.9 per 1000) and Sub-Saharan Africa (2.7 per 1000). This could plausibly be due to true geographic or ethnic differences between populations; several previous studies have found the prevalence of placenta praevia to vary between different ethnic groups (Sheiner et al. 2001; Ananth et al. 2003), including increased prevalence among Asian women relative to White women (Shen et al. 2005). Any biological mechanism for this association is poorly understood; however, ethnic differences in prematurity rates or pelvic structure are possible explanations. It has been observed that normal gestational length is shorter among Black and Asian women than among White European women (Patel et al. 2004). However, it is also possible that the apparent importance of the geographic region variable is a marker for other (unknown) causes of heterogeneity between the studies. The majority of studies identified in this review did not report the distance between the placental edge and the cervical os; subtle differences in national diagnostic protocols or conventions may account in part for the observed regional differences.

This study has investigated the prevalence of placenta praevia in more diverse geographic locations than previously. However, there are a number of limitations. Despite our best efforts, we retrieved a relatively small number of studies from low- and middle-income countries. This partially reflects an imbalance in where studies on placenta praevia are conducted; however, this was also partially due to our a priori decision to only include facility-based studies where skilled birth attendance was >95%. We acknowledge that doing so has meant that we have very few studies from some regions, such as Sub-Saharan Africa, and that consequently our estimates cannot be interpreted as representative of the entire region. Nevertheless, we believe that our strategy was preferable to any alternative to enhance the validity of our estimates.

The true prevalence of placenta praevia is likely to lie somewhere between that estimated by the facility-based and population-based designs. Facility-based studies may overestimate prevalence because of referral patterns. Conversely, estimates from population-based studies may underestimate the prevalence of placenta praevia due to the increased potential for missing cases. Population-based studies are more likely to lack detailed information on the grade of the placenta praevia, and clinical diagnosis procedures and many population-based studies restrict the denominator to live births, which excludes cases of placenta praevia accompanied by stillbirth.

Placenta praevia is a rare condition and may have been subject to differential under-reporting depending on local antenatal routines, which are difficult to ascertain at the aggregate level. For example, one Sub-Saharan study Prual et al. 2000 reported a substantially lower prevalence (0.59 cases per 1000 pregnancies) than other studies in this review. This study predominantly identified cases of placenta praevia that were accompanied by a maternal haemorrhage, and likely missed those with less serious outcomes, which would be identified in well-resourced settings where women may have multiple ultrasounds during pregnancy.

We observed very high levels of heterogeneity in this review. Most of the studies included in this review had large sample sizes that produced very precise estimates. Because within-study variance was very small compared to between-study variance, high I2 values were observed. Statistically significant differences between study estimates were not necessarily clinically significant.

In conclusion, the prevalence of placenta praevia is low, but remains a serious obstetric complication with a high case-fatality rate. There is some evidence to suggest regional variation in the prevalence of placenta praevia, although it is not possible to determine from studies such as this whether this is due to true population differences or some unknown factor.

Acknowledgements

This project was funded by a grant from the Bill and Melinda Gates Foundation to the US Fund for UNICEF to support the Child Health Epidemiology Reference Group (PI: Robert E Black). The authors acknowledge the valuable contributions of Alma Adler, Sara Thomas and Karen Wagner to the project. The authors would like to thank Fernanda Boueri, Kathryn Church, Xing Lin Feng, Sylvia Marinova, Ana Montoya, Yusuke Shimakawa and Katerini Storeng for their help translating articles.

Ancillary