Background

Obstetric ultrasound is a non-invasive, sensitive, accurate and cost-effective imaging technique for the detection and diagnosis of congenital abnormalities, obstetric pathologies, and complications of pregnancy. The use of obstetric ultrasound is proliferating in resource-constrained settings due to its increased affordability, availability, portability, and durability [1]. Identifying foetal and placental abnormalities during pregnancy is essential to identify potential complications during gestation or delivery [2]. Foetal abnormalities, e.g., congenital malformations can necessitate additional provisions for delivery care, for example by requiring a Caesarean Section [3]. Equally, placental abnormalities such as placenta praevia or placenta accreta may require additional precautions during delivery [2]. Placental abnormalities may affect the development of the foetus in utero, requiring additional antenatal or postnatal care and supervision [4]. As such, accurate information on placental and foetal abnormalities that can be obtained from obstetric ultrasounds is vital for adequate antenatal support, delivery planning and reducing maternal and neonatal mortality. In resource-constrained settings, the WHO recommends obstetric ultrasound screenings once during pregnancy, before 24 weeks of gestation [5, 6]. For resource-constrained settings, the WHO questions the overall medical value of obstetric ultrasound but simultaneously promotes its ability to encourage pregnant women to attend routine antenatal care [6]. Nevertheless, reported clinical practices in low- and middle-income countries (LMICs) differ from this recommendation: Late obstetric ultrasound is routinely used to identify foetal and placental abnormalities (such as foetal malpresentation after 36 weeks of gestation or placenta previa) to reduce adverse events during delivery [7–14]. However, as these reports do not represent quasi-experimental or randomised trials, the effect of performing late instead of early obstetric ultrasound in resource-constrained settings to improve patient management and health outcomes remains elusive [1, 9, 11, 12, 15–19], especially given the challenges of administering obstetric ultrasound in resource-constrained settings, e.g., shortage of qualified staff and guidelines, proper supervision, and adequate maintenance of facilities and equipment [20].

In Madagascar, maternal and neonatal mortality rates remain high, with 353 maternal deaths per 100,000 live births in 2017 and a neonatal mortality of 20.6 per 1,000 live births in 2018 [21, 22]. Only approximately half of pregnant women in Madagascar complete 4 antenatal care visits as recommended by the WHO, while 54% of deliveries take place without qualified personnel [23]. Eighty-three percent of Malagasy women report difficulties accessing healthcare services and women living in rural areas are particularly affected [24]. Common barriers to maternal and neonatal care include geographical distance to a healthcare facility, absence of a reliable and efficient mode of transportation, financial situations, cultural beliefs, a lack of family support, and poor quality of care [25–28]. Moreover, to the best of our knowledge, public databases focusing on the proportion of foetal and placental abnormalities are not available for the Malagasy population, which poses a challenge for policymaking and implementations regarding priority setting and design for programs aimed at improving maternal health in Madagascar.

To complement routine antenatal care in line with current WHO recommendations, the non-governmental organisation (NGO) Doctors for Madagascar has made mobile obstetric ultrasound screenings available to all pregnant women at 62 public-sector healthcare facilities in urban and rural regions of Madagascar, leading to a multi-year initiative to promote maternal healthcare and facilitate a data-driven understanding of foetal and placental abnormalities.

The aim of this study was to describe the demographic characteristics of a large group of pregnant women who utilized obstetric ultrasound services provided through an NGO. Specifically, we aimed to determine the prevalence of foetal and placental abnormalities in this study sample and to assess the timing and frequency of ultrasound examinations among these women. We expect these data to guide policymakers and program implementers in refining the design, implementation, and resource allocation of maternal health initiatives in this setting.

Methods

Study setting

This study was conducted in the Analamanga region (3.6 million inhabitants) in southern Madagascar, including Antananarivo, the capital of the island as well as the Atsimo-Andrefana region (1.8 million inhabitants), Anosy region (833,000 inhabitants), and Androy region (903,000 inhabitants) [29] (Fig 1). In Analamanga, 65% of women attend at least 4 antenatal care visits and two-thirds (68%) of deliveries take place in a healthcare facility [30]. Atsimo-Andrefana, Anosy, and Androy are among the most remote regions of the island [29]. In these regions, only around half of all women attend at least 4 antenatal care visits (50%, 49%, and 43% for Atsimo-Andrefana, Anosy, and Androy, respectively) and only approximately one-third of deliveries take place in a healthcare facility (30%, 49%, and 26% for Atsimo-Andrefana, Anosy, and Androy, respectively) [30]. Over 90% of the population in the three regions live below the international poverty line of 1.90 USD per day [28].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Map of study facilities.

Map of Madagascar, depicting sixty-two public-sector primary care facilities where obstetric ultrasound screenings were conducted between July 1, 2017, and September 30, 2020, in urban (Analamanga) and rural (Atsimo-Andrefana, Anosy, and Androy) settings. The base map was provided by OpenStreetMap. OpenStreetMap is open data and is made available under the CC BY-SA license (OpenStreetMap contributors, http://www.openstreetmap.org/copyright) [31].

https://doi.org/10.1371/journal.pone.0311918.g001

Study cohort

Ultrasound reports from all women who had at least one obstetric ultrasound examination at any of the sixty-two public-sector primary care facilities during the study period were included. A total of 38,688 ultrasound exam reports (data entries) were collected, from which we identified 33,332 individuals who had one single ultrasound examination and 2,587 individuals who had more than one ultrasound examination. Accordingly, an individual may have had multiple data entries depending on the number of ultrasound examinations she had. Therefore, the 38,688 data entries belonged to 35,919 women. Of these 38,688 reports, 492 lacked data on the time of the first ultrasound, and 827 lacked data on placenta location. Of these 35,919 women, 762 had missing information on the number of foetuses. Of all foetuses examined, information on amniotic fluid volume was missing in 1,012 cases. Fig 2 below illustrates the data source and study sample in more detail.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Study cohort.

Depiction of study sample of women included in the study between July 1, 2017, and September 30, 2020, in urban (Analamanga) and rural (Atsimo-Andrefana, Anosy, and Androy) Madagascar.

https://doi.org/10.1371/journal.pone.0311918.g002

Results

Sample description

Our analytical sample included 35,919 women with 38,688 ultrasound reports. Data on age was available from 84.04% (30,186/35,919) of women; the median age was 22 years (IQR: 18–28) (Fig 3). We found age heaping at ages that are multiples of ten, which is common in low-resource settings. Most women (84.96%; 30,517/35,919) lived in rural areas; 15.04% (5,402/35,919) of women lived in urban areas.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Age distribution of study participants.

Age distribution of 35,919 pregnant women who received an obstetric ultrasound screening exam between July 1, 2017, and September 30, 2020.

https://doi.org/10.1371/journal.pone.0311918.g003

Distance between the healthcare facility and the village of residence

The median distance between the ultrasound-providing healthcare facility and a woman’s village of residence was 6 km (range: 0.7–14 km). For rural regions the median distance was 8 km (range: 1–16 km); for urban regions, the median distance was 1.9 km (range: 0.5–3.2 km, Fig 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Distance between the nearest healthcare facility and a woman’s village of residence.

Average distance from a woman’s village of residence to the nearest public-sector primary care facility where obstetric ultrasound screenings were performed for a) urban, and b) rural regions of Madagascar. size of bubbles = number of women that visited a healthcare facility; colour intensity of bubbles = average distance from each woman’s village of residence to the nearest public-sector primary care facility where obstetric ultrasound screenings were performed. The base map was provided by OpenStreetMap. OpenStreetMap is open data and is made available under the CC BY-SA license (OpenStreetMap contributors, http://www.openstreetmap.org/copyright) [31].

https://doi.org/10.1371/journal.pone.0311918.g004

Timing of ultrasound exams during pregnancy

Among all women, 7.20% (2,587/35,919) had more than one ultrasound examination. Of those, 6.78% (2,434/35,919) received two ultrasound examinations, and only a few received three or four ultrasound exams during their pregnancy (0.41% (149/35,919) and 0.01%, (4/35,919), respectively). Most visits for obstetric ultrasound to the mobile clinic occurred during the third trimester of pregnancy (68.63%; 26,550/38,688), while 25.57% (9892/38,688) and 4.53% (1754/38,688) of obstetric ultrasound examinations occurred during the second and first trimester of pregnancy, respectively. Most scans were performed between 26 and 37 weeks of gestation (Table 1). The proportion that received a scan in the third trimester was significantly higher among women from rural backgrounds (71.33%, 23,352/32,740) than from urban backgrounds (53.77%, 3,198/5,948; p-value: <0.01).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographics and maternal characteristics.

https://doi.org/10.1371/journal.pone.0311918.t001

For the 2,587 women who had more than one ultrasound examination, we also analysed when the follow-up visits occurred. Among these 2,587 women, inconsistencies between date of facility visit and trimester (e.g., earlier date is associated with a later trimester) or missing information were observed in data from 19 women, leading to exclusion from this analysis. Of the 2,568 women included, 2,418 had two ultrasound examinations (1 revisit; 94.16%), 147 had three ultrasound examinations (2 revisits; 5.72%), and 3 had four ultrasound examinations (3 revisits; 0.12%).

Among the 2,418 women who had two ultrasound examinations, a total of 256, 1433, and 729 first examinations occurred in the first, second, and third trimesters, respectively, and the follow-up occurred in the first trimester in 6 women, in the second trimester in 150 women, and in the third trimester in 2262 women. Of the 147 women who had three ultrasound examinations, 35 had their first examination in the first trimester, 95 had their first examination in the second trimester, and 17 had their first examination in the third trimester. The first follow-up was in the second trimester in 48 women and in the third trimester in 99 women, respectively. A total of 2 women had the second follow-up in the second trimester, and 145 women in the third trimester. Of the 3 women who had four ultrasound examinations, 2 had their first examination in the first trimester and 1 in the second trimester. The first follow-up occurred was in the second trimester for one woman, and in the third trimester for 2 women. The second and third follow-ups were all in the third trimester.

Proportion of foetal and placental abnormalities

Data on foetal presentation between 24 and 43 weeks of gestation was available for 99.62% of all (29,674/29,788) women and missing for 0.38% (114/29,788). From ≥ 36 weeks of gestation, 94.52% (3,035/3,211) of all foetuses were in a cephalic position. Of the remaining foetuses from ≥ 36 weeks of gestation, 3.99% (128/3,211) were in the breech, 0.84% (27/3,211) were in transverse, 0.5% (16/3,211) were in mobile and 0.16% (5/3,211) were in an oblique position. The prevalence of breech presentation decreased with gestational age. Around one in six foetuses (15.67%, 824/5,258), 9.5% (972/10,236), 4.39% (481/10,969), and 3.99% (128/3,211) were in breech position between 24 to 27, 28 to 31, 32 to 35, and from 36 weeks of gestation, respectively.

Placenta previa was observed in 2.31% (875/38,755) cases. Of those, 76,11% (666/875), 7.66% (67/875) and 0.69% (6/875) were placenta previa grade I, II, and III, respectively; 15.54% (136/875) diagnosed with placenta previa where not further categorised (Table 2). We found that 96.85% (34,787/35,919) of women had one foetus and 1.03% (370/35,919) had multiple gestations. Overall, 1.02% (366/35,919) had a twin pregnancy and 0.01% (4/35,919) had three foetuses. Amniotic fluid disorders were diagnosed in 0.66% (234/38,702) of cases. Low volume of amniotic fluid (oligohydramnios) was diagnosed in 12.82% (30/234) of cases and high volume of amniotic fluid (polyhydramnios) was seen in 87.18% (204/234) (Table 2). In terms of foetal pathologies, non-gravid uteruses (0.33%; 118/35,919), foetal death in utero (0.19%; 68/35,919), and foetal malformation (0.06%; 22/35,919) including acrania or hydrocephalus were the most common pathologies (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Foetal and placental abnormalities.

https://doi.org/10.1371/journal.pone.0311918.t002

Additional analyses showed no significant correlation between the number of foetuses, amniotic fluid disorders, or placental and foetal pathologies and the number of ultrasound visits. Pearson’s chi-square test revealed a significant correlation between the grade of placenta praevia and the number of revisits (p <0.01). We have uploaded the tables depicting the results of these analyses in S1 File.

Discussion

Using data collected from over 35,000 women, we estimated the proportion of foetal and placental abnormalities among pregnant women attending obstetric ultrasounds at public-sector healthcare facilities in Madagascar. Our study revealed three main findings: First, most women in our sample came from rural areas. Secondly, less than 10% of women in our sample were diagnosed with a potential complication during pregnancy. Foetal malposition, breech position and abnormal position of the placenta were the most common findings. Lastly, most obstetric ultrasounds were conducted in the third trimester of pregnancy, especially among women from rural backgrounds.

About one-fourth of the Malagasy population live more than 10km from a healthcare facility [47]. This holds especially true for women in rural areas where most of the Malagasy population and individuals in this study reside [47]. We found that women from rural areas had to travel longer distances (median 8km) than women from urban areas (median 1.9km) to reach a healthcare facility providing obstetric ultrasound. No other study has so far investigated the distance to the nearest available obstetric ultrasound in Madagascar; a systematic review that assessed the access to skilled care for childbirth in Sub-Saharan Africa found the distance to the nearest healthcare facility to be 4km and 15km in urban and rural settings, respectively [48]. Given that only around two-thirds of Madagascar´s inhabitants have access to any form of primary health care and that the longer the distance to a healthcare facility, the lower the access to maternal care [49, 50], women who live in villages further from primary healthcare facilities, especially those from rural backgrounds in our sample had to travel to access obstetric ultrasounds are in a challenging situation.

Overall, the proportion of foetal malpresentation and multiple gestations detected by obstetric ultrasound in our study are consistent with findings from other countries in sub-Saharan Africa [35, 51, 52], with a few important exceptions. First, in our sample, 5.48% (176/3,211) of all foetuses were in malpresentation from >36 weeks of gestation. These findings differ from a study from Uganda where foetal malpresentation in the third trimester was found in 15% of pregnancies, but are consistent with a study from Guatemala where the overall rate of foetal malpresentation was 4.49% and the prevalence of breech presentation was 3–4% of all term pregnancies [7, 35, 51]. This might be a consequence of different sampling strategies of the studies, given that the sample in Uganda only included women accessing a rural, private clinic, while our study also included public facilities. As, overall, care at public facilities is cheaper and more accessible in Sub-Saharan Africa and socioeconomic status is a key determining factor in accessing healthcare [27] it might be that women who fear complications are more likely to access healthcare at a private facility.

Secondly, while the overall prevalence of placenta previa in Sub-Saharan Africa is unclear and in a reported range between 0.27% and 1.7% [10, 17, 53], these results do not match the findings in our study where the prevalence of placenta previa was 2.31% (875/38,755). This discrepancy might be explained by the gestational age at the time of ultrasound. Around one-fourth of ultrasounds in our study were performed during the second trimester when placental abnormalities were more common.

Thirdly, accurate data on the prevalence of congenital foetal anomalies is rare. Data from registries of congenital anomalies indicate that they are seen in 2–3% of newborns in Africa [54]. However, only 0.06% (22/35,919) of foetuses had a malformation detected during obstetric ultrasound, which is consistent with findings from other studies on the prevalence of foetal abnormalities during routine third-trimester ultrasound [55]. This might be an indication of the limited diagnostic capacities of obstetric ultrasounds in identifying neonatal abnormalities, especially given their dependence on the training and experience of the sonographer [7, 56, 57].

Lastly, the incidence of extrauterine pregnancy worldwide ranges from 1.02% to 3.9%, which is not in line with our findings of 0.03% (11/35,919) [58]. As extrauterine pregnancy is the main cause of first-trimester maternal mortality, the discrepancy in results is most likely due to the low number of ultrasound examinations during the first trimester (4.53% 1754/38,688) in our study [59].

A central finding of our study is that women in rural and urban Madagascar accessed obstetric ultrasound predominantly during late pregnancy. Two-thirds (68.63%; 26,550/38,688) of women had their first ultrasound examination during the third trimester of pregnancy and a quarter of women (25.57%; 9,892/38,688) during the second trimester. Rural women were significantly more likely to have their first ultrasound during their third trimester compared to women from urban backgrounds. These results are consistent with findings from other resource-restricted settings that show a clear trend towards accessing antenatal care late in pregnancy, despite the WHO recommendation of conducting obstetric ultrasounds before 24 weeks gestation [12, 13, 52, 60]. Evidence from other Sub-Saharan African countries suggests that women accessing obstetric ultrasounds late in pregnancy is a common phenomenon.

Some studies suggest that screening for breech presentation at 36–37 weeks of gestation in resource-restricted settings might reduce the risk of perinatal complications and maternal mortality [54, 55, 61]. Early ultrasounds, on the other hand, are crucial for understanding if a pregnancy is intrauterine, and for identifying essential indicators for reducing early maternal mortality, such as the number of foetuses and their respective vitality [51]. Our findings highlight the need for a better knowledge of the reasons women, especially those from rural regions, only access ultrasound examinations late in pregnancy, despite the WHO recommendation and advantages of early ultrasound. It further calls into question if additional late ultrasounds might be an important diagnostic tool in reducing perinatal mortality, even though the current evidence on this subject is limited [51].

Additionally, our study shows that there was no significant correlation between the number of ultrasounds visits and foetal and placental abnormalities that had been identified during obstetric ultrasounds, apart from the grade of placenta praevia. This is worrying as several of the pathologies identified should be followed up through repetitive ultrasound examinations in pregnancy. Our findings indicate that women in Madagascar with foetal and placental abnormalities do not access repetitive ultrasound examinations in such cases. Further indicate should be conducted to identify the reasons behind these healthcare seeking behaviours.

Our study also has limitations. First, ultrasound examinations were performed by different sonographers, which may have introduced inter-observer bias. However, all sonographers underwent a standardised training program and obtained a diploma in obstetric ultrasound from the National Institute of Public and Community Health to improve inter-observer reliability and reduce bias. Secondly, we based our analyses on primary data that was not collected for research purposes and contained data entry errors and missing data due to limited resources and under-standardised data acquisition guidelines. To enable optimised preprocessing and cleaning, we designed and applied several methods, including crosschecking, screening for duplicate entries, out-of-range values, data entry errors, and overall consistency. In addition, as no unified personal identifier system was employed at the time of this study, it was difficult to detect multiple visits to the healthcare facilities by the same individual. Accordingly, we used several criteria to screen for multiple visits and conducted inspections of the whole dataset through random sampling of data entries, for improved validity and consistency. In addition, some, but not all women with twin pregnancies had two records/data entries per visit, one entry for each foetus. Meanwhile, in 20 women with twin pregnancies, the number of foetuses in the records was inconsistent. Nevertheless, it was not possible to correct these errors due to lack of reliable information given the time lag between data collection and analysis. Fourthly, as we drew on secondary data from an NGO program, our sample may not be representative of all women seeking antenatal care and results may differ for other populations and settings. Nevertheless, our dataset contains data from a variety of cultural, demographic, and geographic contexts within Madagascar, with substantial variation regarding their geography, health indicators, and socio-economic status. We compared the basic demographic characteristics of our sample with the broader population of women of reproductive age in Madagascar and found them to be comparable. Demographic characteristics of our study population were also similar to those in studies conducted in Nigeria and Uganda [10, 13]. We do acknowledge however, that this may not be an accurate comparison, as studies conducted in similarly resource-constrained settings in SSA may have been affect by similar limitations.

Key strengths of our study included its large sample size of rural and urban pregnant women from a resource-restricted setting. Furthermore, the secondary analysis of NGO program data allowed for the inclusion of marginalised or otherwise hard-to-reach communities with restricted access to health services, who are commonly underrepresented in studies and surveys within the fields of obstetrics and prenatal development and diagnosis. To our knowledge, this study represents the largest dataset with obstetric ultrasound data collected from Madagascar, featuring both urban and rural settings.

Conclusion

Our data show that obstetric ultrasound in a resource-restricted setting can lead to the detection and diagnosis of high-risk obstetric conditions including placental abnormalities, malpresentation of the foetus, and multiple gestations, which are highly associated with maternal and foetal morbidity and mortality. In stark contrast to current WHO recommendations, our findings suggest that universally accessible obstetric ultrasound is commonly and mostly utilised by pregnant women, especially those from rural, resource-constrained settings, during the third trimester of pregnancy. This finding highlights the need to improve policy and practice in Madagascar to improve access to obstetric ultrasounds, particularly during the first and second trimesters.

Supporting information

Acknowledgments

We would like to thank Doctors for Madagascar for initiating and running the ultrasound project, and all study participants in Madagascar for taking part in this project. We would equally like to thank Prof. Dr. Dr. Till Bärninghausen for his support of this study.