How twins differ: multiple pregnancy and the use of health care in the 2008 Nigeria Demographic and Health Survey


Corresponding Author Andrea M. McDonald, Department of Primary Care and Public Health, Imperial College London, London, UK. Tel.: +44 20 75942341; E-mail:


Objectives  To (i) document the incidence of multiplicity in Nigeria, (ii) compare healthcare utilization during pregnancy and at delivery for singleton and multiple pregnancies (iii)and investigate whether antenatal care modifies the relationship between multiplicity and likelihood of having skilled attendance.

Methods  This observational study was a secondary analysis of the 2008 Nigeria Demographic and Health Survey and included 17 635 women who gave birth to a live infant between 2003 and 2008. Multivariate logistic regression with adjustment for weighting, clustering and confounding was used to investigate associations and look for effect modification.

Results  There were 18.5 multiple pregnancies per 1000 maternities. Multiple births had nearly six times the chance of neonatal mortality (AOR 5.74). Mothers with multiple births had more skilled attendance (AOR 1.75), but similar antenatal care utilization (AOR 0.95) as women with a singleton pregnancy. Women with multiple pregnancies attending antenatal care had more visits (mean 9.0 vs. 8.2), blood pressure checks (OR 1.52) and urine tests (OR 1.51). Although antenatal care was strongly associated with skilled attendance, there was no evidence that this was more so for twins than singletons.

Conclusions  Multiplicity in Nigeria is not just a common occurrence, but an indicator of a high-risk pregnancy. The fact that the use of antenatal care by women with a multiple pregnancy is not associated with a disproportionately greater use of skilled delivery care raises questions about the quality of antenatal services. Services should encourage and link all women and especially high-risk women to skilled attendance at delivery.


Objectifs:  (1) Documenter l’incidence de la multiplicité au Nigeria, (2) comparer l’utilisation des soins de santé pendant la grossesse et lors de l’accouchement pour les grossesses uniques et multiples, (3) déterminer si les soins prénataux modifient la relation entre la multiplicité et la probabilité d’avoir une assistance qualifiée.

Méthodes:  Cette étude observationnelle est une analyse secondaire de l’Enquête Démographique et de Santé de 2008 au Nigeria et inclus 17.635 femmes qui ont donné naissance à un enfant vivant entre 2003 et 2008. La régression logistique multivariée avec ajustement pour la pondération, le regroupement et les facteurs confusionnels a été utilisée pour étudier les associations et rechercher la modification de l’effet.

Résultats:  Il y avait 18,5 grossesses multiples pour 1000 maternités. Les naissances multiples avaient près de six fois de chance de mortalité néonatale (AOR: 5,74). Les mères ayant eu des naissances multiples avaient plus de présence qualifiée (AOR: 1,75), mais une utilisation similaires des soins prénataux (AOR: 0,95) que les femmes ayant une grossesse unique. Les femmes avec des grossesses multiples visitant des soins prénataux avaient plus de visites (en moyenne 9,0 contre 8,2), de mesures de la pression sanguine (OR: 1,52) et d’analyses d’urine (OR: 1,51). Bien que les soins prénataux soient fortement associés à la présence de personnel qualifié, il n’y avait aucune preuve que cela était plus appliqué pour les jumeaux que les singletons.

Conclusions:  La multiplicité au Nigeria n’est pas seulement un phénomène courant, mais un indicateur d’une grossesse à haut risque. Le fait que l’utilisation des soins prénataux par les femmes avec une grossesse multiple n’est pas associée à une utilisation proportionnellement plus élevée des soins à assistance qualifiée lors de l’accouchement pose des questions sur la qualité des services de soins prénataux. Les services devraient encourager et relier toutes les femmes et en particulier celles à haut risque, à une assistance qualifiée à l’accouchement.


Objetivos:  (1) Documentar la incidencia de multiplicidad en Nigeria, (2) comparar el uso de cuidados sanitarios durante el embarazo y parto para embarazos únicos y múltiples, (3) investigar si los cuidados sanitarios prenatales modifican la relación entre la multiplicidad y la probabilidad de tener una atención cualificada.

Métodos:  Este estudio observacional surgió de un análisis secundario de la Encuesta Demográfica y de Salud de Nigeria, realizada en el 2008, e incluyó a 17,635 mujeres que dieron a luz a neonatos vivos entre el 2003 y 2008. Se utilizó una regresión logística multivariada con ajuste para el peso estadístico, por conglomerados y el factor de confusión, con el fin de investigar asociaciones y buscar una modificación del efecto.

Resultados:  Hubo 18.5 embarazos múltiples en 1,000 maternidades. Los partos múltiples tenían casi seis veces más posibilidad de terminar en mortalidad neonatal (AOR 5.74). Las madres con partos múltiples tenían una atención más especializada (AOR 1.75), pero un uso similar de cuidados prenatales (AOR 0.95) comparadas con las mujeres que tenían un embarazo único. Las mujeres con embarazos múltiples, recibiendo cuidados prenatales, tuvieron más visitas (media 9.0 versus 8.2), controles de presión sanguínea (OR 1.52) y pruebas de orina (OR 1.51). Aunque los cuidados prenatales estaban fuertemente asociados con una atención especializada, no había evidencia de que lo fuese más para los embarazos múltiples que para los únicos.

Conclusiones:  Los embarazos múltiples en Nigeria no solo eran comunes, sino un indicador de un embarazo de alto riesgo. El hecho de que el uso de cuidados prenatales por parte de mujeres con embarazos múltiples no estuviese asociado con un uso desproporcionadamente mayor de cuidados especializados durante el parto, plantea dudas acerca de la calidad de los servicios prenatales. Los servicios deberían promover y facilitar a todas las mujeres, y en especial a aquellas con embarazos de alto riesgo, a recibir cuidados especializados durante el parto.


Nigeria is reported to have some of the highest rates of multiple births in the world (Nylander 1969; Akinboro et al. 2008), yet few population-based estimates are available. Multiple births are known to experience increased morbidity and mortality for both mother and child. In a US study by Kovacs et al. (1989), the antenatal complication rate was 83% for twins and 32% for singletons. In low- and high-income settings, neonatal mortality is up to six times higher for multiple births (Doyle 1996; Justesen & Kunst 2000). Yet for this very reason, twins in studies are frequently excluded. This study seeks to highlight differences in care received by mothers with multiple pregnancy in Nigeria. Twins and singletons are likely to face similar barriers limiting access to health services. Multiple pregnancy, therefore, provides us with a unique window with which to explore how well the health service responds to women with a high-risk pregnancy, in terms of the care she receives.

Delivery with a skilled attendant is widely recommended as an intervention to reduce maternal and neonatal deaths. Nevertheless, rates of skilled attendant coverage in many parts of Nigeria are low, and even women who receive care may not be getting the quality of services that they require Galadanci et al. (2007). Less than 20% of health facilities provide emergency obstetric services, and only 36% of deliveries are attended by skilled personnel.

The heterogeneous health system in Nigeria is limited by pervasive corruption at all levels of the public system, under-funding and critical health worker shortages. There are concerns that the overall availability, accessibility, quality and utilization of health services have significantly fallen or stagnated in recent times (Uneke et al. 2008). There are marked disparities between urban and rural, north and south (Rohde et al. 2008). The northern part of the country has generally worse indicators (USAID 2008), and many urban and rural hospitals in the north do not have consultant and specialist obstetricians in their employment (Monjok et al. 2010). The Ministry of Health usually spends about 70% of its budget in urban areas where 30% of the population resides. In one rural state in the south-east, Ochie and Ugwu (2009) assessed access to ultrasound amongst a sample of antenatal care patients. Only 30% of women in their previous pregnancy had received ultrasound, although 80% were willing and aware of the service.

Antenatal care can benefit babies in terms of growth, reducing risk of infection and enhancing survival; however, the contribution of antenatal care to maternal health has now shifted to an emphasis on the promotion of health-seeking behaviour, detection and treatment of complications and linking women to other services (Campbell & Graham 2006). Many studies, such as Vanneste et al. (2000), Bloom et al. (1999), and Mpembeni et al. (2007), show that women who attend antenatal care are much more likely to have a skilled attendant at delivery. It is difficult to disentangle this relationship from confounding by access and availability of health services (Gabrysch & Campbell 2009), but antenatal advice to deliver in a facility has been associated with increases in skilled attendant delivery.

We used a recent Demographic and Health Survey (DHS) in Nigeria to document the population-based incidence of multiple births and to compare singleton and multiple pregnancies in terms of neonatal mortality, healthcare utilization in pregnancy and at delivery and any differences in number of antenatal care visits. We also investigated whether attending antenatal care modifies the relationship between multiplicity and the likelihood of having skilled attendance at birth. It remains to be determined whether antenatal care in Nigeria, with the opportunity it provides for twin detection and individualized advice, actually improves the uptake of skilled attendance for a multiple pregnancy more than a singleton pregnancy, given that multiple pregnancies are a high-risk subset of the population.


Data were obtained from a cross-sectional population-based household survey, namely the Nigeria DHS (2008), implemented June–October 2008. The survey used a stratified two-stage cluster design. All women aged 15–49 from each household in 888 clusters were sampled, equating to 33 385 interviewed women and 101 977 births.

We analysed all women with a live birth in the 5 years immediately preceding the survey, because data on healthcare utilization were only available for this subset. A woman’s most recent live birth contributed one case for analysis irrespective of whether the newborn delivery was singleton, twin or triplet. A maternity was defined so that a twin pregnancy ends in one maternity but two births. Multiple birth status was the primary explanatory variable.

Outcome variables included skilled attendant at delivery (primary outcome), antenatal care and quality of antenatal care, caesarean section, place of delivery, postnatal care and neonatal mortality. To be classed as a neonatal death, an infant was required to be a live birth. Neonatal mortality was defined as infant death from 0 days up to but not including death at 31 days or 1 month of age, per 1000 live births (n = 28 100). A skilled attendant delivery included a doctor, nurse/midwife or auxiliary midwife at delivery. Antenatal care was described by use (yes or no), number of visits, timing of visits [none, less than recommended (1–3), recommended number or more (4–9), or more than routinely expected (10+)] and components of visits. Components of care were analysed individually and as a total score counting each component, with missing components considered absent. Place of delivery was either at a health facility or at home, the latter including other locations (0.4% overall). Postnatal care was classified as a check on a women’s health by either a healthcare provider or a traditional birth attendant, before discharge or after delivery at home.

Data were analysed using Stata SE 11.0 and multivariate logistic regression. All frequencies were calculated with adjustment for sample weighting and stratification (survey:tab). Because of the weighting and rounding required to obtain whole numbers, some frequencies do not add up exactly to their totals. To account for clustering of the sampling, population average odds ratios (OR’s) were calculated with generalized estimating equations (GEE), weighting and robust standard errors. Wald tests were used to determine P-values for logistic regression models (testparm). Confidence intervals (CI) were calculated to the 95% level.

The relationship between multiplicity and skilled attendance was adjusted for confounding, where there was an important change in the OR after adding the confounding variable in question to the model. Variables investigated for potential confounding were region, urban or rural residence, socioeconomic status, maternal education, maternal age, birth order, ethnicity, religion, BMI and short stature. Confounders of the association between multiplicity and skilled attendance were used to adjust associations between multiplicity and other healthcare utilization outcomes (Table 2).

Categorical antenatal care outcomes were investigated with chi-squared tests using the ‘survey:set commands. Multivariate logistic regression was used to examine antenatal care as an effect modifier of the association between multiplicity and skilled attendance, with adjustment for the confounders identified.


Multiplicity and mortality

Our first objective was to estimate the twinning rates in Nigeria. In the study period (2003–2008), 17 635 women in the Nigeria DHS had live births. Of these, 325 had multiple births, yielding 17 967 infants, with 17 309 singletons, 320 sets of twins and six sets of triplets. This translates into 18.5 (CI: 16.5–20.6) multiple maternities per 1000 maternities (325/17 635) and 36.6 multiple infants per 1000 live births. The highest rates of multiplicity were in the south-east with 25 (CI: 19–35) per 1000 maternities. The south-west rate (20 per 1000 maternities) was only marginally higher than the northern rates (15–18 per 1000 maternities) and not more so than might be due to chance (Table 1).

Table 1.   Multiple births by region, Nigeria DHS 2003–2008
 Total pregnancies n (%)Maternities resulting in a multiple birthMultiple pregnancies per 1000 maternities (95% CI)OR (95% CI)Wald P-value
  1. DHS, Demographic and Health Survey.

 North-central2525 (14)4618 (14, 24)1.19 (0.83, 1.70)0.023
 North-east2751 (16)4115 (12, 19)0.98 (0.70, 1.39)
 North-west5372 (31)8215 (12, 19)1
 South-east1603 (9)4125 (19, 35)1.68 (1.15, 2.47)
 South-south2310 (13)5423 (18, 31)1.56 (1.10. 2.22)
 South-west3075 (17)6120 (15, 26)1.31 (0.91, 1.89)
Overall17 63532518.5 (16.5, 20.6)  

The overall neonatal mortality rate was 40 deaths per 1000 live births. Multiple births had a neonatal mortality rate of 132 (CI: 107–163) deaths per 1000 live births, compared with 37 (CI: 34–39) deaths per 1000 live births for singletons, an adjusted odds ratio (AOR) of 5.74 (CI: 3.97–8.28, P < 0.0001) after taking into account birth order, wealth, maternal education and uptake of antenatal care (unadjusted OR 4.01, CI: 3.12–5.17). Twins and triplets accounted for 7.5% of all neonatal mortality.

Healthcare utilization

Of the 17 635 pregnancies, 10 869 (63%) women made at least one visit to antenatal care, 6977 (40%) had skilled attendance at delivery, 6272 (36%) delivered at a health facility, 7728 (45%) had postnatal care and 347 (2%) had a caesarean section. The greatest coverage of skilled attendance was 81.8% in the south-east and 77.7% in the south-west. The lowest coverage was 11.1% in the north-west followed by 17.1% in the north-east. Other healthcare utilization variables followed a similar pattern with region (Figure 1).

Figure 1.

 Maternal healthcare coverage (%) by region, Nigeria Demographic and Health Survey 2003–2008.

Our second main objective was to explore the utilization of health care for twin pregnancies compared with singletons. Table 2 shows that after adjusting for wealth, education, birth order and ethnicity, women with multiple pregnancies (compared with singleton) were more likely to deliver with a skilled attendant (AOR 1.75, CI: 1.36–2.24), have a health facility delivery (AOR 1.51, CI: 1.13–2.02), undergo a caesarean section (AOR 3.54, CI: 2.08–6.01) and attend postnatal care (AOR 1.37, CI: 1.04–1.81). However, there was no evidence of increased antenatal care use (AOR 0.96, CI: 0.73–1.28).

Table 2.   The effect of multiple birth status on healthcare utilization, Nigeria 2003–2008. Odds ratio (OR) is adjusted for wealth, education, birth order and ethnicity
Outcome variablesExplanatory variableAdjusted OR* (95% CI)Wald P-value
Singleton n (%)Multiple n (%)
Antenatal care
 None6316 (37)88 (27)10.80
 1+ visit/unsure10 860 (63)218 (67)0.96 (0.73, 1.28)
 Missing134 (0.8)19 (6) 
Skilled attendance
 No10 239 (59)141 (43)1<0.001
 Yes6977 (40)180 (55)1.75 (1.32, 2.34)
 Missing93 (0.5)4 (1) 
Place of delivery
 Home10 995 (64)161 (50)10.006
 Health facility6272 (36)159 (49)1.51 (1.13, 2.02)
 Missing43 (0.2)4 (1) 
Postnatal care
 No9433 (55)140 (43)10.025
 Yes7728 (45)180 (55)1.37 (1.04, 1.81)
 Missing148 (1)5 (2) 
Caesarean section
 No16 954 (98)302 (93)1<0.001
 Yes347 (2)23 (7.1)3.54 (2.08, 6.01)
 Missing8 (<0.1)0 (0) 
Total (n = 17 635)17 310 (98.2)325 (1.8)  

Antenatal care

Of the women who attended antenatal care, women with multiple gestations had slightly more antenatal care visits than singletons (P = 0.0098) (Table 3). There was no evidence of a difference between singleton and multiple gestations in the proportion of mothers attending antenatal care in the first trimester, nor was there a difference in the number of components of antenatal care received. However, women with multiple gestation appeared to receive more urine tests (OR 1.51, CI: 1.09–2.09, = 0.013) and more blood pressure (BP) checks (OR 1.52, CI: 0.99–2.32, = 0.056).

Table 3.   The effect of multiplicity on the quality of antenatal care (ANC) for those who attended, Nigeria DHS 2003–2008
OutcomeSingleton pregnancy n (%)Multiple pregnancy n (%)OR (95% CI)Wald/χ2 test P-value
  1. DHS, Demographic and Health Survey.

Number of visits to ANC (if attended)
 1–31673 (15)27 (11)(excludes missing and unsure)0.0098
 4–94746 (43)78 (33)
 10–503002 (27)79 (33)
 Unsure1439 (13)35 (15)
 Missing134 (1)19 (8)
Trimester of first ANC visit
 1st2810 (26)55 (23)10.72
 2nd or 3rd7894 (72)163 (69)1.06 (0.77, 1.46)
 Unsure194 (2)1 (0.4) 
 Missing96 (1)19 (8) 
Weight measured
 No1380 (13)21 (10)10.30
 Yes9464 (87)196 (90)1.24 (0.82, 1.88)
Blood pressure
 No1573 (15)19 (9)10.056
 Yes9268 (85)198 (91)1.52 (0.99, 2.32)
Urine sample
 No2728 (25)36 (17)10.013
 Yes8106 (75)181 (83)1.51 (1.09, 2.09)
Blood sample
 No2809 (26)48 (22)10.38
 Yes8024 (74)169 (78)1.15 (0.84, 1.56)
Told of complications
 No1067 (37)66 (30)10.27
 Yes6672 (61)150 (69)1.17 (0.89, 1.53)
Told where to go if complications
 No211 (3)4 (2)10.55
 Yes6433 (96)147 (98)1.29 (0.56, 2.96)
Took iron supplements
 No1804 (17)21 (10)10.066
 Yes8844 (81)192 (88)1.23 (0.99, 1.54)
Took an anti-malarial drug
 No7102 (65)140 (64)10.68
 Yes3441 (32)66 (30)0.95 (0.72, 1.24)

Women without antenatal care had a very low rate of delivery with a skilled attendant, with 5% of singletons and 13% of multiple pregnancies receiving skilled care at delivery. Conversely, women who received antenatal care had more skilled attendant use at delivery – 62% of singletons and 73% of multiple pregnancies.

The third main objective of this study was to explore whether the presence of antenatal care strengthened the association between multiplicity and skilled attendance. After taking into account wealth, education, ethnicity and birth order, there was no evidence that antenatal care modified the relationship between multiplicity and skilled attendance. The odds ratio interaction parameter had close to no effect; AOR = 0.99 (CI: 0.44–2.21, = 0.98). While women with a multiple pregnancy and women attending antenatal care were each more likely to have skilled attendance at delivery, multiples attending antenatal care did not receive an additional boost that led them to skilled attendance. If anything, it appeared that multiplicity itself was more strongly associated with skilled attendance amongst those women who did not receive antenatal care (AOR 2.54 vs. 1.85) (Table 4).

Table 4.   Test for effect modification shows no evidence that antenatal care (ANC) modifies the relationship between multiple pregnancy and skilled attendance (SA), and therefore, pooled odds ratio (OR) estimate should be used, Nigeria 2008. 1.8% missing data (*Adjusted for wealth quintile, education, birth order and ethnicity). These figures are also adjusted for weighting and clustering of the sample
 Skilled Attendance (%) missing 0.6%Adjusted* OR for skilled attendance (95% CI)Wald P-value
Pooled data
 Multiple551.75 (1.36, 2.24)
Stratified by ANC missing 0.7%
 Without ANC (n = 6368)
  Multiple132.54 (1.16, 5.56)
 Received ANC (n = 11 050)
  Multiple731.85 (1.33, 2.58)
Test for effect modification = 0.98

Response rate

The overall response rate reported was 94.9%, ranging from 94.1% to 98.1% in the different regions. There were no missing information on multiple birth status, and a small proportion (0–0.81%) of women was missing data for the main outcomes. There were more missing data on antenatal care for multiples (6%) than singletons (0.8%). Women missing data on antenatal care had more multiple births than women overall (126 vs. 18 per 1000 maternities) and higher neonatal mortality rates (128 vs. 28 deaths per 1000 live births).


Strengths and limitations

The 2008 Nigerian DHS provides a strong platform and a large sample size from which to investigate healthcare utilization for twin pregnancies. To our knowledge, this is the first time these issues have been explored in a nationally representative population-based sample in a low-income country. The high response rate (95%) and minimal missing data minimize non-response bias. Our multivariate analysis takes into account the majority of known potential confounders.

The Nigeria DHS was not specifically designed to answer questions around multiplicity, so important questions were missed entirely or may not have been probed or recorded accurately. In particular, there was no information on healthcare utilization in pregnancies resulting in stillbirth. Justesen and Kunst (2000) suggest that the DHS registers a twin as a singleton if the second twin is stillborn, because all twin pairs in the DHS consist of two live-born children. This would under-estimate the incidence of multiplicity and could alter the magnitude of the associations seen with multiplicity, if, for example, the surviving twin is recorded as a singleton, but faces the neonatal mortality risk of other twins.

A long interview format poses constraints such as fatigue, inaccurate recording or pressure to give socially desirable answers. However, care was taken to ensure privacy and neutrality, and an interviewer could return to complete the survey at a different time.

Another limitation is that the small proportion of women missing data on antenatal care and skilled attendance had more neonatal mortality and multiple births than women overall. This is unlikely to substantially influence results if missing data are evenly distributed. However, for antenatal care, the proportion missing varied by multiplicity, so that 6% of multiple pregnancies had no antenatal data compared with 0.8% of singletons. This would overestimate the magnitude of the association between multiplicity and antenatal care, which may exaggerate the association between multiplicity and the number of antenatal visits.

Multiplicity and mortality

Nigerian multiple births (18.5 multiple births per 1000 maternities) are statistically more frequent than the 16.4 per 1000 maternities reported in England and Wales (Office of National Statistics 2010). Nigerian twinning is less frequent than might be expected based on earlier publications, which record multiple birth rates of 28–53 per 1000 maternities (Knox & Morley 1960; Nylander 1969; Rehan & Tafida 1980; Aisien et al. 2000; Akinboro et al. 2008). This may be because previous studies sampled hospitals or limited geographical locations and were not population based. However, our data may underestimate the true pregnancy incidence of multiplicity, because twins are registered as singletons if the second twin is stillborn Justesen and Kunst (2000). The young average maternal age in Nigeria (high maternal age is associated with twinning) and infrequent use of IVF and fertility treatment compared with the United Kingdom and the United States also are likely to contribute to a more comparable rate of twinning than expected. Our figure of 37 multiple births per 1000 live newborns (3.7%) is also comparable to the 4% calculated from the 2003 Nigeria DHS by Uthman et al. (2008), the 3.7% in Eastern and Southern Africa (Justesen & Kunst 2000) and the 3.8% reported for the non-Hispanic Black group in the United States (Martin et al. 2010).

Our analysis also shows that the highest rates of multiple pregnancy were in the south-east (25 per 1000 maternities) and not amongst the Yoruba in the south-west as shown in the literature (Akinboro et al. 2008). The high rates of cultural taboo in the south-east previously associated with twinning do not explain our results (Asindi et al. 1993).

The neonatal mortality rate overall was 40 (CI: 37–43) deaths per 1000 live births, but multiples faced markedly increased odds of neonatal mortality (AOR 5.74, CI: 3.97–8.28). This is not dissimilar to the 6.24 (CI: 5.02–7.77) increased odds demonstrated by Justesen and Kunst (2000), who pooled Malawi, Tanzania and Zambia DHS data, and the 6.7 times increased odds of neonatal mortality seen for twins in the UK (CEMACH 2009).

Skilled attendance

Skilled attendance is widely recommended to prevent maternal and neonatal mortality. Delivery by a skilled attendant in the Nigeria DHS (2008) was only 40.6%, a slight improvement from the 35.2% seen in the DHS (2003). The overall caesarean section rate (2%) was lower than the 5% recommended minimum (WHO 1997). Women with a multiple pregnancy consistently utilized more care for delivery (AOR 1.75), with up to three times increased odds of caesarean section compared with singletons.

We have not found any other studies in low-income countries that explore delivery care utilization for twins. However, the increased use of delivery care services with multiples may stem from the higher rate of complications and symptoms that women with multiples are likely to increasingly experience with pregnancy duration. In Guatemala, Glei et al. (2003) demonstrated that women with serious pregnancy complications were about twice as likely to seek biomedical care as their counterparts. Conversely, in southern Tanzania, Mpembeni et al. (2007) showed that women who reported complications during pregnancy did not have increased skilled attendance (n = 974).

Antenatal care

Antenatal care coverage was low at 63%, but also a slight improvement since the 2003 DHS (58%). A similar proportion of women with multiple and singleton pregnancies utilized antenatal care (AOR 0.95), and the timing of the first visit was comparable. Some aspects of content appeared to vary with multiplicity, so that multiples had more visits and more BP and urine tests. Missing data (Table 3) may have contributed to these associations; however, there is a plausible explanation. Women with dizygotic multiple pregnancies have an increased incidence of pregnancy-induced hypertension (PIH) or pre-eclampsia. Symptoms such as oedema, headaches and dizziness may have prompted women to attend antenatal care more frequently and triggered the monitoring of BP and urinary protein.

Tests for interaction were crucial to this study’s objective. We hypothesized that women with a multiple pregnancy who used antenatal care should have increased their chances of using skilled attendance more than the singletons who used antenatal care, because health workers should be able to detect a multiple pregnancy and recommend facility delivery, helping women overcome any barriers if necessary. The interaction term (0.99) showed there was no effect modification; in other words, no evidence that antenatal care was of additional benefit to women with a multiple pregnancy in encouraging the use skilled attendance. Interaction tests often lack power, but our confidence intervals for the interaction parameter were not wide (CI: 0.44–2.21), suggesting power to detect a moderate interaction effect if present.

The availability of skilled attendance, while posing a significant barrier to use in Nigeria, was likely to affect both multiples and singletons similarly. We cannot tell whether the lack of additional benefit for multiple pregnancies attending antenatal care was because health care workers did not assess multiplicity, assessed multiplicity but failed to detect it, detected it but did not advise women with multiples more strongly to seek skilled attendance, or gave advice but had their recommendations ignored.

The literature suggests that all these steps can be implemented, but also indicates that they may be obstacles in some settings. Abdominal examination to assess foetal growth was reported by 95.7% of women who attended antenatal care in a study in north-west Nigeria (Galadanci et al. 2007). In Bangladesh, antenatal diagnosis using abdominal examination detected 68% of all cases of twins (Vanneste et al. 2000), whereas in a Tanzanian study, only two in eight twin pregnancies (25%) were identified before delivery Jahn et al. (1998).

The benefits of quality skilled attendance can also be communicated. In southern Tanzania, antenatal advice to deliver in a health facility increased skilled attendance by 43%, holding other factors constant (Mpembeni et al. 2007). Knowledge of pregnancy risk factors and discussion with a male partner on the place of delivery also increased use of skilled attendance. Another study in southern Tanzania (Jahn et al. 1998) found that as many as 91.5% women with previous caesarean section sought facility delivery and had largely been referred by health providers. In a socio-economically disadvantaged region of eastern Zaire, the referral success rate was 33% (Dujardin et al. 1995).

By contrast, other studies in southern Tanzania found that expert-defined risk status had little influence on a woman’s decision to seek hospital care (Kowalewski et al. 2004), with women relying instead on community perceptions of severity and aetiology and danger signs. Jahn et al. (1998) showed that antenatal care had limited benefit in increasing obstetric care for high-risk mothers.

We suggest that antenatal care should identify Nigerian women with twins and assist them and their families to optimize their use of services, ideally in ways that take into account their perception of risk and illness explanatory models.

Further research

Additional analyses to look at the benefit of antenatal care to high-risk pregnancies in other settings, including settings with fewer resource constraints, would be desirable. Research is also required to verify the benefit and feasibility of client-centred antenatal care approaches in improving skilled attendance. Health system strengthening interventions aimed at improving the quality of antenatal care and skilled attendance should be established, monitored and evaluated. The focus should include not just the coverage but also the quality of care.


Multiple births are not uncommon in Nigeria and experience a substantially higher risk of neonatal mortality, accounting for 7.5% of all neonatal mortality. Despite limited access to health services in Nigeria and multiple barriers to skilled attendance, women with multiple pregnancies did report more delivery care. As far as we were able to measure, the reasons for this do not seem to be related to the presence or quality of antenatal care. This finding calls in to question what is being accomplished through antenatal care services.

For policy-makers, health providers and the wider community, a multiple pregnancy should not just be perceived as a common occurrence, but an indicator of a high-risk pregnancy that requires referral to obstetric services for delivery care. To strengthen antenatal and delivery care, improved coverage of well-staffed, well-resourced and well-trusted health facilities is urgently required with priority to rural and northern regions.


We wish to acknowledge and thank Professor Simon Cousens for assistance with the statistical analysis, particularly with the use of Stata and the analysis of effect modification, and Caroline Free and Andy Sloggett for their support and advice.