Infants Delivered in Maternity Homes Run by Traditional Birth Attendants in Urban Nigeria: A Community-Based Study

Abstract

We explored factors associated with traditional maternity/herbal homes (TMHs) run by traditional birth attendants (TBAs) compared with hospital or home delivery in Lagos, Nigeria, and found that infants delivered at TMHs were less likely to have severe hyperbilirubinemia compared with infants delivered in hospitals or residential homes. These infants were also less likely to be preterm compared with those delivered in hospitals or undernourished compared with infants delivered in residential homes. We concluded that infants delivered at TMHs who survive are unlikely to be at greater risks of some adverse perinatal outcomes than those delivered in hospitals or family homes.

Of the 122 million babies born in developing countries every year, about 44% are delivered outside hospitals and 37% without skilled attendants at delivery (United Nations Children's Fund [UNICEF], 2009). In sub-Saharan Africa and South Asia, which together account for over half (57%) of deliveries in the developing world, up to 65% of births occur outside hospitals and are predominantly delivered by traditional birth attendants (TBAs) (Sibley et al., 2007; World Health Organization [WHO], 2004). Many TBAs are independent (of the health system), nonformally trained and community-based providers of care during pregnancy, childbirth, and the immediate postnatal period (WHO, 2004).

The role of TBAs in maternal and child health care in developing countries has been a subject of seemingly unending debate principally because of the high rates of maternal and child mortality more closely associated with deliveries not supervised by skilled attendants comprising doctors, midwives, and nurses (Campbell, Graham, & Lancet Maternal Survival Series Steering Group, 2006; Costello, Azad, & Barnett, 2006; Sibley et al., 2007). While the services provided by TBAs are inevitable and widespread in rural areas due to the lack of or poor access to orthodox health facilities, there is growing evidence suggesting that regardless of their socio-economic status, many women in urban areas where physical access and financial barriers to facility-based obstetric services are minimal still choose the services offered by TBAs either in their residential homes or in commercial traditional maternity/herbal homes (TMHs) (Izugbara, Ezeh, & Fotso, 2009; Khan et al., 2009; Olusanya, Alakija, & Inem, 2010; Rahman, Tarafder, & Mostofa, 2008; Sreeramareddy et al., 2006). For instance, in Bangladesh, 61% of women in urban areas have been reported to use TBAs (Rahman, Tarafder, & Mostofa, 2008). Similarly, about 40% of women in an inner-city community in sub-Saharan Africa wellserved by several health facilities were reported to have delivered in private TMHs exclusively run by TBAs and accounted for about three-quarters of all nonhospital deliveries (Olusanya, Alakija, & Inem, 2010). In rural areas, TBAs provide their services predominantly in the homes of individual clientele, whereas in urban areas the same services are more commonly provided in private maternity “facilities” owned by TBAs in addition to residential homes (Kamal, 1998).

While there is increasing attention to the scaling up of skilled attendants as part of the current global strategy to curtail maternal and neonatal mortality in developing countries (United Nations [UN], 2008), this goal appears unattainable in many resource-poor countries due to the combination of several sociocultural and economic factors (Kamal, 1998; Koblinsky et al., 2006). Some researchers, therefore, have argued for context-specific strategies for each country that essentially entail that the use of TBAs and skilled attendants should not be viewed as being mutually exclusive (Costello, Azad, & Barnett, 2006; Sibley & Sipe, 2006). For example, every month, at least 5 million people migrate mainly into the already overcrowded cities in developing countries, and the current urban population of 39% in Africa in particular would hit 50% by 2030 at an estimated annual growth rate of 4.7%, the highest worldwide (UN, 2007; United Nations Human Settlements Program [UN-Habitat], 2008). Evidently, the services provided by TMH/TBAs will continue to serve at least as a necessary stop-gap for the more ideal care by skilled attendants in many communities against the backdrop of this rapid urban transition in some countries (UN, 2008).

While the perinatal mortality associated with deliveries by TBAs mainly in rural areas along with the potential impact of targeted training in enhancing their effectiveness has been extensively reported (Jokhio, Winter, & Cheng, 2005; Sibley et al., 2007), the perinatal profile of the survivors of delivery by TBAs in TMHs particularly in urban settings is sparsely documented. Such information is essential to more accurately capture the perinatal quality-of-life outcomes associated with delivery by TBAs. We therefore set out to determine the adverse perinatal outcomes independently associated with use of TMH run by TBAs in an urban setting well served by both private and public hospitals after adjusting for the potential confounding effects of maternal predictors of place of delivery.

METHODS

Setting

This study was conducted in an inner-city area of Lagos, southwest Nigeria, with a population of about 250,000. The community is served by several private hospitals and TMHs as well as one general hospital, one children's hospital, one specialist maternity hospital, and seven primary health care centers, all of which are state owned. Only one-third of deliveries occur in these health facilities (UNICEF, 2009). The city of Lagos, the nation's capital since independence in 1960 until ceded to Abuja in 1991, is still Nigeria's commercial/financial hub and currently about 94% urbanized. Nigeria's population is presently 48% urbanized and it is projected to grow by more than twofold to 289 million by 2050, with a majority of urban dwellers from the year 2020 on (UN, 2007).

Study Population

All women attending four community health centers to obtain Bacille Calmette-Guérin (BCG) immunization for their newborns between July 2005 and April 2008 were enlisted for the study. Routine immunization clinics are widely acknowledged as providing a cost-effective platform for other primary health care services in developing countries, with a unique advantage of attracting babies born outside hospitals who cannot be reached by hospital-based interventions. The uptake for BCG immunization in Lagos is about 75%–98% compared with the national average of 69% (Babalola & Adewuyi, 2005), and the four immunization clinics used under this project accounted for over 75% of BCG vaccinations in this study location. Ethical approval was obtained as part of an extended pilot project on universal infant hearing screening (Olusanya, Wirz, & Luxon, 2008) from the Lagos State Health Management Board, Nigeria, and University College London, United Kingdom, according to the guidelines laid down in the Declaration of Helsinki. Informed consent was obtained from all participating mothers in writing or by thumb printing before enrollment.

Main Outcome Variables

Women who delivered in TMHs were compared with those who delivered in hospitals (private and public) or in family residential homes. Traditional maternity homes (TMHs) also were referred to as herbal homes because of the exclusive use of herbal preparations/medications for all ailments and procedures. These homes are located in close proximity to most of their clients, and a typical set-up consists of one split-room apartment made up of consulting and delivery cubicles. The available facilities are mainly delivery couch, fetoscope, stethoscope, sphygmomanometer, wash-hand sink, baby scale, record books, and herbal dispensary. The owner TBA usually resides in the same premises and is assisted by a few family members or young apprentices. Women who delivered in church premises/faith-based clinics or before arrival in a health facility (usually in cars or public transport) were excluded because of their distinctive antecedents and trajectories beyond the scope of this study (Adetunji, 1992).

Maternal Factors

The factors of interest were guided by evidence from published literature and those that could be reliably elicited from the mothers by trained research assistants (Gabrysch & Campbell, 2009; Stephenson, Baschieri, Clements, Hennink, & Madise, 2006). These included sociodemographic factors such as age, marital status, parity, ethnicity, religion, education, occupation, and social class. The actual maternal age obtained from each participant subsequently was categorized into commonly reported thresholds reflecting

subgroups most at risk of poor obstetric outcome. Social class is a useful proxy for socioeconomic status and was determined based on mother's education and father's occupation as previously validated in this study population (Olusanya, Okpere, & Ezimokhai, 1985). Social class I was termed as “high,” II or III as “middle,” and IV or V as “low.” Maternal factors also included obstetric variables such as antenatal care, use of traditional herbal medications in pregnancy, and pregnancy type. Information on the type of attendant at delivery comprising doctor, nurse, midwife, TBA, auxiliary nurse, or others (relation, neighbor or friend) also was documented. Additionally, practices regarding cord cutting and cord dressing as well as infant feeding were obtained.

Infant Factors/Outcomes

The infant factors or outcomes of interest were gender, gestational age (full term, i.e., 37 weeks and over, or preterm, i.e., less than 37 weeks) and gestation type (singleton or multiple). Gestational age was estimated from the first day of the mother's reported last normal menstrual period. Birth weight and length could not be ascertained as birth records of the

participants, particularly those born outside hospitals, were not available in this community-based setting. Instead, anthropometric measurements for each child were obtained at the time of enrollment by a trained research worker throughout the study period to determine a child's nutritional status. Gender-specific z-scores for weight-for-age (WAZ), body mass index-for-age (BMI), and height/length-for-age (HAZ) expressed as z-scores were obtained from the WHO's Multicentre Growth Reference (WHO-MGR) package using the macro provided by the organization (WHO, 2005). Moderate-to-severe or “significant” wasting, underweight, and stunting were defined as zBMI, WAZ, and HAZ below –2, respectively. The term “any undernourishment” was included in the final analysis as a summary factor to reflect the comorbidity of the three nutritional deficits in this population. Although lack of access to individual birth records did not allow us to consider potentially relevant perinatal outcomes such as Apgar scores, birth asphyxia, and sepsis, history of severe neonatal jaundice in the first week of life necessitating hospital admission for phototherapy, for exchange blood transfusion, or both was documented. Both undernourishment and severe hyperbilirubinemia are prominent causes of long-term cognitive and neurodevelopmental deficits such as mental retardation, cerebral palsy, and sensory disorders in many survivors in the developing world (Committee on Nervous System Disorders in Developing Countries, Board on Global Health, 2001). Hearing screening outcomes also were considered within the context of the primary study (Olusanya et al., 2008). The three screening outcomes follow: pass, refer, or incomplete.

Statistical Analysis

Cross-tabulation of the three places of delivery and explanatory variables was done to provide a descriptive overview of our study population. A multinomial logistic regression model was used to estimate the effect of the independent variables on the probability of a particular place of delivery. The three outcomes were verified to be sufficiently distinct as to satisfy the assumption of independent alternatives based on the Hausman and Small-Hsiao tests (Long & Freese, 2006). Two sets of multinomial analyses were conducted. In the first model, maternal factors significantly (p < .05) associated with place of delivery were first determined and subsequently used as adjustment variables in the second model containing the infant factors. The strength of an association was estimated by odds ratios (ORs) and the corresponding 95% confidence intervals (CIs). The exponentiated coefficients in each model were the conditional ORs of delivery in TMHs versus delivery in hospitals or at home, simultaneously with delivery in homes versus delivery in hospitals. Each independent variable in the model was compared with the specified reference category while holding other variables constant. There were no a priori hypotheses for interaction terms, so these were not investigated. Missing data were managed by exclusion in all of the analyses. All statistical analyses except for the Hausman and Small-Hsiao tests (STATA) were done with SPSS Windows version 16.0 (SPSS Inc., Chicago, IL, USA).

RESULTS

A total of 6,706 mother–infant pairs were enrolled for this study out of which 3,403 (50.7%), 2,847 (42.5%), and 456 (6.8%) delivered in hospitals, TMHs, and family homes, respectively. All eligible mothers consented to participate in the study. The majority of the women were married, from the Yoruba ethnic group, shared Muslim faith, had a minimum of a secondary education as their spouses, were engaged in small-trade or casual work, and belonged to middle social class (Table 1). All but 2.1% of the women had antenatal care, and almost two-thirds (65.5%) used herbal preparations during pregnancy. More than half (52.2%) of the infants were male, and the vast majority were fullterm (Table 2). While 8.4% of all infants were undernourished by at least one measure of nutritional deficit, about 5.1% had hyperbilirubinemia requiring phototherapy and 1.6% also had exchange blood transfusion (EBT). Some 69.7% of all infants who received phototherapy and 71.4% of those who had EBT were delivered in hospitals. About 1.7% of all infants failed the hearing tests, while more than half (51.5%) of those who did not complete the hearing tests were delivered in hospitals.

Table 1 Maternal Profile and Place of Delivery in Lagos, Nigeria

	Hospitals	Herbal homes	Family homes
	n = 3,403	n = 2,847	n = 456	Total
Profile	(%)	(%)	(%)	n = 6,706
Age (Years) [a]
<20	88  (38.9)	122  (54.0)	16  (7.1)	226
20–35	2994  (50.3)	2556  (42.9)	406  (6.8)	5956
>35	314  (61.3)	164  (32.0)	34  (6.6)	512
Marital status
Not married	106  (41.9)	137  (54.2)	10  (4.0)	253
Married	3297  (51.1)	2710  (42.0)	446  (6.9)	6453
Ethnicity
Yoruba	2824  (47.1)	2792  (46.5)	383  (6.9)	5999
Hausa	104  (67.1)	18  (11.6)	33  (21.3)	155
Ibo	418  (88.4)	26  (5.5)	29  (6.1)	473
Other	57  (77.2)	11  (13.9)	11  (13.9)	79
Religion
Islam	1906  (41.2)	2407  (52.0)	316  (6.8)	4629
Christianity	1497  (72.2)	440  (21.2)	140  (6.7)	2077
Education
None	71  (36.6)	88  (45.4)	35  (18.0)	194
Primary	358  (34.7)	561  (54.4)	112  (10.9)	1031
Secondary and above	2974  (54.3)	2198  (40.1)	309  (5.6)	5481
Education of spouse
None	34  (49.3)	17  (24.6)	18  (26.1)	69
Primary	76  (36.5)	101  (48.6)	31  (14.9)	208
Secondary and above	3293  (51.2)	2729  (42.4)	407  (6.3)	6429
Occupation
None	574  (62.5)	277  (30.1)	68  (7.4)	919
Small trade/casual job	1911  (43.2)	2179  (49.3)	329  (7.4)	4419
Full-time job	918  (67.1)	391  (28.6)	59  (4.3)	1368
Social class
Low	591  (40.8)	733  (50.6)	125  (8.6)	1449
Middle	2454  (50.5)	2081  (42.8)	322  (6.6)	4857
High	358  (89.5)	33  (8.2)	9  (2.2)	400
Parity [b]
Primiparous	1471  (53.0)	1153  (41.6)	149  (5.4)	2773
Multiparous	1929  (49.1)	1694  (43.1)	307  (7.8)	3930
Antenatal care
Yes	3334  (50.8)	2781  (42.4)	451  (6.9)	6566
None	69  (49.3)	66  (47.1)	5  (3.6)	140
Herbal drug in pregnancy
Yes	1344  (30.6)	2753  (62.7)	295  (6.7)	4392
None	2059  (89.0)	94  (4.1)	161  (7.0)	2314
Multiple pregnancy
No	3323  (50.5)	2811  (42.7)	449  (6.8)	6583
Yes	80  (65.0)	36  (29.3)	7  (5.7)	123
Missing data: [a] = 12 (0.2%); [b] = 3 (0.0%). Herbal homes = traditional maternity homes.

Table 2 Infant Profile and Place of Delivery in Lagos, Nigeria

	Hospitals	Herbal homes	Family homes
	n = 3,403	n = 2,847	n = 456	Total
Profile	(%)	(%)	(%)	n = 6,706
Gender
Female	1617  (50.5)	1383  (43.2)	205  (6.4)	3205
Male	1786  (51.0)	1464  (41.8)	251  (7.2)	3501
Gestation age [a]
Preterm  (<37 weeks)	68  (73.1)	20  (21.5)	5  (5.4)	93
Full-term  (≥ 37 weeks)	3333  (50.5)	2824  (42.7)	449  (6.8)	6606
Underweight [b]
No	2892  (51.1)	2416  (42.7)	348  (6.2)	5656
Yes	400  (47.4)	360  (42.7)	83  (9.8)	843
Stunted [c]
No	1991  (51.8)	1619  (42.1)	234  (6.1)	3844
Yes	879  (51.7)	688  (40.4)	134  (7.9)	1701
Wasted [d]
No	2637  (53.0)	2033  (40.9)	305  (6.1)	4975
Yes	232  (42.3)	256  (46.6)	61  (11.1)	549
Jaundice with phototherapy
No	3166  (49.7)	2786  (43.8)	414  (6.5)	6366
Yes	237  (69.7)	61  (17.9)	42  (12.4)	340
Jaundice with EBT
No	3328  (50.4)	2829  (42.9)	444  (6.7)	6601
Yes	75  (71.4)	18  (17.1)	12  (11.4)	105
Hearing screening outcome
Pass	3162  (50.8)	2640  (42.4)	418  (6.7)	6220
Refer	51  (43.6)	50  (42.7)	16  (13.7)	117
Incomplete	190  (51.5)	157  (42.5)	22  (6.0)	369
Missing data: [a] = 7 (0.1%); [b] = 207 (3.1%); [c] = 1,161 (17.3%); [d] = 1,182 (17.6%).
EBT = exchange blood transfusion; Herbal homes = traditional maternity homes.

A total of 2,916 births were assisted by TBAs. Whereas delivery in TMHs was exclusively handled by TBAs, only 69 (2.4%) of all TBAs were present at delivery in clients’ homes (Figure 1). Delivery in hospitals was predominantly (77.5%) handled by midwives, followed by doctors (11.9%) and nurses (10.5%). About 36% of home delivery was supervised by nurses and was least (0.7%) attended by doctors. Neighbors, relations, and friends accounted for 25% of all home delivery, and only three women reported not having any attendant at delivery. The average delivery fee charged ranged from the local equivalent (Naira) of U.S.$19 for auxiliary nurses, U.S. $41 for nurses, U.S. $42 for TBAs, U.S. $61 for midwives, and U.S. $315 for doctor-assisted delivery. In the majority (98.5%) of delivery in the hospital, the baby's cord was cut with either a surgical knife or scissors, while, blade was used in 89.6% of deliveries in TMHs. Only 5 (0.1%) women reported that the baby's cord was dressed with mentholated powder rather than alcohol or spirit in the vast majority, and only 40 infants overall did not receive exclusive breastfeeding (data not shown).

Figure 1 Types of birth attendant at place of delivery in Lagos, Nigeria.

Maternal Predictors of Place of Delivery

In the univariate multinomial analyses, all maternal factors except antenatal care were significantly associated with delivery in TMHs compared with hospital or home delivery. Based on biological plausibility, all factors were entered into the multivariable logistic regression models and all except marital status, antenatal care, and multiple pregnancy were predictive of delivery in TMHs (Table 3). Women who delivered in TMHs compared with those who delivered in hospitals were significantly more likely to be teenagers, of Muslim faith, belong to low or middle social class, and used herbal drugs in pregnancy but were less likely to be non-Yoruba. Additionally, these women were also significantly more likely to be primiparous compared with those who delivered in their residential homes. Women who delivered in their homes, compared with hospital delivery, were significantly more likely to belong to the Hausa tribe, with primary or no education and engaged in small trade, casual employment, or none. They were also more likely to belong to the low or middle social class and use herbal drugs in pregnancy but less likely to be first-time mothers. Also, first-time mothers were less likely to deliver outside hospitals. While multiple pregnancy was not predictive of place of delivery, affected women were less likely to deliver in TMHs or their homes compared with hospitals.

Table 3 Maternal Factors Associated With Place of Delivery in Lagos, Nigeria

	Herbal homes	Herbal homes	Family homes
	versus hospital	versus family homes	versus hospital
Profile	OR (95% CI)	OR (95% CI)	OR (95% CI)
Age (Years) [a]
<20	1.88 (1.28–2.78)**	1.31 (0.71–2.42)	1.44 (0.78–2.64)
20–35	1.0	1.0	1.0
>35	0.70 (0.55–0.91)**	0.87 (0.58–1.30)	0.81 (0.55–1.18)
Marital status
Not married	1.0	1.0	1.0
Married	0.92 (0.65–1.30)	0.52 (0.26–1.04)	1.77 (0.88–3.58)
Ethnicity
Yoruba	1.0	1.0	1.0
Hausa	0.49 (0.26–0.94)*	0.26 (0.13–0.53)***	1.87 (1.13–3.10)*
Ibo	0.43 (0.27–0.70)**	0.47 (0.26–0.87)*	0.91 (0.59–1.42)
Other	0.50 (0.23–1.11)	0.41 (0.16–1.05)	1.22 (0.59–2.52)
Religion
Christianity	1.0	1.0	1.0
Islam	1.50 (1.27–1.76)***	1.53 (1.19–1.99)**	0.98 (0.76–1.25)
Education
None	2.24 (1.39–3.61)**	0.83 (0.46–1.47)	2.72 (1.54–4.80)**
Primary	1.67 (1.38–2.02)***	0.75 (0.58–0.98)*	2.23 (1.70–2.92)***
Secondary and above	1.0	1.0	1.0
Education of spouse
None	0.64 (0.29–1.42)	0.40 (0.17–0.96)*	1.60 (0.75–3.42)
Primary	1.04 (0.70–1.56)	0.67 (0.41–1.09)	1.56 (0.96–2.53)
Secondary and above	1.0	1.0	1.0
Occupation
None	1.16 (0.91–1.49)	0.73 (0.48–1.11)	1.58 (1.07–2.34)*
Small trade/casual job	1.51 (1.27–1.79)***	0.87 (0.64–1.19)	1.73 (1.28–2.35)***
Full-time job	1.0	1.0	1.0
Social class
Low	3.57 (2.31–5.52)***	1.29 (0.58–2.88)	2.78 (1.34–5.75)**
Middle	3.72 (2.46–5.62)***	1.35 (0.62–2.93)	2.75 (1.38–5.50)**
High	1.0	1.0	1.0
Parity
Primiparous	0.94 (0.82–1.08)	1.32 (1.05–1.66)*	0.72 (0.57–0.90)**
Multiparous	1.0	1.0	1.0
Antenatal care
Yes	1.0	1.0	1.0
None	0.95 (0.63–1.45)	2.25 (0.88–5.75)	0.42 (0.17–1.08)
Herbal drug in pregnancy
None	1.0	1.0	1.0
Yes	31.13 (24.88–8.95)***	11.35 (8.40–15.34)***	2.74 (2.17–3.46)***
Multiple pregnancy
None	1.0	1.0	1.0
Yes	0.69 (0.42–1.13)	1.08 (0.46–2.53)	0.64 (0.29–1.42)
*p < .05; **p < .01; ***p < .001. Herbal homes = traditional maternity homes. Model adjusted for all covariates.

Perinatal Outcomes and Correlates

After adjusting for all the potential maternal confounders, infants delivered at TMHs were significantly less likely to have severe hyperbilirubinemia

requiring phototherapy compared with infants who were delivered in hospitals (OR:0.61; 95% CI:0.39–0.94) or those delivered at family homes (OR:0.34; 95% CI:0.20–0.59) as shown in Table 4. Similarly, these infants were also significantly less likely to be preterm (OR:0.48; 95% CI:0.24–0.87) compared with those born in hospitals or undernourished (OR:0.70; 95% CI:0.56–0.87) compared with infants delivered in family homes. Additionally, infants delivered in family homes were at increased risk of undernourishment (OR:1.47; 95% CI:1.19–1.84) and hyperbilirubinemia requiring phototherapy (OR:1.77; 95% CI:1.12–2.81) compared with infants delivered in hospitals. In this group of homebirths, hyperbilirubinemia (52.4%) and undernourishment (35.1%) was most common among those delivered by nurses and least among those delivered by doctors (data not shown). Infants delivered at TMHs were not at any significant risk of unfavorable hearing screening outcomes (refer or follow-up default) as well as severe hyperbilirubinemia requiring EBT than infants delivered in hospitals or family homes.

Table 4 Infant Factors Associated With Place of Delivery in Lagos, Nigeria

	Herbal homes	Herbal homes	Family homes
	versus hospital	versus family homes	versus hospital
Profile	OR (95% CI)	OR (95% CI)	OR (95% CI)
Gestation age	p = .016
Full-term (≥37 weeks)	1.0	1.0	1.0
Preterm (<37 weeks)	0.48 (0.26–0.87)*	0.73 (0.26–2.02)	0.65 (0.25–1.67)
Any undernourishment [a]		p = .002	p < .001
No	1.0	1.0	1.0
Yes	1.03 (0.90–1.18)	0.70 (0.56–0.87)**	1.47 (1.19–1.84)***
Jaundice with phototherapy	p = .027	p < .000	p = .015
No	1.0	1.0	1.0
Yes	0.61 (0.39–0.94)*	0.34 (0.20–0.59)***	1.77 (1.12–2.81)*
Jaundice with EBT
No	1.0	1.0	1.0
Yes	0.69 (0.31–1.54)	0.91 (0.34–2.44)	0.76 (0.33–1.73)
Hearing screening outcome
Pass	1.0	1.0	1.0
Refer	1.09 (0.66–1.78)	0.70 (0.36–1.37)	1.55 (0.79–3.04)
Incomplete	1.13 (0.85–1.50)	1.34 (0.82–2.19)	0.84 (0.52–1.36)
*p < .05; **p < .01; ***p < .001, Missing data: [a] = 204 (3.0%).
EBT = exchange blood transfusion; herbal homes = traditional maternity homes. Model adjusted for all maternal factors with statistical significance (p < .005).

DISCUSSION

We set out to examine the adverse perinatal outcomes among infants delivered in TMHs compared with those delivered in hospitals or residential homes against the backdrop of the continuing, inevitable but limited role of TBAs even in urban areas in developing countries. Although the services provided by TBAs are not confined to TMHs especially in rural areas (Bazant, Koenig, Fotso, & Mills, 2009; Izugbara, Ezeh, & Fotso, 2009; Jokhio, Winter, & Cheng, 2005; Sibley et al., 2007; Thind, & Banerjee, 2004), the relatively small proportion of TBAs involved with home delivery in this study would appear to make a distinction between TBAs and TMHs in this population immaterial. Hence, our study, derived from a secondary analysis of existing data, essentially complements earlier reports from this population that examined maternal preference for private hospitals compared with public hospitals among women seeking facility-based obstetric services (Olusanya, Roberts, Olufunlayo, & Inem, 2010) and, more broadly, maternal preference for non facility-based delivery with or without skilled attendants at delivery (Olusanya, Alakija, & Inem, 2010).

One principal finding is that the offspring of women who deliver in TMHs assisted by TBAs were unlikely to be at greater risks of undernourished physical state, severe hyperbilirubinemia, and sensorineural hearing loss compared with infants delivered in hospitals or residential homes. This is quite notable considering that delivery attended by trained and untrained TBAs or other unskilled attendants is more commonly associated with maternal and perinatal mortality necessitating concerted shift and global priority toward skilled attendants even for home delivery, especially where hospital-based intrapartum care is not readily available or accessible (Campbell, Graham, & Lancet Maternal Survival Series Steering Group, 2006; Koblinsky et al., 2006). In contrast, infants delivered in family homes were more likely to be undernourished and at risk of severe hyperbilirubinemia.

The observed perinatal outcomes may not be unrelated to the transition in practices regarding maternal and immediate newborn care by the TBAs. The beliefs and practices of TBAs predominantly are geared toward safe delivery and the health of the mother and have been extensively reported (Ahmed et al., 2005; Fronczak Arifeen, Moran, Caulfield, & Baqui, 2007; Itina, 1997; Lefèber & Voorhoever, 1997; Sreeramareddy, Joshi, Sreekumaran, Giri, & Chuni, 2006). For example, potentially harmful practices commonly reported in this population in the past such as the use of unsterilized scissors or recycled blades for cord cutting, which predisposes to infection/illness in the newborn, and dressing of umbilical cord with mentholated powder, which has been implicated as a significant cause of severe hyperbilirubinemia in this population, were rare (Ransome-Kuti, 1986). This may be attributed to the impact of training for TBAs and public health promotion by local health authorities and nongovernmental organizations over the years. In fact, most of the TBAs in this community belong to an association that promotes self-regulation and periodic accreditation for its members with support from the local health authorities apparently aided by geographical proximity. Aspects covered by such training typically include elements of clean and safe delivery, identification of potential obstetric complications, and timely referral to the hospital. It is pertinent to mention that it is common practice for TBAs to refer/refuse high-risk pregnancies such as first-timers, those with prior cesarean section or preterm delivery, HIV-positive individuals, and those with placenta previa in order to forestall bad publicity from possible unfavorable outcomes thereby unduly compromising their historical claim to safe motherhood (Izugbara, Ezeh, & Fotso, 2009). Moreover, TBAs always encourage women to take their babies to public hospitals immediately after delivery for routine examination and immunization, as they recognize their limitations in providing appropriate neonatal services beyond the first bath, weighing, and basic thermal care to keep the baby warm.

In contrast, infants delivered in health facilities especially in private hospitals have been found to be at a greater risk of severe hyperbilirubinemia as previously reported due to the fact that most hospitals lack in-house facilities for monitoring bilirubin levels, phototherapy, and EBT thus relying more often on visual estimation of the severity of neonatal jaundice (Olusanya, Roberts, Olufunlayo, & Inem, 2010; Olusanya, Akande, Emokpae, & Olowe, 2009). This practice often is exacerbated by the inappropriate but common use of prophylactic antibiotics as the first-line management before referral to a tertiary center. A high proportion of undernourishment in this age group usually is attributed to fetal growth restriction caused by maternal malnutrition underpinned by poverty. Poorer women are more likely to engage in inappropriate infant feeding practices and deliver at home rather than TMHs or the hospital because of financial constraints. Home and TMH delivery was associated with increased odds of hearing screening referral but was not statistically significant. This observation needs to be interpreted cautiously, however, in the light of lack of data on the diagnostic outcomes after referral and the high rate of follow-up default in this population (Olusanya, Wirz, & Luxon, 2008).

Comparison of the proportion of deliveries by TBAs in this inner-city setting with other urban-based studies was hampered by the methodological differences. For example, Bazant and colleagues (2009) reported that 10% (193/1,926) of their study population in urban Kenya delivered in TMHs but did not specify the proportion of their home births (n = 467) assisted by TBAs (Rahman, Tarafder, & Mostofa, 2008). This would have provided better perspective on the higher proportion (24.2%) of home delivery in their study compared with 6.8% in our study. Similarly, while a significantly higher proportion (61.2%) of TBA-assisted deliveries was reported in urban Bangladesh, the proportion attributable to TMH was unknown, although the vast majority was likely to be home delivery, as only 22% of births in urban areas are hospital based (Rahman et al., 2008). In another study from urban Nepal, the authors reported a rate of 19% (62/375) for home delivery and 81% for facility-based delivery but provided no information on TBA-assisted delivery outside the family homes (Bolam et al., 1998). It was reported that TBAs were less commonly used, however, as only 4% of home deliveries were assisted by TBAs compared with 46% in a periurban area.

The high mean delivery fee charged by doctors is to be expected, as their services are often needed for complicated deliveries that more often require surgical intervention. The mean charge by TBAs, however, is not substantially different from trained nurses or midwives' fees for straightforward deliveries. This observation would suggest that the reasons why many women in this inner-city setting without apparent geographical and financial barriers prefer to deliver in TMHs with less attractive ambience are most likely attributable to sociocultural factors including religious/superstitious beliefs as well as perceived interpersonal relationship for the care provided by TBAs (Campbell et al., 2006; Izugbara et al., 2009). For example, Izugbara and colleagues in an in-depth survey among a group of TBAs in urban Kenya identified several factors applicable to many developing countries for the continued attraction for TBAs such as the responsiveness of their services to the sociocultural and economic sensitivities of the women, which goes a long way in developing trust and confidence in their services (Izugbara et al., 2009). The services of TBAs are not limited to delivery, but include nutritional counseling during pregnancy, advice on family planning and contraceptives, menstrual problems, infant circumcision, earlobe piercing, as well as special support for unmarried women and financial accommodation for installmental payments (Izugbara & Ukwayi, 2003). Women also choose to go to TMHs in order to avoid some practices in hospitals such as requirement for the donation of blood, fear of cesarean section (whether indicated or not), rigorous dietary requirements, as well as disrespectful and shameful comments from caregivers especially in public hospitals. In contrast, some women will deliver at home for a variety of economic and noneconomic reasons including convenience, privacy, cost of hospital delivery, lack of escort during labor, and precipitate labor (Sreeramareddy et al., 2006).

The maternal predictors of place of delivery reported in this study accord with earlier findings from sub-Saharan Africa and other developing countries (Bazant et al., 2009; Fronczak et al., 2007; Gabrysch & Campbell, 2009; Rahman et al., 2008; Stephenson et al., 2006; Thind & Banerjee, 2004). The use of herbal drugs in pregnancy was rarely reported in other studies, while antenatal care commonly reported in other studies was not a predictor in this population. Teenage mothers were more likely to use TMHs because of the stigma and embarrassment they were likely to face with caregivers in hospitals especially as unmarried women, while older women

(>35 years) were likely to choose hospital delivery rather than TMHs because of their obviously high-risk status. Non-Yoruba were less likely to use TMHs in this predominantly Yoruba community possibly because of ethnic differences in delivery practices or even discrimination without ruling out the fact that some of these women may behave otherwise in their own cultural settings. Privacy is a universal norm with Muslim women, and this may account for the increased odds of delivery in TMHs compared with their Christian counterparts, consistent with other reports (Thind & Banerjee, 2004). There are multiple often interrelated pathways through which maternal education, occupation, and social class affect health-seeking behavior that are well covered in the literature (Gabrysch & Campbell, 2009; Stephenson et al., 2006). Low level or lack of education, lack of gainful employment, and low social class lead to poor knowledge, awareness, and financial resources, and work in tandem in reinforcing cultural disposition to TMHs. The higher odds associated with use of herbal drugs is not surprising because the services offered by TBAs are firmly rooted in local culture and beliefs. Moreover, TBAs are unable to dispense modern medications otherwise provided in hospitals due to regulatory restrictions, besides the fact that often they lay claims to the superiority of herbal preparations especially when an ailment is attributed to supernatural causes (Izugbara et al., 2009). The harmful side-effects of these herbal medications without standardized/regulated dosages, however, are seldom documented. The lack of correlation between antenatal care and place of delivery may result from the fact that even women who chose to deliver at home had good access to antenatal services from either TBAs or skilled attendants, unlike women in rural areas. It was also possible that the lack of information on the number of antenatal visits resulted in this factor not being sufficiently discriminatory to confirm some other reports that found that insufficient antenatal care was significantly associated with the use of TBAs (Bazant et al., 2009; Gabrysch & Campbell, 2009; Rahman et al., 2008; Thind & Banerjee, 2004). Overall, it is uncertain if, when, or the extent to which current international efforts to scale up skilled attendants at delivery will optimally address the apparently intractable structural, economic, and cultural barriers in a country like Nigeria, which is a leading contributor to the global and regional burden of maternal and child mortality. The evidence from our study would seem to underscore the “transitional” need for trained TBAs where necessary.

Barring the effects of any residual confounding, our study has highlighted rarely reported outcomes among survivors based on the place of delivery or the person assisting at delivery especially in relation to TBAs (Sibley et al., 2007), but prospective studies still are warranted due to the inherent limitations of the retrospective design. Such studies will need to cover a wider range of perinatal outcomes that is feasible in a nonhospital community setting especially those that portend long-term or developmental risks among survivors. It was uncertain whether the results of this study could be generalized to other developing countries considering the variations in the delivery of the services by TBAs in TMHs or the residential homes of their clients and the possibility of selection bias as a result of those who failed to attend the immunization clinics due to neonatal death or other factors. Although trained TBAs often perform better than untrained TBAs, there were no data on how many of the TBAs received training or the full content and duration of such training. Thus, it was not possible to relate the outcomes in this study more objectively among the trained TBAs against the backdrop of the long-standing debate and conflicting evidence in the literature from other developing countries on the effectiveness of such training (Jokhio, Winter, & Cheng, 2005; Sibley et al., 2007; Sibley & Sipe, 2006). Nonetheless, the findings in this study seem to reinforce the renewed call to train and engage TBAs as key partners in maternal health care delivery as an “interim” strategy even in urban areas against the backdrop of the imminent urban transition in many developing countries, particularly in sub-Saharan Africa, and the prevailing multidimensional barriers to optimal scale up of skilled attendants at delivery. It is important regardless still to take appropriate steps to facilitate a reduction in the demand for TBA services with increasing availability of skilled birth attendants.

CONCLUSIONS

Several studies on the impact of TBAs on maternal and child health in developing countries focus on maternal and perinatal mortality especially in rural areas and rarely provide information on the perinatal profile of the surviving newborns. We have shown in this study that whereas infants delivered in family homes were more likely to be undernourished and at risk of severe hyperbilirubinemia, infants delivered in TMHs exclusively managed by TBAs in an urban setting were unlikely to be at greater risks of undernourished physical state, severe hyperbilirubinemia, and sensorineural hearing loss compared with infants delivered in hospitals or residential homes. Although the potential contribution of training for some of the TBAs could not be ascertained objectively, the evidence in this study points to the continuing relevance of TBAs in maternal and child health in some urban areas regardless of the availability and access to better equipped private and public hospitals. Given the retrospective design of the current study, prospective studies are necessary to determine more broadly and objectively the range of adverse perinatal outcomes associated with the activities of TBAs among survivors to complement the vast and growing evidence on maternal and perinatal mortality in the developing world.

Acknowledgments

The authors are grateful for the kind support of all the staff and management of the health centers used in this study, as well as all the women who participated in the wider pilot study on universal infant hearing screening in Lagos, Nigeria, formally launched by the Federal Health Minister and the Lagos State Health Commissioner. The authors are also grateful to the Education Trust Fund (ETF), an agency of the federal government of Nigeria, for educational materials for all participating women; and Hearing International Nigeria, a local charity, for funding support for this project. None of the sponsors were involved in the study design, collection, analysis and interpretation of data; the writing of the manuscript; or the decision to submit these results for publication.