Introduction

Access to emergency obstetric and neonatal care (EmONC) and skilled birth attendance are important elements of strategies to reducing maternal and neonatal mortality [1,2,3]. The fifth WHO Millennium Development Goal (MDG5) aimed to reduce maternal mortality by 75% by 2015 but made the least progress among the MDGs, that is 45% was achieved from 1990 levels of 543,000 maternal deaths [4]. Globally, 289,000 women died from pregnancy related complications in 2013 [4], half of them due to haemorrhage, hypertensive disorders and sepsis [5]. The majority occurred just before, during or after birth [6], signifying that the need for skilled birth attendance and access to emergency obstetric and neonatal care had not been met by far.

EmONC refers to life-saving services for maternal and neonatal complications provided by a professional in a health facility as described in Fig 1 [7,8], and are essential in reducing maternal and neonatal mortality [2]. However, a large proportion of maternal deaths occur in hospitals compared to home [9]. This is related to delays in accessing skilled care as many arrive at a facility in critical condition or more women with complications are referred to hospitals, but also delays in receiving quality care after arriving at the health facility [9].

Download:

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

Fig 1. Basic and Comprehensive Emergency Obstetric and Neonatal Care.

https://doi.org/10.1371/journal.pone.0145196.g001

Monitoring short-term improvement in maternal health using maternal mortality ratio is challenging in low-income countries due to under-reporting, misclassification of maternal deaths, lack of resources to conduct large surveys, and census data are only available every ten years [10]. Output and process indicators have been proposed for monitoring interventions aimed at reducing maternal mortality [7,8]. One such indicator, the unmet obstetric need, quantifies to what extent emergency obstetric care is provided [11]. Thus, it measures the discrepancy between what should be done to deal with obstetric complications and the care actually provided [11].

The unmet obstetric need approach has been used to monitor responsiveness by the health system in emergency obstetric care in resource-limited areas [12,13]. Combining process indicators such as availability, accessibility and functionality of EmONC services, proportion of births in obstetric care facilities, met need for emergency obstetric care, caesarean section rate and direct obstetric case fatality rate is recommended to strengthen the assessment of maternal care services [14]. The Unmet Obstetric Need (UON) concept, developed by the UON Network [11,15], considers unmet need by including major obstetric interventions that are done for absolute maternal indications and the gap between those expected and those actually done. Process indicators do not focus on adequacy of bed capacity, competence and number of health care providers in health facilities. These become necessary as part of improvement to care when deficits are identified.

Zambia had an estimated maternal mortality ratio of 398 per 100,000 live births in 2014 [16]. Estimate from another source has not been included here as the DHS report so far presents estimates for comparison across countries. Although this may not be the most accurate considering the sisterhood method, challenges in reporting maternal deaths and large confidence intervals in the estimates, it is the most reliable for Zambia at present. Though 67% of women utilize health facilities at childbirth [16], urban and rural disparities exist. Our previous study in three districts of Kenya, Tanzania and Zambia revealed less health facility utilization at childbirth among women in rural than urban areas and in women of low socioeconomic status [17]. However, it is relevant to know the extent to which this low utilization encompasses women with life-threatening conditions and the difference in access to life-saving services between rural and urban areas. In Kenya, caesarean section rates were found to be much lower in rural than in urban settings [18]. In addition, poor attitudes towards provision of prompt attention and respectful care by health providers discourage women from seeking care at facilities [19]. Although underutilization of health facilities at childbirth in Zambia is known, less is known of the magnitude of unmet obstetric needs. We aimed to assess and compare the unmet need for life-saving obstetric services in rural and urban areas in a Zambian district by using the unmet obstetric need approach. Specifically, we aimed to obtain information on major obstetric interventions performed. Secondly, we aimed to assess the functionality of EmONC health facilities in rural and urban areas.

Methods

Data Quality

Reliability of the data depends on reproducibility, which implies the ability of health care providers to notify the same complications in the same way. This includes whether the same health care provider (test-retest intra-rater reliability), and other health care providers (inter-rater concurrent reliability) notify complications in the same way. Validity of the data depended upon how reliable the recorded diagnoses were made, such as cephalo-pelvic disproportion which could have low specificity. Cephalo-pelvic disproportion is a common and accepted diagnosis and may be used by default in less clear conditions. However, a gauge of reliability and validity of data poses challenges when dealing with women who pass quickly through the system of care providers. On data completeness, it was possible that some women could have utilized other hospitals neighbouring Kapiri Mposhi district, although the geographical distribution of other hospitals made self-referrals less likely. Whereas Kabwe General Hospital was about 50km from Kapiri Mposhi, Ibenga Mission Hospital and Mpongwe Mission Hospital were 94km and 99km away, respectively, and public motor transport was not easily available for the local community (Fig 2).

Download:

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

Fig 2. Distribution of health facilities with and without EmONC in Kapiri Mposhi, 2010.

https://doi.org/10.1371/journal.pone.0145196.g002

Data obtained from registers mainly agreed with those recorded in case records, such that there was no reason to query the quality of information in the registers.

Information on functionality of basic and comprehensive EmONC facilities was also obtained. The rationale for this was based on knowledge that to reduce maternal deaths, certain obstetric functions in facilities must be available. Certain health facilities may be categorised as EmONC by the health system, yet if conditions are not met this may hide inadequacy in coverage of services. Fig 1 shows signal functions that describe basic and comprehensive EmONC facilities. A standard tool was used to obtain information on functions performed at least once in the previous three months prior to the survey by interviewing the nurse-in-charge of the maternity ward [8]. Reasons for not performing any of the signal functions were recorded. Inspection for availability of equipment and drugs in maternity wards was done. The WHO defines a basic EmONC facility as one that performed all seven signal functions in the last three months prior to a survey [8]. A comprehensive EmONC facility performs caesarean section and blood transfusion plus the seven functions in basic EmONC in the last three months. However, if a function such as assisted vaginal delivery is systematically absent in an area due to policy, the functioning of the facilities is indicated as “basic EmONC minus 1” or “comprehensive EmONC minus 1” [8].

Results

Characteristics of women with childbirth in EmONC facilities in urban and rural areas

A total of 2114 women in urban and 1226 in rural areas had childbirth in EmONC facilities (Table 1). Median age (inter-quartile range—IQR) was 24 (20, 30) years for both urban and rural areas. Primigravida constituted 32% and 25% of women in urban and rural areas respectively (Table 1). More women in rural areas had higher parity (four or more) compared to urban areas, 33% vs. 22% respectively. Seventy-one percent of women in the urban were from within the catchment area for the study facilities compared to 30% in the rural areas (Table 1). Caesarean section was performed on 3.4% of deliveries in both urban and rural areas (Table 1). Missing information on delivery mode was low, 1.8% in rural and 1.5% in urban areas, p = 0.534, rate ratio 1.19 (95% CI 0.69–2.06).

Download:

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

Table 1. Characteristics of women who had facility childbirths in EmONC facilities in Kapiri Mposhi district and referrals from Kapiri Mposhi to Kabwe general hospital, 2010.

https://doi.org/10.1371/journal.pone.0145196.t001

Complications observed during childbirths in the facilities

Cephalo-pelvic disproportion was the most common indication for a major obstetric intervention, 54% and 33% of all complications, in urban and rural areas respectively (Table 2). 3% of women presented with post-partum haemorrhage (Table 2). 7.1% of pregnant women from Kapiri Mposhi Urban Hospital were referred due to complications, whereas Mukonchi Rural Health Centre (RHC) referred 5.3%, Mpunde RHC 3.6%, Nkole RHC 2.6%, and 1.7% from St.Paul’s RHC. Women were referred directly to the comprehensive EmONC general hospital, except Nkole RHC that referred to the urban hospital due to proximity unless a major obstetric intervention necessitated referral to the general hospital (Fig 3).

Download:

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

Table 2. Frequency and proportions of complications in pregnant women who were referred to Kabwe general hospital from urban and rural EmONC facilities in Kapiri Mposhi, 2010.

https://doi.org/10.1371/journal.pone.0145196.t002

Download:

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

Fig 3. Flow of women with complicated institutional deliveries from basic to comprehensive EmONC facility in Kapiri Mposhi in 2010.

https://doi.org/10.1371/journal.pone.0145196.g003

Major obstetric interventions for absolute maternal indications

EmONC facility childbirths constituted 81% of expected births in urban and 16% in rural areas (Table 3). Based on the assumption that 1.4% of women would be in need of life-saving intervention [11], we calculated 142 women should have been referred for a major obstetric intervention, but only 84 had actually been treated. The overall unmet obstetric need in the population was 58 out of 142 women or 41% (95% CI 33–49%), in which the deficit from rural areas was estimated to be 77 out of 106 women or 73% (95% CI 71–75%). This implied that 77 women in the rural areas in 2010 in need of a life-saving intervention did not receive one and probably died or had severe morbidity. No deficit was observed in the urban area. The negative deficit implied that more major obstetric interventions for absolute maternal indications were actually performed than expected. This may reflect the possibility of some rural women providing a temporary address of residence that was close to the urban health facility. Major obstetric interventions for absolute maternal indications were 2.1% (95% CI 1.60–2.71%) of expected deliveries in the urban hospital, i.e. above the reference value of 1.4%, and 0.38% (95% CI 0.27–0.55%) in rural facilities (Table 3), implying unmet needs for women in rural areas. The rate ratio of major obstetric intervention for absolute maternal indication in urban to rural areas was 5.5 (95% CI 3.55–8.76).

Download:

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

Table 3. Major obstetric interventions (MOI) for absolute maternal indications (AMI) referred from urban hospital and rural EmONC centres in Kapiri Mposhi to Kabwe general hospital, 2010 (women aged 12–48 years).

https://doi.org/10.1371/journal.pone.0145196.t003

Discussion

Our study revealed a substantial deficit in the use of emergency obstetric services. However, it was the rural population that suffered the entire deficit, which was 73% implying that 77 out of 106 women had unmet needs in 2010. Further, the likelihood of undergoing a major obstetric intervention was 5.5 times higher in women from urban compared to rural areas, which underscores the unacceptably large inequity between urban and rural societies. In the rural area, only one health facility met the criteria of “basic EmONC minus 1”. The suffix “minus 1” was used due to absence of assisted delivery by vacuum or forceps in all rural facilities. The health policy did not recommend midwives to perform the procedure in health centres. Three rural health facilities did not qualify as “basic EmONC minus 1” since they lacked equipment to perform removal of retained products of conception.

Rural health facilities had few deliveries, 13–45 deliveries/month, and so could raise concern about skills and experience of the health providers [25], or trust from the community. Some facilities had not received and administered parenteral anticonvulsants for eclampsia in the three months prior to the study. This could reflect low utilization of health facilities for childbirth partly associated with constraints of long distance [17,26], and lack of trust due to perceived poor quality of care [19]. Our finding of 71% of women in the urban area residing within the catchment area compared to 30% in the rural area reflects that a higher proportion of women in the rural area probably have less geographical access to EmONC services than in urban areas. Furthermore, 16% of expected births in the rural and 81% in urban areas occurred in health facilities that were supposed to provide basic EmONC. Our previous study in the same area found 37% pregnant women in rural and 77% in urban areas had facility utilization during childbirth [17]. This could indicate that most institutional births in rural areas occurred in non-EmONC facilities, or possibly by-passed to urban facilities.

The possibility of by-passing rural facilities and providing a temporary urban address could not be ruled out, although most women who received a major obstetric intervention were traced within the health referral system. By-pass has been associated with perceived poor quality of care due to lack of emergency obstetric care services in other countries [27]. In Morocco, assessing unmet obstetric need at district level lead to improved quality of care and increased utilization of facilities [12]. Tanzania, Rwanda and Ethiopia also improved utilization of health facilities by women and achieved reduced mortality after upgrading infrastructure and technical competence of their health providers [28].

The substantial rural-urban disparity demonstrated in the present analysis of unmet obstetric needs should alert policy makers and the health care system to improve life-saving services to rural areas. The situation highlighted in this study could have changed due to opening of a district hospital in the study setting after our data collection. However, the extent of improvement especially in the rural area can only be ascertained by a follow-up study of the present situation. Although Kapiri Mposhi district embraced accountability and equity in priority setting, decisions were constrained by limited resources [29]. Sharing ambulance service among basic EmONC facilities located far apart was a challenge for prompt referrals. Using the unmet obstetric need approach, Mali identified problem areas and embarked on a nation-wide improvement in communication and transport, reduced financial barriers and was able to increase institutional deliveries and management of obstetric complications [3]. A greater reduction in maternal deaths was observed in women with haemorrhage, whose survival depended on prompt management and referral [3]. Urban and rural inequity could also be related to socioeconomic differentials, particularly in sub-Saharan Africa where extremely low caesarean section rates are observed among the very poor [30,31].

Population differences between rural and urban areas worsen the rural-urban inequity. Rural areas in Zambia have a higher percentage of women who begin childbearing between 15–19 years compared to urban areas, 36% vs. 20% [16]. This difference would contribute to a higher need for emergency obstetric care in rural areas since women aged 15–19 years are twice as likely to die from causes related to pregnancy and childbirth compared to women aged 25–29 years (Banda, R. et al., submitted), [32,33]. Although our data indicated a higher proportion of primigravida that utilized facilities in the urban than rural areas, this is not likely to reflect the actual differences due to the large urban-rural differential in facility childbirth and thus leading to selection bias.

The commonest indication for major obstetric intervention was cephalo-pelvic disproportion. This was consistent with findings in the region [18,34]. Furthermore, cephalo-pelvic disproportion may be used by default when other causes of prolonged labour have not been properly explored such as dystocia, mal-presentation or inadequate uterine contraction. The levels of caesarean section observed in the urban area were much lower than the suggested 5–15% reasonable to prevent both maternal and foetal complications [8]. Other countries such as China and Brazil are concerned with high caesarean section rates from increased provision of services and nearly universal hospital delivery [35] and demand from women [36]. It was unlikely that women in our context demanded caesarean section. Our study population culturally viewed difficult childbirth and operative delivery as related to marital infidelity and normal childbirth was a sign of strength [19]. In addition, facility childbirth was mostly in public facilities unlike private facilities which have been associated with high caesarean rates [31].

Despite cephalo-pelvic disproportion being the commonest complication, the most common direct cause of maternal mortality in sub-Saharan Africa is post-partum haemorrhage [5,37]. The finding of few women with life-threatening post-partum haemorrhage and ruptured uterus in our study may indicate either poor access to health services such that women died before reaching facilities, or they benefited from prevention of post-partum haemorrhage by active management of third stage of labour and timely referral. The deficits and rates of major obstetric intervention for absolute maternal indications observed in this study were similar to other UON studies in the region [13,38,39].

There were some methodological issues that merit mention regarding accuracy and completeness of facility records. Evaluating maternal health services by use of process indicators in order to make improvements depend very much on accurate records in health facility registers. The reliability may suffer when records are incomplete or duplicate, a danger when using several registers. In this study we attempted to avoid double counting by cross-checking names and file numbers recorded in both admission registers and delivery ward registers during data collection. Incomplete information in delivery registers was followed up by further checking case records. It was also important to have proper records of origin of residence to compare unmet needs between urban and rural areas [38]. Details for women who were referred to the comprehensive EmONC facility were obtained from both the basic EmONC and referral facility, which helped to identify residence of origin. When more than one indication for an intervention was recorded, the absolute maternal indication was utilized for the analysis.

A limitation of our study was the exclusion of abortions and ectopic pregnancies when assessing absolute maternal indications. Although abortions and ectopic pregnancies are important causes of maternal mortality, they may strictly speaking not be appropriate indicators of obstetric care. Abortions are also prone to inaccuracies in reporting. Complications may be over-reported from incorrectly reported incomplete abortions. In our study most women with abortions were managed at basic EmONC facilities and rarely presented with severe haemorrhage or sepsis. Therefore, excluding them would not have affected the estimates of absolute maternal indications requiring major obstetric interventions. Another limitation is that we did not consider quality of care in facilities, such as delays from admission or development of complications before referral to a comprehensive EmONC facility. Further, the diagnosis of cephalo-pelvic disproportion as an absolute maternal indication may be subject to misclassification. Some women may be referred for a ‘borderline’ cephalo-pelvic disproportion but turn out to have an uneventful delivery. This might bias the results towards a low ratio of major obstetric intervention for absolute maternal indication. Midwives diagnosed cephalo-pelvic disproportion by both clinical assessment and the use of a partogram as a tool for monitoring progress of labour, which is adequate.

Despite these limitations, the unmet obstetric need approach has been found to be useful in monitoring health system performance of maternal health services. This approach on its own may not identify which level in the system needs improvement. Therefore, combining assessment of services in facilities becomes important. In this respect, the approach to priority setting in the REACT study focusing on fairness and inclusiveness from different stakeholders may be useful [19]. The present results of rural-urban inequities in the district may be an eye-opener for both local and national authorities. Upgrading non-EmONC to basic EmONC facilities could improve services in rural areas and providing an ambulance to each basic EmONC health facility could ease referral and transfer of women with complications.

Although facility data are limited in generalizability due to population variation in use of services, the results of this study may be extrapolated to districts with similar context in access to and functionality of EmONC facilities.

Conclusion

This study reveals a considerable unmet obstetric need in rural Zambia with a striking rural-urban inequity both in use of EmONC facilities and major obstetric interventions. The likelihood of having a life-saving intervention was 5.5 times higher for urban than rural women. This should alert health policy makers to improve access to life-saving services in rural areas.

Acknowledgments

This study was made possible through funding and support from the European Union Sixth Framework Programme (INCO-2003-A.1.2, contract PL517709) for the Specific Targeted Research and Innovation Project REACT—Response to ACcountable priority setting for Trust in health systems. Full title: Strengthening of fairness and accountability in priority setting for improving equity, and access to quality health care at district level in Tanzania, Kenya and Zambia. We thank the Norwegian government through its Quota Scheme for financial support. Special thanks go to Kapiri Mposhi District Health Management Team Office for the support during data collection, and REACT team members who were involved in planning of the study.