Maternal mortality in Northern Nigeria: a population-based study

Maternal mortality in Northern Nigeria: a population-based study

European Journal of Obstetrics & Gynecology and Reproductive Biology 109 (2003) 153–159 Maternal mortality in Northern Nigeria: a population-based st...

150KB Sizes 0 Downloads 22 Views

European Journal of Obstetrics & Gynecology and Reproductive Biology 109 (2003) 153–159

Maternal mortality in Northern Nigeria: a population-based study Yusuf M. Adamua, Hamisu M. Salihub,*, Nalini Sathiakumarc, Greg R. Alexanderb a

Department of Geography, Bayero University, Kano, Nigeria Department of Maternal and Child Health, University of Alabama at Birmingham, Birmingham, AL 35294, USA c Department of Epidemiology and International Health, University of Alabama at Birmingham, Birmingham, AL 35294, USA b

Received 11 September 2002; received in revised form 6 December 2002; accepted 12 December 2002

Abstract Objectives: To determine the incidence and causes of maternal mortality as well as its temporal distribution over the last decade (1990– 1999). Study design: All maternal deaths recorded within the study period in the State of Kano, Northern Nigeria, were analyzed. Maternal mortality ratios (MMR) were computed using the Poisson assumption to derive confidence intervals around the estimates. A non-linear regression model was fitted to obtain the best temporal trajectory for MMR across the decade of study. Results: A total of 4154 maternal deaths occurred among 171 621 deliveries, yielding an MMR of 2420 deaths per 100 000. Eclampsia, ruptured uterus and anemia were responsible for about 50% of maternal deaths. Conclusion: We found one of the highest maternal mortality ratios in the world. Maternal mortality could be reduced by half at study site with effective interventions targeted to prevent deaths from eclampsia, ruptured uterus and anemia. # 2003 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Maternal mortality; Population-based study; Eclampsia

1. Introduction Over 99% of the annual global estimates of 585 000 maternal deaths occur in developing countries, and over half of these occur in sub-Saharan Africa [1]. A woman in sub-Saharan Africa is 75 times more likely to die as a result of this than a woman in Europe or North America [1,2]. On a risk per birth basis, the countries with the highest levels of maternal deaths (using maternal mortality ratios as index) are all in Africa; the world’s top 11, with MMRs of 1300 or greater are all in Africa, with Rwanda heading the list [1]. This huge burden of maternal deaths makes the reduction of maternal mortality in African countries a global priority. However, in order to formulate effective global strategies that can successfully curtail this immense problem, it is essential to have reliable estimates of the dimensions of maternal deaths in sub-Saharan Africa. Measuring maternal mortality is difficult and complex, especially, in these countries due to numerous socio-economic, cultural and infra-structural barriers. Nevertheless, there has been tremendous progress in the quality of data from these countries due to the development of new ways of measuring maternal mortality [2,3].

*

Corresponding author. Tel.: þ1-205-934-6469; fax: þ1-205-934-8248. E-mail address: [email protected] (H.M. Salihu).

In this contribution, we report maternal mortality estimates from the Northern part of the most populous African nation, Nigeria. A number of features makes this study unique; firstly, it is one of the rare population-based studies on maternal deaths from the African continent. Secondly, the existing old health-care administration zoning system in place assures that almost all deliveries and maternal deaths are routinely recorded and ascertained by the zonal health offices that cover the various administrative units in the state. For this reason, both the numerator (maternal deaths) and denominator (deliveries) for the computation of maternal mortality ratios are minimally biased by the effect of underreporting. Finally, and perhaps most importantly, the level of maternal deaths found in this study is to our knowledge the highest in the world (slightly higher than that of Rwanda; the country credited with the worst maternal mortality ratio on the planet).

2. Materials and methods 2.1. Background of study site The study was carried out in Kano State, one of the 36 states of Nigeria and the second most populous with a population of about 5.6 million people. The State lies within

0301-2115/03/$ – see front matter # 2003 Elsevier Science Ireland Ltd. All rights reserved. doi:10.1016/S0301-2115(03)00009-5

154

Y.M. Adamu et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 109 (2003) 153–159

the Sudan Savannah belt and is bordered to the south by the Guinea Savannah. It has two main seasons; the rainy and dry seasons. The indigenous people of the area are mainly Hausas and Fulanis, with other ethnic groups from various parts of the country concentrated mainly in the urban centers. Within the state, there are 9 specialist hospitals, 10 general hospitals, 7 cottage hospitals, and 592 primary care institutions consisting of 16 primary health centers, 85 primary health clinics, 243 dispensaries, 194 basic health posts and 54 leprosy clinics. There are also over 170 private health facilities [4]. The state counts 180 physicians, 4 dentists, 42 pharmacists, 1800 nurses and midwives. Most of the secondary and tertiary referral facilities and staff are concentrated in urban zones of the state while the rural areas are served with primary health facilities and trained healthcare workers and technicians. Life expectancy is currently estimated at 48 years for men and 50 for women [5]. The crude birth rate is about 46 per 1000 and the crude death rate is circa 16 per 1000. Infant mortality rate is 100 per 1000 [5]. Data on maternal mortality rate (MMR) or ratio for this study area are surprisingly unavailable, reflecting either a lack of interest or awareness of the importance of this index on the part of health administrators in the area.

are reported to the village or local government councils. These councils forward the vital statistic information to the zonal office in charge of that area. It should be noted that these activities are constantly being conducted and verified because they form the basis on which statewide political and economic decisions are taken, and the tendency is that each local government council makes sure that vital event that occurs in its area of jurisdiction is reported for budgetary purposes. A zonal health office oversees a number of hospitals, groups of primary health-care centers, private health institutions and local government councils that are considered to be proximal to each other geographically. The Zonal Health Office serves as a filter to detect inconsistencies, missing information and other errors in the records submitted from its catchments area. It serves not only as a relay station but also, and more importantly, as a quality control unit. Once the data have been certified there, they are then forwarded to the Research and Statistics Department at the Kano State Ministry of Health for policy-making purposes. The data analyzed in this study were obtained from this source at the ministry. 2.3. Statistical analysis

2.2. Study design and data collection This was a retrospective study using information contained in the vital statistics register maintained by the Research and Statistics Department of the Ministry of Health in Kano State. There is in place within the State, a system of data collection on all cases of deliveries and maternal mortality that occur in public health institutions. At the level of hospitals, comprehensive health centers and primary health facilities, there is a record documentation unit denoted officially as ‘‘Record and Statistics Unit’’ responsible for the collection and documentation of data on hospital admissions, diagnoses, and discharges, including deliveries and maternal deaths. In large and tertiary health facilities, a special unit attached to the maternity wards is responsible for documenting all labor ward admissions, deliveries, complications, discharges, maternal and perinatal deaths. When any of these events occur, the staff from this special unit or the Records and Statistics Unit abstracts information from the record charts on the wards onto a register kept permanently at the unit. For the special case of mortality, there is a register code-named ‘‘mortality register’’ that contains data on all deaths that have occurred stratified by period of occurrence. Information abstracted relate to socio-demographic characteristics of the deceased, cause of death as certified by the attending physician, date of admission, time of delivery and accompanying complications. On a regular basis, copies of these records are relayed to the Zonal Health Office responsible for that area. The zonal offices also receive compiled quarterly reports of deliveries and deaths (maternal and infant) that occur in private health facilities. Births and deaths that occur at home

Maternal mortality ratio (MMR) was computed by dividing maternal death counts by the total number of deliveries multiplied by 100 000. To calculate the 95% confidence interval around the MMR estimates, we assumed a Poisson distribution for our data, namely, that the probability (Pr) of maternal deaths (y) per 100 000 deliveries is equal to some number r is given by Prðy ¼ rÞ ¼

lr el r!

where l is the expected value (mean) of y and r! ¼ rðr  1Þðr  2Þ    ð2Þð1Þ . The variance derived using the above assumption was then used in generating the standard errors and confidence intervals around the MMR estimates. To determine the temporal trend in maternal mortality ratios over the decade studied, a scatter plot of maternal mortality ratio versus year was initially produced to have an idea about the pattern of temporal distribution of these indices. We then proceeded to determine the best regression model that fitted the data by considering MMR as a continuous variable that depended on time (year). The linear model gave a very poor fit as portrayed by a very low coefficient of determination (R2). We therefore, added sequentially, a quadratic and finally a cubic term. The best model fit was obtained with all the three terms loaded. Using predicted values generated from the model, the best trajectory curve for the data was constructed alongside scattered points of observed values of MMR. A logistic regression model was constructed to identify maternal socio-demographic characteristics predictive of maternal mortality.

Y.M. Adamu et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 109 (2003) 153–159

All tests of hypothesis were two-sided with a type 1 error rate fixed at 5%.

3. Result A total of 4154 maternal deaths occurred among 171 621 deliveries within the decade under study yielding a maternal mortality ratio (MMR) of 2420 deaths per 100 000 deliveries (CI ¼ 23462494 deaths per 100 000). This average figure, however, hides a very important trend of maternal deaths within the period of investigation. Fig. 1 depicts marked yearly fluctuations in MMR. After smoothening, the curve revealed two significant spikes in the levels of MMR within the decade of study. The first rising segment of the curve reached its zenith in 1992 with the highest maternal mortality ratio ever recorded within the decade studied, a staggering figure of 3638 maternal deaths per 100 000 deliveries. Subsequently, a period of decay ensued

155

characterized by a rapid decline in MMR and reached its lowest point ever at 1026 per 100 000 deliveries by 1995. Following this was a latency period that lasted only 2 years (1996–1997 inclusive). This was probably a transitional and unstable period awaiting a trigger that lead to another skyrocketing of the MMR heralding the second spike on the curve (see also Table 1). The association between maternal socio-demographic factors and maternal deaths from pregnancy complications is displayed in Table 2. Teenage mothers and mothers aged 40 years and above constituted the age groups at greatest risk. A more or less similar bimodal pattern of maternal deaths was observed with parity with elevated risks occurring among primiparae and grand multiparae. By far the strongest predictor of maternal mortality was the place of residence. Mothers living in rural areas and beyond 20 km radius from urban centers were four times more likely to die from delivery complications as compared to urban-dwellers after adjusting for other confounding maternal characteristics

Fig. 1. Temporal pattern of hospital maternal death ratios (MMR) in Kano State, 1990–1999 (ratios are per 100 000 deliveries).

Table 1 Maternal mortality ratios over the decade in Kano State, Northern Nigeria, 1990–1999 Year

Total deliveries

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 Total/average a b

Total number of maternal deaths

MMRa per 100 000

95% confidence interval

Odds ratiob

19117 18400 10584 26699 16606 40455 25207 30320 22512 17248

400 449 385 365 412 415 391 444 321 572

2092 2440 3637 1367 2481 1026 1551 1464 1426 3316

1887–2297 2214–2666 3275–4001 1227–1507 2241–2721 927–1125 1397–1705 1328–1600 1270–1582 3044–3588

1.0 1.17 1.74 0.65 1.19 0.49 0.74 0.70 0.68 1.58

171621

4154

2420

2346–2494



Maternal mortality ratio. Mantel–Haenszel w2 for trend ¼ 10.81 (P ¼ 0:001).

156

Y.M. Adamu et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 109 (2003) 153–159

Table 2 Selected maternal socio-demographic characteristics predictive of maternal deaths Variable

Crude OR Adjusted OR and CI

Age (in years) <20 20–29 30–39 40þ

1.00 0.86 0.23 1.34

1.00 0.73 (0.51–0.90) 0.21 (0.09–0.42) 1.02 (0.85–1.16)

Parity Primiparae Multiparae (2–5) Grand multiparae (>5)

1.00 0.60 0.92

1.00 0.78 (0.48–0.92) 1.06 (0.88–1.29)

Residence Urban (within 20 km radius) Rural

1.0 3.6

1.0 4.0 (3.50–5.21)

Literacy level Class I (secondary education or above) 1.0 Class II (primary education) 1.83 Class III (no formal education) 3.40

1.00 1.81 (1.59–2.00) 2.91 (2.62–3.33)

Previous poor obstetric history Abortion No Yes

1.00 1.18

1.00 1.16 (0.84–1.66)

Stillbirth No Yes

1.00 1.07

1.00 1.04 (0.31–1.11)

Premature delivery No Yes

1.00 1.20

1.00 1.08 (0.80–1.25)

OR: odds ratio. CI ¼ 95% confidence interval. Note: adjusted estimates were generated by loading all the variables in the table into the regression model.

(OR ¼ 4:0 ; 95% CI ¼ 3:505:21 ). The second most significant predictor of mortality was the educational level of mothers. As the level of education rose, the probability of the mother dying from delivery complications diminished

in a dose-dependent fashion (p-value for trend ¼ 0.02). Place of residence and literacy level are markers of socioeconomic status. The two may also have an additive effect on maternal mortality. We therefore, proceeded and explored the possibility of interaction effects of the two variables on maternal mortality. We assessed and compared goodness-offit of models with interaction terms with those containing terms for main effects only using deviance estimates. However, because the effect estimates for the interaction between place of residence and literacy level were not significant we excluded these interaction terms from our final model. Adjusted estimates did not show any significant relationship between previous poor obstetric history (abortion, stillbirth and premature birth) and maternal mortality. Table 3 compares the magnitude of MMR found in other studies with ours. The first half of the table gives a list of the top five countries with the highest level of maternal deaths in the world as reported by the World Health Organization. Strikingly, the MMR we are reporting is even higher than the worst nationally reported MMR from Rwanda. The second half of the table presents a region-by-region breakdown of MMR for Nigeria. The average MMR reported generally for Nigeria is 1000 per 100 000, which clearly masks substantial variations from one part of the country to another as illustrated in the table. It should be pointed out, however, that only two studies [1,3] were population-based, and therefore, comparable to ours. All the rest were conducted on selected populations from teaching hospitals in most instances. The three most common causes of maternal death in the study were eclampsia (31.3%) ruptured uterus (10.1%) and anemia (8.4%), in descending order. The triad accounted for 50% of the total maternal deaths recorded over the past decade at the study site. The causes of maternal deaths were as depicted in Fig. 2. The group labeled as others consisted mainly of maternal deaths attributable to ectopic gestations and complications of cesarean sections as well as indirect

Table 3 Comparison of maternal mortality ratios (MMR) from regions of Africa and geo-political zones of Nigeria with the MMR obtained in this study Study

Year

MMR per 100 000

Region

Comparison with the top five countries with the highest MMR in the world [1] Rwanda (WHO) 1995 Sierra Leone (WHO) 1995 Burundi (WHO) 1995 Ethiopia (WHO) 1995 Somalia (WHO) 1995 This study 1990–1999

2300 2100 1900 1800 1600 2420

Africa Africa Africa Africa Africa Nigeria

Comparison with other recent studies published from Nigeria Harrison, KA [3] 1997 WHO [1] 1996 Aboyeji AP [6] 1998 Ujah et al. [7] 1999 Etuk et al. [8] 2000 Olatunji AO et al. [9] 2001 Okaro et al. [10] 2001 This study –

1000 1546 532 739 600 1700 1406 2420

Nigeria North-East Nigeria Ilorin (North-Central Nigeria) Jos (North-Central Nigeria) Calabar (South-East Nigeria) Sagamu (South-West Nigeria) Enugu (South-East Nigeria) Kano (North-Central Nigeria)

Y.M. Adamu et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 109 (2003) 153–159

157

Fig. 2. Specific causes of maternal deaths by percentages in Kano State, 1990–1999.

causes comprising mainly medical conditions such as hepatitis and malignancy of the cervix.

4. Discussion This study detected very high maternal mortality ratios (MMR) at the study site. These values by far exceed the average of 1000 maternal deaths per 100 000 estimated for Africa [1], the continent with the highest MMR on the planet. It is also higher than that of Rwanda, the country reputed for the highest risk of death for pregnant women in the world. Even after considering a myriad of factors, including heterogeneity in study design and methods, one could at least cautiously describe our finding of 2420 maternal deaths per 100 000 deliveries found in this study as one of the world’s highest population-based MMRs reported within the last decade [1]. When comparison is made with other studies on maternal mortality published from Nigeria within the decade that elapsed [3,6–10], the MMR we are reporting is clearly the most disturbing in magnitude. A probable explanation for this could be the difference in the study design and target population. While data in this study were population-based, almost all the other studies from Nigeria were based on cases of maternal deaths observed among deliveries that took place in teaching hospital settings so that inferences deduced there from might not have been representative of the general population. Patients that deliver in teaching hospitals are likely to be more enlightened with regard to health issues, can afford quality care offered in teaching hospitals, and many of them live in the urban areas so that physical access to ante-natal care and hospital delivery may not constitute a big issue. This contrasts with the socio-economically disadvantaged mothers living in the remote and neglected rural areas where illiteracy and poverty reign supreme, the comprehensive and

primary health centers are poorly staffed and ill-equipped, and complicated deliveries have to be managed on-site due to physical, social and economic barriers which often preclude referrals to higher levels of care. Moreover, in many cases, such referrals are perceived by patients and their relatives as acts devoid of compassion or acts that simply delay the inevitable death of the mother. Most of these complicated deliveries therefore remain and die at these centers. The maternal deaths in these areas not served by teaching hospitals are therefore, more likely to be higher than the figures reported in previous reports. By being population-based, our study took these neglected maternal deaths into account in order to obtain less biased estimates. Another unique aspect of this report is that, to our knowledge, this represents the first study of its kind from Nigeria that covered maternal mortality occurrence over a whole state, which in this case, is the second most populous in the country. It is interesting to note the crescendo–descendo pattern of maternal mortality ratios within the decade studied. It demonstrates that when population data on maternal deaths are employed to describe the temporal occurrence of these events a lot more useful information could be obtained than the popular use of a descriptive average, which may convey the wrong impression of a ‘‘static equilibrium.’’ Our data clearly shows that maternal mortality occurrence in developing areas of the world, such as Nigeria, could exhibit episodes of outbreaks. It is therefore, important that we consider maternal mortality occurrence as a ‘‘dynamic process’’ associated with triggering factors. Unidentified triggers of spikes in maternal death are most likely also the root causes of the unstable baseline values of maternal deaths. Given this concept, it becomes clear that one of the most effective strategies in understanding and consequently curtailing the high maternal mortality ratios in developing countries, especially, sub-Saharan Africa, is the establishment

158

Y.M. Adamu et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 109 (2003) 153–159

of a sustainable ‘‘Surveillance System’’ that constantly monitors this dynamic occurrence of maternal mortality in the population. One possible explanation for the fluctuations in maternal mortality ratios over the decade as apparent in our data is the periodic changes in Government policies regarding service fees in health institutions. Reduction of service fees as well as availability of drugs in Government health institutions in Nigeria, which coincided with the establishment of the Petroleum Trust Fund (PTF) in 1995, might have contributed in alleviating the burden of pregnancy-related maternal deaths. However, in subsequent years, the PTF was abolished and health institutions were again deprived of lifesaving subventions. Other possible factors that might have contributed to these variations in maternal mortality include sporadic fuel shortages causing break-down in transportation, incessant strikes by health-care providers for a higher pay etc. Unfortunately, because of the set objectives of our study (which was grossly limited by availability of resources in conducting it) and its design, we could not quantify these variables and consider their predictive power in a regression model, a procedure that could have provided interesting and additional insights. Subsequent studies may consider the role played by these factors in causing fluctuations in the level of maternal mortality in the area. Eclampsia was the single most important cause of maternal deaths in our study, an interesting finding. While studies carried out in the southern part of Nigeria found hemorrhage and illegal abortion to be the leading causes of maternal mortality [9,11–13], those conducted in Northern Nigeria are in agreement with our result that eclampsia is the leading cause of death among pregnant mothers in that region [14– 16]. It is important to delve into the health belief model prevalent among women in this region of the world in order to understand the non-biologic pathway that makes eclampsia the most common cause of maternal death. Pregnant women affected by eclamptic fits or seizures are believed to be possessed by evil spirits [17] that have gained access into the body system of the patient. Since such ailments are generally considered to be beyond the realm of modern therapy, the expertise of traditional healers are sought first and foremost to exorcize the evil spirits and cure the affected mother. This results in considerable delay in seeking modern obstetric care. While delay in seeking and receiving obstetric care in general, has been identified as a major pre-disposing factor in the causation of maternal mortality in developing countries [18], phase I delay (delay in seeking appropriate treatment), in particular, appears to be the rate-determining step along the non-biologic pathway through which death precipitated by eclampsia occurs. While no claim is being made that the above is the sole indirect mechanism of maternal death caused by eclampsia, nonetheless, and in accord with other authors [12], the role played by phase I delay in the process leading to deaths among African mothers is substantial. In a strongly traditional society with pristine socio-cultural values like the one investigated in this

study, it becomes crucial to study, analyse and understand the socio-cultural contribution of maternal mortality causation otherwise on-going efforts to reduce maternal deaths in these societies will be futile. The five major causes of maternal deaths detected in this study (eclampsia, ruptured uterus, anemia, post-partum hemorrhage and sepsis) are more or less similar with the findings of a most recent investigation conducted within the same socio-cultural zone (eclampsia, ruptured uterus, obstetric hemorrhage, anesthetic complications and sepsis) [14]. The fact that four of these obstetric complications match in both studies is strong evidence that they constitute a constant threat to pregnant mothers in that region, and therefore, deserve to be considered priority targets for intervention. Also, eclampsia, ruptured uterus and anemia were responsible for about 50% of maternal deaths in our study. All three are highly preventable with appropriate interventions. A notable goal of the Safe Motherhood Initiative (SMI) that was launched in 1987 was to reduce maternal mortality at least by half by the year 2000 in developing countries including Nigeria [4]. One possible reason for the collapse of this initiative in Nigeria was the lack of focus of the program. Instead of identifying the most important causes of maternal deaths and allocating adequate resources to avert them, the strategy adopted was that of ‘‘mass attack’’ approach that aims to combat all causes of maternal deaths, a highly ambitious but difficult task. We believe that what is required at the study site is a program that consists of training, deployment and supervision of a large number of professional midwives and traditional birth attendants (TBAs) in villages, an information, education and communication (IEC) system to increase use of midwives and TBAs, and a district-based maternal and perinatal audit (MPA). The trained staff should be skilled in detecting and referring mothers that are at risk for eclampsia, anemia and ruptured uterus during pregnancy. Similar strategies have been found to reduce maternal mortality substantially elsewhere [19–22]. Finally, it is noteworthy to point out that differing definitions for what constitutes maternal death can influence the estimates for maternal mortality ratios and rates. Two main definitions predominate in general: pregnancy-related and pregnancy-associated maternal deaths. Pregnancy-related maternal mortality is death of the mother resulting from pregnancy, its complications, or treatment, while pregnancyassociated maternal deaths are those that would have occurred even in the absence of pregnancy [23–25] in accordance with definitions proposed by the Centers for Disease Control (CDC) and the American College of Obstetricians and Gynecologists (ACOG). Our report covered almost exclusively pregnancy-related deaths, and accidental or incidental causes, e.g. injuries, homicide and suicide were not reported. This omission would have underestimated the maternal mortality ratios we have reported because of undercounts in the numerator. However, since developing countries in general are concerned mostly with causes of death

Y.M. Adamu et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 109 (2003) 153–159

that are related to pregnancy event and how to curtail them, it was only reasonable and practical on our part to have focused on pregnancy-related instead of pregnancy-associated maternal deaths. Another issue of debate relates to the temporal measurement of maternal mortality. We used the conventional definition of maternal deaths that occurred up to 42 days post-delivery instead of the 1-year post-delivery definition currently being advocated [25]. Since our study was conducted in a developing setting with extremely constrained logistic support, it would have been difficult to capture maternal deaths related or associated with pregnancy 1 year post-delivery. Furthermore, use of the latter definition would have made it almost impossible to compare our estimates with those from similar regions as we did.

Acknowledgements This paper was based on a field study conducted for the award of the PhD degree to the first author titled, Spatiotemporal patterns of maternal mortality in Kano State. Our sincere thanks go to Prof. Falola JA for supervising the thesis, and to the Social Science Academy of Nigeria and The Bayero University Kano for their financial support. We also appreciate the co-operation of the Kano State Ministry of Health for allowing us access to the data. The Fulbright Commission awarded a 9-month research fellowship to the first author at the School of Public Health (University of Alabama at Birmingham), a stay that facilitated the production of this paper.

References [1] WHO. Maternal mortality in 1995: estimates developed by WHO, UNICEF, UNFPA. WHO/RHR01.9; 2001. [2] Walraven G, Telfer M, Rowley J, Ronsmans C. Maternal mortality in rural Gambia: levels, causes and contributing factors. Bull World Health Org 2000;78(5):603–13. [3] Vandecruys HIB, Pattinson RC, Macdonald AP, Mantel GD. Severe acute maternal morbidity and mortality in the Pretoria Academic Complex: changing patterns over 4 years. Eur J Obstet Gynecol Reprod Biol 2002;102:6–10. [4] Adamu YM. Distribution of health facilities in Urban Kano. In: Proceedings of the 38th annual conference of the Nigerian Geographical Association. University of Maiduguri (Nigeria); 1997. [5] Danyaro MM. Kano State 30 years of statehood, 1967–1997, 30th anniversary publication. Kano; 1997. p. 3.

159

[6] Aboyeji AP. Trends in maternal mortality in Ilorin, Nigeria 1987– 1996. Int J Gynecol Obstet 1998;63:183–4. [7] Ujah IA, Uguru VE, Aisien CIA, Sagay AS, Otubu JA. How safe is motherhood in Nigeria? The trend of maternal mortality in a tertiary health institution. East Afr Med J 1999;76:436–9. [8] Etuk SJ, Itam IH, Asuquo EE. Morbidity and mortality in booked women who deliver outside orthodox health facilities in Calabar, Nigeria. Acta Trop 2000;75:309–13. [9] Olatunji AO, Sule-Odu AO. Maternal mortality at Sagamu, Nigeria-a 10-year review (1988–1997). Niger Postgrad Med J 2001;8:12–5. [10] Okaro JM, Umezulike AC, Onah HE, Chukwuali LI, Ezugwu OF, Nweke PC. Maternal mortality at the University of Nigeria Teaching Hospital, Enugu, before and after Kenya. Afr J Reprod Health 2001;5:90–7. [11] Ogunniyi SO, Faleyimu BL. Trends in maternal deaths in Ilesa, Nigeria, 1977–1988. Afr J Med 1991;10:400–4. [12] Okonofua FE, Abejide A, Makanjuola RA. Maternal mortality in Ile-Ife, Nigeria: a study of risk factors. Stud Fam Plann 1992;23: 319–24. [13] Asuquo EE, Ottong JG, Olaniran NS, Duke F. Perception of obstetric haemorrhage in two rural communities of Cross River State. In: Women’s Health Issues in Nigeria. Zaria: Tamaza Publishers; 1992. p. 21–30. [14] Onwuhafua PI, Onwuhafua A, Adze J. The challenge of reducing maternal mortality in Nigeria. Int J Gynecol Obstet 2000;71:211–3. [15] El-Nafaty AU, Omotara BA. Perceived causes of eclampsia in four ethnic groups in Borno State, Nigeria. Afr J Reprod Health 1998;2:20–5. [16] Shehu DJ. Socio-cultural factors in the causation of maternal mortality and morbidity in Sokoto. In: Women’s Health Issues in Nigeria. Zaria, Tamaza Publishers; 1992. p. 203–14. [17] Ahmad BM. Women perception of eclampsia in Kano. BS degree project. Department of Geography, Bayero University, Kano, Nigeria; 1999. [18] Thaddeus S, Maine D. Too far to walk: maternal mortality in context. New York: Center for Population and Family Health, Columbia University; 1990. [19] Schaider J, Ngonyani S, Tomlin S, Rydman R, Roberts R. International maternal mortality reduction: outcome of traditional birth attendant education and intervention in Angola. J Med Syst 1999;23:99–105. [20] Maine D, Akalin MZ, Chakraborty J, de Francisco A, Strong M. Why did maternal mortality decline in Matlab? Stud Fam Plann 1996;27:179–87. [21] Ibrahim SA, Omer MI, Amin IK, Babiker AG, Rushwan H. The role of the village midwife in detection of high risk pregnancies and newborns. Int J Gynaecol Obstet 1992;39:117–22. [22] Atrash HK, Rowley D, Hogue CJR. Maternal and perinatal mortality. Curr Opin Obstet Gynecol 1992;4:61–71. [23] Atrash HK, Alexander SA, Berg CJ. Maternal mortality in developing countries: not just a concern of the past. Obstet Gynecol 1995;86:700–5. [24] Jocums SB, Berg CJ, Entman SS, Mitchell EF. Postdelivery mortality in Tennessee, 1989–1991. Obstet Gynecol 1998;91:766–70. [25] Frye V. Examining homicide’s contribution to pregnancy-associated deaths (Editorial). JAMA 2001;285:1510–1.