SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgments
  8. References

Objective  To audit trends in maternal mortality in the Peninsula Maternal and Neonatal Service (PMNS) over a 50-year period, with respect to rates and patterns of causation.

Design  Retrospective and prospective audit.

Setting  The PMNS, an integrated perinatal service composed of primary, secondary and tertiary facilities in Cape Town.

Population All women giving birth in the area of the Cape Peninsula served by the PMNS over the 50-year period.

Methods  Data on maternal deaths were collected for 1953–2002 inclusive, from annual obstetric and gynaecological reports. Three triennia (1954–1956, 1981–1983 and 1999–2001) were selected for a detailed comparison of trends in rates and causes of death.

Main outcome measures  Maternal mortality rates (MMRs). Causes of maternal deaths.

Results  Total deliveries increased from 7315 in 1953 to 27,575 in 2002. The MMR declined from 301 deaths per 100,000 deliveries in 1953 to 31.2 in the triennium, 1987–1989. From 1999, the MMR increased, reaching 112 in 2002. Comparing 1954–1956 (MMR of 253.9) with 1981–1983 (MMR of 43.8), there was a marked decline in the MMR related to hypertension (80.4 to 11.3), haemorrhage (50.8 to 4.2), abortion (55 to 4.2), suspected pulmonary embolism (25.4 to 2.8), pregnancy-related sepsis (8.5 to 4.2) and cardiac disease (21.2 to 2.8). Comparing 1981–1983 (MMR of 43.8) with 1999–2001 (MMR of 59.4), there was a decline in the MMR associated with abortion (4.2 to 0). The MMR for haemorrhage, suspected pulmonary embolism and cardiac disease remained the same. There was a slight increase in the MMR attributed to hypertension (11.3 to 14.5) and pregnancy-related sepsis (4.2 to 7.3). There was a marked increase in the MMR associated with non-pregnancy-related infections/AIDS (4.2 to 18.2).

Conclusions  The MMR for all causes of maternal death declined significantly from 1953 to 1981 as a result of several interventions. From 1999, there has been a non-significant increase in MMR, predominantly due to the burden of HIV/AIDS-related mortality.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgments
  8. References

Deaths of women related to pregnancy and childbirth remain a major public health problem. Developing countries account for 99% of all maternal deaths globally or approximately half a million maternal deaths per year.1 Audits of maternal mortality have contributed to the marked reductions in maternal mortality rates (MMRs), which have occurred in many developed countries. In the UK, the Confidential Enquiries into Maternal Deaths in England and Wales commenced in 1952 and have analysed MMRs and causes in triennial reports.2 The identification of avoidable factors, later termed substandard care, has led to the development of national recommendations and guidelines whose implementation contributed to improvements in maternal outcome. However, it should be noted that, in the UK, a sharp decline in the MMR had already commenced in the 1930s, when the level was 400–500/100,000 deliveries. The decline has continued steadily, to reach 11/100,000 deliveries in the latest report.3 The early decline was related to general medical measures such as antibiotics, safe blood transfusion, the use of oxytocic drugs, improvements in anaesthesia, specific measures related to maternity care and a general increase in the living standards of the population.

Many developing countries have MMRs similar to, or in excess of, those occurring in the UK prior to 1939.4 Accurate determination of the MMR is hampered by incomplete data collection and the large number of home deliveries.5 In 1987,the World Health Organisation launched the Safe Motherhood Initiative at a global conference in Nairobi, Kenya.6 At this conference, the vast differences in MMRs, between developed and developing countries were highlighted, as well as the differing patterns of causes of maternal deaths.7 Very few developing countries have shown the sharp declines in MMR seen in the UK. A follow up Safe Motherhood conference, 10 years later, in fact showed that MMRs had actually increased in many developing countries. This increase predominantly affects countries in sub-Saharan Africa.8

South Africa is the first country in Africa to have introduced an ongoing National Audit and Confidential Enquiries into Maternal Deaths.9 The latter were initiated in 1998, four years after South Africa's first democratic election. A system, in which maternal deaths are identified and investigated, with a systematic assessment of avoidable factors, has been developed and two national reports have been produced. However, due to data collection problems in many provinces of South Africa, there is still no accurate figure for a national MMR because there are no accurate denominators (i.e. the numbers of either live births or deliveries) in many provinces.

The Peninsula Maternal and Neonatal Service (PMNS) in Cape Town is a grouping of public maternity services comprising primary level Midwife Obstetric Units (MOUs), secondary hospitals and a tertiary unit.10 It has served a defined geographical area of the Cape Peninsula for over 50 years. The Department of Obstetrics and Gynaecology of the University of Cape Town has collated and analysed maternal deaths and numbers of deliveries since 1953.11–13 This article analyses trends in maternal mortality over that period.

METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgments
  8. References

The study population included all women giving birth in the area of the Cape Peninsula served by the PMNS. The women were predominantly residing in high density and poorly resourced communities that had been disadvantaged under apartheid being predominantly Coloured (mixed race) and African. The Coloured group comprises mainly permanent urban residents. The African group comprises permanent urban residents, as well as those who have migrated from neighbouring rural provinces. This migration has accelerated, particularly in the last 14 years.

In the 1950s, the obstetric services in the Cape Peninsula consisted of a domiciliary midwifery system and a variety of clinics, backed up by three secondary obstetric hospitals and one tertiary unit. At that stage, the various components of the service were not well integrated or co-ordinated. In the 1970s, MOUs were developed in various ‘townships’ and suburbs. The MOUs have virtually replaced domiciliary midwifery. In 1980, all maternity facilities were formally integrated and rearranged to form the PMNS. This consists of seven MOUs (Khayelitsha, Guguletu, Mitchell's Plain, Retreat, Hanover Park, Heideveld and St. Monica's/Vanguard), two secondary hospitals (Somerset Hospital and Peninsula Maternity Hospital; the latter moving in the 1990s to Mowbray Maternity Hospital) and one tertiary maternity unit (Groote Schuur Hospital). The integrated PMNS is a well co-ordinated system, with agreed upon referral criteria and referral routes. The geographical area covered by the PMNS over the 50-year period has remained constant. However, the population has increased considerably due to influx of people from neighbouring rural provinces, in particular the Eastern Cape. This influx has led to the growth of large peri-urban informal settlement/squatter areas such as Khayelitsha, which formed in 1983 and has grown rapidly since. With the relaxation of influx control in the early 1990s, there was an accelerated urban migration.

Data were collected from the annual obstetric and gynaecological reports, produced by the Department of Obstetrics and Gynaecology, University of Cape Town, which supervises the PMNS. The reports included numbers of maternal deaths with a description of causes. Abortion deaths were included in the Gynaecology reports, which were commenced in 1957. Total numbers of deliveries and births were also recorded.

Maternal deaths were identified by doctors and nurses working at all levels of the service. The World Health Organisation (WHO) definition of a maternal death as described in the Ninth Revision of the International Classification of Diseases (ICD-9) was used.14 Accidental/fortuitous deaths were also documented. MMRs were calculated as the number of maternal deaths per 100,000 maternities. This is the same as the formula used for MMRs in the UK Confidential Enquiries.2 Causes of death were determined by clinical case discussions at audit meetings and postmortems for cases where there was clinical uncertainty about the cause of death. Causes were grouped into categories consistent with those developed in the SA Confidential Enquiries into Maternal deaths.9

The second author (H.A.v.C d G) reviewed the data from 1953 to 1996 (retrospectively until the mid 1970s and prospectively thereafter) and collated them into triennial reports. From 1997, the first author (S.R.F.) collected the data prospectively until 2002 inclusively.

For 1953 and 2002, data are presented for single years.

For 1954–2001 inclusive, data are presented in triennia.

Three triennia (1954–1956, 1981–1983 and 1999–2001) were selected for a detailed comparison of the causes of maternal deaths. The 1954–1956 and 1999–2001 represented the beginning and end of the 50-year period. The 1981–1983 represented a triennium when the MMR was close to its lowest level.

Because the gynaecological reports that included the abortion deaths only commenced in 1957, the number of abortion deaths in the triennia 1954–1956 was estimated from the number of abortion deaths in the 1960–1962 triennium. This was the first accurate measurement of abortion deaths in a triennium.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgments
  8. References

Data were collected from 1953 to 2002. Missing data included data for the triennium 1992–1995. In addition, abortion deaths reporting only commenced in 1957.

In 1953, there were 7315 deliveries within the area of the future PMNS, increasing to 27,575 in 2002.Figure 1 shows the trend in deliveries for triennia from 1954 to 2001.

image

Figure 1. Deliveries in the PMNS for the triennia 1954–2001.

Download figure to PowerPoint

In 1953, the MMR was 301/100,000 deliveries. In 2002, the MMR was 112/100,000 deliveries.Figure 2 shows the MMRs for the years 1954–2001.

image

Figure 2. MMR in the PMNS for the triennia 1954–2001.

Download figure to PowerPoint

There was a sharp decline in the MMR starting in the 1950s and reaching its lowest level of 31.2 in the triennium 1987–1989. Since then, the MMR has risen again and markedly so since the late 1990s to 112/100,000 deliveries in 2002.

Table 1 compares MMRs in the three chosen triennia. It shows that the initial sharp decline was statistically significant but not the latter increase.

Table 1.  Trends in MMRs.
PeriodConfidence interval (95%)
No. of deathsNo. of birthsMMRLowerUpper
  1. The C1 is based on the Poisson distribution.

  2. The χ2 test for comparing proportions (Fleiss, 1972) gives the period 1954–1956 as significantly different from the other two periods (P < 0.0001). The difference between 1981–1983 and 1999–2001 is not significant (χ2= 3.0, df= 1).

  3. Total of maternal deaths excluded accidental deaths and deaths with unknown cause.

1954–19566023,635253.86195.47330.02
1981–19833170,71743.8428.6460.66
1999–20014982,47059.4243.9178.43

Table 2 compares the pattern of causes of maternal deaths, in the triennia 1954–1956, 1981–1983 and 1999–2001. During this time period, the MMR (maternal deaths per 100,000 maternities) declined from 253.9 in 1954–1956 to 43.8 in 1981–1983, and then increased to 59.4 in 1999–2001.

Table 2.  Comparison of the causes of maternal death between 1954–1956, 1981–1983 and 1999–2001.
Causes1954–1956 (deliveries = 23,635)1981–1983 (deliveries = 70,717)1999–2001 (deliveries = 82,470)
No. of deathsMMRNo. of deathsMMRNo. of deathsMMR
  1. MMR = maternal mortality rate (number of maternal deaths per 100,000 deliveries).

  2. * Estimate only.

Direct
Hypertensive disorders1980.4811.31214.5
Abortion13*5534.200
Haemorrhage1250.834.233.6
Suspected pulmonary embolism625.422.833.6
Pregnancy-related sepsis28.534.267.3
Anaesthetic complication0011.411.2
 
Indirect
Cardiac521.222.800
Other medical disorders312.768.5910.9
Non-pregnancy-related infections0034.21518.2
 
Coincidental
Trauma0011.411.2
 
Unknown000022.4
Total60253.93245.35263.1

Data collection of deaths from abortion only began in 1957. For the next six years, there was an average of 13 deaths per triennium. The number of abortion deaths for 1954–1956 is thus an estimation, based on subsequent totals. The numbers for 1981–1983 and 1999–2001 are measured.

Comparing 1954–1956 with 1981–1983,Table 2 shows a very marked decline in the MMRs related to hypertensive disorders, obstetric haemorrhage, abortion, suspected pulmonary embolus and cardiac disease (predominantly rheumatic heart disease).

Deaths in the haemorrhage category included those from postpartum haemorrhage, antepartum haemorrhage and ruptured uterus. In the 1954–1956 triennia, there were three deaths from ruptured uterus secondary to traumatic operative vaginal delivery. Traumatic operative vaginal delivery did not feature as a cause of obstetric haemorrhage in the subsequent triennia. There was a less marked decline in the MMRs associated with pregnancy-related sepsis from 1954–1956 to 1981–1983.

Comparing the causes of maternal deaths in 1981–1983 with those in 1999–2001, the MMRs from haemorrhage, suspected thromboembolism, anaesthesia, cardiac disease and other medical disorders remained largely unchanged. There was a slight increase in the MMR related to hypertension and, more notably, in the MMR associated with pregnancy-related sepsis. There were no recorded deaths from abortion. However, there was a very marked increase in the number of indirect maternal deaths related to non-pregnancy-related infections. In this category, HIV/AIDS was assigned as the cause of death in women who tested HIV positive and had an AIDS defining illness or a CD4 count below 200. Of the 15 deaths in this category, HIV/AIDS definitely accounted for 11 (73%). In a further two cases (13%), the death was possibly due to HIV/AIDS because the patients had tuberculosis and features suggestive of HIV/AIDS but had not been tested for HIV.

Figure 3 is a graphic representation of the previous two tables, in which the six major causes of maternal deaths, in each of the three triennia, are compared. The six major causes are hypertension, haemorrhage, abortion, preexisting medical disorders, pregnancy-related sepsis and non-pregnancy-related infection. The observed changes in patterns of causes were not subjected to statistical significance testing because the numbers for individual causes were too small.

image

Figure 3. Trends in causes of maternal mortality. Comparison of causes of death in three triennia.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgments
  8. References

This 50-year audit of maternal mortality in the PMNS shows a trend towards a U-shaped pattern, with a dramatic decline from 1953 to 1985, followed by a period where the MMR remained at approximately 35 deaths per 100,000 deliveries and ending with a marked rise from 1999. Although this rise relates to only one triennium, 1999–2001, it is likely to be a real trend because the MMR was 112 in 2002 and 85 in 2003 (unpublished data).

There are obviously limitations to an audit of this nature. Although the system of data collection remained constant over the 50-year period, the population served by the PMNS has changed over the 50 years. Since 1983, there has been an influx of rural African women from the Eastern Cape, contributing to the explosive growth of urban townships, such as Khayelitsha. This influx accelerated in the 1990s due to the relaxation of influx control. Due to frequent rural–urban migration of disadvantaged groupings, and the development of informal settlements around Cape Town, it is difficult to accurately estimate population size. In the past half-century, the provision of maternity services has also changed markedly. From the 1950s, domiciliary midwifery had been expanded and starting in 1974 the MOUs were established. It is possible that, as maternity services were made more accessible to women, those who had previously had uncomplicated unsupervised deliveries at home now moved into the formal health sector.

The data for this audit were obtained from the health facilities and district midwifery record of the PMNS. It is therefore possible that some deaths at home, or in facilities outside the PMNS, may have been missed.

Despite these limitations, the trends observed are important and require discussion. The sharp decline in the MMR, from 1953 to 1983, is a source of optimism in providing documentary proof that such a decline is possible in a developing country scenario. Possible reasons for this decline include the following:

  • (i)
    The development of domiciliary midwifery, later to be replaced by the MOUs.
  • (ii)
    The introduction of a ‘Flying Squad’, a dedicated ambulance for obstetric and neonatal emergencies.15 This particularly contributed to the decline in deaths from obstetric haemorrhage.
  • (iii)
    The improvement in critical care management of the hypertensive disorders of pregnancy.

The sharp decline in the number of deaths due to abortion is interesting because it preceded liberalisation of the abortion legislation in South Africa, which was only enacted in 1997. Possible explanations for this observation could be that illegally induced abortions were performed more safely in the Western Cape and also that women presented earlier with complications.

It is also possible that the observed reduction in overall MMR may have been exaggerated by a ‘denominator effect’, in that more low risk women may have been delivering in formal health service facilities in the 1981–1983 triennium as compared with the 1954–1956 triennium. However, if it was purely a ‘denominator effect’, the percentage decline in MMR should have been similar for all causes of death, whereas the data show different percentage decreases for the different causes of death.

Sharp declines in the MMR have been well described in developed countries, such as the UK. The ready availability of safe blood transfusion, antibiotics and oxytocics, as well as the improvement in general living standards, were all thought to be contributory factors.2

Similar marked declines have only been observed in a few developing countries, such as in China, Cuba, Malaysia and Sri Lanka, where progressive primary health care policies were introduced, together with increased stress being placed on female literacy and education.16 In Africa, there has been a decline in MMR in the last 25 years in Egypt17 and also in the Gambia.18

The recent documented increase of the MMR in the PMNS is disturbing. It is predominantly due to HIV/AIDS-related causes although the MMR associated with pregnancy-related sepsis has shown an increase. The latter increase could be related to the increasing HIV prevalence. Voluntary HIV testing commenced in one suburb (Khayelitsha) in 1999. Between 2000 and 2002 it was extended to all areas. The category of death ‘non-pregnancy-related sepsis’ includes 15 women in the 1999–2001 triennia, 11 of whom definitely died from confirmed HIV/AIDS. Because HIV testing was not comprehensive during the study period, it is likely that the contribution of HIV/AIDS to maternal mortality was under-estimated.

Other less important factors contributing to the rising MMR could be the stringent financial cutbacks in health service facilities, and a continuing exodus of trained midwives. There has furthermore been a steadily increasing influx of high risk women from the rural Eastern Cape into the PMNS region.19 The Eastern Cape is an area of higher maternal mortality. This could also have contributed to the increased MMR.

HIV/AIDS is the major cause of maternal mortality in South Africa20 and probably in neighbouring countries as well. So far, programmes tackling the problem of HIV/AIDS have focussed on prevention of mother to child transmission (MTCT). Even though voluntary testing for HIV has increased in the PMNS since 1999, and even became routine in 2001, this has not provided any benefit for the mother's health. The reasons are that midwives and obstetricians lacked protocols on how to medically manage HIV-positive women in pregnancy. Also, until 2004, there have not been programmes in the South African Health sector to treat adults with antiretroviral drugs. From 2004, it is hoped that this will change in view of recent government policy to begin adult treatment programmes. This is starting in pilot sites in the Western Cape. The Burden of Disease Research unit of the Medical Research Council has provided initial estimates, which suggest that HIV/AIDS was the top single cause of mortality in South Africa.21

The most recent Saving Mothers Report has produced guidelines for the management of HIV-positive women in pregnancy.20 This is aimed at promoting more effective care of HIV-positive pregnant women.

‘Where is the M in MCH?’ was the title of an article by A. Rosenfield in the Lancet in 1985.22 The aim was to focus global attention on the problem of maternal mortality, which had been so neglected. Currently, ‘Where is the M in MTCT?’ may be a more relevant question for developing countries where MTCT programmes have focussed predominantly on preventing transmission of HIV to the fetus and less on measures to promote the health of the mother. There is concern that the increasing prevalence of HIV/AIDS is reversing the gains made by Safe Motherhood programmes.23 Indeed, HIV/Aids has contributed to an increase in maternal mortality in Malawi.24

In conclusion, this 50-year audit of maternal mortality in the PMNS has demonstrated that an integrated, regionalised maternal and neonatal service, based on strong primary care units (MOUs), can achieve a remarkable improvement in maternal outcome. However, the emergence of HIV/AIDS has reversed this trend and has presented us with new challenges, which will require comprehensive strategies to regain much lost ground.

Acknowledgments

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgments
  8. References

The authors would like to thank all the doctors and midwives who contributed to the collection of these data over the 50-year period and Dr George Draper for help with the graphics.

References

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgments
  8. References
  • 1
    Abouzahr C, Wardlaw T. Maternal mortality at the end of the decade: signs of progress? Bull World Health Org 2001;79: 561573.
  • 2
    Department of Health Report on Health and Social Subjects. Report on confidential enquiries into maternal deaths in England and Wales 1982–1984. HMSO 1989;34: 140151.
  • 3
    Department of Health. Why mothers die? 1997–1999. The Fifth Report of the Confidential Enquiries into Maternal Deaths in the United Kingdom. London: RCOG Press, 2001.
  • 4
    Graham WJ. The global problem of maternal mortality: inequalities and inequities. In: MacleanAB, NeilsonJ, editors. Maternal Morbidity and Mortality. London: RCOG, 2002.
  • 5
    Royston E, Armstrong S. Preventing Maternal Deaths. Geneva: WHO publications, 1989: 1418 [chapter 2].
  • 6
    Mahler H. The Safe Motherhood Initiative: a call to action. Lancet 1987: 268270.
  • 7
    World Health Organisation. Maternal mortality: helping women off the road to death. WHO Chronicle 1986;40(5):175183.
  • 8
    Hill K, Abouzahr C, Wardlaw T. Estimates of maternal mortality for 1995. Bull WHO 2001;79: 182193.
  • 9
    Saving Mothers. First Report on Confidential Enquiries into Maternal Deaths in South Africa. Pretoria: Department of Health, 1998.
  • 10
    Van Coeverden de Groot HA, Davey DA, Howland RC. The Peninsula Maternity and Neonatal Service, an urban community perinatal programme. S Afr Med J 1982;61: 3539.
  • 11
    Van Coeverden de Groot HA. Trends in maternal mortality in Cape Town 1953–1977. S Afr Med J 1979;56: 547552.
  • 12
    Van Coeverden de Groot HA. Deaths in gynaecological wards at Groote Schuur hospital, Cape Town 1957–1977. S Afr Med J 1979;56: 553557.
  • 13
    Van Coeverden de Groot HA. Maternal mortality in Cape Town, 1978–1983. S Afr Med J 1986;69: 797802.
  • 14
    World Health Organisation. International Classification of Diseases. Manual of the international statistical classification of diseases, injuries and causes of death. Ninth revision. Geneva, 1977.
  • 15
    Gunston KD, van Coeverden de Groot HAM. The Cape Town obstetric flying squad. S Afr Med J 1975;49: 11471148.
  • 16
    Royston E, Armstrong S. Preventing Maternal Deaths. Geneva: WHO Publication, 1989: 96100 [chapter 5].
  • 17
    Campbell O. Egypt 1992 to 2000 in Reducing Maternal Mortality. In: Koblinsky, editor. Washington DC: World Bank Publication, 2003: 93111 [chapter 6].
  • 18
    Walraven G, Telfer M, Rowley J, Ronsmans C. Maternal mortality in rural Gambia: levels causes and contributing factors. Bull WHO 2000;78: 603613.
  • 19
    Human I, Kroon M, Bergman N, Fawcus S, Ntwana B. Patient population movement in a Cape Town obstetric service. S Afr Med J 2003;93: 634.
  • 20
    Saving Mothers. Second Report on Confidential Enquiries into Maternal Deaths. South Africa 1999–2001. Pretoria: Department of Health, 2003.
  • 21
    Bradshaw D, Groenewald P, Laubscher R, et al. Initial burden of disease estimates for South Africa, 2000. S Afr Med J 2003;93: 682688.
  • 22
    Rosenfield A, Maine D. Maternal mortality—a neglected tragedy: where is the M in MCH? Lancet 1985;2: 8385.
  • 23
    Mcintyre J. Mothers infected with HIV. Br Med Bull 2003;67: 127135.
  • 24
    Adamson S, Phiri A. Did maternal mortality ratio increase in Malawi between 1992–1998? Review of Malawi Demographic and Health Surveys and other data sources. Trop Doct 2003;33: 182185.

Accepted 18 November 2004