Perinatal mortality survey in Jamaica: aims and methodology

zyxwvutsrq zyxwvu Paediatric and Poinatal Epidemiology 1994, 8, Suppl. 1 , 6 1 6 Original articles Perinatal mortality survey in Jamaica: aims and methodology zyxw zyx Deanna Ashleya, Affette McCaw-Binnsb.', Jean Golding', Jean Keelingd, Carlos Escofferyb, Kathleen Coardb and Karen Foster-Williamsb uMinisty of Health and bLIniuersity of the West Indies, Kingston, Jamaica, CInstituteof Child Health, University of Bristol, UK and dRoyal Hospital for Sick Children, Edinburgh, UK. zyxwvu Summary. The Jamaican Perinatal Mortality Survey was designed to identify the true perinatal mortality rate, and assess the factors which could contribute towards a reduction in perinatal mortality on the island. All births in a 2-month period (n = 10527) were compared with all perinatal deaths occumng over a 1.2-month period (n= 2069). Over half the deaths (n= 1058)received a detailed post-mortem examination. Use of the Wigglesworth classification identifies the major component of perinatal death in this country to be associated with intrapartum asphyxia (44% of deaths). Deaths due to congenital malformations and miscellaneous causes contribute relatively little (<10%) to the overall mortality rate. Over a quarter of deaths apparently occur before the onset of labour, and a fifth are prematurely liveborn but die of causes related to immaturity. zy zyxwvutsrqp Background Jamaica is an island nation situated in the Caribbean sea, with many rivers and with a mountainous interior, rising to 2500 metres. The terrain makes transportation and communication difficult in some areas and this is often aggravated by flooding and Address for correspondence: Dr Deanna Ashley, Ministry of Health, 10 Caledonia Avenue, Kingston 5, Jamaica. 6 zyxw zyx zy Aims and methodology 7 landslides that follow torrential rains, especially in the months of September, October, April and May. The climate is tropical, with average maximum temperatures ranging from 90°F in August to 84°F in February. Like many other tropical areas, Jamaica is at risk of hurricanes (cyclones) each year between the months of June and November. The island gained its independence from Britain in 1962, and is now a member of the British Commonwealth. Its economy is largely dependent on tourism and bauxite mining, with agriculture playing a much smaller but important tertiary role. The gross domestic product in 1986 was US$2124 million with a per capita income of US$909.’ The island is divided into 14 administrative units called parishes (Fig. 1). The parishes of Kingston and St Andrew operate as a single unit for most administrative matters, and are often referred to as Kingston/St Andrew (KSA).Over 50% of the population live in urban areas mainly in the parishes of Kingston/St Andrew, St Catherine and St James (whereMontego Bay is located) whilst the other 10 parishes are largely rural. zyxwvut zyxw zyxwvu zyxwvu WEST N O R M EAST zyxw ., SOUTH EAST Figure 1. Jamaica:Parishes, regions and location of public hospitals. hospitals (Type A/B); 0,location of Level-I1 hospitals (Type C). location of Level-I The health service infrastructure is quite well developed and distributed throughout the 14 parishes. A network of 364 primary health care centres provides basic antenatal, postnatal and child health care, and curative and domiciliary midwifery services in 62 health districts. Approximately 90% of the population reside within 10 miles of a health centre with a mean distance of 2.4 miIes.2 Health centres are graded into Types I to V. Type I is the simplest,providing basic maternal and child health, family planning and domiciliary services under a registered midwife assisted by two community health aides. In addition to these personnel and servicesthe Type I1 usually carries a registered staff nurse, public health nurse, public health inspector, visiting nurse practitioner and a visiting doctor. Services include simple treatment provided by a nurse. The Type I11 health centre is the 8 zyxwvuts zyxwv zyxwvu zyxwvut zyxwvu 0.Ashley et al. headquarters for the health district and carries a public health nurse, public health inspector, a nurse practitioner, the district medical officer with dentist and dental nurse, providing a wider range of clinical services on a regular basis. When the parish health department (administrative office) and a Type I11centre are combined, this unit is designated a Type IV centre. Large polyclinic centres located in urban areas serving populations of 50000or more are designated Type V centres. They provide specialist ambulatory services in addition to the services provided at the other types of primary care centres. Hospital services are delivered through 24 public and six private hospitals; at least one public hospital is located in each parish. In the public sector, basic maternal, general medical and surgical services are provided at Type C hospitals; additional specialist obstetrical and paediatric services are provided at regional Type B hospitals whilst Type A institutions located in the cities of Kingston and Montego Bay, provide tertiary obstetric, neonatal and paediatric care as well as other specialised services. Reasons for the study It had been reported that under-registration of infant deaths in Jamaica ranged from 33-54%, but it is thought that there is an even greater deficiency in registration of stillbirths and early neonatal deaths. In one parish in 1982,63% of early neonatal deaths were shown not to have been registered.3The stillbirth rate in 1984based on registered stillbirths was reported to be 5.9 per 1000.This was recognised as a gross underestimate as the rate for public hospital deliveries alone was more than double this figure. The Registrar General's Department ceased publication of data on stillbirths and infant deaths in 1984 because of the unreliable nature of their information. Where adequate data were available there was little cause for complacency. Lowry and colleagues4 showed that perinatal mortality rates for the University Hospital in Kingston had not reduced over a 10-year period between 1963and 1965 the rate was 35.1 per 1000, and from 1973 to 1975 the rate was 38.8 per 1000. There was no information on mortality rates outside the capital city. Nor was there any information on the causes of perinatal death, or information concerning identilication of high-risk women within the island. It seemed important therefore that a study be conducted to estimate more accurately the real level of perinatal mortality, to identify causes of death and to determine the maternal, social and environmental factors predictive of fetal and early infant death. To do this, a series of aims was set out? those related to mortality being: 1 To determine the stillbirth and neonatal mortality rates for the country. 2 To determine the actual clinical and pathological causes of these deaths. 3 To identlfy maternal and environmental characteristics associated with increased risk of stillbirth or neonatal death. Aims and methodology 9 To identify clinical characteristics (in mother or newborn) predictive of death of the infant or fetus. From the results, to revise the recommended methods of identifying mothers at high risk of such outcomes. From the results of the survey, to develop appropriate risk scores or check lists for use at different levels of care. In association with this, and as a result of the survey, to revise the norms and procedures for use at primary care level. To develop norms and procedures for use at the secondary care level. To revise the national strategy and programme to reduce perinatal mortality. Study design zy zyxwv zy zyxwvu In order to achieve the objectives of this study and in particular to have sufficient numbers to identify specific types and causes of death, it was important to have a stratified sampling mechanism. The design chosen was similar to that developed for the First British Perinatal Mortality Survey.6Two observation strata of different lengths were required in order to obtain adequate sample sizes: (1)The main cohort (2months); and (2) stillbirth and neonatal deaths (12 months). Full details of the way in which the study was organised and implemented have been described el~ewhere.~ (1) The main cohort All pregnant women who had a live birth or stillbirth during the 2-month period from 1 September to 31 October 1986, regardless of the place of delivery, were included. These women were interviewed and their babies examined, usually within the first 48 hours after delivery. In all, 94%of the births in the 2-month period were identified and included in the study.2 (2) Mortality component zyxw zyxwvu All stillbirths (fresh or macerated)and neonatal deaths he. < 28 days) over500g for babies born between 1 September 1986 and 31 August 1987 were included in the mortality study. Cadavers were transferred where possible to one of three institutions (one in Montego Bay and two in Kingston) for post-mortem examinations by the team of study pathologists. All babies were transported to Kingston for post-mortems after the Main Cohort Study ended. Mothers or next of kin were asked to give consent for necropsy and for disposal of the body. Parents wishing to bury their infants had the bodies returned after post-mortem (there were only three such requests). 10 zyxwvuts zyxwvu zyxwvu D. Ashley et al. The questionnaires There were three questionnaires relevant to the assessment of factors related to perinatal mortality: (1)The Main Questionnaires. These were administered to all mothers, whether in the Main Cohort or Mortality Study. They were divided into eight sections and contained 133 questions. Information elicited related to past obstetric history, social and environmental factors, the antenatal period, labour and delivery. Antenatal care data were supplemented by information obtained from hospital, clinic or doctois records as necessary. Information relating to the infant was obtained by direct examination of the baby and supplementary information extracted from records. (2) Perinatal Death Questionnaires were designed for all perinatal deaths. They summarised information about the death including place and time and a brief clinical history of the events surrounding labour, delivery and the immediate postpartum period which may have contributed to the outcome. (3) The Pathology Questionnaires had been devised by Dr Jean Keeling for a study of perinatal deaths in Britain and adapted for use in the Jamaican study. Babies sent for necropsy had relevant clinical and maternal information included in a Perinatal Death Questionnaire,and a clinical extract form summarising information taken from the Main Questionnaire. These documents were to assist the pathologist in interpreting and assessing the post-mortem findings. Whether or not the body was sent for necropsy, a Main Questionnaire and a Perinatal Death Questionnaire were completed. zyxwvut Post-mortem examination Prosectors were instructed in methodology appropriate to performance of a necropsy on a stillbirth or neonate, including the recording of external measurements and body weight, recording and photography of dysmorphic features when possible, and full dissection of internal organs including the cranial cavity. Printed proformas were used to record all findings, including organ weights, and an estimate of gestation from the gyral pattern of the brain. As specific questions were asked, a clear statement of all negative findings was obtained. The system also permitted recording of descriptive details of individual abnormalities. Tissue samples were fixed in buffered formalin for histological examination and the placenta was examined whenever possible. Incidence of perinatal death zyxw zyx zyxwv A perinatal death was defined as either a fetal death weighing 500 g or more or the death of a live birth occurring within 7 days of delivery. There were 10527 births within the cohort period and 430 perinatal deaths, giving a perinatal mortality rate zyxw zyx zy zy Aims and methodology 11 of 40.8 per 1000 total births during this 2-month period [95% confidence interval (CI)37.1-44.6 per 10001. During the 12 months of the death study, there were an estimated 54400 births; 2069 fetal and early neonatal deaths on the island were notified by the coordinators during the study year, giving an estimated perinatal mortality rate of 38.0 per 1000 (95%CI 36.4-39.6 per 1000).Notwithstanding the overlap of these CIS, it is a matter of conjecture whether the slightly higher rate in the 2-month period may represent an increased risk during this period or a subsequent reduction in case ascertainment. Validation zyxw zyxw zyxw In order to assess how complete the sample of cohort deliveries was, a complex comparison of information obtained by the study with registrations of birth and deaths was camed out? Of the 10227 live births in the cohort, 94% had been registered within 12 months. Thus, 597 had not been registered. Only in 47% of the neonatal deaths identified in the study, had the actual birth been registered and in only 9.2%(17)of the 184deaths had thedeatk been registered. Of the stillbirthsin the study, only 23 (9%)had been registered. During the exercise of matching the registrations with births in the cohort, a further 652 live births were identified as being registered, but not known to the study. Thus, the total sample of live births in thisstudy, appears to be of the order of 94% (10227/10879) of the population of liye births on the island. Post-mortem rates In all, post-mortem proformas were completed on 1057(51%)of the 2069 perinatal deaths. The post-mortem rate increased as the study proceeded, being only 48% during the cohort months but 58%during the rest of the year. There were sigruficant variations in the post-mortem rate with sex of the baby (boys being more likely to be examined), with time of death (neonataldeaths being more likely than stillbirths), and among twins compared with singletons (twins more likelyL7In addition, there were substantial variations between institutions at which the death occurred. Consequently there were sigruficant variations in post-mortem rate with both .~ of these biases it parish of delivery and parish of residence of the r n ~ t h e rBecause was necessary to use a classification system that was largely independent of whether or not post-mortem examination occurred. Classification of deaths Wigglesworth" devised a form of analysis for perinatal deaths which was designed to facilitate rapid identification of problems within the total death population. The 12 zyxwvut zyxwvut zyxwvut D.Ashley et al. underlying concept is that the aim of any analysis of deaths is prevention. The scheme divided deaths into only five groups, so that its use is appropriate for relatively small numbers of deaths. Another aim was that it should not be dependent on performance of necropsy examination so that it could be used in centres or countries where facilities were limited. The analysis could easily take birthweight into consideration, either in the primary analysis or later when looking at an area of major concern in more depth. This classification assigns deaths to one of five mutually exclusive groups: antepartum fetal deaths (APFD),major congenital malformations (CM),conditions associated with immaturity (IMMAT), asphyxia1conditions arising during labour and delivery (PA) and other miscellaneous specific conditions (MISC). It changes little whether necropsy has been undertaken or n ~ t ,although ~ , ~ the CM and M I X groups are slightly larger after necropsy at the expense of the other three groups. This classificationwas applied both to deaths coming to necropsy and those where post-mortem examination was not undertaken. The following modifications to the Wigglesworth classification were applied9 in the interests of clarity and reproducibility: APFD. Macerated stillbirths were assumed to have occurred prior to the onset of labour, unless there was clear clinical evidence to the contrary. CM. Infants were put in this category if there was a major malformation that would have resulted in death or severe morbidity, or if multiple minor malformations in more than one system were present, making a syndrome diagnosis likely. IMMAT. Comprised live births weighing between 1500g and 2499g dying after the first day of life and all live births weighing under 1500g, providing they did not fall into the CM or M I X categories. Babies of 2500g or more who were clearly preterm were also included provided they survived the first day. IPA. Included here were all fresh stillbirths, those macerated stillbirths where there was evidence that death had occurred during labour, live births weighing 1500g or more dying on the first day of life, and all normally formed live births of 2500g or more who died after the first day but had clinical evidence of birth asphyxia. MISC. Included specific causes of death such as Rhesus isoimmunisation or congenital syphilis, and mature babies dying with disorders normally associated with preterm delivery such as intraventricular haemorrhage. zyxwvut zy zyxw zy Post-mortem bias When necropsy is not undertaken, it is thought that the CM and MISC groups will be inappropriately small. When sophisticated investigations are undertaken either anteparturn, intraparturn or in the neonatal period and detailed necropsy investigations are performed, then the CM group will be maximal and the MISC group zyxwv zyx zy Aims and methodology 13 will increase at the expense of the IPA and IMMAT groups, and to a lesser extent, of APFD? The time of death of stillbirths was frequently not known and, when stated, reservations were expressed about the accuracy of some of the observations because of discrepancies between clinical data. For these reasons, stillbirths were often allocated to APFD or IPA categories based on necropsy observation of maceration. It is appreciated that the macerated and fresh stillbirth groups are not exactly equivalent to antepartum and intrapartum death but it will hold true for the majority of cases. The use of this classification has the advantage that all deaths can be categorised. Discrepancies were observed between classification of death as 'intraparturn' by the midwife and other observations which suggested that the baby was live born but died within a few minutes of birth. This would result in spurious enlargement of the fresh stillbirth group at the expense of the first day deaths, but considering both together as the IPA group gets around this problem. The Wigglesworth classification was applied to deaths both with and without necropsy (Table 1). The only Wigglesworth group where there is a significant difference between deaths with post-mortem and those without, is the antepartum fetal deaths. This is largely because some study personnel thought these babies not worthy of examination and permitted disposal. As the study proceded, effortswere made to discourage this practice. Interestingly there were no significant differences in the Wigglesworth classification distribution in the months when the postmortem rate was low, compared with months when the post-mortem rate was high (third and fourth columns of Table 1). zyxwv Antepartum fetal deaths (APFD) These are assumed to be deaths occurring before the onset of labour. In all, over one quarter (29%) of all perinatal deaths fell into this category. It was a working hypothesis of this study that improved antenatal care may help to reduce mortality in the antepartum period. Congenital defects (CM) zyxw Only 6% of the perinatal deaths had major congenital malformations. Even among those receiving post-mortem examination the rate was only 8%.1°Since the overall perinatal mortality rate was 41 per 1000, that attributable to congenital defects is likely to be only of the order of three per 1000 (8% of 41). Among specific lesions studied, there has been shown to be a particularly low rate of anencephaly,'' but a cluster of four cases conceived within a very short time 14 zyxwvut zyxwvut zyxwvut zyxwvu zyxw D.Ashley et al. Tabie 1. Classification of perinatal deaths according to (1) whether post-mortem examination carried out, (2) whether born in period with high post-mortem rate Wigglesworth classification Post mortem performed No HighPM months Restofyear All 220 (21%) 372 (37%) 208 (28%) 383 (29%) 592 (29%) 88 (8%) 226 (21%) 26 (3%) 166 (16%) 53 (7%) 126 (17%) 60 (5%) 252 (20%) 114 (6%) 392 (19%) 484 (46%) 40 (4%) 420 (41%) 27 (3%) 326 (45%) 20 (3%) 569 (43%) 43 (3%) 904 (44%) 1058 1011 733 (100%) 1307 2069 (100%) (100%) . Yes Antepartum fetal death Majormalformations Conditions association with Date of birth immaturity Intraparturn asphyxia Other specific 67 (3%) conditions Total (100%) (100%) P < o.Oo01 N.S. [29 deaths did not have date of birth recorded] N.S.: P > 0.05 period in an isolated area was observed.I2There were also two babies with complex disruptions and deformations as manifestations of the amnion rupture sequence, the first to be reported from the English-speaking Caribbean.13 At post-mortem the proportion of deaths with fatal malformations increased with increasing birthweight; fatal defects were found more often in neonatal deaths than in stillbirths. Even though spina bifida and anencephalus were relatively rare, when the deaths with congenital defects were analysed by the system involved, the central nervous system was most common, followed by renal, gastrointestinal and cardiovascular systems. Of those defects identified at necropsy, it was thought that the deaths from gastrointestinal defects were the most likely to be preventable given appropriate surgical services.*O Deaths due to immaturity (IMMAT) This category involves only live births dying in the early neonatal period. The classificationused here has generally taken the birthweight of the infant as being an important criterion for admission to this category. There were 392 deaths in this group, comprising 19%of the perinatal deaths. Only 29%of these babies had been admitted to a neonatal unit prior to death. Babies of birthweight over 2500g or gestation 37 weeks or more but dying of conditions normally associated with zyxw zyx Aims and methodology 15 immaturity (e.g. hyaline membrane disease or intraventricular haemorrhage) have been counted among the M I X group. Intrapartum asphyxia zyx zyxw zyxw Use of the Wigglesworth classification emphasises the large number of babies, both mature and immature, who die as a result of asphyxia1conditions, the vast majority of which arise during labour. This group comprises almost half (44%) of the perinatal deaths in Jamaica and should be the group most likely to be preventable by changes in obstetric care during labour and delivery. Discussion Prior to this study, assessments of perinataI mortality in Jamaica were hospital based. The most recent study reported a rate of 27.9/1000 for 1982.14 Examination of national statistics showed serious under-reporting of perinatal deaths. The perinatal mortality rate from the present study is 30% higher than previous estimates. It is, however, similar to the rate of 34.5/1000 births reported for 1984-85 from Curacao (another Caribbean island).15 The average post-mortem rate of 51% was lower than we had hoped for, but is partly related to there being more deaths than anticipated, difficulties in transporting bodies to Kingston and problems with preserving cadavers in a tropical climate. The Wigglesworth classification and post-mortem findings identified intrapartum asphyxia as the most important cause of perinatal death in Jamaica. The next largest group of deaths identified comprised antepartum fetal deaths. These two groups should provide the focus for preventative action. Twenty per cent of deaths were identified as conditions associated with immaturity whilst only 6% were related to major malformations, and 3% to miscellaneous causes. These three groups account for a smaller proportion of deaths than is currently found in Europe and North America. Although identrfying intervention strategies to deal with these conditions is not a priority at this time, these conditions will gradually increase in significance as interventions succeed in preventing deaths from intrapartum asphyxia and antepartum fetal death. zyxwvu Acknowledgements This vital and extensive study has been funded by the International Development Research Centre of Canada. The statistical analyses were supported by the Science and Technology for Development Programme of the Commission of the European Community Contract No. TS2-0041-UK.We are grateful to our funders, but also to the teams of pathologists, paediatricians, study coordinators and their assistants, 16 zyxwvut zyxwvu zyxwvut D.Ashley et al. the interviewers, and especially the Jamaican mothers and their babies who made the study a reality. zyxwvut zyxwvutsr zyxwvu zyxw zyx References 1 Planning Institute of Jamaica. Economic and Social Survey 1986. Kingston, Jamaica: 1988. 2 McCaw-BinnsA. Does antenatal care make a difference? An examination of antenatal care in Jamaica and its relationship to pregnancy outcome. University of Bristol, UK: PhD thesis, 1993. 3 Figueroa JP, McCaw AM, Wint BA. Review of primary health carein Jamaica 1977-1982. PAHO Project Consultancy Report. Washington DC:PAHO, 1983. 4 Lowry M, Hall J, Sparke B. Perinatal mortality in the University Hospital of the West Indies: 1973-1975. West Indian Medical Journal 1976; 2592-100. 5 Ashley D,McCaw-Binns A, Foster-WilliamsK. The perinatal morbidity and mortality survey of Jamaica 1986-1987. Paediatric and Perinatal Epidemiology 1988; 2138-147. 6 Butler NR, Bonham DG. Perinatal Mortality: The First Report of the British Perinatal Mortality Survey. Edinburgh: E & S Livingstone, 1963. 7 Coard K, Codrington G, Escoffery C, Keeling JW, Ashley D, Colding J. Perinatal mortality in Jamaica. Acta Paediatrica Scandinavica 1991; 80749-755. 8 Wigglesworth JS.Monitoring perinatal mortality. A pathephysiological approach. Lancet 1980; ik684486. 9 Keeling JW,MacGillivray I, Colding J, et al. Classificationof perinatal death. Archives of Disease in Childhood 1989; W1345-1351. 10 Coard K, Codrington G, Keeling JW, et al. Fatal malformations in Jamaica. Pediatric Pathology 1990; 10729-742. 11 Coard K, Escoffery C, Colding J, Ashley D. InGdence of anencephaly in Jamaica. Teratology 1990; 41:173-176. 12 Golding J, Foster-WilliamsK, Coard K, Ashley D. A cluster of central nervous system defects in Jamaica.Human Experimental Toxicology 1990; 913-16. 13 Escoffery CT,Coard KCM. Amnion rupture sequence in Jamaica. West Indian Medical Journal 1989; 38:164-170. 14 Ashley D,Gayle C, Fox K. A retrospective study of perinatal and neonatal mortality at the Victoria Jubilee Hospital in 1982. Kingston, Jamaica: Ministry of Health (Mimeo), 1985. 15 Wildschut HIJ, Tutein Nolthenius-PuylaertMCBJE, Viedijk V, et al. Fetal and neonatal mortality, a matter of care? Report of a survey in Curacao, Netherlands Antilles. British Medical Journal 1987; 295894-898. The Main Questionnaire used in this study is reproduced in Paediatric and Perinatal Epidemiology 1994;8: Suppl. 1.pp. 174-188.