Perinatal mortality audit


Associate Professor JF King, Department of Perinatal Medicine, Royal Women's Hospital, 132 Grattan Street, Carlton, Vic. 3053, Australia. Fax: +61 3 9347 1761; email:

Perinatal mortality 1982−2001 at Vila Central Hospital, Vanuatu, RF Grace and LB Everard

In this edition of the Journal, two clinicians from the Vila Central Hospital in the developing Pacific island nation of Vanuatu, report on a 20-year survey of the hospital's perinatal deaths. The hospital has approximately 1300 births per annum. They report that the perinatal mortality rate (PMR) over this period was 27 per 1000 births with some significant year-to-year variation. Not surprisingly, as this was a retrospective survey, they describe difficulties in obtaining consistent and reliable data, and acknowledge the likelihood of under-ascertainment.

The authors are to be commended on undertaking this initiative in perinatal mortality audit and rightly point out that the exercise has educational value and will provide a stimulus for improved local data collection.

Comparing perinatal mortality rates between jurisdictions is fraught with difficulties because of differences and inconsistencies in definitions, in ascertainment and in reporting practices. As the authors point out, the WHO recommends for international comparisons, a definition using an inclusion threshold of 1000 g for stillbirths and a 7 day cut-off for neonatal deaths, and this is probably the best definition for a hospital or a region in a developing country to report. This definition has the advantage that it is amenable to accurate and consistent ascertainment and avoids inclusion of infants of (usually) previable gestational ages, and therefore it is a more reliable indicator of reproductive health and the quality of maternal and neonatal services than PMR based on broader definitions. For a developing country, the lower down the weight and gestational age ladder one goes with the inclusion criteria for PMR, the less useful the PMR becomes as a perinatal indicator. The definitions used in this report from Vila Hospital render comparisons with other institutions or jurisdictions difficult. The inclusion criteria were fetuses and infants of ≥20 weeks and/or 500 g, and neonatal deaths up to 28 days. The Australian Bureau of Statistics (ABS) has (since 1997) used a definition of PMR as ‘deaths of infants and fetuses weighing at least 400 g or having a gestational age of 20 weeks’, but ABS still presents, for purposes of comparison, the PMR using the old definition of ‘deaths of infants and fetuses weighing at least 500 g or having a gestational age of 22 weeks’1. Neither of these definitions matches the one used in Vila, but the second is closer. These caveats notwithstanding, the PMR for Vila Hospital in the year 2000 of 30 per 1000 was about four times higher than the rate for Australia (7 per 1000)1. (It is worth noting however, that the PMR for the Indigenous population of Queensland in 1997 was 24.3 per 1000)2. In addition, the authors quote a 1996 WHO estimation of Vanuatu's PMR as 50 per 1000, supporting the authors’ contention that the PMR for the hospital is lower than that for the whole country. (Using the WHO international definition, the rate for Australia in 1996 was 5 per 1000)1.

The authors also cite a Vanuatu 1999 census estimation of the Infant Mortality Rate (IMR) of 25 per 1000 live births, which compares with the Australian IMR of 5.2 per 1000 live births1. The IMR has fewer definitional and ascertainment problems than the PMR, and is a very useful comparative public health indicator, particularly for interregional comparisons.

The authors also present data on some other outcomes which are useful indicators of basic reproductive health and access and quality of perinatal services, although the results seem counterintuitive. For example, it is hard to understand how a referral hospital in a developing country with a high infant mortality rate would have such an unusually low preterm birth rate (i.e. birth less than 37 weeks gestation), of 3.7%. By comparison, the preterm birth rate in Australia for 2000 was 7.9%3. This suggests either serious under-ascertainment and/or major differences between the hospital population and the population at large. It would be interesting to see the hospital's rate of low birth weight (infants born weighing less than 2500 g). This is a useful and readily ascertained index of perinatal health status, which would be a valuable statistic to record and report over time.

The authors have taken an important first step in acknowledging the value of accurate recording and reporting of perinatal deaths. Once the definition is standardized, Vila hospital and its clients will benefit from the monitoring of its performance over time for this important indicator. The setting up of a systematic approach to investigating all perinatal deaths will enhance the ability to assess preventability and identify categories where systematic improvements can be made. The next step in perinatal mortality audit will involve the classification of perinatal deaths according to antecedent, which invites the opportunity of considering and applying preventive measures. The Perinatal Society of Australia and New Zealand has recently endorsed such a system for national reporting4. However, this ideally involves incorporation of information from expert perinatal autopsy services, which are scarce enough in Australia, and probably nonexistent in Vanuatu. But before that, standard definitions are required and systems must be in place for accurate and systematic ascertainment of all perinatal deaths, with regular multidisciplinary review. This should apply in all maternity institutions, in both developing and well-developed countries, and the results synthesized and presented in regular institutional, regional and national reports.