Perinatal mortality rate (PMR)

The number of perinatal deaths per 1000 total births

A perinatal death is a fetal death (stillbirth) or an early neonatal death.

The perinatal mortality rate is calculated as:

(# of perinatal deaths / total # of births (still births + live births)) x 1000

A stillbirth is the death of a fetus weighing 500g or more, or of 22-weeks gestation or more if weight is unavailable (ICD 10).

An early neonatal death (END) is the death of a live newborn in the first 7 days (i.e., 0-6 days) of life.

Great variation exists both between and within countries on how the stillbirth component of perinatal mortality is recorded, particularly for early stillbirths that occur at 22- to 27-weeks gestation. For international comparisons, WHO suggests including only deaths of fetuses weighing at least 1000g, or of 28-weeks gestation or more if weight is unavailable. Presentations of the PMR should include a clear statement of the definition of perinatal mortality used. In practice, in most developing countries accurate data on birth weight or gestational age are difficult to obtain.

Number of perinatal deaths in a given population in a given reference period(i.e., 12 months) and number of births (live births + stillbirths) in the same population and reference period

Population-based surveys; vital registration; service statistics

Routine HIS may collect data for this indicator to obtain estimates of the PMR for facilities. Facility data are not recommended for estimating the PMR for the general population because in many settings, many perinatal deaths and live births occur outside the health system, which will cause substantial selection bias.

The PMR is a key outcome indicator for newborn care and directly reflects prenatal, intrapartum, and newborn care. It has also been proposed as a proxy measure of maternal health status and mortality, but a recent study has cast doubt on its use as a proxy for maternal mortality (Akalin et al., 1997).

Because the PMR includes both fetal deaths and deaths in the first week of life, it avoids conflicting judgments as to whether a fetus exhibited signs of life and variations in administrative practice regarding whether or not a death should be counted. In many countries, however, vital registration data are not sufficiently complete to allow reliable estimation of the PMR. Techniques now exist for collecting data on stillbirths, live births, and early neonatal deaths in population-based surveys (pregnancy histories) and applied in surveys including the DHS. However, there has been relatively less experience with pregnancy histories than with birth histories because of concerns about the quality of retrospectively reported pregnancy histories. Common problems with data quality include:

  • Omission of stillbirths and early neonatal deaths;
  • Difficulty in obtaining accurate information on gestational age or birth weight leading to the misclassification of stillbirths as late spontaneous abortions; and
  • Heaping of the reported age at death of live births on 7 days, leading to the misclassification of early neonatal deaths as late neonatal deaths.

Prospective population-based surveys of pregnant women provide better quality data, but are expensive to undertake.

Evaluators typically calculate the PMR obtained from large population-based surveys at a national level and may aggregate data across countries to obtain a global or UN subregion statistic. Evaluators may also obtain sub-national estimates if sample sizes are sufficiently large.

The early neonatal component of the PMR may respond relatively quickly to programmatic interventions, for example, following the introduction of elements of the WHO Essential Newborn Care Package. The stillbirth component may decline more slowly because it depends more on interventions that influence primarily maternal health and on the availability of technologies such as cesarian section. Survey-based estimates are generally subject to relatively large sampling errors, so detecting changes over short periods of time is impossible unless the changes are quite large. Also, retrospective survey-based estimates are often based on a five-year period prior to the survey. Therefore, evaluators should collect survey-based estimates of the PMR not more than every three to five years.

The following caveats bear mention. The PMR is sensitive to changes in the quality of data. For example, a rise in the PMR may indicate deterioration in perinatal outcomes, or it may indicate an improvement in the reporting of perinatal deaths. Therefore, an assessment of data quality is an essential component of analysis. In this context, evaluators often find it useful to separate the PMR into its two components: stillbirths and early neonatal mortality. Data quality is generally more problematic for stillbirths than for early neonatal deaths, because the problems of obtaining gestational age and ambiguity over the definition of stillbirths and fetal deaths are much less likely to be reported than deaths of live births (WHO, 1996b).

Evaluators should interpret facility-based estimates of the PMR with caution. The PMR in a facility is very sensitive to the types of deliveries occurring in the facility. Consequently, it may rise or fall in response to changes in the complexity of deliveries in the facility. In small facilities, the PMR will be very unstable because of the small number of deliveries and perinatal deaths; thus, the PMR is ineffective for monitoring change over time within the facility.

share this