Neonatal mortality rate (NMR)
The number of neonatal deaths per 1000 live births
A neonatal death is defined as a death during the first 28 days of life (0-27 days).
|# of neonatal deaths x 1000
Total # of live births
The NMR is often broken down into early and late mortality rates. The Early Neonatal Mortality rate (ENMR) is calculated as follows:
|# of neonatal deaths 0-7 days x 1000
Total # of live births
The late neonatal mortality rate (LNMR) is calculated as follows:
|# of neonatal deaths 8-27 days x 1000
Total # of live births
Number of neonatal deaths in a given population and reference period and number of live births in the same population and reference period
Vital registration; population-based surveys; services statistics
Where data on the numbers of live births for the denominator are unavailable, evaluators can calculate total estimated live births using census data for the total population and crude birth rates in a specified area. Total expected births = population x crude birth rate
The gold standard for NMR data collection is a well-functioning vital registration system. Routine HIS may collect data for this indicator to obtain estimates of the NMR for facilities. Facility data are not recommended for estimating the NMR for the general population, because in many settings, many neonatal deaths and live births occur outside the health system, which will cause substantial selection bias.
The NMR is a key outcome indicator for newborn care and directly reflects prenatal, intrapartum, and neonatal care. In addition, as infant mortality rates decline, the proportion of infant deaths that occur in the neonatal period typically increases. The NMR differs from the perinatal mortality rate in that it focuses only on deaths among live births and covers a longer period after birth. Information on live births is generally thought to be easier to obtain than information on non-live births and is more widely available, because many population-based surveys such as the DHS typically only collect information on live births. Early neonatal deaths are more closely associated with pregnancy-related factors and maternal health, whereas late neonatal deaths are associated more with factors in the newborn‘s environment.
In many countries, vital registration data are not sufficiently complete to allow reliable estimation of the NMR. The standard techniques for collecting data on live births and neonatal deaths in population-based surveys have been widely applied in programs such as the World Fertility Survey and DHS. Data quality is an important issue; common problems include omission of deaths, particularly very early neonatal deaths, and heaping of the reported age at death on 7, 28, or 30 days. Heaping on these digits is particularly problematic because it will lead to the misclassification of early neonatal deaths as late neonatal death (7 days) or late neonatal deaths as post-neonatal deaths (28 and 30 days).
Evaluators typically calculate NMR at a national or international level. They may also obtain sub-national estimates if sample sizes are sufficiently large. The NMR is sometimes calculated at a facility level to monitor the outcome of delivery and newborn care in health facilities. Reliable estimates for individual facilities can only be obtained for very large facilities with large numbers of deliveries and neonatal admissions.
The NMR may respond fairly quickly to programmatic interventions, for example, immunizing all pregnant women in areas of high tetanus prevalence. However, survey-based estimates are generally subject to relatively large sampling errors, so it is impossible to detect changes over short periods of time unless the changes are quite large. Also, survey-based estimates are often based on a five-year period prior to the survey. Therefore, we recommend collecting survey-based estimates of the NMR not more than every three to five years.
One limitation of note is the NMR‘s sensitivity to changes in the quality of data. For example, a rise in the NMR may indicate deterioration in newborn health outcomes, or it may indicate an improvement in the reporting of neonatal deaths. Therefore, assessing data quality is essential to analysis.
Also, evaluators should interpret comparisons of facility-based estimates of the NMR very carefully because the NMR in a facility is very sensitive to the case mix of deliveries and neonatal admissions. One should not interpret a higher NMR in one facility as suggesting that the quality of neonatal care is worse in this facility because the NMR may rise or fall in response to changes in the case-mix. Additionally, improvements in prenatal and intrapartum care and advances in medical technology may increase the NMR because babies who may otherwise have been stillbirths may survive delivery only to die in the neonatal period. For these reasons, we recommend that evaluators break down facility-based estimates of the NMR by birth weight (see Birth weight specific mortality rate) and by admission status (direct admission or transfer-in) as a proxy for case mix.