Is The New WHO Classification System For Stillbirths Feasible In Low-Resource Settings?

Every year, an estimated 2.6 million stillbirths occur, which is over 7,000 deaths every day. The majority of these deaths (98%) occur in low- and middle-income countries (LMIC), with sub-Saharan Africa and South Asia carrying the highest burden.

One of the major challenges of tackling the burden of global stillbirths is the lack of an internationally-agreed classification system for categorizing causes of these deaths. This would allow for aggregation of data and enable comparison between and within settings to inform a more strategic approach to the problem.

The problem is not a lack of classification systems for stillbirth, but that there are too many of them. A previous systematic review indicated that there are over 30 of these classification systems currently in use for stillbirth alone, making any form of global comparison of the cause of death difficult.

Consequently, the World Health Organisation (WHO) developed a new classification system which, if adopted internationally, would finally solve the problem of our inability to compare causes of stillbirth at national and international levels. Known as “the WHO Application of ICD-10 to Deaths during the Perinatal Period” (abbreviated as “ICD-PM”), this new classification system is designed to categorize the causes of both stillbirth and newborn death.

ICD-PM uses a simple approach and ensures that even cases for which only limited information is available, as is the case for many stillbirths occurring in LMIC, would be classifiable at least to some degree. It also allows healthcare providers to assign a cause of death by considering conditions that could affect both the mother and the baby.

There is at the moment very limited information on whether ICD-PM could be feasibly applied in LMIC settings, where clinical records are often poorly kept, incomplete, or both.

To generate this vital information, we, researchers at the Centre for Maternal and Newborn Health (CMNH), Liverpool School of Tropical Medicine (LSTM), carried out a multi-country study (in Kenya, Malawi, Sierra Leone, and Zimbabwe), testing ICD-PM using real-life data.

The study, which was published by PLOS ONE journal, has demonstrated that only a small proportion (16%) of stillbirths could be classified by conditions directly affecting the baby (“fetal cause of death”). However, the majority of cases (76%) could be classified using information relating to conditions affecting the mother (“associated maternal conditions”).

The study also found that half of the stillbirths (50.7%) occurred during labor and childbirth (intrapartum death). This information is essential in designing interventions to tackle the problem.

One of the major challenges we faced in this study was establishing the time of death – that is whether the baby died before or after the onset of labor. This information is required as the first step to the application of ICD-PM before proceeding to classify the cause of death. In low-resource settings, monitoring of fetal heart sound is not always done adequately. Healthcare providers often rely on the physical appearance of the baby to determine the time of death. However, this is unreliable. Thus, the time of death is unknown for many stillbirths. This could be a stumbling block for the application of ICD-PM.

To circumvent this problem, at least partly, we used the combination of information on the fetal heart sound at the time of labor admission as well as the physical appearance to determine the time of death. This reduced the proportion of cases for which ICD-PM could not be applied. Furthermore, we created new categories to capture information about cases whose time of death could not be determined. This helped to avoid further loss of information.

Nonetheless, the techniques we used above do not replace the need for improved clinical care and clinical documentation of care. Healthcare facilities should develop or adapt existing diagnostic protocols and guidelines to improve care. It is also recommended that future revisions of ICD-PM should include an additional category to enable capture of cases for which time of death is unknown.