Giving a voice to millions: developing the WHO application of ICD‐10 to deaths during the perinatal period: ICD‐PM

In global health monitoring mechanisms, perinatal deaths have long been largely invisible and poorly counted. Millions of stillbirths occur annually unrecognised by death certification, or vital classifications.1, 2 In addition to this, neonatal deaths contribute an enormous burden to mortality in the under-5s.3 The outcomes of both of these groups are linked with care around the time of childbirth, and by extension, condition and care of the mother.4, 5 Advancing progress on perinatal mortality therefore requires us to embrace a unifying perinatal death classification system that recognises stillbirths and neonatal deaths together with the contributing maternal conditions in these cases. Accurately capturing and classifying the causes of these deaths across the globe will lead to better identification of trends, gaps and efforts for prevention and management of conditions. To meet this need, The World Health Organization (WHO) application of ICD-10 to deaths during the perinatal period: ICD-PM has been developed to be released in August 2016.6 This globally applicable system for classifying perinatal mortality brings together the causes of perinatal death, separated by timing of the death, with the maternal conditions that contribute to perinatal death. This way the mother–baby dyad is recognised, and areas where programmatic interventions that could benefit both mother and baby become more clear. In addition to explaining the development of ICD-PM, this paper will introduce the associated mini-series by highlighting the take-home messages from each paper in the series (Box 1). Development of ICD-PM and our pilot testing is described in this mini-series of four articles. Paper 1 details the development of the ICD-PM.7 Paper 2 demonstrates the application of ICD-PM, using two perinatal death databases from South Africa and the UK.8 In paper 3, we use preterm neonatal deaths as an informative case study of a challenging condition where ICD-PM allows better use of data around the causes of death and contributing factors.9 Paper 4 explores the contributing maternal conditions and benefits of changes to the coding rules in the upcoming ICD-11, such that ICD-PM and ICD for maternal mortality (ICD-MM) would use the same maternal codes, which is currently not the case.10 The WHO Application of ICD-10 to both maternal and perinatal deaths was first conceptualised in 2008 ‘to facilitate the consistent collection, analysis and interpretation of information’ on causes of maternal and perinatal deaths.11 The need for an international classification system for perinatal mortality that incorporates key characteristics such as recognising both the maternal condition and the perinatal condition in all perinatal deaths was called for in a 2010 review.12 The WHO Application of ICD-10 to maternal deaths during pregnancy, childbirth and the puerperium—the ICD-MM—was developed first and published in 2012.11 Following this, the ICD-PM was further developed through a consultative process over 2014–15. A technical working group of experts in perinatal death classification was convened in July 2014, and resulted in a consensus decision on the underlying structure of ICD-PM. This working group agreed that the system should be globally applicable, follow existing ICD-10 rules, and be compatible with the upcoming ICD-11. As part of the background work for the development of ICD-PM, a systematic review of existing classification systems was undertaken (S. H. Leishe, Z. Teoh, H. Reinebrant, E. R. Allanson, H. Blencowe, J. J. H. M. Erwic, et al., under review-a), with more than 80 systems identified as being used (regardless of when they were created) for the period between 2009 and 2014. In addition to this, a Delphi survey of experts from 21 countries was conducted, which identified 17 key characteristics necessary for a perinatal death classification system.13 The relationship between these characteristics and existing classification systems was also explored (S. H. Leishe, Z. Teoh, H. Reinebrant, E. R. Allanson, H. Blencowe, J. J. H. M. Erwic, et al., under review-b). Following the initial development of the ICD-PM, pilot testing was undertaken on two perinatal death databases in South Africa and the UK, which served several purposes: 1. allowed verification and detailed description of the data collection procedures to be developed, 2. provided insights into the ease of use of the proposed system, and 3. aided in the identification of any specific issues that may require detailed outlining for users of the final ICD-PM.8 In November 2015, WHO convened a larger technical working group of obstetricians, neonatologists, epidemiologists and public health professionals from low-, middle- and high-income countries to finalise the ICD-PM and prepare the ICD-PM standard guide for capturing information relating to perinatal deaths. The first step in developing ICD-PM was to agree on the importance of separating perinatal deaths into three distinct time periods; antepartum, intrapartum and neonatal. ICD-PM is mainly designed to be used for all antepartum, intrapartum and early neonatal deaths, and can be used for late neonatal deaths. This information is the first step in global comparability, as accurate capture of the time of death can allow for assessment of trends, even if no other information is captured, which may be the case in poorly resourced settings where the burden of disease is the greatest,14 and this can be used to make international comparisons as well as programmatic decisions to focus local efforts and interventions. It can also help to address the documented challenges to distinguishing intrapartum stillbirths from early neonatal deaths,15 ensuring that both are captured and correctly classified in a range of settings. Moreover, delineating intrapartum from antepartum stillbirth (which in many settings is possible but not always done,2, 16 can improve the accuracy of estimates of the stillbirth burden, which are likely to be largely preventable if quality of care can be improved.2, 12 As a second step, ICD-PM groups the main condition in the fetus or infant into a limited number of cause-of-death categories under the headings of timing of death (Table 1). There are six groups of antepartum causes of death, designated by a leading ‘A’; seven groups of intrapartum causes of death, designated by a leading ‘I’; and 11 groups of neonatal causes of death, designated by a leading ‘N’. All of the ICD-10 codes that can be assigned to the perinatal cause of death on a death certificate are represented in these new groupings. The ICD-10 codes have been re-ordered and clarified to better represent the pathologies at different times of perinatal death. Codes that are not considered to be a cause of perinatal death in these sections have been excluded from the ICD-PM groupings. Finally, the maternal condition contributing to the perinatal death is captured for every case. The five existing ICD-10 groups of maternal conditions in perinatal death have been reordered in to groups 1–4, denoted with a leading ‘M’: the complications of placenta, cord and membranes, maternal complications of pregnancy, complications related to labour and delivery, and the medical and surgical conditions that may or may not be related to the present pregnancy (e.g. pre-eclampsia or pre-existing hypertension). A fifth group (M5) was added to this; ‘no maternal condition’, as we considered it critical to capture the scenario where there is no maternal condition identified contributing to the perinatal death. Capturing the maternal condition for each perinatal death ensures that providers reviewing a death always consider the role of the maternal condition or lack thereof. Moreover, the latter may highlight the potentially preventable factors related to the care provided around the time of childbirth. In any perinatal death classification system, there are disease processes and conditions within perinatal mortality that may challenge clinicians, researchers and developers of classification systems alike. Through consultation with experts and feedback through the pilot testing, the ICD-PM document, which will be launched in August 2016, provides clear guidance on how to address such issues as growth restriction, prematurity, preterm labour, hypoxia, obstructed labour, maternal HIV/AIDS and termination of pregnancy. Moreover, ICD-PM will be launched along with a WHO guide on perinatal audit. Both of these documents aim to strengthen the data captured around perinatal deaths, allowing for targeted comparison, assessment and planning of interventions. As part of the development, pilot testing of the ICD-PM system on databases from South Africa and the UK demonstrated that the system is applicable in differently resourced settings, where different levels of investigation are available to determine the cause of a perinatal death and the contributing maternal conditions. In the first analysis,8 the timing of perinatal deaths was clearly identified in each setting, and the main perinatal causes of death were highlighted. Moreover the application of ICD-PM to these deaths allowed the linking of the maternal conditions (commonly hypertensive disorders of pregnancy, and placental abruption), to perinatal outcomes such as hypoxia, problems of fetal growth and prematurity. The second analysis in our pilot testing was an in-depth exploration of a challenging condition where better use of data around the causes and contributing factors is essential given the high burden. Therefore, preterm neonatal deaths from the South African and UK databases were analysed to demonstrate the additional benefits of applying ICD-PM to extract detailed information about the causes of deaths and the contribution of maternal conditions.9 The final step was a demonstration that an alignment between the codes used for maternal condition in ICD-MM and ICD-PM increases the clarity, specificity and applicability of ICD-PM.10 The ICD codes for the maternal condition in a perinatal death are broad and largely non-specific. As such, there are only 53 codes for maternal conditions in ICD-PM in comparison with the 702 codes used for maternal conditions in ICD-MM. The ICD-MM codes are applicable to the maternal conditions in perinatal death, and the ultimate aim is that ICD-PM and ICD-MM use the same classifications and codes for maternal conditions regardless of whether the outcome was a maternal or a perinatal death. This requires a change in ICD rules, which we advocate to see introduced in the upcoming revision to ICD-11. The ICD-PM uses the existing ICD-10 as a framework for a system that is globally applicable, has ease of use across settings, and addresses the needs of clinicians, programme officers, public health workers, mothers and babies. It is ultimately a programmatically driven system that focuses attention on the areas where intervention is more likely to change outcomes and lead to a triple return on investment by focusing care on both mother and baby around the time of childbirth.17 As we move towards the sustainable development goals, where the aim is not only to survive but also to thrive,18 this is an investment that we cannot afford to ignore. Full disclosure of interests available to view online as supporting information. EA and OT drafted the manuscript. JG and RCP provided access to the tested databases. EA, OT, JG, AF, RCP, JPV, JJHM, VJF, JFF, JN, AQ, DC, MM, LS, AMG reviewed the drafts and approved the final version of the manuscript. Emma Allanson is a PhD candidate funded by the University of Western Australia with an Australian postgraduate award, and an Athelstan and Amy Saw Medical top-up scholarship, and by the Women and Infants Research Foundation with a Gordon King Doctor of Philosophy scholarship. We would like to acknowledge the input of the following participants of the technical consultation, which took place on 10–11 November 2015 in Geneva, Switzerland: Hannah Blencowe, Rogelio Gonzalez, Robert Jakob, Kate Kerber, Edward A. Liechty, Cynthia Pileggi-Castro, Cleo Rooney and Khalid Yunis. The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the institutions to which the authors are affiliated. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.