INVITED COMMENTARY No resuscitation orders and withdrawal of therapy in paediatric intensive care units CA Ryan Department of Paediatrics and Child Health, University College Cork, Cork, Ireland ‘‘After the first death, there is no other.’’ A Refusal to Mourn the Death, by Fire, of a Child in London. Dylan Thomas The study by Martinot et al. (1), published in this issue of Acta Paediatrica, which was prospective and multicentred, is important in that it provides empirical evidence for what is current practice in end-of-life dilemmas in a selection of paediatric intensive care units (PICUs) in France. Com- bining previous studies with their own experience, Marti- not et al. demonstrated that among the children who died in PICU, death was acknowledged according to brain death criteria in 24%. Over one-quarter (30%) fail to respond to cardiopulmonary resuscitation (CPR) and almost half (46%) have their treatment limited or withdrawn. The similarity in the distribution of modes of death in these studies which include PICUs in the United States, Canada, Britain and mainland Europe, is interesting. The fact that almost half of the children who die in PICUs die from limitation or withdrawal of support suggests that paedia- tric intensive care physicians, throughout the Western world, are willing to make judgements as to when additional or further therapy is considered futile and to act on those judgements. Premature infants were excluded from the French study. However, we are aware that limitation or withdrawal accounts for up to three-quarters of deaths in some neonatal intensive care units (2). Of the three modes of death described above, deci- sions about withdrawal or limitation of support are the most agonizing, mainly because they are made in the absence of prognostic certainty. In order to help practi- tioners and families make informed choices, efforts have been made to develop predictive scoring systems. The Paediatric Risk of Mortality (PRISM) score, an objective validated severity of illness measure, was originally developed [and recently updated (3)] from PICU data- bases in the United States to identify patients with high mortality risks. In their study, Martinot et al. (1) pro- spectively applied this score to deaths in nine PICUs in France, postulating whether it might identify the mode of death. Secondly, these investigators questioned whether children with a pre-existing chronic illness or moderate to severe disability were more likely to die in a certain manner, presumably, more likely to have support limited or withdrawn. Their results showed similar PRISM scores among critically ill children who died of brain death, failed CPR, and limitation or withdrawal of support. It is not surprising that the PRISM score, a predictor of mortality, was not associated with a particular mode of death. While some physicians use prognostication scores as an adjunct in determining futility (4), others are of the opinion that severity scores have no role in clinical decision making, owing to their low sensitivity (5). Most would agree that outcome of paediatric intensive care should incorporate not only survival but should also take into account quality of life, morbidity and disability. It is likely that these and other issues such as patient diagnosis, physician and family agreement on the prognosis (6), communication and family values that determine the mode of death. The French study also indicated that a pre-existing handicap or chronic illness were not associated with a particular mode of death in the PICU. It is tempting to interpret this finding as an indication that quality of life issues are not important considerations in death and dying in PICU settings. Such an interpretation is likely to be incorrect and was rightly avoided by the investigators. One explanation for the observed lack of association between illness or disability and mode of death may be the PICU gatekeeper. Children dying of chronic conditions (e.g. cystic fibrosis, malignancies) and severely disabled children with life-threatening intercurrent illnesses are frequently not supported by intensive care facilities and technology. These ethical dilemmas frequently remain out- side the domain of the PICU. Nevertheless, quality of life assessments are part of the decision making process in adult (4) and neonatal intensive care (2). Unquestionably, similar balancing of the complex tensions between life extension, quality of life and family values also occur in paediatric intensive care settings. While we are aware that withdrawal and limitation of support are an integral part of paediatric intensive care, we need more information on the events surrounding limita- tion or withdrawal of care in PICUs, similar to those described in adult (7) and neonatal intensive care settings (8). Investigators should now address the process, timing and mechanics of dying in the PICU, the involvement of the team and the quality of communication with the family. With this information, consensus approaches might be developed that best protect the interests of the patient, and also help bereaved families to mourn the death of a child. Acta Pædiatr 87: 721–2. 1998  Scandinavian University Press 1998. ISSN 0803-5253