BackgroundCardiac arrest is the final common step in the dying process. In the right context, resuscitation can reverse the dying process, yet success rates are low. However, cardiopulmonary resuscitation (CPR) is a highly invasive medical treatment, which, if applied in the wrong setting, can deprive the patient of dignified death. Do-not-attempt-cardiopulmonary-resuscitation (DNACPR) decisions provide a mechanism to withhold CPR. Recent scientific and lay press reports suggest that the implementation of DNACPR decisions in NHS practice is problematic.Aims and objectivesThis project sought to identify reasons why conflict and complaints arise, identify inconsistencies in NHS trusts’ implementation of national guidelines, understand health professionals’ experience in relation to DNACPR, its process and ethical challenges, and explore the literature for evidence to improve DNACPR policy and practice.MethodsA systematic review synthesised evidence of processes, barriers and facilitators related to DNACPR decision-making and implementation. Reports from NHS trusts, the National Reporting and Learning System, the Parliamentary and Health Service Ombudsman, the Office of the Chief Coroner, trust resuscitation policies and telephone calls to a patient information line were reviewed. Multiple focus groups explored service-provider perspectives on DNACPR decisions. A stakeholder group discussed the research findings and identified priorities for future research.ResultsThe literature review found evidence that structured discussions at admission to hospital or following deterioration improved patient involvement and decision-making. Linking DNACPR to overall treatment plans improved clarity about goals of care, aided communication and reduced harms. Standardised documentation improved the frequency and quality of recording decisions. Approximately 1500 DNACPR incidents are reported annually. One-third of these report harms, including some instances of death. Problems with communication and variation in trusts’ implementation of national guidelines were common. Members of the public were concerned that their wishes with regard to resuscitation would not be respected. Clinicians felt that DNACPR decisions should be considered within the overall care of individual patients. Some clinicians avoid raising discussions about CPR for fear of conflict or complaint. A key theme across all focus groups, and reinforced by the literature review, was the negative impact on overall patient care of having a DNACPR decision and the conflation of ‘do not resuscitate’ with ‘do not provide active treatment’.LimitationsThe variable quality of some data sources allows potential overstatement or understatement of findings. However, data source triangulation identified common issues.ConclusionThere is evidence of variation and suboptimal practice in relation to DNACPR decisions across health-care settings. There were deficiencies in considering, discussing and implementing the decision, as well as unintended consequences of DNACPR decisions being made on other aspects of patient care.Future workRecommendations supported by the stakeholder group are standardising NHS policies and forms, ensuring cross-boundary recognition of DNACPR decisions, integrating decisions with overall treatment plans and developing tools and training strategies to support clinician and patient decision-making, including improving communication.Study registrationThis study is registered as PROSPERO CRD42012002669.FundingThe National Institute for Health Research Health Services and Delivery Research programme.