Aims: To evaluate end of life practices among hospitalised children who died of HIV/AIDS. Design: Retrospective chart review. Setting: A public, secondary and tertiary children's teaching hospital in a developing country. Patients: A consecutive series of in-patient deaths among HIV-infected children. Main Outcome Measures: Identification of patients as dying, presence of do not resuscitate (DNR) orders, documentation of comfort care plans, whether end of life decisions were discussed with parents or caretakers, nature of diagnostic and therapeutic interventions in the last 24 hours of life, and presence of pain and distress in the last 48 hours of life. Results: 165 out of 167 in-patient deaths were reviewed. 79% of patients died in the general wards. The median age of patients was 4 months. The median length of hospitalisation was 6 days. 30% of patients fell in Category B. Patients with shorter lengths of stay were more likely to fall in Category B (median 4 days versus 7 days, P=0.0000). About 1 quarter of patients had a median length of stay of 25 days. 84% of patients had a DNR order, with a median of 4 days between admission and documentation of the order. DNR orders appeared simultaneously in only 41 % of medical and nursing entries. 39% and 63% respectively of doctors did not document their justification for the DNR order or whether it had been discussed with parents. 50% of patients were identified as dying. Terminology such as 'TLC' and 'keep comfortable' designated 44% of patients to receive comfort care only. The median time between admission and identifying a patient as dying and documenting a comfort plan was 5 days and 7 days respectively. In 44% of folders there was no indication of whether the comfort plan had been discussed with parents. 73% and 62% respectively of patients with comfort plans received IV fluids and IV antibiotics in their last 24 hours of life. 55% of patients who died in general wards experienced pain and distress in the last 48 hours of life. Respiratory symptomatology and oral and oesophageal candidiasis accounted for most discomfort. 2 in 5 patients with a comfort plan failed to receive analgesia, despite pain and distress. Conclusions: Despite extreme diagnostic and prognostic uncertainty, doctors made key end of life decisions. Doctors' practices often failed to meet procedural and ethical requirements in professional guidelines. Failure to discuss DNR orders or comfort plans with parents ignores their role as principal decision makers for their children. The low rate of comfort plans, compared to DNR orders, suggests doctors had difficulty making the transition from curative to palliative care. Many comfort plans were incoherent and included interventions neither meant for, nor likely to promote patients' comfort. Whilst fear of hastening death may explain doctors' reluctance to prescribe adequate analgesia, undertreating pain and distress in a dying child is of more concern morally and medically than the risk of suppressing respiratory effort. To achieve better end of life care for HIV-infected children, it will be necessary to improve practice patterns. A structured medical treatment plan that focuses on goals of care is proposed to manage transitions from life-sustaining treatment to palliation.

End-of-life (EOL) care decisions present a challenge for family caregivers. Despite the increasing number of terminally ill patients in need of pain management and comfort care, there is limited qualitative data about how populations in the developing world can access culturally appropriate resources and EOL support. In this phenomenological study, 17 Garifuna family caregivers in southern Belize were interviewed about their experience caring for terminally family members. The conceptual frameworks were Kübler-Ross’s hospice approach and Watson’s theory of human caring. The two theories were selected based on their significance to this process: Kübler Ross’s hospice approach and its impact on the family system during the end stage of life and Watson’s theory of human caring for its emphasis on the impact of the importance of meeting the basic needs of individuals. NVivo 12 was used to code and generate themes for further analysis. Caregivers who said they would not utilize support outside of the home were those who were committed not do so at the request of the patient. Caregivers who cared for their family member and those who could financially afford to hire caregivers in their home to assist with their relative said they would not utilize nonfamily support. All others, regardless of relationship to the patient, indicated they would have accepted care if the environment were safe, caring, and culturally sensitive. Additionally, the caregivers saw their needs as secondary and insignificant compared to the comfort and care of the patient. This study may contribute to positive social change by revealing strategies and services that could be included in the design of a health services delivery system to meet the needs of individuals facing EOL decisions.