Mortality records were obtained from the New Zealand Health Information Service (NZHIS) for all children who died aged 0–17 years between 1 January 1996 and 31 December 1998. Hospitalisation data for these children were obtained from NZHIS for the period 1995–1998, using linkage via encrypted National Health Index (NHI). These data include any inpatient discharges from public hospitals and publicly-funded discharges from private hospitals. Each case was then classified as requiring palliative care or not using two separate analyses.
Analysis 1 For the primary analysis, a subjective methodology was used to estimate the frequency of children aged 0–17 years requiring palliative care. Information about cause of death was supplemented with hospitalisation data to provide additional clinical information.
All children who died under the age of seven days were excluded, as were those who remained in hospital from birth until death. The following causes of death were excluded: motor vehicle accidents; injuries and poisonings (International Classification of Disease (ICD) codes 800–999); and SIDS (ICD 7980). However, exceptions were made for cases in which there was evidence of sustained brain injury and palliative care may have been required. Deaths due to infection (ICD 001–139) were also excluded, except one case of herpetic meningoencephalitis and one case of aspergillosis. Mental disorders (ICD 290–319) were also excluded after first checking the hospital admission records for any evidence of co-morbid life-limiting conditions.
The remaining cases were assessed on an individual basis, guided by the UK definitions for life-limiting conditions and palliative care3 (Table 1). Provisional judgments were made by the research team; in equivocal cases the final decision was made by a paediatrician with experience of paediatric palliative care. Assessment was based on a summation of all the available evidence, including death registration information, any other conditions identified from hospital admission data, and information about any procedures undergone in hospital.

Cases were considered in four broad disease categories using death registration ICD codes:

Two of these groups presented considerable difficulty in assessing need for palliative care – cancer and cardiac conditions. After consultation with a paediatric oncologist, the records of children with cancer who died in hospital were examined in more detail, in view of the fact that some of these children may have died of an acute, unexpected complication. This resulted in the exclusion of 24 children: 21 with acute lymphoid leukaemia, two with benign neoplasms and one with a neoplasm of uncertain behaviour. It was also decided to exclude six children who died after receiving a bone marrow transplant.
For the group with cardiac anomalies, it was often difficult to determine from the available information whether or not they would have been considered palliative prior to death or would still have been receiving active treatment. A paediatric cardiologist was consulted and reviewed 48 cases randomly chosen from the list of 85 children who died due to cardiac anomalies. There was disagreement on nine of these cases; our estimate of 43% of cardiac cases requiring palliative care compared to the cardiologist’s estimate of 33%. The results presented here reflect our judgments rather than those of the cardiologist.
In order to further test the validity of our analysis, 50 cases were chosen at random from the 2122 mortality records and were reviewed by a general paediatrician. This resulted in agreement on 98% of the cases, providing support for the methodology.
Palliative care need was analysed by age, gender, ethnicity and NZDep96 deprivation decile. Population rates were calculated using 1996 census data; ‘prioritised’ ethnic group was used to assign ethnicity. Other analyses examined seasonality, place of death (home vs hospital), and the proportion of children’s lives spent in hospital. Place of death was determined from combined hospitalisation and mortality data. The time spent in hospital was calculated as a percentage of the time the child was alive during the period 1995 to 1998.
Analysis 2 This analysis was performed in order to allow a direct comparison between New Zealand’s paediatric palliative care requirement and overseas estimates. A recent UK study3 examined deaths of children aged 1–17 years, with palliative care need assessed from the ICD code alone. Need for palliative care was based on a proposed list of life-limiting incurable diseases developed by an expert advisory group. We used this list of disease ICD codes (ICD-9, to the level of three digits) for the purpose of this analysis. Cases in which the underlying cause of death was on the list were deemed to have required palliative care.