Children and young people who die from cancer: epidemiology and place of death in England (1995-9)

BMJ 2003; 327 doi: (Published 28 August 2003) Cite this as: BMJ 2003;327:478
  1. Irene J Higginson, head of department (irene.higginson{at},
  2. Margaret Thompson, statistician1
  1. 1Department of Palliative Care and Policy, Guy's, King's, and St Thomas's School of Medicine, King's College London, London SE5 9RJ
  1. Correspondence to: I Higginson
  • Accepted 5 August 2003


Survival from cancer among children and young adults has improved, but a need remains to care for the one in four who cannot yet be cured.13 One component of quality care is to provide it in the place of choice,3 usually home; we analysed factors affecting place of death.

Descriptive statistics for cancer deaths registered in England and Wales in 1995-9, for children and adolescents (age range 0-15) and young adults (age range 16-24). Values are numbers (percentages) unless otherwise indicated

View this table:

Participants, methods, and results

We derived data from death registrations for all cancer deaths (international classification ofdiseases, 9th revision, codes 140-239) in England and Wales, for 1995-9, for ages 0-24. Age, sex, social class, country of place of birth, geographical location, underlying cause of death, and place of death were available directly. Further potential explanatory variables were a classificationof local authority type(rural, urban, inner London, etc) from the Office for National Statistics and deprivation indices for 2000, from the Department of Environment, Transport, and the Regions, at parliamentary ward level(treated as continuous variables, higher scores indicated greater deprivation). We defined two further variables: diagnosis (main primary tumour) and whether or not the cancer was a solid tumour.

Response variables indicated whether death took place at home or in hospital, or in a hospice or palliative care unit. We analysed age groups 0-15 (children and adolescents) and 16-24 (young adults) separately. Social class, recategorised on a scale (I-VI, table) and treated as a continuous variable, was based on parents for age range 0-15 years. We used binary logisticregression models in our exploratory analysis to examine associations between place of death and potential explanatory variables. We used multiple logistic regression models to determine the jointeffect of variables identified as significant or borderline (P = 0.05-0.1) during exploratory analysis. Odds ratios with 95% confidence intervals are presented.

During 1995-9 a total of 3197 deaths from cancer in young people aged 0-24 were registered in England. Of the small number of deaths in hospices, 23 (42.6%) among children and adolescents and 34 (24%) among young adults were from brain cancer (table).

A multivariate model for home death, for children and adolescents, indicated that death at homerather than in hospital was less likely for those at the bottom of the social scale (odds ratio 0.93, 95% confidence interval 0.87 to 0.99)–home deaths were 65% in social class I, 49% in V, 44% inVI), with leukaemia or lymphoma rather than solid tumours (0.46, 0.37 to 0.58) less likely in inner London (0.54, 0.32 to 0.92–for example, 62% “prosperous” England, 58% rural, 37% inner London) and in areas with high rates of child poverty (0.99, 0.99 to 1.00), all P < 0.05. This was consistent with associations found in univariate analysis. The only potential explanatory variable thatwas significantly associated with dying in a hospice was a diagnosis of brain cancer (2.52, 1.45 to 4.38, P = 0.001).

Among young adults multivariate analysis found that home death was less likely with increasing age (0.91, 0.80 to 0.95), less likely for young women (0.73, 0.56 to 0.94), patients with leukaemia or lymphoma rather than solid tumours (0.22, 0.97 to 0.30), and where access to local services (school, shops, general practitioner) was good (1.25, 1.05 to 1.49), all P < 0.05.


Home is an important place of death; 52% of children and adolescents, and 30% of young adults died at home. This is higher than for the United States (20%)4 and for adults (26%).5 Primary care and community servicesare therefore critical, although individual services encounter patients rarely. Lower social class,living in inner London, or living in an area of highchildhood poverty reduced the likelihood of home death. Although relatively few children died in hospices, brain tumours accounted for around half of these. These findings are relevant to service planning and need investigation in prospective research.


We thank the Office for National Statistics for its helpful collaboration, making available these datasets, and guiding us in their interpretation.


  • Contributors IJH conceived the idea for the study, was responsible for day to day management, analysis plan, literature review, and interpretation; MT was responsible for liaison with the Office for National Statistics, data checking, recoding, merger of datasets, and data analysis. Both contributed to the writing of the manuscript and act as co-guarantors for the paper.

  • Funding King's College London and the Interdisciplinary Research Group on Palliative and Person Centred Care funded MT to undertake the analysis

  • Competing interests None declared


View Abstract