Intended for healthcare professionals


Mortality in young people admitted to hospital for diabetes: database study

BMJ 2004; 328 doi: (Published 25 March 2004) Cite this as: BMJ 2004;328:741
  1. Stephen E Roberts (stephen.roberts{at}, epidemiologist1,
  2. Michael J Goldacre, professor of public health1,
  3. H Andrew W Neil, reader in clinical epidemiology2
  1. 1Unit of Health-Care Epidemiology, Department of Public Health, University of Oxford, Oxford OX3 7LF
  2. 2Department of Public Health, University of Oxford
  1. Correspondence to: S E Roberts
  • Accepted 27 November 2003


People with type 1 diabetes mellitus have an increased risk of death at a young age. We aimed to quantify this risk in a population of people admitted to hospital for type 1 diabetes and to determine whether mortality in the three year period after admission has fallen in recent years.

Methods and results

From 1968 to 1996, for people aged under 30 years in the area covered by the Oxford record linkage study (population of 0.9 million in 1968, 1.9 million in 1974, and 2.5 million in 1987) we analysed all NHS hospital admissions for diabetes mellitus (ICD-9, code 250) in which that condition was coded as the principal diagnosis. The database included linkage to death certificate data to 1999.

The coding of type 1 diabetes is uncommon on hospital records. We chose age 29 as our upper age limit because almost all people under 30 admitted with diabetes mellitus would have had type 1 diabetes. We analysed multiple admissions for the same person as follows. Each person's first admission was identified and followed for three years. At the end of a person's three year follow up, any subsequent admission was included for a “new” period of three year follow up. Standardised mortality ratios were calculated for the study population.

There were 4992 admissions (2603 (52.1%) male) for diabetes among people aged under 30 years. There were 58 deaths during the three year follow up period (standardised mortality ratio 8.5; 95% confidence interval 6.5 to 10.8; table), including 32 in the first year (14.1; 9.6 to 19.4) and 15 during the first hospital admission. Standardised mortality ratios were 27.9 (14.8 to 45.2) at one year and 12.9 (7.6 to 19.5) at three years for the 1010 patients who had been recorded at admission as having diabetic ketoacidosis or coma.

Number of hospital admissions for diabetes in people aged <30 years, 1968-1996, with number of deaths, mortality per 1000 admissions, and standardised mortality ratios during three year follow up period. Values in parentheses are 95% confidence intervals

View this table:

The underlying cause of death or coroners' verdicts were diabetes mellitus (29 cases), other diseases (14), suicide (9) and accidents (6). Three year standardised mortality ratios were 818 (547 to 1143) for diabetes mellitus, 12.6 (9.1 to 16.6) for all natural causes, 11.7 (5.3 to 20.6) for suicide, and 2.2 (0.4 to 5.5) for accidents. Sex specific standardised mortality ratios for suicide were 5.0 (0.9 to 12.3) in men and 35.2 (12.7 to 69.1) in women (1). All cause death rates per 1000 admissions declined a little, but not significantly, over the study period; standardised mortality ratios were largely unchanged.


Young people admitted to hospital for diabetes have an increased risk of death in the following three years, not only from natural causes but also from suicide. Although, in absolute terms, death in young people with diabetes is uncommon, standardised mortality ratios showed that death within three years of hospital admission was nine times more common than in the general population. These ratios were higher than those of 2 to 4 recently reported from population based diabetes cohorts and registers.13 This suggests that people with type 1 diabetes who need hospital admission are at much greater risk of mortality, at least in the short term after care, than the population of people with type 1 diabetes as a whole. Increased risk of suicide has previously been reported in men with diabetes,3 4 but we found an even higher risk in young women.

Other studies have reported improvements in prognosis in recent decades for people with type 1 diabetes.1 5 We found no appreciable improvement, however, in young people admitted to hospital for diabetes in the past 30 years. Because methods for glycaemic control and the delivery of insulin therapy have improved over time, the proportion of people admitted with diabetes whose condition is difficult to control is unlikely to have increased. Survival of young people with type 1 diabetes whose disease was serious enough to warrant admission is therefore not likely to have improved much.

See also p 750

Embedded ImageFurther details about the study's methods and results are at


Leicester Gill, Glenys Bettley, and Myfanwy Griffith built the database. We thank Myfanwy Griffith and David Yeates for programming.


  • Contributors SER and MJG designed the study and wrote the first draft of the manuscript; SER analysed the data; and HAWN contributed to study design, interpretation, and further drafts. SER and MJG are guarantors for the paper.

  • Funding The Oxford record linkage study was a project of the former Oxford Regional Health Authority. The Unit of Health-Care Epidemiology is funded by the Department of Health to analyse the linked data.

  • Conflict of interest None declared.

  • Ethical approval The historical data files were built with approval from the Oxford Region Data Protection Steering Group and the Health Authorities' Caldicott Guardians; and are wholly anonymised. Ethical approval was not needed for analysis of anonymised statistical datasets.


View Abstract