- Michael J Goldacre, director (, )
- Stephen E Roberts, statistician,
- David Yeates, computer scientist
- Unit of Health-Care Epidemiology, Department of Public Health, Institute of Health Sciences, University of Oxford, Oxford OX3 7LF
- Correspondence to: M J Goldacre
- Accepted 21 March 2002
The death rate within one year of fractured neck of femur is typically reported as between 20% and 35%.1–3 Performance indicators based on mortality after hospital admission for such fractures have been promoted.4 The only measure of mortality in routine hospital statistics, however, is “in-hospital mortality”—death during the initial admission for the fracture. We analysed inpatient statistics that had been linked to death registration data in the former Oxford NHS health region (population 2.5 million) from 1994 to 1998.
Methods and results
We selected emergency admissions in people aged 65 years and over who had been admitted to eight main acute trusts and for whom fractured neck of femur was the principal diagnosis (international classification of diseases, ninth revision (ICD-9), codes 820, 821.0, and 821.1). We calculated standardised mortality ratios after fracture by applying the age and sex specific mortality in five-year age groups in the whole population of the region (“standard” population) to the number of people with fractured neck of femur in the equivalent age and sex strata, in successive months up to one year after fracture. For each hospital, we calculated case fatality rates for in-hospital deaths within 30 days and for all deaths within 30, 90, and 180 days of admission. We calculated age and sex standardised case fatality rates for each hospital by applying the age and sex specific rates in each hospital to the number of people in each age-sex stratum in the total inpatient population. We calculated case fatality rates separately for deaths certified as fractured femur and for all deaths.
In total, 8148 people aged 65 and over were included (80.2% women; mean (SD 7.2) age 82.2 years). In the first month after fracture the standardised mortality ratio was 1246 (95% confidence interval 1164 to 1331; general population 100). The standardised mortality ratios, adjusted for person months at risk, were 451 (397 to 509) in month 3, 238 (197 to 283) in month 6, and 187 (149 to 230) in month 12. Fractured femur was certified as the underlying cause in 16%, and as a cause anywhere on the death certificate in 43%, of deaths occurring in the first month.
As the table shows, the mortality ranking of hospitals varied with definitions and time frames. Death rates for all causes showed that three hospitals (B, C, and H) had significantly lower rates than hospital A for in-hospital mortality within 30 days, and two (B and C) had significantly lower rates for 30 day mortality regardless of place of death. By 90 and 180 days, differences between hospitals were not significant. Hospitals also changed rank depending on whether deaths from all causes or only those certified as fractured femur were included (table).
The standardised mortality ratios show that mortality is much higher in people after fractured neck of femur than in the general population of comparable age, and they remain raised for many months after fracture. The persistently increased standardised mortality ratio may indicate continuing sequelae of the fracture or that people fracturing their neck of femur are more frail and ill than the general population of similar age.
Measures of prognosis after fracture and comparisons between hospitals are substantially affected by whether death registration data are included, whether time intervals are extended beyond 30 days, and whether deaths that are not certified as fractured femur are included. When death registration data are available, one option is to confine analyses of mortality to the deaths attributed by the certifying clinician to the fracture. Our study confirms, however, that the fracture is often not recorded on death certificates even when death occurs soon after fracture.5 Studies of mortality after fractured femur will be misleading unless they include deaths after discharge from the initial admission and consider all causes of death.
We thank Pamela Evans for preparing the manuscript.
Contributors MJG designed the study and jointly wrote the manuscript. SER contributed to the design, analysed the data, and jointly wrote the manuscript. DY extracted the data, contributed to the design, and commented on the manuscript. MJG and SER are guarantors for the paper.
Funding SER receives funding from the Department of Health (the views expressed in this paper are those of the authors and not necessarily those of the Department of Health). The Unit of Health-Care Epidemiology is funded by the South East regional office of the NHS Executive.
Competing interests None declared.
A table with further data is available on bmj.com