Age specific trends in asthma mortality in England and Wales, 1983-95: results of an observational studyBMJ 1997; 314 doi: https://doi.org/10.1136/bmj.314.7092.1439 (Published 17 May 1997) Cite this as: BMJ 1997;314:1439
- M J Campbell (), reader in medical statisticsa,
- G R Cogman, MSc studenta,
- S T Holgate, MRC professor of immunopharmacologyb,
- S L Johnston, senior lecturer in medicineb
- a Southampton University Department of Medical Statistics and Computing, Southampton General Hospital, Southampton S016 6YD
- b Southampton University Department of Medicine, Southampton General Hospital
- Correspondence to: Dr Campbell
- Accepted 6 March 1997
Objective: To determine trends in asthma mortality by age group in England and Wales during 1983-95.
Design: Observational study.
Setting: England and Wales.
Subjects: All deaths classified as having an underlying cause of asthma registered from 1 January 1983 to 31 December 1995.
Main outcome measure: Time trends for age specific asthma deaths.
Results: Deaths in the age group 5-14 years showed an irregular downward trend during 1983-95; deaths in the age groups 15-44, 45-64, and 65-74 years peaked before 1989 and then showed a downward trend; and deaths in the age group 75-84 years peaked between 1988 and 1993 and subsequently dropped. Trends were: age group 5-14 years, 6% (95% confidence interval 3% to 9%); 15-44 years, 6% (5% to 7%); 45-64 years, 5% (4% to 6%); 65-74 years, 2% (1% to 3%). Deaths in the 75-84 and 85 and over categories plateaued.
Conclusions: There are downward trends in asthma mortality in Britain, which may be due to increased use of prophylactic treatment.
Asthma mortality in England and Wales is dropping by about 6% a year in people aged 5-64 years
It is changing only slowly in those aged 65 and over
It is now over 10 years since trends and seasonality in asthma mortality were investigated in England and Wales,1 2 3 and the long term trends over the period 1974-84 were found to be increasing. We aimed at updating these findings in the light of changing views on the appropriateness of treatment for asthma, increasing concerns in relation to environmental effects on asthma, and also concerns on admission rates for asthma.4
The data were obtained from the Office of Population Censuses and Surveys (now the Office for National Statistics). They consisted of all deaths classified as having an underlying cause of asthma (International Classification of Diseases, ninth revision (ICD-9), codes 493.0 to 493.9 inclusive) registered in England and Wales from 1 January 1983 to 31 December 1995. Some coding procedures were changed in 1983, so the data were analysed including and excluding that year to see if the change affected conclusions. Data included the date of death, cause of death, sex of subject, and age at death. Annual age specific population sizes and death rates for all respiratory deaths (ICD-9 codes 460-519) were obtained from Office of Population Censuses and Surveys publications.
For analysis the data were split into seven age groups, coinciding with the Office of Population Censuses and Surveys age classifications 0-4, 5-14, 15-44, 45-64, 65-74, 75-84, and 85 years and over. The number of people alive during each year in these age groups was obtained from census projections. Deaths were aggregated into years. The appendix describes the method of analysis.
In total, 23 311 asthma deaths were registered between 1 January 1983 and 31 December 1995. The proportions of these deaths in each age group 0-4, 5-14, 15-44, 45-64, 65-74, 75-84, and 85 and over were 0.5%, 1%, 12%, 27%, 26%, 24%, and 10% respectively.
Figure table 1) shows the yearly asthma death rates plotted on a log scale for each age group. Table table 1) gives the results of the Poisson analysis. Little could be made of the data for the under 5s because the numbers were too small. For all age groups except 5-14 there was a significant quadratic term. For each age group between 15 and 74 both linear and quadratic terms were negative, implying an accelerating decline which started before 1989. For subjects aged 75-84 the linear term was positive, suggesting that mortality initially rose and then either peaked or plateaued after 1989. Visual inspection of the data suggests that for those aged 75-84 there was a drop in 1994 and 1995, but for those aged 85 and over no decline was evident.
Excluding 1983 made little difference to the results. As the model was on a log scale, disregarding the quadratic term we can interpret the linear coefficients as a proportionate drop. Hence for deaths between 5 and 14 years of age the drop was about 6% a year (95% confidence interval 3% to 9%), for deaths at ages 15-44 years it was also 6% (5% to 7%), for deaths at ages 45-64 years it was 5% (4% to 6%), and for deaths at ages 65-74 years it was 2% (1% to 3%). For deaths in subjects aged 75 and over the rate was flat.
Total respiratory deaths also decreased after about 1991 in subjects aged 5-14 and 15-44. In 1991 as a proportion of all respiratory deaths asthma deaths accounted for 44% among subjects aged 5-14 years, 31% among those aged 15-44, 11% among those aged 45-64, and 2% among those over 65.
In contrast with studies up to the mid-1980s1 2 3 which showed increasing mortality, we have shown that since the late 1980s asthma mortality in England and Wales seems to have dropped except among people aged over 85. Other countries have had different experiences. In Scotland mortality was stable between 1975 and 1989 for 5-44 year olds.5 In France a peak in mortality for both the under 35s and over 35s was observed from 1985 to 1987; this was attributed to influenza epidemics, and though death rates were lower subsequently there was no evidence of a trend.6 In contrast, deaths in New Zealand at ages 5-54 years showed a downward trend from 1986 to 1992.7
Trends in asthma mortality among children are reflected in the trends in asthma admissions to hospital. Routine data for asthma admissions are not available for England after 1985, but for Wales admissions showed a steady rise from 1983 to 1988 and then a drop in 1989 and 1990.4 Age standardised death rates seem to reflect this trend.
There has been considerable concern in the medical community and in the public domain over reports of increasing asthma prevalence.8 It is encouraging to note that even with a background of increasing prevalence there were downward trends in mortality in the under 75s in the five to seven years before 1995. It is possible that these trends were a result of increased awareness among physicians and patients of the inflammatory basis of asthma and the need for prophylactic treatment, particularly in view of the increased prescribing of corticosteroids as a proportion of all prescriptions for asthma.9 Diagnostic transfer is a possibility, but it is reassuring that respiratory deaths are also dropping in younger people. The accuracy of death certification in asthma was not good in 1979 but has been shown to be more accurate for younger people.10 In old people asthma deaths form only a small proportion of all respiratory deaths and it would be impossible to quantify the extent of diagnostic transfer.
The trends in asthma mortality may be related to the increased use of prophylactic treatment, the use of which should continue to be encouraged.
We thank the Office for National Statistics (formerly the Office of Population Censuses and Surveys) for the data and Mr S A Julious for additional references.
Conflict of interest: None.
Analysis was by Poisson regression in STATA.11 The midyear population was included as an “offset,” which ensured the analysis was based on a rate and allowed for changes in the age structure of the population. Checks for overdispersion and serial correlation of the residuals were carried out but it was not found necessary to make allowances for them. Trend terms were centred on the midyear 1989. When the linear and quadratic terms were negative this implied that the fitted model peaked before 1989. When the linear term was positive and the quadratic term was negative this implied that the model had peaked or plateaued or was expected to peak or plateau after 1989. When both coefficients were positive an increasing and accelerating rate was implied.