A new way of coding health statistics will reveal a change in mortality rates in England. Alison Moore reports

This year you will be less likely to die from pneumonia than in 2000, but more prone to succumb to long-term chronic conditions and strokes.

Overall, mortality statistics will show a fundamental shift, with deaths from pneumonia expected to drop by between 20 per cent and 67 per cent.

A new wonder drug? Improved treatment? The welcome side-effects of the extra£50 heating allowance for pensioners? Alas, no - the number of deaths will remain the same, but the way in which they are coded and reflected in national statistics will undergo its most radical change for 50 years.

From 1 January, England and Wales has been using a new version of the World Health Organisation-approved international classification of diseases.

ICD-10 has already been in use in Scotland and is used for cancer registries and hospital episode statistics in England and Wales, replacing the 21-year-old ICD-9.

ICD-9 needed updating, as it ignores some diseases and the coding allocated to others does not reflect modern understanding of the diseases and their processes.

In 1979, when ICD-9 was adopted, HIV/AIDS was unknown, for example: now it and its manifestations can be more accurately represented through a revised coding system.

But by themselves these changes would be relatively minor. What is going to alter the face of mortality statistics is a change in the rules governing how the underlying cause of death is selected from the information a doctor provides for a death certificate.

This should switch the emphasis in mortality statistics from the immediate mode of death - for example, pneumonia or acute bronchopneumonia - to the underlying condition, such as Parkinson's disease or chronic bronchitis.

Similar rule changes in the past have led to significant shifts in the apparent cause of death - one was in force in England and Wales between 1984-92, for example. 'One of the most striking things when a similar change was done was a more than doubling of the deaths due to senile dementia, ' says Dr Cleo Rooney, a medical epidemiologist with the Office For National Statistics.

She expects the number of deaths attributed to chronic underlying conditions to increase.

The death certificate will not change, nor will the information required of the GP or hospital doctor who provides the initial certificate of the causes of death - the coding is done by the ONS, not the doctor.

The doctors most affected are likely to be directors of public health and their staff, who will soon start to receive information using the new coding.

The first documents they will receive using the new data will be the public health mortality files and the NHS rapid estimates of deaths.

But will such changes skew the public perception of causes of death, or even affect funding? Dr Rooney thinks not - she does not expect significant changes in the numbers of deaths attributed to key areas, such as cancer or accidents, which are national priorities.

And future press releases from the ONS may come with a warning that comparisons with earlier years are difficult because of the changes. This is likely to be most important in the early days of the new system - for example, if the ONS-estimated numbers of deaths are issued weekly during the winter months.

These may show a dramatic drop in deaths from flu, pneumonia and other respiratory causes if compared with figures from this year or previous years - but this is likely to be illusory.

Dr Rooney says the proportion of 'excess deaths' in winter attributed to pneumonia will drop - studies using ICD-10 in other countries have shown drops of 20 and 67 per cent in pneumonia deaths, although using a slightly different basis.

'But the main thing in the excess winter deaths is the overall total - which will not change, ' she says.

But former Liberal Democrat health spokesman Dr Evan Harris - a public health specialist - says the government misuses health statistics, and results should be viewed with caution.

'It could allow a well-resourced government spin system to seize on changes that may be due to the way the data is collected, ' he says.

By early 2002, when the first yearly routine statistics using ICD-10 are published, a new method of comparing deaths across time should be in use, allowing a comparison between statistics produced under ICD-9 and ICD-10.

But until then politicians trumpeting big drops in deaths from many common diseases should be treated with extreme caution.