How many diabetes-related deaths are avoidable?
6th Nov 2012
"Up to 24,000 patients with diabetes are dying needlessly each year"
Daily Mail, 6 November 2012
"Every year, 24,000 people with diabetes die simply because their disease has not been effectively managed"
Daily Star, 6 November 2012
Margaret Hodge MP, Chair of the Public Accounts Committee (PAC), was on truculent form recently during an oral evidence hearing with top NHS officials on the subject of diabetes services for adults.
During a lively cross examination she suggested that in his interview with the committee Sir David Nicholson, Chief Executive of the NHS in England, made it sound as if he "didn't give a toss that 24,000 people die unnecessarily" of the disease each year. The same figure, as it happened, made it into several headlines this week.
Sir David strongly rejected Mrs Hodge's accusation, arguing that it was "unclear" how many of the 24,000 deaths could be described as "avoidable" and that he regarded the PAC report into diabetes treatment as "unbalanced".
It's clear that there's some disagreement about how many deaths caused by diabetes are in fact avoidable and to what extent the NHS is failing these patients. So are the headlines misleading?
The origin of the '24,000 deaths' figure
The source for this statistic, which the PAC understood to be broadly accepted, is the National Diabetes Audit (NDA).
In 2008 the NDA produced a report on mortality rates among diabetes sufferers. This research group, operating out of the Department of Health, tracked a sample of diabetes patients - some 1.4 million people registered with GP practices in England.
It calculated how many of them had died over the course of a year and the likelihood of their death being a result of diabetes. (A death might be "attributable" to diabetes even if it's not recorded as the primary cause of death.) The NDA then compared the mortality rates for diabetes sufferers with the standard estimate for the general population.
It established that there were 16,000 "extra" deaths among diabetes sufferers. But the NDA's survey only collected data on 68% of diabetes patients in England. When it extrapolated its findings, it estimated that there were some 24,000 excess deaths each year for people diagnosed with diabetes. This means that an extra 24,000 people (with diabetes) die each year compared to an equivalent population who don't suffer from the disease.
The NDA is not the only organisation to quote a figure in the region of 24,000. The World Health Organisation (WHO) has developed a formula to calculate how many deaths are likely to be the result of diabetes in a particular age cohort, where there's a certain prevalence of the disease. For each age group and gender, it's possible to calculate a "hazard ratio" that quantifies the extra risk of death if you happen to suffer from diabetes.
The Yorkshire and Humber Public Health Observatory, which uses the WHO metholodology, calculated that in 2005 there were up to 26,300 "excess" deaths among diabetes sufferers aged 20-79. This serves as a reasonable check on the NDA's estimate of 24,000 deaths.
What do we mean by "avoidable" deaths?
If we return to the PAC report, Sir Bruce Keogh, Medical Director of the NHS, addresses this question directly:
"There are 24,000 excess deaths. That means that there are 24,000 people more with diabetes who die than an equivalent population who do not have diabetes. It does not mean that they are avoidable, and we need to be very, very clear about that."
However, the charity Diabetes UK clearly refers to these 24,000 deaths as just that - "avoidable" - both in the evidence it submitted to the PAC and on its website. So who's correct?
In his evidence to the PAC, Sir Bruce implied - crucially - that not all of these early deaths are necessarily the result of inadequate treatment and that, by the very fact of being diseased, some of these patients are more likely to die early.
Diabetes is a disease that can be managed effectively, but at the moment a person with diabetes is still looking at a lower life expectancy than somebody without the disease - a member of the "general population".
In describing these deaths as "avoidable", Diabetes UK suggest that people with diabetes could be treated more effectively. Indeed, the PAC report includes various recommendations as to how patients' life expectancy might be extended. However, it's not altogether surprising that diabetes sufferers fare worse than the general population overall.
What we appear to be dealing with here is ambiguity over terminology. Many people might assume that an "excess" death is one that's avoidable, but for statisticians this is a specific term with a precise application. Diabetes UK, a charity which campaigns for better treatment of the disease, might be inclined to argue that nobody ought to be dying from diabetes and that any death from the disease is avoidable.
Neither Diabetes UK nor Sir Bruce Keogh is wrong, but they disagree over the meaning of the word "avoidable" in this instance. Nevertheless, there's broad agreement that if the NHS helps patients to manage their condition more effectively, fewer lives will be cut short.