Assumptions used in estimating a population benchmark
The assumptions used in estimating a population benchmark of 0.75% endoscopies per annum are based on the following sources of information:
- epidemiological data estimating the prevalence of symptoms, correlated against those needing endoscopy as described in NICE guidance
- activity data showing the numbers of endoscopies currently being performed
- data from examples of current practice where referral guidelines are being followed.
It is also assumed that H.pylori testing will be carried out to reduce referral rates.
Epidemiology - or prevalence of symptoms in the population - provides an estimate of the numbers of people who are likely to meet the referral criteria.
Using the British Society of Gastroenterology definition of dyspepsia, which is particularly relevant to primary care, it is estimated that annually:
40% of the adult population may suffer from dyspepsia, of whom:
- 5% consult their GP
- 0.5% have alarm symptoms
Figure 1: Dyspepsia and alarm signs: prevalence
There is an age-related rise in the consultation rate for dyspepsia. In a population with an average age composition, a GP with a list of 2000 patients can expect approximately 100 to consult with dyspepsia related illness (that is, 5%).
The minimum rate for upper GI endoscopies will therefore be 0.5%, representing the core group of patients with alarm symptoms.
Activity data gives an indication of the current rates at which endoscopy is being performed. Average rates do not necessarily correlate with the rate that would be expected if NICE clinical guidelines referral criteria were uniformly followed.
Analysis of data on referral for endoscopy, which was collected by the Health and Social Care Information Centre as part of the hospital episode statistics returns, shows that the mean directly standardised rate for all English PCTs for the 3 years 2004/5 to 2006/7 is 950 per 100,000.
The average rate hides large variations. Referral rates vary among PCTs from fewer than 200 per 100,000 to nearly 2000 per 100,000 population. Thus there is a 10-fold variation in rates across England.
Some of the variation is likely to be due to disease or symptom prevalence and other clinical reasons. However, this alone is unlikely to account for all of the variation, and there may be other factors that influence it, such as variations in service capacity and differences in referral criteria. Those with very high referral rates are likely not to be operating within expected referral criteria, and will have artificially inflated the overall mean figure.
Examples of current referral practice
The most accurate benchmark estimates would be expected from practices known to be following appropriate referral criteria.
One study has been identified that measured rates of referral after introducing rigorously applied referral criteria. In this study, researchers from the Aintree Centre for Gastroenterology and Liver Disease established a rapid access upper GI cancer service. As a consequence of setting up this service, the referral rate for those with alarm symptoms was 0.54%. If patients referred by an open access service (also following standard referral criteria) are also included, the referral rate for endoscopy increases to 0.75% per year.
H. pylori testing
A study by the Bristol Helicobacter Project found a 15.5% prevalence of H. pylori infection among asymptomatic people and a 90.7% eradication rate after standard treatment. This was a population-based cohort study of people aged 20-59 in Bristol. The study indicates that it may be possible to reduce endoscopy referrals among this group by investigating and treating for H. pylori before referral, and not performing endoscopy in those who respond to treatment.
A recent study by the Health Protection Agency on the variation in the use of H. pylori tests in UK general practices found that serology submission rates varied 600-fold (between 0.1 and 59 per 1000 population per year). In this study, the proportion of people with dyspepsia who tested positive for H. pylori was between 32% and 53%, varying with the ethnic composition and deprivation level of the population. It was also noted that low-testing practices were less aware of the benefits of H. pylori testing and had shorter endoscopy waiting times. It is possible that shorter waiting times act as a disincentive to prioritise endoscopy testing, thus causing some of the variation in referral patterns.
A randomised controlled trial involving 47 general practices demonstrated a statistically significant 18.8% reduction in referrals for endoscopy from GP practices that were provided with information and access to H. pylori testing. A systematic review and meta-analysis of data from five other trials has also indicated the usefulness of the 'test and treat' strategy in the management of patients with dyspepsia.
Serology testing has lower specificity and lower positive predictive value than a stool antigen test or urea breath test. See the NICE clinical guideline CG17 on dyspepsia (page 149). Hence the use of the serology test may result in large numbers of false positive results, especially in areas with low prevalence of H. pylori infection. One study has indicated that practices may prefer stool tests to breath tests, because they impact less on practice budget and time. The wider use of H. pylori testing may lead to a reduction in the variability of testing pathway and protocols.
Primary care trusts should develop systems to provide reliable tests for H. pylori (either breath test or stool test) for patients with dyspepsia in primary care. A detailed primary care guide for H. pylori testing, and cost comparison of the various tests available, is available from the Health Protection Agency.
From the information above, we can conclude that:
- if following the criteria above, the referral rate based on the prevalence of alarm symptoms is likely to be 0.50-0.54% per year
- the current average rate is 0.95%, although there is wide variation and no indication of whether referral criteria are being followed.
Because the majority of referrals for upper GI endoscopy are likely to result from the presence of alarm symptoms, a benchmark estimate of 0.75% has been selected. This is equivalent to the rate identified in the Aintree Centre study, in which an open access service was available that followed expected criteria for referral.
This page was last updated: 02 March 2012