This site is intended for health professionals only


Researchers develop computer model to assess prevalence of patients with ‘medically unexplained symptoms’

Computer models can predict the prevalence of patients with medically unexplained symptoms (MUS) at GP practices, but require validation before they can be used by GP commissioners to plan better services, say UK researchers.

Their study found computer programs developed to search GP records for patients with MUS showed ‘adequate goodness of fit' when compared with GP observed levels in general practice.

But the researchers cautioned that more work was needed, as often all symptoms were not coded by GPs, and that severe MUS was not necessarily associated with higher GP consultation rates.

The researchers said this was the first study to look at developing computer models to estimate the prevalence of MUS in general practice. It was welcomed as ‘potentially highly productive' by leading GP commissioners.

It looked at the prevalence of symptoms recorded by 17 GP practices – such as anxiety, stress, psychiatric referrals and asthma – in ‘consecutive consulters' and made an estimate of the prevalence of MUS in the registered population.

The researchers found the models had a c-statistic of 0.70 and 0.76, suggesting both had a ‘reasonable ability to distinguish cases from non-cases'. But they also found a ‘surprising lack of association between MUS cases and the number of somatic symptoms recorded by GPs'.

They concluded: ‘The most likely explanation for the lack of association in this study is that GPs do not record a complete list of all the patient's bodily symptoms in either the system codes or the free text.'

They also found patients with serious MUS may also be unwilling to seek medical help or demonstrate ‘excessive reassurance-seeking'.

The Government announced last year that the Improving Access to Psychological Therapies programme would receive £400m in additional funding up through 2014/15 to extend it to new groups such as those with MUS.

Study leader Professor Richard Morriss, professor of psychiatry at the University of Nottingham, said the study showed models for estimating the prevalence of MUS ‘should have only a very limited use for exploring overall workload and potential resource use in terms of additional psychological treatment'.

But GP commissioners welcomed the work as potentially very useful for those seeking to plan services.

Dr Shane Gordon, a GP in Tiptree, Essex, and clinical commissioning co-lead at the NHS Alliance, said: ‘This is a potentially highly productive area of work. We need not only good models of prevalence, but also good methods to identify specific patients and evidence-based interventions to reduce any inappropriate use of health resources.'