What is it about?

It can be difficult for healthcare professionals to be sure if someone has asthma (or not). Unlike other conditions, there is no single test to prove, or disprove, a diagnosis of asthma. Instead a diagnosis is made by gathering information about an individual and using that information to weigh up how likely they are to have asthma. Information can be gathered from the symptoms described, past or family history, findings from a physical examination, lung function and other clinical tests.  This project, led by Dr. Luke Daines, wanted to make it easier for doctors or nurses to weigh up the likelihood of asthma by identifying the most important bits of information that can be gathered and use them to make a mathematical ‘model’ to predict asthma. The first step was identifying the features that can be used to predict who has asthma by developing a mathematical model using a research dataset with information from 11,972 children and young people (up to 24 years old).  Then, we tested the model in anonymous data available from routine general practice consultations to check the model worked in other children/young people. From the research dataset, we found 11 features that could be used to work out how likely a child or young person, has asthma. The features were: - Symptoms: wheeze, cough and breathlessness - Other health conditions: hay fever, eczema and allergy to food or drink - Aspects relating to family life: social class and exposure to cigarette smoke - If the child/young person’s mother had asthma. - Treatment with a symptom relieving inhaler (*e.g.* salbutamol) - A past test for lung function. When the prediction model was tested in the research dataset it worked well, and the model was good at identifying the children with asthma from those that didn’t have asthma. However, when the model was tested in a different dataset, the predictions made by the model underestimated the probability of asthma for each individual. One of the reasons that the model did not work as well was that a lot of information had not been coded in a form that could be used in the analysis, so the model had less information to work on. This might not matter in real life, because a health professional could ask the patient or collect any missing information that was needed.

Featured Image

Why is it important?

Using 11 features easily available in general practice, the prediction model provides an evidence-based way of helping doctors and nurses weigh up the likelihood of asthma in children and young people. The prediction model will make it easier for health professionals to be sure that someone has asthma (or not). If used widely in general practice, the prediction model could help to reduce variation in the way that asthma is assessed between different health professionals and improve the experience of people with undiagnosed asthma (and their parents or carers). If children/young people and their parents accessed the prediction model in advance, it could allow them to prepare for a consultation, helping them to know what to expect and improving the chance that they are actively involved.

Perspectives

I personally only made a small contribution to this piece of work, but if you'd like to learn more about the research led by Dr. Daines this is his University of Edinburgh profile: https://www.research.ed.ac.uk/en/persons/luke-daines

Holly Tibble
University of Edinburgh

Read the Original

This page is a summary of: Deriving and validating an asthma diagnosis prediction model for children and young people in primary care, Wellcome Open Research, September 2023, Faculty of 1000, Ltd.,
DOI: 10.12688/wellcomeopenres.19078.2.
You can read the full text:

Read
Open access logo

Contributors

The following have contributed to this page