The content of this website is intended for healthcare professionals only

Previous Posts

1 2 

How easy is it to diagnose sub-types of uveitis?

Evidence-based medicine

Gerry Morrow

Monday, 15 October 2018

uveitis_AdobeStock_34671485.jpgUveitis is defined as inflammation of the uveal tract, which comprises the iris, ciliary body, and choroid. Inflammation may be more generalized to include the retina, the optic nerve, and the vitreous humour, known as panuveitis. ​

The inflammation may be acute, recurrent or chronic. The most commonly affected age group is between 20-60, but it can also affect children. There is a multiplicity of causes of uveitis, including autoimmune disorders, cytomegalovirus and other infections.

Symptoms and signs can be insidious and subtle and include redness of the eye, pain, photophobia, a watering eye and sometimes (but not always) reduced vision associated with sluggish pupillary reactions. The diagnostic problem for clinicians is three-fold:

  1. There is a long list of differential diagnoses, including relatively innocent causes of red eye,  including viral conjunctivitis.
  2. A slit lamp examination is required (and the associated expertise) to make a substantive diagnosis.
  3. This is a relatively rare condition; a GP may see fewer than one person a year with a new diagnosis of uveitis.

An early diagnosis is crucial to prevent the possible complication of visual loss due to secondary macular oedema, cataract and glaucoma.

Given all of these problems how easy is it to make a definitive subtype diagnosis of this important and problematic condition?

According to a paper in the American Journal of Ophthalmology the answer is ‘not very easy'. Authors set out to measure how likely uveitis experts were to agree on a diagnosis of sub-types of uveitis in 5,766 patients. Seventy-six investigators and nine committees reviewed case notes and images for all these patients and reported on how much agreement on a specific diagnosis there was between reviewers.

Their analysis relied on the concept of measuring interrater reliability, which is a correlation statistic called kappa (k). If k= 1 this means that both clinicians agree exactly on a diagnosis. If k= minus 1 then there is complete disagreement. Most research is based on figures of k= 0.81 to 1 as providing ‘almost perfect’ agreement.

In the uveitis interobserver agreement paper it was found that even amongst uveitis experts variability was only ‘moderate at best’ as the whole series of 5,766 patients k= 0.39, with a range of 0.23-0.79.  It was only when teleconferences were organised to create a consensus between clinicians that the variability correlation was moved closer to complete agreement. 

Crucially this agreement has fed into a final dataset for machine learning in order to assist the development of clearer diagnostic criteria for each diagnostic class of uveitis. It is hoped that this work will help GPs in their diagnostic challenge in managing patients with visual symptoms in order to exclude or refer people with possible uveitis.

Related OnMedica content:

  • The red eye - part 1: This course covers the presentation and management of this common ophthalmological condition and includes some case histories.
  • The red eye - part 2: This course elaborates on certain additional causes of red eyes that may present in a primary care setting.

Author's Image

Gerry Morrow

With a keen interest in evidence-based medicine and patient involvement, Gerry has over twenty years’ experience working as a GP, based initially in Worcester and then in the rural practice of Allendale. Now Medical Director of Clarity Informatics, a leading IT healthcare solutions provider, Gerry is leading a globally recognised team of clinicians and researchers, and is also directly responsible for the production and delivering of Clarity’s clinical guidance which forms the clinical content of NICE’s Clinical Knowledge Summaries (CKS) service. Gerry is also a clinical non-executive director of the North East Ambulance Service, responsible for operating patient transport and ambulance response services across a region covering 3,200 square miles and a population of 2.7 million people.
Registered in England and Wales. Reg No. 2530185. c/o Wilmington plc, 5th Floor, 10 Whitechapel High Street, London E1 8QS. Reg No. 30158470