Using electronic health records to help support the diagnosis of hard-to-suspect conditions in primary care associated with diagnostic delay

My project will support efforts to improve the quality and safety of medical diagnosis.

I will focus on exemplar conditions known for their diagnostic difficulty - resulting in long periods between patients first complaining to their GP about symptoms relating to their condition and them receiving the correct diagnosis.

It may be possible that some pathways to diagnosis can be improved by better decision-making by doctors, empowered by new knowledge that will arise from my research.

I plan to focus on certain exemplar diseases (e.g., endometriosis, ankylosing spondylitis and chronic obstructive pulmonary disorder), known for their long diagnostic intervals.

A patient’s electronic health record (EHR) contains lots of information about their medical history – for example, when, and how often, they visited the doctor and what symptoms they had.

Using large collections of anonymous EHRs, I will identify different presenting features that patients shared before they were diagnosed and related healthcare utilisation events – so called ‘pathways’ to diagnosis.

Having done that, I can examine patients who take similar pathways, and analyse whether different pathways have different diagnostic intervals.

Associations between different presenting features and risk of as-yet-undiagnosed illness will also be examined in the latter part of the project.