Development of a diagnostic support tool for predicting cervical arterial dissection in primary care.

Objectives: Cervical arterial dissection (CAD) is an important cause of stroke in young people which may be missed because early features may mimic migraine or a musculoskeletal presentation. The study aimed to develop a diagnostic support tool for early identification of CAD.

Design: Retrospective observational study.

Setting: Tertiary hospital.

Participants: Radiologically confirmed CAD cases ( = 37), non-CAD stroke cases ( = 20), and healthy controls ( = 100).

Main Outcome Measures: The presence of CAD is confirmed with imaging. Predictive variables included risk factors and clinical characteristics of CAD. Variables with a p-value <0.2 included in a multivariable model. Predictive utility of the model is assessed by calculating area underthe ROC curve (AUC).

Results: The model including four variables: age 40-55 years (vs < 40), trauma, recent onset headache, and > 2 neurological features, demonstrated excellent discrimination: AUC of 0.953 (95% CI: 0.916, 0.987). A predictive scoring system (total score/7) identified an optimal threshold of ≥ 3 points, with a sensitivity of 87% and specificity of 79%.

Conclusions: The study identified a diagnostic support tool with four variables to predict increased risk of CAD. Validation in a clinical sample is needed to confirm variables and refine descriptors to enable clinicians to efficiently apply the tool.Optimum cutoff scores of ≥ 3/7 points will help identify those in whom CAD should be considered and further investigation instigated. The potential impact of the tool is to improve early recognition of CAD in those with acute headache or neck pain, thereby facilitating more timely medical intervention, preventing inappropriate treatment, and improving patient outcomes.Wordcount: 3195.