National swine flu adult assessment guidelines: Retrospective validation of objective criteria in three proxy datasets

Objectives: To validate the objective criteria in the Department of Health Adult Swine Flu Assessment Tool against proxy datasets for pandemic influenza. Design: Comparative validation study with 3 datasets. Setting: Urban Emergency Department (group 1) and prehospital care (groups 2 and 3). Participants: Adults with community-acquired pneumonia (group 1, n=281), shortness of breath (group 2, n=211) or any respiratory diagnosis (group 3, n=300). Outcome measures: Hospital admission (group 1), hospital admission or intravenous therapy (group 2) and transfer to emergency department (group 3). Results: Sensitivity and specificity of the tool were 0.73 (95% CI 0.67 to 0.8) and 0.83 (0.72 to 0.9) in group 1, 0.64 (0.55 to 0.71) and 0.63 (0.52 to 0.73) in group 2 and 0.84 (0.75 to 0.9) and 0.55 (0.48 to 0.62) in group 3. Analysis of individual components of the tool and a summative score is presented. Conclusions: The objective criteria of the proposed DH assessment tool do not perform particularly well in predicting relevant clinical outcomes in feasible proxy conditions for pandemic influenza.