The presence of any of the clinical symptoms increases the likelihood of a genetic basis. Clinical symptoms are graded as never, sometimes, often, always. A score of 1.32 or higher in this system determined a child with PFAPA-like febrile attacks to be at high risk of having one of the three hereditary forms of non-CAPS periodic fever syndromes and therefore appropriate for genetic testing. This gave a very high sensitivity (>90%) meaning very few children with a proven gene mutation would be missed. The sensitivity (59-82%) was lower so a significant number of children would be tested and no mutation found.
Of those PFAPA-like children deemed to be at high risk of having an inherited periodic fever, the clinical features that helped direct the order of gene testing were:
Mean duration of fever
Enlarged spleen (splenomegaly)
The duration of fever discriminated between TRAPS and FMF, but not between FMF and HIDS, which were then separated by vomiting and splenomegaly which were more likely to be present in HIDS than FMF. Regression tree analysis using these features accurately predicted the correct mutated gene in 77%.