Development and Validation of a Clinical Diagnostic Scoring System for the Diagnosis of IPF.

Rationale Interpreting the radiological data in conjunction with an objective clinical score could help to harmonize IPF diagnostic and improve accuracy. Objectives We sought to establish and validate a multivariable objective scoring model based on clinical parameters, by stratifying the risk for patients to be diagnosed as IPF versus other forms of interstitial lung disease (ILD). Methods A clinical score was derived from review of patients evaluated at the Inova Fairfax ILD program and validated in three distinct cohorts. Based on known IPF clinical characteristics, a multivariable model was created and assessed by receiver operator curve (ROC) characteristics. Results There were 844 ILD patients diagnosed as either IPF (n=347, 41%) or non-IPF ILD (n=497, 59%). Based on calculated odds ratios, a score was assigned to each of the following clinical parameters: age, sex, smoking history, ethnicity, ILD family history, exposures, presence of CTD signs or symptoms, and Velcro crackles. The final Fairfax IPF clinical score (FICS) ranged from 1 to 25. The clinical diagnostic score system was accurate in predicting IPF, as measured by the area under the curve (AUC 0.88) in the derivation cohort with similar AUCs of 0.91, 0.81 and 0.71 in the respective validation cohorts. Conclusion The FICS appears to be an accurate tool for estimating the pretest probability of IPF in patients with ILD. How the FICS performs in conjunction with the various HRCT patterns remains to be determined. This model could ultimately be useful for increasing the degree of confidence in the final diagnosis and help to obviate the need for lung biopsy in cases of non-UIP patterns on HRCT.