To evaluate several alternative measures of forced expiratory volume in 1 second percent predicted (FEV₁ %pred) variability as potential predictors of future FEV₁ %pred decline in patients with cystic fibrosis.

We included 13 827 patients age ≥6 years from the Epidemiologic Study of Cystic Fibrosis 1994-2002 with ≥4 FEV₁ %pred measurements spanning ≥366 days in both a 2-year baseline period and a 2-year follow-up period. We predicted change from best baseline FEV₁ %pred to best follow-up FEV₁ %pred and change from baseline to best in the second follow-up year by using multivariable regression stratified by 4 lung-disease stages. We assessed 5 measures of variability (some as deviations from the best and some as deviations from the trend line) both alone and after controlling for demographic and clinical factors and for the slope and level of FEV₁ %pred.

All 5 measures of FEV₁ %pred variability were predictive, but the strongest predictor was median deviation from the best FEV₁ %pred in the baseline period. The contribution to explanatory power (R²) was substantial and exceeded the total contribution of all other factors excluding the FEV₁ %pred rate of decline. Adding the other variability measures provided minimal additional value.

Median deviation from the best FEV₁ %pred is a simple metric that markedly improves prediction of FEV₁ %pred decline even after the inclusion of demographic and clinical characteristics and the FEV₁ %pred rate of decline. The routine calculation of this variability measure could allow clinicians to better identify patients at risk and therefore in need of increased intervention.