Predicting Asthma Development in Children Using a New Personalized Asthma Risk Score

RATIONALE: The ability to predict children at risk of asthma has remained difficult. This has hindered advancements of preventative interventions. We sought to develop an algorithm to predict asthma risk in young children using information easily collected and without requiring laboratory tests.

METHODS: The Personalized Asthma Risk Score (PARS) was constructed by integrating demographic and clinical data from the CCAAPS (Cincinnati Childhood Allergy and Air Pollution Study, n=762) birth-cohort. Asthma status was determined at age 7yrs. Using logistic regression, backward selection was used to develop the PARS (p<0.05). Informative variables were converted into a score by converting odds ratio (OR) of each and summing to calculate a continuous risk score. The sensitivity, specificity, negative (NPV) and positive predictive value (PPV) of PARS was compared to the Asthma Predictive Index (API).

RESULTS: PARS was derived from asthma associations with parental asthma (OR=1.92; 95% CI [1.17 – 3.17]), eczema (OR=1.84 [1.09 – 3.06]), early wheezing (OR=2.87 [1.52 – 5.37]), wheezing apart from colds (OR=2.66 [1.39 – 5.17]), African-American race (OR=2.02 [(1.18 – 3.43]), and ≥2 positive skin prick tests (OR=2.7 [1.54 – 4.87]). PARS reliably predicted asthma with an improved sensitivity (0.64) and PPV (0.38) with similar specificity (0.8) and NPV (0.92) when compared to the API. Notably, given its continuous nature, PARS had an improved ability to predict asthma in children with intermediate risk.

CONCLUSIONS: PARS was able to predict asthma with improved sensitivity, similar specificity to the API without requiring laboratory tests, thus it may be more useful as a research and clinical tool.