External Validation of the Predicting Asthma Risk in Children (PARC) tool in a clinical cohort
Authors:
- Daria Olena BergerDBDaria Olena Berger
- Eva S L PedersenEPEva S L Pedersen
- Maria Christina MalletMMMaria Christina Mallet
- Claudia E KuehniCKClaudia E Kuehni
- the SPAC Study TeamTTthe SPAC Study Team
Preprint
Abstract
Rationale: The Predicting Asthma Risk in Children (PARC) tool uses questionnaire-based respiratory symptoms collected from preschool children to predict their risk of asthma 5 years later. The tool was originally developed and externally validated in population-based settings and has not yet been validated in a clinical setting. Objective: To externally validate the PARC tool in children seen in paediatric pulmonology clinics. Methods: The Swiss Paediatric Airway Cohort (SPAC) is a prospective study of children seen in respiratory outpatient clinics across Switzerland. This analysis included children seen at ages 1-6 years for cough or wheeze at baseline and who completed the follow-up questionnaire 2 years later. The outcome was defined as current wheeze plus use of asthma medication. In sensitivity analyses, we explored effects of varied inclusion criteria and outcomes. We assessed performance by describing sensitivity, specificity, negative and positive predictive value (NPV, PPV), area under the curve (AUC), scaled Brier’s score and Nagelkerke’s R 2 scores and compared performance in SPAC to that achieved in the original population, the Leicester Respiratory Cohort (LRC). Results: Among the 346 children included, 125 (36%) reported the outcome after 2 years. At a PARC score cut-off of 4, sensitivity was higher (95% vs 79%) but specificity lower (14% vs 57%) in SPAC compared to LRC. NPV was comparable (0.84 vs. 0.87) as was PPV (0.37 vs.0.42). Discrimination was lower in SPAC (AUC of 0.71 vs 0.78), as were Nagelkerke’s R 2 (0.18 vs 0.28) and scaled Brier’s scores (0.13 vs 0.22). When the outcome was changed to moderately severe asthma (>4 attacks plus use of asthma medication), there were improvements in AUC (0.74), sensitivity (0.97), specificity (0.22) and NPV (0.99), but some deterioration in PPV (0.13), R 2 (0.15) and scaled Brier score (0.09). Conclusion: While the PARC tool performs well in a population-based setting and has some clinical utility, in particular for ruling out the development of asthma, this study highlights the need for new prognostic prediction tools to be developed specifically for the clinical setting. Funding: SNSF:320030_182628, SLA2019-03_641670Keywords
All Related Versions
- Published version: Pediatric Pulmonology, 57 (11), 2715.
This publication has 26 references indexed in Scilit:
- The Novel 10-Item Asthma Prediction Tool: External Validation in the German MAS Birth CohortPLOS ONE, 2014
- A simple asthma prediction tool for preschool children with wheeze or coughJournal of Allergy and Clinical Immunology, 2014
- Exhaled nitric oxide in symptomatic children at preschool age predicts later asthmaAllergy, 2013
- Predicting asthma in preschool children with asthma symptoms: study rationale and designBMC Pulmonary Medicine, 2012
- A Simple Tool to Identify Infants at High Risk ofMildtoSevereChildhood Asthma: ThePersistentAsthma Predictive ScoreJournal of Asthma, 2011
- Discriminative properties of two predictive indices for asthma diagnosis in a sample of preschoolers with recurrent wheezingPediatric Pulmonology, 2011
- Predicting the long-term prognosis of children with symptoms suggestive of asthma at preschool ageJournal of Allergy and Clinical Immunology, 2009
- Cohort Profile: The Leicester Respiratory CohortsInternational Journal of Epidemiology, 2007
- Substantial effective sample sizes were required for external validation studies of predictive logistic regression modelsJournal of Clinical Epidemiology, 2005
- A Clinical Index to Define Risk of Asthma in Young Children with Recurrent WheezingAmerican Journal of Respiratory and Critical Care Medicine, 2000