Development and validation of a risk score to identify children at risk of life-threatening asthma

Abstract

Objective

To develop and validate a prediction risk score for identification of children at risk of developing life-threatening asthma (LTA).

Methods

Our study utilized existing medical records and retrospective analysis to develop and validate a risk score. The study population included children aged 2-17 years, admitted with a primary diagnosis of asthma, to Sydney Children’s Hospital between 2011-2016. Children admitted in the intensive care unit with asthma at risk of LTA (cases) and those admitted into general ward (comparison group), were randomly divided into a derivation and a validation cohort. Candidate predictors from derivation cohort were selected through multivariable regression, which were used to estimate each child’s risk of developing LTA in the validation cohort. Predictive performance of the risk score was evaluated by the area under the receiver operating characteristic curve (AUROC) and Hosmer-Lemeshow goodness-of-fit test.

Results

The study population comprised of 1171 children; 586 in the derivation and 585 in the validation cohort. Four independent candidate variables from derivation cohort (age at admission, socioeconomic status, a family history of asthma/atopy and previous asthma hospitalizations) were retained in the predictive model (AUROC 0.759; 95% CI, 0.694–0.823), with a sensitivity of 78.5% and specificity of 46.6%.

Conclusions

Our risk algorithm based on routinely collected clinical data may be used to develop a user-friendly risk score for early identification and monitoring of children at risk of developing LTA.

KEYWORDS:

• Asthma
• paediatrics
• fatal outcome
• risk factors
• risk management

Acknowledgements

Thank you to the Respiratory Department at the Sydney Children’s Hospital for their timely and constructive feedback and the health data unit at the Sydney Children’s Hospital for access to the medical records. I would also like to thank the Sydney Children’s Hospital Foundation for their support in our research endeavour and the Rotary Club of Sydney Cove for their funding.

Declaration of interests

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of this article.

Additional information

Funding

This work was funded by Rotary Club of Sydney Cove.