A respiratory alert model for the Shenandoah Valley, Virginia, USA

David M. Hondula, Robert E. Davis, David B. Knight, Luke J. Sitka, Kyle Enfield, Stephen B. Gawtry, Phillip J. Stenger, Michael L. Deaton, Caroline P. Normile, Temple R. Lee

Research output: Contribution to journalArticlepeer-review

19 Scopus citations


Respiratory morbidity (particularly COPD and asthma) can be influenced by short-term weather fluctuations that affect air quality and lung function. We developed a model to evaluate meteorological conditions associated with respiratory hospital admissions in the Shenandoah Valley of Virginia, USA. We generated ensembles of classification trees based on six years of respiratory-related hospital admissions (64,620 cases) and a suite of 83 potential environmental predictor variables. As our goal was to identify short-term weather linkages to high admission periods, the dependent variable was formulated as a binary classification of five-day moving average respiratory admission departures from the seasonal mean value. Accounting for seasonality removed the long-term apparent inverse relationship between temperature and admissions. We generated eight total models specific to the northern and southern portions of the valley for each season. All eight models demonstrate predictive skill (mean odds ratio = 3. 635) when evaluated using a randomization procedure. The predictor variables selected by the ensembling algorithm vary across models, and both meteorological and air quality variables are included. In general, the models indicate complex linkages between respiratory health and environmental conditions that may be difficult to identify using more traditional approaches.

Original languageEnglish (US)
Pages (from-to)91-105
Number of pages15
JournalInternational journal of biometeorology
Issue number1
StatePublished - Jan 2013
Externally publishedYes


  • Air quality
  • COPD
  • Classification tree
  • Respiratory health
  • Virginia
  • Weather

ASJC Scopus subject areas

  • Ecology
  • Atmospheric Science
  • Health, Toxicology and Mutagenesis


Dive into the research topics of 'A respiratory alert model for the Shenandoah Valley, Virginia, USA'. Together they form a unique fingerprint.

Cite this