Cluster analysis applied to building-related illness

Abstract

Identifying remediable causes of occupant symptoms in building-related illness is frequently difficult. This is particularly true when the building-wide prevalence of symptoms is comparable to that reported in non-problem buildings. This analysis applied an epidemiological approach to an assessment of a problem building, allowing investigators to visually identify an area of apparent increased symptom density. A cluster analysis approach permitted biostatistical confirmation of the visual cluster. Building-related symptom reporting was statistically significantly associated with a prior physician diagnosis of dust and/or mold allergy. The likely etiology of building occupant symptoms was identified within the region implicated by the cluster analysis. This approach may be useful to focus building evaluations on both the likely physical source and general characteristics of suspect etiologic agents.