CLUSTER-ANALYSIS APPLIED TO BUILDING-RELATED ILLNESS

Identifying remediable causes of occupant symptoms in building-related illness is frequently difficult, This is particularly true when the b uilding-wide prevalence of symptoms is comparable to that reported in non-problem buildings. This analysis applied an epidemiological approa ch to an assessment of a problem building, allowing investigators to v isually identify an area of apparent increased symptom density. A clus ter analysis approach permitted biostatistical confirmation of the vis ual cluster. Building-related symptom reporting was statistically sign ificantly associated with a prior physician diagnosis of dust and/or m old allergy. The likely etiology of building occupant symptoms was ide ntified within the region implicated by the cluster analysis. This app roach may be useful to focus building evaluations on both the likely p hysical source and general characteristics of suspect etiologic agents .