Interrater Reliability of Historical Virtual Audits Using Archived Google Street View Imagery

Abstract

Impaired mobility occurs in up to half of community-dwelling older adults and is associated with poor health outcomes and high health care costs. Although the built environment impacts mobility, most studies of older adults lack information about environmental-level factors. In-person observational audits can be utilized but cannot assess the historical environment. We applied a 78-item checklist to archived Google Street View imagery to assess historical residence access and neighborhood characteristics. Interrater reliability between two raters was tested on 50 addresses using prevalence-adjusted and bias-adjusted kappa (PABAK). The mean PABAK for all items was .75, with 81% of the items having substantial (PABAK ≥ .61) or almost perfect (PABAK ≥ .81) agreement. Environmental assessment using archived virtual imagery has excellent reliability for factors related to residence access and many neighborhood characteristics. Archived imagery can assess past neighborhood characteristics, facilitating the use of historical environment data within existing cohorts.