Validity of Self-Reported Drinking Before Injury Compared with a Physiological Measure: Cross-National Analysis of Emergency-Department Data from 16 Countries * (Report‪)‬

ALTHOUGH MANY STUDIES CARRIED OUT in the emergency department (ED) have relied on patient self-reports of alcohol consumption before the injury event to identify those injuries that may be considered to be alcohol related, to identify those patients who may benefit from a brief intervention for problem drinking, or both, relatively few studies have examined the validity of self-report data compared with objective measures in this setting. Although verbal self-reports have generally been found to have reasonable validity within the context of a structured interview (Babor et al., 1990; Del Boca and Noll, 2000; Midanik, 1982, 1988), validity has been found to vary considerably according to a number of variables, including the population studied, the criterion used, and the quantity and frequency of usual drinking (Babor et al., 1987; Dawson, 2003; Embree and Whitehead, 1993; O'Farrell and Maisto, 1987). As quantity and frequency of drinking increase, the variance in quantity and frequency also increases and could be a source of systematic bias related to accurate recall (Gruenewald, 1996). Analysis of national alcohol survey data found less consistency in self-reported heavy drinking among those who reported more frequent heavy drinking (Greenfield et al., 2006). Studies comparing self-reports with estimates of blood alcohol concentration (BAC) have not found good validity among those arrested for drunken driving (Fine et al., 1978; Jalazo et al., 1978), although BAC estimates may not correlate well with self-reports, regardless of the population studied, because of the amount of alcohol consumed, the period during which consumption occurred, and the time lapsed following consumption (O'Farrell and Maisto, 1987).