, 2009). This study has limitations, such as those related to the measurement of tobacco Tofacitinib FDA use. Although the ASSIST is a valid and reliable screening instrument, it is not a routinely used tobacco use measure, and therefore, comparisons with other studies are difficult. For example, the ASSIST assesses general tobacco use (cigarettes, pipes, chewing tobacco, and cigars combined), while the CHIS statewide survey assesses cigarette smoking specifically. Therefore, the differences in use between patients and the general population described in this study may be overestimated. On the other hand, the use of tobacco products other than cigarettes is relatively low in California (Al-Delaimy et al., 2008); therefore, we think that the differences we report are useful.

In addition, differences in the wording of the ED interview items and those on the statewide survey required us to make some assumptions of equivalence. For example, the lifetime use item in the ED interview was, ��In your lifetime, have you ever used tobacco products.�� In the CHIS sample, the item was, ��Altogether, have you smoked at least 100 or more cigarettes in your entire lifetime?�� However, although we believe that the lack of comparability between the ED measures and the CHIS measures is a concern, the very large differences seen give us confidence that the finding of ED patients having considerably higher tobacco use is real and of public health significance for targeted interventions.

Another measurement shortcoming was the lack of widely-used measures of cigarette use and addiction��number of cigarettes smoked, the latency to first cigarette of the day, the smoking of mentholated cigarettes, and more detailed cessation history items��all measures that have shown Black�CWhite differences in other studies (Muscat et al., 2002; Okuyemi et al., 2004). Another important covariate, socioeconomic status, was not reliable and therefore not included in our models. A sensitive reliable socioeconomic measure would be important to include because of its likely association with race/ethnicity and health behaviors. The present ED data were not collected specifically for tobacco research but rather came from opportunistic screenings primarily for alcohol and drug misuse. Although the number of involved EDs reflected an impressive representation of all County EDs and the sample size of patients was large, the present study used a convenience sample rather than a probability sample.

Health interviewers attempted universal screening of all capable patients; however, to the degree that the sample does not represent all patients in the area, we cannot assume that the findings are generalizable to all ED patients. Finally, for the purpose of confidentiality and rapport building, Dacomitinib HEs avoided asking patients about their immigration status and their insurance coverage, thereby limiting our ability to describe and analyze data by these characteristics.