Author Affiliation: Department of Psychiatry and Behavioral Sciences, University of Washington, Seattle.
We welcome the review of our article, “Prevalence of Alcohol Use Disorders Among American Surgeons,” by Crandall and Tuttle-Newhall. However, we do not agree that “the methodological flaws of their study limit the ability to draw definitive conclusions from the data.” Although it is certainly true that our survey response rate of 28.7% was low, it did not, per se, demonstrate bias in our study for the following reason: our sample size provides abuse/dependence estimates within 1.2% of the population rate with 95% confidence. If we had obtained a response from most of the surgeons, the accuracy would only have been within 0.63%. Although this error rate is based on the assumption that data are missing at random, even if a systematic bias existed such that those who did not respond had an 8% rate of abuse/dependence (eg, approximately half the rate of respondents), it would decrease our abuse/dependence estimate for the population by only 0.33%. Given our large sample size, our estimate of the abuse/dependence rate is therefore likely to have relatively small bias, even under a scenario of major nonresponse bias. In addition, 2 studies1,2 failed to identify significant differences between responding and nonresponding physicians in cross-sectional surveys.
Oreskovich MR. Alcohol Use Does Not Equal Abuse—Reply. Arch Surg. 2012;147(8):786. doi:10.1001/archsurg.2012.1275