Has underreporting of cigarette consumption changed over time?

Abstract

According to survey data, the prevalence of Americans' self-reported cigarette smoking is dropping steadily. However, the accuracy of national surveys has been questioned because of declining response rates and the increasing stigmatization of smoking. We used data from 2 repeated, cross-sectional, nationally representative health surveys (National Survey on Drug Use and Health (NSDUH), 1979-2014; and National Health Interview Survey (NHIS), 1965-2015) to determine whether self-reported cigarette consumption has changed over time as a proportion of federally taxed cigarette sales. From each survey, we calculated national equivalents of annual cigarette consumption. From 1979 to 1997, the amount of cigarettes that NSDUH and NHIS respondents reported corresponded to an average of 59.5% (standard deviation (SD), 2.3%) and 65.6% (SD, 3.2%), respectively, of taxed cigarette sales. After 1997, respondents' reported smoking data corresponded to the equivalent of an average of 64.2% (SD, 5.9%) and 63.3%(SD, 2.5%), respectively, of taxed cigarette sales. NHIS figures remained steady throughout the latter period, with a decline during 2013-2015 from 65.9% to 61.1%. NSDUH figures increased steadily, exceeding those of the NHIS after 2002. Given the consistent underreporting of cigarette consumption over time, these surveys are likely not less accurate than they were previously. The recent decrease in NHIS accuracy, however, gives pause about the magnitude of the reported decline in smoking prevalence in 2014 and 2015. Improvement in the accuracy of NSDUH data is encouraging.