How Many Americans Smoke?

The U.S. Centers for Disease Control and Prevention (CDC), provider of “official” statistics on American smokers, has generally reported a downward trend in adult smoking. For example, in 1998 the CDC said 24.1% of the population (47.2 million adults) smoked. By 2005, the CDC said the number had declined to 20.9% (45.1 million).

Are CDC smoking estimates accurate?

UAB epidemiologist Philip Cole and I asked this question in a new study, published in the European journal Public Health. In contrast to the decline reported by the CDC, we found evidence from another federal source that the number of adult smokers in the U.S. has been stable for about a decade. In 2005, for example, the CDC estimate was 45.1 million smokers; our analysis revealed that number could be as high as 54.2 million.

The CDC Counts Smokers Using the National Health Interview Survey

The CDC uses two questions from the National Health Interview Survey (NHIS) to establish the smoking status of participants. The first question is, “Have you smoked at least 100 cigarettes in your entire life?” Subjects who answer “no” are never smokers, and those who answer “yes” are ever smokers. Ever smokers are asked a second question: “Do you now smoke cigarettes every day, some days, or not at all?” Respondents who answer “every day” or “some days” are current smokers; those who answer “not at all” are former smokers.

We Counted Smokers Using the National Survey on Drug Use and Health

We used the National Survey on Drug Use and Health (NSDUH), which is sponsored by the Substance Abuse and Mental Health Services Administration (SAMHSA). Although NHIS and NSDUH survey data on adult smoking have been available for many years, no government agency has ever directly compared them. The Office of Applied Studies at SAMHSA acknowledged differences in the two surveys, but it offered no explanation for the discrepancy.

NSDUH asks a 100-cigarette lifetime question that is identical to NHIS. But NSDUH does not ask about every-day or some-day smoking. Instead, it asks questions about smoking in the past 30 days, which the CDC counts as current smoking. If subjects reported that they had smoked on all 30 days we called them every-day smokers; if they had smoked on 1 to 29 days, we called them some-day smokers.

NHIS and NSDUH Surveys Show Different Pictures of American Smoking

We found that NHIS and NSDUH describe an increasingly divergent picture of smoking in the U.S. over the 1998-2005 period. In 1998, smoking prevalence in both surveys was 24%. By 2005, NHIS prevalence had declined to 21%. But the NSDUH estimate was 25%, suggesting no change. The difference is not trivial: NSDUH’s 4.5% higher estimate in 2005 amounted to 9.1 million more smokers than the NHIS estimate.

Higher smoking prevalence estimates in NSDUH are largely due to much higher estimates for some-day smoking. This is not surprising, because NSDUH surveys included as some-day smokers persons who smoked as infrequently as one day in the past month. In the 2005 NSDUH, one-quarter of 18-34 year old some-day smokers and almost one-third of 35+ year-old some-day smokers smoked only on 1 to 4 days in the past month. If these very occasional smokers had been subjects in the 2005 NHIS survey, perhaps they would not have been counted as smoking at all.

Our findings are consistent with a recent study of college students who smoked at least one cigarette in the past 30 days, 50% of whom denied being smokers. Among the deniers, about 70% smoked on only 5 or fewer days in the past month.

Cigarette smoking has been described in many government publications as “the single most preventable cause of disease and death in the United States.” But in order to estimate accurately the consequences of smoking, it is necessary for the government to count smokers accurately. It is surprising that the CDC counts smokers using only one source, the NHIS survey, and that little research has been conducted to assess its accuracy. Our study raises doubt about the CDC’s claim of a recent decline in smoking. Investigation of this discrepancy by federal survey administrators may lead to better surveys and an improved understanding of American smoking trends.