Article Text

Download PDFPDF

Origin and use of the 100 cigarette criterion in tobacco surveys
  1. S J Bondy1,2,
  2. J C Victor2,
  3. L M Diemert2
  1. 1
    Dalla Lana School of Public Health, University of Toronto, Toronto, Ontario, Canada
  2. 2
    Ontario Tobacco Research Unit (OTRU), Toronto, Ontario, Canada
  1. S Bondy, Dalla Lana School of Public Health, University of Toronto, 155 College Street, 6th Floor, Toronto, Ontario M5T 3M7; sue.bondy{at}utoronto.ca

Abstract

Truly global standards and definitions will likely never exist for tobacco control surveillance. One difference across definitions of smoking status is whether or not a lifetime consumption of 100 cigarettes is a necessary criterion for ever and current smoking. Frequently asked questions about this measure demonstrate a need for information on its development and appropriateness in different settings. This commentary attempts to assemble information on the origin and adoption of this measure and provide some critical commentary on its usefulness.

The question has been traced to Canadian and American mortality cohort studies from the mid-1950s. From there it has spread to inconsistent use in many settings. To our knowledge, it was not originally (or since) empirically defined as a threshold of exposure related to health consequences or future smoking risk when used in youth.

Anecdotal evidence over several decades, however, shows the question has pragmatic utility in self-report data collection. It is a useful, if somewhat arbitrary, screener for “never regular” tobacco use among adults, where never smoking needs to be defined in data collection. Use of the criterion may lower prevalence estimates somewhat. Definitions must always be considered when creating time-trends or international comparisons.

There are also circumstances where it is inappropriate to exclude individuals who do not meet this criterion from further data collection, or reports. For research in youth, the criterion typically should be used only with more detailed information about experimentation, but it may be a useful additional indicator of established smoking.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

National and international programmes in tobacco control should be supported by high quality surveillance data14 and many groups have sought to standardised definitions of smoking status for surveillance purposes.1 5 6 There are many differences among definitions of current smoking, including whether occasional or non-daily smokers are included and whether an explicit time frame is used. Definitions of lifetime smoking vary as well, and include everything from a single puff through to daily smoking for a year.

The present work focuses on just a single source of disagreement between definitions: whether or not one requires a lifetime consumption of 100 cigarettes for current and lifetime smokers. We and colleagues at the Ontario Tobacco Research Unit have been involved in several Canadian national7 and provincial8 efforts to establish surveillance indicators for tobacco use and cessation. We have found the 100 cigarette criterion to be a source of debate, and of much curiosity. Similar questions about the criterion have appeared on discussion boards in tobacco control and public health.9 10

In our experience, questions about the origin of the 100 cigarette rule are motivated by the need for information on its initial psychometric development and validity, as would be expected for any other measure. Researchers and public health personnel are concerned about the validity of the measure for different purposes. Finally, professionals in surveillance are under considerable pressure to use “exactly the same questions”11 as other sources to maintain comparability and trend data, and so feel compelled to find the original, or definitive, versions of specific measures.

Here, we try to answer questions frequently asked about the 100 cigarettes rule. We have searched for the information on the earliest and subsequent uses of the 100 cigarettes question and how it was developed. In addition, we provide some commentary on the validity and usefulness of this criterion.

WHAT IS THE ORIGINAL SOURCE FOR THE QUESTION; HOW WAS IT USED?

The earliest examples we have found of this measure is a cluster of 3 studies from 1954 and 1955 using a similar threshold of 100 cigarettes to define lifetime smoking: 2 Veteran’s mortality cohorts (an American12 one and a Canadian13 14 one); and, the 1955 smoking supplement to the US Census Bureau’s Current Population Survey (CPS).15 This CPS shouldn’t be confused with the Cancer Prevention Studies16 17 which also began in the 1950s17 but did not use the question.

Although all three early sources used a similar question, it was applied differently. The CPS (fig 1) asked about “5 to 10 packs” in a lifetime, but only if a person did not smoke cigarettes at the time.18 Thus, the criterion segregated former from never users, but did not define current use. The American Veteran’s12 questionnaire seems to have been similar. An ever smoker had consumed “at least 5 to 10 packs of cigarettes or 50 to 75 cigars or 3 to 5 packages of pipe tobacco”.12 Other information was still captured and reported for people with lower or unknown total consumption.

Figure 1

Smoking use items in questionnaire for US Current Population Survey, 1955. Source: http://www.cdc.gov/nchs/data/series/sr_10/sr10_059.pdf.

The Canadian Veterans study differed:13

“A general question on smoking history also identified the individual as a non-smoker or as a smoker at any time during his life. The reported smoking of a total of 100 cigarettes or 10 cigars or 20 pipes of tobacco during the lifetime of the respondent qualified him as a smoker. For persons reporting a smoking history, the reverse side [captured] smoking history for cigarettes, cigars and pipe”.13

This may be the first example of the 100 cigarette consumption question being used as a skip pattern to determining whether other smoking behaviour is recorded, and its use to define current smoking.

The 100 cigarette question was subsequently used in the 1964 Adult Use of Tobacco Survey19 and the first US National Health Interview Survey (NHIS, 1965)20 21 providing a 10-year update on American smoking behaviour.18 The NHIS use (fig 2) differed from the CPS. Here, the 100 cigarette question appeared first, and formed a skip pattern.20 Only those who said “yes” were asked about current or past smoking. At that moment, 100 cigarettes became a de facto criterion for lifetime, former and current smoking alike, in countless broader health studies modelled after the NHIS.

Figure 2

Smoking use items from US National Health Interview Survey of 1965. Source: http://www.cdc.gov/nchs/data/series/sr_10/sr10_059.pdf.

HOW CONSISTENTLY HAS THE CRITERION BEEN USED?

Subsequent cycles of the NHIS and CPS tobacco supplements use the 100 cigarette question, as does the Behavioral Risk Factor Surveillance System from 1984 through to 2009.22 In 1994, the US Centers for Disease Control formally included the criterion, for surveillance, for lifetime and current smoking,5 but it is not a universal American standard. For example, the 1996 California Tobacco Survey uses the 100 cigarette criterion to distinguish former from never smokers, but current smoker status does not require this cumulative exposure.23

Canadian population studies did not adopt the 100 cigarette question early on. It appears in none of the Surveys of Smoking Habits (1966 to 1986),24 the 1978 Canada Health Survey,25 or other sources providing trend data.24 2631 These studies used self-reported status (eg, “Do you smoke daily, occasionally or not at all?”).

The 100 cigarette question reappears in Canadian surveys with a 1994 Health Canada workshop7 to establish common questions and definitions for smoking research and surveillance. The lead author shares credit (or blame) for bringing the 100 cigarettes question to that table. The workshop report concluded former and current smokers had to have smoked 100 cigarettes, recognising this as a change in practice.7

This workshop did not result in uniform definitions, even from Health Canada and Statistics Canada. The 100 cigarettes question was concurrently adopted in tobacco-specific surveys, namely: Survey of Smoking in Canada (SoSIC, 1994–1995)24 and Canadian Tobacco Use Monitoring Surveys (CTUMS, 1999–).32 In contrast, Federal Health Indicators (health care focused) make no mention of the 100 cigarette rule in defining who is a current smoker.33 Federal public health reports often cite figures where the criterion has not been applied.27 34 35

In the tobacco surveys that adopted the question (above), none used the item used as a hard skip pattern. Analysts can decide whether or not to apply the criterion.

A more recent Canadian advisory group study36 recommended that indicators for youth, specifically, included a single puff for experimentation but 100 cigarettes for “established smoking”.36 Technical notes on indicators for cessation, however, refer to categories (daily, occasionally, etc) without specifying if the 100 cigarette rule is applied.

Outside North America, the 100 cigarette rule has not been adopted extensively, although there are many examples of its use. In the 1998 World Health Organization (WHO) Guidelines for controlling and monitoring the tobacco epidemic,1 definitions for current and occasional smoker explicitly do not require a consumption of 100 cigarettes. Instead, this subdivides the non-smokers; “never” and “ex-occasional smokers” have smoked fewer than 100 cigarettes.1 Consistent with this, the current WHO Statistical Information System37 does not mention the 100 cigarette criterion for prevalence. European countries tend to use self-report smoking status.38

Other countries outside North America have at least occasionally introduced the 100 cigarette question and, in doing so, altered the definitions for smoking status in major surveillance surveys. For example, the Finnish annual adult health behaviour survey did not use the indicator from 1978 to 1995. From 1996 onward, the instrument included the item and it became required for the definition of ever regular daily smokers.39 For another example, national trend data for Australia have been presented combining data from various sources dating back to 1945.40 The Australian National Drug Strategy Household Survey, which started collecting smoking data using standardised measures in 2001 does include the 100 cigarette question,40 41 but in a unique way. Respondents who report they smoke daily are skipped the 100 cigarettes question, but respondents who smoke at least once a week, less often, or not at all, are asked the question about lifetime consumption.40 41 In 1997, a validation study examining the impact of the change in measures was undertaken within the survey sample for Victoria.41 In the evaluation study, 13% of smokers (ages 16 and older), by the original self-report definition, said they had not smoked 100 cigarettes and another 9% said “can’t say”.

In terms of multinational tobacco surveillance initiatives, recent surveys aligned with the International Tobacco Control Project42 apply the criterion to current smoking as the default.43 In contrast, the core questionnaire for the Global Adult Tobacco Survey (GATS) did not include a 100 cigarette item,44 although it does offer this as an optional item and country representatives have disagreed regarding whether the question should be adopted (GA Giovino, School of Public Health and Health Professions, Buffalo, New York, USA; personal communication).

WHAT WAS THE ORIGINAL PURPOSE OF THE CRITERION; HOW WAS IT ORIGINALLY VALIDATED?

Examination of publications on the health effects of smoking from the 1950s and 1960s16 17 4548 yielded no indication that 100 cigarettes (5 packs) was assessed at the time as a threshold below which increased risk was not found. In contrast, (where used) it was incorporated into the definition of never smoker used as the reference category against which other exposure levels were compared. The first Surgeon General’s reports47 48 make no mention of the 100 cigarette criterion and refer to “never regular smokers” in discussion of all studies reviewed, without stating that the definitions for this status varied across studies.32 Questions about ever regular smoking, and regular smoking for variable periods of time, were the most common means to approximate never smoking in studies up to the 1950s.49 50

WHY HAS THE 100 CIGARETTE QUESTION BEEN USED IN MANY STUDIES?

Available evidence points to the conclusion that the original question was simply meant to be useful in self-report data collection (CA Schoenborn, CDC National Center for Health Statistics, Hyattsville, Maryland, USA; personal communication), and that it amounted to a reasonable definition of never smokers. The NHIS questionnaire (and possibly the US CPS) are thought to have been written by Ron Wilson with these pragmatic goals in mind (CA Schoenborn, CDC National Center for Health Statistics, Hyattsville, Maryland, USA; personal communication. T Stephens, Thomas Stephens and Associates, Manotick, Ontario, Canada; personal communication).

When we recommended the item to Canadian researchers,7 we were retracing history. Our immediate concern was reliability of definitions for former and lifetime smoking, as well as the quit ratios.1 51 Telephone interviews prior to testing showed that general questions on lifetime smoking history are difficult for respondents and interviewers. Adults asked, “Have you ever smoked cigarettes?” frequently stop the interview for an explanation (eg, “Do you mean, was I a smoker, or that one cigarette when I was 12?”).

Doll and Hill45 expressed the same concern 40 years earlier.

“it was necessary to define what was meant by a smoker. Did the term, for example, include the woman who took one cigarette annually after her Christmas dinner, or the man of 50 who as a youth smoked a couple of cigarettes to see whether he liked it and decided he did not? If so, it is doubtful whether anyone at all could be described as a non-smoker”.45

Doll and Hill settled on having smoked every day for at least 1 year and concluded this was “reliable enough to indicate general trends and to substantiate material differences between groups”.45 This reminds us of the (non-pejorative) expression, “close enough for public health”, describing screening measures which are highly cost-effective for surveillance, but which can never satisfy every possible research purpose.

Survey methods literature also shows that explicit definitions of behaviour (eg, “even one puff of one cigarette”) yield more reproducible data.52 In longitudinal studies, former substance users sometimes become lifetime never users with time;53 it is likely they are unmotivated regarding long-past behaviour. Definitions improve face validity of questions and support respondent motivation (meaningful questions are worth answering).

Clear questions are also answered quickly but skip patterns have a huge advantage in reducing interview length for never regular smokers. In our own adult surveys,54 55 never regular smokers (by varying definitions) are spared some questions about past behaviour. The 100 cigarettes question serves this purpose very well, and provides a “not applicable” category comparable to external sources in our context. Alternatives, such as “never regular” smoking, may accomplish the same goal and may be preferred in populations with less emphasis on numeracy (GA Giovino, School of Public Health and Health Professions, Buffalo, New York, USA; personal communication). Mullins and Borland41 reported more confusion arising from questions asking about past “daily” smoking, relative to asking about 100 cigarettes.

One clear argument, therefore, for including a screener for never smoking is to reduce respondent burden and costs, as appropriate. However, when choosing to skip respondents past certain questions, an assumption is being made that the answer would be “not applicable” (or “no”) in the vast majority of cases. Little is lost by asking one or two pivotal questions about current behaviour before skipping the respondent out altogether and losing forever the opportunity to capture this information and to test this assumption.

WHAT DIFFERENCE DOES IT MAKE?

Figure 3 presents reanalysis of data from the CTUMS surveys indicating how current smoking prevalence varies by definition. Adding the 100 cigarette rule to either 30 day prevalence or self-report of every day or occasional smoking lowers prevalence estimates by 1 to 3 percentage points fairly consistently. In these data, lowest and highest prevalence estimates, within each year, were not significantly different. However, even small differences could be misinterpreted as change, or different prevalence differences, where definitions change. In the Australian validation study,41 the change in definition caused a reduction of less than 1% in prevalence estimates for regular current smoking (where the 100 cigarette question does not override self-reported current daily smoking).

Figure 3

Percentage of “current smokers” by varying definition of current smoking status. Canadians aged 15 and older in Canadian Tobacco Use Monitoring Surveys (1999 to 2006). ♦, 30 day prevalent smoker (1 or more cigarettes in past 30 days) (19.5% in 2006); , 30 day prevalent smoker +100 cigarette criterion (19.0% in 2006); ▾, self-reported “daily” or “occasional smoker” (18.9% in 2006); ×, Self-reported “daily” or “occasional” smoker +100 cigarette criterion (18.5% in 2006). Source: secondary data analysis, Statistics Canada. Estimates are population weighted. Pooled sample  = 171 948.

WHAT DO WE KNOW ABOUT ADULT CURRENT SMOKERS WHO HAVE NEVER SMOKED 100 CIGARETTES?

Table 1 presents the prevalence of a group of adults who had smoked within 30 days prior to the interview, but who did not report having smoked 100 cigarettes in their lifetime. The data come from CTUMS, which identifies this group in data reports as “beginning smokers”.32 Although this category is more common at younger ages; a few individuals appear to be taking up smoking past the age of 60. It seems unlikely that a consistent 1% of all those aged 45 years or older who smoked in the previous 30 days were caught into random samples just as they took up smoking.

Table 1 Percentages of Canadian current smokers (smoked 1 or more cigarettes in the past 30 days), who report having not achieved a lifetime consumption of 100 cigarettes

Some “no” answers among otherwise current smokers may be in error. The occasional respondent may interpret the question as 100 cigarettes in a day (a five pack a day smoker). Mullins and Borland41 suggested the same possibility. It is also interesting that the NHIS surveys added capitalisation on the words “entire lifetime” in 1997. Emphasis is typically added where fieldwork shows clarification is needed. In our CTUMS analysis, “no” answers (among otherwise current smokers) were more common at lower levels of education, (p<0.05, age-adjusted).

Our Ontario Tobacco Survey (OTS) longitudinal study of smokers provides limited data on current smokers,7 who would be never smokers when the rule is applied. Out of the first 3807 smokers recruited (had smoked within 6 months at baseline), we found a mere 56 individuals in this category (data not shown). These respondents included people in every category of self-report smoking status (“daily” through to “not at all”). Most (35/56; unweighted and treated as a convenience sample) both described themselves as “occasional smokers” and averaged between 1 and 3 cigarettes per day. However, 7 of the 56 averaged 4 or more cigarettes per day (roughly 114 cigarettes) that month; so, their answers to the 100 cigarettes question lack concurrent validity. Follow-up data (again limited), show they may be poorly motivated participants with a 50% loss to follow-up in 1 year (less than 20% overall).

The relative rarity of these inconsistent response patterns and the need for large sample sizes make these groups difficult to study. Additionally, there are no perfect questions on smoking status, or perfect respondents. Even subtle matters such as missing data can affect estimates to a small degree.41

WHAT ABOUT YOUTH SURVEYS?

Adoption of the measure into youth and school survey questionnaires happened at about the same time as for adults. The US Teenage Smoking Surveys (1968–1979)56 used the question without a skip pattern until the 1979 survey57 and added a skip pattern in 1979. In contrast, the National Household Surveys on Drug Abuse used a cut-off of five packs of cigarettes in 1979,58 but not in subsequent years. The Ontario Student Drug Use Surveys (1977–)59 added this as supplemental information in 1987, and it was included in the 1994 and subsequent Youth Smoking Surveys60 not in a leading skip pattern.

Use of 100 cigarettes (or 5 packs) to define prevalence categories for youth smoking has also varied, and this has complicated comparisons over time.61 It was suggested in the 1989 Surgeon General’s report that the 100 cigarette question might be under-reported, leading to overestimation of the prevalence of never and experimental smoking.61 As above, some Canadian surveillance reports have used the term “beginning smoker”, as has been done for adults.32 62

Research is limited with respect to whether a threshold of 100 cigarettes per se is a discriminating breakpoint in uptake of smoking. Nicotine dependence might be experienced with much less exposure63 64 and individuals may be distinguished in terms of liability for dependence with their very first puff.65 66 Several studies have looked at categories of prior smoking exposure in youth, up to 100+ cigarettes, and found that increasing levels of prior exposure were related to concurrent signs of dependence67 and risk of future smoking68 although increases in risk were also found across categories at lower levels of exposure than 100 cigarettes. Rapid movement through level of experience categories up to 100 cigarettes has also been related to susceptibility to dependence.69 Choi et al68 also found that the association between prior accumulated exposure and later risk of being an established smoker was mediated by cognitive factors. Even if somewhat arbitrary, the 100 cigarette level has been suggested as an intermediate outcome variable in research on the uptake of regular smoking,68 70 as a potential means to identify adolescents at greatest risk,69 and as a potential advance indicator of future changes in adult prevalence.71 72

Importantly, studies focused on youth should rarely rely on just the 100 cigarette threshold to measure experimentation. Far more detailed instruments, addressing the complexity of exposure and reaction, are more appropriate and more common. For example, The Global Youth Tobacco Survey initiative (GYTS)73 recommends several measures; none of which are the 100 cigarette criterion. Again, however, definitions used vary across studies.56 62 74 75

DISCUSSION

Most multi-purpose health surveys include any number of practical questions whose original source is unclear. Often these cannot feasibly be replaced with more detailed measures. However, it is useful to remind oneself of the limitations of such measures and not use them where research objectives are better served by more rigorous assessments.

From our review and experience, we suggest the following conclusions regarding the use and applicability of the 100 cigarette rule:

  • The indicator is common, but not universal. Important sources have included the criterion for current and former smokers, only to separate former from never smokers and for neither. So, following tradition is often a weak argument in choosing to use it or not use it.

  • One should not assume this cut-off point is valid for all research purposes; measures should be chosen based on research objectives and setting.

  • This measure is, no doubt, like other screeners used for the same purpose reliable enough to indicate general trends and to substantiate material differences between groups.45 It should at least be preferred over no definition at all, in asking about lifetime exposure.

  • Screening questions for ever smoking have practical advantages for data collection. However, one is cautioned against using them in such a way that it becomes impossible to apply other definitions. Examples from California, Canada and Australia (above), which also capture recent smoking behaviour, seem to be good models.

  • Adult smokers who say “no” to this question are not well understood. For youth studies, this indicator (alone) is too coarse to address the complexities of early experimentation well. However, it may also be of interest to document which young smokers who meet this adult criterion. Further research on the transition from experimentation to established smoking is needed.

Ideally all measures used in health studies should have known reliability, validity and source. Practically, however, it is sometimes difficult for researchers and public health staff to obtain information about where familiar measures come from, and which critically appraises available screeners and indicators.

What this paper adds

  • This article contributes to the literature by pulling together archival information, experience of researchers and limited secondary data analysis to answer questions frequently asked regarding a commonly used self-report question in tobacco use surveillance and research on smoking history: whether or not a person has smoked 100 cigarettes in a lifetime.

  • The provenance of simple questions used in multi-purpose health surveys is often not known and difficult to track down.

  • This summary was developed in response to repeated appearance of questions about the 100 cigarettes measure and criterion in the fields of epidemiology, public health and tobacco control research, as well as questions brought directly to us in the Ontario Tobacco Research Unit.

Acknowledgments

Roberta Ferrence, Ontario Tobacco Research Unit. Gary Giovino, University at Buffalo School of Public Health. Murray Kaiserman, Health Canada. Shawn O’Connor, Ontario Tobacco Research Unit. Charlotte Schoenborn, National Center for Health Statistics, US Centers for Disease Control. Thomas Stephens, Thomas Stephens and Associates and OTRU. Dennis R Wahlgren, Center for Behavioral Epidemiology and Community Health San Diego State University.

REFERENCES

Footnotes

  • Funding: This work was carried out at Ontario Tobacco Research Unit (OTRU) and The Dalla Lana School of Public Health, University of Toronto. OTRU receives funding from Ministry of Health Promotion as the research component of the Smoke-Free Ontario Strategy.

  • Competing interests: None.

  • Ethics approval: Ethical approval for the Ontario Tobacco Survey has been obtained from the University of Toronto and the University of Waterloo, Ontario, Canada.