Customize your JAMA Network experience by selecting one or more topics from the list below.
Clark PI, Natanblut SL, Schmitt CL, Wolters C, Iachan R. Factors Associated With Tobacco Sales to MinorsLessons Learned From the FDA Compliance Checks. JAMA. 2000;284(6):729–734. doi:10.1001/jama.284.6.729
Context Tobacco products continue to be widely accessible to minors. Between
1997 and 1999, the US Food and Drug Administration (FDA) conducted more than
150,000 tobacco sales age-restriction compliance checks. Data obtained from
these checks provide important guidance for curbing illegal sales.
Objective To determine which elements of the compliance checks were most highly
associated with illegal sales and thereby inform best practices for conducting
efficient compliance check programs.
Design and Setting Cross-sectional analysis of FDA compliance checks in 110,062 unique
establishments in 36 US states and the District of Columbia.
Main Outcome Measure Illegal sales of tobacco to minors at compliance checks; association
of illegal sales with variables such as age and sex of the minor.
Results The rate of illegal sales for all first compliance checks in unique
stores was 26.6%. Clerk failure to request proof of age was strongly associated
with illegal sales (uncorrected sales rate, 10.5% compared with 89.5% sales
when proof was not requested; multivariate-adjusted odds ratio [OR], 0.03;
95% confidence interval [CI], 0.03-0.04). Other factors associated with increased
illegal sales were employment of older minors to make the purchase attempt
(adjusted ORs for 16- and 17-year-old minors compared with 15-year-olds were
1.52 [95% CI, 1.46-1.63] and 2.43 [95% CI, 2.31-2.59], respectively), attempt
to purchase smokeless tobacco (adjusted OR, 2.16 [95% CI, 1.90-2.45] vs cigarette
purchase attempts), and performing checks at or after 5 PM (adjusted OR, 1.28
[95% CI, 1.21-1.35] vs before 5 PM). Female sex of clerk and minor, Saturday
checks, type of store (convenience store selling gas, gas station, drugstore,
supermarket and general merchandise), and rural store locations also were
associated with increased illegal sales.
Conclusions This analysis found that a request for age verification strongly predicted
compliance with the law. The results suggest several ways in which the process
of compliance checks might be optimized.
After more than a decade of efforts to reduce youth access to tobacco,
tobacco products remain widely available to adolescents through retail sources.
In 1999, it was estimated that 3.76 million daily smokers aged 12 to 17 years
consume an estimated 924 million packs of cigarettes per year, generating
a retail value of $1.86 billion.1 Surveys consistently
show that minors believe they can easily obtain cigarettes,2-4
and that adolescents can readily purchase tobacco in retail outlets.5 Curtailing easy youth access to tobacco is a crucial
component in the primary prevention of tobacco use, and restricting retail
sales is an important element of reducing youth access. Given that tobacco
control resources are limited, it is important to understand the predictors
of sales to minors and thus design efficient compliance check programs to
identify retailers who sell tobacco to minors. This analysis of 110,062 compliance
checks performed by the US Food and Drug Administration (FDA) was undertaken
to determine what elements of the compliance check process are most likely
to result in illegal sales and therefore might be used in formulating best
practices for efficient checks.
Previous articles focused on youth access issues have identified factors
associated with illegal sales, including sex of the minor and the clerk, age
of the minor, ethnicity of the minor, and type of store visited. These studies
have reported conflicting findings about the direction and magnitude of sales
predictors, making it difficult to use the previously published literature
to determine the best practices for conducting checks.6
Older minors buy more often than younger minors.7-15
The sex of the minor has a mixed influence on illegal tobacco sales rates,
with some investigators finding girls could buy more often than boys,14,16-18 others
finding that boys buy more often than girls,19
and some investigators have found no sex difference.12,16,17
Studies that have reported the likelihood of sales by retail outlet type are
also mixed. At least 2 studies have shown lowest sales rates in pharmacies,20,21 1 found mid-level sales rates in
pharmacies,10 and 1 found high sales in pharmacies.22 Inconsistencies among previous studies may be the
result of small samples, few minors employed in the checks, and wide ranges
in the minors' ages. Employing few minors may be particularly problematic
given that the apparent age and maturity of a particular youth and that youth's
experience with conducting compliance checks can affect his or her ability
to purchase tobacco. These factors could partially account for the varying
sales outcomes when minors of the same age and sex try to buy tobacco. For
instance, DiFranza et al 23 reported widely
varying buy rates for three 16-year-old boys they employed to attempt to buy
smokeless tobacco. Purchase rates for each of the 3 ranged from 26.5% to 88.4%.
This would suggest that a large sample of compliance checks performed by a
large number of minors is required to offset the differences between minors
that might affect their ability to successfully purchase tobacco. The large
FDA data set, composed of data from compliance checks performed by more than
3000 minors, provides such information.
In 1996, the FDA asserted jurisdiction over cigarettes and smokeless
tobacco products and issued a rule regulating youth access to these products.24 The youth access provisions made it illegal for retailers
to sell cigarettes or smokeless tobacco to anyone under the age of 18 years
and required that retailers verify the age of anyone under the age of 27 years
by checking photographic identification (ID). To enforce the rule, the FDA
contracted with the states and territories to perform unannounced compliance
checks in which undercover minors visit retail cigarette outlets and attempt
to purchase tobacco products.
By December 1999, when this analysis was begun, more than 150,000 compliance
checks had been completed in 43 states and territories, providing the largest
number ever performed under a relatively uniform protocol.
The FDA trained and commissioned state officials to conduct compliance
checks on its behalf under protocols prescribing such things as ages of the
minors attempting the purchase, the procedures for conducting the purchase,
and the handling of evidence. In brief, trained and commissioned adult investigators
accompanied minors to the stores. The FDA required the adult agents be in
the store when the minors made purchase attempts unless the agent believed
that his or her presence in the store would signal a compliance check (usually
in very small stores). The minors attempted to buy the cigarette brands or
smokeless tobacco products frequently used by young people in their area and
had the option of purchasing other items, such as gum or chips, at the same
time. They were encouraged to carry valid photographic ID, if owned, and were
required to produce it at the request of a clerk. In addition, they were not
allowed to lie about their ages or for whom the tobacco purchase was made.
During the first year of the program, only 15- or 16-year-old minors
were employed. Subsequently, 17-year-olds were included, with the FDA requiring
a substantially equal mix of the 3 age groups. Minors were instructed to maintain
their normal, everyday appearance. If they typically wore make-up or facial
hair, they were permitted to do so during compliance checks, but were not
permitted to alter their appearance to appear older. States were instructed
to select minors who reflected the ethnic and racial characteristics of the
communities in which they conducted the checks. For reasons of safety, minors
did not conduct checks in their own communities. Minors were typically paid
by the states for their participation.
The FDA generally required that states conduct a minimum of 375 checks
per month. States that license tobacco retailers used their licensure lists
as a basis for identifying outlets to check. If a state could not generate
a list of tobacco retailers, the FDA provided a retailer list from a commercial
source. There was no attempt by the FDA to systematically sample stores in
the states; rather, the intention was to do a complete census of all tobacco
retail outlets. The store checks analyzed here represent checks of approximately
10% of the retail tobacco outlets in the United States.
This analysis used the first compliance check done in each unique store
(n = 110,062). The outcome variable was the outcome of the purchase attempt
(sale or no sale) at that first check. The relationships among potential explanatory
variables and between independent and dependent variables were explored through
frequency tables, appropriately stratified.
Logistic regression analysis was used to investigate the contribution
of the independent variables to the probability of illegal sales to minors.
Variables included in the model were those significantly associated with sales
in an exploratory analysis (including 2- and 3-level interaction terms) and
those reported as associated with sales in the published literature. Dummy
variables were constructed when appropriate. Odds ratios (ORs) were constructed
to reflect the ratio of the odds of a sale while controlling for the simultaneous
effects of the independent variables. Ninety-five percent confidence intervals
for ORs were calculated using SEs estimated by the Wald statistic.
To account for potential clustering of stores within communities, SUDAAN
(Research Triangle Institute, Research Triangle Park, NC) software was used
to account for any effect of clustering within states or within ZIP codes.
Measures of precision for the model parameters were approximately the same
as those generated by a simple log-linear model, suggesting that the degree
of homogeneity between stores within clusters was small and did not affect
The Compliance Check Record. Agents supervising compliance checks were required to complete a written
report immediately following each buy attempt. This written report documented
the outcome of the check, the type of establishment visited (convenience store,
convenience plus gas station, gas station, drugstore, general merchandise,
supermarket, tobacco store, or other), the date and time of the check, the
sex of the store clerk, the type of tobacco the minor attempted to buy (cigarettes
or smokeless tobacco), the minor's ID code, whether the minor was asked for
proof of age, and whether the minor carried a valid ID card.
The FDA's contract with the states did not require that they report
age and sex of the minors, and not all states had the resources to abstract
the sex and age of the minors from their records, particularly from their
earliest checks for the purpose of this analysis. The sex and age of the minors
was available for 81,181 checks (74% of first compliance checks). Table 1 shows a comparison of all first-time
checks and checks for which minor data were available.
Urbanicity of the Community in Which the Check Occurred.
The actual store address, including the postal ZIP code, was available
for most stores. Postal ZIP codes often cross over urban, suburban, and/or
rural areas. An urbanicity variable was constructed to represent the largest
proportion of households within each ZIP code, using 1990 US Census Bureau
data. Because uncorrected sales rates were lowest in predominantly urban ZIP
codes, dummy variables were constructed for suburban and rural areas, with
urban areas as the reference category.
Approximately 8000 ZIP codes were missing because a business address
was different than the address of the store location. After using commercial
business lists to locate as many missing store location ZIP codes as possible,
3733 (3.4%) remained missing.
The FDA completed 151,301 compliance checks in 110,062 unique tobacco
retail outlets between 1997 and 1999. The postal ZIP codes and minor characteristics
were available for 78,812 (72%) of the compliance checks in 36 states and
the District of Columbia. The rate of sales for all first compliance checks
in unique stores was 26.6%. The rate was 27.7% for checks in which there were
complete data. Characteristics of the checks are shown in Table 1 and results of the logistic regression modeling are shown
in Table 2.
Older age of the minor was associated with illegal tobacco sales, with
the odds of buying increasing with each year of age. Girls were more likely
to be able to successfully buy than were boys, and female clerks were more
likely to sell than were male clerks. The percent sales, corrected only for
minor sex and age and clerk sex, are shown in Table 3. Two- and 3-way interaction terms for sex of the minor,
age of the minor, and sex of the clerk were not significant in the exploratory
analysis and are not included in the model.
Clerks' requests for proof of age were highly associated with denial
of sales to the minors. However, some sales (10.5%) occurred even though the
clerk requested proof of age. The minors had been trained to produce valid
photographic ID for inspection if requested, although it is not known how
many actually did show an ID. The model was reproduced with the reduced data
set of only checks done by minors who carried their own valid ID cards (n
= 39,726). Results were similar, with 9.8% of sales completed when an ID card
Minors attempted to buy either cigarettes or smokeless tobacco. Only
2.3% of all buy attempts were for smokeless tobacco products. The corrected
buy rate was significantly higher when the attempt was for smokeless tobacco
compared with cigarettes.
Only 17% of the checks occurred after 5 PM (16% by boys and 18% by girls).
The rate of sales was flat until that hour, and then rose so that completed
sales were significantly higher after 5 PM compared with before 5 PM. Sales
were significantly higher on Saturdays than on any other day of the week.
The type of store was categorized by the agent, using his or her best
judgment. Consistent with past studies, there was variation in the sales rate
by the type of retail store visited. Convenience stores not selling gasoline
had the lowest rate of sales in the exploratory analysis, so that category
was set as the reference value. Sales rates were highest in gas stations.
Only the category "other" did not have a significantly higher sales rate than
did convenience stores.
More than half of buy attempts were in urban areas. In the logistic
model, both suburban and rural areas were significantly associated with increased
sales compared with urban areas.
This was an analysis of the largest available set of compliance checks
conducted under a relatively uniform protocol. The analysis suggested several
ways in which the process of compliance checks might be optimized. As with
most previous studies, older minors were more likely to be able to buy tobacco
products than were younger minors.7-15
To determine which retailers are more diligently complying with age restriction
laws, it is vital that older teens be included in the mix of teens doing compliance
Some of the most conflicting results in previously published reports
have been the effects of the sex of the minor.10,16-19
As previously discussed, these disparate results may be due in part to the
relatively few minors used in those studies, such that differences in perceived
maturity by sex may have contributed to the variation in results. Approximately
3172 minors contributed to the compliance checks for the FDA, providing stable
estimates of sales by age and sex.
In keeping with past research, this study found that a request for age
verification strongly predicted compliance with the law.15,23,25,26
It is not clear how often asking for an ID card is a serious request for proof
of age eligibility and how often it is the verbal mechanism that merchants
use to terminate the transaction when they have already decided that the buyer
is too young.
Interestingly, in 10% of the compliance checks in which clerks asked
for proof of age, they still sold to minors. Sales rates under these conditions
have ranged from 6% to 33% in previous studies.15-23
These findings may suggest an incomplete understanding of the "carding" or
age-verification process. The process necessitates 3 actions by the clerk:
requesting the ID card, inspecting the card, and calculating or verifying
the age of the buyer. If retailers merely request an ID card, without the
requisite inspection and calculation or verification of age, then age verification
may not actually occur. Clerks may sell tobacco even after requesting proof
of age because they cannot calculate age eligibility from a birth date. Training
programs for clerks typically teach them that they must request an ID card
and know how to spot a fake one, but they may not teach the clerk how to make
a correct decision about age eligibility once a card is presented.27,28
Previous studies have shown slightly lower or similar success rates
for smokeless tobacco purchases compared with cigarettes.21,23,29-32
In this analysis, attempts to buy smokeless tobacco products were almost twice
as successful as attempts to buy cigarettes. This phenomenon deserves more
study to determine why it occurs, and smokeless tobacco should more frequently
be included in compliance check programs.
The time of day when purchase attempts occur has been suggested as a
potential confounding variable.6 O'Grady et
al33 also found higher sales among checks performed
later in the day, with 6% sales among checks performed in the morning, 18%
in the afternoon, and 21% among checks performed after 6 PM. It is not known
why late afternoon sales rates were significantly higher in this and the previous
study, but it is possible that clerks who are on duty after 5 PM are younger
than daytime clerks and more inclined to sell tobacco products to their contemporaries.
Evening clerks may have less training, or they may be supervised and monitored
less closely than their daytime counterparts. Sales were higher on Saturday
than other days of the week, perhaps for the same reasons. Two conclusions
are clear: it is important to conduct some compliance checks during later
hours and on Saturdays, and owners and managers should improve the training
and supervision of evening and weekend clerks.
The sales rates in this sample were highest for gas stations and convenience
stores that also sold gas. The different sales rate by outlet type, however,
was not so great that any particular store class should be excluded from compliance
Sales rates were higher in rural and suburban areas compared with urban
areas, but more than half of the checks were performed in urban areas. These
results suggest that a better mix of stores will be an important component
of future compliance check programs, even though the relative proximity of
stores in urban areas makes urban checks more efficient to perform.
As with other analyses of administrative data sets, some additional
cautions are warranted. It is not known whether the loss of almost 30% of
the compliance checks because of missing data (primarily missing sex or age
of the minor and missing ZIP codes) may have introduced some bias. The stores
were not randomly sampled, and no national estimates of sales rates can be
inferred. It is not known if the sampling of stores for inclusion produced
any biases in the results. Also, when data are not acquired under a research
protocol, misclassification errors may be more likely. The most likely source
of error in this data set was in classification of the store type, which was
left to the judgment of the agent. In addition, the use of postal ZIP codes
for definition of urbanicity does not provide the precision that would exist
if the store neighborhoods were identified at the census block level. Whether
or not the adult accompanied the minor into the store was not consistently
reported; another important predictor of sales, the ethnicity of the minor,
was not identified. There is need for further research that explores the interaction
between clerk, minor, and neighborhood characteristics and the effect of witnesses
to the transaction on illegal tobacco sales.
This analysis explored a wide range of variables related to the illegal
sale of cigarettes or smokeless tobacco from a very large number of compliance
checks conducted across the country. These results provide important guidance
for public health officials responsible for curbing illegal sales to minors.
Retailers interested in stopping illegal sales in their own stores can
also learn from this analysis. First, all retailers need to understand that
no type of store is risk free. This study should serve as a wake-up call to
pharmacy employees, for example, who may not realize how easy it is for minors
to buy tobacco from them. Also, retailers must recognize the need to train
their clerks that a tobacco sale to a 17-year-old minor is just as illegal
as a sale to a 15-year-old minor. By the same token, a sale in which a clerk
asks for a photographic ID but sells anyway is as illegal as one in which
the clerk does not ask for an ID card. The results suggest that additional
training and monitoring of clerks who work in the evening and the weekends
may be needed. All clerks must be trained that selling smokeless tobacco products
to minors is illegal.
Between 1997 and 1999, the FDA completed more than 150,000 compliance
checks in about 110,000 retail establishments throughout the United States.
While that is the largest number of checks ever conducted by a single entity,
it represents only about 10% of the approximately 1 million retailers selling
tobacco in this country. On March 21, 2000, the Supreme Court ruled that the
FDA lacked the authority to regulate tobacco as customarily marketed.34 As a result of that decision, the FDA will no longer
be conducting compliance checks. Further, it is unlikely that sufficient funding
will be available from other sources to conduct compliance checks in every
store even once a year. As a result, efficient compliance check programs are
needed to conserve limited resources, while reducing illegal sales of tobacco
Create a personal account or sign in to: