Center for Regulatory Effectiveness (CRE) assessment of
the following research report:
“Lower quit rates among African American and
Latino menthol cigarette smokers at a tobacco
K. K. Ghandi, J. Foulds, M. B. Steinberg, S. –E. Lu, J. M. Williams
a. Research Report: “Lower quit rates among African American and Latino menthol
cigarette smokers at a tobacco treatment clinic” The International Journal of Clinical
Practice, March 2009, pp. 360-367, by K. K. Ghandi, et al.
b. Research Report: “Relationship between menthol cigarettes and smoking cessation
among African American light smokers” Society for the Study of Addiction, No. 102,
May 2007, pp. 1979-1986, by Dr. Kolawole S Okuyemi, et al.
c. Clinical Study: “Factors associated with quitting smoking at a tobacco dependence
treatment clinic.” American Journal of Health Behavior, 2006, pp. 400-412, by J. Foulds,
K. K. Ghandi, M. B. Steinberg, et al.
On June 22, 2009, President Obama signed into law the Family Smoking Prevention and
Tobacco Control Act, which gives the Food and Drug Administration the power to further
regulate the tobacco industry. One element of the law imposes new warnings and labels on
tobacco packaging, with the goal of discouraging minors and young adults from smoking. The
bill bans flavored cigarettes, including cloves, cinnamon, candy, and fruit flavors, with a special
exception for menthol cigarettes. There is a need to investigate possible health hazards of
smoking menthol cigarettes as well as cessation (quitting) efforts.
The Tobacco Products Scientific Advisory Committee (TPSAC) provisioned under the bill is to
submit a recommendation on menthol cigarettes to the United States Secretary of Health and
Human Services no later than March 23, 2012. The intent of this CRE assessment is to consider
the merits and shortfalls of the study as well as present the reader with topics for further
discussion and investigation.
The report at reference (a) was identified for review and public discussion due to its focus on
smoking cessation rates among African American and Latino menthol smokers. The researchers
presented the following primary results:
Compared to White menthol smokers, African American and Latino menthol smokers
had lower quit rates at a four-week follow-up point;
African American and Latino menthol smokers had significantly lower odds (approx 1-
to-3) of quitting than their non-menthol counterparts;
Compared to African American non-menthol smokers, African American menthol
smokers had half the odds of being abstinent at a six-month follow-up point.
The researchers concluded “[d]espite smoking fewer cigarettes per day, African American and
Latino menthol smokers experienced reduced success in quitting as compared with non-menthol
smokers within the same ethnic/racial groups.” [p. 360]
The authors cite reference (b) findings as a motivator for further exploring the supposed
association between light-smoking (less than 10 cigarettes per day) minority group menthol
smokers and significantly lower quit rates. CRE would like to point out that reference (b) has
been assessed as having significant shortfalls (i.e. disproportionate comparisons between
menthol and non-menthol treatment and control groups; and the lack of accounting for no-show
patients by week 26). See Okuyemi, 2007.
The CRE conducted a limited assessment which comprised a review of the report and the
referenced primary clinical study (cohort data). Under the Data [Information] Quality Act, the
FDA is prohibited from using any information from a third-party, such as TPSAC, unless it
meets the requirements of the DQA. CRE has reviewed the study by Ghandi et al., and has
identified the following shortcomings, which if valid after outside peer review, would deem it
non-compliant with the DQA. CRE is requesting public comment for the material set forth
Summary of Findings and Issues
The cohort study described in reference (c) lacked control and replication, key elements to a
valid experiment. A data utility concern.
This report, performed by Ghandi, was described as a …“retrospective cohort analysis of 1688
consecutive patients who set a quit date and attempted to quit smoking between January 1, 2001
and June 30, 2005. They all attended a specialist tobacco dependence treatment outpatient clinic
in New Jersey.” This cohort study group scheme had very similar study characteristics as the
clinical trials that were used by the secondary Okuyemi (2007) report. However, unlike in the
clinical set-up re-used by the Okuyemi report, the cohort patients were largely observed (as it
should be with a cohort study). The cohort group of patients were given the opportunity to
receive medications, but no formal statistical treatment and control groups (i.e. such as an
administered placebo) were established. Furthermore, review of reference (c) does not include
statistics on accepted treatments within menthol versus non-menthol smokers.
Regarding replication (the ability of an independent researcher to re-create the experimental
setting), the Ghandi report cites 1688 consecutive patients studied between the period January
2001 and June 2005. However, a review of reference (c) indicates only an “…analysis of the
first 1021 patients, …from [the study’s] inception in January 2001.”
Are the statistical models valid?
The researchers reported that “African American and Latino menthol smokers had significantly
lower odds of quitting (Odds Ratio of 0.34; 95% Confidence Interval of 0.17 to 0.69 for African
Americans, and Odds Ratio of 0.32; 95% Confidence Interval of 0.16 to 0.62 for Latinos). All
confidence intervals are built with the assumption of data being Normally distributed or that a
researcher has a sufficient data size in its mean (average) tends to be Normal. In the Statistical
vernacular, the Central Limit Theorem is used to establish confidence interval conditions. In
particular, if the pedigree of a data set is unknown, then the Theorem states that the sample size
must be greater than or equal to 30 in order for the mean to approximate a Normal-like
CRE reviewed reference (c), the source of the clinical data, and found the following statistics for
Latino smokers (African American statistics are also shown for comparison sake):
4 Download full-text
Taken from Reference (c), Table 1, p.404 Abstinence at 4
Abstinence at 26
Notice that the 4-week and 26-week sample sizes (n) for the Latino group are smaller than the
required size of 30, in order to form confidence intervals. The main point is that when the
sample size is small (less than 30), then confidence intervals cannot be relied upon, regardless of
whether one is expressing intervals in terms of odds ratios or sample means.
What should be the most meaningful milestone for assessing cessation results, 4 weeks or 6
The researchers computed statistics for reported abstinence responses, across three ethnic groups,
at 4 weeks as well as 6 months. Their results were similar to that of the Okuyemi 2007 report
(e.g. lower abstinence rates among menthol smokers versus non-menthol smokers). However, as
was the case in the Okuyemi report, there appears to be a higher relapse rate among non-menthol
smokers over the longer period of time. This should be investigated further.
valid number N