The U.S. Food and Drug Administration on Jan. 23, 2017, published a proposed smokeless tobacco regulation that was based on erroneous calculations of smokeless tobacco risks.

The agency’s rules would require a radical reduction in N-nitrosonornicotine (NNN), a tobacco-specific nitrosamine, in smokeless tobacco products. Most American ST products today have very low NNN levels: about 2-10 millionths of a gram per-gram of product—that is, 2 to 10 parts per-million (ppm)—while a limited number of products contain 10-15 ppm.  The FDA proposes to limit the NNN level in ST products to 1 ppm or lower.

As is typical with FDA action on ST (see here and here), this proposal is based on obscure calculations and flawed assumptions and interpretation.  Here is a summary of the FDA logic:

  • Compared with never users, ST users have an excess risk of mouth and throat cancer (Relative risk, RR = 2.16).
  • ST use causes 276 deaths from mouth and throat cancer in the United States each year (this number was rounded to 300 in the proposed rule) – 268 men and 8 women.
  • ST products contain NNN, a human carcinogen that causes mouth and throat cancer.
  • Reducing NNN levels to no more than 1 ppm will reduce the number of deaths by about 115 per year.

While the latter two statements are by no means scientific certainties, in this post, I will focus on the demonstrably inaccurate first assertion, which is the flawed basis for the FDA’s calculation of “300” deaths due to ST use.

The FDA’s use of an RR of 2.16 is in error. The figure, based on a 2008 report, combines radically different RRs for men and women. In 2002, I documented that men, who generally use moist snuff (also called dip) or chew, do not have significantly elevated risk for mouth and throat cancer (RR ~ 1).  In contrast, women, who primarily use powdered dry snuff, have significantly elevated risk (RR ~ 4-6).

My research is in general agreement with risk estimates published last year by the National Institute of Environmental Health Sciences’ Annah Wyss and colleagues at the National Cancer Institute and over a dozen universities and health centers in the U.S. and beyond. They determined odds ratios (ORs, interpreted the same as RRs) among snuff and chew users for head and neck cancer, which includes voice box cancer in addition to mouth and throat, but the risk estimates are applicable.  The combined OR among snuff users, men and women combined, was 3.0, which is similar to the FDA estimate.  However, the OR among men was 0.86, while the OR among women was 8.89.  The ORs for chewing tobacco were not significantly elevated for men or women.

I will use the Wyss ORs to show how the application of risks specific to men and women change the FDA estimates. First, the following table (adapted from Table 5 in the FDA’s proposed rule) summarizes the data the FDA used (ST prevalence, RRs and total numbers of deaths from mouth-throat cancer in the U.S.) and the agency’s estimates for deaths attributable to ST (last column). The FDA erred in applying a single RR (2.16) for men and women when, in fact the RR is 0.86 for men and 8.89 for women.

Table 1. FDA Estimate of Annual Deaths from Mouth and Throat Cancer Attributable to Smokeless Tobacco (ST), Men and Women in the U.S.
Prevalence of ST Use (%) Relative Risk All Mouth/Throat Cancer Deaths Percentage Attributable to ST Number Attributable to ST
Men
Age 35-64 years 4.6 2.16 2,770 5.1 140
Age 65+ years 3.9 2.16 2,997 4.3 128
Women
Age 35-64 years 0.2 2.16 832 0.3 2
Age 65+ years 0.3 2.16 1,699 0.3 6
All 8,298 276

In the next table, I illustrate how the death estimates change if the specific ORs for men (0.86) and women (8.89) from the Wyss study are used.

Table 2. Revised Estimate of Annual Deaths from Mouth and Throat Cancer Attributable to Snuff Use, Men and Women in the U.S.
Prevalence of ST Use (%) Relative Risk All Mouth/Throat Cancer Deaths Percentage Attributable to ST Number Attributable to ST
Men
Age 35-64 years 4.6 0.86 2,770 0 0
Age 65+ years 3.9 0.86 2,997 0 0
Women
Age 35-64 years 0.2 8.89 832 1.56 13
Age 65+ years 0.3 8.89 1,699 2.31 39
All 8,298 52

As noted earlier, my revised estimates are specific to snuff use – powdered dry snuff for women and moist snuff for men. However, the risks for chewing tobacco in the Wyss study are not significantly elevated for men or women, so that product would add zero deaths.

In summary, the FDA failed to use RRs specific to men and women, resulting in overestimation of cancer deaths in the former and underestimation in the latter. Using the same FDA formula and assigning the proper risks, I estimate that 52 women using powdered dry snuff die each year from mouth-throat cancer, but the number of deaths among men who use dip or chew is zero. The proposed regulation is therefore unjustified for dip and chew products.


Image by vichni