| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
LETTERS TO THE EDITOR |
1 Department of Medicine, School of Medicine, University of Louisville, 3rd Floor, Brown Cancer Center, 529 South Jackson Street, Louisville, KY 40202, USA;
2 Department of Epidemiology, School of Public Health, University of Alabama at Birmingham, USA;
* corresponding author, brad.rodu{at}louisville.edu
To the Editor:
Fisher et al.(2005) claim to have published the first report of an association between smokeless tobacco (SLT) use and periodontal disease, based on odds ratios of about 2 that are only marginally statistically significant. However, their analysis is deficient. They did not control the reported association for education and socio-economic status (SES), two of the strongest correlates of periodontal disease that are also strongly and inversely correlated with SLT use. It is a certainty that control for education and SES will reduce the odds ratios and render them non-significant.
For over two decades, it has been standard practice to control for education and SES in studies of periodontal disease (Ismail et al., 1983). In fact, Fisher et al. included at least seven references that address this matter, the most relevant of which was derived from the same dataset that they used (Tomar and Asma, 2000).
Fisher et al. did not entirely ignore education. They reported that study subjects lacking a high school education were about three to four times more likely to have periodontal disease than were graduates, making education the third most important (after age and diabetes) correlate of periodontal disease in their own study. Thus, it is surprising that education was omitted from the multiple logistic regression analysis that they performed.
The Fisher et al. paper appeared as a Rapid Communication, a category the Journal designates for "definitive reports of findings of unusual significance" (Journal of Dental Research Instructions to Authors). Until the deficiencies in their analysis are corrected, their report is not definitive or significant.
FOOTNOTES
[NB: The authors are supported in part by unrestricted gifts from smokeless tobacco manufacturers to their respective institutions. The sponsors have no knowledge of this document.]
REFERENCES
Fisher MA, Taylor GW, Tilashalski KR (2005). Smokeless tobacco and severe active periodontal disease, NHANES III. J Dent Res 84:705–710.
Ismail AI, Burt BA, Eklund SA (1983). Epidemiologic patterns of smoking and periodontal disease in the United States. J Am Dent Assoc 106:617–621.[Abstract]
Journal of Dental Research Instructions to Authors, available at http://jdr.iadrjournals.org/misc/ifora.pdf
Tomar SL, Asma S (2000). Smoking-attributable periodontitis in the United States: findings from NHANES III. J Periodontol 71:743–751.[ISI][Medline]
3 Case School of Dental Medicine, Department of Orthodontics, 10900 Euclid Avenue, Cleveland, OH 44106-4905, USA;
4 University of Michigan School of Dentistry, Dept. of Cariology, Restorative Sciences & ENDO, 1011 N. University, Ann Arbor, MI 48109, USA
5 University of Alabama at Birmingham School of Dentistry, Department of Diagnostic Sciences, 1530 3rd Avenue South, Birmingham, AL 35294-0007, USA
* corresponding author, maf35{at}case.edu
We appreciate Rodu and Cole’s interest in our study and would like to address their concerns. Socio-economic status (SES) is a complex construct involving education, income, wealth, occupation, and a variety of other cultural factors. While variables for education and income are included in the NHANES III dataset, many of the other elements comprising SES are not available. We did not include the specific variables for education and income in our multivariable models, because of our concern for data sparseness in expanding the number of variables beyond those we presented in our paper. The problem we worked to avoid was the introduction of bias in our estimates of the association between smokeless tobacco use and severe active periodontal disease, while adequately controlling for confounding in our models to the degree that the data would allow. Unlike the Tomar and Asma report (2000) to which Rodu and Cole refer in their letter, our inclusion of smokeless tobacco use, analysis of interproximal periodontitis, and inclusion of analyses restricted to ‘never-smokers’ added extra limitations in sample size that Tomar and Asma did not experience.
However, we did not completely omit capturing important dimensions of SES in our multivariable models. Our considerations for selection of variables to control for in testing the association between smokeless tobacco use and severe active periodontal disease led us to include the variable ‘having a dental visit in the past year’, which is directly related to socio-economic status and closely associated with educational levels (Burt and Eklund, 2005), captures access to dental treatment, and influences periodontal status—established demographic factors that are also closely related to SES and periodontal status in the US (age, gender, and minority status), as well as established biologic risk factors for periodontal disease (cigarette smoking and diabetes).
Rodu and Cole mention that "it has been standard practice to control for education and SES in studies of periodontal disease". While we agree that it is important to adjust for SES in multivariable analyses where the data are available and sufficient to allow for appropriate adjustment, we believe that it may be debatable that it has been "standard practice to control for education and SES". A recent and very relevant critical review of the analytical epidemiology of periodontitis for the period 1994–2004 (Borrell and Papanaou, 2005) includes a useful summary (although not intended to be exhaustive) of reports of risk factors associated with the onset and progression of periodontitis. Education or income was not reported as being considered in 29 of the 40 studies of associations with periodontal disease presented in the two tables in that paper. Of the remaining 11 studies, 6 studies considered education and income, 2 studies considered education only, and 3 studies considered income only.
We would also like to clarify the statement by Rodu and Cole that we "claim to have published the first report of an association between smokeless tobacco use and periodontal disease". We actually reported:
"To our knowledge, this is the first report of the association between smokeless tobacco use and severe active [emphasis added] periodontal disease. Similar results were found for the association between smokeless tobacco use and interproximal severe active periodontal disease. When we removed the effect of smoking by restricting the analysis to never-smokers, the strength of the association between current smokeless tobacco use and both severe active periodontal disease and interproximal severe active periodontal disease was similar to that found for all adults, after smoking was taken into account in the multivariable analyses."
We believe that our study design and analytical approach are substantially supported by the current knowledge base and scientific evidence.
REFERENCES
Borrell LN, Papapanou PN (2005). Analytical epidemiology of periodontitis. J Clin Periodontol 32(Suppl 6):132–158.
Burt BA, Eklund SA (2005). Dentistry, dental practice and the community. 6th ed. St. Louis: Elsevier Saunders.
Tomar SL, Asma S (2000). Smoking-attributable periodontitis in the United States: findings from NHANES III. J Periodontol 71:743–751.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| IADR Journals | Advances in Dental Research ® |
| Journal of Dental Research ® | Critical Reviews (1990-2004) |
