LETTER TO THE EDITOR
National prevalence of oral HPV infection and related risk factors in the US adult population reply
Article first published online: 2 JUL 2012
© 2012 John Wiley & Sons A/S
Volume 19, Issue 1, page 106, January 2013
How to Cite
Sanders, A., Slade, G. and Patton, L. (2013), National prevalence of oral HPV infection and related risk factors in the US adult population reply. Oral Diseases, 19: 106. doi: 10.1111/j.1601-0825.2012.01954.x
- Issue published online: 7 DEC 2012
- Article first published online: 2 JUL 2012
- Accepted manuscript online: 6 JUN 2012 09:52AM EST
We appreciate these readers’ interest in our paper (Sanders et al, 2012) and confine our response to comments in their letter that address the methods, results, and interpretation of findings of our study.
In their letter, Prabhu et al state that our work ‘extrapolates data from 4846 adult subjects’ and they caution against over-interpretation. It is not necessary to study all people in a population to estimate prevalence and in practice that almost never happens. In our paper, oral HPV infection prevalence was estimated from the 2009–2010 National Health and Nutrition Examination Survey (NHANES). NHANES uses probability sampling methods to select people and to produce prevalence estimates that are generalizable to the community dwelling US population. In fact, the ability to estimate prevalence is the primary purpose of NHANES. Because prevalence estimates in NHANES are from a sample of study participants, it is essential that the estimates be interpreted in view of sampling error that is inherent in such surveys, which is why we report 95% confidence intervals throughout the paper. We made no attempt to extrapolate beyond the data.
Prabhu et al express concern about the use of the word ‘infection’ in the title. Throughout the manuscript, our use of ‘infection’ is consistent with Stedman’s Medical Dictionary (28th edition; Stedman, 2006), which defines infection as ‘Invasion of the body with organisms that have the potential to cause disease.’ In accordance with this definition, we do not imply the presence of disease.
Prabhu et al note that saliva is not recognized as a vehicle for transmission. While this consideration is beyond the scope of the objectives of our paper, we point out that in this study, DNA was purified from oral exfoliated cells, not from saliva.
A fourth criticism concerns our interpretation of attributable risk for oral squamous cell carcinoma (OSCC) associated with HPV infection. Last defines attributable risk as ‘the rate or proportion of a disease or other outcome of interest among exposed persons that can be attributed to an exposure of interest’(Last, 2007). This conventional measure can be readily calculated using two pieces of information: (1) the prevalence of exposure to a risk factor in a defined population; and (2) the strength of association between that risk factor and an outcome of interest in that population. In the context of this study, the exposure is oncogenic oral HPV infection and the outcome is OSCC. We combined the estimate of oncogenic oral HPV infection prevalence of 3.1% (95% CI: 2.4, 3.9) observed in the 2009–2010 NHANES with published findings from a case–control study of OSCC that estimated the strength of association between oral HPV infection and OSCC (Smith et al, 2004). Of note, that study used the same methods as used in 2009–2010 NHANES to evaluate oral HPV infection, yielding an odds ratio of 2.6 (95% CI: 1.5, 4.2) as the estimate of the association after adjusting for age, tobacco use, and alcohol consumption. If the measure of HPV exposure was either a necessary or sufficient cause of oral cancer (as might be the case, for example, if the study had measured altered protein expression in the host as a consequence of HPV integration into DNA), the corresponding odds ratio would be infinity. The attenuated odds ratio of 2.6 reflects, in part, the fact that HPV infection of oral exfoliated cells is an upstream marker of causal events contributing to cancer, as Prabhu et al explain. Its usefulness is as a practical measure to gauge excess risk in the population. Such markers need not be necessary or sufficient causes, and in practice, they almost never are. When we used the estimated association from Smith et al (2004) with our own estimate of prevalence, the point estimate for population attributable risk was 4.7%. The significance of this contribution to public health is that it permits interpretation that there would be a 4.7% reduction in incidence of head and neck cancer in the US adult population (exposed and unexposed) if exposure to oncogenic types of oral HPV infection was eliminated. Given that this estimate is based on the best quality data presently available, we respectfully disagree with Prabhu et al who suggest that this finding is premature.
We thank these readers for the opportunity to clarify these points. We trust that our reply might help to enhance a fuller understanding of epidemiologic principles and methods for the diverse readership of Oral Diseases.
- 2007). A Dictionary of Public Health. Oxford University Press: New York. (
- 2012). National prevalence of oral HPV infection and related risk factors in the U.S. adult population. Oral Dis18: 430–441. , , (
- 2004). Human papillomavirus in oral exfoliated cells and risk of head and neck cancer. J Natl Cancer Inst96: 449–455. , , et al (
- 2006). Stedman’s Medical Dictionary. Lippincott & Wilkins: Philadelphia, PA. (