Interpretation of the findings in relation to other studies
In the present study, based on a school sample of 4,277 adolescents aged 14-22 years, males appeared to be consistently less conscious about the risks of infectious diseases and mandatory hygienic norms when performing body art and were less likely to refer to a certified parlour to undergo body art and to seek medical advice in the event of related medical complications. By contrast, students attending the higher school years had a better knowledge of infectious diseases correlated with body modifications, were more likely to refer to certified body art parlours, and to know the hygienic norms mandatory in such salons. Schorzman  compared attitudes toward body piercing and awareness of the potential for health related problems between participants aged 17 to 20 years and those aged 21 to 25 years, between men and women and between those who had body piercing and those who had not. There were no statistical differences regarding age, sex, or race/ethnicity. Likewise, in a convenience sample of 225 participants, aged 12-21 years, recruited from an urban adolescent clinic, Gold  found that neither race, gender or age had a significant effect on the awareness of health risks. Houghton  instead, found that the highest level of awareness was among the group at greatest risk, boys with tattoos, suggesting that some males may be attracted to tattoos because of their known risks. The rationale of these conflicting findings is unclear, although an insufficient statistical power in the studies carried out by Gold,  Schorzman, and Houghton  could explain this in view of the small numbers involved.
In our study, differences in awareness/education on the safest ways of practicing body art to avoid unwanted complications were found across some geographical areas. Interviewees residing in the provinces of Rovigo and Vicenza (areas more rural) had a lower perception of body art risks. We would have also expected the Provinces of Padua, Verona, and Venice to be significantly more aware of the risks associated with body art, when compared with other parts of Veneto, as the latter three Provinces are served by large Universities, thus suggesting that people living in these areas are, on average, more educated. Similar geographical analyses have never previously been carried out, due to study groups being small and spatially restricted,[11–13, 26, 28] whereas population-based studies failed to either investigate the awareness of health risks [1, 18, 29] or the geographical distribution of the findings [18, 29].
In a Canadian study most secondary school students with body modifications indicated that they used the services of a body art professional, a result similar to Carroll . Houghton  instead, in a sample of Australian secondary school students, found that the majority of the tattooed participants had self-administered tattoos. An overestimation of the health risks associated with body art was reported by Schorzman, despite pierced participants estimating these risks at 30% with non-pierced participants estimating them at 43% (p < 0.001). Conversely, health risks as related to body piercing and tattooing were not seen as a threat by most participants, as the majority of respondents believed they had been pierced and/or tattooed in a safe, clean environment . Furthermore, Huxley  reported that a significant proportion of pierced and tattooed participants had not considered the health risks, while those who had were often unaware of potentially serious health problems. In our study, the majority of the respondents had a reasonable knowledge of related infectious diseases and hygienic norms to be applied in body modifications, considered it important to refer to a certified body art practitioner, would refer to a health-care professional in case of complications. However, students with tattoos were found to be less aware of the risks of the blood-borne infectious diseases potentially transmittable by body art tools. Furthermore, adolescents with a positive attitude toward body piercing (already having or considering it) were less keen to refer to a certified parlour to undergo body art, and those with piercings were less likely to seek medical advice in the event of related complications. Lastly, as body art is illegal for people younger than 18, the remarkable percentage of underage (roughly 50%) among those with piercings or tattoos could imply that both forms of body modification were performed illegally (in an unauthorized environment), or that they were carried out by the adolescents themselves or by their friends.
Such findings are important for family physicians (to better serve the body art population in their medical needs, offering preventive and tertiary interventions) and school educational staff (who can take an influential proactive role by sharing information, realistic concerns and care guidelines on tattooing and body piercing). Educators should be helped in assisting adolescents to become better informed decision makers, prevent risks and (where appropriate) dissuade them from tattooing and body piercing.
Strengths and weaknesses
A potential weakness of this study is the fact that the schools were not randomly selected. Instead of constructing a representative random sample, a stratified convenience procedure was employed to ensure representation of all the 7 Provinces of the Region and each of the 6 types of Italian public secondary schools. Moreover collaboration and engagement with the schools' head teachers depended on individual negotiations. Convenience and ad-hoc sampling strategies are the ones most commonly used in current research on body art among adolescents [11–13, 26, 28].
The study base (schools) presented both advantages and disadvantages. On one hand students were given time off during school hours to complete the questionnaire, whereas had the study been based on a random sample of individuals, respondents would have been required to complete the questionnaire in their own time which would probably have reduced the response rate. On the other hand, currently education in Italy is mandatory until the age of 16, and the sample collected for this study does not include adolescents who have dropped out of school. This may lead to a slight underestimation of the phenomenon, as drop-out and street youths are proportionately more likely to adopt risk-taking behaviours and thus undergo some types of body modification . Moreover our results cannot be generalised to include older populations.
As the sample exhibited a preponderance of female over male individuals, in order to improve its representativeness of the target general population the regression analysis was weighted by age and gender, using 2007 Census data of the Veneto Region.
Our definition of infectious disease knowledge is somewhat arbitrary, as we awarded the respondents a score according to what we thought were the main blood-borne diseases transmittable by body art. Each underlying disease was assigned the same weight (one point). Other infections could have been included,  however we selected diseases which we felt were appropriate to the educational level of the respondents.
We did not have any information on the socio-economic level of the parents as this was difficult to assess and to enquire about; however we felt that the educational level of the parents was a good proxy for socio-economic status.
The strength of the relationship (pseudo R2 statistic) between the various outcomes and the explanatory variables included in the relative regression models was quite scarce. Nonetheless, thanks to the large sample size, the findings were highly significant (Wald tests).
All the four regression models had similar AICs, including the model fitted for the Outcome 1 (despite having the smallest number of complete observations).
Every variable (explanatory and dependent) with missing values was transformed into a binary term being 0 if data were missing and 1 otherwise. These binary terms and the original variables of table 1 with complete information were fitted into four models of multivariable logistic regression (where the dependent variables were the outcomes 1 to 4, separately) in order to test for the missingness pattern. As no structural pattern was noted from the distribution of missing values, imputation was not felt to be necessary. Moreover, as the maximum percentage of missing information was around 10% (solely in one variable, father's age), complete-case analysis was considered a reasonable approach.
Despite these weaknesses, our study surveyed a large number of adolescents, whereas much of the available literature is based on substantially smaller samples of undergraduate students [11–13, 26]. A further strength is that our study focused on secondary school adolescents who have seldom been interviewed/questioned about body modifications [19, 21, 28, 29] and might be a better target for health education programs on body art. Lastly, the present study is the first to investigate the health risks associated with both piercing and tattoo in this age band.