What about vaping and chewing tobacco?  These are stealthier methods of obtaining nicotine and often more popular than cigarettes.  The questionnaire should at least ask about them.

Dear Dr. Robertson,

the questionnaire monitors tobacco use of non-conventional and conventional tobacco products in detail. How would you validate cigarette smoking biochemically?

Thanks,

Titus Brinker

Dear Titus,

I am not an expert on the chemical sides of things but I was wondering whether a) and b) are truly a problem if you take these into account in your statistical model? E.g. in a) if you conduct an ANCOVA to test for group differences, varying baseline levels could be factored in as covariates as long they are documented? b) unless you do not expect a bias for certain food intake in one of your group, shouldn't this confound cancel out in any group comparison?

Best wishes,

David

I do see that in the fine print of part c, sorry about that.  So it sounds like you want to know the prevalence of cigarette smoking as opposed to other methods.  Test c is the only reliable way to detect nicotine but is not specific to intake route.  I would say you can use Bayesian reasoning to deduce a prevalence.  

- What percent of students say they smoke but do not have cotinine?

- What percent of students say they use other nicotine but do not have cotinine?

- What percent of students have cotinine but do not indicate any nicotine use?

- Among stated nicotine users, what percent say they smoke?

All of these data can be combined to deduce a reasonably accurate estimate in a Bayesian model.

i.e. We want to add up all possible cases:

1. Admitted cigarette smokers with detected cotinine

2. Admitted cigarette smokers with no detected cotinine

3. Non-admitted cigarette smokers with detected cotinine

4. Non-admitted cigarette smokers with no detected cotinine

I would do hair testing for nicotine, which would give you the statistics for their last 3 months of usage. It would detect smoking, vaping and chewing, however is fairly expensive. If you didn't want to detect vaping, that is possible too - you can test for metabolites specific to tobacco in the hair.

see for NKK too (http://tobaccocontrol.bmj.com/content/early/2012/09/01/tobaccocontrol-2011-050298.full.html)

I think you can combine CO and cotinine

CO positive measurement in the morning seems fine for specificity and sensibility to diagnose daily smokers

cotinine negative to confirm no smoking no vaping

Validation is not considered necessary for large-scale trials in which the adolescents would not have a clear motivation to lie-- for example, tracking studies which have no cessation component.  Research generally indicates self-reports are adequately valid in this situation.

Biochemical testings among adolescents can include cotinine, thiocyanate, or carbon monoxide and the type of sample can be considered are air, blood, saliva.

One can consider cotinine cutoff values in saliva, serum, blood and urine. Validation of Self-Reported Smoking status using saliva or serum or blood or urine cotinine can be carried out by confidential interval, one of the most popular statistical approach.