There will be something. Basically, from the sounds of it you need to get it down to an IV and a DV and build up from there. Get a working hypothesis together essentially.
Even as simple a question as, are teenagers who smoke more likely to have parents who smoked before conception would be a start. It's essentially a retrospective study but you could build from there and gather additional information to see if you could identify any additional predictors of their likelihood of smoking, such as household income, age parents quit etc. Basically, you'd be looking to develop the picture from the basic correlation and actually identify the factors that increased the likelihood that a teenager would smoke.
It may not seem like the most interesting study but it would be moving the research picture on a little and there'd be some hefty data analysis to be done there that would use all the skills that the EPQ is designed to test. If you read some literature about uptake of smoking in your target population you will start to get a picture of what factors are predictive of uptake and you can start looking at which ones you'd like to test in your own study. Maybe come up with something entirely new.