Prompted by Jurek's off. Doesn't matter who was at fault in the collision, so if you went into the back of a parked car, include it.
EDIT: to make this scientific, I'd need to adjust for location, type of roads travelled, typical time of day, hi-viz, lighting, years of cycling experience, I'm paranoid/I am like Colonel Kilgore in "Apocalypse Now" and I know I am going to get through this without a scratch yadda yadda, but it's years since I worked with multivariate linear regressions and I would need a sample size as big as the CTC.