Imputation Sequences in Cannabis Study
I have several questions for the authors of the recently-published paper in BMJ entitled, 'Prospective cohort study of cannabis use, predisposition for psychosis, and psychotic symptoms in young people.'(1) Before I enumerate my questions, I would like to assure the authors that I am asking them out of ignorance and for no other reason than my attempt to understand the methodology and statistical analysis of their work.
Here is what I do not understand:
1. The authors stated (page 2) that they used sensitivity analyses to examine whether differential attrition in the sample as a whole could have biased the findings... by multiple imputation of missing values of cannabis use at baseline, ..." It is my understanding that this was done to explain why there was an approximately 19.3% attrition rate (Of the original cohort of 3,021, 584 patients were lost to follow-up) in the study. I would appreciate if the authors would explain how the imputation of missing values gave them confidence that their 19.3% attrition rate did not bias their conclusions?
2. My second question is related to my first. I do not understand how (at the end of page 2) "Based on 1000 imputation sequences...(where the authors)..stochastically imputed missing values of cannabis use at baseline and psychotic symptoms...at four year follow-up in the whole sample.." gave confidence to their conclusions about the estimated average additive interaction between predisposition for psychosis at baseline and cannabis use at baseline? How is that possible?
The relationship between cannabis usage in early life and subsequent predisposition to psychopathology---or any human pathology (or the lack thereof)---remains a vitally important area of research. I appreciate the authors’ research.
(1)C Henquet et al. Prospective cohort study of cannabis use, predisposition for psychosis, and psychotic symptoms in young people.BMJ 2004; 0: bmj.38267.664086.63v1
Competing interests: None declared
Competing interests: No competing interests