In the commentary of Rentzsch et al. (Reference Rentzsch, Koller and Kronenberg2016) concerns are addressed to our article ‘Cannabis use and transition to psychosis in individuals at ultra-high risk: review and meta-analysis’ in Psychological Medicine (Kraan et al. Reference Kraan, Velthorst, Koenders, Zwaart, Ising, van den Berg, de Haan and van der Gaag2016). The authors raise questions about our methodology and the interpretation of our findings. We hope that the more detailed information of our study presented below can clarify our methods used and overcome the uncertainties.

The authors’ first concern is that in only two of the five studies used in the second meta-analysis the time-frame of Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV; American Psychiatric Association, 1994) abuse and dependence disorder was specified (Phillips et al. Reference Phillips, Curry, Yung, Yuen, Adlard and McGorry2002; Buchy et al. Reference Buchy, Perkins, Woods, Liu and Addington2014). In the studies of Auther et al. (Reference Auther, McLaughlin, Carrión, Nagachandran, Correll and Cornblatt2012) and Auther et al. (Reference Auther, Cadenhead, Carrión, Addington, Bearden, Cannon, McGlashan, Perkins, Seidman, Tsuang, Walker, Woods and Cornblatt2015), the Kiddie Schedule for Affective Disorders and Schizophrenia – Epidemiologic Version (Orvaschel & Puig-Antich, Reference Orvaschel and Puig-Antich1987) was used. This questionnaire is developed to measure cannabis abuse or dependence according to DSM-IV criteria. Using this questionnaire both lifetime cannabis use and cannabis use 6 months prior to baseline can be assessed, as has been described in the study of Auther et al. (Reference Auther, McLaughlin, Carrión, Nagachandran, Correll and Cornblatt2012). The study of Valmaggia et al. (Reference Valmaggia, Day, Jones, Bissoli, Pugh, Hall, Bhattacharyya, Howes, Stone, Fusar-Poli, Byrne and McGuire2014) used the Cannabis Experience Questionnaire. In this study data on lifetime use, current use and frequent cannabis use were reported. We agree with the authors that some of these studies did not provide current cannabis use. However, as we have specified in our Method section, our criterion for current use was defined as having used cannabis in the previous 12 months.

Their second concern is that the studies of Auther et al. (Reference Auther, McLaughlin, Carrión, Nagachandran, Correll and Cornblatt2012) and Auther et al. (Reference Auther, Cadenhead, Carrión, Addington, Bearden, Cannon, McGlashan, Perkins, Seidman, Tsuang, Walker, Woods and Cornblatt2015) have overlapping samples. Before we conducted our meta-analysis we contacted Dr Auther for specific details of the study samples. Only 42 of the 370 participants from the Auther et al. (Reference Auther, Cadenhead, Carrión, Addington, Bearden, Cannon, McGlashan, Perkins, Seidman, Tsuang, Walker, Woods and Cornblatt2015) study were used in the Auther et al. (Reference Auther, McLaughlin, Carrión, Nagachandran, Correll and Cornblatt2012) study. After comprehensive consideration we therefore decided to include both studies in our meta-analysis.

The authors correctly commented that in the study of Valmaggia et al. (Reference Valmaggia, Day, Jones, Bissoli, Pugh, Hall, Bhattacharyya, Howes, Stone, Fusar-Poli, Byrne and McGuire2014) cannabis use was not defined according to DSM-IV criteria. Although we used the data of the frequent cannabis users, we re-examined our meta-analytical data without this study to determine whether this would significantly alter the results. Excluding the study of Valmaggia et al. (Reference Valmaggia, Day, Jones, Bissoli, Pugh, Hall, Bhattacharyya, Howes, Stone, Fusar-Poli, Byrne and McGuire2014) our findings on the effect of current cannabis abuse or dependence remain significant (odds ratio 1.88, 95% confidence interval 1.17–2.59, p = 0.009).

We agree that our findings should be interpreted cautiously, because there are several confounding factors that were not controlled for. The authors argue that for instance alcohol use was not controlled for while five studies did report on alcohol use. However, in only two of the five studies that we could include in our second meta-analysis data on alcohol use were specified. The same applies for the other confounding factors mentioned by the authors; these were not systematically specified in the included studies and were therefore not controlled for in our analysis. To conclude, we fully agree that the relationship between cannabis use and psychosis is in need of further research and we believe that our meta-analysis contributed a small but important part in unravelling this relationship.