Piper 2016.

Study characteristics
Methods	Study design: factorial RCT Country: USA Recruitment: smokers attending primary care clinics were invited to participate in a research programme to help them quit smoking
Participants	637 smokers; aged ≥ 18 years, ≥ 5 cigarettes per day for 6 months, motivated to quit 45.4% men; average age 45.8 years; average cigarettes per day 17.7; mean FTND 4.8; baseline CO 20.3 ppm; HSI 3.1
Interventions	2 x 2 x 2 x 2 x 2 x 2 factorial design. There were 6 intervention components tested (detailed below) that were tested in different combinations resulting in 32 study groups. 1) Nicotine patches for 3 weeks prior to quit date (patch preloading) versus no preloading patches 2) Nicotine gum for 3 weeks prior to quit date (gum preloading) versus no preloading gum 3) Preparation counselling versus no preparation counselling 4) Intensive cessation in‐person counselling versus minimal in‐person counselling 5) Intensive cessation telephone counselling versus minimal telephone counselling 6) 16 weeks of nicotine patch and gum from quit date versus 8 weeks of nicotine patch and gum from quit date For the purposes of this review, we are interested in comparisons 1, 2 and 6.
Outcomes	Self‐reported 7‐day PPA at 6‐month post‐quit date Self‐reported 7‐day PPA at 16‐week post‐quit date Validation: none Adverse events: measured in visits at weeks ‐1 and 4, and in calls at weeks 8, 16, and 26
Notes	This study had a factorial design, and an interaction between interventions was detected. However, results of a regression accounting for this have been presented in the publication and authors supplied group‐by‐group data. We checked to see if the odds ratios generated from these raw data were significantly, clinically different from those generated for the model adjusting for interactions in the paper, for comparisons 1, 2 and 6. Odds ratios were similar in all cases, and in all cases CIs indicated statistically non‐significant results. We have therefore entered raw data, supplied by authors, into meta‐analyses. This results in wider confidence intervals than the models accounting for interactions, but does not affect interpretation. The study was funded by grants 9P50CA143188 and 1K05CA139871 from the National Cancer Institute. Conflicts of interest: "The authors have received no direct or indirect funding from, nor do they have a connection with, the tobacco, alcohol, pharmaceutical or gaming industries or anybody substantially funded by one of these organizations. W.‐Y.L. is partially supported by a grant from Eli Lilly and Company for research that is unrelated to smoking or tobacco dependence treatment."
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “Participants were randomized to treatment conditions via a database that used stratified permuted block randomization”
Allocation concealment (selection bias)	Low risk	Quote: “Staff were blinded to randomization until eligibility was confirmed; participants were blinded until consent was provided.”
Blinding (performance bias and detection bias) All outcomes	High risk	No placebos. Quote: “assessed by staff who were not involved in treatment, but were not blind to treatment assignment”
Incomplete outcome data (attrition bias) All outcomes	Low risk	Loss to follow‐up < 50% overall (263/637), and similar for each of 6 study comparisons