Extended Data Table 8. Protocol Amendments in the TriMaster randomised three way crossover trial.

Amendment no., protocol version, and date	Description
SA1 v3 06.07.16	Amendment to randomisation process to allocate individual bottles rather than ‘packs’ of 3 bottles to allow for shorter expiry dates, and clarification of safety reporting procedures
SA4 v4 20.03.17	Amendment to exclusion criteria to allow patients who have previously tried the study drugs to be included, as long as this has not been in the previous 3 months. The original criteria was unnecessarily strict and did not reflect real-world prescribing habits. The amendment also removed the blanket exclusion for patients in concurrent clinical trials, providing sufficient washout period between IMPs.
SA6 v5 01.08.17	Amendment to eligibility criteria to include patients taking metformin-only, or metformin and a sulfonylurea. This was adjusted due to the change in guidelines and prescribing trends leading to decline in use of sulfonylureas. At the time of study design sulfonylureas were the most commonly prescribed second line therapy in the UK. Subsequent decline in their use in favour of DPP4-inhibitors and SGLT2 inhibitors 22, resulted in the inclusion of patients currently treated with either metformin and sulfonylureas or metformin only. We will perform a sensitivity analysis to determine if the difference in study “epoch” (before/after this amendment) has any impact on the main study outcomes. Altered exclusion criteria also added ‘limb ischaemia’ due to updated safety information for Canagliflozin, and an upper limit of HbA1c >110mmol/mol.
SA9 v6 15.05.18	Amendment to sample size due to over-cautious sample calculations (alpha changed to 0.05), extension to recruitment period due to delays in regulatory approvals at study set-up and slow early recruitment, and additional secondary analysis included on the advice of the Data Monitoring Committee.
SA10 v7 22.02.19	Amendment to study analysis plan. Following advice from the Trial Steering Committee statistician, the protocol was amended to analyse only those completing at least 12 weeks on therapy, as this will determine whether the strata result in differences in response (we cannot adequately measure glycaemic response by HbA1c if the patient has been on the drug for less than 12 weeks). A separate analysis will be performed to determine whether the strata influence tolerability by assessing whether the proportion completing at least 12 weeks on therapy differs by drug and strata.
SA12 v8 20.03.20	Amendment to ensure ongoing participant safety and study integrity during Covid-19 pandemic. Urgent safety measures included (i) extension of visit windows to 14-18 weeks to allow greater flexibility for participants who are unwell/isolating, (ii) provision for remote visits with sample collection outside the usual research setting, (iii) ensuring participants remained on study therapy when only a remote visit is possible, by allowing an additional ‘continuation’ bottle of the same IMP to be issued, or when no other option, transfer to the next IMP without collection of blood samples.