Rosenstock 2002.

Methods	Parallel randomised controlled clinical trial Randomisation ratio: 1:1:1 Superiority design
Participants	Inclusion criteria: people with type 2 diabetes on insulin therapy (≥ 30 IU/day) for 4 months, 30‐75 years, HbA1c ≥ 8.0%, fasting C‐peptide > 0.7μg/l Exclusion criteria: history of ketoacidosis; unstable retinopathy, nephropathy or neuropathy; impaired hepatic and renal function; anaemia; unstable cardiovascular or cerebrovascular condition; concomitant lipid lowering medication must be taken for a stable dose > 60 days Diagnostic criteria: not stated
Interventions	Number of study centres: 79 Treatment before study: pioglitazone 15: 70.2 (34.0) insulin, pioglitazone 30: insulin 72.3 (38.5), placebo: insulin 70.7 (33.5) U/day (insulin regimen: 88% monotherapy) Titration period: insulin monotherapy: 2 weeks screening + 1 week single‐blind placebo insulin with oral anti‐diabetic (OAD) medication: 2 weeks screening + 4 weeks single‐blind placebo (= washout)
Outcomes	Primary outcome(s) (as stated in the publication): HbA1c, glucose, C‐peptide, lipids, insulin dose, safety profile: chemistry, haematology, urine analysis, vital signs, ECG, adverse events
Study details	Total study duration: 16 weeks Run‐in period: 3 weeks for insulin users and 6 weeks for insulin + OAD users (last group discontinued oral agent at the beginning of the screening period Study terminated early (for benefit/because of adverse events): no
Publication details	Language of publication: English Funding source: Takeda Pharmaceuticals Publication status: peer‐reviewed journal
Stated aim of study	Quote from publication: "To evaluate the ability of two doses of pioglitazone in combination with a stable insulin regimen to improve glycaemic control in patient whose type 2 diabetes remained poorly controlled despite current insulin therapy."
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Comment: not possible to judge whether the sequence generation was adequate.
Allocation concealment (selection bias)	Unclear risk	Comment: not possible to judge whether the allocation to the intervention and control group was concealed
Blinding of participants and personnel (performance bias) Adverse events	Low risk	Quote from publication: "... the 16‐weeks double‐blind treatment period" Comment: Probably the participants and the personnel were blinded
Blinding of participants and personnel (performance bias) HbA1c, FPG, lipids	Low risk	Quote from publication: "... the16‐weeks double‐blind treatment period" Comment: probably the participants and the personnel were blinded
Blinding of participants and personnel (performance bias) Insulin dose	Low risk	Quote from publication: "During the single‐blinded period, patients received their stable insulin regimens" Quote from publication: "... the16‐weeks double‐blind treatment period" Comment: the single‐blind period refers to the run‐in period and the 16‐weeks to the treatment period. Probably the participants and the personnel were blinded
Blinding of outcome assessment (detection bias) Adverse events	Unclear risk	Comment: unclear if the outcome assessor was blinded
Blinding of outcome assessment (detection bias) HbA1c, FPG, lipids	Unclear risk	Comment: unclear if the outcome assessor was blinded
Blinding of outcome assessment (detection bias) Insulin dose	Unclear risk	Comment: unclear if the outcome assessor was blinded
Incomplete outcome data (attrition bias) Adverse events	Low risk	Comment: data on weight gain and drug‐related adverse events were collected and reported
Incomplete outcome data (attrition bias) HbA1c, FPG, lipids	Low risk	Comment: outcome data were collected and reported
Incomplete outcome data (attrition bias) Insulin dose	Low risk	Comment: data on insulin dose were collected and reported
Selective reporting (reporting bias)	Low risk	Comment: all outcomes of interest were reported
Other bias	Unclear risk	Comment: funded by a pharmaceutical company