Arjona Ferreira 2013.

Methods	Study design: double‐blind, parallel RCT Study time frame: not reported Duration of follow‐up: 54 weeks
Participants	Countries: multinational (number of countries not reported) Setting: multicentre (number of centres not reported) Inclusion criteria: patients with type 2 DM; moderate to severe chronic kidney insufficiency (eGFR < 50 mL/min/1.73 m2 using the MDRD equation); not on dialysis and unlikely to require dialysis for the duration of the study; HbA1c 7.0 to ≥ 9.0%, and were ≥ 30 years of age at the screening visit Number: treatment group (211); control group (212) Mean age ± SD (years): treatment group (64.8 ± 10.6); control group (64.3 ± 9.2) Sex (M/F): treatment group (80/55); control group (78/64) Exclusion criteria: taking insulin within 12 weeks of the screening visit; type 1 DM; history of ketoacidosis, AKI, kidney transplant or liver disease; recent (within 3 months) cardiovascular event; hepatic transaminase levels ≥ 2 times the ULN; thyroid stimulating hormone outside the reference range; triglycerides > 6.78 mmol/L; met one of the following prespecified glycaemic criteria: visit 2, FBG > 14.44 mmol/L, unlikely to improve with diet/exercise; visit 3, FBG > 13.89 mmol/L consistently (i.e. measurement repeated and confirmed within 7 days); visit 4, FBG > 13.33 mmol/L consistently; and visit 5, finger‐stick glucose > 13.33 or < 6.67 mmol/L
Interventions	Treatment group Patients with moderate kidney insufficiency received 50 mg/d of sitagliptin (2 x 25 mg tablets). The dose of sitagliptin was reduced from 50 mg/d to 25 mg/d for patients whose kidney status changed from moderate to severe Patients with severe kidney insufficiency received 25 mg/d of sitagliptin (1 x 25 mg tablet) After maximally titrating the matching placebo to glipizide, patients had insulin rescue therapy initiated, with the regimen and dose determined by investigator, if they met the following criteria: confirmed FBG > 13.33 mmol/L any time from randomisation to week 6; confirmed FBG > 12.22 mmol/L from week 6 to 12; confirmed FBG > 11.11 mmol/L from week 12 to 24; and confirmed HbA1c > 8% after week 24. Once insulin rescue therapy was initiated, patients continued to take blinded sitagliptin or matching placebo, but discontinued the matching placebo to glipizide. Control group Glipizide was administered at a starting dose of 2.5 mg/d, prior to the morning meal, and electively titrated to a maximum of 20 mg/d as considered appropriate by the investigator based on the patient’s glycaemic control. The dose of glipizide could also be reduced or interrupted to prevent hypoglycaemia. Patients received a placebo for sitagliptin. After maximally titrating glipizide, patients had insulin rescue therapy initiated, with the regimen and dose determined by investigator, if they met the following criteria: confirmed FBG > 13.33 mmol/L any time from randomisation to week 6; confirmed FBG > 12.22 mmol/L from week 6 to 12; confirmed FBG > 11.11 mmol/L from week 12 to 24; and confirmed HbA1c > 8% after week 24. Once insulin rescue therapy was initiated, patients continued to take blinded sitagliptin or matching placebo, but discontinued blinded glipizide
Outcomes	Change from baseline in HbA1c at week 54 FBG, fasting serum insulin and proinsulin, and plasma lipid profiles (total cholesterol, LDL, HDL, non–HDL‐cholesterol, triglycerides) Homeostasis model assessment–B‐cell function, HOMA‐IR, proinsulin/insulin ratio were calculated from fasting measurements of FBG, insulin, and proinsulin Proportion of individuals whose HbA1c values met glycaemic goals (< 7.0% as primary; < 6.5% as secondary) at week 54 Post hoc analysis evaluated the effect of sitagliptin versus glipizide on a composite end point consisting of glycaemic control (reduction in HbA1c > 0.5%), no body weight gain, and no hypoglycaemia Adverse events, physical examination and vital signs, and ECG Laboratory safety studies included serum chemistry, haematology, and urinalysis Hypoglycaemia were considered of special interest Events of hypoglycaemia requiring (non‐medical) assistance of others, requiring medical intervention, or exhibiting markedly depressed level of consciousness, loss of consciousness, or seizure were considered severe Change in body weight and GI adverse events (nausea, vomiting, diarrhoea, and abdominal pain)
Notes	The study included a 1‐week screening period, a diet/exercise and oral glucose‐lowering agent wash‐off period of up to 14 weeks, a 2‐week, single‐blind placebo run‐in period, and a 54‐week, double‐blind treatment period. At screening, patients not taking glucose‐lowering agents for ≥ 12 weeks with an HbA1c of 7–9% directly entered the single‐blind placebo run‐in period and those with an HbA1c > 9% entered a 6‐week diet and exercise period. Patients taking oral glucose‐lowering agents with an HbA1c of 7–9% entered an 8‐week drug wash‐off and diet and exercise period (those taking thiazolidinediones underwent a 10‐week wash‐off period), and those with an HbA1c of 6.5 to < 7% entered an 8–12‐week drug wash‐off and diet and exercise period (those on thiazolidinediones underwent a 10–14‐week wash‐off period). Patients received diet and exercise counselling throughout the study, consistent with American Diabetes Association recommendations and appropriate for their kidney insufficiency status. Following the placebo run‐in, eligible patients were randomised (1:1) using a computer‐generated randomisation schedule to receive sitagliptin or glipizide and their matching placebo. Data from one study site (3 patients) were considered potentially unreliable due to lack of compliance with Good Clinical Practice and excluded from all analyses.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	“Patients were randomised (1:1) using a computer‐generated randomisation schedule to receive sitagliptin or glipizide. Randomization was stratified based on: 1) kidney insufficiency status (moderate or severe), 2) history of cardiovascular disease (yes or no), and 3) history of heart failure (yes or no)".
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Sitagliptin and glipizide matching placebos were used to maintain blinding
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	The study was described as double‐blind, but the methods to ensure blinding of outcome assessment were not reported
Incomplete outcome data (attrition bias) All outcomes	High risk	22.3% of the sitagliptin group and 19.8% of the glipizide group dropped out
Selective reporting (reporting bias)	Low risk	The prespecified outcomes were available on a clinical trials database. All outcomes were reported
Other bias	High risk	Conflicts of interest: The study was sponsored by Merck Sharp & Dohme Corp., a subsidiary of Merck & Co., Inc., Whitehouse Station, NJ, the manufacturer of sitagliptin. J.C.A.F., H.G., G.T.G., C.M.S., K.D.K., and B.J.G. are employees of Merck Sharp & Dohme Corp., a subsidiary of Merck & Co., Inc., and may have stock or stock options in the company. N.B. has served on the National Diabetes Advisory Board. M.M. is a consultant for Association Diabete Risque Vasculaire, serves on the Merck global advisory board and the French subsidiary advisory board, and is a speaker for Merck, Sanofi, Novo Nordisk, Servier, and Abbott Diagnostics. No other potential conflicts of interest relevant to this.