Sodium Glucose Co-transport 2 Inhibitors in the Treatment of Type 2 Diabetes Mellitus

A Meta-analysis of Randomized Double-Blind Controlled Trials

Asres Berhan; Alex Barker


BMC Endocr Disord. 2013;13(58) 

In This Article


Search Strategy

Electronic based literature search was conducted in the databases of MEDLINE, HINARI, EBASE and The Cochrane Library by both authors (AB and AB). The literature search was further strengthened by searching relevant articles from the reference list of retrieved articles. During searching the following search terms were alternatively combined using the Boolean logic (AND, OR and NOT): sodium glucose co-transport (SGLT) inhibitors, dapagliflozin, canagliflozin, ipragliflozin, empagliflozin, sergliflozin etabonate, remogliflozin etabonate, tofogliflozin and type 2 diabetes.

Inclusion Criteria and Study Selection

The predetermined study inclusion criteria were: (1) randomized double-blind controlled trials of SGLT2 inhibitors in patients with type 2 diabetes mellitus; (2) studies which recruited patients with type 2 diabetes irrespective of their antidiabetic drug exposure history (naïve or drug experienced) but with an inadequate glycemic control (HbA1c ≥7.0); (3) studies written in English and (4) studies with a minimal duration of therapy for 12 weeks. The study selection of the retrieved literature was conducted in two steps: First, all the retrieved literature titles and abstracts were reviewed and then grouped either under "eligible for full document review" or "ineligible for full document review". Second, all literatures that were grouped under "eligible for full document review" were reviewed in detail and then grouped as "eligible for the meta-analysis" or "ineligible for the meta-analysis".

Data Extraction and Quality Assessment

Data extraction from the selected studies was conducted by both authors independently with the same data extraction template. Standard Excel spreadsheets were used for the data extraction. The following information was abstracted from the included studies: name of the first author, year of publication, sites of the study, the study design, duration of therapy, antidiabetic drugs used by the patients before they were recruited in the studies, antidiabetic drugs used in combination with SGLT2 inhibitors, dose, change in HbA1C(%) from baseline, number of patients with HbA1c < 7.0%, change in FPG, change in body weight, change in blood pressure, number of patients with adverse events, And number of patients who discontinued medication due to adverse events, experienced serious adverse events, experienced hypoglycemia, experienced urinary tract infection, and experienced genital tract infection.

Risk of bias in every of the included studies was assessed by the Cochrane risk of bias assessment tool. The predefined key domains were: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting and other bias. Based on the included articles, each domain was judged as "low risk of bias" or "unclear risk of bias" or "high risk of bias".

Data Synthesis and Statistical Analysis

Before the pooled analyses were conducted, some statistical transformations and unit conversions were performed. In case of continuous variables, where the standard deviation (SD) was not reported in the included studies, we computed the SD from standard errors (SE), 95% confidence intervals (CI) or P-values. When the value of FPG was reported in mmol/L, it was converted to mg/dl using an online converter.[17]

All effect sizes in this meta-analysis were computed using the random effects model. SMDs and the 95% CIs were computed for the changes in HbA1C (%), FPG, body weight, and blood pressure from baseline using the inverse variance method (IV). For dichotomous variables (adverse events, discontinuation of medication due to adverse events, serious adverse events, hypoglycemic events, urinary tract infection, and genital tract infection) ORs and 95% CIs were computed with Mantel-Haenszel method (M-H). When the 95% CI does not include zero for the SMDs and one for ORs, it was considered as statistically significant. Sensitivity analysis was performed by removing a study with a specific dose from the analysis at a time to evaluate the stability of the pooled values.

The consistency of the included studies was evaluated by the heterogeneity test (I2 statistics); when the value of I2 is greater than or equal to 50%, the included studies were considered as statistically inconsistent. To identify the possible sources of heterogeneity subgroup analysis and meta-regression were conducted. Publication bias was assessed by funnel plots and by funnel plot asymmetry test (Egger's test). All the statistical analyses were performed with OpenMetaAnalyst software and Review Manager (RevMan) Version 5.1 software. We reported the meta-analysis by following the PRISMA checklist.