Sixteen studies with a total of 351 patients (Table 4 includes only 345 patients). Forty-four studies were identified by the search, of which 16 were found to be homogeneous using the omnibus test, and 28 were excluded from the analysis. One study was excluded due to its lack of homogeneity with the pooled glucose data, 13 because they were not placebo-controlled; 3 because metformin was used; 5 because only abstracts could be obtained; 4 for miscellaneous reasons such as use of proinsulin rather than insulin, no English language translation, no evaluation, and a comparison with the combination therapy and therapy with the oral therapy alone; and 2 because data reported were insufficient, or median rather than mean values were reported.
Analysis of the 16 included studies found improved metabolic control (up to at least 16 weeks) with the combination therapy, as reflected by a significant lowering of fasting serum glucose (difference in treatment group -2.5 plus or minus 0.4, compared with -0.6 plus or minus 0.7 in control, p<0.01) and glycohaemaglobin concentrations (difference in treatment group -1.1 plus or minus 0.2, compared with -0.25 plus or minus 0.25 in control, p<0.025). Moreover, improved metabolic control was achieved with a significantly-smaller daily insulin dose (difference in treatment group -12 plus or minus 6, compared with -1 plus or minus 2.5 in control, p<0.01) and without a significant reduction in body weight. Finally, the combination therapy enhanced the endogenous insulin secretion, as expressed by an increase in fasting serum C peptide concentration (difference in treatment group 0.15 plus or minus 0.06, compared with -0.01 plus or minus 0.80 in control, p<0.05). The effect on lipids could not be clearly assessed because data were either non-uniform or insufficient in the selected studies.
Euglycaemic concentrations were not achieved in all studies because the goal was to observe the change in metabolic concentration, rather than to achieve euglycaemia or normalisation of haemoglobin A1c levels. Significant improvements in metabolic control together with a decrease in insulin levels were observed in the concurrent trials, but not the crossover trials. The authors suggest that the lack of uniformity may be explained by the differences in the study designs, especially in terms of daily insulin dosage. Most of the studies lacked distinct guidelines for adjusting insulin regimens.