Initiating Insulin in Patients Uncontrolled on Oral Agents 
There are two different approaches to initiating insulin therapy in patients with T2DM who are inadequately controlled on oral agents: either using insulin in combination with oral agents or using insulin by itself. Regimens using combination therapy with oral agents plus bedtime neutral protamine hagedorn (NPH) insulin have been proved to be as effective at achieving improved glycaemic control as regimens using insulin as monotherapy, even when two or more daily injections are given.11 Studies have shown potential advantages of continuing metformin in particular when insulin is added. These include improved control with less weight gain and hypoglycaemia.¹² The logic of combining basal insulin and oral agents is that the basal insulin suppresses hepatic glucose production and therefore controls the fasting glucose, which allows the oral agents to be more effective in controlling the post-prandial glucose and maintaining glucose control during the day.²NPH insulin has traditionally been the basal insulin used. Recent trials have evaluated the efficacy of adding long-acting basal insulin analogues to oral agents. In the Treat To Target Trial, Riddle et al. randomised patients inadequately controlled on oral agent therapy (HbA1c 7.5–10%) to bedtime NPH or glargine, and continued the oral agent therapy in both groups.¹³ A dose-titration algorithm was used to try to achieve fasting glucose levels below 100mg/dl. After 24 weeks, both groups achieved similar HbA1c values of about 6.9%, and about 60% from each group reached an HbA1c of <7%. More patients achieved an HbA1c of <7% without documented nocturnal hypoglycaemia with glargine than with NPH (p<0.05), and the rates of other categories of hypoglycaemia were also significantly lower with glargine. However, the authors noted that severe hypoglycaemia was uncommon with both treatment groups.Fritsche et al. randomised patients poorly controlled on oral agents (mean HbA1c was 9.1%) to glimepiride in combination with glargine in the morning or bedtime or bedtime NPH and used an insulin titration algorithm to try to achieve fasting glucoses below 100mg/dl.14 The greatest HbA1c reduction was seen with morning glargine (p=0.001 versus NPH and 0.008 versus bedtime glargine). Regardless of the injection time, glargine was associated with significantly less nocturnal hypoglycaemia than NPH (p<0.001), although the number of patients experiencing hypoglycaemia or severe hypoglycaemia was similar in both groups.In the Lantus®+metformin versus NPH+metformin (LANMET) study, Yki- Jarvinen et al. randomised patients uncontrolled on oral agents (HbA1c >8%) to bedtime glargine plus metformin or bedtime NPH plus metformin and titrated the dose to try to achieve fasting glucose values below 100mg/dl.¹⁵

After nine months, HbA_1c values were reduced similarly in both groups to about 7.1%. The frequency of symptomatic hypoglycaemia was significantly lower in the first 12 weeks in the glargine plus metformin group (p<0.05), but not significantly different thereafter. There were no differences in biochemical hypoglycaemia and no episodes of severe hypoglycaemia.Another long-acting basal analogue, detemir, has also been studied in combination with oral agent therapy.¹⁶ Patients inadequately controlled on oral agents (HbA1c 7.5–10%) were randomised to twice-daily detemir, or NPH with a dose-titration algorithm aimed to reach AM and PM glucose goals of <108mg/dl. After 24 weeks, both groups achieved an HbA1c level of <7%. The proportion achieving this level without hypoglycaemia was significantly higher with detemir than NPH (p=0.008), and less hypoglycaemia, including less nocturnal hypoglycaemia, was seen with detemir than with NPH (p<0.001).Pre-mixed insulin preparations that contain short- and intermediateacting insulins are an alternative way for patients to start insulin. Several recent studies compared initiating insulin with pre-mixed versus basal insulin in patients not achieving adequate control with oral agents. In the INITiation of Insulin to reach A1c TargEt (INITIATE) study, Raskin et al. randomised patients inadequately controlled onoral agents (HbA1c >8%) to an oral agent regimen of metformin with or without a thiazolidinedione (TZD) combined with either twice-daily biphasic insulin aspart (BIAsp) 70/30 or bedtime glargine.17 Insulin doses were titrated to target fasting glucoses (and pre-supper glucoses for the group on the pre-mixed insulin) of 70–110mg/dl by an algorithm-directed titration. After 28 weeks, HbA1c was significantly lower in the BIAsp 70/30 group than in the glargine group (6.9 versus 7.4%, p<0.01), and more reached an HbA1c of less than 7% (66 versus 40%, p<0.001). Of note, a significant difference in HbA1c reduction between the groups was seen only if the baseline HbA1c was >8.5% (p<0.05). Hypoglycaemia occurred significantly more often in the BIAsp 70/30 group (p<0.05). The authors stated that major hypoglycaemia was rare.