ORIGINAL ARTICLE doi: 10.1111/j.1463-1326.2008.00976.x Long-acting insulin analogues vs. NPH human insulin in type 1 diabetes. A meta-analysis M. Monami, N. Marchionni and E. Mannucci Unit of Geriatrics, Department of Cardiovascular Medicine, Azienda Ospedaliero-Universitaria Careggi, Florence, Italy Aim: Basal insulin in type 1 diabetes can be provided using either NPH (Neutral Protamine Hagedorn) human insulin or long-acting insulin analogues, which are supposed to warrant a better metabolic control with reduced hypoglycaemic risk. Aim of this meta-analysis is the assessment of differences with respect to HbA1c (Glycated hemoglobin), incidence of hypoglycaemia, and weight gain, between NPH human insulin and each long-acting analogue. Methods: Of 285 randomized controlled trials with a duration > 12 weeks comparing long-acting insulin analogues (detemir or glargine) with NPH insulin in type 1 diabetic patients identified through Medline search and searches on www.clinicaltrials.gov, 20 met eligibility criteria (enrolling 3693 and 2485 in the long-acting analogues and NPH group respectively). Data on HbA1c and body mass index at endpoint, and incidence of any, nocturnal and severe hypoglycaemia, were extracted and meta-analysed. Results: Long-acting analogues had a small, but significant effect on HbA1c [-0.07 ( 0.13; 0.01)%; p ¼ 0.026], in comparison with NPH human insulin. When analysing the effect of long-acting analogues on body weight, detemir was associated with a significantly smaller weight gain than human insulin [by 0.26 (0.06;0.47) kg/m 2 ;p¼ 0.012]. Longacting analogues were associated with a reduced risk for nocturnal and severe hypoglycaemia [OR (Odd Ratio, 95% Confidence Intervals) 0.69 (0.55; 0.86), and OR 0.73 (0.60; 0.89) respectively; all p < 0.01]. Conclusions: The switch from NPH to long-acting analogues as basal insulin replacement in type 1 diabetic patients had a small effect on HbA1c, and also reduced the risk of nocturnal and severe hypoglycaemia. Keywords: glycaemic control, insulin analogues Introduction Basal insulin can be provided using either NPH (Neutral Protamine Hagedorn) human insulin or long-acting insulin analogues (i.e. glargine and detemir). Analogues, which are more expensive than NPH insulin, should provide greater reproducibility of absorption after subcutaneous injection, better metabolic control and reduced hypoglycaemic risk [1,2]. Although the use of glargine [3 5] or detemir [6,7] as basal insulin has been reported to improve metabolic control in type 1 diabetes, most available trials did not highlight any difference in HbA1c (Glycated hemoglobin) in comparison with NPH insulin [8 17]. In some studies of type 1 diabetes, the use of long-acting analogues has been associated with a lower risk of nocturnal hypoglycaemia [6,8,11 14,17], but not in other studies [3,7,10]. Only a few of the available studies detected a reduction in the overall incidence of hypoglycaemia [6,17]. Severe hypoglycaemia was reported to be reduced with analogues, in comparison with NPH, in only one study [11]; however, the small number of events recorded in each trial could Correspondence: Edoardo Mannucci, MD, Department of Cardiovascular Medicine, Section of Geriatric Cardiology, Azienda Ospedaliero- Universitaria Careggi, Via delle Oblate 4, 50141 Florence, Italy. E-mail: edoardo.mannucci@unifi.it 372 j Diabetes, Obesity and Metabolism, 11, 2009, 372 378
M. Monami et al. Long-acting insulin analogues in type 1 diabetes j OA have prevented the observation of clinically relevant differences. The combination of results of all available trials comparing long-acting analogues with NPH insulin could shed more light on the different effects on metabolic control and hypoglycaemic risk of alternative strategies for replacement of basal insulin in type 1 diabetes. The aim of the present meta-analysis is the assessment of differences with respect to HbA1c, incidence of hypoglycaemia, and weight gain, between NPH human insulin and each long-acting analogue in type 1 diabetes. Research Design and Methods A meta-analysis was performed including all randomized clinical trials, either with a cross-over or a parallel series design, enrolling patients with type 1 diabetes, with a duration of at least 12 weeks, comparing a long-acting insulin analogue (detemir or glargine) and human NPH insulin, combined with a prandial insulin. Trials with a shorter duration were excluded, due to the fact that they could not yield relevant information on glycated haemoglobin, which had been chosen as the principal outcome variable. Trials in which prandial insulin was not comparable between treatment arms (e.g. long-acting plus shortacting analogues vs. NPH plus regular human insulin) were also excluded, as well as those in which a long-acting analogue was compared with a formulation of human insulin different from NPH (e.g. ultralente). An extensive Medline search for detemir and glargine was performed, collecting all randomized clinical trials on humans up to 1 April 2008. The identification of relevant abstracts, the selection of studies based on the criteria described above, and the subsequent data extraction were performed independently by two of the authors (E.M., M.M.), and conflicts resolved by the third investigator (N.M.). The quality of trials was assessed using the parameters proposed by Jadad et al. [18]. The web site http://www.clinicaltrials.gov was also searched for identifying unpublished trials with the same characteristics reported above. The main results of unpublished Fig. 1 Trial flow diagram. CSII: continuous subcutaneous insulin infusion; RCT: randomized clinical trial. Diabetes, Obesity and Metabolism, 11, 2009, 372 378 j 373
OA j Long-acting insulin analogues in type 1 diabetes M. Monami et al. Table 1 Characteristics of the studies included in the meta-analysis Study (ref.) Insulin (type) Trial duration (weeks) No. of patients (LA/NPH) # Admin Setting Random Drop-out Definition of hypogl. ITT Sponsor Kolendorf [12] DT 16 130/130 2 MC NA NA <3.1 mmol/l Yes NN Pieber et al. [7] DT 16 271/129 2 MC A A <2.8 mmol/l Yes NN Home et al. [6] DT 16 276/132 2 MC A A <2.8 mmol/l Yes NN NN304-1604 [21] DT 24 56/27 1 or 2 MC NR A <3.1 mmol/l Yes NN Standl et al. [16] DT 26 154/134 2 MC NA A <2.8 mmol/l Yes NN Robertson et al. [13] DT 26 232/115 1 or 2 MC A A <3.1 mmol/l Yes NN Russell-Jones et al. [14] DT 26 275/157 2 MC A A <2.8 mmol/l Yes NN Vague et al. [17] DT 26 301/146 2 MC A A <2.8 mmol/l Yes NN NN304-1582 [21] DT 26 75/38 1 or 2 MC NR A NR Yes NN NN304-1476 [21] DT 48 196/98 1 or 2 MC NR A <3.1 mmol/l Yes NN De Leeuw [8] DT 52 216/99 2 MC A A <2.8 mmol/l Yes NN NN304-1595 [21] DT 104 331/164 1 or 2 MC NR A NR Yes NN Hassan et al. [20] GL 12 23/19 1 or 2 SC A A <2.7 mmol/l Yes SA Radman et al. [19] GL 12 26/26 1 SC NA A <3.9 mmol/l NR None Chatterjee et al. [3] GL 16 60/60 1 SC NA A <2.8 mmol/l No NN/SA Raskin et al. [10] GL 16 174/114 1 MC A A NR NR SA Schober et al. [15] GL 26 262/258 1 MC NA NA <2.8 mmol/l NR SA Home et al. [9] GL 28 264/270 1 MC A A <2.8 mmol/l NR SA Ratner et al. [11] GL 28 214/208 1 MC NA A <2.0 mmol/l Yes SA Porcellati et al. [5] GL 52 61/60 1 SC A A <4.0 mmol/l Yes None Total 30.0 3693/2485 # Admin: number of daily administrations of long-acting analogues; LA: long-acting analogues; Hypogl.: hypoglycaemia; ITT: intention to treat; DT: detemir; GL: glargine; MC: multicentre; SC: single centre; NA: not adequate; A: adequate; NR: not reported; NN: novo-nordisk; SA: Sanofi-Aventis. trials with detemir were retrieved from the web site, http://www.clinicalstudyresults.org, and included in the meta-analysis. The analysis was performed on summary data for each study. The principal outcome was HbA1c at the end of trial in patients receiving long-acting analogues, compared with NPH human insulin. Secondary outcomes included body mass index (BMI) at the end of the trial. Furthermore, data on the incidence of symptomatic, nocturnal, severe, or any hypoglycaemia (number of patients with at least one event) were extracted. Whenever possible, separate analyses were performed for trials with different insulin analogues (detemir and glargine). Standardized mean differences were calculated for HbA1c and BMI and a random effects model was used for the meta-analysis. Mantel Haenszel odds ratio with 95% confidence interval (MH-OR) was calculated for hypoglycaemia, using a random effects model. All analyses were performed using Comprehensive Meta-analysis Version 2, Biostat (Englewood, NJ, USA). Results The trial flow is summarized in figure 1. The Medline search allowed the retrieval of 16 trials; five more unpublished trials (one with glargine and four with detemir) were identified on www.clinicaltrials.gov. The main results of the four detemir trials were retrieved from http://www.clinicalstudyresults.org; a query to the sponsor (Sanofi-Aventis, Bridgewater, NJ) for the only unpublished trial with glargine (NCT00046501 HOE901/4030) received no answer. Of the 20 retrieved trials, all with a parallel series design, with the exception of two studies [3,12], enrolling 3693 and 2485 in the long-acting analogues and NPH group respectively (tables 1 and 2), 7 reported a significant improvement of HbA1c with analogues [3,5 7,19 21], while in the remaining 13 studies no difference was detected between groups with respect to HbA1c. Prandial insulin was provided as regular human insulin in four trials [6,11,15,16]; in the remaining 16 studies patients were treated with short-acting insulin analogues. Of the studies retrieved, two were designed as non-inferiority trials [8,13], while seven were aimed at assessing the superiority of analogues over NPH insulin [3,5 7,11,12,14]; this point was not specified in the remaining 11 trials. Overall, a significant reduction of HbA1c was observed with analogues [ 0.07 ( 0.13; 0.01)%; p ¼ 0.026], with different long-acting analogues yielding similar results on glycaemic control (figure 2). Nine trials with detemir [6 8,13,14,16,17,21], but none of those with glargine, reported a significantly smaller 374 j Diabetes, Obesity and Metabolism, 11, 2009, 372 378
M. Monami et al. Long-acting insulin analogues in type 1 diabetes j OA Table 2 Moderators and outcome variables in individual studies included in the meta-analysis Study (ref.) Age* (years) Duration of DM* (years) BMI baseline* (kg/m 2 ) BMI endpoint (kg/m 2 ) HbA1c baseline* (%) HbA1c endpoint (%, LA/NPH) Severe hypogl. (n, LA/NPH) Any hypogl. (n, LA/NPH) Nocturnal hypogl. (n, LA/NPH) Kolendorf [12] 39.2 16.6 25.3 NR 7.9 7.3/7.3 NR 116/118 58/81 Pieber 40.0 15.0 25.2 25.1/25.4 8.1 7.6/7.7 10/4 192/100 111/60 et al. [7] Home 40.0 17.0 25.2 25.1/25.5 8.6 7.7/7.9 17/10 228/107 106/64 et al. [6] NN304-1604 13.5 5.3 20.6 NR 7.4 7.5/7.7 5/3 53727 NR [21] Standl 41.0 16.0 15.4 25.1/26.1 7.7 7.9/7.8 18/14 135/113 102/94 et al. [16] Robertson 11.9 5.1 NR 19.2/19.4 8.8 8.0/7.9 37/23 223/113 174/101 et al. [13] Russell-Jones 40.9 17.1 25.1 25.0/25.5 8.3 8.3/8.4 24/17 251/140 190/110 et al. [14] Vague 40.0 17.2 24.5 24.3/24.8 8.2 7.6/7.6 24/21 271/138 198/110 et al. [17] NN304-1582 41.0 NR NR NR 8.4 NR NR NR NR [21] NN304-1476 42.2 13.2 22.3 NR 7.4 7.3/7.3 2/3 178/95 133/78 [21] De Leeuw 40.5 17.1 24.5 24.4/25.0 8.1 7.5/7.6 30/21 NR 180/87 [8] NN304-1595 35.0 13.0 24.7 25.3/25.6 8.4 7.4/7.6 NR 309/159 NR [21] Hassan 11.0 0.3 21.0 21.1/21.3 6.8 6.7/7.6 0/0 0/3 NR et al. [20] Radman 36.7 12.0 24.3 NR 8.3 8.0/7.1 NR NR NR et al. [19] Chatterjee 42.9 18.2 27.0 27.2/27.3 8.5 8.1/8.3 1/1 50/49 21/27 et al. [3] Raskin 39.0 18.5 25.6 NR 7.7 7.5/7.6 20/16 NR 214/195 et al. [10] Schober 11.6 4.2 18.8 NR NR NR 40/33 NR NR et al. [15] Home 39.0 16.0 24.9 NR 8.0 8.1/8.1 31/44 NR 178/179 et al. [9] Ratner 38.5 17.4 25.8 NR 7.7 7.5/7.5 5/15 NR 45/100 et al. [11] Porcellati 35.0 14.0 23.0 NR 7.1 6.7/7.1 NR NR NR et al. [5] Total 33.9 13.3 23.5 24.4/24.9 8.0 7.6/7.7 264/225 2.006/1.162 1.710/1.286 DM: diabetes mellitus; BMI: body mass index; HbA1c: Glycated hemoglobin; LA: long-acting analogues; Hypogl.: hypoglycaemia; NR: not reported. *Mean value between the two treatment groups. weight gain in comparison with NPH insulin. BMI at the end of the trial was reported in eight trials with detemir [6 8,13,14,16,17,21]; combining the results of those studies, detemir was associated with a significantly smaller weight gain than NPH human insulin [by 0.26 (0.06 0.47) kg/m 2 ;p¼ 0.012]. Such analysis could not be performed for glargine, as BMI at end of trial was reported only in two studies [3,20]. The total number of patients reporting at least one episode of hypoglycaemia was reported in only 12 studies [3,6,7,12 14,16,17,20,21], as summarized in table 2; meta-analysing those results, long-acting analogues were not associated with a significant reduction of hypoglycaemic risk in comparison with NPH insulin (figure 3). The incidence of severe hypoglycaemia was reported by all trials except three [15,19,21]; one more trial reported the number of events, but not the number of patients with events, and therefore could not be included in the analysis [12]. The total number of patients experiencing at least one episode of severe Diabetes, Obesity and Metabolism, 11, 2009, 372 378 j 375
OA j Long-acting insulin analogues in type 1 diabetes M. Monami et al. Overall Glargine Chatterjee Home Porcellati Raskin Ratner Radman Hassan Detemir De Leeuw Russell-Jones Home 2004 (b) Kolendorf Pieber2005 Robertson Vague Standl NN304-1476 NN304-1595 NN304-1604 Standardized mean differences (95%CI) -1.5-1.0-0.5 0.0 1.0 0.5 1.5 Fig. 2 Differences (with 95% CI, Confidence Intervals) between long-acting analogues and NPH insulin in the effects on HbA1c (Glycated hemoglobin) at endpoint. hypoglycaemia was 264 and 225 in the analogue and NPH group respectively [OR 0.73 (0.60; 0.89), p ¼ 0.002; figure 3]. The incidence of nocturnal hypoglycaemia, which was described for all trials except seven [5,15,19,20], was similarly reduced with long-acting analogues [OR 0.69 (0.55; 0.86), p ¼ 0.001; figure 3]. Detemir was associated with a significantly reduced risk of severe and nocturnal hypoglycaemia in comparison with NPH. In trials with glargine, which enrolled a smaller number of patients, the point estimated (MH-OR) was similar to detemir, although it was not statistically significant (figure 3). Discussion Insulin therapy in type 1 diabetes is based on the administration of a long-acting insulin formulation, for the replacement of basal insulin secretion, combined with short-acting insulin at each meal. Basal insulin replacement was originally provided by either NPH or ultralente insulin. Long-acting insulin analogues have been suggested as an alternative in the treatment of both type 1 and type 2 diabetes [1]. Currently available long-acting analogues are glargine (usually administered once a day) and detemir (usually administered twice daily). The higher cost of these drugs should be justified by a Mantel-Haenzel Hazars Ratio (95%CI) Any hypoglycaemia Severe hypoglycaemia Nocturnal hypoglycaemia 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.5 3.0 15.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Overall Glargine Chatterjee Home Raskin Ratner Hassan Detemir De Leeuw Russell-Jones Home 2004 (b) Kolendorf Pieber 2005 Robertson Schober Vague Standl NN304-1476 NN304-1595 NN304-1604 Fig. 3 Differences (with 95% CI, Confidence Intervals) between long-acting analogues and NPH (Neutral Protamine Hagedorn) insulin in the incidence of any, severe, and nocturnal hypoglycaemia. 376 j Diabetes, Obesity and Metabolism, 11, 2009, 372 378
M. Monami et al. Long-acting insulin analogues in type 1 diabetes j OA greater reproducibility of absorption, leading to better glycaemic control and reduced risk of hypoglycaemia. Some trials reported a significant reduction in HbA1c with long-acting analogues in comparison with NPH [3 7]. The meta-analysis of all available trials confirms a significant, although small, advantage of long-acting analogues over NPH insulin with respect to HbA1c. Once the desired glycaemic target has been reached, the drug to be preferred should be the one associated with the lowest hypoglycaemic risk. The overall incidence of hypoglycaemia does not appear to be different between long-acting analogues and NPH insulin. However, the present meta-analysis shows that long-acting analogues are associated with a significant reduction in the rate of nocturnal hypoglycaemia, as observed in some trials in patients with type 2 [22] and type 1 [6,8,11 14,17] diabetes. This confirms the results obtained in type 1 diabetic patients meta-analysing five trials with glargine for which patient-level data were available [23]. Furthermore, the combination of results of available studies showed a reduction of the risk of severe hypoglycaemia, which could not be detected in most clinical trials due to the small number of events. The risk of severe and nocturnal hypoglycaemia is reduced with long-acting analogues by approximately 30%. The two long-acting analogues seem to be associated with a similar reduction of hypoglycaemic risk in comparison with NPH, although statistical significance was reached with detemir, but not glargine, due to the smaller sample size. The switch from NPH to detemir seems to be associated with a modest, although statistically significant, weight loss. This result is consistent with what reported in type 2 diabetes in direct comparisons of detemir with NPH insulin [24] or glargine [25]. The clinical relevance of such differences in type 1 diabetic patients is questionable. Further studies are needed to establish whether differences in body weight are limited to the months immediately following the beginning of treatment with analogues. The cost of long-acting analogues is higher than that of NPH insulin (the Italian price for the penfill formulation is 5.48 and 2.34 Euro cents per Unit for analogues and NPH respectively). The decision to use these more expensive drugs should be balanced with the clinical benefits obtained. A cost-effectiveness analysis goes beyond the aims of the present paper. It should also be considered that the cost of analogues is different across countries. In any case, a correct economic analysis should take into account the reduction of direct and indirect costs (including the impact on the frequency of blood glucose self-monitoring) of the decreased hypoglycaemic risk associated with long-acting insulin analogues. Some limitations of this meta-analysis should be recognized. Most of the trials were sponsored by manufacturers of long-acting analogues. Differences in the criteria used for the definition of hypoglycaemia should also be considered when interpreting results on the rates of nocturnal hypoglycaemia; the only consistent definition of hypoglycaemia across trials was that of severe hypoglycaemia, that is, low blood glucose requiring help from a third party. Furthermore, in some trials the number of daily administrations of NPH insulin was different from that of long-acting analogues (notably glargine). On the other hand, a strength of this meta-analysis is represented by the retrieval and inclusion of all unpublished trials, except one. In conclusion, the switch from NPH to long-acting analogues as basal insulin replacement in type 1 diabetic patients has a small effect on HbA1c, and also reduces the risk of nocturnal and severe hypoglycaemia. References 1 Valensi P, Cosson E. Is insulin detemir able to favor a lower variability in the action of injected insulin in diabetic subjects? Diabetes Metab 2005; 31: 4S34 4S39. 2 Peterson GE. Intermediate and long-acting insulins: a review of NPH insulin, insulin glargine and insulin detemir. Curr Med Res Opin 2006; 22: 2613 2619. 3 Chatterjee S, Jarvis-Kay J, Rengarajan T, Lawrence IG, McNally PG, Davies MJ. Glargine versus NPH insulin: efficacy in comparison with insulin aspart in a basal bolus regimen in type 1 diabetes the glargine and aspart study (GLASS) a randomised cross-over study. Diabetes Res Clin Pract 2007; 77: 215 222. 4 Kudva YC, Basu A, Jenkins GD, et al. Randomized controlled clinical trial of glargine versus ultralente insulin in the treatment of type 1 diabetes. Diabetes Care 2005; 28: 10 14. 5 Porcellati F, Rossetti P, Pampanelli S, et al. Better longterm glycaemic control with the basal insulin glargine as compared with NPH in patients with Type 1 diabetes mellitus given meal-time lispro insulin. Diabet Med 2004; 21: 1213 1220. 6 Home P, Bartley P, Russell-Jones D, et al. Insulin detemir offers improved glycemic control compared with NPH insulin in people with type 1 diabetes: a randomized clinical trial. Diabetes Care 2004; 27: 1081 1087. 7 Pieber TR, Treichel HC, Hompesch B, et al. Comparison of insulin detemir and insulin glargine in subjects with Type 1 diabetes using intensive insulin therapy. Diabet Med 2007; 24: 635 642. 8 De LI, Vague P, Selam JL, et al. Insulin detemir used in basal-bolus therapy in people with type 1 diabetes is associated with a lower risk of nocturnal hypoglycaemia Diabetes, Obesity and Metabolism, 11, 2009, 372 378 j 377
OA j Long-acting insulin analogues in type 1 diabetes M. Monami et al. and less weight gain over 12 months in comparison to NPH insulin. Diabetes Obes Metab 2005; 7: 73 82. 9 Home PD, Rosskamp R, Forjanic-Klapproth J, Dressler A. A randomized multicentre trial of insulin glargine compared with NPH insulin in people with type 1 diabetes. Diabetes Metab Res Rev 2005; 21: 545 553. 10 Raskin P, Klaff L, Bergenstal R, Halle JP, Donley D, Mecca T. A 16-week comparison of the novel insulin analog insulin glargine (HOE 901) and NPH human insulin used with insulin lispro in patients with type 1 diabetes. Diabetes Care 2000; 23: 1666 1671. 11 Ratner RE, Hirsch IB, Neifing JL, Garg SK, Mecca TE, Wilson CA. Less hypoglycemia with insulin glargine in intensive insulin therapy for type 1 diabetes. U.S. Study Group of insulin glargine in type 1 diabetes. Diabetes Care 2000; 23: 639 643. 12 Kolendorf K, Ross GP, Pavlic-Renar I, et al. Insulin detemir lowers the risk of hypoglycaemia and provides more consistent plasma glucose levels compared with NPH insulin in type 1 diabetes. Diabet Med 2006; 23: 729 735. 13 Robertson KJ, Schoenle E, Gucev Z, Mordhorst L, Gall MA, Ludvigsson J. Insulin detemir compared with NPH insulin in children and adolescents with Type 1 diabetes. Diabet Med 2007; 24: 27 34. 14 Russell-Jones D, Simpson R, Hylleberg B, Draeger E, Bolinder J. Effects of QD insulin detemir or neutral protamine Hagedorn on blood glucose control in patients with type I diabetes mellitus using a basalbolus regimen. Clin Ther 2004; 26: 724 736. 15 Schober E, Schoenle E, Van DJ, Wernicke-Panten K. Comparative trial between insulin glargine and NPH insulin in children and adolescents with type 1 diabetes mellitus. J Pediatr Endocrinol Metab 2002; 15: 369 376. 16 Standl E, Lang H, Roberts A. The 12-month efficacy and safety of insulin detemir and NPH insulin in basalbolus therapy for the treatment of type 1 diabetes. Diabetes Technol Ther 2004; 6: 579 588. 17 Vague P, Selam JL, Skeie S, et al. Insulin detemir is associated with more predictable glycemic control and reduced risk of hypoglycemia than NPH insulin in patients with type 1 diabetes on a basal-bolus regimen with premeal insulin aspart. Diabetes Care 2003; 26: 590 596. 18 Jadad AR, Moore RA, Carroll D, et al. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials 1996; 17: 1 12. 19 Radman M, Jurisic D, Ljutic D, Jerkovic R, Kovacic N, Hozo IS. Assessing glycemia in type 1 diabetic patients using a microdialysis system for continuous glucose monitoring. Ann Saudi Med 2007; 27: 166 170. 20 Hassan K, Rodriguez LM, Johnson SE, Tadlock S, Heptulla RA. A randomized, controlled trial comparing twice-a-day insulin glargine mixed with rapid-acting insulin analogs versus standard neutral protamine Hagedorn (NPH) therapy in newly diagnosed type 1 diabetes. Pediatrics 2008; 121: e466 e472. 21 Novo-Nordisk web site. http://www.clinicalstudyresults. org. 2008. 22 Horvath K, Jeitler K, Berghold A, et al. Long-acting insulin analogues versus NPH insulin (human isophane insulin) for type 2 diabetes mellitus. Cochrane Database Syst Rev 2007: CD005613. 23 Mullins P, Sharplin P, Yki-Jarvinen H, Riddle MC, Haring HU. Negative binomial meta-regression analysis of combined glycosylated hemoglobin and hypoglycemia outcomes across eleven Phase III and IV studies of insulin glargine compared with neutral protamine Hagedorn insulin in type 1 and type 2 diabetes mellitus. Clin Ther 2007; 29: 1607 1619. 24 Hermansen K, Davies M, Derezinski T, Martinez RG, Clauson P, Home P. A 26-week, randomized, parallel, treat-to-target trial comparing insulin detemir with NPH insulin as add-on therapy to oral glucose-lowering drugs in insulin-naive people with type 2 diabetes. Diabetes Care 2006; 29: 1269 1274. 25 Rosenstock J, Davies M, Home PD, Larsen J, Koenen C, Schernthaner G. A randomised, 52-week, treat-to-target trial comparing insulin detemir with insulin glargine when administered as add-on to glucose-lowering drugs in insulin-naive people with type 2 diabetes. Diabetologia 2008; 51: 408 416. 378 j Diabetes, Obesity and Metabolism, 11, 2009, 372 378