This review shows that insulin analogs are cost-effective versus human insulins based on pharmacoeconomic models and retrospective database analyses.
Objective: To examine the cost-effectiveness of analogs versus human insulins, citing primarily studies conducted in the United States.
Study Design: The use of insulin analogs in type 1 and type 2 diabetes mellitus provides a better balance between glycemic control and hypoglycemia compared with human insulins, with the resultant potential to reduce the costs of treatment of hospitalization and chronic complications. The lower incidence of hypoglycemia seen with analogs versus human insulins may help to overcome barriers to insulin acceptance among patients with type 2 diabetes mellitus. In type 2 diabetes mellitus, prompt initiation or intensification of insulin therapy could save costs by delaying the development of complications.
Methods: The cost-effectiveness of analogs was analyzed through a literature review. Searches were conducted in PubMed to identify articles in the past 5 years with the name of any insulin analog plus the word cost or economic in the title or abstract. American Diabetes Association abstracts for 2005 to 2007 were also searched.
Results: Pharmacoeconomic modeling studies have consistently shown that insulin analogs provide gains in quality-adjusted life-years at costs well below accepted cost-effectiveness limits. In these studies, increased prescription costs were offset by reductions in complications. Retrospective analyses of healthcare databases have also shown cost-effectiveness for analogs versus human insulins, primarily because of lower inpatient care costs.
Conclusion: Treatment with insulin analogs has been demonstrated to be cost-effective versus other options over time and is an appropriate investment of healthcare dollars.
(Am J Manag Care. 2008;14(11):766-775)
Treatment of type 1 and type 2 diabetes mellitus with insulin analogs is cost-effective versus other options over time.
It is well known that the costs of diabetes mellitus (DM) are high. The American Diabetes Association (ADA)1 estimated that in the United States the 2007 costs for DM were $174 billion, composed of $116 billion in direct medical expenditure and $58 billion attributed to lost productivity. The ADA also estimated that healthcare costs are more than doubled when DM is present.
Much of this increased expenditure is accounted for by the complications of DM.2 Landmark clinical trials such as the Diabetes Complications and Control Trial and follow-up studies and the United Kingdom Prospective Diabetes Study have shown that improving glycemic control reduces microvascular and macrovascular complications.3-6 For example, it is estimated that every 1% decrease in glycosylated hemoglobin (A1C) level is associated with a 37% reduction in microvascular events and a 21% reduction in DM-related mortality (P <.001).6 These are epidemiological associations and not causality studies; nevertheless, on the basis of these and similar studies, leading entities such as the ADA7 and the American Association of Clinical Endocrinologists8 have recommended target maximum levels for A1C. However, surveys have shown that in the United States many patients fail to achieve these targets.9,10
The costs of DM include direct and indirect costs. Among the direct costs, the cost of the drugs alone is estimated to account for only 10% to 20%, with a much larger proportion of the direct costs being contributed by the costs of healthcare services directly related to DM or associated with complications of DM. For example, a breakdown of DM costs in the United States in 2007 assigned amounts as follows: $27.7 million for outpatient medication and supplies, $65.8 million for institutional care, and $22.7 million for outpatient care.1
Indirect costs of DM cover aspects such as lost productivity. Drug treatment that improves glycemic control can reduce the costs of DM, primarily by delaying or preventing the complications of the disease and the associated morbidity and mortality. This was illustrated by a study11 in which the authors reviewed cost-benefit analyses for 17 DM interventions, with reduction of A1C level shown to be cost-effective. In that study, Klonoff and Schwartz11 noted that intensive insulin therapy (≥3 insulin injections/d, with dose adjustments based on frequent glucose monitoring) was cost-effective versus conventional therapy (1-2 insulin injections/d, with less stringent glucose goals), based on the results of interventional studies12,13 that used human insulins (analogs were unavailable at the time).
Many patients with type 2 DM will eventually need insulin as their disease progresses and as they become unable to maintain glycemic control using oral antidiabetic drugs (OADs). However, patient acceptance of insulin use is still low, due in part to injection phobias and concerns related to adverse effects such as hypoglycemia or weight gain.14,15 In recent decades, insulin analogs have been developed with the aim of overcoming some of the disadvantages of conventional human insulins. Analog molecules resemble human insulins except for small changes in amino acid sequence or the addition of a fatty acid chain, which results in different pharmacokinetic and pharmacodynamic profiles. The analogs can be used in regimens that mimic the action of endogenous insulin more closely and provide a less variable and more consistent effect than conventional human insulins. The available insulin analogs are listed in .
In this article, evidence is reviewed about the cost-effectiveness of insulin analogs compared with human insulins, primarily using pharmacoeconomic studies conducted in US settings. Despite the well-established acceptance of insulin analogs by clinicians and patients, such studies are important because the higher initial cost of analogs versus conventional insulins means that their value has to be demonstrated to payers and to authorities. This review presents a brief description of the clinical benefits of insulin analogs and how these may help to overcome barriers to insulin acceptance, followed by an overview of pharmacoeconomic studies that were available as of 2007.
Advantages of Insulin Analogs
Insulin analogs have consistently been shown to improve the balance between glycemic control and tolerability (in terms of hypoglycemia) in type 1 and type 2 DM compared with human insulins. A detailed overview was published by Gough.16 Some representative studies17-29 comparing insulin analogs with human insulins are summarized in . These trial durations ranged from 12 to 26 weeks.
In type 1 DM, lower A1C and postprandial glucose levels were achieved with rapid-acting analogs versus human insulins, and hypoglycemia risk was reduced. With basal analogs, A1C levels were generally similar to those achieved with conventional insulins, as the studies were treat-to-target trials in which insulin doses were titrated upward to reach prespecified target glycemic values. However, hypoglycemia risk was reduced by treatment with analogs.
In type 2 DM, studies were generally treat-to-target trials, and A1C levels were similar for analogs versus conventional insulins, but hypoglycemia was lower with analogs. Therefore, cost-savings can result if the incidence of hypoglycemia requiring treatment is reduced (discussed herein in the “Retrospective Analyses of Databases” section) and, in the longer term, if preventing complications of DM is obtained through better glycemic control.
In type 2 DM, patients may be unwilling to initiate or intensify insulin therapy, exhibiting so-called psychological insulin resistance. This reluctance to use insulin is based on factors that include a perception of failing in their disease control efforts, fear of hypoglycemia and weight gain, and needle anxiety.30 However, prompt initiation or intensification of insulin therapy is vital to delay or reduce the development of complications. Analogs may have a role in overcoming psychological insulin resistance, as discussed herein.
The reputed potential for hypoglycemia is widely believed to be a major barrier to insulin acceptance by patients.30-32 For example, 41% of respondents in a study33 agreed that possible hypoglycemia might explain the reluctance to begin insulin therapy. However, analogs are associated with less frequent and less severe hypoglycemia,16,34,35 which may encourage patients to accept and adhere to insulin therapy from an earlier stage. In terms of weight gain, the insulin analog insulin detemir has been associated with slightly less weight gain than conventional insulins. In studies25,36,37 of patients with type 2 DM, weight gain was slightly but significantly less with detemir than with neutral protamine Hagedorn (NPH) insulin.
Observational studies of insulin analogs reveal an important aspect of treatment outcomes not fully illuminated by controlled clinical trials, namely, adherence to therapy. Adherence to DM therapy has been shown to be cost-saving: a 10% increase in adherence is estimated to be associated with up to a 29% decrease in annual healthcare costs, primarily because better adherence is associated with fewer hospital visits.38
Compact, prefilled, disposable pen devices are available for use with all insulin analogs. Human insulins are available in prefilled and refillable pens. These pen devices can help to overcome needle anxiety because they are easier to use, are more discreet than needles and syringes, are typically not painful, and are preferred by patients compared with standard needles.39 A recent observational study40 in the United States found that adherence improved and rates of hypoglycemia and total annual costs per patient decreased after patients switched from vial and syringe to an insulin analog pen device. Total annual cost-savings were $1590 per patient, of which $788 were hypoglycemia-related cost-savings and $600 were other DM-related cost-savings (P <.01 for all compared with the use of vials and syringes before the switchover).
Insulin analogs have also been shown to improve patients’ health-related quality of life (QOL) relative to human insulins.41,42 Because QOL is an important determinant of adherence, 43 this is another factor that may result in improved adherence to analog use and may lower expenses related to poor compliance with or delayed adoption of insulin therapy. Finally, the rapid-acting insulin analogs are suitable for pump therapy, which allows patients to achieve similar or better glycemic control relative to human insulins.44,45
Insulin analogs are more expensive to the payer than human insulins. Although the cost of medication may be higher, there may be reductions in more expensive long-term expenditures such as the costs related to treatment of hypoglycemia or chronic complications of DM.
Selection and Types of Studies
Unlike clinical end points, cost-effectiveness cannot readily be measured using clinical trials because the required time span is too long and there are too many confounding variables that cannot be controlled. Therefore, pharmacoeconomic models are frequently used to compare the effect of 2 or more strategies (eg, drug A vs drug B or drug A vs no intervention) on long-term clinical and cost outcomes. The models may assess direct costs, indirect costs, or both and may calculate costs from different perspectives such as that of the payer (patient, managed care organization, healthcare authority, etc) or of society. The results of models may show that an intervention (drug A) is cost-saving when its use actually results in lower costs than the use of an alternative (drug B or no intervention). Alternatively, the results may suggest that drug A is cost-effective versus drug B or no intervention when its use results in higher costs but in an improved outcome for patients. In the latter case, the number of life-years gained by an intervention is adjusted to allow for QOL and is expressed as quality-adjusted lifeyears (QALYs). Cost-effectiveness is then expressed as an incremental cost-effectiveness ratio (ICER) (ie, the cost per QALY gained with drug A).
ICERs are relative values and can only be quoted for the intervention of interest (drug A) and not for the reference standard (drug B or no intervention) against which it is being measured. The lower the cost per QALY, the more cost-effective the treatment. In the United States, an ICER of $50,000 to $100,000 per QALY is typically deemed as acceptable, although these limits are not fixed and can vary according to many factors such as disease burden and societal expectations. For example, payers may decide that it is acceptable to provide a drug that costs more than $100,000 per QALY if it is effective in a rare disease that affects young patients. For an intervention that will be widely required by older patients, a more stringent ICER cutoff such as $30,000 per QALY may be applied.
Pharmacoeconomic models have several disadvantages. There may be a lack of good-quality clinical data, requiring the use of results from small trials or expert opinion. The observational data used may be biased. Assumptions have to be made, and these may be (intentionally or not) too conservative or too optimistic. Finally, it may be difficult to extrapolate data from one patient population to another. A good model must be transparent (ie, users must have access to how it works and the data sources used, and any assumptions made must be clearly explained). It should also be externally validated (ie, shown to predict accurate results in scenarios in which the real outcomes are known). The uncertainty arising from the need to make assumptions must be explored through sensitivity analyses in which the effect of varying the assumptions is explored.
Many of the modeling studies that were identified for the present analysis examined the cost-effectiveness of different strategies in type 1 and type 2 DM using the CORE Diabetes Model,55 an interactive Internet-based computer model that allows the calculation of long-term outcomes in different patient populations in realistic clinical settings. Users of the model are able to enter data about the baseline characteristics, history, and future management of the population under study. Development of complications, life expectancy, QALYs, and total costs are then calculated by the model. The key treatment effects included in the model are clinically relevant results seen in trials, primarily reductions in A1C levels and hypoglycemia rates. The CORE Diabetes Model was validated by comparing results from model simulations with observed outcomes from published epidemiological and clinical studies in type 1 and type 2 DM.56
It is important to note that model results cannot always be directly extended to other countries because of differences in healthcare systems and costs. Despite these limitations, strong similarities in trends are often mirrored across different countries, as would be expected if one intervention provides real clinical advantages versus another. The results of the pharmacoeconomic modeling studies analyzed in the present review follow.
Analogs Versus Human Insulins. Three pharmacoeconomic modeling studies (1 study46 of type 1 DM and 2 studies 51,52 of type 2 DM) were identified that compared analogs with human insulins in a US context (Table 3). Representative studies from other countries are also given in Table 3. The data reported for US studies and for non-US studies demonstrate that analog treatment regimens were associated with ICER values that indicate cost-effectiveness.
In type 1 DM, detemir was shown to be cost-effective versus NPH insulin and glargine when considered during a 35-year horizon in the United States.46 The cost per QALY gained with detemir relative to NPH insulin was $14,974 when direct costs were considered.
In type 2 DM, it was calculated that detemir would reduce costs by $2020 versus NPH insulin during 10 years in a US context, largely through anticipated reductions in complications, notably nephropathy and retinopathy.51 Projected during a 35-year period, an ICER of $6269 per QALY was calculated for patients switching to detemir (with or without OADs) from NPH insulin (with or without OADs).52 In these studies, the treatment effects of detemir were based on results from the German cohort of the Predictable Results and Experience in Diabetes Through Intensification and Control to Target: An International Variability Evaluation (PREDICTIVE) study,57 a large real-life observational trial that included more than 1800 patients with type 2 DM.
Six pharmacoeconomic modeling studies47-50,53,54 of insulin analogs in type 1 or type 2 DM conducted in Canada, the United Kingdom, or 7 European countries show ICER values associated with cost-effectiveness at levels comparable to those measured in the US studies (Table 3). Because healthcare delivery in those countries is not similar to the US system, the collective pharmacoeconomic modeling data strengthen the point that insulin analogs are costeffective.
Analogs Versus OADs. The cost-effectiveness of insulin analogs used to replace or enhance OAD regimens in type 2 DM has been examined using pharmacoeconomic modeling in a US context ().51,52,58-60 An ICER of $657 per QALY was calculated for patients switching to detemir from OADs during a 10-year period.51 During a 35-year period, a cost of $7412 per QALY was calculated for patients switching to detemir (with or without OADs) from OADs alone. Similar to the comparisons versus NPH insulin, these models used results from the German arm of the PREDICTIVE observational study,57 and the low ICERs resulted largely from reductions in DM complications.
Two studies59,60 considered the use of biphasic insulin aspart 70/30. Regimens of biphasic insulin aspart 70/30 plus an OAD were compared with OAD regimens over a lifetime perspective. In the first study,59 total lifetime direct costs were higher with biphasic insulin aspart 70/30 plus metformin than with an optimized OAD regimen. However, the use of biphasic insulin aspart 70/30 was projected to result in reduced complications, particularly nephropathy (one of the costliest DM-related complications); this resulted in an ICER of $8487 per QALY gained for biphasic insulin aspart 70/30 plus metformin versus the optimized OAD regimen.
When a regimen of biphasic insulin aspart 70/30 plus a thiazolidinedione was compared with a sulfonylurea plus a thiazolidinedione, the cost per QALY was $25,400.60 In another study,58 the addition of biphasic insulin aspart 70/30 to metformin plus pioglitazone hydrochloride was cost-effective versus metformin plus pioglitazone hydrochloride during a 35-year period, with a projected cost per QALY of $22,209.
In all of these studies versus OADs, the use of an insulin analog resulted in increased prescription costs, but these were offset by reductions in complications. This finding underlines the importance of achieving effective glycemic control as early as possible.
Retrospective Analyses of Databases
Two studies69,70 compared an analog with OADs in type 2 DM using a different approach. Direct drug costs were calculated based on the mean daily doses of the different drugs at the end of a clinical trial.
The first of these was a 24-week study 69 evaluating the efficacy and safety of add-on glargine versus rosiglitazone maleate in insulin-naive patients with type 2 DM inadequately controlled on therapy with a sulfonylurea plus metformin. The improvements in A1C level from baseline were similar in both groups, but there was less weight gain with glargine (P = .02). The mean cost of all antihyperglycemic medications and resources during the 24 weeks was $1368 with glargine and $1603 with rosiglitazone; therefore, the mean cost of glycemic control was $235 lower with glargine.
In the second study,70 costs were estimated based on drug dosages at the end of a 16-week clinical trial of patients receiving biphasic insulin aspart 70/30 plus metformin versus patients receiving 1 or more OADs titrated by their clinicians to optimize glycemic control.71 Direct pharmacy costs were higher for the biphasic insulin aspart 70/30 group. With the biphasic insulin aspart 70/30 regimen, significantly more patients achieved an A1C level of 7.0% or less (P <.02). These patients were considered successfully treated; the mean annual cost for each successfully treated patient was $896 lower with biphasic insulin aspart 70/30.70
2. Caro JJ, Ward AJ, Oâ€™Brien JA. Lifetime costs of complications resulting from type 2 diabetes in the US. Diabetes Care. 2002;25(3):476-481.
4. Writing Team for the DCCT/EDIC Research Group. Sustained effect of intensive treatment of type 1 diabetes mellitus on development and progression of diabetic nephropathy: the Epidemiology of Diabetes Interventions and Complications (EDIC) study. JAMA. 2003;290(16):2159-2167.
6. Stratton IM, Adler AI, Neil HA, et al. Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ. 2000;321(7258):405-412.
8. AACE Diabetes Mellitus Clinical Practice Guidelines Task Force. American Association of Clinical Endocrinologists medical guidelines for clinical practice for the management of diabetes mellitus. Endocr Pract. 2007;13(suppl 1):1-68.
10. Greisinger AJ, Balkrishnan R, Shenolikar RA, Wehmanen OA, Muhammad S, Champion PK. Diabetes care management participation in a primary care setting and subsequent hospitalization risk. Dis Manag. 2004;7(4):325-332.
12. Diabetes Control and Complications Trial Research Group. Lifetime benefits and costs of intensive therapy as practiced in the Diabetes Control and Complications Trial [published correction appears in JAMA. 1997;278(1):25]. JAMA. 1996;276(17):1409-1415.
14. Mollema ED, Snoek FJ, Pouwer F, Heine RJ, van der Ploeg HM. Diabetes Fear of Injecting and Self-Testing Questionnaire: a psychometric evaluation. Diabetes Care. 2000;23(6):765-769.
16. Gough SC. A review of human and analogue insulin trials. Diabetes Res Clin Pract. 2007;77(1):1-15.
18. Home PD, Lindholm A, Riis A; European Insulin Aspart Study Group. Insulin aspart vs. human insulin in the management of longterm blood glucose control in type 1 diabetes mellitus: a randomized controlled trial. Diabetic Med. 2000;17(11):762-770.
20. Ratner RE, Hirsch IB, Neifing JL, Garg SK, Mecca TE, Wilson CA; U.S. Study Group of Insulin Glargine in Type 1 Diabetes. Less hypoglycaemia with insulin glargine in intensive insulin therapy for type 1 diabetes. Diabetes Care. 2000;23(5):639-643.
22. Hermansen K, Fontaine P, Kukolja K, Peterkova V, Leth G, Gall MA. Insulin analogues (insulin detemir and insulin aspart) versus traditional human insulins (NPH insulin and regular human insulin) in basal-bolus therapy for patients with type 1 diabetes. Diabetologia. 2004;47(4):622-629.
24. Riddle M, Rosenstock J, Gerich J; Insulin Glargine 4002 Study Investigators. The treat-to-target trial: randomized addition of glargine or human NPH insulin to oral therapy of type 2 diabetic patients. Diabetes Care. 2003;26(11):3080-3086.
26. Anderson JH Jr, Brunelle RL, Keohane P, et al; Multicenter Insulin Lispro Study Group. Mealtime treatment with insulin analog improves postprandial hyperglycemia and hypoglycemia in patients with nonâ€“insulin-dependent diabetes mellitus. Arch Intern Med. 1997;157(11):1249-1255.
28. Boehm BO, Home PD, Behrend C, Kamp NM, Lindholm A. Premixed insulin aspart 30 vs. premixed human insulin 30/70 twice daily: a randomized trial in type 1 and type 2 diabetic patients [published correction appears in Diabet Med. 2002;19(9):797]. Diabet Med. 2002;19(5):393-399.
30. Korytkowski M. When oral agents fail: practical barriers to starting insulin. Int J Obes Relat Metab Disord. 2002;26(suppl 3):S18-S24.
33. Polonsky WH, Fisher L, Guzman S, Villa-Caballero L, Edelman SV. Psychological insulin resistance in patients with type 2 diabetes: the scope of the problem. Diabetes Care. 2005;28(10):2543-2545.
35. 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;(2):CD005613.
37. Hermansen K, Davies M. Does insulin detemir have a role in reducing risk of insulin-associated weight gain? Diabetes Obes Metab. 2007;9(3):209-217.
39. Korytkowski M, Bell D, Jacobsen C, Suwannasari R; FlexPen Study Team. A multicenter, randomized, open-label, comparative, twoperiod crossover trial of preference, efficacy, and safety profiles of a prefilled, disposable pen and conventional vial/syringe for insulin injection in patients with type 1 or 2 diabetes mellitus. Clin Ther. 2003;25(11):2836-2848.
41. Bott U, Ebrahim S, Hirschberger S, Skovlund SE. Effect of the insulin analogue insulin aspart on quality-of-life and treatment satisfaction in type 1 diabetic patients. Diabet Med. 2003;20(8):626-634.
43. Hanestad BR, Albrektsen G. Quality of life, perceived difficulties in adherence to a diabetes regimen, and blood glucose control. Diabet Med. 1991;8(8):759-764.
45. Raskin P, Holcombe JH, Tamborlane WV, et al. A comparison of insulin lispro and buffered regular human insulin administered via continuous subcutaneous insulin infusion pump. J Diabetes Complications. 2001;15(6):295-300.
47. Grima DT, Thompson MF, Sauriol L. Modelling cost effectiveness of insulin glargine for the treatment of type 1 and 2 diabetes in Canada. Pharmacoeconomics. 2007;25(3):253-266.
49. Palmer AJ, Roze S, Valentine WJ, Smith I, Wittrup-Jensen KU. Costeffectiveness of detemir-based basal/bolus therapy versus NPH-based basal/bolus therapy for type 1 diabetes in a UK setting: an economic analysis based on meta-analysis results of four clinical trials. Curr Med Res Opin. 2004;20(11):1729-1746.
51. Palmer AJ, Cobden D, Koenen C, et al. Projecting the economic and health outcome effects of basal insulin among type 2 patients in clinical practice settings [abstract]. Diabet Med. 2006;23(suppl 4):368.
53. McEwan P, Poole CD, Tetlow T, Holmes P, Currie CJ. Evaluation of the cost-effectiveness of insulin glargine versus NPH insulin for the treatment of type 2 diabetes in the UK. Curr Med Res Opin. 2007;23(suppl 1):21-31.
55. Palmer AJ, Roze S, Valentine WJ, et al. The CORE Diabetes Model: projecting long-term clinical outcomes, costs and cost-effectiveness of interventions in diabetes mellitus (types 1 and 2) to support clinical and reimbursement decision-making. Curr Med Res Opin. 2004;20(suppl 1):S5-S26.
57. Meneghini CF, Rosenberg KH, Koenen C, Merilainen MJ, Lilddeke HJ. Insulin detemir improves glycaemic control with less hypoglycaemia and no weight gain in patients with type 2 diabetes who were insulin naive or treated with NPH or insulin glargine: clinical practice experience from a German subgroup of the PREDICTIVE study. Diabetes Obes Metab. 2007;9(3):418-427.
59. Minshall M, Ray J, Lammert M, et al. Estimating the long-term cost-effectiveness of biphasic insulin aspart plus metformin versus optimization of oral hypoglycemic agents in insulin-naive patients with type 2 diabetes in a US cost setting [poster]. Diabetologia. 2005;48(suppl 1):A335.
61. Bullano MF, Al-Zakwani IS, Fisher MD, Menditto L, Willey VJ. Differences in hypoglycemia event rates and associated cost-consequence in patients initiated on long-acting and intermediate-acting insulin products. Curr Med Res Opin. 2005;21(2):291-298.
63. Zhang Q, Menditto L. Incremental cost savings 6 months following initiation of insulin glargine in a Medicaid fee-for-service sample. Am J Ther. 2005;12(4):337-343.
65. Miller DM, Gardner J, Hendricks A, Fincke G, Zhang Q. Changes in health care utilization and expenditures associated with insulin glargine use [abstract]. Diabetes. 2005;54(suppl 1):A302.
67. Hall JA, Summers KH, Obenchain RL. Cost and utilization comparisons among propensity scoreâ€“matched insulin lispro and regular insulin users. J Manag Care Pharm. 2003;9(3):263-268.
69. Rosenstock J, Sugimoto D, Strange P, Stewart JA, Soltes-Rak E, Dailey G. Triple therapy in type 2 diabetes: insulin glargine or rosiglitazone added to combination therapy of sulfonylurea plus metformin in insulin-naive patients. Diabetes Care. 2006;29(3):554-559.
71. Ushakova O, Sokolovskaya V, Morozova A, et al. Biphasic insulin aspart 30 (NovoLog Mix 70/30), a premix analogue, is an effective and well tolerated starter insulin in type 2 diabetes [abstract]. Diabetes. 2005;54(suppl 1):A502.