Insulin Delivery in Type 2 Diabetes: To Pump or Not to Pump?

Written By

Brooklyn Nelson, PharmD
Rick Hess, PharmD, CDE, BC-ADM

Reviewed By

Amanda Schartel, Pharm.D., BCACP
Ashley Johnson, Pharm.D., BCPS


Reznik Y, Cohen O, Aronson R, et al. Insulin pump treatment compared with multiple daily injections for treatment of type 2 diabetes (OpT2mise): a randomized open-label controlled trial. Lancet 2014; 384: 1265-72.

The role of continuous subcutaneous insulin infusion (CSII) in Type 2 Diabetes Mellitus (T2DM) is not clearly defined.1 In patients with type 1 diabetes mellitus, CSII improves glycemic variability, lowers the risk of hypoglycemia, reduces hemoglobin A1C (A1C), and increases treatment satisfaction as well as quality of life when compared with multiple daily injections (MDI).2,3 However, studies in patients with T2DM have produced conflicting findings.4,5,6,7 See Table 1. The American Diabetes Association and European Association for the Study of Diabetes (ADA/EASD) guidelines acknowledge that “CSII is a less commonly used and more costly alternative” to basal-bolus therapy in patients with T2DM.A large, robust study is needed to clarify the role of CSII in patients with T2DM.


Table 1. Comparison of Efficacy and Safety of CSII to MDI in T2DM




# of Patients

Type of insulin

Baseline HbA1C

Change in HbA1C

Safety Parameters



6 months


CSII: (aspart)

MDI: (aspart + NPH)

CSII: 8.2%
MDI: 8.0%

CSII:  -0.6±1.22%

MDI: -0.5±1.08%*


No difference in weight or hypoglycemia



12 months


CSII: (lispro)

MDI: (lispro + glargine)

CSII: 8.4%
MDI: 8.1%

CSII: -1.7±1.0%

MDI: -1.6±1.2% (P=0.20)

No difference in weight or hypoglycemia


Randomized crossover

3 months


CSII: (lispro)

MDI: (lispro + NPH)

CSII: 8.6%
MDI: 9.3%

CSII: -1.3±0.8%

MDI: -0.4±1.6% (p<0.03)

No difference in hypoglycemia


Randomized crossover

4.5 months


CSII: (lispro)

MDI: (regular +NPH)

CSII: 10.2%
MDI 10.3%



MDI: +0.4±1.3%  (p<0.007) 

No difference in weight or hypoglycemia


*No p-value reported

CSII: continuous subcutaneous insulin infusion; MDI: multiple daily injections; T2DM: Type 2 Diabetes Mellitus



The Opt2mise trial was an international, randomized controlled trial that included 331 patients with T2DM who were poorly controlled (HbA1C ≥ 8% and ≤12%) on MDI regimens.8 Patients were 30-75 years of age, not pregnant, and had insulin doses of 0.7-1.8 units/kg or total daily dose ≤ 220 units at baseline. Patients were excluded if they had significant diabetes-related complications, ≥ 2 hypoglycemia-related seizures or comas in the previous 6 months, or if they self-measured blood glucose (SMBG) < 2.5 times daily. Prior to randomization, patients underwent a two-month run-in phase so that insulin therapy with MDI using long-acting (glargine or detemir) and rapid-acting analogues (lispro, aspart, or glulisine) could be optimized. TDD was increased 10-40% from the baseline regimen to reach fasting glucose values 70-130mg/dL and post-prandial values <180 mg/dL. Patients discontinued all oral antidiabetic agents except metformin. Of note, 33% of patients who entered the run-in phase were excluded before randomization for various reasons most often inadequate SMBG infrequency.


After the run-in phase, patients were randomly assigned to continue their current MDI regimen (n=163) or begin CSII with the Medtronic MiniMed Paradigm Veo system (n=168). Patients were trained to use CSII and training could continue for up to 3 weeks during the randomization phase. The primary endpoint was change in mean A1C between baseline and 6 months. Safety endpoints included number of severe hypoglycemic and ketoacidosis events.  No qualitative outcomes were measured.


Mean A1C at baseline was 9% in both groups. At 6 months, mean A1C had decreased by 1.1% with CSII and 0.4% with MDI, resulting in a treatment difference of -0.7% (95% confidence interval: -0.9 to -0.4, p<0.0001).  The largest reduction in A1C was observed in patients with higher baseline values. The A1C reduction was independent of diabetes duration, body-mass index (BMI), education level, Montreal Cognitive Assessment Score, and SMBG frequency. At the study conclusion, mean TDD was 97 units with CSII vs. 122 units for MDI (p<0.001).  There was no significant difference in weight gain (1.5 kg vs. 1.1 kg, p=0.322).


Five patients developed hyperglycemia not requiring hospitalization in the CSII treatment group. Hyperglycemia resulting in hospitalization occurred infrequently, but more often in the CSII treatment arm (2 vs. 1). There was only one episode of severe hypoglycemia, which occurred with MDI. Data for self-reported mild hypoglycemia and hyperglycemia were not reported; however, 6-day masked continuous glucose monitoring system (CGMS) data at baseline and 6 months suggested no significant difference existed in non-severe hypoglycemia between MDI and CSII. Other adverse events occurring more frequently with CSII included cellulitis (7 vs. 0), infection (7 vs. 0), and site irritation/allergy (6 vs. 1).


Treatment difference in A1C between CSII and MDI therapy in the Opt2mise trial (-0.7%) was smaller than previous studies with cross-over designs (-0.9% and -1.2%) but larger than parallel-group studies (-0.1%). Consistent with previous studies, there were no significant differences in weight gain or hypoglycemia.4,5,6,7 Given that 38% of patients receiving CSII in the Opt2mise has mild cognitive impairment as measured by the Montreal Cognitive Assessment Score, CSII may be a viable treatment option for some patients not previously considered good candidates. Furthermore, while carbohydrate counting is generally recommended for patients using CSII, it was not required in this study. The majority (59%) of patients used the bolus calculator <25% of the time, indicating CSII may be beneficial for patients with T2DM, irrespective of their ability to carb count.


Several limitations to this study exist. Enrollment required nearly three years to identify 331 patients. This suggests that choosing suitable candidates for CSII remains crucial. The two-month run-in period and three-week CSII training portion may have introduced treatment bias and limits external validity.  Adherence was not measured in the MDI treatment arm, which may have impacted the primary outcome when compared to CSII. Furthermore, patients receiving CSII performed SMBG more often than patients receiving MDI, particularly in the final 3 months of the study (3.8 vs. 3.1 tests/day, respectively). Data related to quality of life outcomes and bolus dosing were not reported.  Lastly, the long-term benefits (and risks) of CSII glycemic management remain unknown. The Opt2mise study was only 6 months long.


There are several potential barriers to using CSII in practice. Firstly, patients must commit to frequent SMBG and working closely with the healthcare team for pump training and ongoing management.  Affordability is another disadvantage.  Pumps typically cost $5,000-$7,000 and pump supplies another $2,500 per year or more.1 While any US clinician can legally prescribe CSII, only a small number of providers have been adequately trained to initiate and manage insulin pumps. Too often patients must rely on the product manufacturer to initiate and manage this challenging technology. Unfortunately, several inappropriate candidates without adequate training and guidance have started CSII, resulting in tragic outcomes.9 Pharmacists, particularly in ambulatory care settings, can help alleviate this gap by becoming certified pump trainers.


The Opt2mise study suggests that CSII is a feasible and effective option for patients with poorly controlled T2DM when intensification of MDI has failed — with three important caveats. First, ideal candidates are those who are motivated and willing to SMBG at least three times a day.  Second, they must be able to afford the out-of-pocket costs not covered by insurance. Third, they have access to a health professional with expertise in CSII training and management. If a patient meets these criteria, are you likely to suggest CSII? Does your practice have the appropriate resources to support a patient with CSII? If not, are you willing to step up and become a certified pump trainer?


Lucas G Hill's picture

Thank you for an excellent review and discussion of this potentially practice-changing trial. As you noted, the large portion of patients with mild cognitive impairment and lack of a carb-counting requirement were compelling regarding external validity. I'm not particularly concerned with the short duration of the study given this is not a new drug, but rather a well-established delivery system not commonly used in the study population. I recently had the opportunity to care for children with DM1 and discover how easy pumps are to manipulate and maintain. However, the process for pump and supply procurement is still a mystery to me. I would like to see additional pragmatic studies detailing the implementation of this approach in primary care (including cost considerations) before investing my time in establishing such a practice.

I have a lot of patients with type 2 diabetes that would be great pump candidates; however, we often struggle with getting insurance companies to approve them. Hopefully, this will add to the evidence that insulin pumps aren't just for type 1 diabetes patients.