Insulin Pumps 

Updated: Nov 10, 2014
  • Author: Se Min Kim, MD; Chief Editor: George T Griffing, MD  more...
  • Print


Since the first human trials of insulin pumps in the late 1970s, [1] insulin pump therapy, also known as continuous subcutaneous insulin infusion (CSII), has been used for more than 30 years. As technology has developed, insulin pumps have become more user-friendly and smaller. Modern pumps weigh less than 4 ounces and are the size of a pager or cell phone.

The number of patients using CSII has been increasing. The US Food and Drug Administration (FDA) reported that approximately 375,000 adults with type 1 diabetes used external insulin infusion pumps in 2007, compared to 130,000 in 2002. [2]

The FDA classifies insulin infusion pumps as class II devices, which need more than “general controls” (ie, provisions that relate to controlling the quality of the device, preventing issues such as adulteration or misbranding, and tracking components and devices) to ensure reasonable safety and effectiveness. [2] As with any class II device, the “special controls” required may include additional labeling requirements, mandatory performance standards, and postmarket surveillance. [2]

The chief benefit of insulin pump therapy is customized flexible basal and bolus dosing to meet patients’ individual requirements. Insulin pumps allow users to program different basal rates to allow for variations in lifestyle and bolus doses to allow for variations in diet. Insulin delivery via a pump is more consistent and precise than via syringe or injection pen. [3]

There are newer electronics with complex algorithms capable of calculating insulin bolus doses necessary to maintain glucose level within a set range as measured by a continuous glucose monitor (CGM). Ultimately, this technology will likely lead to an “artificial pancreas” that automatically senses glucose level and its rate of change and adjusts insulin dosing accordingly. However, this technology is still under development and is not ready for clinical use.


Insulin pump

Device details

Accu-check Spirit (Roche Insulin Delivery system)

Dana Diabecare IIS (Sooil Development)

Minimed Paradigm Real-Time Revel (Medtronic Diabetes)

Omnipod (Insulet Corp)

OneTouch Ping (Animas Corp)


Design Features

Most insulin pumps on the market consist of 3 parts: the pump itself (which is a programmable electronic device that includes a user interface, electronic processor, worm screw to control dosing, and batteries), a disposable reservoir that stores the insulin, and a disposable infusion set (which is the tubing that connects to the reservoir and which terminates in a cannula or needle through which the insulin is infused). The needle or cannula is inserted under the skin, and the insulin pump itself is worn externally. Newer pumps have convenient features to track trends of glucose levels; some also have lighted screens or audio cues for patients with visual impairment.

Other insulin pump features

The latest models of insulin pumps include new features to help patients manage insulin delivery.

Bolus calculator

The physician prescribes bolus insulin based on the carbohydrate content of a meal and for the correction of any elevation of blood sugar levels. An insulin-to-carbohydrate ratio is usually programmed into the pump (eg, 1 unit of insulin per 15 g of carbohydrate). However, the patient must estimate the carbohydrate content of the meal.

Similarly, correction insulin can be programmed into the pump (eg, 1 unit of insulin for every 50 mg/dL glucose >100 mg/dL). The patient tests his or her blood sugar level, and the glucose level is either entered manually or automatically communicated to the pump, depending on the pump model. The calculator can also adjust the bolus amount based the amount of insulin still in the patient’s system from a previous bolus.


Alarms can be set to remind the user if a bolus was missed (eg, if no bolus is given at the usual lunch time, the pump will alert the user), if a timed glucose test was missed, or if various other user-defined parameters are met or unmet.

Associated glucose meters

Some insulin pumps facilitate direct communication between the patient’s glucose meter and insulin pump.

The Medtronic Minimed Paradigm series of pumps can receive data transmitted wirelessly from a glucometer (Lifescan in the United States; Bayer in the rest of the world), which then allows the pump to use glucose measurements to calculate boluses.

The Accu-Chek Combo Insulin Pump System allows the glucometer to be used to control the pump itself, as well as to communicate glucose readings to the pump. The Insulet OmniPod remote has an integrated glucometer, which allows the remote to calculate boluses based on glucose readings and then transmit them to the pod connected to the patient’s body.

The DANA Diabecare IISG pump works similarly but has a glucometer integrated into the pump itself.

Continuous glucose monitors

The Medtronic Minimed Paradigm series of pumps can interact with the Guardian Real-Time CGM, which is a device that takes real-time measurements of the glucose concentration in the subcutaneous fluid; this parallels the serum glucose levels. The CGM can then wirelessly communicate measurements to its associated insulin pump, which then displays the blood glucose and its rate of change. Alarms can also be set to warn the user when glucose levels are too high or too low or are changing too quickly. Although other CGM systems exist, the Guardian Real-Time is the only CGM that communicates with an insulin pump.

CGM technology has been gaining prominence in optimized CSII, although frequent self-monitored blood glucose (SMBG) using a glucometer remains the foundation of optimized insulin therapy.

CGM has a potential role to revolutionize insulin delivery via a closed-loop insulin delivery system, the so-called “artificial pancreas.” Closed-loop systems were shown to reduce the risk of nocturnal hypoglycemia in children and adolescents with type 1 diabetes and improved overall glycemic control.

Three randomized studies of overnight closed-loop systems were performed in young patients with type 1 diabetes. An analysis of pooled data showed that CGM increased patients’ duration of normal glucose levels and decreased hypoglycemic episodes. [29] A study in adult patients demonstrated similar findings. The same study group performed a randomized crossover study with eating-in and eating-out scenarios in adult type 1 DM. The amount of time patients spent in the target glucose range increased to 77%, compared to 51% in the conventional CSII group. The amount of time patients spent in a hypoglycemia state overnight was also significantly reduced. [30]

However, closed-loop control during waking hours is complicated by various factors, such as diet, lifestyle, and the intrinsic limitation of current technologies. Safety during the postprandial period is an issue because of the delays in rapid-acting insulin analogue absorption. A study comparing fully closed-loop control to hybrid closed-loop control that used small manual premeal “priming” boluses in addition to dosing via the closed-loop system found that the hybrid closed-loop system improved daytime and postprandial glycemic control in patients with type 1 diabetes. [31]

Overall, closed-loop insulin delivery systems are currently limited and are still under development. CGM devices should be used to complement SMBG until the FDA approves them as a replacement for SMBG.

Communication with personal computers

Many insulin pumps have mechanisms to transmit and receive data between the pump and a personal computer. The computer interface can be used to manage pump settings and/or to document and analyze data from the pumps and associated glucose meters. They can also assist physicians in monitoring patients’ glycemic control and behavior related to managing diabetes.



Despite the increasing popularity of insulin pump therapy, there is no official guideline for the initiation of insulin pump use. The American Diabetes Association (ADA) published a position statement in 2004, [5] and the American Association of Diabetes Educators (AADE) published its Guideline for Successful Outcomes in 2009. [3] The American Academy of Pediatrics published its position statement in 2006. [6] In 2010, the American Association of Clinical Endocrinologists (AACE) released a statement regarding CSII. [7]

There is a consensus that appropriate patient selection is critical and is a key to realizing the potential advantages of CSII, since insulin pump therapy requires more training than other forms of insulin delivery. Patients must be motivated and willing to work with providers to succeed in using this complex therapy.

The AADE recommends considering an insulin pump in the following patients: [3]

  • Patients whose hemoglobin A1c (HbA1c) level is greater than 7%, accompanied by frequent severe hypoglycemia (< 55 mg/dL)
  • Patients who have hypoglycemic events that require third-party assistance or that interfere with work, school, or family obligations
  • Patients with frequent and unpredictable fluctuations in blood glucose levels
  • Patients who perceive that diabetes management impedes the pursuit of personal or professional goals

Prospective pump users or caregivers must be able to handle infusion pumps and must be prepared to troubleshoot problems. [3] In addition, they should be competent in assessing the nutritional value of meals and monitoring blood glucose levels frequently (preferably 6-8 times but at least 4 times per day). [3]

The psychological aspect of insulin pump use is also important. One study showed that patients who feel more responsible for the process of CSII achieve better glycemic control than patients who feel less responsible. [8] The patient’s emotional maturity and the stability of his or her life situation need to be considered before starting insulin pump therapy. [7]

Clinically suitable insulin pump candidates

The 2010 AACE statement proposed the following patients as clinically suitable insulin pump candidates. [7]

Class 1

Patients are classified as class 1 if they have type 1 diabetes mellitus (DM) and do not reach glycemic goals despite adherence to a maximum multiple daily injection (MDI) and are on a non-CSII program (≥4 insulin injections and ≥4 self-monitored blood glucose measurements daily), especially if they have the following:

  • Very labile DM (erratic and wide glycemic excursions, including recurrent diabetic ketoacidosis [DKA])
  • Frequent severe hypoglycemia and/or hypoglycemia unawareness
  • Significant “dawn phenomenon” (increase in blood glucose levels, usually from 2 am to 8 am, resulting from increased secretion of counter-regulatory hormones, particularly growth hormone)
  • Extreme insulin sensitivity
  • Special circumstance (eg, preconception, pregnancy, children, adolescents with eating problems, competitive athletes)

Class 2

Patients are classified as class 2 if they have type 1 DM and are on a maximized basal-bolus MDI insulin regimen, defined as more than 3 daily injections, regardless of their level of glycemic control and who, after investigation and careful consideration, feel that CSII would be helpful or more suitable for lifestyle reasons.

Class 3

Patients are classified as class 3 if they have insulin-requiring type 2 DM and satisfy any or all of the following:

  • Positive C-peptide results but with suboptimal control on a maximal program of basal/bolus injections
  • Substantial “dawn phenomenon”
  • Erratic lifestyle (eg, frequent long-distance travel, shift work, unpredictable schedules that disrupt maintaining timing of meals)
  • Severe insulin resistance, candidate for U500 insulin via CSII (eg, type A and type B insulin resistance syndrome, congenital and acquired generalized lipodystrophy, hyperandrogenism–insulin resistance–acanthosis nigricans [HAIR-AN], Rabson-Mendenhall syndrome)
  • Selected patients with other DM types (eg, DM due to pancreatectomy)


The following patient factors contraindicate insulin pumps:

  • Unable or unwilling to perform multiple daily insulin injections (≥3 daily), frequent blood glucose monitoring (≥4 daily), or carbohydrate counting
  • Lack of motivation to achieve tighter glucose control and/or history of nonadherence to insulin injection protocols
  • History of serious psychological or psychiatric condition(s) (eg, psychosis, severe anxiety, depression)
  • Reservations about pump usage interfering with lifestyle (eg, contact sports or sexual activity)
  • Unrealistic expectations of pump therapy (eg, belief that it eliminates the need to be responsible for diabetes management)

Clinical Trial Evidence

In 1993, the Diabetes Control and Complications Trial (DCCT) demonstrated the beneficial effects of intensive glycemic control in type 1 diabetes with an insulin pump or multidose insulin in slowing the progression of microvascular complications, such as retinopathy, nephropathy, and neuropathy. [9]

Since then, numerous randomized and nonrandomized studies have supported the efficacy and safety of CSII.

The large “5 Nations Trials,” including 272 patients in 11 European centers, also showed a statistically significant difference in HbA1 reduction (-0.23%, P < 0.001) in the CSII group compared to MDI with neutral protamine hagedorn (NPH) as a bolus insulin. CSII treatment resulted in lower HbA1c levels (7.45% vs 7.67%, P < 0.001) and less fluctuation in blood glucose levels than MDI. [10]

Other randomized trials by Hanaire-Broutin et al, DeVries et al, and Weintrob et al also reported that CSII yielded better glycemic control than MDI. [11, 12, 13]

A meta-analysis from 2008 assessed 12 randomized trials with adult type 1 diabetes and showed a statistically significant difference of HbA1c reduction (-0.6%; 95% CI, -0.87 to -0.22) in favor of CSII over MDI. [14] However, all of the studies included in this meta-analysis used NPH rather than long-acting insulin analogues in the MDI group.

In 2010, a Cochrane review [15] compared CSII with MDI in type 1 DM. This review included 23 randomized studies involving 976 participants. Of 23 studies, 5 trials compared CSII to MDI with a long-acting insulin analogue (glargine) and short-acting insulin analogue (lispro or aspart). There was a statistically significant difference in HbA1c favoring CSII (weighted difference, -0.3%; 95% CI, -0.1 to -0.4). In addition, reduced severe hypoglycemia was reported in those using CSII, and CSII users demonstrated greater improvement in quality-of-life measures.

MDI with a long-acting insulin analogue was also compared with CSII. One randomized study compared a group using MDI with once-daily glargine and premeal/snack insulin aspart and another using CSII with insulin aspart. This study demonstrated that the CSII group achieved lower HbA1c level, from 8.1% to 7.2% after 16 weeks of therapy (P < 0.02 vs baseline and P < 0.05 vs glargine group). Premeal and bedtime glucose levels were significantly higher in the MDI group, although fasting glucose levels did not differ in the two groups. [16]

This short-term study showed that a lack of compliance may be the intrinsic obstacle that contributes to poorer glycemic control in the MDI group. This would be consistent with the findings from a long-term study by Schiaffini et al in which both MDI and CSII improved glycemic control, but the improved HbA1c levels lasted longer in the CSII group. [17]

However, other studies showed that MDI with glargine as the basal insulin was not inferior to CSII, even though the insulin requirement in the CSII group was significantly less than in the MDI group. [18, 19]

The psychosocial benefit of CSII has been also noted. Parent-reported or self-reported diabetic-specific quality of life increased significantly with CSII use. [20] One study also reported that the improvement in glycemic control with CSII was accompanied by improvements in mood, behavior, and cognition in children with type 1 DM. [21]

The use of CSII in type 2 DM is not well established. Two randomized trials showed similar efficacy of CSII to MDI in terms of HbA1c reduction. Those studies used NPH with aspart and glargine with lispro, respectively. [22, 23]

However, patients on CSII showed a lower incidence of hyperglycemic excursions and achieved more consistent glucose control. [22] A small randomized crossover study showed a reduced hyperglycemic period using continuous glucose monitoring. In this study, CSII with lispro was slightly more efficacious than MDI with NPH as the bolus insulin, and HbA1c levels decreased from 9±1.6% to 8.6±1.6% with multiple injections and to 7.7±0.8% when using an insulin pump (P < 0.03). [24]

In pregnant women with diabetes, CSII has been compared with MDI to evaluate decreases in neonatal complications associated with DM, especially macrosomia (neonates born at >90th percentile of weight for gestational age or sex or >2 standard deviations). A few studies have suggested that CSII is safe and effective during pregnancy with type 1 DM. [25, 26] However, a Cochrane review in 2007 reported that a significant increase in mean birth weight was associated with CSII, although the authors concluded that the result is not clinically significant since the difference in the rate of macrosomia was not significant. [27]

More robust evidence is needed to assess the advantages of CSII over MDI in pregnancy, but there are anecdotal observations from healthcare professionals that women who have been very troubled by hypoglycemia during pregnancy have improved by a switch to CSII. [28]


Clinical Implementation

Insulin pumps differ from manufacturer to manufacturer, and even models from the same manufacturer may have substantial differences. Thus, patients need to be trained on the use of their specific pump.

An important consideration in prescribing an insulin pump is the patient’s (or caregiver’s) willingness and ability to perform diabetes self-management, because, once patients are trained in how to use an insulin pump, daily management of the pump needs to be under the control of the patients themselves or their caregivers. The physician’s role is to provide guidance on pump settings and other aspects of diabetes care.

For the successful use of CSII, there must be regular communication between the patient and the healthcare team. Thorough evaluation and training is needed before initiating this therapy to ensure that the patient is a suitable candidate for insulin pump use and that he or she has the knowledge needed to manage the device safely and effectively.

As with any technology, insulin pumps have advantages and disadvantages. While insulin pumps are not appropriate for all patients, they offer tremendous benefits for the right candidates—improved glycemic control, reduction of hypoglycemic events, and increased flexibility in managing diabetes. This explains the increasing use of CSII technology.

Insulin for the pump

To use the insulin pump, the patient must first fill the reservoir with insulin, typically a rapid-acting insulin analog such as glulisine, lispro, or aspart. Some insulin pumps use a prefilled cartridge. For all pumps except the Insulet OmniPod, the reservoir is then attached to the infusion set. The infusion set is primed by the insulin pump to load insulin to the end of the cannula. This is done with the pump disconnected from the body to prevent unintended insulin delivery. The cannula surrounds a stainless steel needle.

The infusion set is inserted into the infusion site, typically a site on the patient’s abdomen, buttock, or thigh, and the needle is removed, leaving the cannula taped down in place at the infusion site. Again, with the exception of the Insulet OmniPod, the specifics of filling the reservoir, connecting the infusion set, and inserting the device at the infusion site vary among different manufacturer’s products, but the basic process remains similar.

The Insulet OmniPod consists of two components—a “pod,” which injects the insulin it contains into a subcutaneous infusion site, and a separate electronic device that wirelessly controls how the pod delivers insulin. Unlike other insulin pumps on the market, the OmniPod connects directly to the body instead of connecting via an infusion set. To prepare the OmniPod, the reservoir within the pod is filled via a syringe. The pod is then attached at the infusion site via adhesive. The patient then initiates pumping by pressing Start on the electronic controller, and the pod automatically inserts a cannula subcutaneously and begins pumping.

Insulin dosing

Conceptually, there are two different dosing regimens administered via insulin pump: basal and bolus insulin. Both use the same rapid-acting insulin contained in the pump’s reservoir.

Basal insulin is given at a specified (usually hourly) rate over time. As with long- and intermediate-acting insulins, basal insulins are intended to meet the insulin requirements that the body has for basal activities. The dose of insulin administered by most pumps can be calibrated and varied hour by hour. Thus, unlike long-acting insulins, in which the dose of insulin is fixed for the day, the basal rate of an insulin pump can be set to change throughout the day to account for variations in the patient’s metabolic needs.

In theory, a well-calibrated basal rate should maintain the patient’s glucose level with minimal variation as long as the patient does not eat or undergo activities that would change the serum glucose level. For example, early in the morning, many patients require more insulin because of the normal physiologic rise of cortisol and growth hormone (“the dawn phenomenon”). Settings on the pump can be set to account for this variability in basal insulin needs. Contemporary insulin pumps also allow users to set a temporary higher or lower basal rate for when their insulin needs differ from the norm (eg, during exercise or illness). Once the device is programmed, basal insulin runs automatically.

Bolus insulin is given to cover the food that a patient eats or to correct elevated blood glucose levels. A standard bolus is infused in less than a minute, similarly to a typical subcutaneous rapid-acting insulin injection. Unlike basal insulin, which is given automatically once dosing parameters are set, bolus insulin must be manually entered by the patient for each bolus given.

Many pumps also allow what is called an “extended” or “square wave” bolus, which pumps the dose over an extended period, with the time range typically lasting from a half hour to 3 hours, depending on how the patient sets the bolus. This is useful for high-fat, high-protein meals, which take longer to digest and which raise the blood glucose level for an extended period. It can also be useful in patients who digest their food more slowly, such as in those with gastroparesis. Combination standard/extended boluses are also possible to help manage high-carbohydrate, high-fat meals.



An important consideration in prescribing an insulin pump is the patient’s (or caregiver’s) willingness and ability to perform diabetes self-management, because, once patients are trained in how to use an insulin pump, daily management of the pump needs to be under the control of the patients themselves or their caregivers. The physician’s role is to provide guidance on pump settings and other aspects of diabetes care.

For the successful use of CSII, there must be regular communication between the patient and the healthcare team. Thorough evaluation and training is needed before initiating this therapy to ensure that the patient is a suitable candidate for insulin pump use and that he or she has the knowledge needed to manage the device safely and effectively.



Insulin pumps are generally used to prevent complications associated with insulin injections (ie, frequent hyperglycemia or hypoglycemia). However, patients who use insulin pumps can also still become hyperglycemic or hypoglycemic. In addition, insulin pumps present unique complications of their own.


Multiple studies have shown that insulin pumps reduce the frequency of severe hypoglycemic events relative to multiple daily insulin injections. This has been shown in randomized control trials [10, 13, 32, 33] and in before/after studies. [34, 35, 36] This reduction largely results from the patient no longer needing to use long- or, especially, intermediate-acting insulins, which require the patient to predict caloric intake, activity, and other issues that affect serum glucose levels ahead of time. Intermediate-acting insulin can especially be problematic overnight, when its activity peaks at a time when the patient is not eating.

However, despite the advantages of insulin pump use, hypoglycemia may still occur on an insulin pump if the patient is given too large a bolus or the basal rate is set too high. These problems can occur with human error in programming the pump or with device malfunction leading to excess insulin delivery. There has been concern expressed about unintentional and uncontrolled insulin delivery, or “pump runaway.” Although this has occurred, [37] improvements in insulin pump technology and alarms on modern pumps mean that it is an extremely unlikely event, and no cases of this have been reported in the United States in more than decade.

Hyperglycemia and diabetic ketoacidosis

Studies have also shown improved glycemic control in patients using insulin pumps compared with patients using multiple daily insulin injections. [16, 38] Another study showed that the incidence of DKA is equivalent for pump users and patients on MDIs. [39] In particular, nocturnal and predawn glycemic control is improved on insulin pump therapy, as hour-by-hour preprogrammed basal rate changes facilitated by the pump help counteract prebreakfast blood glucose increases (the “dawn phenomenon”) that are not easily addressed with injection therapy.

However, hyperglycemia and DKA can occur with insulin pump use, whether due to user error in programming or to device malfunction. Studies show that device problems leading to hyperglycemia include diminished insulin delivery due to a depleted or malpositioned battery, occlusion or crimping of the cannula, disruption of the infusion set (eg, via torn tubing), or complete pump failure. DKA in particular occurs more frequently early after starting insulin pump use, suggesting that acclimation to the device has a learning curve.

Finally, continuous use of the same infusion set for an extended period (ie, >2.5 days) increases the risk of cannula occlusion and changes the physiochemical environment at the delivery site, altering the rate of insulin absorption [40] and increasing the risk of hyperglycemia, [41] as well as other problems described below. One study has found that there was a slow but steady increase in average daily serum glucose concentrations as patients wore an infusion set continuously beyond 3 days. [41]

Irritation and infection at the infusion site

Insulin pump manufacturers recommend changing infusion sets and infusion sites every 48–72 hours to prevent irritation and/or infection at the infusion site. Wearing an infusion set for an extended period increases the risk of bacterial infection at the site, contact dermatitis from the infusion set adhesive, swelling and erythema at the infusion site, cannula occlusions, and hyperglycemia.

One study in which subjects wore an infusion set for increasing durations found that an increasing incidence of these problems began to occur on the third day of infusion set use. [41] According to one prospective descriptive study, with normal use, infected infusion sites occur once in every 27 patient-months. [37]

Infusion-site infections most often involve Streptococcus bacteria that have been introduced from the skin by insertion of the cannula, but Staphylococcus species and other pathogens may also be involved, particularly in staph carriers. [42]

Pump discontinuation

A number of studies [43, 44, 45, 46] have examined the rate of discontinuation of insulin pump therapy. Discontinuation rates found in these studies varied widely, from 18%-49% of patients in the studies. The reasons reported for discontinuing insulin pump therapy included interference with lifestyle, lack of improvement in glycemic control, or discomfort/infection at the infusion site.

Subjects were more likely to discontinue CSII if they were female, were younger, were pregnant, had a shorter duration of diabetes, had more frequent prior episodes of DKA, or had psychological comorbidities. Interestingly, one study [46] found that even the patients who discontinued insulin pump usage had decreased frequency of DKA after discontinuing insulin pump therapy than they had before initiating insulin pump use.