While not a new tool, insulin pump therapy remains the gold standard of insulin delivery for type 1 diabetes (figure 9). CSII is the most precise way to mimic normal insulin secretion because basal rates can be programmed in half-hour segments throughout a 24-hour period. Essentially, the CSII pump may be thought of as a computerized mechanical syringe automatically delivering insulin in physiologic fashion. Patients can accommodate metabolic changes related to eating, exercise, illness, or varying work and travel schedules by modifying insulin availability on an hour-to-hour basis. Basal rates can also be adjusted to match lower insulin demands at night (e.g., between approximately 11 PM and 3 to 4 AM) and higher requirements between 3 to 4 AM and 9 AM. Although hour-to-hour variability is possible, most patients use only 1 to 3 basal rates per day. On the other hand, they may use totally different basal rates on weekends versus weekdays, summer versus winter, or while on vacation.Various studies comparing glycemic control during CSII versus intensive insulin injection regimens have been published. A meta-analysis of 12 randomized controlled trials of CSII versus multiple injection regimens showed a weighted mean difference in blood glucose concentration of 16 mg/dL (95% CI 9-22) and a difference in A1C of 0.5% (95% CI 0.2-0.7) favoring CSII. The slightly but significantly better control in patients on CSII was accomplished with a 14% average reduction in daily insulin dose.

Interestingly, a widely perceived disadvantage of CSII—i.e., that it predisposes patients to hypoglycemia—is one of the most common misconceptions about its use. In fact, a large body of evidence suggests that hypoglycemia is significantly less common in CSII than injection therapy, even when the injection regimen is not intensive.

Modern insulin pumps are much smaller and easier to use than the pumps of a decade ago. Most weigh around 115 grams (4 oz), and are approximately the size of a small cellular phone (figure 12). Each houses an insulin-filled cartridge or syringe connected to a 23- to 24-inch, 31-inch, or 42- 43-inch length of plastic tubing. At the end of the tubing is a 25- or 27-gauge needle or a soft Teflon cannula that can be inserted into the subcutaneous tissue at a 30- to 45- or 90-degree angle, depending on the type of infusion set used. The abdomen is the preferred infusion site because placement of the catheter there is convenient and comfortable and insulin absorption is most consistent in this region. However, the upper outer quadrant of the buttocks, upper thighs, and triceps fat pad of the arms may also be used.

Many infusion sets allow removal of the insertion needle, leaving only the soft cannula in place subcutaneously. Early adapters of CSII were required to use straight or bent needles but those for the most part have disappeared. After the syringe is placed in the pump, a lever mechanically pushes down the plunger of the syringe, and the insulin travels through the infusion tube, entering the subcutaneous tissue through the soft, flexible catheter. In current models, infusion lines have a “quick-release” mechanism, allowing the pump to be temporarily disconnected from the tubing going to the insertion site. This quick-release feature makes dressing, swimming, showering, and other activities more convenient. A newer version of CSII involves a disposable “pod” which is discarded every three days (figure 12). The insulin is infused directly from the pod through a catheter without the use of any tubing. Both basal and bolus insulin dosing is communicated to the pod though radio frequency via a separate “personal diabetes manager”. It is clear that choices of CSII continue to grow for patients with type 1 diabetes.

With the introduction of rapid-acting analogues, there is no reason to use regular insulin with CSII. Indeed, all three rapid-acting analogues are approved in the United States for use in insulin pumps. The basal rate of the insulin pump provides continuous delivery of microliter amounts of insulin thus providing basal insulinemia and replacing the need for separate basal insulin such as NPH insulin, insulin glargine, or insulin detemir. The boluses activated before each meal provide the incremental prandial insulin, just as would be done with separate subcutaneous injections of insulin lispro, insulin aspart, or insulin glulisine. The pump allows programming of many different basal infusion rates (usually ranging from 0.4 to 2.0 units/hour) to meet non-prandial insulin requirements, though it is unlikely that the average patient will require more than 3 different rates. As with MDI, correction doses can be provided before or between meals.

However, there are many fundamental differences between CSII and MDI. First, as noted above, the basal dose can be titrated for each individual throughout the day. There are various recommendations for how to best do this, but we prefer to perform “basal checks”. This is accomplished on a night the patient has a bedtime glucose level within target (e.g., 90 to 130 mg/dL). The patient then measures at least one but preferably two glucose levels throughout the night. The next morning breakfast is skipped and glucose is measured every hour, if possible until early afternoon. If hypoglycemia occurs at any time during this time period, food is consumed and the experiment is completed. An insulin adjustment can then be made. Alternatively, if glucose levels tend to rise at any time during this period, basal rates need to be increased. It is important to appreciate the insulin action times with basal rates, just as with correction doses noted above. For example, if glucose levels are found to increase around 5 AM, it would be appropriate to increase the basal dose around 4 AM.

Another major difference between CSII and MDI is that most current pumps can accurately track the insulin-on-board for safer use of correction doses (figure 11). As noted above, doing this accurately can have a major impact in preventing insulin stacking. Modern-day pumps also calculate insulin required based on carbohydrates anticipated to be consumed. So the seemingly complicated mathematics to best utilize MDI are done automatically with CSII. Finally, pumps can be programmed for individual boluses to be administered over an extended period of time (“extended” or “square wave” bolus). This feature may be particularly helpful for very high-fat meals, or those patients with delayed gastric emptying, seen with gastroparesis or those receiving pramlintide (see below).

From a practical point of view, the first and most important insulin dose to provide in a correct amount is the basal rate. If the basal dose is set incorrectly, neither the bolus doses nor the correction doses will be appropriate. The most common mistake we see (as with MDI) is the basal dose set too high, making even small correction doses cause hypoglycemia. The greatest advantage of CSII is it allows more flexibility and titration of the basal doses. The other major advantage of CSII is it allows the use of “temporary basal rates”, i.e., an insulin dose increase (such as during intercurrent illness) or an insulin dose decrease (such as during exercise) for a specific period of time. Again, due to the time action of the rapid-acting analogues, sufficient time must be incorporated when using a temporary basal rate. For example, we find that for exercise, the insulin rate must be decreased one to two hours prior to the activity.

There are several risks with CSII. The first is an abrupt stoppage of insulin delivery either from an occlusion or dislodging of the catheter. For most patients who measure glucose levels at least 4 times daily the problem can be discovered and rectified quickly. However, for the occasional patient who tests infrequently or misses several glucose tests the discontinuation of the insulin infusion can result in ketoacidosis. Fortunately, this is rare. When glucose levels are found to be elevated for no apparent reason, it is appropriate to bolus the appropriate correction dose and if after 1 to 2 hours glucose levels are not improved, a separate subcutaneous injection of a rapid-acting insulin analogue is recommended; the pump infusion site also should be changed.

Another potential complication is infection, sometimes even abscess formation, at the infusion site. This is also rare and can be minimized by meticulously cleaning the pump site prior to catheter or needle insertion. Although not as severe, inflammation at pump sites can be problematic. This can be minimized by changing the infusion set every 24 to 72 hours and rotating pump sites. Similarly, some patients develop lipohypertrophy from repetitively infusing the insulin in the same area. This can result in extreme variability in insulin absorption. Again, frequent rotation of pump sites can alleviate this problem which is under-reported. Clinicians should therefore make site observation part of every clinic visit.