Building and Implementing Insulin Infusion Order Sets and Protocols

Andrew Ahmann, Richard Hellman, Greg Maynard

Introduction and Getting Started

Implementing a medical center-wide standardized insulin infusion order set with supporting policies, protocols, monitoring programs, and educational programs is a daunting task, but can be manageable by continuing to use the performance improvement framework presented earlier in this resource room. Whether you are far along in implementing your infusion order sets or just starting, following the steps outlined here could greatly enhance your chances of long-term success and allow for substantial improvements.

The multidisciplinary steering committee needs to do the homework carefully here and reaffirm the institutional support for implementing uniform standards for insulin infusions and the infrastructure needed to achieve good glycemic control in all crucial critical care and perioperative arenas. Once this assurance is confirmed, the team should revisit the institutional assessment items pertaining to insulin infusion in Analyze Care Delivery.

All insulin infusion orders and policies/procedures presently used in the institution should be identified and examined. The stakeholders (physicians, nurses, and pharmacists) for these order sets should be engaged in a dialogue and encouraged to share their experiences about the order sets they use and the attendant work flow, glucose monitoring, and reporting. Any data about glycemic control and hypoglycemia rates should be scrutinized and then compared against the metrics outlined in Track Performance.

As you go through this process, you should be able to identify current and potential local champions in the physician, nursing, and pharmacy realms. You will also uncover barriers, dysfunctional and inconsistent practices, and individuals who will pose challenges to your efforts. Identifying these issues should not discourage the team; rather, it should guide your interventional strategies and help to build a consensus that change is needed.

In the next subsection, we encourage you to step back and look at your infusion order sets in the context of the published protocols outlined that follow and in the context of their many variations in common clinical use. Some examples are available for your perusal in the Clinical Tools section. Look for the annotated examples of Intravenous Infusion Order Forms and Transition to Subcutaneous Regimens. Although there are no head-to-head comparisons of protocols of different models, your team will probably come up with an infusion protocol model that you prefer to others, given your own unique history, resources, and infrastructure.

Once you pick the basic structure for your insulin infusion order set, we encourage you to enhance the basic version with a variety of measures designed to enhance reliability of use and safety in your institution.

Review of Published Protocols

The evidence base for improved outcomes from enhanced control of hyperglycemia in the hospital has been spearheaded by studies in cardiac surgery patients and intensive care units. Although there are many reasons to believe benefits will come with broader, hospital-wide application of improved glucose measures, it is acute care settings that have led the way. There has been little question that intravenous insulin infusions are the most effective means of expeditiously and safely gaining glucose control in acute, labile situations. Accordingly, protocols for intravenous infusions have been quickly popularized in many intensive care units. The evolution of this process has been marked by publication of a few protocols and local development of many others, often modifications of the originals. There have been no published comparisons of various protocols for such factors as rapidity and consistency of reaching targets or frequency of hypoglycemia.

Most GIK (glucose-insulin-potassium) infusion protocols don't have glucose control as the primary objective and are not recommended for the specific purpose of glucose control. This is not to be confused with the common addition of potassium to IV insulin infusions, but without the intent to infuse large amounts of glucose. Glucose control is the issue, not delivery of substrate to various tissues.

Several general principles have consistently characterized these protocols. It is worth considering the published protocols to understand the various approaches to intravenous insulin and to view the overall progress in this area.

Protocol algorithms can first be classified as those meant to track according to insulin sensitivity (or resistance) and those intended to adjust rapidly according to present and last glucose without use of a specific sensitivity modifier.

Protocols that track by level of insulin sensitivity generally use the column method, with individual columns representing different categories of insulin sensitivity, generally presented with the highest insulin sensitivity category on the left and the highest insulin resistance category on the right. These protocols use a multiplier to adjust for sensitivity and in essence are constructed according to the "rule of 1500 or 1800," adjusting for changes in insulin sensitivity that follow surgery or other alterations in physiologic stress in the acute setting. Three examples of this type of protocol follow:

1. Markovitz LJ, Wiechmann RJ, Harris N, et al. Description and evaluation of a glycemic management protocol for patients with diabetes undergoing heart surgery. Endocr Pract. 2002;8:10-18.

   This protocol was used in cardiac surgery patients. It was the first published article form of the column method and incorporated a higher glycemic goal than typically proposed today. Many parties have modifieid the column method by adjustment of goals. It offers a specific infusion rate and directed sensitivity adjustments for nursing personnel.
   
   

2. Trence DL, Kelly JL, Hirsch IB. The rationale and management of hyperglycemia for in-patients with cardiovascular disease: time for change. J Clin Endocrinol Metab. 2003;88:2430-2437.

   This protocol was adapted from the Markovitz protocol. It uses four columns represent algorithms of increasing insulin resistance. Although this article reported success with this method, specific results were not included. It has been used in critical care units and on medical-surgical floors.
   
   

3. Osburne RC, Cook CB, Stockton L, et al. Improving hyperglycemia management in the intensive care unit: preliminary report of a nurse-driven quality improvement project using a redesigned insulin infusion algorithm. Diabetes Educ. 2006;32:394-403.

   This article reports a pilot study implementing a 10-column method based on 10 levels of insulin sensitivity called the "Columnar Insulin Dosing Chart." It was used in the ICU. Only 0.9% of readings were < 60mg/dL. It took about 12 hours to reach goal (80-110 mg/dL).
   
   

4. Braithwaite SS, Edkins R, MacGregor KL, et al. Diabetes Technol Ther. 2006;8:476-488.

   Dr. Braithwaite is an author of the original Markovitz protocol, above, and has continued to evolve this protocol as have others. In this article, the authors publish their 6-column protocol, designed to attain glucose levels < 110 mg/dL in trauma patients. The hypoglycemic rate (< 70 mg/dL) was <1.0% of all POC glucose determinations and about 2.4% once goal glucose levels were obtained. No values were <50 mg/dL. For patients who eventually achieved goal glucose (< 110 mg/dL), the median time was 11 hours.
   
   

The above implement "paper" orders. The same concepts have been used to develop computer-assisted insulin infusion protocols. One such method has been published using the Glucommander, but a number of institutions use similar, locally developed computer-assisted protocols.

1. Davidson, PC, Steed RD, Bode BW. Reduction in severe hypoglycemia with long-term continuous subcutaneous insulin infusion in type I diabetes. Diabetes Care. 2005;28:2418-2473.

   This method has been at least as good as paper methods. It uses a continuously updated insulin sensitivity algorithm and should nearly eliminate nursing error. The nurse is asked to enter the glucose level twice (for confirmation). The prevalence of hypoglycemia < 40 mg/dL was relatively low, 2.6%. It also can help in the transition to subcutaneous insulin. Its commercial availability is not clear at this time.
   
   

Protocols that adjust to any situation according to present glucose and change from last glucose generally require more calculations by the nurse. It has been suggested that these protocols are more agile and more flexible, but there have been no direct comparisons with column-method protocols in effectiveness, nursing error, or hypoglycemic risk.

1. Furnary AP, Wu Y, Bookin SO. Effect of hyperglycemia and continuous intravenous insulin infusions on outcomes of cardiac surgical procedures: the Portland Diabetic Project. Endocr Pract. 2004;10(suppl 2):21 - 33.

   This protocol was used in cardiac surgery patients. It has been intensified at least four times since its original version in 1992. This particular version was used from 2001 from 2003 with a stated glucose goal of 100-150 mg/dL. It uses a combination of fixed adjustments (eg, 0.5 units/hour) and relative adjustments (eg, 50%). Goal glucose is reportedly attained in 3 hours. Hypoglycemia (< 60 mg/dL) was reported to occur in 0.5% of "patients in whom the protocol was strictly followed." Changes over the years have included longer duration of use with more conservative goals for patients once they have left the unit. A more intensive protocol with a goal below 110 mg/dL is now used, but this protocol has not been published. Although the Portland protocol publications proposing updated protocols can be found at the Portland Diabetic Project Web site, at present no protocols are listed. The published material on this protocol is quite detailed about the morbidity and mortality benefits but is much less so about complications and ease of use for nurses. The hypoglycemia data are also minimal. Recent versions include a postmeal dose of a rapid-acting insulin analog with a crude adjustment for meal size but no adjustment for insulin sensitivity. The unpublished, updated protocol reportedly includes additional adjustments with IV bolus insulin when postprandial glucose is high, but the risk of hypoglycemia has not been reported with this approach.
   
   

2. Van den Berghe G, Wouters P, Weekers F, et al. Intensive insulin therapy in the critically ill patients.

   N Engl J Med. 2001;345:1359-1367.

   This protocol was used in surgical ICU patients whenever glucose rose above 110 mg/dL. The same protocol was later used in a study of medical ICU patients with slightly different results. The glucose goal is 80-110 mg/dL. A nurse makes frequent adjustments using rate of change to calculate relative rate adjustments. Prevalence of hypoglycemia < 40 mg/dL was 7.2% in the SICU study but was higher in the MICU study (19%). This is the best prospective outcomes study reported, but the protocol may be too intensive, with relatively high rates of hypoglycemia.
   
   

3. Goldberg PA, Siegel MD, Sherwin RS, et al. Implementation of a safe and effective insulin infusion protocol in a medical intensive care unit. Diabetes Care. 2004;27:461-467.

   The original report on this protocol gave a goal of 100-139 mg/dL for medical ICU patients; when republished it listed a goal of 90-119 mg/dlL (Endocrine Pract. 2006;12(suppl 3):79-85). It uses two tables. The first helps the nurse identify the infusion rate delta (Δ"), or change, according to the present BG and rate of glucose change. The second table converts the " to an insulin infusion rate change, which is dependent on the current infusion rate. Hypoglycemia at 40 mg/dL was reported to be 4.3% in the MICU and 0% in the cardiothoracic surgery ICU.
   
   

4. DeSantis AJ, Schmeltz LR, Schmidt K, et al. Inpatient management of hyperglycemia: the Northwestern experience. Endocr Pract. 2006;12:491-505.

   Recently published, this protocol uses three tables to adjust rates. Its goal of 80-110 mg/dL is reached in 10.6 hours. Prevalence of hypoglycemia < 60 mg/dL is reported to be 1.6%.
   
   

These methods can be computerized as well.

1. Vogelzang M, Zijlstra F, Nijsten MWM. Design and implementation of GRIP, a computerized glucose control system at a surgical intensive care unit. BMC Med Inform Decis Mak. 2005;5:38.

   This method uses a computer program that receives information directly from POC testing and selects an ongoing infusion rate and interval for the next glucose test. It looks at rate of change over a period of hours in making calculations and recommendations. Glucose levels < 60 mg/dL occurred in about 11% of patients. The authors reported that this protocol will be freely available online in the future.
   
   

Review these published protocols and their variants in the Clinical Tools section. Compare and contrast them with your own currently used order sets and decide which basic type of insulin infusion order set you'd like to implement widely.

Making this selection is only the beginning. Indeed, plucking an insulin infusion order set "off the shelf" and expecting to plug it in without paying attention to implementation, safety, and local institutional issues could lead to problems and increased resistance to your efforts.

We suggest that you address the multiple issues surrounding the use of insulin infusion, define your institutional practice for different situations, and then augment your infusion order set by building as much guidance, automation, and high-reliability design into your order set and attendant processes as possible, following the principles set forth in the Layer Interventions sections: Protocols and Reliability, Advance Reliability, and in the Continue to Improve section: Plan-Do-Study-Act. The next section is designed to help you do just that.

Implementation and Safety Issues

Safety issues
The use of insulin/glucose infusions has completely changed the standard of care for the hyperglycemic inpatient. However, with new tools come new potential hazards of which hospitalists and others caring for these patients should be aware. In this section, we cover some key points to consider when using these methods in order to minimize accidental injury to patients being given insulin therapy via insulin/glucose infusions. It should be understood that the authors view the infusions as a great asset, but it is their power to rapidly change glucose levels that carries the risk of overshooting and causing hypoglycemia. Conversely, the infusions may be considered so effective that their underutilization is not noted or responded to effectively, leaving a patient severely hyperglycemic despite an infusion being in place.

Initiation of insulin/glucose infusion' patient safety issues
The initial assessment of a patient's clinical status is key, because the context of the clinical setting in which the infusion occurs may be unclear. For example, choosing the initial rate of insulin/hour of the infusion depends on an expert assessment of the probable degree of insulin resistance and overall clinical status of the patient and does not primarily depend on the initial glucose level alone. In clinical settings, the initial insulin requirement may be as little as <0.5 units/hour for a severely dehydrated, mildly azotemic Addisonian patient or >30 units/hour for an obese, extremely insulin-resistant patient with respiratory distress who is on beta agonists for lung disease and high-dose steroid therapy. Both may have an initial glucose level of 450 mg/dL, but the initial choice of infusion of insulin and the rate and increment of change will vary greatly. This could lead to iatrogenic hypoglycemia in the first case and to an unacceptably long duration of severe hyperglycemia in the second. Although general guidelines incorporated into orders at the point of care are desirable, the use of ironclad rules for assessing initial insulin requirements is usually a mistake: there is no substitute for directly observing a patient, reviewing the patient's status, and using expert judgment to begin the infusion.

System issues
Unfortunately, it is often the case in a hospital that key people involved in translating order into action may be either unsympathetic to or unknowledgeable about the orders. Knowing the steps required to achieve the goal is key, and time spent beforehand rehearsing and teaching key staff may make the difference in whether there is safe and effective implementation of the excellent insulin protocols attached to the document. Nurses who do not understand the orders may mis?handle the infusions in ways that expose the patient to unexpected hyperglycemia or hypoglycemia.

It is important to have an algorithm that allows the nurse to contact the physician for help if the infusion is not working as expected. It is equally important for the physician to look at the continuing results of the infusion with a jaundiced eye and not assume that just because the orders are exceptional that users of the protocols are equally clear as to how they work and are sympathetic to the goals of the physician.

"Field trips" and interruptions in nutrition or infusion
Patients on insulin infusion often need to go to the OR, for imaging studies, or to other places where caloric intake could be interrupted or where it is difficult for glucose values to be monitored and monitored as frequently as they would be in the unit. It is critical to think about these issues ahead of time and to arrive at a common institutional procedure for addressing them. Again, guidance for handling these issues should be incorporated into your order sets to the extent it is possible to do so.

For example, many institutions use D10 solution to replace the carbohydrate calories lost when tube feedings have to be interrupted. Anesthesia, interventional radiology, and OR and recovery room staff must all be included in crafting solutions that will continue a consistent approach to glycemic control as the patient goes on "field trips" to their respective areas, and nursing procedures need to reflect this.

Another common scenario is when insulin infusion is interrupted unbeknownst to the ordering physician. This may occur for many reasons, such as the IV site becoming clogged or the patient pulling out the infusion accidentally or intentionally. Also, doctors and nurses may deliberately stop the protocol for many other reasons, including using the only IV site to deliver blood or there being incompatible medication, and they may mistakenly or deliberately not restart the infusion. It is important to realize that the algorithm for insulin infusion should have clear-cut orders that if the infusion is stopped for any reason, the ordering physician must be immediately notified. Otherwise, in-hospital ketoacidosis and death could well be the patient's outcome. Unfortunately, this preventable scenario, although uncommon, has occurred in clinical settings.

Another common situation is a drastic change in a patient's insulin requirements occurs, but the ordering physician does not know this. This could be a decreased in insulin requirements; for example, when the pulmonologist abruptly stops intravenous steroid therapy, the insulin requirements plummet and so does the glucose unless the physician monitoring the insulin infusion has anticipated this and the insulin dose decreases in locked step. An example in the other direction may be a marked increase in insulin requirements from a change in medication, the onset of severe pain, sepsis, or early vascular collapse, all with very different consequences and different time courses. Rarely, a pharmacy or nursing error may induce a pseudo-change in insulin requirements, and this possibility can be explored by hanging a new infusion.

A regrettably common clinical error is not asking why a drastic change in insulin requirements has occurred, because the answer may be far more important than just the glucose level itself. It is important to keep in mind that, just as capillary oxygen concentration may mirror important clinical events, so can glucose level both mirror important events and be critically important for a patient's health in its own right.

More benign and common causes of variation in glycemic control for patients on insulin infusions are deviations from the protocols. This can occur because of how the insulin rate is changed or, if the patient's clinical status has sufficiently improved to be fed, because of changes in delivery of medication such as oral medication whose intensity and time course varies, requiring being compensated for by the infusion rates, which may not do so perfectly. It is important, from a safety perspective, to keep in mind that looking at the glycemic level alone is much less likely to be as useful a guide about what to do as is the story behind the glucose level. Too often a nurse may misread an order and the wrong rate is going, or the insulin concentration is different than what was ordered.

This is particularly important with lower insulin requirements, because if the insulin infusion rate is < 1 unit/hour, either the delivery system has to give <1 cc/hour accurately, or the patient will become hypoglycemic on the lowest infusion rate. Also, if the rate of glycemic monitoring occurs less often than the speed at which the glucose is changing - for example, if the glucose is dropping by 100 mg/hour and the glucose level is 180 mg/dL but is being checked only every 2 hours - hypoglycemia may occur between checks. Trend analysis and algorithms that compensate for speed of drop may be needed.

Other Implementation Issues

Authors' note: In the Implementation Issues section, we often refer to " Selling Root Canals: Lessons Learned from Implementing a Hospital Infusion Protocol" (Goldberg PA, Inzucchi SE. Diabetes Spectr. 2005;18:28-33).

Choosing a glycemic target and overcoming implementation barriers
Insulin infusion order sets generally have an inherent glycemic target range, so the model you adapt should have a glycemic target consistent with your own. Although the national guidelines call for a glycemic target with an upper limit of 110 mg/dL in critical care areas, there is some room for debate on this issue, and at your institution you may run into legitimate concerns about safely achieving this target or even downright refusal to accept an insulin infusion protocol.

Methods to address the glycemic target and insulin infusion barriers:

1. Compromise - start with a higher glycemic target (eg, glucose upper limit of 130 mg/dL, desired range 90-130 mg/dL). You can probably get support to lower the threshold when you can demonstrate that achieving the target can be done safely and is well accepted.

2. Identify a local champion in nursing and physician areas, as discussed above.

3. Pilot your protocol in an area with the least resistance and the strongest champions, thereby gaining momentum with a demonstration of success.

4. Allow different targets in certain units but require the target be uniform within a unit. Example: in a trauma ICU, the target may justifiably be a little more lax, but the same order set/protocol should be used with all trauma unit patients, and the basic structure and instructions given or implied in the order sets should be uniform across the institution.

5. Shore up your institutional support and make the protocol the default with an automatic trigger. If you really have the institutional support you need, after negotiation and discussion with key medical staff, your institution should back up the idea of insulin infusion being the default at a certain trigger glucose level (which may even be as high as 150 mg/dL in some places). Once the trigger glycemic excursion is noted and confirmed, insulin infusion according to the protocol is begun. The physician may opt out with good cause, but otherwise, the default premise that hyperglycemic patients in the ICU setting get placed on the insulin infusion protocol is upheld and automatic.

Need for various protocols and various targets for control
Insulin infusion protocols are most common in the intensive care setting. However, the insulin infusion protocols most frequently used for the management of hyperglycemia in the ICU patient are not suitable for those with acute hyperglycemic emergencies such as DKA. You must use caution in rapidly correcting those with DKA to the usual ICU target levels (eg, 110 mg/dL), although this practice is desirable in most other uses of an IV insulin protocol.

In some institutions IV insulin protocols are safely and effectively used outside the ICU. Obviously, the number of nurses and other personnel who must be familiar with the protocols is much higher outside the ICU, and protocol errors are somewhat more likely to occur.

Suggestions for safe implementation of insulin infusion regimens outside the critical care setting include:

Choose an infusion protocol with a higher glycemic target.

Select one general medical surgical unit in which to concentrate the expertise and focus on insulin infusion.

Consider simplified infusion protocols.

Look into automated and closed-loop computerized assistance, which can relieve some of the nursing burden from maintaining insulin infusion.

Educating and engaging nursing staff
The nursing staff generally bears the brunt of implementing intensive insulin infusion protocols on the front lines. Educational efforts for nurses should explain the rationale for why intensive infusion is importance, in addition to the particulars of how to properly administer your insulin infusion protocol. Frontline nurses then become some of the strongest advocates for intensive insulin therapy, instead of reluctant partners. Organizing regular sessions during early-implementation efforts to elicit their feedback on ease of use, clarity, and difficulties encountered is essential.

Your team should listen to and address the concerns of the frontline nursing staff. Some examples, as outlined by Phil Goldberg and Silvio Inzucchi in "Selling Root Canals: Lessons Learned from Implementing a Hospital Infusion Protocol" (Diabetes Spectr. 2005;18:28-33), are:

BARRIER: Inadequate number of glucose meters to match the increased frequency of testing.

SOLUTION: Purchase more glucose meters. Some newer meters yield test results more quickly, and units that simultaneously transmit results wirelessly to your hospital database are becoming available.

BARRIER: Checking the glucose and making changes in the insulin infusion takes a lot of time for busy nurses in a critical care unit.

SOLUTIONS: Get ancillary help to check glucose values (nurse assistants), and focus on eliminating duplicate work in the process of recording glucose values. You will need to retain important alert triggers to call a physician, but you can eliminate needless physician calls by allowing infusion to be predominantly automated and nursing driven.

BARRIER: Frequent point-of-care glucose testing leads to frequent uncomfortable sticks to the patient and eventually sore finger tips.

SOLUTION: Intensive care unit patients often have central lines or arterial lines that can provide samples for hourly testing, a practice that is acceptable because venous, arterial, and capillary blood glucose testing usually vary from each other by less than 10%. As long as a consistent source of sampling is used during each shift, any of these three sources should be adequate to guide therapy. Note that in situations such as hypotension and shock, capillary blood glucose testing can be unreliable, so the other sources would be preferred.

Hypoglycemia and dispelling hypoglycemia myths
Fear of hypoglycemia is one of the most potent barriers to intensive insulin infusion implementation. Because hyperglycemia is so common in critical care units, the nurses and physicians there may have developed a skewed view of what is considered hypoglycemia, at times fearing for a patient whose glucose level is about 100 mg/dL. Poll each nurse on her or his self-estimated fasting glucose level and then measure it - the nurses might be surprised that that scary 80 mg/dL reading is the same as their own.

Hypoglycemia does occasionally occur with infusion protocols, no matter how carefully you craft your protocol and how well it is administered. Therefore, your hypoglycemia protocols should be incorporated directly into your infusion order set.

Insulin infusion preparations and administration
Insulin infusions should generally be centrally prepared with a standard concentration of regular insulin in the pharmacy, and the infusion concentration should be included in your infusion order set. When nurses change IV tubing or initially set up an insulin drip, education/instructions on priming new tubing with 30-50 mL of insulin infusion to saturate the binding to the PVC tubing should be included.

Monitoring, recording, and analyzing glycemic control data
Frequently measuring glucose level is necessary for the safe infusion of insulin. Guidance for how often monitoring is required must be explicit and included in the infusion order set. Occasionally auditing for compliance in the frequency of glucose testing is good practice. Attention should be paid to how glucose levels are obtained, recorded, and made available to the health care team in your institution and to recording the response to the glucose reading. All glucose readings should be recorded electronically for ongoing analyses and retrieval, and ideally, this can be done in an automated or single-step method. Try to eliminate duplication of effort, such as asking nurses to record glucose data and comments both on paper and electronically. Your team should also provide guidance about the potential problems of using point-of-care (POC) glucose testing for patients with hypotension, using pressors, and other conditions that might impair the accuracy of POC readings.

Steroids
Steroid boluses can lead to glycemic excursions that result in rapidly varying insulin requirements. Your team should investigate promoting the use of steroid infusions in situations when a bolus is not absolutely necessary.

Dealing with an eating patient and other sources of carbohydrate-induced glycemic excursions
The blood glucose level of a patient who is eating while on insulin infusion can be difficult to control, as the infusion chases the glycemic by rapidly going up and then back down. We recommend providing bolus nutritional insulin to cover the expected glycemic excursion caused by carbohydrate ingestion. Counting carbohydrates and using a unit of insulin for each 10-15 g of carbohydrate consumed can smooth out the rapid fluctuations in glucose. Guidance for this should be incorporated into your order set.

Transitioning off insulin infusion
One transition of care that deserves special mention is the transition off insulin infusion. Rational strategies for dealing with this transition are covered in the subcutaneous insulin section of this resource room, the SHM PowerPoint educational module available in the Education Resources section of the resource room, and several examples are also available in the Clinical Tools section of the resource room, but we will review this important transition briefly here as well. Guidance for managing this transition should be integrated into your insulin infusion and subcutaneous order sets. The transition to subcutaneous insulin may represent a separate order set but is sometimes integrated into the IV insulin infusion set.

This transition is a time of high risk of loss of glucose control. Consider having a pharmacy or nursing check of all orders to discontinue insulin infusions for all patients with type 1 diabetes, DKA, or those remaining on infusions of 0.8 units per hour or more. The goal of this check is for scheduled subcutaneous insulin to be ordered before the infusion is discontinued and to call the physician to request orders for the recommended subcutaneous insulin. Most often intravenous insulin infusions should be followed by a basal-prandial insulin regimen.

For more details and guidance on the transition off of intravenous insulin infusion, please refer to the Transitions and Perioperative section and the Clinical Tools section.

Insulin infusions remain a very powerful clinical tool in the inpatient setting and must be used regularly. Optimally, they are used throughout the hospital whenever they are needed. To accomplish this goal and for safety reasons, much greater emphasis should be placed on collaboration between physicians, nurses, and pharmacists. Each of these groups makes errors, and the key to developing a culture of safety is cooperation, backup checks, and, most of all, robust training of all those involved in the care of a patient on an insulin infusion. The physician needs to always ask why a particular glycemic response is occurring, and most often a careful review of all the medications and clinical problems of the patient will be most useful both initially and for continuing management. The belief that an insulin drip can just run by itself on the proper algorithm puts the patient at risk and at present is simply an illusion, not good care. Nevertheless, with new techniques, the use of insulin glucose infusions is very likely to improve patient both outcomes and quality of care for the patient with hyperglycemia.

The final point is a reminder to track the use of your insulin infusion protocol, note variations from your protocol, and learn from them. In some cases, you will find that your infusion protocol was too difficult to use or was not well suited to some patient populations, suggesting areas where you might improve it. Alternatively, you may uncover some unexpected pocket of uneven education or buy-in among your staff that needs to be addressed.