Glucommander is a computer based system for controlling blood glucose by directing an intravenous infusion of insulin in response to the measurement of glucose at the patient’s bedside.
Paul Davidson and Dennis Steed collaborated to computerize and enhance insulin infusion protocols used by Dr. Davidson in the early 1980’s. Further details of the history of the Glucommander are provided at the end of this FAQ.
The Glucommander algorithm is typically implemented on a computer available at the patient’s bedside or nearby nursing station. A physician orders Glucommander therapy by specifying initial parameters. The nurse caring for the patient enters these parameters and the current glucose. The Glucommander then recommends an insulin infusion rate and a time to check the next blood glucose. At the recommended time, the nurse checks the blood glucose and enters it into the Glucommander. The Glucommander again recommends an insulin infusion rate and the time to check the next glucose. This process is repeated indefinitely.
The insulin infusion is always according to the formula:
Insulin per hour = multiplier x (blood glucose – 60)
where “multiplier” is a parameter which is automatically adjusted based on the glucose pattern and response to insulin.
To order the Glucommander, the physician must specify five parameters:
Various targets are used. In general, a higher target range will result in less risk of hypoglycemia although at the expense of a higher average glucose. We have used targets as low as 80 to 95 for healthy patients being stabilized overnight in preparation for testing to as high as 140 to 200 for patients undergoing surgery with anesthesiologists uncomfortable with glucoses in the normal range.
There are several rules of thumb that can be used. For an adult with normal insulin sensitivity, a multiplier of 0.01 is a good initial choice. Although this tends to be on the low side, it will be automatically adjusted upwards as needed. Also, if your clinical experience suggests an appropriate initial insulin rate for a given patient and glucose, it is possible to solve for the corresponding multiplier = insulin rate / (bg – 60). For example, if you know a patient on an insulin pump typically has required 0.5 units per hour to maintain a glucose of 100, a multiplier of 0.5 / (100 – 60) = 0.0125 could be used.
This depends on your perceived risk of the patient having a sudden shift in the blood glucose. In general, 120 minutes is reasonable for most patients, although this can decrease to 60 minutes for patients in surgery or critically ill. It should be noted that blood glucoses can be entered more frequently than the Glucommander requests.
This depends on the infusion device. In general, it is good to be able to adjust the infusion rate by increments of less than 0.1 units per hour for hospitalized adults. We are currently using 0.5 units per cc solution in infusers that are adjustable to the nearest 0.1 cc per hour. This means the infusion can be adjusted by 0.05 unit per hour increments. Higher concentrations can be used in volume critical patients with insulin resistance.
In addition to the algorithm parameters, the physician must order an appropriate IV or tube feeding. To ensure a constant glucose flux, a controller should be used. The insulin drip is typically “piggy backed” into a maintenance IV at a side port close to the infusion catheter to eliminate lag in insulin delivery.
If the blood glucose is above target and failing to decrease, the multiplier will increase. If the blood glucose is below target and failing to increase, the multiplier will decrease.
Anywhere from every twenty minutes up to the Maximum Interval specified in the initial orders. Typically the monitoring interval is about an hour, increasing when glucoses stabilize in the target range and decreasing if sugars are low or falling rapidly.
In general, the Glucommander can only control the blood glucose when the rate that glucose enters the patient’s system from the gut or by IV is constant. This can be accomplished by ordering a certain rate of infusion of dextrose containing IV solutions or tube feedings. Intravenous boluses of glucose and meals containing significant carbohydrate should be avoided. The Glucommander is quite capable of stabilizing blood sugar even if the patient is fasting and receiving no IV dextrose or food.
Glucommander has been used in many clinical settings including:
The Glucommander cannot respond quickly to sudden changes in glucose. Because the blood sugar checks are typically an hour or more apart, abrupt changes in the rate at which glucose enters the system can result in large changes in blood sugar before the system can respond. Therefore, patients should not eat discrete meals on the Glucommander, although a constant tube feeding is well handled. Abrupt cessation of IV dextrose or tube feeding may result in hypoglycemia. On the other hand, an incremental increase in IV dextrose or tube feeding is usually handled well, with only a transient increase in the blood glucose.
We have accumulated well over 100,000 hours of experience in over 5000 cases with the Glucommander. This data collection represents only a fraction of the actual experience. We have found it to be safe and effective. In all cases, the Glucommander’s recommendations have been appropriate and consistent with what we would recommend. Any deviations have resulted from things such as interruptions in IV fluids or tube feedings, improper entry of initial parameters, disconnected insulin infusion pumps, patients receiving trays of food when they should not, etc. In general, the Glucommander has made appropriate and reasonable recoveries from these operating errors.
We do not claim that an expert endocrinologist, constantly at the patient’s bedside, might not do better in controlling an insulin drip. However, we do believe that the Glucommander is an expert system that accurately captures at least one approach used by practicing endocrinologists. We are also comfortable with it because it is an evolution of protocols already in clinical use. In fact, we believe that it eliminates any ambiguity in the complex orders we used to write for insulin drips, and actually results in a much more accurate implementation of our intentions.
Glucommander had its beginnings in 1983 when Paul Davidson,
then practicing at
Davidson had read a paper [White et al, “Practical
closed-loop insulin delivery. A system for the maintenance of overnight
euglycemia and the calculation of basal insulin requirements in
insulin-dependent diabetics.” Ann Intern Med. 1982 Aug;97(2):210-3] describing
a protocol for administering intravenous insulin to patients in the clinical
research unit at
Insulin per hour = multiplier x (blood glucose – 60).
When the multiplier is equal to 0.02, this was a close fit to the sliding scale of White.
When Steed heard about Davidson’s work, he proposed that a device be created that would infuse insulin automatically using Davidson’s formula. A nurse at the patient’s bedside would need only to enter a glucose into a computer, and the computer would drive an infusion device which would deliver the correct amount of insulin. The first implementation of this approach was in early 1984. In this form, a Tandy TRS80 model 100 laptop computer was joined to an Autosyringe AS2c insulin pump. This initial version of the program used a constant multiplier that was entered manually.
By this time, Davidson had begun writing orders for adjusting the multiplier based on the patient’s response to IV insulin. Davidson felt that for the Glucommander to be clinically useful, that these adjustment rules must be incorporated into the Glucommander program. For this reason, the initial version was never used clinically.
For the next version of the Glucommander, Steed incorporated features of the adjustment protocol that Davidson had been using as well as several new features. Clinical use of the Glucommander was begun in December, 1984.
Glucommander was an immediate success and rapidly became the standard of care for managing intravenous insulin in our institution. It became apparent that there were some problems with the rules Davidson had been using for adjusting the manual insulin drips. In the spring of 1985, Steed modified the rules for multiplier adjustment to improve stability.
Eventually, it became apparent that the original reason for the creation of the Glucommander (the ability to automatically drive an insulin pump based on the input of sugar alone) was less valuable than the ability of the Glucommander to automatically adjust the multiplier by “learning” the patient’s insulin sensitivity.
In 1992, Glucommander was modified to actually eliminate the direct link to the insulin infuser in favor of directions to the nurse at the bedside to manually set the infusion rate for the insulin containing IV.
The Glucommander algorithm has been periodically updated to accommodate new devices. The most recent adaptation has been to the Palm computing platform.
Copyright © 2005 All rights reserved.