During the June 2017 episode and in this episode, when I heard discussion about insulin pumps, I noticed that there were a few misunderstandings about insulin pumps and their function. Insulin pumps were painted as all automatically reactive to blood sugar levels, but (unfortunately!) this is not the case. As an emergency physician who personally wears an insulin pump 24/7/365, I wanted to clarify some points.
It is most important to understand that insulin pumps are not (yet) fully autonomous. Users set a basal rate, which is a continuous subcutaneous infusion of insulin. Users also must bolus insulin every time they eat: they must tell the insulin pump how many carbs they plan on eating each and every time. Based on this, the pump will give insulin based on the user’s own insulin to carbohydrate ratio (which is different for every person and based on their own insulin sensitivity). For instance, I take 1 unit of insulin for every 13 g of carbs that I eat. Every time that I eat, every single meal, I need to estimate how many grams of carbohydrates that meal is and tell the pump how much I plan on eating. The pump will then suggest a bolus amount that is manually confirmed by the user; this bolus is given prior to eating.
Based on this, you can easily see how hypoglycemia could occur: because users of insulin pumps must “eyeball” the amount of carbohydrates they are eating and then tell the pump how much they are eating, if they overestimate they number of carbohydrates and their insulin bolus is too large for that meal, this can trigger hypoglycemia. Many insulin pumps have no idea what your blood sugar is unless you’ve supplied it with a finger stick blood glucose; some integrate with a continuous glucose monitor (CGM) that tracks blood glucose levels, but bolusing for meals and for hyperglycemia correction is user-driven. Some pumps that integrate with a CGM can automatically and temporarily suspend insulin delivery for hypoglycemia, but this is not that common. For the insulin pump that I personally use, for instance, it will continue to give me my basal insulin at the same exact rate regardless of whether my blood sugar is 30 or 300 mg/dL. It has no idea what my blood sugar is unless I tell it and will not make any automatic adjustments or withhold insulin (I must manually make changes).
Dr. Willis is right in that some pumps will automatically suspend insulin delivery during hypoglycemia, but this is not yet even close to a universal feature of insulin pumps. The statement made “If they have hypoglycemia and are on a pump, they probably have a pump problem” is not correct. If a user boluses too much for their meal, even suspension of insulin delivery (if the pump happens to have that feature; mine does not) will not be enough to prevent hypoglycemia.
There are no fully automated closed-loop systems available; the MiniMed 670G is the closest we have and was just released this year, and this pump still requires the user to estimate the carbohydrates and bolus for each meal but then subsequently the pump will use CGM data to make “micro-adjustments” to the insulin delivery and does automatically suspend for hypoglycemia. However, insulin pumps would never give an automatic large bolus of insulin for hyperglycemia without the user specifically triggering it in any scenario.
A good overview of basics of insulin pump therapy can be found here: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4730130/
Hey Ryan! This is some fantastic nuanced excellence! Thanks for sharing your thoughts and knowledge. We'll be putting this in an upcoming mailbag so everyone can benefit.
The HELLP/eclampsia was scary stuff, glad I wasn't there. While I understand it's not exactly the point of the segment, it might be worth mentioning that some of the distress of the situation may have been avoidable...Magnesium in this scenario can be infused at a rate of 4-6g in 15-20 minutes, alleviating some of the time crunch. Alternatively, you could give a whopping dose IM and push them out the door.
I had one other thought, regarding the afib RVR/fitbit segment...We know the patient's rate spiked at the time of the seizure, but technically they could have been in afib, not RVR, for longer than the 48 hour window. I'm sure the risk of thrombus is much higher with a rapid rate, and it is probable that the afib started at the time of the spike, but I think it's important to acknowledge that we are operating on couple of assumptions if we cardiovert based on the fitness tracker data.
To join the conversation, you need to subscribe.
Sign up today for full access to all episodes and to join the conversation.
Ryan P. - September 11, 2017 9:18 AM
During the June 2017 episode and in this episode, when I heard discussion about insulin pumps, I noticed that there were a few misunderstandings about insulin pumps and their function. Insulin pumps were painted as all automatically reactive to blood sugar levels, but (unfortunately!) this is not the case. As an emergency physician who personally wears an insulin pump 24/7/365, I wanted to clarify some points.
It is most important to understand that insulin pumps are not (yet) fully autonomous. Users set a basal rate, which is a continuous subcutaneous infusion of insulin. Users also must bolus insulin every time they eat: they must tell the insulin pump how many carbs they plan on eating each and every time. Based on this, the pump will give insulin based on the user’s own insulin to carbohydrate ratio (which is different for every person and based on their own insulin sensitivity). For instance, I take 1 unit of insulin for every 13 g of carbs that I eat. Every time that I eat, every single meal, I need to estimate how many grams of carbohydrates that meal is and tell the pump how much I plan on eating. The pump will then suggest a bolus amount that is manually confirmed by the user; this bolus is given prior to eating.
Based on this, you can easily see how hypoglycemia could occur: because users of insulin pumps must “eyeball” the amount of carbohydrates they are eating and then tell the pump how much they are eating, if they overestimate they number of carbohydrates and their insulin bolus is too large for that meal, this can trigger hypoglycemia. Many insulin pumps have no idea what your blood sugar is unless you’ve supplied it with a finger stick blood glucose; some integrate with a continuous glucose monitor (CGM) that tracks blood glucose levels, but bolusing for meals and for hyperglycemia correction is user-driven. Some pumps that integrate with a CGM can automatically and temporarily suspend insulin delivery for hypoglycemia, but this is not that common. For the insulin pump that I personally use, for instance, it will continue to give me my basal insulin at the same exact rate regardless of whether my blood sugar is 30 or 300 mg/dL. It has no idea what my blood sugar is unless I tell it and will not make any automatic adjustments or withhold insulin (I must manually make changes).
Dr. Willis is right in that some pumps will automatically suspend insulin delivery during hypoglycemia, but this is not yet even close to a universal feature of insulin pumps. The statement made “If they have hypoglycemia and are on a pump, they probably have a pump problem” is not correct. If a user boluses too much for their meal, even suspension of insulin delivery (if the pump happens to have that feature; mine does not) will not be enough to prevent hypoglycemia.
There are no fully automated closed-loop systems available; the MiniMed 670G is the closest we have and was just released this year, and this pump still requires the user to estimate the carbohydrates and bolus for each meal but then subsequently the pump will use CGM data to make “micro-adjustments” to the insulin delivery and does automatically suspend for hypoglycemia. However, insulin pumps would never give an automatic large bolus of insulin for hyperglycemia without the user specifically triggering it in any scenario.
A good overview of basics of insulin pump therapy can be found here: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4730130/
Rob O - September 22, 2017 10:18 AM
Hey Ryan! This is some fantastic nuanced excellence! Thanks for sharing your thoughts and knowledge. We'll be putting this in an upcoming mailbag so everyone can benefit.
Christopher H. - October 11, 2017 8:32 PM
Great episode, as always!
The HELLP/eclampsia was scary stuff, glad I wasn't there. While I understand it's not exactly the point of the segment, it might be worth mentioning that some of the distress of the situation may have been avoidable...Magnesium in this scenario can be infused at a rate of 4-6g in 15-20 minutes, alleviating some of the time crunch. Alternatively, you could give a whopping dose IM and push them out the door.
I had one other thought, regarding the afib RVR/fitbit segment...We know the patient's rate spiked at the time of the seizure, but technically they could have been in afib, not RVR, for longer than the 48 hour window. I'm sure the risk of thrombus is much higher with a rapid rate, and it is probable that the afib started at the time of the spike, but I think it's important to acknowledge that we are operating on couple of assumptions if we cardiovert based on the fitness tracker data.