Fasting on Yom Kippur — Part 4: What Should Your Blood Sugar Target Be During the Fast?

BS"D

Raising the target feels safer. The opposite is true. The number to set, why, and the one question to ask yourself before the fast begins.

In this post:

• “Rabbi, If My Basal Isn’t Reduced, I’ll Go Low”
• The Basal Is Coming Down Whether You Like It or Not
• This Argument Has Been Had Before
• Your Doctor May Have Told You to Use It
• Confidence — and the First-Year Rule
• What Changed for Me This Year
• The Target Question
• The Target: 120
• The Choices for the Tandem
• The First Night
• Two Edges

In Part 3 I laid out the two conflicting risks of a fast. Too much insulin and you go low. Too little and you are on the way to DKA. On a fast day both are live at the same time, and the whole question of how to set your pump is really the question of which of the two you move toward.

Much of what follows was written last year as well. It has been refined and restructured here into something you can follow, and it is written to stand on its own, so you do not need to have read anything before it.

“Rabbi, If My Basal Isn’t Reduced, I’ll Go Low”

I hear this every year, and the answer is short. Then you break the fast.

A low is fixed in ten minutes with a drink. DKA is not fixed in ten minutes, and sometimes not outside a hospital. Between the two risks, the one you are permitted to take is the low. Risking DKA is not on the table. It is not an option, and everything in this post assumes that from here on.

Which means this: if there is no way for you to avoid the low without cutting insulin too much, the fast is over. Not “be careful.” Over.

And if that is where you already stand — if the only way you can see to get through the day without a low is to take off a lot of insulin — then the decision was made before Yom Kippur started. Fasting is not meant for you this year. I will come back to what that means, and how to know, in the second half of this post.

I have written this before and I will write it again. I have witnessed trips to the emergency room at the end of Yom Kippur with good blood sugars — and DKA.

One of them I will tell you about. Many years ago I went to see the rav of a wealthy community. I was there to raise money for Friends With Diabetes, and we had a long conversation about what the organization had done and was doing. Fasting came up. He told me a woman in his community has Type 1, and he would have her call me. She did. I sent her our instruction guide, and we went through it together on the phone. On Motzaei Yom Kippur the phone rang again. She was in the hospital.

That was the end of my fundraising in that community. And I have never been able to look that rav in the face since.

The Basal Is Coming Down Whether You Like It or Not

There is no such thing as a fast day on which the basal stays where it was. It cannot — as the day goes on the liver’s stores of glucose run down, and the fast itself does that. That is what the Excel file was for, in the first version of our guidance — you entered your basal rates, it gave you the reductions for the day, and the special profile was where you put them. That is what the temporary basals were for after it. We had to reduce, so we built tools to reduce by.

Now our sophisticated pumps do it. Out of sight, some of them every five minutes, without a profile and without asking. But it does it. The basal on Yom Kippur afternoon is not the basal you had on Tuesday, on any pump, in any mode.

Now put the whole day together. No boluses for twenty-five hours, because there is no food. The basal shrinking hour by hour. The pump cutting the insulin further on its own every time the number drifts, because the only thing it sees is the number. That is the baseline. That is before anyone has touched a setting.

Now raise the target to 140 to 160. The pump takes off more insulin still, to get you up there and keep you there. Every one of those is a reduction, and they do not replace each other. They stack. By the middle of the afternoon a person in exercise mode is running on less insulin than at any hour of any other day of the year, and there is no bolus coming to interrupt it.

That is what exercise mode does on a fast. It pushes you toward the second risk — and it does it in the hours when you are already closest to it.

This Argument Has Been Had Before

In the years when we reduced basal by hand, I had one hard rule and I gave it to everyone: never reduce basal below fifty percent. And even that only for a few hours, not the entire day.

A doctor fought me on it. She told her patients to keep reducing as far as the blood sugar warranted — if the number is fine, the insulin is fine. She told me there is no such thing as DKA when the blood sugar is normal. All I had to do was send her a few reports of euglycemic DKA. It is known. It cannot be refuted, and it was not.

That argument disappeared for a while, because nobody reduces basal by hand anymore. We handed it to the pump. But the argument did not go anywhere. It is back, in different clothes, and its name is exercise mode.

Your Doctor May Have Told You to Use It

This is the hardest section of the series for me to write, and I have rewritten it more than any other.

I am not a doctor. So how can I disagree with what a doctor tells his patient?

This is what is pushing me. Doctors do tell people to put the pump in exercise mode for the fast, and I disagree with them. Euglycemic DKA is real. Exercise mode aggravates it. We are not allowed to do that, and I am not going to stay quiet because of who is saying it.

If people hear the advice from one another, it is an easy target. If they hear it from their doctor, it is a hard one. I know that. And I know what some will say: who are you, and what makes you think you are in a position to say this? I do not care. People are landing in a dangerous zone, and I am here to protect you.

Two conversations I had. One was with a doctor who gives this advice. I asked him straight out: are you aware of euglycemic DKA? He was. He agreed with me that exercise mode is not a safe option on a fast. And he still tells his patients to use it. For whatever reason, that conversation did not stay with him. So it falls to me to set the record straight. This is my duty.

The other was with a rav who is very well versed in medicine. He was about to rule that people with diabetes may fast on the other fast days of the year, not only Yom Kippur — the reasoning being that we now have pumps that prevent low blood sugars. When I explained euglycemic DKA to him in detail, he told me openly: I have never heard of it. And he retracted.

Not everyone knows about it. And the ones who know do not always think about it when it matters. What made me think about it is that I have watched it happen.

Confidence — and the First-Year Rule

Why does our own Rav not allow a person to fast within the first year of diagnosis? It was never about the number on the screen. It is that a person in the first year has not yet learned whether the system he uses will hold him. He has no long-lasting track record. That is the whole rule, and it applies to everyone, not only to the newly diagnosed — to the person who switched from injections to a pump, and to the person who moved from an old-fashioned pump to a sophisticated one. The rule is not a calendar. It is a track record. On a new system you have less of one, and the question is whether you have enough to know it holds you. The fast is not the day to find out.

Confidence in your system means one specific thing. You can keep your target low enough that you are getting enough insulin, and you can hold it there without the fear of going low. Both at once. A person who is only safe from lows because his target is high does not have confidence. A person who gets plenty of insulin but drops too low does not have it either. Confidence is when the system lets you have both, and a year on it either proves that or it does not.

People write to me: Rabbi, this is complicated, just tell me what to do. I understand the request. But this is the honest answer. A condition of being allowed to fast is that you understand diabetes, you understand your body, and you understand your device. If you read this post and could not explain to someone else why a higher target is the dangerous side, then this year, fasting is not for you. That is not a punishment. It is the same rule as the first year: you fast when you know what holds you.

If you have the confidence, you can try to fast. If you do not have it, the only tool left to you is a higher target — which means less insulin, which is the DKA risk. And that is not a tool you are allowed to pick up. So: fasting is not meant for you. Not “fast carefully.” Not this year.

There is still time to build some of it, and here the pumps differ. On Tandem and on twiist the algorithm works from your own settings — basal, ratios, correction factor — so getting them right is worth doing, and a settings review with your team now is part of passing the test. Not because either system does a poor job. Because a fast day has no slack: on an ordinary day the system has hours to catch up to a setting that is a little off, and on a fast day the catching up is done with insulin you cannot afford to lose. On Medtronic, Omnipod and the iLet, the target is the setting, as detailed below; the algorithm builds its own picture and there is nothing else to tune. There, the rest of the confidence comes from the year behind you.

What Changed for Me This Year

Two things happened over the last year that made me think all of this through again. I ran two systems I had never used before: twiist, and the Medtronic 780G. Both showed me a level of stability I had not seen. I would call it confidence in the pump, and I mean the word the way I defined it above.

They gave me the confidence, year-round, to push my target as low as each one allows. On the Medtronic that is 100, the lowest it will take. On twiist I keep it lower still — my correction range is set at 90 to 105. And I do it without being afraid of going low. The lows I did have on twiist came from one place: I miscalculated a meal.

A miscalculated meal cannot happen on a fast day. So the one cause of lows I actually have is gone for twenty-five hours — and with it goes the reason to raise the target. Having that confidence in a pump is what makes a safer fast: keep the target close to what it is all year, get the insulin, and not be scared of the low. (The first night is the exception, and last year’s post on first-night lows covers why.)

These systems deserve the confidence. I am speaking from my own year on them, not from a study, and your year may look different — which is the whole point of this post.

So in terms of confidence during the year, I have it in both. That is one category, and both pumps earned their place in it.

The fast day is a second question, and there I have a choice to make, because I have both pumps at home. I would choose twiist. The 780G learns — its basal is built from what it saw over the previous days, and the days before Yom Kippur are exactly the ones I do not want it learning from: the Erev Yom Kippur meal, the Yom Tov eating. It corrects every five minutes, which is why it held me all year, but it walks into the fast carrying a picture of a body that eats. twiist does not learn across days. It looks at the last half hour, predicts the next six, and recomputes from my settings every five minutes. It has no memory of Tuesday. On a fast day, that is the one I want.

The Target Question

If only we could keep the target exactly where it is the whole year, we would have an ample supply of insulin.

But numbers do not run in a straight line. They wobble, they drift, they dip for no reason you can name. On an ordinary day the dip runs into the next meal. On a fast day there is no next meal, and a low target with nothing coming in is exactly where a low happens.

That is where raising the target on a fast came from. It was never a foolish idea. It was the answer to a real problem. But the answer may cost more than the problem does, and this post is about what to do instead.

The Target: 120

It is hard to give an exact number for a safe target. But two things I can say without hesitation. A target of 140 to 160 is too high, full stop — it takes off too much insulin. And 120 would be best for most people. It is comfortably above a low, and low enough to keep the insulin flowing.

Every system reaches it, and on most of them it is the default, not an aspiration:

Omnipod 5: set the target to 120. Medtronic 780G: 120. iLet: “Usual,” which is 120. twiist: my own is set at 90 to 105 all year, and for the fast I plan to raise it a tiny bit, to 110–120. This will be the first fast I go into feeling safe and secure, bez"H.

Exercise mode, activity mode, a high temp target — whatever your pump calls it — on any system, during the fast: no. It cuts insulin to get you to 140 to 160, and that is the DKA side. On Tandem exercise mode does something worse than raise the aim: it moves the point where the pump starts cutting insulin from 112 up to 140. On a fast day, where the numbers are drifting down anyway, that is the pump taking insulin off at 140. And it will not put any insulin back until it predicts 160. Between 140 and 160 it does nothing; below 140 it cuts.

The Choices for the Tandem

Tandem is different from the other pumps in that it does not aim at a target. There is no target you can set that affects the automation — not in the profile, where the Target BG feeds only the bolus calculator, and not through a temp basal, which shifts what the pump starts from but not what it aims at. Whichever mode you are in, that mode’s numbers are the numbers. It holds a range, and it does that in one of two ways, and each way has one downside.

Originally we advised Sleep Mode for the fast, and the reason was a good one. Sleep Mode pushes basal insulin in as soon as it predicts you above 120, to bring you back into its range, where Regular Mode waits until 160. On the DKA side, that makes it the mode that gives the most insulin, and fewer people landed in DKA on it. But the feedback that came back was too many lows. Sleep Mode cuts harder on the way down and tends to reach a lower number. For some people their own history says it drops them, and they will not be confident in it.

So the three modes are an insulin ladder. Sleep gives the most, Regular the middle, Exercise the least. On the insulin-cutting side, Sleep and Regular are the same — both leave you alone until 112 — so Sleep Mode’s risk is the low, not DKA. Exercise is the only one that cuts early, and it is off the table.

Regular Mode has a second tool: an automatic correction bolus, delivered when the pump predicts a rise past 180. On an ordinary day that is the feature. On a fast it might be a problem, because on a fast day a bolus can work stronger than it usually does. The same correction that brings you down gently on Tuesday can bring you down too far on Yom Kippur — and a bolus cannot be taken back, where a basal change is gradual and can be walked back five minutes later.

The middle way, which is what last year’s guidance settled on, is Regular Mode with the correction factor softened. Instead of 1:40, set 1:60. Then if the pump does predict 180 and delivers a correction, it is a small one, easy to tolerate on a fast; and when the prediction is only mid-range, the pump does not bolus at all, it only adjusts the basal, and the descent is gradual.

And if your year shows you something else — if you run Sleep Mode around the clock and it holds you — then you have passed the test on it, and it is yours to use. That is the whole point of the test: it is answered by your history, not by me.

The same can apply on twiist, one step over: its insulin sensitivity is a setting the algorithm reads, so review it with your team the way a Tandem user reviews the correction factor.

The First Night

The only time we ever allowed exercise mode was the first night, and the reason still holds. After the last meal there is plenty of insulin on board, and the first hours are where most people who break a fast break it. A higher target there is not cutting insulin you need. It is easing off insulin you already have.

But not all night. Until four to six hours after the last bolus — that is the window the meal insulin is actually working in. After that, the risk is the fast, and a higher target is doing the wrong job.

And every system can be told to end it by itself. Omnipod’s Activity feature takes a duration. Medtronic’s temp target takes a duration. The iLet’s “Higher” target can be set as temporary with an automatic end. Tandem has no higher target to set, so there the tool is the temp basal in Control-IQ+, which runs on a timer and works in whichever mode you are confident in — Exercise Mode on Tandem has no timer, which is one more reason it is the wrong tool. Set it for four to six hours and let it expire. Nobody has to touch a pump at three in the morning, and nobody wakes up at eight still in a mode that is cutting insulin at 140.

One honest limit. None of these levers reaches the insulin already on board from the last meal. The bolus is spent; a higher target or a reduced basal changes only what the pump adds after it. So these are real tools, and modest ones. The bigger lever is the meal itself, and that is the first-night post linked above.

Two Edges

Low body weight raises the risk of DKA, so a thin young teenager should not attempt a full fast. That is not new, and it is not negotiable. They will need strong coaching from their parents to understand why. Explain to them that eating less than a shiur is still, halachically, fasting. And most of all, tell them that a year or two or three down the road, bez"H, they will be able to do a full fast.

Pregnancy raises the DKA risk further still. There the answer is simply no.

And anyone taking Jardiance or another SGLT2 inhibitor has a separate conversation to have with the doctor who prescribed it — Part 3 has it.

· · ·

When people fast, what they see in front of their eyes is the blood sugar. Will I go low. That is the concern that fills the day.

There is a different question that belongs in front of it. Am I doing enough to stay away from DKA? If the answer is yes — if your target is where it should be, your system holds you there, and you know what you would do the moment it did not — then you can try to fast. If the answer is no, you cannot. Not this year.

Review your own plan with your doctor before the fast.

Wishing you a kesivah vechasimah tovah, and an easy fast.

— Rabbi Hirsch Meisels