The One Insulin Pump Setting That Might Matter More Than You Think

If you or someone you love uses an automated insulin delivery (AID) system, like the Tandem t:slim X2 with Control-IQ, you already know these devices are a game-changer. They adjust insulin automatically every five minutes, correct for highs, and take some of the mental math out of daily diabetes management. 

But here's something that doesn't get talked about enough: how the system is programmed still matters, a lot. A large real-world data analysis points to one setting in particular that may offer an outsized benefit with surprisingly little downside.

The Study: Messer LH, Breton MD. Therapy Settings Associated with Optimal Outcomes for t:slim X2 with Control-IQ Technology in Real-World Clinical Care. Diabetes Technol Ther. 2023 Dec;25(12):877-882. doi: 10.1089/dia.2023.0308. Epub 2023 Oct 25. PMID: 37751154; PMCID: PMC10698772.

What pump settings predict better blood sugar control?

Researchers analyzed data from over 20,000 people using Control-IQ in everyday life. They wanted to know: do pump settings predict better blood sugar control?

Three settings were in the spotlight:

• Correction Factor (CF):  how much insulin is given to bring down a high blood sugar

• Carb-to-Insulin ratio (CR): how much insulin is dosed per gram of carbohydrate eaten

• Basal rate (BR): the steady background insulin delivered throughout the day

All three, when set more aggressively (i.e. more insulin), were linked to more time spent in the healthy blood sugar range (TIR) of 70–180 mg/d (3.9-10 mmol/L). That part isn't shocking. The more insulin coverage generally means tighter control.

The Surprise: Not All "Aggressive" Settings Are Created Equal

Here's where it gets interesting. Pushing basal rates and carb ratios harder did improve time-in-range, but it also came with a modest increase in low blood sugar (hypoglycemia) events. That's the tradeoff most people expect: tighter control, slightly more risk.

The Correction Factor broke that pattern. Strengthening it improved time-in-range by as much as 14% between the most and least aggressive groups, but without a meaningful increase in hypoglycemia. Researchers believe this is because Control-IQ's automatic basal adjustments act as a built-in buffer, quietly reducing background insulin in the hours after a correction dose to prevent overcorrection.

In plain terms: of the three dials you can turn, the Correction Factor seems to be the one that gives you more reward for less risk.

And yet, roughly two-thirds of users in the study had Correction Factor settings that were more conservative than standard clinical guidelines recommend, meaning a lot of people may be leaving easy improvement on the table.

There was a Surprise: 

The single strongest predictor of good outcomes overall wasn't a pump setting at all,  it was simply how often someone manually bolused for meals. More manual boluses, better control. Technology helps, but it doesn't replace the basics.

What This Means If You Use an AID System

• If your Correction Factor setting hasn't been revisited in a while, it might be worth a conversation with your care team. The CF appears to be a relatively low-risk, high-reward adjustment.

• Small, incremental changes (10–20% at a time) with follow-up a few days later remain the safest way to fine-tune settings.

• Bolusing for meals is the still very important lever to pull for better control. 

Automated insulin delivery has already transformed diabetes care. But these findings are a good reminder that the algorithm is only half the equation. 

What do you think? Does this help you?  Let us know.

References:

This post summarizes findings from Messer & Breton (Diabetes Technology & Therapeutics, 2023) and Shah et al. (Diabetes, Obesity and Metabolism, 2026). It is intended for general education and is not a substitute for individualized medical guidance.

Messer LH, Breton MD. Therapy Settings Associated with Optimal Outcomes for t:slim X2 with Control-IQ Technology in Real-World Clinical Care. Diabetes Technol Ther. 2023 Dec;25(12):877-882. doi: 10.1089/dia.2023.0308. Epub 2023 Oct 25. PMID: 37751154; PMCID: PMC10698772.