DailyDoctor Archive Specialties Get app
Back to the 21 August 2026 edition

Pearl · 07 of 07

Read the CGM download by day of the week, not just as an average

Before changing a dose on the strength of a CGM download, split the report into weekdays and weekend days — a pattern confined to two days a week needs a behavioural fix, not a bigger basal dose.

Most reviews of an ambulatory glucose profile collapse two weeks into a single 24-hour curve. That is the right place to start, but it hides a pattern almost every clinician has seen and few write down: control that is steadier on working days and drifts at weekends, with the evening and overnight hours carrying most of the trouble.

It matters because two quite different fortnights can produce the same average time in range. If the shortfall is concentrated on Saturday and Sunday evenings, the fix is behavioural — meal timing, alcohol, an unstructured day — and not a change in basal dose. Increase the basal to cover a weekend pattern and you buy hypoglycaemia on Tuesday.

So split the report before you touch the regimen. Look at weekdays and weekend days separately, and look at the evening and overnight window separately from the daytime. Then ask the patient to describe what a Saturday actually looks like. The answer usually explains the curve faster than any dose calculation does.

  • Ask for at least 14 days of data with 70% or more sensor wear before drawing any conclusion from a pattern.
  • Compare weekdays and weekend days separately before changing a dose.
  • Look at the 6 pm to 6 am window on its own — it is where most of the lost time in range usually sits.
  • Ask the patient to walk you through a typical weekend day, including meal times and alcohol.
  • If a pattern is confined to two days a week, change behaviour or bolus timing, not the basal dose.

The statistics, in plain English

Fourteen days of wear with at least 70% sensor coverage is the usual minimum for an ambulatory glucose profile to represent longer-term control. Splitting that period by day of week leaves only four or five days in each arm — enough to spot a consistent pattern, not enough to trust a small difference. If weekday and weekend curves differ by a couple of percentage points of time in range, treat it as noise. If the weekend curve sits visibly lower and does so across two separate sensor periods, it is real.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

QR code to install Daily Doctor
Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.

type2insulinglp1obesitylipidspregnancysglt2cgmtechnologytype1

Tomorrow morning, before your first patient

One edition a day for diabetes & endocrinology, written by the desk, every claim tied to its paper. Six minutes.

Get the app — free
Daily Doctor All 27 specialties, every morning. Free.
Get the app