Continuous glucose monitoring replaced about half of inpatient fingersticks in a small protocol study

Design

Single-site prospective evaluation of a hybrid sensor protocol, paired sensor and fingerstick values

Population

54 noncritically ill inpatients (62 visits) on general medicine or with renal impairment

Primary outcome

Sensor accuracy (mean absolute relative difference), hypoglycaemic events, fingersticks avoided

Effect

MARD 11.9% (10.0% after protocol exclusions); 53% of fingerstick tests replaced; 24 events below 70 mg/dL

A hospital team evaluated a hybrid continuous glucose monitoring protocol in noncritically ill inpatients, extending earlier work on patients admitted with infections to a general medicine and renally compromised population. The protocol aimed to replace most point-of-care glucose tests when making treatment decisions.

Fifty-four patients across 62 admissions were evaluated. Across 732 paired sensor and fingerstick values, the mean absolute relative difference was 11.9% (SD 10.8). After removing 258 pairs under protocol rules, it was 10.0% (SD 8.1). Twenty-four hypoglycaemic events below 70 mg/dL were recorded, and the sensors replaced 53% of fingerstick tests, short of the 75% originally aimed for.

This was a small, single-site accuracy study. It did not compare outcomes such as hypoglycaemia rates with fingerstick-only care, and the abstract does not describe how many events were missed by the sensor. Accuracy is known to be weaker at low glucose and in people with poor perfusion, so a protocol with rules for when to confirm by fingerstick is essential.

• Keep a fingerstick check for symptomatic patients and any reading that does not fit the clinical picture.
• Confirm low or rapidly changing sensor readings before treating.
• Do not use sensors in unstable or poorly perfused patients without a local protocol.
• If your hospital considers this, agree rules for when a sensor value can and cannot guide insulin dosing.

Why it matters

It asks whether sensors can cut the nursing burden of ward glucose checks without making dosing less safe.

Don't overread it

This was a small accuracy study with no comparison of clinical outcomes against fingerstick monitoring.

The statistics, in plain English

A mean absolute relative difference of about 10% means the sensor value differed from the fingerstick by about a tenth on average. That is acceptable for trends but can matter at low glucose, where a small absolute difference is a large proportion.

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.

Get Daily Doctor — free

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