Most routine blood tests showed no meaningful change with fasting

Design

Retrospective single-centre cross-sectional study with within-person comparisons

Population

101,148 adult outpatients; 9.76 million results across 372 tests

Primary outcome

Differences by fasting duration exceeding analytical performance specifications

Effect

<8 h v longer: glucose +10.8%, triglycerides +20.1%, GGT +14.8%, lipase +36.4%; none between 8–12 h and >12 h

A single-centre study in JAMA Internal Medicine (28 September) analysed 9.8 million results from 101,148 adult outpatients at a Korean university hospital, across 372 tests, grouped by hours since the last meal: under 8, 8 to 12, and over 12.

Comparing groups flagged 27 tests with differences larger than accepted analytical error, but only 10 held up when the same patients were compared with themselves on different visits — the rest reflected who fasts rather than fasting itself. The consistent effects of a short fast were higher glucose (about 11%), triglycerides (20%), GGT (15%) and lipase (36%). Between an 8–12 hour fast and a longer one, no test differed meaningfully.

For ward and clinic practice this supports ordering fasting only where it changes interpretation — fasting glucose, triglycerides when the level matters, and lipase with a recent meal in mind — and not keeping inpatients nil by mouth overnight for a routine panel. There is also no reason to insist on more than 12 hours.

• Consider dropping the fasting requirement for routine renal, liver and blood count panels.
• Glucose, triglycerides, GGT and lipase read higher within eight hours of a meal; note the timing.
• An 8–12 hour fast is enough where fasting matters; longer adds nothing.
• Avoid keeping inpatients nil by mouth solely for morning bloods.

Why it matters

It removes a common reason for delayed breakfasts, missed doses and repeat visits for tests that do not need it.

Don't overread it

This is one Korean hospital's outpatient data; it does not address specific diagnostic protocols such as oral glucose tolerance tests.

The statistics, in plain English

The comparison standard was total allowable error — how much a result can vary before it matters, based on normal biological variation. The within-person analysis compares each patient against themselves, which removes most confounding; that step cut the list from 27 tests to 10.

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.