Accuracy Lives At The Intersection Of The Right Time And The Right Place.

The day the GP’s “mild” COPD turned “severe”

I still remember the phone call.
“Sanjeev, your clinic letter says Mr. Barnes has severe COPD, but our spirometry last month called it mild. Did we miss something?”
My GP colleague sounded worried—rightly so. Mr. Barnes had already booked his bucket-list trip to the Lake District on the strength of that “mild” label.

We repeated the test in our pulmonary-function lab at Wishton Hospital. Same man, same effort, same technique… yet his FEV1 dropped from 68 % predicted to 43 %. One cup of coffee later, the holiday was cancelled and the inhalers upgraded.

That moment in 2011 was the spark for the small study we eventually took to the European Respiratory Society congress the following year. If community and hospital numbers don’t match, whose numbers do we trust?

Why bother testing in the community anyway?

COPD is like high blood pressure: you can’t spot it without a cuff. Spirometry is our cuff, but hospitals are busy, parking is a nightmare, and half the patients never turn up. If GPs could do reliable spirometry on the spot, we’d catch thousands of missing cases earlier and spare patients years of breathless mornings.

The Dutch had already hinted that community readings run “a tad high,” but no one had looked at a chunky UK cohort. So we did what any nosy clinician-researcher does on a rainy Merseyside afternoon: we pulled every abnormal community spirometry done in St Helens and Knowsley during 2010–11 and invited the owners to repeat it in our lab. Simple, retrospective, zero extra cost—ethics committees love that phrase.

405 lungs, one spreadsheet

Who were they?

• Age 22–78, average 54 (basically my Saturday football squad plus dads)
• 46 % still smoking, 41 % ex-smokers, 13 % “never touched the stuff”
• 82 % already labelled “obstructive” by their GP
• 45 % moderate, 32 % severe, 11 % very-severe COPD (GOLD stage speak)

What did we compare?

Same patient, same week, two machines:

1. GP surgery (hand-held spirometer, usually the nurse who also does smear tests)
2. Hospital lab (body-box, three experienced technicians who live on caffeine)

And the headline?

Mean FEV1 in the community: 1.52 L
Mean FEV1 in the lab: 1.49 L

A meagre 30 mL difference—about a tablespoon of air—yet statistically significant (p < 0.01). In plain English: community numbers were reliably optimistic. No malice, no broken kit, just the way it is.

Three nuggets nobody puts in the abstract

1. The “good-day” effect
   Patients try harder when they know a white-coated technician is watching. In the lab we give verbal encouragement—“keep going, keep going, almost there…”—like a bargain-basement spin-class instructor. That extra second of blast empties the lungs further and drops the FEV1. In the GP room it’s usually “blow out… thanks, you can stop.”

2. The filter fudge
   Most community devices use a heated sensor; hospital booths are ambient. Warm air is less dense, so the sensor over-reads by ~2 %. Engineers yawn at this; clinicians panic when the patient’s holiday hangs on it.

3. The selection skew
   GPs only refer “abnormal” tests to us. Normals never see a hospital. So our study was full of people who were already flagged, amplifying tiny differences. If we repeated the exercise on 1,000 random shoppers at Tesco, the gap would probably shrink.

So, can we trust community spirometry?

Yes—but treat it like a screening thermometer, not a verdict.

A community FEV1 that looks “mild” needs a lab confirmation before you tell someone to trek the Inca Trail.

Likewise, a “severe” community result is probably at least severe; it rarely gets better in hospital.

Practical rule I now teach trainees:

• FEV1 ≥ 60 % predicted in the community → repeat in lab before labeling mild
• FEV1 < 40 % predicted in the community → accept and act; the number won’t bounce back up

Everything in between? Use the lower number for counselling; patients forgive pessimism, they sue for optimism.

What changed in St Helens & Knowsley after 2012?

1. We stopped asking GPs to interpret flow-volume loops—just the raw FEV1/FVC ratio. Interpretation lives with respiratory nurses now.
2. We bought the same model spirometer for every surgery (calibration parties every St. David’s day).
3. We created a “traffic-light” letter: green if numbers match, amber if borderline, red if mismatch > 200 mL. Red means automatic hospital repeat within four weeks.

Result: mismatched severe cases fell from 11 % to 3 % in two years. Mr. Barnes wouldn’t have packed his rucksack so early.

Quick patient takeaway

If your GP says your COPD is “mild” on the basis of a surgery spirometry, ask for a copy of the numbers and, if you still smoke, politely request a hospital repeat before you plan that marathon. Your lungs—and travel insurance—will thank you.

Parting shot

Data is only as good as the context it’s breathed in. Community spirometry is a brilliant front-door, but every front-door needs a sturdy frame. Build the frame—training, calibration, feedback—and the numbers behave. Until then, keep a tablespoon of scepticism handy.

And if you ever bump into Mr. Barnes in Ambleside, tell him the view’s still gorgeous from the lowland trails.