- Design
- cluster-randomised diagnostic trial, 2:1 allocation, in Dutch general practice
- Population
- 827 patients presenting to primary care with acute chest pain, 740 assessed with the decision rule
- Primary outcome
- diagnostic performance for ruling out acute coronary syndrome and myocardial infarction
- Effect
- sensitivity 98.3% (95% CI 90.8-100), negative predictive value 99.7% (98.0-100); specificity 49.1%
POB-HELP was a cluster-randomised diagnostic trial across Dutch general practices, allocating 2:1, in which general practitioners assessing acute chest pain applied a rule combining the Marburg Heart Score with a point-of-care high-sensitivity troponin I. Eight hundred and twenty-seven patients were included, 740 assessed with the rule.
The rule missed one acute coronary syndrome: sensitivity 98.3% (95% CI 90.8-100) and negative predictive value 99.7% (98.0-100). For myocardial infarction alone, both sensitivity and negative predictive value were 100%. Specificity was 49.1% and positive predictive value 14.1% - it rules out well and rules in badly, which is the correct trade for this decision.
The negative finding matters as much. Referral rates did not differ from usual care, 54.6% against 47.1% (p=0.39), and the control arm had only 87 patients, far too few to detect a difference. So this trial shows the rule is safe, not that it reduces referral. Before importing it, note what it assumes: a point-of-care high-sensitivity troponin assay in the clinic, and a Dutch population with that population's prevalence of coronary disease. Indian primary care largely lacks the first, and the trade-off shifts with prevalence. Where a point-of-care troponin does exist, this is the trial to point at for how to use it - inside a clinical score, not instead of one.
- Use a point-of-care troponin inside a validated score, never as a standalone rule-out.
- The rule excludes well and includes poorly - a positive result means refer, not diagnose.
- Referral rates did not fall, and the control arm was too small to test that properly.
- The assay must be high-sensitivity; an older point-of-care troponin does not carry this performance.
- Predictive values shift with local prevalence of coronary disease - do not import the numbers unchanged.
Why it matters
It puts a measured number on the decision general practitioners make most anxiously and most often, and the number is reassuring.
Don't overread it
A safety finding, not an efficiency one - referral rates did not fall, and the control arm was too small to show whether they could.
The statistics, in plain English
A negative predictive value of 99.7% is the number that matters for a rule-out, and its confidence interval reaching 98.0% means that in the worst case compatible with these data, two in a hundred cleared patients would still have an event. Note also that predictive values depend on how common disease is: in a population with more coronary disease, the same test's negative predictive value falls. The 100% sensitivity for infarction rests on a small number of events, which is why its interval starts at 92.6%.
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.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for family medicine, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free