A real thermometer for a real fire — and not the fire itself
hs-CRP predicts mortality reliably and causes none of it.
Evidence, by endpoint
| Endpoint | Grade | Citation |
|---|---|---|
| Coronary heart disease (prediction) | B | Emerging Risk Factors Collaboration 2010, Lancet 54 studies, ~160,000 people. Per 1-SD higher log CRP, HR ~1.37 for CHD, falling to ~1.23 after adjustment for conventional risk factors. Also predicts non-vascular mortality. |
| CRP as a causal agent | X | CRP CHD Genetics Collaboration 2011, BMJ; Zacho et al. 2008, NEJM Mendelian randomization on CRP-locus variants shows no increase in CHD risk from genetically elevated CRP. |
| Cardiovascular events via IL-1β inhibition | A | Ridker et al. 2017, NEJM (CANTOS) 10,061 post-MI patients. Canakinumab cut major cardiovascular events 15% (HR 0.85) with no change in lipids. |
| All-cause mortality via anti-inflammatory therapy | X | CANTOS (Ridker et al. 2017, NEJM) All-cause mortality was flat. Fatal infections rose significantly in the treatment arm. FDA rejected the cardiovascular indication. |
What it predicts
hs-CRP sits downstream of IL-6, which sits downstream of IL-1β. The Emerging Risk Factors Collaboration (2010, Lancet; 54 studies, ~160,000 people) found that per 1-SD higher log CRP, the hazard ratio for coronary heart disease is about 1.37, falling to about 1.23 after adjustment for conventional risk factors. It also predicts non-vascular mortality.
Why prediction is not causation here
Mendelian randomization on CRP-locus variants (CRP CHD Genetics Collaboration 2011, BMJ; Zacho et al. 2008, NEJM) shows no increase in CHD risk from genetically elevated CRP. By contrast, Mendelian randomization on IL-6 receptor variants is not null — those variants do associate with lower coronary heart disease risk, so IL-6 signaling appears causal where CRP itself does not.
What happens when you intervene upstream
CANTOS (Ridker et al. 2017, NEJM) randomized 10,061 post-MI patients to canakinumab, an IL-1β inhibitor. It cut major cardiovascular events by 15% (HR 0.85) with no change in lipids — but all-cause mortality was flat, fatal infections rose significantly in the treatment arm, and the FDA rejected the cardiovascular indication. COLCOT (Tardif et al. 2019, NEJM) and LoDoCo2 (Nidorf et al. 2020, NEJM) both found low-dose colchicine reduced cardiovascular events; LoDoCo2 showed a non-significant increase in non-cardiovascular death.
What this means for the number on your lab report
hs-CRP is a reliable thermometer for a real fire elsewhere in the body, not the fire itself, and not a validated target for treatment on its own.
What argues against this
Incremental value over an established risk model is very small — adding CRP moves the C-statistic by roughly 0.00–0.01. The Emerging Risk Factors Collaboration's 2012 NEJM analysis estimated it would prevent about one additional event per 400–500 people screened over ten years. Roughly a third of circulating IL-6 comes from adipose tissue, so once waist-to-height is known, CRP is largely restating it.
Gotchas
- Values above 10 mg/L are uninterpretable for cardiovascular risk — that indicates acute illness. Repeat in two weeks.
- Within-person variability is high enough that two measurements are the standard recommendation.
- Runs higher in women, with estrogen or oral contraceptives, with age, and — well documented and awkward for fixed cutpoints — higher in Black populations at equivalent risk.
- AHA/CDC 2003 cutpoints: below 1.0 mg/L low, 1.0–3.0 average, above 3.0 high.
Educational content only — not medical advice, diagnosis, or treatment. Not a medical device; not FDA evaluated. Consult a qualified healthcare professional about your own health, and call emergency services for urgent symptoms. Full medical disclaimer →