← Visualization gallery
Provider profiling & quality measurement · Reporting

Provider-Profiling Funnel Plot

A statistical process control chart that plots each institution's observed (indirectly standardized) event rate against its caseload. Hyperbolic 95% and 99.8% control limits narrow with increasing volume; institutions outside the 99.8% band are flagged as statistical outliers warranting review....

Provider-Profiling Funnel Plot: A statistical process control chart that plots each institution's observed (indirectly standardized) event rate against its caseload. Hyperbolic 95% and 99.8% control limits narrow with increasing volume; institutions outside the 99.8% band are flagged as statistical outliers warranting review. Distinct from the meta-analysis funnel plot (slug: funnel-plot), which plots study effect estimates vs their standard errors to detect publication bias.
When to use it

To identify high- or low-performing institutions (hospitals, surgical centres, prescribers) in quality-of-care or outcomes research. Event rates should be indirectly standardized for casemix before plotting. Requires sufficient caseload per institution to have meaningful statistical power.

How to read it

Institutions inside the 95% limits are within expected random variation. Institutions between 95% and 99.8% bands warrant monitoring. Institutions beyond the 99.8% limits are statistical outliers at the three-sigma level — a signal for deeper casemix review, data-quality audit, or quality-improvement investigation. Volume-dependent limits mean that low-volume institutions can have wide bands and are harder to classify.

Worked example

40 hospitals reporting 30-day readmission after elective hip arthroplasty. Overall (target) rate p₀ = 5.2% (0.052). Caseloads range 50–800/yr. Control limits at each caseload n: 95%: p₀ ± 1.96√(p₀(1−p₀)/n); 99.8%: p₀ ± 3.09√(p₀(1−p₀)/n).

40 hospitals; one outlier at n=120, observed rate=11.8% (above 99.8% UL). 99.8% UL at n=120: 0.052+3.09×√(0.052×0.948/120) = 0.052+0.063 = 0.115. 11.8% > 11.5% → flagged. ✓

Result: One hospital (n=120) exceeds the 99.8% upper control limit (observed 11.8% vs limit 11.5%), flagged as a high-rate outlier. No hospitals fall between the 95% and 99.8% bands. The remaining 39 hospitals fall within expected variation.

Produced by

Reference: Gatto NM, Wang SV, Murk W, et al. Visualizations throughout pharmacoepidemiology study planning, implementation, and reporting. Pharmacoepidemiol Drug Saf. 2022;31(11):1140-1152.