PrevIQ Docs
Open app

Prevalence methodology

How PrevIQ defines and computes a prevalence rate — and why the denominator choice matters.

This page is the definitive explanation of what a PrevIQ prevalence rate is. If a client ever asks "where does this number come from?", everything you need is here.

The definition

Prevalence rate

prevalence_rate = detected / tested

  detected = specimens in which the organism was DETECTED
  tested   = specimens that were actually TESTED for that organism
             (within one syndrome × panel)

A rate is always scoped to a single (syndrome, panel, organism) context, for a single laboratory and date range.

Why the denominator is "all tested specimens"

This is the most important — and most defensible — design choice in PrevIQ.

PrevIQ (all-tested denominator)

The denominator is every specimen tested for the organism, whether the panel came back positive or negative. This answers the real clinical question: "how often do we detect this organism when we look for it?"

Historical (positive-panel denominator)

Some legacy spreadsheets divided by only the panel-positive specimens. That inflates apparent rates and isn't comparable across panels with different positivity.

Important

Because PrevIQ uses the broader denominator, its rates are often lower than older positive-only spreadsheets. This is expected and correct. Every report carries this methodology footnote:

Prevalence rates are computed as the fraction of all tested specimens in which the organism was detected. Some historic analyses may use a positive-panel-only denominator, producing higher apparent rates.

The within-panel rule

Note

Prevalence is computed within a syndrome/panel, never aggregated across panels for the same organism. Strep B in respiratory ≠ Strep B in vaginosis.

If you want a pooled view of an organism across contexts, PrevIQ provides explicit cross-vendor and compare rollups — but those are deliberate, separate calculations, not an accidental blend.

What counts, and what doesn't

Each result cell is interpreted before it touches the math:

Raw result Interpretation In denominator? In numerator?
Detected (incl. High/Med/Low) Positive ✅ Yes ✅ Yes
Not-Detected Negative ✅ Yes ❌ No
N/A Not tested (skip) ❌ No ❌ No
IND, INV, RER Invalid (exclude) ❌ No ❌ No

Note

"Not tested" and "invalid" results are removed from both the numerator and denominator — they never silently deflate or inflate a rate. See Result statuses.

Quantification levels (High / Med / Low) are preserved as metadata, but prevalence is computed on the binary collapse: detected vs. not.

Specimens vs. organism-result cells

One specimen on a 20-target panel produces 20 organism-result cells. To report a meaningful specimen count, PrevIQ estimates it per syndrome as the sum of the maximum tested count within each panel — never the raw sum of cells (which would massively overcount).

Every rate comes with precision

A point estimate alone is misleading at small samples, so every rate ships with:

  • a sample-size tier (A / B / C), and
  • a 95% Wilson score confidence interval.

See Sample-size tiers for both.