Three Stopping Points: What Positive Predictive Value Actually Measures

PPV1, PPV2, and PPV3 ask a single question about patients with a positive breast imaging finding: how many produced a tissue diagnosis of cancer within one year?

Look at how the ACR writes the three formulas and the answer is sitting in the notation [1].

PPV1 = TP / (TP + FP1)
PPV2 = TP / (TP + FP2)
PPV3 = TP / (TP + FP3)

TP carries no subscript. FP carries one in every formula. Same numerator, three different false positive counts, because each measure asks about a different set of examinations.

Where each one stops to ask

PPV1 counts positive screening examinations, usually BI-RADS 0, 3, 4, and 5 [1]. Every one of those is a recall. Category 0 returns for additional imaging. Category 3 returns at a four to six month interval for surveillance. Categories 4 and 5 return immediately for an interventional procedure. Acceptable range 3% to 8% [1].

PPV2 counts examinations recommended for tissue diagnosis, categories 4 and 5. Acceptable range 20% to 40% [1]. Reported for a screening practice it tracks the patient forward from her screening examination, so a BI-RADS 0 followed by a diagnostic recommendation for the same lesion belongs here [1].

PPV3 counts biopsies performed. The ACR places it in the diagnostic table, at 20% to 45% for workup of abnormal screening [1]. It is also called the biopsy yield of malignancy.

The numerator finds you. The denominators do not.

Cancers arrive on their own. Pathology comes back, registries match, and a tissue diagnosis is difficult to overlook. Most facilities can produce their cancer count.

The false positives have to be built, and there are three of them. A false positive is a positive finding whose workup was completed and resolved as negative. That means an answer exists. A patient nobody heard from again is not a false positive. She is unresolved, and an unresolved patient cannot be audited at all.

So FP1 requires a known outcome for every positive screening finding, across categories that come due on three different clocks. FP2 requires knowing which tissue recommendations resolved negative, including recommendations a patient acted on somewhere else. FP3 requires the benign pathology reports themselves. That denominator is built from biopsies with a record in the pathology file [2].

Who owns the number

PPV1 belongs largely to the screening interpreter. PPV2 does not belong to her alone, and the ACR says so directly. Reported at screening it measures screening practice in general, because the tissue recommendation may come from a different physician at the diagnostic examination [1]. By PPV3 the number has passed through scheduling, navigation, the procedure, and pathology reporting.

Seeing it move

The MammoToolbox PPV Calculator takes true positive and false positive counts, the shape your audit already reports, and builds each denominator from them. Three scenarios show how one change moves a single measure and leaves the others still.

The arithmetic takes a second. The correlation behind it takes a year.

Which leaves the question the calculator cannot answer for you. You can see exactly what the formula needs. How would your facility produce those three false positive counts?

---

References

1. ACR BI-RADS Atlas, Breast Imaging Reporting and Data System, 5th edition, Follow-up and Outcome Monitoring. American College of Radiology; 2013. Positive predictive value definitions at item 12. Acceptable ranges for screening at Table 7 and for diagnostic mammography at Table 8.
2. Breast Cancer Surveillance Consortium. BCSC Data Definitions, version 3; 2020. Defines FP2 and FP3, and documents that the PPV3 denominator is built from biopsies with a record in the pathology file.

About the Author

Richard D. Lippert Jr. is the founder and CEO of Mammologix LLC. He has more than thirty years of experience in breast imaging program operations and is clinically trained in radiologic technology and mammography. He has tracked FDA MQSA National Statistics monthly since December 2002.