Faster Is Not Always Better: Why Mammography Audit Outcomes Need Time to Mature

MammoComply · Opinion · Mammologix · Breast Imaging Operations since 1995

A case sitting in your false-positive column right now might be a true positive that nobody has diagnosed yet. That isn't a data-entry mistake. It's how the math works. Some of the most-watched numbers in a mammography audit are built on cancer outcomes that haven't finished happening, and a value you pull too early can quietly flip once they do.

Editorial Disclosure: This piece reflects Mammologix's operational and compliance-support perspective. It is intended for quality-improvement and audit-readiness discussion, not as medical, legal, or regulatory advice.

---

That puts a fair amount of audit reporting in an awkward spot. Administrators want dashboards that refresh now, radiologists want feedback while the cases are fresh, and quality teams want to see the program moving the right way. All reasonable. We support timely reporting, and we build it. But some mammography audit measures shouldn't be handed over as if they're final, because by definition they aren't yet.

The metrics that can't be final on day one

A recall rate you can calculate almost immediately. It rests on the screening interpretation and whether the patient was called back. Nothing about it waits on a diagnosis.

Cancer detection rate is a different animal. So are positive predictive value and its variants (PPV1, PPV2, PPV3), sensitivity, specificity, and the counts underneath them: true positives, false positives, true negatives, false negatives. Every one of those depends on knowing whether breast cancer was diagnosed inside a defined follow-up window. Until that window closes, you're working with a number that's still settling.

The one-year window isn't a reporting habit, it's the definition

This is the part that gets missed. The follow-up period isn't a convention laid on top of the metric. It's built into what the metric means.

Federal regulation defines positive predictive value as the share of patients with positive mammograms who are diagnosed with breast cancer within one year of the exam. It defines screening cancer detection rate using patients diagnosed within one year of the initial screening mammogram, expressed per 1,000 patients. The benchmark literature uses the same one-year frame. In the Breast Cancer Surveillance Consortium screening work, cancer follow-up ran 12 months past the mammogram, and exams without a full 12 months of follow-up were excluded so the diagnosis-and-reporting window could close. A true positive there is a positive screening read followed by invasive cancer or ductal carcinoma in situ within 12 months. A false positive is a positive read with no cancer diagnosed in that same window.

Read those definitions closely and the problem shows up on its own. A case that looks like a false positive today becomes a true positive if cancer surfaces inside the year. A true negative today becomes a false negative the same way. The numerator moves. The denominator moves. Calculate before the window closes and you've measured something that's still changing under you.

Why this hits PPV, CDR, sensitivity, and specificity specifically

Every outcome-dependent measure is built from the true and false, positive and negative classifications. When those classifications are still in flux, so is the measure.

Metric	Why an early number can mislead
Cancer detection rate	Needs to know which screening exams led to a cancer diagnosis inside the defined window.
PPV	Needs to know which positive mammograms ended in a cancer diagnosis within one year.
PPV1	Rests on cancers found after an abnormal screening interpretation.
PPV2	Rests on cancers found after a biopsy recommendation.
PPV3	Rests on cancers found after a biopsy is performed.
Sensitivity	Needs both true positives and false negatives.
Specificity	Needs both true negatives and false positives.
False-negative rate	Needs the cancers that show up later, after a negative exam.

The BCSC reports the full set: cancer detection rate, sensitivity, specificity, false-negative rate, PPV1, PPV2, and PPV3. It defines PPV1 by abnormal interpretation, PPV2 by recommendation for tissue diagnosis, and PPV3 by biopsy performed. So when a facility asks for the "current" CDR or the "current" PPV, the honest answer is that the figure may be useful for operations but isn't final. It should carry that label.

The benchmark comparison trap

Here's where an early number does real damage. Facilities want to see how they stack up against national benchmarks, which is a good instinct. Benchmarking points teams toward the right problems.

But national benchmark data are built on standardized definitions, defined audit periods, and matured follow-up. The BCSC describes its screening benchmark tables as drawn from a large slice of U.S. screening practice, built on millions of digital mammography and tomosynthesis examinations. Set a facility's three-month or six-month outcome report next to that, and you're not comparing like with like. You're comparing a number that's still moving against one that has stopped.

Interim reports still earn their keep. They flag operational drift early, and that's worth having. They just aren't audit outcomes, and treating them as if they are is where a team talks itself into a problem that isn't there, or misses one that is.

What this means for MQSA readiness after September 10, 2024

The amended MQSA regulations took effect September 10, 2024, with changes to the medical outcomes audit, reporting, and recordkeeping. The amended rule requires each facility to run a medical outcomes audit, follow up positive assessments, correlate pathology with the interpreting physician's findings, and review outcome data, individually and collectively for interpreting physicians. At minimum it requires positive predictive value, cancer detection rate, and recall rate, for each interpreting physician and for the facility as a whole.

In our experience, facilities should be ready to show those procedures and metrics were performed for the inspection period in question. When the request involves outcome-dependent metrics, choosing an audit period far enough back for the one-year window to mature tends to produce results that hold up better, both clinically and under review. The point isn't to slow quality work down. It's to keep from reporting precision you don't actually have yet.

What we recommend

Plainly: outcome-dependent mammography audit measures shouldn't be treated as final until the one-year cancer outcome window has matured and the available pathology, diagnosis, and follow-up information have been reconciled. When you do calculate them early, label them.

A workable disclaimer reads:

Outcome-dependent mammography audit measures, including CDR, PPV, PPV1, PPV2, PPV3, sensitivity, specificity, TP, FP, TN, and FN, may change as pathology, cancer diagnosis, and follow-up information becomes available. Values calculated before the one-year outcome window has matured should be treated as preliminary and should not be compared directly with national benchmarks without qualification.

That sits with how we report generally: audit reports are built from client-supplied clinical, operational, and administrative data, findings reflect what's available when the report is prepared, and conclusions can change if new or updated data come in.

The number worth waiting for

Real-time data earns its place on the dashboard. A matured outcome number is the one that can carry weight in an MQSA review or a benchmark comparison, because it's the one that has stopped changing. A facility should know which of its numbers are preliminary, which are mature, and which are fit for which use.

In a mammography audit, the most responsible number is often the one that's had enough time to become true.

---

References

1. Electronic Code of Federal Regulations. 21 CFR §900.12: Quality standards. Accessed June 10, 2026.

2. US Food and Drug Administration. Important information: final rule to amend the Mammography Quality Standards Act (MQSA). Accessed June 10, 2026.

3. US Food and Drug Administration. Mammography Quality Standards Act and Regulation Amendments: Small Entity Compliance Guide: Guidance for Industry and Food and Drug Administration Staff. Published August 2024. Accessed June 10, 2026.

4. Food and Drug Administration, Department of Health and Human Services. Mammography Quality Standards Act. Final rule. Fed Regist. 2023;88(47):15126–15172.

5. Rosenberg RD, Yankaskas BC, Abraham LA, et al. Performance benchmarks for screening mammography. Radiology. 2006;241(1):55–66. doi:10.1148/radiol.2411051504

6. Breast Cancer Surveillance Consortium. Screening performance benchmarks. Accessed June 10, 2026. bcsc-research.org

7. Breast Cancer Surveillance Consortium. Performance measures. Accessed June 10, 2026. bcsc-research.org

8. Lee CI, Abraham L, Miglioretti DL, et al; Breast Cancer Surveillance Consortium. National performance benchmarks for screening digital breast tomosynthesis: update from the Breast Cancer Surveillance Consortium. Radiology. 2023;307(4):e222499. doi:10.1148/radiol.222499

9. Lehman CD, Arao RF, Sprague BL, et al. National performance benchmarks for modern screening digital mammography: update from the Breast Cancer Surveillance Consortium. Radiology. 2017;283(1):49–58. doi:10.1148/radiol.2016161174

10. D'Orsi CJ. The clinically relevant breast imaging audit. J Breast Imaging. 2020;2(1):2–6. doi:10.1093/jbi/wbz080

11. Funaro K, Ataya D, Niell B. Understanding the mammography audit. Radiol Clin North Am. 2021;59(1):41–55. doi:10.1016/j.rcl.2020.09.009

12. American College of Radiology. Breast Imaging Reporting & Data System (BI-RADS). Accessed June 10, 2026. acr.org

13. D'Orsi CJ, Sickles EA, Mendelson EB, Morris EA, eds. ACR BI-RADS Atlas: Breast Imaging Reporting and Data System. 5th ed. American College of Radiology; 2013.

14. Mammologix. Mammologix Report Disclaimer. Internal company document. Mammologix; 2026.