Why this project exists
Breast cancer screening is often summarized as a rate: the percentage of women who are up to date with mammography. That rate is important. But by itself, it compresses many different situations into one number. A woman who has no insurance, a woman who has a regular physician but was never referred, and a woman who received an order but never managed to schedule an appointment can all end up on the same side of the screening statistic.
The Mammography Gap Explorer starts from a different place. Instead of asking only how high is the screening rate?, it asks what we can learn about the women who are not represented in that completed-screening number. How many are there? How many are still insured? How many report having a personal healthcare provider or a recent routine checkup? Which groups make up the overdue population? Where do county-level estimates appear lower? And when women themselves are asked why a mammogram did not happen, what do they report?
The goal is not to turn public survey data into a precision it does not have. The goal is to organize existing evidence around the practical problem of screening completion: to separate questions that public data can answer from questions that would require claims, electronic health records, scheduling data, interviews, or new research.
The question we're asking
The central question is simple: when a woman has not had a mammogram within the measured screening interval, what do we know about where the pathway may have broken down?
In an ideal dataset, we could follow the full chain: eligibility, recommendation, order or referral, scheduling, attendance, completion, and follow-up. Public survey data do not observe that chain directly. What they can do is help narrow the possibilities. If most women in the measured gap are insured, lack of insurance cannot be the whole explanation. If many report a regular healthcare provider or a recent routine checkup, some portion of the gap exists among women who are not entirely disconnected from care. If women nationally report reasons related to perceived need, clinician recommendation, cost, time, fear, or follow-through, those answers can help us formulate better questions about what happens between contact with healthcare and completed imaging.
That is why the Explorer is organized as a set of questions rather than a single composite score. Each section is meant to answer one narrow question, preserve its denominator and limitations, and make clear where the evidence stops.
What “the mammography gap” means
In this project, “the mammography gap” is a plain-language label for the population captured by a specific measurement rule. In the Illinois state analysis, the working measure is women ages 40–74 who report no mammogram in the previous two years. The state percentage comes from the 2024 Behavioral Risk Factor Surveillance System (BRFSS) and is applied to a population denominator from the 2024 American Community Survey (ACS).
We sometimes use words such as “overdue,” “unscreened,” or “being missed” to make the analysis readable. Those terms should not be interpreted as individual clinical determinations. BRFSS records whether a mammogram occurred; it does not establish whether the mammogram was screening or diagnostic, whether an individual woman follows a different clinical schedule, or whether she was continuously eligible throughout the full two-year window.
For the Illinois prototype, the state-level gap is the estimated number of women ages 40–74 who reported no mammogram in the past two years. It is a population estimate used to study patterns and possible points of non-completion—not a diagnosis, a patient list, or a claim that every person in the estimate is clinically overdue.
How we use the data
The Explorer deliberately combines several public sources because no single dataset answers all of the questions on the site. Each source is used for the kind of statement it can support, and the different datasets are not treated as if they describe the same individual women.
The statewide population denominator comes from the U.S. Census Bureau’s American Community Survey. State-level mammography, insurance, personal-provider, and routine-checkup measures come from the CDC’s Behavioral Risk Factor Surveillance System. Demographic composition of the estimated state gap is calculated from the same survey microdata using survey weights. County geography uses modeled estimates published by State Cancer Profiles. The reasons women report for not completing mammography come from the National Health Interview Survey (NHIS) and are presented as U.S.-level evidence rather than assigned to Illinois.
When the site shows an approximate number of women rather than a survey percentage, that number is a derived estimate: a survey-weighted proportion is applied to a population anchor. It should not be mistaken for a Census-published count of women with a particular screening history.
Four rules keep the different datasets from becoming one fictional dataset.
Every percentage is tied to a defined population. We do not switch silently between all women, women in the estimated gap, a survey subsample, and county populations.
Survey percentages are estimates from survey data. Approximate headcounts are calculated by applying those percentages to a population denominator and are labeled accordingly.
County modeled estimates are used to explore geographic variation. They are not treated as a geographic allocation of the statewide count of women with no mammogram in two years.
Education, income, race and ethnicity, rurality, insurance, and healthcare contact can describe patterns. They do not by themselves tell us why an individual mammogram did not happen.
What this project cannot tell us
The Explorer is intentionally descriptive. It can estimate the size and composition of a measured screening gap and show where certain patterns appear. It cannot reconstruct an individual patient journey from survey responses, and it cannot identify the cause of non-completion for a particular person, clinic, health system, or county.
Most importantly, the public datasets used here do not tell us whether a clinician recommended mammography, whether an order was placed, whether the patient tried to schedule, whether an appointment was available, whether the appointment was missed or canceled, or whether screening was completed elsewhere and not reflected in a given healthcare system. Those are operational questions that require different data.
Age-based survey measures cannot reproduce an individual screening recommendation, risk profile, symptoms, or shared decision-making.
A demographic or access characteristic can be associated with non-completion without being the reason the mammogram did not occur.
Survey data do not observe recommendation, referral, scheduling, cancellation, no-show, completion, or closed-loop follow-up.
Modeled county screening estimates are useful for orientation, but they are not a direct disaggregation of the statewide overdue estimate.
Why we're talking to experts
A public-data model can be internally consistent and still frame the real-world problem badly. That is why this prototype is being reviewed as it is built. Clinicians can challenge whether a definition matches screening practice. Survey-methods experts can challenge weighting, denominators, uncertainty, and interpretation. Screening-program leaders and patient navigators can point out operational steps that public surveys do not capture. Public-health researchers can identify better data sources or evidence that changes how a pattern should be read.
Expert review is especially important here because the most interesting questions begin exactly where the public data become thin. A recent routine checkup among women with no mammogram in two years may suggest an opportunity for closer study, but it does not prove a missed recommendation. A lower county estimate may identify a place worth examining, but it does not tell us which local mechanism produced the difference. The project is designed to keep those distinctions visible rather than hide them behind a polished chart.
Definitions, age ranges, survey-variable choices, treatment of missing responses, confidence-interval rules, wording, visual emphasis, the questions we ask next, and even whether a particular analysis belongs on the site at all.
Who is behind the project
The Mammography Gap Explorer is being developed by Masha Galenko, a graduate student in public policy at the University of Chicago working on early breast cancer detection and the practical problem of getting recommended screening completed.
The project grew out of a broader interest in what happens between awareness and action. Mammography is a useful case because the endpoint is concrete—a completed imaging test—but the path to that endpoint crosses individual behavior, clinical practice, insurance and cost, scheduling, transportation, reminders, referral systems, and local availability. The Explorer is a way to map what public evidence can already tell us before making claims about which intervention would work.