Data Methods & Resources

This page explains how data shown on ChicagoHealthMap.com is collected, processed, and reported. It is written for a general audience, followed by a researcher-focused technical section at the end. We believe communities have a right to understand the data that describes them — including its strengths and its limits.

Primary Source: CAPriCORN

Our health condition data comes from CAPriCORN (Chicago Area Patient-Centered Outcomes Research Network) — a network of major hospitals and clinics across the Chicagoland area that pool anonymized patient records for public health research. Website: capricorncdrn.org

Seven health systems contribute diagnosis data to CAPriCORN

• Rush University System for Health
• Northwestern Medicine
• UChicago Medicine
• University of Illinois Hospital & Health Sciences System
• Cook County Health
• AllianceChicago — a network of federally qualified health centers serving low-income neighborhoods
• Endeavor Health

What information we receive from CAPriCORN

• Basic patient information: age, sex, race/ethnicity
• Diagnosis codes indicating what conditions a patient has
• Visit information: when they were seen and what type of visit
• Approximate neighborhood: census tract of residence (not exact address)

What we do not receive

• Name or any directly identifying information
• Exact home address
• Social Security number
• Clinical notes, lab values, or test results

Acknowledgments

We acknowledge CAPriCORN's partners: the Medical Research Analytics and Informatics Alliance (MRAIA), which serves as the network's honest data broker ensuring each person is counted only once across all systems; and the Chicago Area Institutional Review Board (CHAIRb), which serves as the central IRB of record for CAPriCORN-supported research. CHAIRb study number: STUDY2025-0712: CONSCIENCE.

Population Reference: American Community Survey

To calculate rates and assess how representative our patient population is, we use population estimates from the American Community Survey (ACS) 5-Year Estimates, produced by the U.S. Census Bureau. Source: data.census.gov

Geographic Boundaries

We define neighborhoods using census tracts — small geographic areas set by the U.S. Census Bureau, each containing roughly 1,200 to 8,000 people. Census tracts are small enough to capture neighborhood-level patterns, stable enough to allow comparisons over time, and directly linked to Census demographic data.

For Chicago specifically, we also aggregate data to the 77 Chicago community areas — a set of historically defined neighborhoods widely used for city planning and public health reporting. Source: Census Bureau TIGER/Line Shapefiles

Data Coverage and Time Period

Deduplication: One Person, Counted Once

People often seek care at more than one health system. Without linking records across hospitals, the same person could be counted multiple times — once at Rush, once at Northwestern — inflating our numbers and distorting neighborhood-level counts.

How We Identify Conditions

Clinicians record diagnoses using ICD-10 codes — the international standard system for classifying medical conditions used across all health systems. We use these codes to count how many people in each neighborhood have each of the 38 tracked chronic conditions plus firearm injury.

Because coding practices can vary across systems and providers, we use validated ICD-10 code sets developed through clinical consensus for each condition. See the Condition Reference section below for the complete code list.

How We Calculate Diagnosis %

Raw counts alone can be misleading. A neighborhood with 200 people with diabetes may sound worse than one with 100 — but if the first neighborhood has 10,000 adults and the second has 1,000, the second actually has a higher rate. We use % to allow more comparable values across neighborhoods of different sizes.

The Fewer-Than-10 Rule

We never display data when fewer than 10 people have a condition in a given area. This requirement follows federal privacy standards under HIPAA and is consistent with NCHS Data Presentation Standards for small-area health data.

Secondary suppression

In some cases, a suppressed value could be calculated from other visible numbers. We apply secondary suppression where needed to prevent this.

All data on this site is group data — counts and rates for populations, never information about any individual person.

Not all data on this map is equally reliable. How useful a given figure is for understanding a neighborhood's health depends on two things: how many people from that area are in our data, and whether those people reflect the neighborhood's actual population. Our reliability flags communicate both dimensions at once.

Why Reliability Varies

CAPriCORN hospitals and clinics are not evenly distributed across the region. Neighborhoods closer to participating facilities tend to have higher coverage. Some areas have 8 out of 10 adults represented in our data; others have 3 out of 10 or fewer.

Coverage alone doesn't tell the whole story. Even in well-covered areas, the patients we see may skew older, more insured, or demographically different from the neighborhood as a whole. Our equity flags identify when this demographic misalignment is significant enough to affect how race- and ethnicity-specific estimates should be interpreted.

The Two Dimensions: Coverage and Alignment

Reliability Tiers

Each area receives one of four reliability labels based on its capture rate, with or without an equity note:

Capture rate thresholds reflect the competitive multi-system healthcare market in Chicagoland. In a market with eight participating health systems, a single system capturing 10–20% of a tract's adult population represents strong market presence. There is no universal benchmark in the published literature for 'good' capture rate — these thresholds were calibrated to the Cook County context and validated against demographic alignment data.

Equity Note: What the Flags Mean

The equity note does not mean the data is wrong. It means the patient population in that area does not fully reflect the neighborhood's demographic composition — and that race- and ethnicity-specific estimates should be interpreted with caution rather than taken at face value.

How Reliability Flags Are Assigned

Flags are assigned at the census tract level using 2023 as the reference year, applied consistently across all map years. Analysis showed less than 2 percentage point variation in tier distribution from 2021 to 2024, confirming that a single reference year is stable and prevents flag instability when users toggle between years.

Understanding the limits of this data is just as important as understanding what it shows. No dataset is perfect — and being transparent about limitations is part of using data responsibly and equitably.

Coverage varies by neighborhood: This data does not include all healthcare visits for all residents. We only have data on people who visited a CAPriCORN hospital or clinic at least once between 2019 and 2024. Coverage varies by neighborhood, as CAPriCORN facilities are not evenly distributed across the region. Use the reliability flags to understand coverage in any given area.

This data is shared publicly to support community understanding, research, and planning. Please use it appropriately:

Do

• Use rates and reliability flags together when interpreting any figure
• Supplement with other data sources, especially in lower-reliability areas
• Cite the source when sharing maps, charts, or figures
• Reach out to us with questions about methodology or specific areas

Don't

• Describe this data as representing all Chicagoland residents — it covers approximately 4 out of 10 adults with healthcare encounters in the region
• Use this data to draw conclusions about specific individuals
• Treat suppressed ('<10') values as zero
• Remove data source attribution when sharing maps or charts
• Use for commercial purposes without permission

This section is intended for researchers and analysts who need additional methodological detail.

Data Standard: PCORnet Common Data Model

All seven health systems share data using the PCORnet Common Data Model (CDM) — a standardized format that allows patient records to be combined across institutions while preserving data quality and privacy. The PCORnet CDM includes structured tables for demographics, diagnoses, encounters, and address history.

Key CDM tables used in CONSCIENCE:

• DEMOGRAPHIC — age, sex, race/ethnicity
• DIAGNOSIS — ICD-10 codes and dates
• ENCOUNTER — visit types and dates
• LDS_ADDRESS_HISTORY — census tract of residence

Full CDM specification: pcornet.org/pcornet-common-data-model

Representativeness Assessment: CONSCIENCE Reliability Framework

The CONSCIENCE reliability framework measures two independent dimensions of data quality at the census tract level:

Chi-square goodness-of-fit testing confirmed that the capture rate and demographic alignment dimensions carry independent information — areas flagged for demographic misalignment show substantially higher distributional divergence from the census reference within every capture tier.

Age Adjustment Method

Direct standardization to the 2000 U.S. Standard Population, consistent with CDC and NCHS reporting standards. Age strata: 18–34, 35–44, 45–54, 55–64, 65–74, 75–84, 85+.

Key Methodological References

1. Austin, P.C. (2009). Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples. Statistics in Medicine, 28(25), 3083–3107. [SMD threshold]
2. Parker, J.D., et al. (2017). Proportions with confidence intervals. National Center for Health Statistics. [N thresholds for suppression and reliability]
3. Weiskopf, N.G., & Weng, C. (2013). Methods and dimensions of electronic health record data quality assessment. Journal of the American Medical Informatics Association, 20(1), 144–151. [EHR representativeness framework]
4. Kho, A.N., et al. (2014). CAPriCORN: Chicago Area Patient-Centered Outcomes Research Network. Journal of the American Medical Informatics Association, 21(4), 607–611.
5. PCORnet Common Data Model, current version. pcornet.org.

Key terms used across this page, the Chicago Health Map, and CONSCIENCE methodology.

Common abbreviations used throughout this page and the Chicago Health Map.

Complete list of all 38 tracked chronic conditions plus firearm injury, with their ICD-10 diagnostic code definitions.

Overview of all data sources used in ChicagoHealthMap.com, including clinical data, demographics, geographic boundaries, and supplemental layers.

Use the citation that matches where you're using the data.