FAQ

How much will my project cost?

Costs vary from project to project, depending on sample size and requested services. Contact the research team for details.

Will all people in my study have a HOUSES Index?

The Housing-Based Socioeconomic Status (HOUSES) Index is available for nearly all residential addresses in the U.S. The HOUSES Index will not be available for P.O. box addresses or nonresidential addresses, such as hospitals or schools. We have HOUSES data dating back to 2004.

What methodology is used to calculate the HOUSES Index?

Each property item corresponding to a person's address was standardized into a z-score within each county. These data are then aggregated into an overall z-score, which is converted to quartile, decile and percentile. A higher HOUSES score indicates a higher socioeconomic status.

What does the HOUSES Cloud provide?

The HOUSES Cloud is a customer-centric service providing users with the HOUSES Index and other socioeconomic status tools for exploring health disparities. Also, the HOUSES Cloud is compliant with the Health Insurance Portability and Accountability Act of 1996 (HIPAA).

Should subject identifiers be uploaded to the cloud?

No. The HOUSES Cloud only requires full address information and index year. The system does not save address information. Also, only verified end users can see the address information.

How easily accessible and usable is the HOUSES dataset?

The HOUSES Index is up and running through a HIPAA-compliant cloud platform. Nationwide assessment property data, which is updated annually for tax purposes, and HOUSES algorithms are stored in Amazon Web Services. Users inside and outside of Mayo Clinic obtain HOUSES Index data via a publicly accessible, password-protected HOUSES Cloud dashboard or application programming interface (API) access model.

How does the HOUSES Index compare to the Area Deprivation Index?

The main difference between the Area Deprivation Index and the HOUSES Index is that the HOUSES Index is an individual-level measure or composite z-score for individual housing value, housing size, and the number of bedrooms and bathrooms. In contrast, the Area Deprivation Index assigns a single score for all subjects living in the same census block group. Since the HOUSES Index is normalized within a county, it measures a ranking for relative social determinants of health within a county.

What other services in addition to the HOUSES Index does the program provide?

Apart from the HOUSES Index, the HOUSES Program provides these services to users:

• ADI.
• Rural classification.
• Distance to a reference point, such as a hospital or clinic — a tool for assessing physical access to healthcare that is important for time-sensitive health outcomes, such as heart attack or stroke, especially in rural areas.
• Geospatial report.

How does the HOUSES Index reflect the housing value differences that exist between rural and urban areas?

The HOUSES Index is normalized within a county. The index uses the property information for all residential parcels and measures a relative SES ranking within a county. This standardization, which may not be available for income or educational attainment, makes socioeconomic comparisons easier and fairer among populations residing in different regions, such as urban versus rural.

Housing value differs across the counties or regions. Our program's experience has been that when housing values vary between urban and rural areas within a given county, the resultant HOUSES Index reflects different socioeconomic status. For example, unlike other regions, in the community of Olmsted County, a mixed urban-rural setting in southeastern Minnesota, one study found that people living in rural areas have overall higher socioeconomic status, as defined by HOUSES, than socioeconomic status in urban areas. Therefore, the people had better health outcomes, such as a lower prevalence of obesity and mood disorder, as well as better preventive healthcare utilization compared with urban residents.

How feasible is it to map a HOUSES Index output to individual patients, such as a Medicare beneficiary?

The HOUSES Index is based on individual residential addresses by matching to publicly available real property assessment data for each housing unit in the U.S. HOUSES may not be calculated for a very small portion of participants, such as those with P.O. box or invalid addresses. In general, about 95% or greater of previous study cohorts or practice patients had HOUSES Index matched with their addresses, including rental properties such as apartments.

Similarly, the Area Deprivation Index is often missing for some census block groups due to either low population or a high group quarters population. For example, in our program's ongoing risk adjustment model study using the Mayo panel cohort (n=120,622), the Area Deprivation Index was missing in 7,008 (5.8%) participants; whereas, HOUSES data were missing in 6,507 (5.4%) of the study cohort. In another study, 15% of the study cohort missed Area Deprivation Index data.

With valid addresses available in electronic health records, there should be no major challenges mapping HOUSES Index to Medicare beneficiaries.

How does the HOUSES Index account for cost-of-living disparities between different areas of the country?

There is comparatively little variation by region relative to costs of food, energy transportation and services. However, variations in housing costs have been reported to be a major driver for differential cost of living across regions. Therefore, our program conceptualizes that each county's cost of living is reflected in its housing market.

The HOUSES Index reflects regional housing markets and provides a within-market comparison, enabling a between-market comparison. Given that the county also is the general geographic unit for property taxation, we chose county as the geographic unit of standardization. As housing prices are significant sources of cost-of-living differences across counties, normalizing HOUSES at a county level is likely to address the concern about differential cost of living across different regions.

How is HOUSES data applied to different apartment or institutional structures, such as nursing homes?

Ownership is not part of HOUSES based on our original study (Juhn et al., 2011). Conceptually, even for a renter who lives in a very expensive apartment unit in New York City, that renter's socioeconomic status may be reflected by the HOUSES index as paying higher rent. We don't have individual apartment unit information on size, value, etc. For a condo that has an individual unit price and size, we assign a separate HOUSES index for each condo if our source data indicate it is a condo.

The HOUSES Index is not designed to be calculated for institutional addresses such as nursing homes or assisted living facilities. That's because property-level attributes such as square footage or number of bedrooms or bathrooms do not reflect the socioeconomic status of individual residents in these settings.

In our cloud, we used a curated list of such facilities and excluded these addresses accordingly. When available patients' prior residential (noninstitutional) addresses are used to derive the HOUSES Index, thereby preserving individual-level socioeconomic status information for patients transitioning into institutional care.

In electronic health record data, placeholder addresses of people experiencing housing instability or homelessness are often used. For example, the addresses of social service organizations or churches, hospitals or health care centers, or shelters may be used. Do you have a way to separate out those folks since those statistics for those buildings, such as square footage, might be misleading?

Placeholder or proxy addresses such as shelters, hospitals, social service organizations may appear in electronic health record data for patients experiencing housing instability. Applying property-based socioeconomic status measures to such addresses could be misleading.

In our implementation, we restricted HOUSES Index calculations to properties classified as "residential" within the underlying property database. Addresses that were identified as nonresidential, such as commercial buildings, institutions or shelters, were not assigned a HOUSES value and were instead treated as missing.

This approach helps mitigate the risk of misclassifying people based on nonrepresentative property characteristics. However, we acknowledge that this strategy does not fully resolve the challenge. Some people experiencing housing instability may still be linked to residential addresses that do not accurately reflect their true living conditions, such as in the case of temporary stays or outdated addresses.

Is there no adjustment for family size?

The HOUSES index does not explicitly adjust for household or family size. Therefore, the index assumes that larger or higher-valued housing reflects greater socioeconomic resources. While this assumption is generally valid at the population level, it may not hold in all cases, such as in larger or multigenerational households where per capita resources differ.

The HOUSES Index is intended as a household-level proxy for socioeconomic status rather than a per capita measure. To further address this limitation, we are conducting external validation studies in Sweden, where national registry data include detailed information on family size. This ongoing work will allow us to directly evaluate the impact of household size on house-based socioeconomic status classification and refine the index accordingly. These efforts aim to strengthen the conceptual and empirical validity of this index across diverse settings.

Do you have any way to account for renters versus owners?

Renters and homeowners may differ in socioeconomic profiles, but housing tenure does not always align perfectly with socioeconomic status. Property-level characteristics, such as housing value, square footage and amenities, reflect the overall resource environment associated with a residence, regardless of ownership status. For example, renters residing in higher-value or amenity-rich properties may have better access to desired resources associated with better health outcomes, even if they do not own the property. Conversely, ownership does not necessarily imply higher socioeconomic status.

We acknowledge that we do not have complete, systematically available data on renter versus owner status in this cohort, which limits our ability to perform stratified analyses as suggested. Nonetheless, prior validation studies suggest that the HOUSES Index performs robustly across varied housing contexts.