Defining Homelessness in Public Health Data Collection: Considerations and Examples

Understanding whether a case of infectious disease has occurred in a person who is experiencing homelessness is necessary to recognize transmission patterns, plan interventions, and prepare for potentially severe disease outcomes. As such, homelessness is a critical piece of information to gather during public health surveillance and investigations.

The US Department of Housing and Urban Development (HUD) defines homelessness here: HUD’s Definition of Homelessness: Resources and Guidance (HUD Exchange). However, the experience of homelessness can be variable, which may complicate defining a data element specific to public health outcomes. Furthermore, data on homelessness come from multiple sources, such as homeless service records, electronic health records, or interviews. Data collected passively or from multiple sources through system matching or integration requires a definition described in “inclusion” criteria. Data collection through interviews requires specific questions tailored to the epidemiologic question of interest. Data collection from both sources requires a predetermined conceptual definition and a plan for implementation.

The following provides considerations for creating the conceptual definition of homelessness for data collection in case report forms and survey instruments for infectious disease data collection. This resource is designed to inform discussion prior to public health data collection and is not designed to determine eligibility requirements for housing or other social support services.

Considerations

Considerations when defining homelessness for public health data collection:

• Homelessness vs housing instability
  People experiencing housing instability are not necessarily experiencing homelessness. Housing instability is often defined to include rent cost burden, risk of eviction, or frequent moves. Therefore, these data element should be separated in data collection. However, some people who are experiencing housing instability may access homeless services like meal programs. Accessing these programs might be of interest in data collection, in which case questions should focus on whether an individual spent time in particular homeless service facilities or settings.
• Time period and duration of homelessness
  The incubation period of the disease of interest can inform the relevant time period for the experience of homelessness. For example, gathering data on a foodborne illness outbreak would likely require asking about recent homelessness (current or in the past two weeks). Gathering data on HIV might require asking about homelessness over the past year or ever experiencing homelessness in their lifetime.  It may also be important to evaluate the duration of homelessness, such as whether it would be considered chronic (often one year or more of homelessness), episodic (experiencing homelessness multiple times but not consistently experiencing homelessness), or transient (a shorter time period of homelessness without recurrence).
• Denominator data
  If you will be using a data source that enumerates the population of people experiencing homelessness, be sure to use the same definition of homelessness in public health data collection as is used in the denominator data source. More information on denominator data sources and their data use considerations is available here: Data Sources That Enumerate People Experiencing Homelessness in the United States: Opportunities and Challenges for Epidemiologic Research (PubMed at nih.gov)
• Type of homelessness
  Some people experiencing homelessness sleep in an emergency shelter, while others sleep outside (or in a place not meant for human habitation). Furthermore, people may have spent different amounts of time in different locations. For example, over a previous month, people may have spent some nights in a friend’s apartment, some nights in a shelter, some nights in an encampment, and some nights in a vehicle. Decisions about how to document these varying experiences should be discussed when developing data collection and analysis plans.
• Potential transmission pathways vs general risk factors
  If the purpose of data collection is to identify where infectious disease transmission could have occurred, the data collection tool may need to include specific homeless service facilities or types of facilities. If you are considering homelessness as a general risk factor, questions might cover the experience of homelessness more broadly, regardless of access to specific facilities. This could entail using a definition that includes doubling up with other families or staying in hotels.

Examples

Case report form: short

In a general case report form, the following question may be used to collect information on sheltered and unsheltered homelessness. The data to answer these questions may come from electronic medical records, case interviews, or be based on the site of testing for the infection. The following is a short example that aligns with the Housing and Urban Development (HUD) definition of homelessness.

Case report form: long

The following case report form question includes expanded information related to homelessness:

Example surveillance system documentation

The following is an example of how to implement integrating housing status into a surveillance system. The National Violent Death Reporting System (NVDRS) assesses both homelessness and housing instability. In this example, a clear definition is provided for what constitutes homelessness with possible examples of how it may appear when submitted to the system. This specific example also provides instruction to the person submitting these data that if someone is identified as homeless, they cannot also select that the individual was also experiencing housing instability later in the system. These instructions help ensure this information is captured consistently across multiple people submitting data, and that it is accurate and reliable.

NVDRS Variable for Homelessness

Discussion:

Marking this variable “Yes” means that there was clear evidence in a document that the victim was experiencing homelessness, such as living in a car.

Examples of coding Homeless “Yes”:

• Victim had been living in his car since his wife discovered he had relapsed on meth and kicked him out of the family home.
• Victim had been staying at a local shelter for persons experiencing homelessness for the past 3 months.
• Victim lived in an abandoned house or building along with several other individuals experiencing homelessness.
• Victim was residing in a tent on a local campground.

Examples of coding Homeless “No”:

• Victim had been staying at a motel after being evicted 2 weeks ago.
• Victim had recently retired and had been residing and traveling in a motor home (i.e., RV) that she owns.
• Victim was a long-distance trucker, and lived in a cabin within his truck when he was not driving.
• Victim and her husband had recently been evicted and were staying with a friend until they could find housing.
• Victim had recently sold her house and was staying at a hotel until a new house she had recently purchased was ready.
• Victim was due to be evicted from his apartment in 3 days and did not have any place to stay once evicted.

Code Homeless “Unknown” when the residential address is stated “Unknown” and homeless status is not otherwise known. Otherwise, mark this variable “No.”

Note: Homeless and Housing instability are mutually exclusive and should NOT be coded together as “Yes” to indicate a victim’s current housing status. The coding of Homeless as (1) “Yes” automatically means Housing Instability should be coded as (0) “No.” The Homeless variable is intended to represent victims who have no fixed address AND have no place to live.”

Example interview questions

The following are examples of questions that have been included in existing surveys. Depending on the objectives of your data collection, it may be possible to model interview questions on these examples.

1. Youth Risk Behavioral Surveillance System (YRBSS)

YRBSS surveys 9^th through 12^th grade students. This question is specific towards adolescents and minors and is focused on the last 30 days. This question allows for distinction between sheltered and unsheltered homelessness.

2. National HIV Behavioral Surveillance (NHBS)

For a question tailored towards adults that captures homelessness during the past year and not all residence options, NHBS asks:

3. Centers for Disease Control and Prevention Medical Monitoring Project (MMP)

Questions about very specific experiences related to homelessness and housing instability may also help understand more specific risk factors and differences in health outcomes related to these experiences.

4. California Healthy Kids Survey (CHKS)

While this survey is intended to be administered to adolescents in grades 7 and 9 throughout California, this question assesses where people live at the present moment. It specifies what constitutes as a “home” and includes options to capture residing with others. This question does not distinguish unsheltered from sheltered homelessness, but will allow for answer options ranging multiple living arrangements.