skip to content
Primary navigation

Which projects require IRB Review?

Which projects require IRB Review?

Many research and evaluation projects conducted on DHS participants or data require review by DHS' Institutional Review Board. This describes the types of projects that require full review, expedited review as well as the types that do not require review.


Vulnerable populations include: people who are mentally ill, chemically dependent, disabled, cognitively impaired, or pregnant. Vulnerable populations also include participants of mandatory programs (e.g. child protective services) and people in institutions (e.g. correctional facilities, residential treatment, nursing homes, group homes, foster care). If participants are receiving the services, program, or are participating in the study because they are members of one of these categories, then they are considered vulnerable. For example, a study of prenatal service use would only include pregnant women and thus the study involves a vulnerable population. In contrast, some participants of a cash assistance program may be pregnant, but this is not considered a vulnerable population unless their pregnancy is a criteria for their participation. Children under the age of 18 are a special vulnerable population, and any form of data collection with them requires a full IRB review.

An identifiable dataset includes data fields which uniquely identify particular individuals (e.g. name, SSN, Recipient ID), or it includes indicators which in combination could identify individuals (e.g. information on Census block, race, age, and gender could together identify some individuals). Data is considered de-identified if it is not identifiable.

By highly sensitive the IRB means a topic which might place participants at risk of criminal or civil liability or might be damaging to their financial standing, employability, or reputation. Personally identifiable data on a person's chemical or mental health or child protection involvement is always considered to be highly sensitive.

The following are examples of using a study's findings in a generalizable or non-generalizable way. If an evaluation of a homeless program will be used to improve the program itself, this is not generalized. If the same findings are applied to a different program or a different population served, this is generalized. Similarly, if you present your findings to people who serve a population other than the one you studied (e.g. at a conference, in a journal) this is also generalizing it. Describing your findings in a grant proposal, on your website, or in any other mode of dissemination is only considered generalizable if you intend your findings to be applied to populations and programs other than the one you studied. Note that if a study is not originally designed to contribute to generalizable knowledge, but later you decide it would be worthwhile to do so, the authors can submit an IRB application at that time.

Data is in an individual-level format if each participant's information can be linked together (even if the participant can't be identified). This might be in a dataset with each participant's data on a single row, or in a database where each participant’s information can be linked by some indicator (e.g. their recipient ID).

back to top