Advancing Natural Experiment Research at NIH

Rigorous evaluation of natural experiments is increasingly recognized as important for advancing the NIH’s health research mission. Although there has been a more recent increase in interest in this area of research over the last two decades, especially in relation to the COVID-19 pandemic, the importance of natural experiment research for improving public health has long been recognized. For example, the cholera outbreak in London in the mid-1800s went into rapid decline after John Snow’s discovery of systematic variation in disease susceptibility based on the use of a specific public water pump.

The events, policies, programs, or infrastructure changes being evaluated though natural experiment research are often not designed as true experiments. The term “natural experiment” generally refers to the evaluation of an exposure or change that is not directly manipulated by the researcher where the population can be divided as exposed and unexposed, and the outcome of interest can be assessed based on this variation in exposure. This area of research often focuses on opportunities where the data would either be impossible, very difficult, or potentially unethical to collect in a randomized experiment. In addition to assessing changes in the outcome of interest, researchers also can often assess potential unintended consequences, sub-group differences in response and other factors that might influence the outcomes of interest such as implementation fidelity and variability.

As with many types of research, natural experiment research has the potential for bias, confounding, and threats to causal inference. The conclusions drawn from this type of research can be strengthened by rigorous design and careful consideration of alternative explanations. Many of these outcome and design considerations are outlined in the NIH’s Pathways to Prevention (P2P) Program materials on Methods for Evaluating Natural Experiments in Obesity. Although this P2P initiative was focused on obesity, the research methods and design recommendations are generalizable beyond obesity research.

In an effort to further advance this area of research, OBSSR recently released a funding opportunity announcement (FOA) in partnership with NIAAA, NIAMS, NICHD, NIDA, NINR, NIMHD, NCI, ORWH, and ODP (PAR-22-233) titled, “Time-Sensitive Opportunities for Health Research (R61/R33 Clinical Trial Not Allowed)”.

The FOA includes an accelerated review and award process to support behavioral and social science research to better understand health outcomes related to an unexpected and/or time-sensitive event (e.g., emergent environmental threat; pandemic; change in local, state, or national policy; natural disaster). Applicants must demonstrate that the research proposed is time-sensitive and must be initiated with minimum delay due to a limited window of opportunity to collect baseline data, answer key research questions, and/or prospectively evaluate a new policy or program. The determination of time-sensitivity is focused on the need for accelerated data collection. As such, the urgency of the public health problem being studied is not, on its own, sufficient justification for time-sensitivity.

Key definitions in the FOA include:

  • An event is a unique and often unanticipated issue that arises in a particular community/population, due to emerging environmental threats, other public health threats, disaster events, or policy changes.
  • Policy is broadly defined to include both formal public policies at local, state, and federal levels of government, and organizational level policies, such as those implemented by large organizations, worksites, or school districts.
  • Program is defined as a set of activities such as implementation of system-level interventions, tools, or guidelines initiated by governmental or other organizational bodies, within public or private entities in local, state, or federal jurisdictions.
  • Infrastructure changes are alterations to the built environment such as housing, roads and other aspects of transportation systems, retail environments, and building of parks or green spaces.

The FOA encourages partnerships and collaborations between researchers and the impacted community. Expected study methodologies may include, but are not limited to, interrupted time-series, difference-in-difference designs, regression discontinuity, or propensity scoring. The distinguishing features of a study that is responsive to the FOA include:

  • The unpredictable and unanticipated nature of the research opportunity.
  • The clear scientific value and feasibility of the study.
  • A feasible plan for collection of baseline data and primary data collection (although use of existing data is allowed, a plan for collecting important and new data rapidly should be provided).
  • A justification for why expedited review and funding (substantially shorter than the typical NIH grant review/award cycle) are required for the scientific question(s) to be addressed and the research design to be implemented.

If this FOA is of interest, I encourage you to read the full document for more details and reach out to the relevant scientific contacts listed to discuss the fit and scope of your research for that Institute or Center. Also, OBSSR and our Institute, Center and Office partners held a technical assistance webinar to help investigators better understand the intent and scope of the FOA. The webinar can be viewed here:

Although randomized controlled trials will always have an important place in advancing health research, increased use of alternate approaches such as natural experiments will enhance our ability to more broadly understand what works and doesn’t work for whom, under what circumstances, and why. This type of non-randomized research advances the NIH mission and is complementary to the large body of randomized controlled research supported by the NIH.