Ecologic studies
Ecologic studies differ from the other types of studies we have considered in that instead of using individual-level data they use aggregated data, such as data about classes in a school, factories, or geographic areas.1,9 Like other types of studies we have looked at, they may consider associations between an exposure variable and an outcome.
Types of variables considered in ecologic studies include: aggregate measures relating to individuals within a group, such as median income; environmental measures reflecting a geographical location of the group, such as air pollution; and global measures relating to the group but not reducible to individual characteristics, such as the type of healthcare system in place in a given area.1
A study of correlations between levels of opioid consumption and opioid-related mortality represents an example of an ecologic study.24 Data were collected for each year from 2005 to 2009 for the Canadian provinces of British Columbia and Ontario. Exposure data consisted of rates of use of a given prescription opioid (e.g., hydromorphone) in each year for each of the two provinces, and outcome data consisted of rates of death related to each opioid in each province. The study reported strong correlations between consumption rates for each of four strong opioids and deaths related to each opioid, raising concern about the association between use of prescription opioids and mortality.
A potential limitation of ecologic studies is that group-level associations may not apply at the individual level, which is known as the ecologic fallacy.1,10 For example, even if an association existed between average consumption of dietary fat and incidence of breast cancer in a country based on aggregate measures, it could be possible that the individuals who developed breast cancer consumed low levels of dietary fat.10 Another limitation is that ecologic studies usually cannot adequately control for factors which could confound the association between an exposure and outcome.9 Due to the limitations of ecologic studies, they tend to be viewed as producing preliminary findings that could be investigated further with more rigorous designs.1