Inclusion Criteria

Inclusion criteria are a set of predefined characteristics used to identify subjects who will be included in a research study. Inclusion criteria, along with exclusion criteria, make up the selection or eligibility criteria used to rule in or out the target population for a research study. Inclusion criteria should respond to the scientific objective of the study and are critical to accomplish it. Proper selection of inclusion criteria will optimize the external and internal validity of the study, improve its feasibility, lower its costs, and minimize ethical concerns; specifically, good selection criteria will ensure the homogeneity of the sample population, reduce confounding, and increase the likelihood of finding a true association between exposure/intervention and outcomes. In prospective studies (cohort and clinical trials), they also will determine the feasibility of follow-up and attrition of participants. Stringent inclusion criteria might reduce the generalizability of the study findings to the target population, hinder recruitment and sampling of study subjects, and eliminate a characteristic that might be of critical theoretical and methodological importance.

Each additional inclusion criterion implies a different sample population and will add restrictions to the design, creating increasingly controlled conditions, as opposed to everyday conditions closer to real life, thus influencing the utility and applicability of study findings. Inclusion criteria must be selected carefully based on a review of the literature, in-depth knowledge of the theoretical framework, and the feasibility and logistic applicability of the criteria. Often, research protocol amendments that change the inclusion criteria will result in two different sample populations that might require separate data analyses with a justification for drawing composite inferences.

The selection and application of inclusion criteria also will have important consequences on the assurance of ethical principles; for example, including subjects based on race, gender, age, or clinical characteristics also might imply an uneven distribution of benefits and harms, threats to the autonomy of subjects, and lack of respect. Not including women, children, or the elderly in the study might have important ethical implications and diminish the compliance of the study with research guidelines such as those of the National Institutes of Health in the United States for inclusion of women, children, and ethnic minorities in research studies.

Use of standardized inclusion criteria is necessary to accomplish consistency of findings across similar studies on a research topic. Common inclusion criteria refer to demographic, socioeconomic, health and clinical characteristics, and outcomes of study subjects. Meeting these criteria requires screening eligible subjects using valid and reliable measurements in the form of standardized exposure and outcome measurements to ensure that [Page 590]subjects who are said to meet the inclusion criteria really have them (sensitivity) and those who are said not to have them really do not have them (specificity). Such measurements also should be consistent and repeatable every time they are obtained (reliability). Good validity and reliability of inclusion criteria will help minimize random error, selection bias, misclassification of exposures and outcomes, and confounding. Inclusion criteria might be difficult to ascertain; for example, an inclusion criterion stating that “subjects with type II diabetes mellitus and no other conditions will be included” will require, in addition to clinical ascertainment of type II diabetes mellitus, evidence that subjects do not have cardiovascular disease, hypertension, cancer, and so on, which will be costly, unfeasible, and unlikely to rule out completely. A similar problem develops when using as inclusion criterion “subjects who are in good health” because a completely clean bill of health is difficult to ascertain. Choosing inclusion criteria with high validity and reliability will likely improve the likelihood of finding an association, if there is one, between the exposures or interventions and the outcomes; it also will decrease the required sample size. For example, inclusion criteria such as tumor markers that are known to be prognostic factors of a given type of cancer will be correlated more strongly with cancer than unspecific biomarkers or clinical criteria. Inclusion criteria that identify demographic, temporal, or geographic characteristics will have scientific and practical advantages and disadvantages; restricting subjects to male gender or adults might increase the homogeneity of the sample, thus helping to control confounding. Inclusion criteria that include selection of subjects during a certain period of time might overlook important secular trends in the phenomenon under study, but not establishing a feasible period of time might make conducting of the study unfeasible. Geographic inclusion criteria that establish selecting a population from a hospital also might select a biased sample that will preclude the generalizability of the findings, although it might be the only alternative to conducting the study. In studies of rheumatoid arthritis, including patients with at least 12 tender or swollen joints will make difficult the recruitment of a sufficient number of patients and will likely decrease the generalizability of study results to the target population.

...