‘Adjusted’ associationSee Association and Logistic regression.
AssociationRelationship between two variables. For binary variables this means that the occurrence of a particular value of one variable, in an individual, is associated with (more likely to be in conjunction with) a particular value of the other variable.
Binary variableHas only two possible values (e.g. accept screening or not, female or not).
Categorical variableHas a set of distinct values, such as gender, recruitment setting.
Explanatory variableA feature potentially associated with outcome.
Logistic regression (LR)LR estimates and tests association between a binary outcome and one or more explanatory variables, with association being summarised as ORs. In univariate LR there is only one explanatory variable and, if that is binary, only one OR. If there is more than one explanatory variable then multivariable LR is needed (MV LR). This estimates the association of outcome with each explanatory variable ‘adjusted’ for the joint associations with other explanatory variables in the model.
Multi-variable modellingA mathematical combination of data for a sample, where there is both an outcome variable, and a number of potential explanatory variables for that outcome. There are various mathematical forms, but where the outcome is a binary variable, not a rare event, then the usual form is multi-variable logistic regression.
Null hypothesis (NH)A statement, prior to testing, of no effect (in this case, no association). See also Significance probability.
Odds ratio (OR)The OR is the ratio of the odds of outcome (unmet need) in those with the explanatory feature (previous unwanted pregnancy), relative to the odds in those without the explanatory feature (no such pregnancy). When there is no association, the two odds should be approximately equal and the ratio approximately 1, or ‘null’. The more extreme the OR (away from 1, i.e. 0.4 vs 0.7, or 3.2 vs 1.8) the greater the degree of association.
Outcome variableIndicates status with respect to the condition of interest (e.g. unmet contraceptive need).
Significance probabilityThe probability of obtaining the observed cell counts, or more extreme, if the NH is true.