1.1 Review of data types

A variable is an attribute used to characterise a data unit. They are called variables because their values vary from one data unit to another and may change over time. Commonly used variables for AMR data may include date of admission, sex, species, production type, sample type and minimum inhibitory concentration (MIC).

Variables can be classified into numeric (quantitative) and categorical (qualitative) variables. The classification of variables as numeric or categorical has implications for how data is analysed and visualised.

Numeric variables contain only numbers and have meaning as a measurement or a count. Numeric data can be represented as integers (1, 2, 3), fractions (½, ¼), decimals (377, 39.134) or percentages (20%, 50%).

Numeric data can be further defined as either discrete or continuous.

Discrete data represent items that can be counted and can only take on finite values. Examples of discrete data include the number of hospitalised patients, the number of deaths attributable to resistant pathogens and the number of different antimicrobials to which resistance is identified.

Continuous data represent measurements, and their values cannot be counted, but they can be measured in units. Examples of continuous data include MIC, zone diameter and milligrams of antimicrobial ingredient per kilo of bodyweight.

Numeric data can be split into categories by applying one or more cut-off values to the data. For example, to determine the susceptibility status of isolates to an antimicrobial, you could apply susceptible and resistant clinical breakpoints to the MIC measurements. This yields three categories: susceptible, intermediate and resistant.

Categorical variables represent characteristics of distinct groups. Categorical data is represented by a name, a string of alphanumeric characters or numeric values. A numerical code may be given to a categorical variable for analysis purposes (e.g. 1 for female and 0 for male), but these numbers have no mathematical meaning.

Categorical data can be further defined as nominal or ordinal.

Nominal data are entirely qualitative and are unordered. That is, the meaning of the data does not change if the categories are reordered. Examples include sex, production type and antimicrobial class. Nominal variables are often mutually exclusive such as female/male or alive/dead.

Ordinal data represents values that are discrete and ordered. These values are ranked, and the order of the ranking is important. The distance between values is not necessarily equal. Examples of ordinal data include socioeconomic status (low income, middle income, high income) or antimicrobial resistance status (susceptible, intermediate, resistant).