Understanding terminology

The following terminology is important in critical appraisal and will be helpful before proceeding further with this section.

You can also open and print the entire Glossary of terms for the CIAP Modules.

Applicability Applicability relates to whether a particular treatment or form of care that demonstrated an overall effect in a study can be expected to provide the same effect for an individual or group in a specific clinical or population setting
Bias Bias is the deviation of a measurement from the ‘true’ value leading to either an over or under-estimation of the treatment effect. Bias may originate from different sources: allocation of patients, measurement, interpretation, publication, and review of data
Chance Chance is the absence of any cause of events that can be predicted, understood, or controlled. It is the unknown and unpredictable element in happenings that seem to have no assignable cause
Confidence interval (CI) Confidence Interval (CI) is the interval within which the population parameter (the ‘true’ value) is expected to lie with a given degree of certainty, for example 95%
External validity External validity is the degree to which the results of a clinical study can be applied to clinical practice in a specific setting
Generalisability The ability to reliably apply the results of a study to other populations, based on the characteristics of the subjects, size of the sample, the setting, and trustworthiness of the study
Intention-to-treat Intention-to-Treat analysis is a method of analysis for randomised controlled trials in which all patients randomly assigned to one of the treatments are analysed together, regardless of whether or not they completed or received that treatment, in order to preserve randomisation
Internal validity Internal validity relates to the quality of the study design in terms of the methods. A study has internal validity if it is free from bias or systematic error and the results seen are due only to the intervention
p value p value is the probability that a particular result would have happened by chance
Power Statistical power is related to sample size. It is the ability of a study to detect a difference where a difference really exists. For example, a sample size of 3 is unlikely to be able to detect a true difference between treatments, whereas a sample of 3000 may be large enough to show a difference. For rare events or uncommon outcomes the sample size needs to be quite large to detect differences with confidence.
Relative risk (RR) Relative Risk (RR) is the ratio of the rates of outcome in the treatment and control groups. This expresses the risk of the outcome in the treatment group relative to that in the control group
Reliability Reliability relates to the trustworthiness of the results

To find out more about Confidence Intervals and p-values, read Davies & Iain Crombie's paper, which explains the terms in user-friendly language. [3]

Calculating risk when an outcome is relatively uncommon
In situations where an outcome is uncommon or has not occurred, for example in a medical trial that has uncovered no major side effects, it is difficult to explain the risk to patients. In these cases the outcome is known as a zero numerator. However, a zero numerator does not mean ‘no risk’.

In order to calculate a 95% confidence interval for the potential risk the ‘rule of three’ may be useful. The 95% confidence limits are between 0 and 3 divided by the number of patients.

For example: zero side effects in a study of 10 patients means the 95% confidence limits are between 0 and 30% whereas zero side effects in a study with 100 patients has 95% confidence limits between 0 and 3%.

Read more about this calculation in the following paper by James A. Hanley & Abby Lippman-Hand. [4]