Clinical Trials

❘❙❚■ Chapter 34 | Presenting Baseline DataWhy should baseline data be described?Baseline data describe the characteristics of the population participating in thetrial. This should allow you to assess the generalizability of the trial by comparingthe characteristics of the trial’s patient population with the overall patientpopulation with that disease, thus confirming the external validity of the trial.In large studies where randomization has been performed correctly, the baselinevariables should be well balanced between treatment groups, confirming thatrandomization has been successful. If the treatment groups are imbalanced withrespect to their baseline variables then the results obtained might be biased –particularly when such variables might affect the outcome (eg, disease severity).Reporting baseline data also allows for the identification of imbalances betweentreatment groups that might subsequently become confounders. Such potentialconfounders can create an apparent difference in outcome between groups wherenone really exists, or they can even mask an effect that truly exists.If the investigator is aware in advance of which baseline characteristics mightact as potential confounders then, when designing the study, procedures such asstratification or minimization can be incorporated into the randomizationalgorithm to balance these characteristics. This will help to ensure that anyimportant variables that might affect the outcome are balanced across groups.For example, in a trial of a treatment for lung disease in cystic fibrosis it mightbe important to ensure that the groups are balanced for baseline lung functionmeasurements. This is particularly important in smaller studies, where there is anincreased risk of imbalance in baseline variables due to chance.What should be included in baseline data?It is essential to provide baseline data when you report randomized controlledtrials. Most major medical journals now require you to follow the CONSORT(Consolidated Standards of Reporting Trials) guidelines [1], which are a set ofrecommendations aimed at improving reporting through the use of a checklist and aflow chart. The checklist pertains to the content of the title, abstract, introduction,methods, results, and discussion. Item 15 of the CONSORT guidelines states thatbaseline demographics and clinical characteristics should be described for each group.Baseline data should always be measured prior to randomization, as close to thetime of randomization as possible. This is particularly important when the variablebeing measured (eg, disease severity) could affect the outcome of the study.386