Clinical Trials

Clinical Trials: A Practical Guide ■❚❙❘incorporating results should still contain sufficient information for the reader tojudge key points of internal and external validity. The CONSORT statementprovides a checklist to help authors cover all the relevant points (Table 1).CONSORT StatementThe CONSORT statement has been very influential in providing a standard forthe quality of a clinical trial report in medical publications. A full explanation canbe found on the CONSORT website, which is freely accessible [1]. In this section,we highlight key points of the CONSORT statement.In order to enable the reader to evaluate the flow of patients through a study,a flow chart (as shown in Figure 1 of Chapter 33 of this book) is a helpful graphicalsummary. A checklist provided by CONSORT (Table 1) describes several itemsthat should be included when reporting a clinical trial. Items 1–14 and 16–18relate to the internal validity of the study, while items 2–4, 6 and 15 areparticularly important to address the generalizability of the findings; these shouldbe highlighted by the authors themselves, as mentioned in item 21.RandomizationRandomization is essential to ensure that there is no confounding. Hence, itshould be mentioned in the title and abstract of the study (item 1), and furtherexplained in detail in the methods (items 8–10). Different randomization schemescan be used depending on the numbers of patients allocated to a particulartreatment in different study centers. Reporting the method of allocationconcealment and its implementation helps the reader to identify potentialproblems in masking and blinding (item 11).Sample sizeA trial needs to include sufficient numbers of patients in order to detect the effectof interest (item 7). The sample size calculations depend on the form of theplanned statistical analysis and on making reasonable assumptions about theexpected treatment effect, summary statistics of the primary endpoint in thecontrol group, etc (see Chapter 9).Defined outcomeStatistical analyses are tailored to a defined hypothesis of interest (item 5 and 12).If statistical analyses are used for hypotheses other than those originally plannedthen there is a danger of false-positive chance findings. Therefore, the primaryoutcome has to be defined in advance (item 6).369