Appraising the validity of a randomized controlled trial

The appraisal of an individual randomized controlled trial also addresses issues of internal and external validity. The main difference compared with the appraisal of a systematic review is that the reader can appraise the trial directly, rather than relying on the author of the review to have made an adequate assessment of the quality of the studies included. The most important questions to answer during the appraisal of the article are those relating to methodological isues that have been randomly shown to affect the reliability of the results (JableS).

1. Wfa^lri^fWmtfl^iKitnu ration

IWun t#mi4> hicvwM'

A VitfT mV+pwpy-* Ah-:h ijifrvn 1-iKfcrTmHiS1

S Wwr :krt-:lmil blrrf Ebuhth LiriliflHW -Id btrf

4 Oh Linn aw rtWFji PHPH ntft Iht panTi DfcBbl ^Jft i W(Tt iN-fOuM.! ^jrjl L1* Hit 0l Vi '

Table 3 Questions that must be answered during the appraisal of an article

This study(24) was a randomized controlled trial. Randomization was by a computer-generated schedule, although no details are provided about the concealment of treatment allocation. Concealment of allocation is one of the most important features to appraise in a randomized controlled trial, and refers to how well the treatment allocation of the next patient was concealed from the participating clinician.(28) If a clinician has definite knowledge of, or can reasonably predict, the next allocation, he may decide not to enter a patient if he favours either treatment in that specific case. This would obviously lead to a subversion of randomization and biased results. This is why methods of quasi-randomization, such as alternate allocation, are often unsatisfactory. The most satisfactory methods of random allocation are when allocation is performed by a third party following entry of the patient into the study, for example using a centralized telephone service. Concealment of allocation should be distinguished from blinding of treatment allocation. Blinding refers to whether the patient (single blind), or both the patient and the clinician (double blind), are kept unaware of which arm of the study the patient has been allocated to following randomization. Blinding protects against bias in the treatment during the trial and in the assessment of subjective outcomes, but can be difficult to maintain when the experimental treatment has characteristic side-effects (e.g. antidepressants). This study was reported to be double blind; however, it is difficult to tell how effective the blinding was and whether this affected the treatment that the patients received or the ratings of outcome.

Of the 410 patients who were randomized, 310 (75.6 per cent) completed the trial. The reasons for drop-out are clearly stated and all the patients who entered the study are accounted for. Patients who drop out of a study may be different from those who complete it—this is a particular problem if there is differential drop-out between the arms of the study. For this reason, the most statistically reliable and clinically useful method of analysis is to include all patients who were randomized in the analysis. This is called an intention-to-treat analysis (ITT). In the study by Thase et al.(24) the primary analysis was done on an ITT basis (with the exception of two patients who dropped out before the first follow-up assessment), and the patients were analysed in the groups to which they were randomized. However, follow-up analysis was not obtained after patients left the study and so the ITT analysis had to use the last available measure of outcome. This is a last observation carried forward analysis and introduces considerable uncertainty about the actual subsequent progress of the patients who left the study. The degree of this uncertainty depends on the proportion of patients who leave the study; generally, if fewer than 80 per cent of patients complete the trial, one should be cautious about the overall reliability of the results. An alternative approach to analysis that is often encountered in randomized controlled trials in psychiatry is the endpoint or completers analysis. This type of analysis includes only those patients who completed the trial and may give an idea of the comparative outcome in these patients. However, it may be misleading when there is a high drop-out rate, and when drop-out from the study is non-random, i.e. when there are systematic differences between drop-outs and completers, and perhaps also different reasons for drop-out between the comparison groups.

Lastly, although randomization will avoid bias in treatment allocation, it is possible that, by chance, the groups will be unbalanced on some key prognostic factors such as age, sex, and duration and severity of illness. Therefore it is important to assess the baseline characteristics of the patients to identify any obvious differences. In Thase et al.(24) the patients were reasonably similar on baseline characteristics at entry into the trial.

5 Secrets to Lasting Longer In The Bedroom

5 Secrets to Lasting Longer In The Bedroom

How to increase your staying power to extend your pleasure-and hers. There are many techniques, exercises and even devices, aids, and drugs to help you last longer in the bedroom. However, in most cases, the main reason most guys don't last long is due to what's going on in their minds, not their bodies.

Get My Free Ebook


Post a comment