What is randomization

Randomization, by definition, randomly assigns patients to the treatment groups. Often compared to a coin toss, the actual randomization process is usually accomplished by use of random number tables or computer programs. It is important that neither the investigator nor the patient knows the random assignment prior to the patient's decision whether or not to enroll. In the older clinical trials literature, one can find examples of trials that allocated participants to treatment groups...

Is it necessary to be a biostatistician to interpret scientific data

Many clinical trials employ statistical methods that are rarely taught in medical school or even in introductory biostatistics courses. It is easy to understand why a reader may be intimidated by the data analyses in clinical trial reports. While collaboration with an experienced biostatistician is critical in designing, analyzing and interpreting a clinical trial, a good command of mathematics, clinical experience and common sense usually suffice for reviewing trial publications. Readers...

How is symptomatic improvement measured

Only patients can accurately assess how they really feel. One should question the findings of any clinical trial that depends on a physician's assessment of the severity of a patient's symptoms such as itching, heartburn or insomnia. Unfortunately, many investigators and sponsors are reluctant to accept patient self-reports as primary outcomes. However, when the objective of treatment is to alleviate symptoms or to improve well-being, a subjective but relevant measure has more meaning than one...

What determines interchangeability for efficacy

Similar effects on a surrogate marker represents a very unreliable indicator that one drug may be safely substituted for another. The limitations of these markers are discussed in Chapter 19. The fact that all ACE inhibitors lower elevated systolic blood pressure does not mean that all ACE inhibitors are interchangeable. Antihypertensive treatment is most meaningful when it is documented to have actual health benefits, i.e., reduction in risk of fatal and non-fatal stroke, myocardial infarction...

What are the challenges in attributing causation

It is sometimes difficult to decide what constitutes a true adverse drug effect. As an extreme example, a patient in a diabetes trial of glycemic control was involved in a car accident that caused the death of the other driver. Is it possible that the accident could have been caused by the study subject suffering a hypoglycemic attack Should this fatality be classified as an adverse effect and the consequence of tight control of blood sugar If the universe of all possible adverse effects were...

Was the result for the primary outcome clearly stated in the conclusion and supported by the confidence interval

The major conclusion(s) of any clinical trial should be based on the findings of the pre-specified primary outcome. Because deviations from this fundamental principle are common, journal readers should be on their guard when reviewing trial results. Non-significant statistical findings for the primary outcome may be overinterpreted. Nominally statistically significant findings for one or more of several pre-specified secondary outcomes may be highlighted rather than drawing attention to a...