5. Design of a Clinical Trial for a New (Analgesic) Drug

A clinical trial for a new agent is carried out during phase II or phase III of the drug’s development

Preclinical development involves animal studies into aspects such as safety, efficacy and mutagenicity. Phase I involves small group studies of fewer than 100 healthy volunteers, looking at pharmacokinetics, pharmacodynamics and adverse effects. Phase II recruits larger numbers of patients, typically 200–300, in which the findings of the phase I studies are refined. Phase III involves still larger numbers of patients, usually in the thousands, who are entered into definitive RCTs. Phase IV occurs after the drug has been licensed for use, and involves post-marketing surveillance of its effects in much greater numbers of individuals.)

No clinical trial can proceed without the approval of an appropriately constituted ethics committee, which will include laypeople amongst its members. The scrutiny of applications by these committees is increasingly rigorous, and in essence each committee seeks to preserve the full protection of the rights of every potential participant. Individuals must receive full information about all aspects of the trial before they consent, and they must be free to withdraw at any stage without compromising their future care. Committees will examine particularly intently any trial in which financial inducements are involved.

: the best-designed clinical trials seek to answer a single simple question: in the case, for example, of a novel painkiller, whether this new analgesic is superior to established treatments. It is essential to have a control in the study, which in this instance would be an analgesic in clinical use that was of proven benefit. Trial design must therefore involve defining endpoints for efficacy, and must also ensure that data relating to adverse effects are collected The use of placebos in trials of analgesics is considered to be unethical, and so the drugs in all limbs of the trial will be pharmacologically active.

that the groups are matched as far as possible. Such matching should include age, gender, American Society of Anesthesiologists (ASA) status and racial characteristics If the drug is to be used for treatment of chronic pain, then the trial can be a double-blind crossover trial (see the following) in which the patient can act as his or her own control. Sufficient time must elapse between administrations of the two drugs to ensure that the first one the patient has received is no longer exerting any effect.

conclusions of any trial can be erroneous. type I error The study can determine either that there is a difference between treatments when none exists, type II or it can determine that there is no difference between treatments when a difference does in fact exist. In other words, the power of a study is its ability to reveal a difference of a particular size. The power calculation allows the investigator to determine the sample size necessary to demonstrate this difference. It is calculated from 1-β, where β is the type II error. Trials are usually designed with a power of 80% (β = 0.2) or better, 90% (β = 0.1). The investigators must also decide the magnitude of the difference that is sought

randomization of patients to one or other limbs of the trial is intended to remove bias. The bias may be unconscious or hidden Simple methods such as tossing a coin are valid, although it is more common to use computer-generated randomization.

: it is ideal for the trial to be double blind, so that neither the patient nor the investigator knows to which group they have been assigned. This is of particular importance when the outcome data are subjective, as in a comparison of analgesic drugs or techniques.

: obvious considerations apply to the scrupulous collection of data. Inherent variation can be avoided by minimizing the number of investigators involved in the process

the appropriate statistical tests must be chosen for the question that is being asked The investigators must also decide whether the data are continuous and normally distributed, in which case a parametric test is appropriate The evaluation of an analgesic would almost certainly involve the use of visual analogue scales, about which statisticians may disagree. Some argue that response to pain is a biological variable with a normal distribution; others contend that the data are not normally distributed and that non-parametric tests should be applied

assumption made at the start of any investigation that there is no difference between the populations, treatments and samples that are being compared This is usually 0.05 (which means that there is a 5% likelihood of the difference occurring purely due to chance).

Type I or α error: In this case the null hypothesis is wrongly rejected, and a difference is found when there is none. This is a false positive. type I error is reduced by requiring a higher probability value (making P smaller), by increasing the sample size, or both. By convention, a 5% probability of making a type I error is accepted, and the confidence level is given by (1  α).

Type II or β error: in this instance the null hypothesis is wrongly proved, and so no difference is found when one does in fact exist his is a false negative. Type II errors are easier to avoid than type I, and their commonest cause is a sample size that is too small

Power: the ‘power’ of a study is the measure of its likelihood of detecting a difference between groups if a difference really does exist.

Sensitivity: this is a measure of how good is a clinical test at excluding false positives and is defined by the proportion of positives that are correctly identified by the test. It is determined by the proportion of patients who test positive in relation to the numbers who actually are positive.

this is an alternative means of determining whether an abnormal result predicts a genuine abnormality. It is defined by the numbers of patients who both test positive and who are genuinely positive as a proportion of the total of correct positive tests.

This is a measure of how good is a clinical test at excluding false negatives and is defined by the proportion of negatives that are correctly identified by the test. It is determined by the proportion of patients who test negative in relation to the numbers who actually are negative.

: this is an alternative means of determining whether a normal result precludes a genuine abnormality. I t is defined by the numbers of patients who both test negative and who are genuinely negative, as a proportion of the total of correct negative tests