Controlled trials in single subjects. 1. Value in clinical medicine.

Abstract

Randomised controlled group trials are the gold standard in clinical research and provide information about the average effect of a treatment in a target population. Such information, however, cannot necessarily be transferred to a single patient because finding a significant effect of treatment does not mean that all the study subjects responded. Besides, in real life clinicians are often confronted with vague symptoms and poorly defined diseases; this is a different case from the strict criteria used in scientific studies. In an attempt to adapt the general knowledge about a new treatment to a particular patient the clinician tries out drugs. Such an uncontrolled trial and error approach is, however, often biased towards a false positive effect as most disorders are self limiting, the placebo effect may be appreciable, and the patient is eager to please the doctor. All these aspects tend to contribute to the myth that treatment with drugs is a prerequisite for regaining health.Controlled trials in individual patients (n of 1 trials) have a long tradition in behavioural medicine' and have recently attracted attention in clinical medicine. Such trials may improve the certainty of a treatment decision in an individual patient, and a series of such trials may permit more general inferences to be drawn about a treatment. Method of studyThe main principle in the n of 1 trial is that the patient serves as his or her own control in a study of the efficacy of a new drug compared with placebo or another drug. The treatments are administered to the patient in a double blind, randomised, multiple crossover sequence (fig 1). Each treatment period is randomly assigned and the trial design is tailored to the clinical condition, the properties of the drug(s), and the statistical requirements. Separate measures of response are obtained for each treatment period, forming the basis for the evaluation of effect.If similar prerequisites are applied for series of n of 1 trials as for a conventional group trial-strict entry criteria, uniform treatment procedures, consensus scales for outcome measures, and acceptable statistical tests-individual responders may be identified, common characteristics detected, and conclusions generalised to the target population.4 Experiences in clinical medicineNon-ulcer dyspepsia is a heterogeneous disorder in which patients are unlikely to benefit uniformly from one type of treatment. H2 receptor antagonists are often prescribed for this condition, although conventional clinical trials show conflicting results." In an attempt to indentify individual responders to H2 receptor antagonists in non-ulcer dyspepsia I and my colleagues devised a multicrossover model in which the effect of cimetidine was compared with that of placebo. We used treatment periods of one, two, and four days. The crossover model was later revised to improve its validity,1 and the current version requires that six tablets of cimetidine (400 mg or 800 mg) and six tablets of placebo are given in a double blind, randomised order. To ...

Randomised study of n of 1 trials versus standard practice

“…These common problems lead to huge numbers of encounters between patients and physicians. Despite this and despite endorsements of the technique25 26 27 28 29 30 31 n of 1 trials are rarely used. This is probably because of the extra effort they demand from patients and physicians—which could be justified if randomised studies showed that n of 1 trials result in clinically important benefits over standard practice.…”

Designing robust clinical trials for orphan drugs

“…N-of-1 trials compare different interventions given in random order within a single patient (Fleming 1982;Johannessen 1991;Zucker et al 1997). In order to make more general conclusions, a series of n-of-1 trials may be conducted to defi ne trends (Zucker et al 2010).…”

A pilot study of ondansetron plus metopimazine vs. ondansetron monotherapy in children receiving highly emetogenic chemotherapy: a Bayesian randomized serial N-of-1 trials design

“…This randomized, double-blind, placebo-controlled trial employed a novel study design and statistical methodology: serial N-of-1 trials analysed using Bayesian statistical modeling. Traditionally, N-of-1 trials have been used to evaluate the effectiveness of an intervention in individual patients [13]. We adapted a statistical technique published by Zucker et al which allows for the combination of data from multiple N-of-1 trials to obtain an estimate of treatment effectiveness for a population [23].…”