Simon Optimal and Minimax Designs

As is clear, from the chosen upper limit of a 75 CI used by Gehan (1961) to determine the number of patients to enter Stage 2 of his design, rather arbitrary assumptions are made when developing statistical designs for Phase II trials. Thus Simon (1989) describes two, two-stage designs with somewhat different properties from those of Gehan. He describes a Phase II design that is 'optimal' for Stage 1, in that the sample size is minimised for that stage if the regimen has low activity. The...

Complete Sample

In some circumstances the sample may consist of all the members of a specifically defined population. For practical reasons, this is only likely to be the case if the population of interest is not too large. Clearly complete sampling of a large population will be very expensive and time-consuming. Example - complete sample - reinfection with Lyme borreliosis Bennet and Berglund (2002) studied all patients diagnosed with erythema migrans (EM) following vector-borne infection by Lyme borreliosis...

Types Of Trials

The objectives of each phase, in a somewhat idealised pathway, of a typical development programme for a new drug are summarised in Figure 7.1 and range from studies to determine the pharmacokinetic profile of a drug in healthy volunteers to large-scale randomised comparative trials. However, some of these steps may be taken in parallel and even simultaneously in the same subjects. The nomenclature of Phase I, II and III has been developed for drug development but they essentially apply...

References

Sample size tables for exact single-stage phase II designs. Statistics in Medicine, 20, 859-866. 8 Altman DG, Gore SM, Gardner MJ and Pocock SJ (2000). Statistical guidelines for contributors to medical journals. In DG Altman, D Machin, TN Bryant and MJ Gardner (eds), Statistics with Confidence (2nd edn). British Medical Journal, London, pp. 171-190. 1, 3 Altman DG, Machin D, Bryant TN and Gardner MJ (eds) (2000). Statistics with Confidence (2nd edn). British Medical Journal...

Hospital or Community Controls

Since cases often arise from hospital records, it makes sense to recruit controls from hospitals. Hospitals are a convenient and cheap source of controls, especially in situations where a clinical procedure, such as a blood sample is required. There are two major risks of bias in the use of hospital controls. A risk factor for the study disease may be a risk factor for admission to hospital for the controls. For example, in a study of lung cancer and smoking, the cases are incident lung cancer...

Case Study

To motivate the discussion in this chapter, we use the prognostic factor study concerned with patients with inoperable hepatocellular carcinoma conducted by Tan, Law, Ng and Machin (2003). The aim was to develop a prognostic index (PI), not the very best possible using sophisticated measures, but rather one of (easy) practicable utility. This study comprised two components. In the first, several potential variables were investigated from information provided from 397 inoperable patients with...

Introduction

Cross-sectional designs occur in all areas of medical research, from preclinical, to clinical to epidemiological studies, although surveys, which are one type of such design, are most common in epidemiological studies. Essentially, a cross-sectional design describes a single group or compares two or more groups of subjects with respect to a particular characteristic or characteristics at one point in time. The groups themselves may be formed through an intervention on the investigators' part in...

Matched Design

Matching can either be 1 1 or with more controls than cases. Thus in contrast to the unmatched design, every control is now linked to a corresponding case. The main purpose of matching is to permit the use of efficient analytical methods to control for confounding variables that might influence the case-control comparison. In addition, it can lead to a more careful consideration of appropriate controls. Despite these advantages, matching can be wasteful and costly if the matching criteria lead...