Power statistical

Growing experience with complex disease genetics has made clear the need to minimize type I error in genetic studies [41, 109]. Power is especially an issue for SNP-based association studies of susceptibility loci for phenomenon such as response to pharmacological therapy, which are extremely heterogeneous and which are likely to involve genes of small individual effect. Table 10.2 shows some simple estimation of required sample sizes of cases needed to detect a true odds ratio (OR) of 1.5 with 80% power and type I error probability (a) of either 0.05 or 0.005. 

Allele frequency Dominant model31 Recessive model4  

2) Exposure (= prevalence) in controls assuming a diallelic locus with a dominant or recessive allele at Hardy-Weinberg equilibrium. 

3) OR of 1.5 between cases and controls for possession of at least one copy of disease-associated SNP by case. 

---

Standard deviation, 227—228 Standard error of the difference, 230 Standard values, 249—251 Statistical power, 253 Statistical significance, 227 Statistics descriptive... 

The number of partidpants must be chosen so that the trial wUl have suflEdent statistical power , particularly in the case of confirmatory trials. As a general prindple, the greater the number of data, the greater the confidence there is in demonstrating that a statistical difierence between two groups exists, or not. However, numbers of participants will be constrained by cost considerations, the availability of suitable subjects and, above all, by the ethical prindple that subjects should not be enrolled in clinical trial unless they add scientific value. 

The number of subjects planned to be enrolled, if more than one site the numbers of enrolled subjects projected for each trial site should be specified. Reason for choice of sample size include calculations of the statistical power of the trial, the level of significance to be used and the clinical justification. 

Much of today s instrumentation in principle allows for the rapid acquisition of vast amounts of information on a particular sample. However, the instruments and the highly trained staff needed to run them are expensive. Often, samples are not cheap either this is particularly true if they have to be pulled to confirm the quality of production lots. [See (A).] Then each point on the graph represents a four-, five-, or even six-digit investment in materials and manpower. Insisting on doubling the number of samples N to increase statistical power could easily bankrupt the company. For further factors, see Section 4.38. 

Reconfiguration of Data. Drug safety data from different sources are often pooled or combined in databases. Reasons for combining data vary. In the case of premarketing studies, data from different sites are routinely combined because one site may not be able to recruit enough patients for a study. Data from different studies are often combined to increase sample size and therefore statistical power for detecting an uncommon adverse event. 

Sturm R, Uniitzer J, Katon W (1999). Effectiveness research and implications for study des n sample size and statistical power. Gen Hosp Psychiatry 21, 274—83. 

The eombined ineidence of stomach, liver, prostate, and lymphohematopoeitic cancers was increased among 2,050 male and 1,924 female Finnish workers occupationally exposed primarily to trichloroethylene (Antilla et al. 1995). The workers were exposed principally through inhalation, although there was some dermal contact. The statistical power of this study was low. 

Statistical power is an expression of the risk that the researcher is willing to accept that the study will fail to find significance when an effect is truly present. Usually the number of subjects utilized will provide 80% or 90%... 

One of the main determinants of the number of subjects required to reach the desired statistical power is the precision of the measurement tool utilized. More precise measurements will reduce the number of subjects required. As an example, if a study is being conducted to assess the influence of a dietary supplement on body fat, several measurement tools could be used to assess this outcome. These tools range from low levels of cost and precision (e.g. skinfold measurements) to moderate levels (e.g. bioelectrical impedance) to high levels of cost and precision (dual x-ray absorptiometry - DXA). A study that uses skinfold measurements to measure the outcome will require many more subjects than one which employs DXA. Therefore, it is often less expensive in total to utilize a more expensive measurement tool, because the more precise tool will allow the study to have sufficient power with a smaller number of subjects. 

The number of subjects needed so that a study is likely to have an acceptable statistical power depends on a number of factors, including analytical parameters (precision, etc.), subject selection and control, and protocol design (cross-over, parallel). 

The two major statistical issues in asthma pharmacogenomics relate to population stratification and statistical power. 

Durbin-Watson statistic Works Objective Is statistically testable Can be computerized Has fatal flaw Requires large number of samples Low statistical power... 

Normal distribution of residuals Objective Can be computerized Uses standard statistics Very insensitive Very low statistical power Not specific for nonlinearity... 

---