In this study, the population was 1269 and 791 workers returned the questionnaires, which was over the predicted sample size of 364.
Given that our predicted sample size for patients (N=400) was double that for nurses (N=200), it was decided to double-weight the patient model.
Assuming a prevalence of 50%, an alpha of 0.05 and a 95% confidence interval, the predicted sample size was 104 for the training cohort [ 29].
As this power calculation assumes a normal distribution, and we expected skewed samples, we doubled the predicted sample size from 30 to 60 to account for non-normality of the ratings.
These prevalence estimates can be used in predicting sample size requirements for future epidemiologic studies to evaluate genetic determinants of susceptibility to chronic and infectious diseases, the severity of disease, and interactions with other risk factors.
Participants who fulfil all the eligibility criteria and provide written informed consent are allocated to either RYGB combined with optimised clinical treatment (OCT) or OCT alone, in a 1 1 randomisation ratio basis, with a predicted sample size of 60 patients, 30 in each group.
Even if the current trial continued until the predicted sample size was achieved (100 per arm) and the remaining ~40 patients per arm that were recruited during this period, demonstrated a 20% rate of healing in the clinic and a 40% rate of healing in the home, the final healing rates of ~52% for clinic and ~45% for home, would not be statistically significant.
Based on a predicted sample size of 150 170 cases (expected number of cases based on STEC cases notified nationally in two preceding years) and an attempted case to control ratio of 1 3, there was sufficient statistical power (at least 80%) at a confidence level of 95% to detect an odds ratio of 3.0, using 5%, 20% and 80% as the expected frequencies of exposure among controls.
Finally, we show that theoretically predicted sample sizes using the so-called S-estimates provide a good lower bound.
The predicted sample sizes to detect a gene conferring an Odds ratio of 3.0 were 99 for case control and TDT methodology, 148 for S-TDT, and 107 to 173 for an ASP study assuming three major loci.
