Using a prevalence estimate of 8% VI, power of 80%, 20% precision on either side of the prevalence estimate (6.4 9.6%), 95% confidence limits and 10% non-response, the sample size required was 345 individuals.
Accounting for possible missing data and non-responses, the sample size was set to 1,000 per area.
The performances differ according to the nature of the response variable, the sample size, the number of pollutants involved, and the strength of exposure-response association/interaction.
Assuming that exposure distributions are reasonably approximated by lognormal distributions and the strength of correlations among pollutants is moderate, the performances of competing methods differ according to the nature of the response variable, the sample size, the number of exposure variables involved, and the strength of exposure-response association.
Experiments have shown that the internal length scale of the microstructure starts to affect the overall stress strain response when the sample size decreases to the micron scale.
With adjustment for non response (10%) the sample size became 423.
Despite the response rate, the sample size obtained was still large and representative enough for the creation of reference curves.
Given a potential of four hospitals per arm and a 70% response rate, the sample size needed was 150 patients (with a single medical record) per arm.
Our data indicates age as significant factor for treatment response, however the sample size is limited and no final conclusions can be drawn.
Whilst this is lower than anticipated, it does represent a substantial response and the sample size achieved was in excess of that required to provide a confidence level of 95% in the primary outcome, self-assessed knowledge of palliative care.
