Power and Sample Size: Core Principles of Mental Health Research PDF

Power and Sample Size Core Principles of Mental Health Research Learning outcomes By the end of this session you should be able to: Explain the importance of sample size calculations Define Type I and Type II errors List the key factors which determine sample size and explain how they imp...

Power and Sample Size Core Principles of Mental Health Research Learning outcomes By the end of this session you should be able to: Explain the importance of sample size calculations Define Type I and Type II errors List the key factors which determine sample size and explain how they impact on the number of participants required for a study Calculate sample sizes for simple studies with continuous and binary primary outcomes Adjust sample sizes to account for loss to follow up Approaches to sample size Statistical / scientific – How many patients are required to obtain reliable evidence of a treatment effect – if it exists – and to estimate any effect precisely Economic / pragmatic – How many patients are available? – How long will it take to recruit the required number? – How much will the study cost? Ethical – How soon can a trial be stopped to avoid some patients getting an inferior treatment (e.g. placebo)? Credibility – If a trial is very small it may be regarded as unreliable Small studies Will not statistically detect clinically important, realistic, moderate sized treatment effects – Clinically significant, but not statistically significant Produce imprecise estimates of effects – Wide confidence intervals for estimated difference Findings less likely to be true in the population as a whole More likely to result in publication bias – Small studies with p0.05 Inconclusive and misleading for clinicians and researchers Unethical for participants to spend time in a study that cannot deliver a robust result Results from trials of different sizes Two trials of the drug streptokinase for reducing mortality after a heart attack Total Drug Placebo Risk Trial sample deaths deaths ratio 95% CI p value 1st Australian 517 26 32 0.78 0.48 to 1.27 0.32 ISIS-2 17,187 791 1,029 0.77 0.70 to 0.84