Planning to do retrospective chart analysis to compare 2 drugs for a common medical condition in terms of there efficacy and safety. What designs should I use, how to decide sample size and what statistical test should I use ?
You can use two samples of patients with a size of 30 and give each type of medication and record each patient's healing time then you can test by using T-test for two independent samples or one-way variance analysis using either spss or minitab
You'd have to have a hypothesis to determine a sample size - i.e. what kind of difference in efficacy do you want to be powered to detect? Just saying you want to compare efficacy and safety is very vague.
With that said, any retrospective study will have a lot of limitations, and there aren't many "design" types for retrospective chart reviews - you just collect data. The only decisions are how much and what data to gather. Sometimes it's as simple as determining a study time period and the inclusion criteria will be "All subjects who receive Drug A or Drug B from 2010 - 2015" and then list all the data you plan to collect.
Depending on your pool of subjects with the two drugs, you could try to do some matching - e.g. age/gender/other clinical characteristics, if possible so that you eliminate as many confounders as possible and the only difference is the drug used.
You can use two samples of patients with a size of 30 and give each type of medication and record each patient's healing time then you can test by using T-test for two independent samples or one-way variance analysis using either spss or minitab