I would like to conduct a clinical trial to do a head to head comparison between two drugs for safety and efficacy. How can I estimate the appropriate sample size?
Hi Famil, you'll first need to decide what your primary endpoint is - usually a measure of efficacy. It's usual to calculate sample size based on your primary endpoint only. You'll need to examine the sample size you need to have to get a good chance of detecting a difference between treatments if a difference truly exists. This is termed the "power" of your test (or study). Commonly you'll want at least 80% power to detect a difference between treatments at the 5% level of significance. To estimate this, you'll need to have an estimated measure of the variability (standard deviation) in the endpoint data you expect to see (maybe from a previous study you've conducted, or you may be able to estimate this from a published study measuring the same endpoint in a similar patient population if they report enough information associated with their statistical tests). There are formulae available to calculate sample size based on the power and significance level you're looking for, but you'll also need to have an estimate of the magnitude of difference between treatments you're looking to detect. You should consider this by thinking about the size of difference that would be meaningful to patients (sometimes called the minimal clinically important difference) - they'll be no point detecting something smaller than this, but you may elect to look for a difference larger than this value. You may want to read Article Sample size calculation
as a starting point.In terms of safety - many safety measures - such as adverse event incidence - are unlikely to be powered to detect differences between treatments, unless they are reasonably frequent. Often, safety data are combined across studies to give better abilities to tease out incidences - particularly in less common AEs.
I think the best next step would be to chat with a statistician at your institute as they will be able to help you work through this, and will be able to help calculate the sample size for your study and estimate the inputs you need for that calculation based on previous study data or published studies.
Good luck with your project.
You can use the following formula:
Sample size from finite population can be estimate by:
n= N / ( 1+ N * e^2 )
and from infinite population by:
n= ( Z^2 * S^2) / E^2
where : Z ( 1.96 for 0.05 and 2.58 for 0.01)
S = standard deviation from previous studies or pilot study
E = significant level
Also, you can use software to calculate sample size (for ex. SPSS).
My approach is: You need roughly 20 cases in each treatment group to get a reliable result. However, in one of the treatment groups all cases must reach the primary end point. So, what you need is the number of cases that will give approximately 20 cases reaching the end point. Large studies (
Sorry, it should have been >1000 cases in my response above. In some trials with nearly 3000 cases (anti-fracture trials), the number of cases reaching the end point (femoral neck fractures) have been only just sufficient to reach a statistcally significant difference between the treatment and control groups. One case more or less in one of the groups would have caused the firm loose a lot of money.
The clinical trial objective and design rather broadly dictates the sample size. The important basic considerations would be:
1. Therapeutic efficacy: Superiority or non-inferiority? (vis-a-vis comparator drug)
2. Safety: Better tolerated or similar to comparator drug?
One can rely on previously reported trials, if available to construct the suitable hypothesis/questions for the study.
And of course, as mentioned by previous experts, once you have clarity about the study objective/hypothesis to be tested, statistician's help can be sought to arrive at the exact number of patients required for the study.
Actually both the safety and efficacy is checked in animal models before approval by the FDA then there is post marketing surveillance for any unwanted effects. How will you measure safety in human clinical trial what further you want to see.
Safety is measured either by therapeutic index or therapeutic window . Both of which are not allowed in human because 1. minimum effective dose will not be sufficient for all participants while minimum toxic dose will not be allowed(Therapeutic window)
Effective dose 50% and Lethal dose 50% are not allowed in human studies(Therapeutic Index).
Comparison of efficacy may be conducted but it needs large sample size .
My sincere advise is to make it as a pilot project on some what 25 for each drug group (Total 50)
That can further be extended in case of some positive findings
Preclinical studies in animals are designed to evaluate pharmacodynamic/pharmacokinetic and toxicity profile of a drug. These battery of tests however do not assess clinical efficacy and tolerability. For this, one needs to embark on clinical development studies in human which are usually done in phases:
Phase I - Initial safety/tolerability studies in human volunteers (First in humans). These include human pharmacokinetic studies.
Phase II - Pilot/ Exploratory trials in small number of patients for efficacy and safety. (First in patients) as well as dose-finding studies. These may also include placebo-controlled trials.
Phase III - These are confirmatory trials in a larger number of patients to assess efficacy and safety.
Phase IV - These are post-marketing studies/surveillance undertaken after approval/marketing and are intended to assess efficacy and tolerability in real-life clinical practice situations.
The above phases of clinical trials are so structured that both therapeutic efficacy and clincal safety are evaluated in a step-wise manner backed by clinical expertise.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size