Hello everyone. In this lecture, we will talk about the ethics involved in conducting power and sample size analysis. We will talk about basic ethical concepts that underlie our study designs, the three key concepts when considering sample size, understanding the dangers regarding ethics when studies had too much or too little power, and why it is important to plan for sample size and power analysis early. First, let's go over the ethical issues we need to consider in our study designs. Informed consent is essentially telling your participants what is going on with the study, what is the purpose, what are the risks and benefits, the value. It also tells participants about the sample, especially if it is from a difficult to recruit population or if researchers are playing on recruiting too small or too large sample sizes and the reasons behind it. Equipoise means that we as scientists are genuinely uncertain about the treatments we're studying. It would be unethical if we conducted a study where this was not the case, where we were convinced of our answer before we present a question. It is also worth noting that equipoise is different from the null hypothesis which is the assumption made for the purpose of statistical analysis such as the effect of treatment A is equal to the effect of treatment B. So, equipoise can also be described as a state of genuine belief that the null hypothesis could be true. Next, let's talk about the notion of therapeutic misconception. This occurs when participants confuse treatments with research. We know the difference, treatments are proposed solutions to problems that we want to test while research is conducted to develop solutions to problems in advanced society in scientific knowledge. Obviously, it's not always black and white, but that's the divide that we begin with. When we look at the difference between treatment and research, treatments are not always known to be effective necessarily especially not regarding specific individuals at times. Often, treatments are provided simply because there's a good reason to hope they might be effective. But that's still different from research, where there must be equipoise as we just discussed. The bottom line is that it is important that participants understand the research does not promise benefits to the participants. Will lead to therapeutic misconception. One factor could be that an investigator may be overly optimistic about the benefits of participation in her research. Another could be that a participant may lack scientific background necessary to fully understand risks, benefits, and limitations of the research. This is why a well-written informed consent is imperative to a study. Essentially, ethical balance is maintained if the value of research to society is greater than or equal to the burden of the individual research participant. The burden to the individual is the projected risk, discomfort, or inconvenience to the participant. The value is the projected value of the research to society as a whole by advancing scientific knowledge, discovering new treatments plus any potential benefits to individual participants. Maintaining this ethical balances is important responsibility of researchers. It is important to describe the benefits that research participants can accrue. These can include the overall notion that research can help people in the future and benefits society. The possibility that the treatment may be effective for the individual participant and there are can even be payment for participants for certain studies. Studies providing burden to the purchase fence with minimal or no benefit to society are unethical. In the contexts of sample size, when the sample is not large enough to answer the scientific question, research generates net burden when the total projected value of research is less than the value projected burden to the individual. Ethical balance depends on the cumulative clinical or scientific value of research to the individual and to society. A study generates an individual net burden may still maintain ethical balance if the cumulative benefit to society is significant. If there is very minimal or no projected burden to the participant conserved minimal risk research, then sample sizes ethical that is appropriate to answer the research question statistically in real-world research, as the ethical impact of sample size is less. Most proposed research studies involving human participants must be approved by an IRB. Institutional Review Board or IRB is responsible for reviewing scientific efficacy and ethical concerns including safety and proposed studies. Benefit risk balance is the basis of Institutional Review Board approvals and rejections. The IRB evaluates the risk to benefit ratio of each study. Often, the IRB will make recommendations for how to make adjustments to the study protocol to achieve balance rather than simply rejecting or approving a study. Most commonly these adjustments will relate to the informed consent process. It is also possible though less common for an IRB to make recommendations regarding sample size or other issues with the protocol. Now let's discuss ethical dilemmas related to under-powered and overpowered studies, starting with under-powered studies. Under-powered studies may result from a failure to distinguish between efficacy and effectiveness. Efficacy is a measure of mean treatment effect in a perfect world. Effectiveness is a measured outcome. Note that non-interference of participant to the sign regimen is just one potential reason for difference in measured outcomes in theory and in the real world. Also note, there are other reasons why study might be under-powered: early results that were promising but end up being wrong, grantsmanship considerations favoring optimistic assumptions, funding restrictions and more. More cynical reasons might include to avoid detecting differences one doesn't want to find such a safety differences that don't favor a new drug for example. For example, if power is calculated assuming perfect adherence to a signed treatment such as an assigned drug for a medical study, the study will be under-powered as adherence is almost never perfect. To make sure the study is ethical and not under-powered, the researcher should calculate power using real world adherence rates. Why is this important? Because under-powered studies are unethical. If you reduce the potential to find to clinically significant differences, you can't answer the research question. Therefore, there's only risk to individuals with no benefit to society. You are placing burden on participants without adequate benefits to society. As we talked about in this course, it is imperative to account for potential loss to follow-up in your consideration of recruitment and sample sizes. Remember that it typically is unethical to choose to do a trial that you know is under-powered. It could be difficult or impossible to get large enough sample sizes such as the situation where a population doesn't trust the researcher or researchers. In this case, a different study design may be necessary. As you can see from this quote from Lenth's 2001 paper, underpowered studies are still common. Researchers conduct underpowered studies because it can be difficult to recruit participants. It can still contribute to scientific knowledge in ways especially if it can be combined with results from other studies and it can at least inspire discussion for future studies. If your plan is to combine results with other studies, it really needs to be designed with that in mind. There are also ethical issues regarding studies that are overpowered. The average projected burden per participant remains constant as the sample size increases. As the number of individual burden increased, the cumulative burden increases, while the value to society does not increase further. This results in an unethical imbalance between value to society and burden to individuals. Overpowered studies can result in result that seem to be significant, at least statistically, but they don't really matter in terms of benefiting society. Overpowered studies can also be found especially in drug development. Why? For cynical reasons, which include the desire to detect statistically significant difference for FDA approval and marketing purposes. This is of course, unethical and abuse of responsibilities held by researchers. As we know, studying multiple aims and recruiting the maximum sample size of the aims can result in an increased sample size. As we have discussed in this lecture about handling multiple aims, researchers must consider how they should best balance the risk and rewards in dealing with multiple aims and making decisions regarding sample size and power. Getting input from participants can be helpful if there's unethical issues to consider. In conclusion, studies that are conducted ethically have sample sizes that are adequate for answering the research question. Studies are not conducted ethically when you have a sample that is too small to result in significant value for society as the researcher doesn't answer the research question while there's still a burden on individuals, or a study is conducted on ethically if it has a sample that is too large resulting in a larger burden on individuals. That's it for this lecture. Thank you for your time.
