Evaluation of different sample size calculations in clinical trials and app development

• Grant number: 20/00607-6
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: February 01, 2020
• End date: January 31, 2021
• Field of knowledge: Physical Sciences and Mathematics - Probability and Statistics - Applied Probability and Statistics
• Principal Investigator: Fausto Feres
• Grantee: Camila Cristina Francisquetti
• Host Institution: Instituto Dante Pazzanese de Cardiologia (IDPC). Fundação Adib Jatene (FAJ). Secretaria da Saúde (São Paulo - Estado). São Paulo , SP, Brazil
• Associated research grant: 17/50342-6 - Institutional development plan in research (PDIp), AP.PDIP

Abstract

schooling and the prevalence of some diseases in the city of São Paulo are, someway, associated, the inhabitant of this city are the small units. The collection of all these small units compounds the entire population, which is the main object of the study. So, to achieve some conclusions about the population, we should evaluate all of its small units. However, in rare situations it is possible the access all these individuals. We are usually capable to recruit only a small number of them. This portion of individuals, the so-called samples, is the instrument that allows us to make inferences about the population. But what is a good sample size? From one side, a small number of individuals will possible not achieve reasonable conclusions. From the other side, really large sample size is often impossible to obtain. It is necessary to find an optimal and practicable number, which permits, with a good probability, to obtain substantial information on the population. The optimal sample size is related to the study design and its cost. It is also necessary to consider the time available for the study and how much individuals we are technically capable to recruit. This information permits us to define a maximum number of individuals that could compound the sample size. At this very moment, it is natural to make the question: is there a sample size, smaller than this maximum number previously defined, capable to give us strong answers about the hypothesis under the population? The answer to this question depends on some important suppositions. The first one is a preview of the statistical analysis we pretend to apply to the data. For each type of analysis, there is a different sample size recommendation. So, without this information, it is not possible to find an optimal sample size. Despite this fact, a wrong formula is commonly used for this calculation, supposing a statistical analysis that will not be done. The misapplication of these formulas does not allow us to obtain the optimal sample size we should take for making strong and reliable inferences on the population. The main objective of this research project is to evaluate the efficiency of different manners to obtain sample sizes and making easy access to this information through an app. (AU)