 Welcome back to a new session on dentistry and more. So today's topic is matching which is coming under case control study and further under epidemiology. So in epidemiology we have descriptive and analytical in the observational study and experimental study. So analytical study we have two types that is case control and cohort study. So in case control study one of the step was matching. So today's video is about matching, its definition, the types of matching, the problems associated with matching. So let's get started what is matching? Matching we have two definitions. So let's study this definition given by Leon Godis. It is a process of selecting controls so that they are similar to cases in certain characteristics such as age, sex, race, socioeconomic status and occupation. So why we are doing matching is we need to keep equal parameters such as age, sex, race, socioeconomic status and occupation in both cases and controls. So it is not easy way because the age, sex, race, socioeconomic status and occupation such parameters would be different but we need to keep in such a way that we select controls similar age or similar sex or similar such parameters so that we can assume that they are different only in the presence or absence of a disease. Otherwise what happens is all these characters or all these variables or all these attributes will act as a confounder or the third variable. If a confounder is there in the study it will distort the result. The result we get will not be the actual result. So we need to control the confounders. So we need to keep the same effect of all these variables both in cases and controls so that we can assume that they are different only in presence or absence of disease because anyway control is people without the condition or the without the case. So let's see what is design of matching. So matching variables such as most commonly we do age and gender matching. If we are doing a matching based on the age that is if you have a case of 5 years we need to always make sure that we select controls with that same age. So controls can be most commonly matched in two types. One is individual matching and another one is pair matching. So individual matching is very simple sorry individual matching and group matching. Individual matching is just like we are selecting a person with a disease. So we can select a person without disease but with the same age or same gender the same socioeconomic status or same race that is known as individual or paired matching. So we can do paired matching when there is two controls which are usually matched to each case. So we can always we take more number of controls. The ideal ratio is 1 is to 4 but minimum we need to have 1 is to 1. But group matching is a different strategy. We select a population of controls such that the overall characteristics of the case. Suppose if 15% of the cases are under age 20 ok so we are considering only age parameter so the controls also should be under 20. Because same group that is 15% controls we select under 20 age because cases are 15% or under 20. So that process is known as group matching or frequency matching. So we have two types of matching individual or pair matching and group matching. So what are the mistakes we face or the problems we face that is over matching. That is matched only on factors not to be caused of the disease. Suppose we are thinking that age could be the factor which can act as a confounder. So we are not addressing other parameters such as gender socioeconomic status. So we purposefully neglect all those factors and we concentrate only on one parameter which we would have thought that could be the contributing factor or that could be the confounder. So that process is known as over matching. We need to match as much as possible variables. So what happens is if we do power analysis by matching more than one control in general the number of controls should be less than four because of there is no further gain of power above that because maximum is one is to four. So there is no point of keeping more number of controls. So over matching has no statistical power also because the maximum power we obtain with four controls per case. And so other problem is individual matching on too many variables is time consuming costly and cumbersome and may lead to two less controls. So if we are trying to match with too many variables like age, gender, race, socioeconomic status income what happens is it will become very confusing and may lead to less number of controls because too many variables are coming up and which we cannot explore possible association of disease with any variable on which cases and controls have been matched. So what we have to do is therefore only factors which are not to be associated with disease are studied. So we don't need to match for all the variables. And over matching also have problem like matching on variables other than those are risk factors of the disease either in a plant or inadvertently we are doing over matching. Just one example like in a study of OCPUs that oral contraceptive pills as a risk factor for cancer if you use best friend friends as controls it is most likely that the controls would also be OCPUs. In effect we would have matched for the very factors we want to study. Suppose the controls if we are selecting they should be without this factor but if you are selecting controls from the best friends they also will not be considered as controls. And we can say that one another example that is if we have neighborhood control in a study on nutrition and tuberculosis we would be inadvertently matching for socioeconomic status and this nutrition. So always we should select neighborhood controls for a study on nutrition and tuberculosis. So what happens is we are matching for socioeconomic status and this nutrition. So it's are a few examples about matching mostly it comes under a short note for four marks or three marks. So only thing we need to write the definition and the types of matching that is pair matching and group matching and little bit about its problems that is over matching problems. And over matching has more than four members of controls means there is no power statistically. So one or two examples also you can write. So that's all about matching which comes under a case of control study step and further in epidemiology. So I will come up with a new session on dentistry and law thank you.