 everyone so welcome back to our classes on epidemiology so today we'll be dealing bias in research so we we have covered all the basic chapter of descriptive analytical and experimental epidemiologies so let's see what is bias in research so as the name as the title suggested bias is an unavoidable error in a research so you just cannot do a perfect research a perfect scientist or a perfect investigator is is always at a risk of producing error so it is unavoidable 100% perfect results you will not get in any study so let's see the bias in research so in this class actually we are seeing bias in case control and cognitive study so those studies are more subjected to the attack of bias and descriptive and experimental so descriptive study there is no comparison group so that itself is a big bias so an experimental group also there are many types of bias will arise but the most commonly affected study designs are case control and cognitive study okay so bias is nothing but it is an error but it is in a systematic error not a random error so there will be basically two types of errors the one is systematic and one is random error so suppose let let me tell you one very simple example so you'll get get a idea about bias so we are trying to measure blood pressure of 100 people okay so it's part of our study so blood pressure is being checked for 100 people so by mistake you get a phone call in between and you you didn't record one person's reading properly so that is in a random error so it may it might have happened more that chances of there in all the studies are all the investigation and all the reading measurement but what happens or what if the machine itself is wrong so the entire hundred people are giving wrong results the machine itself is wrong the BP apparatus you're using is wrong it is giving less 20 millimetre mercury extra measurement so the all hundred participants is getting or producing different BP so that is bias it is a systematic error the error will be systematically repeated okay so basically we have three types of bias the first one this are the three common types okay so the first one is selection bias second one is information and third one is confounding bias so selection bias is nothing but when we select participants into case and control or into cohort group if there is any problem arises there it will result in bias second one is information there will be a lot of collection of information in all the studies so if anything goes wrong in information that becomes information bias and last one is confounding our third variable bias okay we'll come into detail one by one so first we are seeing the bias in bias case control study the first one is selection bias in case control study so this is known as prevalence incidence or selective survival bias this is commonly seen in case control study as a result of selective survival among the prevalent cases okay the problem is when we are doing a case control study okay so today we are starting a case control study so it is about cancer and its outcome something so all the cancer patients will be included so the prevalence incidence bias in cancer type of studies the cases will be of the recent type and a very old type that is an incident case and a prevalent case okay so it is a since it is a case control study the cases which we included must be of a recent origin or a long-standing chronic disease which he must have suffering for 10 years or 7 years okay or it could be diagnosed with cancer last week okay so all will be considered as case so what happens is when we include both the type of cases in same study that is this incidence case and prevalence case in the same study the response will be different so this type of mass introduced into case control as a result of selective survival among the prevalent cases okay so prevalent cases means the cases which are being cases for a particular longer period of time okay that duration is a factor producing bias so in selecting cases we have we are having a late look at the disease so the disease will be seen among the patients or diseases occurred among the patients at very different point of time so that will create a bias in our case control study that is known as selection bias that is particularly prevalence incidence or selective survival bias okay so the second one is second type of selection bias is admission rate of bercan Sonia bias so this bias is named after Dr. Joseph Baxson okay Joseph Baxson bias Baxson in bias so who recognizes problem because most of the case control studies will be done in hospital because the cases will be always at hospital cases will not be at public once his diagnosed as case he will be admitted or will be going to hospital but actually it will not represent the general population scenario because you cannot expect the same number of cases in the public okay because the hospitals are always over represented with the cases so such type of bias is known as Baxson's bias or Baxsonian bias so second type is information bias the bias which I raised when we collect information from our case and control so the first one is memory or recall so we ask questions to controls and cases same questions we asked about their process but what happens is the cases are having more chance to respond to your questions because they are actually having the disease not the controls the same questions you asked to the controls they will not have much to say about the disease because they are not having any disease so that way it creates a problem that type of bias is known as memory or recall bias and the information bias another type is telescopic bias this is a psychological phenomena telescopic effect when people ask about a reason incidents we may tell them the events which might have occurred very fast so the telescopic effect is can be seen up to three years so it will be a shift in events between these three years that is ego things which happened past three years will be reported as recent events and the recent events will be reported as very past events okay so that is a psychological phenomena when people are asked about the recent past they might report the events which occurred very long ago okay that is mass telescopic bias another type of information analysis interviewer's bias okay or exposure to suspicion bias if the interviewer is doing study knows about hypothesis he will definitely try to change the result because he is up to a mission of proving his hypothesis so ultimately he wanted to prove something so in that direction the responses will go off definitely he might be changing some responses or there are high chances of changing the responses of the cases so that type of bias is known as interviewer's bias because interviewer knows the hypothesis means it will lead him to question the cases more thoroughly than control he knows whose cases whose controls and he knows the hypothesis definitely there will be chance of bias and the last one is Hawthorne effector observer bias this is also a psychological phenomena when we know that we are being watched we give or more effort in our classroom also when teacher is watching us we try to study well or study more at least we pretend that we study well that is the observer bias when the case and control when a participant is known as they are being studied they automatically try to alter their response because they are being watched or they are being observed that is known as observer bias or Hawthorne effect when human subjects of experiment change their behavior simply because they are being studied so the last one is bias due to conforming of third variable bias this is a very crucial bias so suppose a person who is consuming alcohol chance of conscious team heart disease congenate congestive heart disease okay so we are studying our independent variable is alcohol consumption and this is our dependent variable this is cause this is outcome okay so we are studying this case control study we will get an odds ratio and it says that seven times risk of people who consumes alcohol to get this disease but what is we are missing here is the same person in the habit of smoking that we are not taken into consideration okay the smoking has an effect on this heart disease at the same time people who consumes alcohol tend to smoke more regularly so this effect we didn't take so this effect is concealed here the total seven odds ratio the strength of association is not true but it is concealed the actual effects because third variables are into action so this third variables effects are to be considered when we are doing a research so if you are removing or if you are not including the variables which could affect both cause and outcome that might create a bias which is known as confounding bias so these all third variables which has an effect which has an effect on in a cause and outcome on us confounding factors and the bias arrays due to the confounding factors as well as confounding bias or third variable bias so in case control study what we do is we do matching to avoid this bias so we know matching age and gender matching individual and group matching so case control study we have to follow matching otherwise this bias will arise so next is bias in cohort study okay the same by a selection information and confounding bias so under selection bias we have non-consent bias and missing data bias so in this class I'm dealing only very few bias bias is a very long chapter it can be taken for maybe 10 10 20 hours so that much biases are there but I'm dealing only with case control and cohort study very few biases action bias in cohort studies non-consent bias and missing data bias the non-consent concern biases we know God study it is a follow-up study a same group of cohort will be followed up for a particular period of time so what happens is this may arise because originally selected members of cohort may refuse to participate okay so they have given permission to be under study at the beginning but later they did not give consent so that becomes non-consent bias and missing data bias when we collect when we study records on some individuals are missing or incomplete okay so we take information their complete information is not available or there because it's being followed up at we'll be checking the data at regular intervals so they might not be giving consent or they might not be participating throughout the study so the data will be missing so in such cases the information no such cases the selection of participants will be a problem so such loss of participants will create the selection bias which is known as non-consent bias bias and missing data bias okay so it is due to the follow-up period they give non- consent or the data will be missing so second one is information bias the information which we collect from people is different one example is a diagnostic bias so diagnostic bias is also known as diagnostic suspicion so if you know the knowledge of subjects prior exposure so in cohort study there will be exposure and the beginning of the study will be free of disease then they will be exposed and they'll develop disease after a period of time so if you know subjects prior exposure that will cause a diagnostic bias so such information will be different between the comparison and this cohort group so that create a information bias because the information we collect will be different because of the prior knowledge subjects prior knowledge about subjects I mean knowledge about subjects prior exposure so the same confronting mass will also arise here the factors which affect both both exposed and unexposed group so such factors has to be dealt very cautiously in any study otherwise the relative risk or odds ratio which we get will be totally misinterpreted so the confounding or third variable bias or confounding factors has to be taken care so cautiously in any study so if it is not taken otherwise the results what we get will not be proper so the last bias is postdoc bias postdoc means nothing but we are not getting some association or the result we wanted but we keep on doing dredging we keep on arranging or keep on trying to keep on doing and the dredging or the data dredging and then we try to find out some data to test its significance okay so that is nothing but finding an association by trata data and data dredging and then using the same data to test its significance so the actual data is not used but instead of the data which might give us a positive result is being used from the original data that is known as data and data dredging so it is known as postdoc bias postdoc means after the effect so such bias is also there in God study so we'll have a just to recap the case control study bias is commonly three type selection information and confounding the selection bias in case control or prevalence incidence bias or selective survival then admission or back and sony and bias information which has memory recall or telescopic bias or interview bias or or tone effect or observe a bias and the third one is biased due to confounding or third variable bias in court studies the selection bias is non-consent or missing data bias information biases diagnostic bias then the confounding as and the last one postdoc bias as data dredging bias okay that's all about biasing this is a very brief idea about bias which occurs in case control and court study neither the descriptive or experimental study are clear of any biases all the study reasons will be having n number of biases but we are dealing with very important biases in case control and first study okay so we'll come up with a another class on ophthalmology thank you