00 : 00 / 05: 00 (Preview)

This discussion has ended. Watch the recording on Medflix app,

Study my question - How to Choose the Correct Design ?

Feb 25 | 12:30 PM

In the 3rd module of the President's Program on Dissertation-Guidance by IRIA, Kerala, Dr Praveen Nirmalan addresses a common problem faced by new researchers- what is the correct design to study my research question? Join us Live to explore the answers using tips and illustrative examples from real world situations

[Music] good evening my name is dr tanvi and we are gathered here uh for the iria session and i welcome you all uh from the behalf of netflix uh we are gathered here uh on the behalf of ira kerala uh where dr praveen nirmalan would be talking about uh the topic study my question how to choose the correct design uh uh giving it over to dr josie to take it forward this is our third the dissertation program uh by dr praveelan ravine and i'm happy because the ira has taken initiative for uh teaching the residents because most of the residents don't know how to write a dissertation so we are very happy to have dr uh praveen and dr josie is going to chair the session today who is hrd of aleppo medical college which is a government institution and who have got a lot of resistance there so dr josie i welcome you to this platform to check the session and introduce dr praveen good evening everybody thank you man and today the presenter is dr uh nermal uh normal he is uh really he is a super specialist in ophthalmology he is a metro retinal surgeon basically and after that he has taken uh mph from john john hopkins and shifted his watch to research in of talmudgy uh after that in obg anesthesia and finally to radiology a critical care also was one of his interesting areas and the focus is uh more on improving the research methods for clinicians so that the clinicians and the residents helped the residents with their thesis trying to develop a clinician scientist that is what he said and these uh clinician scientists they know the real world problems or who will think and find local solutions for these their local problems and um that is that has been his aim for the last uh 15 years or so and on the topic actually the thesis methods of research methods are usually neglected neglected in almost all conferences now it is very appreciable that uh from the side of the ira officers to include this topic for this issue and uh this may help to enlighten our pgs to have a direction to do their thesis in a better um and interesting manner actually they are doing it uh just for uh for some uh somebody so just for finishing this because that's all this may make them uh these topics make them uh they the um writing of this industry i think so so i welcome dr mullen and all the audience to this scientific session thank you dr jose and uh thank you dr gomez for the introduction thank you nvm team this is the third session of the series 12 module series that we are looking at dissertation guidance as part of the kerala irif residence program and today in this third session we will look at how do you design how do you choose the design for a research question and we will look at a basic introductory approach to this the study designs are usually like an ocean we won't get into the complete ocean but we will start the process of trying to understand how do we choose a design for a specific question for those of you who are new here just a straight recap we looked at the key elements of the introduction how to write the introduction chapter of your dissertation in the first session and in the second session in the previous session we looked at the key elements of a good research question hypothesis and aims and objectives what are the learning objectives for this particular session the basic objective is to understand the basic elements that you can use to choose an appropriate design for your research boost now why is that important why is a steady design important it's only when we understand the importance of the study design that we would look at ways to choose that appropriately and carefully now every question has got a specific design that is ideal to answer that specific question and if something is ideal given that we live in a world of regards there will be an alternate design that's also practical if the ideal design cannot be applied so there is an ideal design there is an alternate design but every question every question that you ask has got a specific design that will help you get the answer for that particular question in available and valid so the study design determines the path for you to follow to get the answer for your question and if that is not appropriately chosen you will end up in going through different violence not on the national highway but to small violence and then if you are using google maps you will get stuck somewhere where there's supposed to be a road there might be a point so this will be like that so the design is important for you to determine the path to find the answer to your research question and that affects the validity and reliability of your results so it's a very important component of any step and it's a very important component of your research what specifically when i say it's important what is it specifically important for so it's important in the sense that every design gives you an idea fair idea of how the study population is identified the study population can be identified as a section of a population maybe in a specific design it includes it involves selecting people with the disease and compare them with people without the disease in another design it involves selecting people who are exposed to a particular factor and looking at what happens to them later on based on the exposure to this fact and comparing that to people who are not being exposed to that side study population can also involve people who have been subjected to one particular intervention comparing that to people with a different intervention each of these blocks of study population fit into a particular design so if you are looking to study something else and you pick a different study population or a different design it doesn't work well it doesn't align so the design is important for us to say that this would be the way this would be the study population that we are looking at the way we are planning to select the study population the study population that we are looking at and that comes from the design the steady design is also important in that it then determines the ascertainments and the time of ascertainment of measures or study measures that are going to be done as part of this step for example it can be a single time assessment on a patient for the study it can be that you will do multiple assessments over a period of time longitudinal you will follow up a particular subject and you will have multiple assessments and multiple periods of time you can't do that you can't do this multiple assessments in all steady distance you can't do a single one-time assessment as part of all study results the study design is important as it relates to the way that we use the methods to find the answers to our questions as well very importantly the study design has a great role to play in the determination of estimation of sample size now sample sizes where we usually end up frowning and whining and feeling sad about everything but the sample size is important in a result story and i'll just very briefly tell why it's important you won't get into an integrity of that now we will look at that in a later session but very briefly why is the sample size important why should we look at it if we do not have an adequate number of subjects if we do not have an adequate sample size we are not able to determine if a lack of difference between groups a lack of significant difference between groups is actually truly a lack of difference or is it that we are seeing a lot test difference because they have not studied enough subjects so if we study less number of subjects we cannot be sure that what we are seeing as an insignificant finding is really insignificant maybe if we have studied enough number of subjects it might become significant so we can't be sure of that on the other side if we study more subjects than what is needed the first thing that we must consider is we are then subjecting an extra number of people to be part of the study when they need not be part of the study for instance if we can if we estimate the sample size as 50 and we study 100 people the extra 50 people actually need not be part of the study but we are subjecting them to this study and that is undertaking we would have got the results valued results with 50 people but we are subjecting this extra 50 people to a study without any particular reason they are not going to add anything to this term it's only trouble for them and trouble for us to try to complete those numbers yeah the second point is that if we study more than necessary subjects more than the necessary number of subjects any small difference can also end up within looking significant so it adds more noise and you might find that even minor differences then become statistically significant so both lower sample size and higher sample size are problematic lower sample size is more problematic than a larger sample size but both are problematic and the steady design is the one that helps you estimate the sample size because the formula to estimate your sample size is design specific and the assumptions that you make for that estimation of the sample size are also designed specific to a large text so that's another importance again the other importance is that every design can give you a certain set of results and we'll look at that as we pass through this particular talk when we talk of results it can be we want to know the magnitude of a condition we want to know what is associated with this condition maybe we are looking at prognosis maybe we are looking at effectiveness of interventions maybe you are looking at what causes this particular condition but for each of these factors we must apply a specific design so if we apply a wrong design we won't be able to comment and reach a conclusion on these factors for instance if i want to look at effectiveness but i designed a study where it's only possible to estimate the magnitude i am not going to be able to comment on the effectiveness effectively comment on the effectiveness and so the design becomes important even for us to understand what results can be obtained from a particular result design and how can we interpret that again design is important because of the bias that comes with particular designs now bias is a systematic error it's an error that occurs in the process of measurement it's a systematic error and different designs have got different inbuilt biases that can occur and you aim to minimize those biases we won't get into the details of that now we'll do that in a later session several other terms come into play an observational study and analytical study a descriptive study and experimental study and we have design specific elements now we have all these terms but let's keep it real for this discussion as i said in the previous sessions the learning is going to be more of a playing rather than theoretical concept so let's keep it clear you have basically initials you have loads of patience time is a factor there's you don't want to spend too much time trying to break all of this down when you feel that you could possibly use that time to indeed to take care of a patient so you want something that can be done easily you want to make sure that it is appropriate yet more easier for you you might be a busy resident lots of work and the whole world is dumping on you everything is dumped on you especially when you are a new restaurant and we have all gone through that phase so as a clinician the key aspect is can i have something i can use in my work to pick a design fast i'm saying this problem i feel like i want to study it what design should i pick and that's the first step what design should i do so can i have something that i can use in my work to pick a design fast let's try to see if we can do that so let's walk through some clinical scenarios and the first clinical scenario is i see a patient with an abnormal finding that i have not seen before this is different from what i have learned and what i expect to see based on my experience and available knowledge and evidence in the literature it's different it's a new find i want to share this with my peers and my teachers and my friends and my juniors of course to share knowledge to clarify to make sure pressure myself that i'm on the right path and without saying of course to also show them that yes i am indeed intelligent you thought i am not intelligent but i really am intelligent watch this so can this be a study so this comes in the realm of a very basic design the case report where we are looking at a single patient a single image or image from a single patient samples from a single patient we are looking at something new and it's an observation we have observing this we are observing this in this single patient and this is a new finding so those are the three three key elements of it is a single person or a set of samples from a single person images from a single person there's something new and it's an observation what does the case report not give us it it does not repeat something that's already known so if it's already known it's not a candidate for a case report case report design it is not useful to report on any interventions it's not accepted to use a case report to report on an intervention unless it's a very very rare condition and you don't get enough numbers but ideally no most journals now refuse to accept a case report based on interventions you really have to justify that the numbers are not there it's a very rare condition and hence and very importantly the case report does not provide conclusive evidence it only generates a hypothesis you are only able to say in this particular person i saw this finding and this is a description of that signing and i can speculate that this is why that is happening and i'm generating a hypothesis for further study i cannot provide any conclusive evidence and the reason i cannot provide any conclusive evidence is because it's a report from only one patient and each person differs from the next person so i cannot generalize and therefore i cannot provide a conclusive evidence there is no formal statistical testing or anything that happens in the case the case report is not part of a dissertation but the case report has value and i will discuss that value a little later clinical scenario 2 it's the same scenario where you are finding something abnormal but it's in a series of patients different from what you learn different from what you expect to see different from what is reported in the literature which naturally assumes that you are keeping in track of the literature and you are reading and reviewing the literature you want to share this now this becomes part of a case series so it's a series it's a report single patient a series of patients a case series again something new it's an observation the question that comes is how large is a series if i take 30 patients is that a series or is that an original research study there is no nothing set in stone there's nothing set in stone that a series is so much so many numbers and after that it shifts to an original research the pragmatic practical way to look at it would be let's estimate a sample size and see for that particular problem and if we have less numbers and that estimated sample says it can become a series if you are meeting that sample size estimated sample size it can shift to an original research step now the k series is like the report it does not repeat something that's already known it does not report on interventions it does not provide evidence conclusive evidence it cannot tell you about the effectiveness of interventions again the caveat there is that if it is a very rare condition then you can use the k series to report on interventions with this by stating the limitation that this is a series we are not able to provide conclusive evidence but we are generating this hypothesis that this particular intervention is suffering so that is the case report and the case series now at the useful when you're setting out to do your study when you're setting out to do your research actually it starts from what you see in a single creation most often but ideally it should start from what you see in a patient and that is your case report and then you look over and see a few more patients of that and then you're looking at your k series and then you're thinking okay i'm seeing this is this study that is this something that i should pursue this is a question that i need to answer so it starts with a single question it moves through a few patients so the case report and the case series are important for you as you are trying to develop your question whether you're a clinician or you're registered your questions come from your patients your samples images that you see they should come from that idea and we talked that in the previous session they should come from that idea and the case look what is the first step the k series even if you're not publishing it the k series is the next step and then you are saying yes there is something in this that i must look at and now you are thinking of taking it up further as a original research step in terms of publication the case report and the case series are transferred lower down on the evidence pyramid but everything on that pyramid from your observational study designs to your systematic reviews your meta analysis if your clinical trials everything on that pyramid actually builds on the case report and the case and the key point is if the foundation is not strong the pyramid cannot destroy if the foundation that is setting out the hypothesis is not strong if the initial observation of the case is not strong then everything after that is not going to be strong it cannot be stopped now the case support and the case series are something that every clinician who sees a patient everyone who sees an image everyone who deals with a sample can actually engage it there comes a time and you see something new there comes a time when you see something new in a series of people and that is something that you can then take yes that is some this is something that can be done by every clinician there is no problem whether you are in a remote area you are in a primary setting you are in the most advanced setting whichever setting you are wherever you are the case report and the case it is directly linked to your clinical observation so this is something that everyone commends and it's a skill that must be learned the description of the case is a skill that must be learned by every resident and young clinician calls you in good step so let's move to the next scenario the third scenario now here the question is that i am seeing a lot of patients of condition x in my practice and having a opiod and having a diagnostic lab maybe i'm having an imaging center name i'm seeing a lot of condition x in my practice i want to know the magnitude of conditions and the reason i want to know the magnitude of condition x can be because for reasons of providing healthcare it also has pragmatic practical applications in the sense that if i know the magnitude i'm able to plan my workload it tells me whether i need to invest more in terms of personal in terms of infrastructure in terms of resources so in a sense i am trying to show you that research in healthcare is not related only to a disease it's also related to the daily routine planning that goes around the running of your setup your clinical setup your public health setup your diagnostic center your laboratory center if you know that you are seeing a lot of condition x you are able to then plan what is needed to identify conditions maybe you need more people maybe that condition x needs a person to be taken to multiple places for further investigations blood tests clinical assessment multiple places maybe so then you need personnel to take them around you need a teamwork you need to invest more and personally maybe you need to invest more on infrastructure maybe you need to invest more on resources it also tells you well if this is if i'm seeing more of this do i need to upskill myself so it takes you to the realm of do i need to improve education my educational skills my training decisions do i need to attend a cme do i need to attend a workshop all those come into play these are things that we do today and expansion decisions do we need to more open more branches in certain other areas because there are pockets of this magnitude in different areas so i'll just come into the picture now the case what do i really want to know i want to know how many people have the condition and when i say how many people have the condition it's not just the people who had the condition in the past even the new cases so i'm looking at equivalence old plus new tests how many people have when i see hundred people how many people have hypertension how many people have type 2 diabetes how many people have tuberculosis how many pregnant women have preeclampsia magnitude in which groups is this condition more prevalent that gives me the associations and why is that important for me so that i can identify the highest groups and say okay these are the groups i need to focus more on for these particular conditions and then streamline my investigation streamline my management processes pathways along that line now how do you study this in a clinical setting in a busy clinical setting so you can study a section of the population in a defined location in a different time period and you want to just know this section of the population in this place in this time period how many people have this condition so it's a one time assessment of each person a single time assessment against the criteria that's being used to define these one time assessment and that's the key element of this particular design it's a single time assessment for each subject what we are not looking to study is what happens over a period of time we don't want to know that now you want to just know what's the magnitude and what what's the association so we're looking at a specific period of time specified location specified population and a one-time assessment we are also not looking to see what causes this condition we're just looking at the magnitude how many people have and which groups are more likely to have so what is associated with this condition and that is a cross-sectional design now the cross-sectional design tells in in its name itself it is across across a section of people across a location across time and it's a one-time assessment so you can understand the magnitude of the problem you can understand which are the highest subgroups and equally important you can understand which test will help me identify the condition better so if i want to compare two tests if i want to compare two imaging modalities if i want to compare two tests diagnostic tests this is the design a cross-sectional list i take people with the disease without the disease i apply test a i apply test b one time assess i look to see how many have the disease how many do not have the disease in the two groups and then i reach a conclusion that test a is better than test b but test b is better than tested or both are nearly similar but the design is a cross-sectional design because it's a one-time assessment the ideal design or the most often use design as a cross-sectional if we have to repeat the test multiple times then the design changes but otherwise one time assessment cross sectional disc and that's a key point to remember in a cross-sectional design it's a simple time assessment for each subject and at say time it's not in terms of am or pm but that for each subject in the study their assessment diagnostic assessments are done once so the cross-sectional design when will i use it to understand prevalence then i want to look at associations when i want to test for diagnostic effectiveness when is it not useful if i want to compare interventions no no no no no no cross-sectional design if i am looking at longitudinal follow-up repeat assessments no no cross-sectional list if i am trying to find out the incidence of a disease no not idea the only place where it's ideal is when the condition is very acute and the duration of disease is very short let's say it's like a food poisoning it's here today it goes off tomorrow in that case because the duration of disease is very short the prevalence will be equal to the incidence so that is you can use a cross-sectional just but otherwise you have no incidence you cannot use a cross-sectional decision why is that so because we are doing a single time assessment at that point of time we are not doing a repeat assessment so people what is incidence new cases so we are looking at people who don't have the disease on that one time assessment and then we follow them up to see who gets the disease that's not possible in a cross-sectional descent that's not a cross-section so that goes out of the picture [Music] can we say that factor x causes this is why no we can't because again we are looking at a one time assessment at that one time assessment all that i can say is that when i found out that this person had this disease they had sacraments i cannot say that fat drugs came before the disease or the disease came before factors all i know is that at that particular time both of them are there and there is an association that's all that i can do so that's the limitation of a cross-sectional research so if we want to look at causality we cannot use so it's important to know if your question is looking at comparing interventions rule out a cross-sectional if it's looking at follow-up repeat assessments longitudinal data no cross-sectional if it's looking at causality no cross-sectional risk if your question says i want to understand the prevalence think of a cross-sectional list i want to look at an association think of a cross-section i want to look at diagnostic effectiveness of tests think of a cross-sectionals where can we do it we can do it in a hospital we can do it in a community in a population in a lab diagnostic facility screening camps outreach literally anywhere and it's a favorite dessert for the dissertations one of the most common designs that are used for a dissertation is a cross-sectional design and that's a very pragmatic choice it's a pragmatic choice because it's a single time assessment you don't have the worry of not it running behind follow-ups not getting follow-ups missing that all those things don't come into picture if you devote your time at that single point that one-time assessment you get enough data and you can run a clean cross-sectional study decision but it also limits the type of questions that you are then able to address as part of your dissertation so that's something that you have to keep in mind but it's among the most favorite designs for a dissertation because it's something that can be completed during the duration of the research let's look at the fourth clinicals here i see a lot of patients with clinical symptoms and signs suggestive of fatty liver i would like to have a more quantifiable method process category i'm using clinical scores i'm using biochemical markers not helping me biochemical markers turnaround time is high not all of them are able to do it clinical scores are not available i want something more quantified so that i can monitor progress with the therapeutic advice that i provide my workplace has got an ultrasound unit but they're all over because they do the usgs for all the cases in the hospital so what i want to know is whether i can choose high risk groups to send for an ultrasound rather than sending everyone for an ultrasound can i choose a certain subgroup of these people that are clinically suspect to have fatty liver and send them for an ultrasound [Music] i want to know what clinical factors are associated with the positive ultrasound exam so that i can do a targeted referral of high risk patients so how can i study this we can do a prospective study we can say okay let me do that prospectively and i send all patients for usd a certain number of patients for usd and study factors that are associated with usd positivity for fatty liver the problems there are problems related to time i need to run it prospectively i have to send everyone there there is a time factor it might take months it may take a year depending on the case load so it's a problem sometimes there is an element of cost who pays for all of this do i make the patient pay that does the department pay for this where do i get the money for this ultrasounds who pays for it do patients agree with it when they say oh you're just sending me there so that you can make some money on this side all those kind of questions come into it equally important there is a workflow overload that comes in for this oncologist as well as for the refereeing clinician the sonologist has to do a lot more now ultrasounds the reference clinician has to look through a lot more reports most of which may or may not be normal so there is an increased workload so let's be pragmatic and we say we are doing ultrasound societal liver already we have a pool of patients who have had ultrasound for a fatty liver why don't we dive into that food and then see what's happening how do we dive into that pool we say okay i'll pick people who have fatty liver on ultrasound that's my cases so what's happening here the outcome has already occurred the outcome is sonographically identifiable fatally that's okay that's your case and then look at who does not have fatty liver on ultrasound and i call them as my controls so where the outcome has occurred already that's my case and the key point here is the key phrase here is outcome has already occurred and that's my start point for this step i now go back into our records and say okay among the people who are my cases what all clinical factors are present among the people who have identified as controls what are the clinical factors that are present and a look at an association which is more in this particular group compared to that particular for less than this particular group compared to the other group and i'm looking at an association the advantage is that i don't have to study all patients i can choose a specific number of cases and controls for this and that's an advantage or a cross section so in a cross-sectional study i will choose a cross-section of the population maybe there are 20 cases let's say i've chosen 100 people maybe there are 20 cases and the remaining 80 are normal controls i will end up doing 80 ultrasounds and normally if i do a case control design again i am just putting out numbers there i might end up doing 20 cases and equal number of controls so 20 20 overall 40 ultrasounds as opposed to having to do 100 ultrasound so that there is an advantage there for the same validity of research there's an advantage there are problems with this particular design is the case control design and one of the major problems is that choosing the control is not easy it's easy to tell that we choose a person without sex without a disease or without this without that in practice it's a very very difficult process to choose an appropriate fund and it leads to a lot of problems so the case control design sounds very simple i identify cases people without the disease controls what's there when you start doing it it's i get the cases but who are the controls here are my controls and becomes a real problem and becomes problem for several reasons including the progression of disease are there truly without the disease of that particular problem is it subclinical all those things come into picture so where can it be done where can we do a case control study you need a healthcare setting with good documentation to do it you need a healthcare setting with good documentation because you are going back into the case of course and you are going to look at that case request to see what's been done and what are the clinical factors infrastructure diagnostic definitions criteria for identification assessments all those things coming you need good documentation so when can you use a case control design outcome has occurred that's a start point think of a case control design if you are looking for an association what are the risk factors for this particular output think of a case control list if you are looking to say that this risk factor causes this outcome no so where can you not use it if you're looking for causation no because the outcome is so good and you don't know at that particular point all that you know is that this risk factor is associated with that outcome you don't know which came first in the case control distance so you can't talk about causation if you're looking for magnitude if you also want to report the equivalence or the incidence you can't because you are choosing a set number of cases and control so you can't really tell the incidence or prevalence in the population and this is an error that keeps coming in dissertations we are in a case control design you will have the statement that the prevalence of the condition or the incidence of the condition is x percent not correct you can't do that in a case control decision can you report on interventions no why because it's retrospective you're going back into records interventions will be unstandardized who is getting allocated to what intervention is not controlled is not standardized so it's it's a problem you don't use it for interventions longitudinal data you can't really do it because you don't know whether the exposure whether the risk started first this factor came first for the disease longitudinal data no let's look at the fifth clinical scenario so we had looked at the case report we looked at the case series we looked at the cross-sectional design we have looked at a case control design the basic elements only the key important elements i am telling out here the fifth clinical scenario is something that we see quite often now i see a child whose parents bring him saying that he has disturbed sleep and drops his eyes and blinks more in data they tell me this has started in the past couple of months after online classes started they say he is almost always on a digital device school work homework no social interaction i talk to my friends using the device for relaxation watching movies playing games all those things your life is now embedded in a digital device parents advise him to reduce screening and they're brought in disturbed sleep he keeps on blinking rubbing his eyes and the child asks me what happens if i wash the screen for long why are all of you telling me to reduce my screen what happens if i watch the screen for now so in a sense the child is asking me what are the multiple outcomes that can occur if i watch the screen for a long time now you can apply that conceptually to any question any problem that if i am exposed to ask this course what happens what are the multiple outcomes that can happen if i am exposed to tobacco what are the multiple outcomes that can happen so what are we looking for we are looking for a single exposure passing multiple outputs and the exposure occurs before the outcome so at the time of starting the study all that we have is that this particular person is exposed to this fact the disease has not occurred and we compare exposure with non-exposure we are looking again at an association now in this case we are following up longitudinally we can say that exposure occurred first the c is occurred later and the exposure causes the disease and you can look at multiple outcomes and you know if you're looking at tobacco maybe we can say there is lung cancer maybe we can say there is increased gastric acidity maybe we can say there is something else maybe we can say that sleeplessness maybe we can say that there is oral gravity oral problems all those kind of things we can just multiple outputs equally important this is the design where we can say at the beginning of the study the person did not have the disease and later developed the disease so we can really look at incidence so we are looking at two key points here one is causation incidence and association of course so when can we do a cohort design if you are looking at multiple outcomes from exposure tick cohort design if you are looking to report on incidence of outcomes your incidence of disease stick cohort this if you are looking to state conclusively that exposure to this factor causes these diseases it's a covert disease choose a cohort you're looking at longitudinal data choose a performance if you're going to report on prevalence no the core design does not help you if you're looking at interventions again the covert design is not the first choice if you are not able to do a clinical trial then you can look at a cohort design as an alternative but it has its limitations so these are the places where you can use a copper design you can use it in a healthcare setting you can use it in a community or population based setting the sixth clinical scenario is i have been prescribing medicine x for disease y i hear a talk by an expert that says medicine b is new and better medication or it can be an in any other intervention now should i shift to medicine b how do i decide can i study that on my patients do a small study on my patients and see what's happening then how do i design them and this is something that we see more and more coming diagnostic effectiveness of this is something that we see coming in more often as technology advances as intervention substances medications advances this is also a very commonly recurring scenario in clinical practice at whatever level you are so what design can we do what we want to check is whether medicine be the experimental medicine work spectrum or the effectiveness of medicine b compared to medicine x which is a current standard for disease y it's a condition of interest so that's what we want to do and for any interventional study the gold standard is the clinical trial the clinical trial has got multiple ways to multiple pathways multiple ways to do a clinical trial multiple types of medical trials but we'll do that we'll do a separate session on clinical trials alone multiple ways to do a clinical drive that is the goal standard for any intervention if you are looking at an inter reporting and intervention trying to convince someone that this intervention is better than that intervention you have to do you must do a clinical trial if you are not able to do a clinical trial then your next alternate best design is a cohort study but it's not as good as doing a clinical now maybe you're just taking this particular intervention on all subjects in that case you can still do a pre-post pre-intervention compared to post intervention first intervention compared to pre-intervention an interventional study design pre-post intervention study design where all subjects get the same intervention so that's also possible but for interventional studies clinical trial is the world standard and many most journals now don't accept if you don't do a clinical trial for that you'd have to spend quite some time trying to convince the reviewers why you did not do a clinical most journals of a good standard so if the key points here to note if your question is saying is bringing in that you're going to compare an intervention you can't think of anything much other than a clinical trial if you're looking at longitudinal data yes you can use a clinical longitudinal data as part of an intervention after the intervention causality yes you can look at causality in terms of this intervention causes these outcomes instance of outcomes yes you get it because this particular intervention causes these outcomes and these are new outcomes and hence i can get the incidence of outputs can i report on the prevalence of the condition no i can't really do that using a clinical term because i am choosing a section of the population selecting a section of the population for inclusion in the clinical trial so i can't really generalize that number that i get to the general publishing skill is incredible and uh actually it is a highly productive talk um and one question you please accept that okay you said one number of findings for k series that i couldn't understand can you can you repeat that one patient for k series you put one k one patient with number of findings that's why i couldn't follow that sentence k series k series is one a series of patients with new findings with new findings series of cases [Music] any other questions any other questions no no no so sir you please accept the sincerest thanks for the great presentation um that you have done and uh the lecture that you have dealt with anjali dubai has asked a question if the study is retrospective and prospective both you can't uh you can't really do a retrospective on prospective study at the same time what's your what about i'm perspective what happens in some instances is that when we are when we have a wealth of retrospective data we start from a retrospective period and then we continue prospecting yeah yeah so that is positive that's coming as part of the newer designs that are possible now you can do the same thing with the case control design the conventional attitude towards the case control design is that outcome has occurred and you go back into a guest record so it's retrospective and you're looking at it you can now define the case and then go prospectively we can say okay this is my definition of the case and then start i'm starting from now and i'm going to pick my cases and when i pick my cases i will pick my controls also at the same time so i do need not go retrospectively i can go prospecting and one reason why this is coming in is because the way we are diagnosing is changing very fast and the classification and criteria used for diagnosis is also changing very fast so if you want to go with perspective let's say even something like hypertension the definitions used five years back differ from the definitions used one year back the methods of assessment differ so we are not able to go back in records for so much time earlier it was not a problem there was no internet there was no faster renovations we could go back five years we could go back six years we could go back ten years but now it's not like that we are getting a lot of biochemical markers coming in we are getting lot of newer techniques coming in something like rear clamps are the definition changes fetal growth restriction definition changes immediately changes so then going back retrospectively becomes a problem so what we then tend to do is that we take a shorter period of the prospective one where we get good data and then we can move prospective those are part of the newer adaptations that are coming because the clinical scenario is changing around it okay so one more question is there but i think this uh not uh reliable not related to this topic so how can we calculate uh the sample size i think that will be you'll be doing that in a separate session we have a session plan where we are looking at uh how to calculate the sample size okay [Music] we are applying tumor markers on our previous histological samples if you have a way to do that and you have a way to do that reliably that can be done we can look at a study that is doing that and then we can report on that but again it would be more in the realm of why are you doing it it can be in the realm of a cross-sectional study where you are trying to say that i want to know if i apply this particular tumor markers now that's the magnitude of the condition or my pickup of the conditional diagnosis change can be the realm of that or you can say that applying this on the previous samples and comparing with the previous method that i used you can look at the diagnostic effectiveness so again it can go in the volume of cross-sectional steady decide okay okay sir i i think uh that's the end of the questions yeah so we may wind up you please as a series of lectures so thank you so much thank you everybody

BEING ATTENDED BY

Dr. Murtuza Zozwala & 293 others

SPEAKERS

dr. Praveen Nirmalan

Dr. Praveen Nirmalan

AMMA Center for Diagnosis and Preventive Medicine | Research Mentor, AMMA Healthcare Research Gurukul, Kochi, Kerala

+ Details
dr. Ramesh Shenoy

Dr. Ramesh Shenoy

Consultant Radiologist | Kochi

+ Details
dr. Rijo Mathew

Dr. Rijo Mathew

Consultant Radiologist | Kochi

+ Details
dr. Gomathy Subramaniam

Dr. Gomathy Subramaniam

Professor and HOD, Radiology, Malabar Medical College, Kozhikode

+ Details
dr. Praveen Nirmalan

Dr. Praveen Nirmalan

AMMA Center for Diagnosis and Preventive Medi...

+ Details
dr. Ramesh Shenoy

Dr. Ramesh Shenoy

Consultant Radiologist | Kochi

+ Details
dr. Rijo Mathew

Dr. Rijo Mathew

Consultant Radiologist | Kochi

+ Details
dr. Gomathy Subramaniam

Dr. Gomathy Subramaniam

Professor and HOD, Radiology, Malabar Medical...

+ Details

About Medflix

Medflix is a new platform by PlexusMD, India's most active and trusted doctor community. On Medflix, you can discover live surgeries, discussions, conferences and courses from some of the top doctors and institutions across the world. Join clubs in your areas of interest and access hundreds of amazing live discussions everyday.