Computer generated transcript

Warning!
The following transcript was generated automatically from the content and has not been checked or corrected manually.

Hi, everyone. My name is somewhat selfish. I'm a third year medical student at the University of Manchester. I have a strong interest in academic medicine, Uh, mainly for the reason that you can influence the patient population as a whole, but at the same time, Uh, still stay in touch with clinical practice and more importantly, translate all of the work that you're doing, Um, in the lab to your patient's on the clinical side. So, um, in the journey to guess getting publications and building a strong enough CV for the foundation program as well as applying for a PhD, I feel as if I've had enough experience. Um, that would be worth sharing with all of you. Um, so this initial video will be about that would be a second. Yeah, there we go. This initial video will be about research terminology and types of research, But this video series, um, it's just gonna be two videos. Um, will be about the basics of clinical research and portfolio development. So I think this would be really useful for anyone who's interested in applying for, uh, either a PhD or, um, an interrelated degree or even a probably SFP or Academic foundation program, as it was previously known. Um, so I think, um, it will also be useful for anyone who's wanting to get more publications to gain points on their foundation program application, because that can definitely boost your score and help you get into the Dean ary or hospital that is your first choice or your top choices. So the importance of taking part in research Um, I just mentioned that I want to be able to translate the work that I do in the lab to the patient's, because I I feel as if that's extremely satisfying. So there's obviously the aspect of having a broader impact on the population as a whole, but at the same time having the individual interaction with patient's um, that's also equally does fulfilling. You can also get an in depth understanding about the topic which interests you. So it's a way to follow your passions. Uh, you can develop your scientific, rising ability, um, as you will be publishing at some point along with the writing ability. You'll also be developing data analysis skills and UM, software programming skills such as R and Python, which is going to be extremely useful in the real long one as, uh, academia tends to use larger and larger data sets. And in general, they're emphasizing the use of reliable data and drawing conclusions from large data sets, so such skills would be extremely useful. I think most importantly, we'll be building a network. You can make new friends and also meet mentors. You can guide you, um, according to your specific goals passions. And it helps you build a strong application foundation, years core training years and even beyond, even if you're going abroad, Uh, publications and experiencing research account for quite a lot, So I want to. I won't cover for too long on the contents page, but it's split up in this video, split up into two different sections. The first is basic terminology, and the second is the different types of research, their advantages and disadvantages. Um, so a deal. Is it a good understanding of the basic terminology like, for example, what primary secondary researches and what prospective and retrospective studies are? Understanding of such terms is very useful for, um, understanding how to interpret especially types of studies. Primary research. As the name suggests, it is essentially original reports of research studies. They're generally authored by the researchers themselves. Uh, it has original research data, which is the main subject of the paper. Um, and it's almost always published in peer review, General, because, uh, the funding required for primary research is obviously quite high. As you need your own team, you need a lab access to resources. So not a lot of people can publish primary research. And when it is published, um, it will most likely be published in a peer review general. For that reason, examples include lab research and okay, so dissertations, technical reports and conference proceedings. That's secondary research. Um, let's take an example where, uh, perhaps there's a lot of primary or original research on stem cell therapy. It will be extremely useful for the researchers or the people who are interested in the field. If all the information was on one paper as a summary. So literature reviews um uh, one approach to doing this, and it's just a useful summary of all of the research that's out there so that a clinician or academic than refresh or update their knowledge, um, on a particular topic so it could be a systematic review or literature review, and it could be critical appraisals on original research exposures and interventions. As the images suggest, they could be environmental factors, or they could be, um, treatments or therapies. So the independent variable that you're measuring the effect of is generally going to be the exposure or intervention in clinical trials. Taking these envelope retrospective data The data accumulated from the past is what restaurant retrospective data is. Uh, there may be multiple exposures or interventions that could be linked to specific outcomes. A study main focus on, um one exposure or intervention. For example, let's say the water supply to a small town was contaminated with heavy metals. What will be the long term effects of this on the town's population? So in this case, and the exposure or intervention is heavy metal contaminated water. So the prospective or retrospective studies? Um, firstly, let's take the prospective studies. They generally ask how a specific exposure or intervention affects certain outcomes. They recruit participants, um, and subject them to an intervention or exposure. Then they measure the outcomes of interest in these people over the following months or years. Um, so retrospective studies um, looking at people's medical records or asking about past events? Um is one way of collecting data from the past, Um, essentially, by interviewing them or just looking at the medical records, it begins after outcomes have occurred and look looks for possible past exposures or interventions that may help explain the outcomes. As you probably know, having a control variable or control group is extremely important for any experiment so that you can see whether the independent variable, um, has a significant effect or the variable changing has has a significant effect. Um, so this is done so that well, ideally, um, sorry, just a something. Yeah, So you need to have something to compare the intervention group to. Ideally, the group should demographically be, uh, similar to the intervention group. Um, such that only the variable, um, that you're measuring the independent variable is the one that you can, um, measure or isolate the A pretzel. Confounding variables. If I was to explain this initially using an example um, using the diagram that's already on the slide. Uh, let's say you collect data on sunburns and ice cream consumption. You find that higher ice cream consumption is associated with a high number of sunburns. Uh, so does that mean that ice cream consumption causes sunburns? Obviously, here, the confounding variables temperature Hot temperatures cause people to both eat more ice cream and spend more time outdoors under the sun, resulting in more son blaze. I guess an important thing to do is always eliminate confounding variables and think of ways that you can account for them, um, in your experiments. So obviously the first way of doing this is too. Think of the possible confounding variables. I'm sure that you have data on them, um, so that you can include them as control variables in the models or correlations that you identify. Most of the time, these will be regression models. Then you can compare the models and isolate the impact of the independent valuable. The other way of accounting for confounding variables is to randomize the values of the independent variable. Let's say you randomly assign people to the control group and the treatment group. The idea of this is that you ensure the effects of the confounding variables are the same for all people. Therefore, I guess the effect of the independent variable can be identified. Um, so there may be other ways of doing this. These are just a couple of the simpler approaches blindly trials. Um, a double blind study is where the participants I don't know if they are in the treatment group or the control group. This is to prevent the patient's expectations influencing the results. Um, these may be known as psychophysiological effects. And likewise, the researchers doctors, as statistical analysts, also don't know which groups participants side. This prevents groups being treated differently or having their outcomes interpreted differently. Um, single blinded trials. Obviously, if only the researcher or the participants are blinded so that one's types of recessional case reports, uh, sort of bottom of the pack when it comes to the power of the evidence, Um, as they're not heavily weighted when it comes to, I guess, forming treatment protocols or management protocols. They generally describe the medical history of a single patient anecdotally, and that's why, uh, since they don't have a lot of evidence backing them, they don't go on to, uh, contribute a significant amount to change. In clinical practice, however, they are highly informative about the patient's condition. They can be written up in a short period of time and our generally easy to understand. Um, but they may sometimes influence clinicians perspective on managing the patient so they can be useful in that sense. And then the condition may feel it's necessary to do um, like further studies, um, to prove their theory or opinions so they can be They can provoke burden research quite useful. In that sense, a series have multiple patient's, uh, multiple similar patient's with similar problems or outcomes of interests. Obviously, this means that there's more data, uh, more clinical data, which means more statistical power. It will be more reliable than the case report, but it's still anecdotal the advantages and disadvantages of case series. Oh, the disadvantages includes the fact that the number of patient's in case there is usually small enough. Sorry, it's usually small. It's not enough for meaningful statistical analysis. Um, because obviously, collecting detailed data on a lot of patient's can be quite difficult. Um, there's no control group, so any observed association between an exposure and an outcome of interest may be coincidental. I won't spend too long on the advantages and disadvantages slides, because obviously this is something that This is where you can pause the video, I guess, And you can read through the starve yourself. So waste too much time on these ecological studies. But these could look at changes in mortality. Um, due to covid, for example, Um, um, this could happen over time, and this would be called a time series ecological study. It could also be used to compare the prevalence of the disease between different regions a single point in time. So this would be a geographical study. Uh, generally speaking, ecological studies that involve a retrospective analysis of population data as a whole. And since there's a lot of data, um, there is more statistical power associated with such studies. Uh, this advantage is generally a lack of control for confounding variables because, um, you may have a lot of data on all the different variables, but you have no way to explain or link them confidently, Which is why, uh, compounding variables usually not accounted for. Um so it's also difficult to identify who had both the exposure and the outcome when using population data as gross statistics will be used. Um, of course, sexual surveys, as the diagram suggests, you look at the characteristics of the patient population that as a point in time, um, and you can do these spaced intervals, uh, in order to visualize the trend. So multiple prosection surveys may be useful. Um, for visualizing a friend and getting greater statistical power, you do have a lot of disadvantages. Um, they tend to over represent cases with a long duration and under represent those with a short duration of illness. Um, um, exposure and outcomes are measured, but current exposure information is often, uh, not relevant because it takes time for specific exposures to have an effect. Um, can't establish causation. Only an association, um, a subjective bias recall bias and confirmed labels that, for example, across sexual survey may be used to determine the body mass index of a male of of male and female adolescents. Um, but it gives no information as to why a proportion of the samples are overall underweight. Case control studies, as a diagram, suggests you initially split patient group into the cases, so those with the outcome of interest and a control group, the cases are then obviously further further divided into the patient's who were exposed and unexposed and the same goes for the controls. Um, you compare the medical history to find the exposure. Ideally, the group of patients who are in the cases, um group and the control group should be demographically similar. Um, uh, They would be need control variables again. Advantages and disadvantages. Um, disadvantages. They find associations they don't prove causality. Rely on records to determine exposure. Difficult to eliminate family variables called per halt studies. Uh, the best way to probably explain this is, uh, take an example. Let's say hospital recruits new patient's onto the cohort study to monitor how their illness progresses at the time. However, like patient's, may tend to drop out for several reasons, including reduced participation. They may develop illnesses, um, which limits their participation or just excludes them from the study because of, uh, specific inclusion criteria that's needed, um, uh, remain in the study. And it's also difficult to study up for long enough to give significant amounts of data because the facilitators have to follow the patient's over time, and it's difficult to regulate their behavior. And this is particularly difficult for smaller institutions. Two groups in the diagram must otherwise be, uh, pretty similar demographically as I mentioned, or case control studies, especially in terms of the co morbidities that's there between the two groups, as the other diseases can obviously have an impact on the outcomes, advantages and disadvantages. Um, yeah. Pearl studies on as reliable as our CTS Um, because the groups and the invest investigation may differ in other ways than in the very warm to study. So it's important to make sure that, uh, I guess they're Democratic is similar again, especially in terms of any comorbidities that they might have. They they can take a long time, and they can't exclude unknown confounding variables. Randomized control trials. These are typically a gold standard when it comes to research, especially please pinnacle trials of specific treatments. So they tend to have close monitoring protocols which allow for the rules to be enforced, which means that the data tends to be more reliable. They're generally for shorter durations, so patient's can be followed up more easily. Uh, yeah, the purpose of randomization is such that is so that, uh, confounding the effects of the confounding variable are sort of distributed amongst both the treatment group and the control group, so that you're only able to observe the effects of the independent variable, which is the intervention, typically and again, the demographics of the control and treatment group, uh, should be as similar as possible. Okay, they have a lot of advantages, but very few disadvantages. So disadvantages include random allocation because it may be impossible in practical or unethical. They're usually expensive and time consuming. Um, so our CTS may use two few patient's in the trial. Um, or the trial may be too short lived to your significant results. This is obviously because, um in order to get to the randomize controlled trial base in humans, um, specific treatments or, uh, interventions need to go through several prior phases. And this takes a lot of funding and resources? Certainly, uh, enough of patient population, maybe quite difficult. The crossover trials used when trying to assess the effects to different treatments. Um, so, generally speaking, uh, as the diagram shows, there is initially a randomization process, uh, which is obviously a very good for accounting for confounding variables. And the patient group is split up into two, um, where one of them undergoes one treatment, followed by the other one and then, uh, conversely, uh, for the other group. So the washout period is very important in order to remove the effects of the initial treatment that patient's had. This is either to remove the drug from the body or just remove any of the side effects that they persist. Um, and it's important to have two groups because having one intervention followed by the other, um, having treatment a followed by treatment be may have different outcomes to, uh, if you were to have treatment be followed by treatment. Oh, yeah, that's an important point. Um, yeah, patient surrounding yeah, allocated. And they're generally prospective studies. Yeah, a couple of the two disadvantages that they have Are that the washout period? Maybe lengthy or unknown? It's difficult to know exactly how long the washout period should be established for. Pardon me. As the effects of the drugs or interventions can persist past that period, crossover design is can't be used when treatment effects are permanent or long lasting, systematic reviews. This is probably the best way to summarize all the research that's out there on a particular topic. It's very reliable and a high quality form of research and It generally relies upon a comprehensive literature, um, search to understand what literature is already out there, what's missing and whether your ideas for publication are novel or if, uh, similar literature already exists. So it's very useful to perhaps contact a librarian who's trained in performing comprehensive literature searches when starting a systematic review. They're obviously retrospective in nature because you need to look at data that's already there. Um, yes. So disadvantages include that there may be insufficient high quality primary studies using a specific methodology, so you may end up with only a few papers that you can analyze. In the end, Uh, the interventions examined may not reflect current practice. That's right. Meta analyses. These are often included as part of systematic reviews because it's a nice way to visualize all of the data that's out there. This is typically done using a forest plot, um, which shows the outcome of interest that's proposed by each paper. Uh, it's also generally done when you can pull, uh, more than one data set, which have which have similar data, and this obviously yields greater statistical power. The disadvantages are that, uh, there's heavy reliance on available public studies, and it may create it may create exaggerated outcomes, uh, due to publication advice. And it's more difficult to publish studies which showed, uh, negative or inconclusive results. Thank you for watching, uh, feel free to email me if you have any questions. Um um, And again, this video was all about the different types of research and, uh, the technology that you should be aware of before going on to interpret time seen studies. Your next video will be about how you can get involved in research. Thank you very much.