Evidence Based Medicine PDF

Evidence Based Medicine How to read the quality of the research paper? Janak Awasthi, PhD Evidence Based Medicine How to read the quality of the research paper? Janak Awasthi, PhD  What is Evidence Based Medicine? Evidence based medicine Evan (Evidence Medici...

Evidence Based Medicine How to read the quality of the research paper? Janak Awasthi, PhD Evidence Based Medicine How to read the quality of the research paper? Janak Awasthi, PhD  What is Evidence Based Medicine? Evidence based medicine Evan (Evidence Medicine) (EBM) is the conscientious, cooks (Clinical Expertices) Brownies (Best and Evidence) explicit, judicious and pastas /Patient values reasonable use of modern, best evidence in making decisions about the care of individual patients. 5 A’s of Evidence based medicine All Amazing Apples Are " Awesome Step 1: Ask-The types of clinical questions we can answer with EBM Types of clinical questions: questions about an intervention questions about a condition’s etiology and risk factors questions about diagnosis (e.g. most appropriate tests, etc.) Cronostic o questions about prognosis and prediction question about phenomena questions about frequency and rate of a condition question about cost-effectiveness  od ? Step 1: Ask- PICO Format Go tion Qu es P Patient - Include the most important characteristics of the patient and their health status Intervention- This can be a treatment or even a test that we I are interested in assessing for the current patient/population of interest C Comparison- In order to assess the efficacy of the intervention, it should be compared to another to see how it performs. This can be to placebo in some cases or even to the current gold standard treatment. O Outcome- A testable or measurable outcome. Essentially what you are trying to achieve, measure or affect. We should always carefully consider the outcomes being used and whether or not they are the most relevant ones. Step 2: Acquire the Best Evidence Once you have a question, you will need to look for information to answer the question. One of the most challenging aspects of looking for evidence is sifting through all the information out there to find what is most reliable. Too much information can be majorly challenging!! Step 2: Acquire the Best Evidence- Where should we look for evidence? Traditional sources (e.g. textbooks) ○ Decent starting point for someone new to a field ○ However info can be outdated or disorganized Ask a colleague or “expert” The quality of information you get here is variable Primary resources Articles that appear in peer-reviewed journals and are found by searching reliable online databases e.g., Medline Pubmed Clinical Queries https://www.ncbi.nlm.nih.gov/pubmed/clinical Cochrane Library- https://www.cochranelibrary.com/ Secondary sources Summaries and analyses of evidence derived from and based on primary sources Quality is linked to the quality of the primary sources used However since secondary sources compile primary ones, the quality can be high since they are pulling from multiple studies (provides corroboration as well as a more complete picture).  Sources to AVOID and/or treat with caution Social media Anyone can post anything on social media This is the modern-day equivalent of the guy yelling on a street corner. Sure he might be right, but he also might be crazy (or just uninformed). Any random website or place where you can’t determine its source Once again, anyone can make websites If a source like this is to be believed you need to confirm where it is coming from E.g. Nutrition blogs/web sites Dude with no science degree to speak of, who sells his own supplements USDA website Newspapers, New websites, etc. Sometimes overstate/oversimplify findings from scientific papers Always go back and look at the original source yourself Basic scientific research These are research articles published concerning experiments done primarily in a laboratory setting like a test tube or animal model (aka not in human patients)  Evaluar Step 3: Appraise the Evidence- Basic questions to ask when appraising a paper Was the research peer-reviewed? In order to publish in a peer-reviewed journal, authors must submit their paper to be edited by other experts in their field Is the study published in a reputable journal? Has all the appropriate information been given? Every study should include enough details on how the study was performed: aka a description of the population of interest, an explanation of the process used to select and gather data on study subjects, definitions of key variables and concepts, descriptive statistics for main variables, and a description of the analytic techniques so that you are able to understand what was done and can evaluate it How was the study designed? There are different types of ways to design a scientific study, and each design style has its own strengths and weaknesses Step 4: Apply the Evidence If the data in the paper looks valid, I still need to think about a few key factors before I bring it to my patient. Are the people in the study like my patient(s)? You want a study in which the patients are as alike as your patients as possible, in terms of variables such as: ○ Age ○ General state of health ○ Type and severity of disease process ○ Time in the course of the disease You will rarely find a study with patients exactly like yours, but if they are too different you may want to spend some time looking for another study. My patient is more likely to have a similar outcome to those in the study if they are like the population in the study performed. Step 4: Apply the Evidence Did the study cover all aspects of the problem important to my patient? Most medical problems have a lot of different aspects that you take into account when deciding on a treatment or course of action for a patient. Look for studies that deal with all the aspects that are important to you i.e., treatment side effects. Or it may show that one treatment gives patients better pain relief than another, but not show which of the treatments is better at treating the underlying condition. Be aware that you need to fill in some of the gaps using your own judgment. Step 4: Apply the Evidence After critical appraisal of the evidence is deemed valid and important, consider: ○ Can the evidence be applied to our individual patient or population? You must take into account the patient’s own personal values and circumstances. Within a clinical trial both efficacy and risks should have been fully discussed with the patients ‘‘therapeutic alliance.’’ The fundamental principle of EBM: the integration of good evidence with clinical expertise and patient values. Cost analysis and applying treatment  Study You can enter a subtitle Design here if you need it Types of study designs · & · # Case series and case reports Case series and case reports consist either of collections of reports on the treatment of individual patients, or of reports on a single patient. For example: One of your patients has a condition that you have never seen or heard of before and you are uncertain what to do. A search for case series or case reports may reveal information that will assist in a diagnosis. However, for any reasonably well-known condition you should be able to get better evidence. Case series and case reports, since they use no control group with which to compare outcomes, have no statistical validity. Randomized control clinical trial (RCT) Experimental Group Study design where treatments, interventions, or enrollment into different study groups are assigned by random allocation. With certain research questions, randomized controlled studies cannot be done for ethical reasons. For instance, it would be unethical to attempt to measure the effect of smoking on health by asking one group to smoke two packs a day and another group to abstain, since the smoking group would be subject to unnecessary harm. Randomized controlled trials are the standard method of answering questions about the effectiveness of different therapies. Cohort Study A Cohort Study is a study in which patients who presently have a certain condition and/or receive a particular treatment are followed over time and compared with another group who are not affected by the condition under investigation (Prospective). Like case control studies, cohort studies are confiable not as reliable as randomized controlled studies, since the two groups may differ in other ways. For example, if the subjects who smoke tend to have less money than the non- smokers, and thus have less access to health care, that would exaggerate the difference between the two groups. A Case Control A case-control study is used to investigate the causes of a particular health outcome or disease. Individuals who have a specific outcome or disease (cases) are compared with individuals who do not have the outcome or disease (controls). The primary objective is to identify factors or exposures that may be associated with the development of the outcome or disease Case control studies are not as reliable as randomized controlled studies, since the two groups may differ in other ways. Patient recall may also be inaccurate or biased. Systematic review and meta-analysis Follow rule A systematic review is a comprehensive a survey of a topic in which all of the primary studies of the highest level of evidence have been systematically identified, appraised and then summarized according to an explicit and reproducible methodology. A meta-analysis is a survey in which the Each one is a systematic review results of all of the included studies are similar enough statistically that the results are combined and analyzed as if they were one study. In general a good systematic review or meta-analysis will be a better guide to practice than an individual article. What kind of answer do you want? Validity and Minimizing biasness Key questions to ask when appraising a paper Is the study designed in a valid way? Valid studies answer the questions they pose in a scientifically rigorous manner. When analyzing research, we should assess the following: Internal validity Asks whether the outcome measured is truly due to the independent variables or some other factor that wasn’t properly accounted for. E.g., I want to study the effect of convalescent plasma as a treatment for COVID. I see that patients treated with this treatment tended to show an improvement, but the treatment group was also given some antiviral drugs. I cannot be sure the effect I see therefore is due to the plasma treatment. The study is not internally valid. Relate to how well the study is conducted Cont’d- External validity Asks whether the results obtained in this study can be generalized to other setting, people and places. E.g., I want to study the effect of a treatment for pancreatic cancer on patients living in New York who have an EGFR mutation. Will this generalize to all other pancreatic cancer patients? relates to how applicable the study findings are in the real world Construct validity Asks whether the test, instrument, method I used to measure my variable of interest in the study appropriate. E.g., I am teaching a course on genetics and want to see how well my students are learning. So, I give them an algebra test to measure their performance. Is this an appropriate measure? Problems that exist in the literature Publication bias: Conflicts of interest Poor study design Improper controls Misleading statistics Improper, incorrect, or overstated conclusions 1. Publication Bias This is the tendency of journals to publish positive results and not publish negative results ○ Positive results: the intervention/treatment shows an effect ○ Negative results: the intervention/treatment does not show an effect E.g., A study researching a particular therapeutic agent where the data showed little/ negative impact, will less likely be published. Since scientists need to publish papers in order to get funding, this can lead to them overemphasizing minimal effects in order to get published. 2. Conflicts of Interest Research should always be performed from an objective, unbiased standpoint Research should be motivated by a desire to answer a question correctly However, if a scientist/physician has other motivations when performing the study, this can introduce bias E.g. A drug company funding a scientist to do a clinical trial on a their drug 3. Poor study design: Improper controls In order to draw the correct conclusions, using the proper control is essential. ~100% of time This twin has this twin has disease X For example: Let’s say I discover a new disease X that I disease X believe is caused by a genetic mutation, but I don’t know the MZ twins: we share all genes mutation. I want to do a study to see if there might be a genetic link. I decide to do a study in identical twins, who share all their This person ~20% of the time this person DNA in common. If a mutation is the sole cause of this has disease has disease X X disease, I expect that if one identical twin has the disease, Unrelated strangers the other should too (they will share same mutations). For a control I compare random unrelated individuals. What is wrong with this design? Improper controls From this data, I cannot conclude disease X is This twin ~100% of time caused by a mutation. this twin has has disease disease X MZ twins ALSO share an environment whereas X unrelated strangers do not MZ twins: we share all genes Disease X could be genetic, environmental or BOTH. This control does not let me make the ~20% of the time This person correct conclusion. has disease X this person has disease X Unrelated strangers Misleading statistics Choosing the correct statistical analysis is vital towards being able to draw and illustrate meaningful conclusions Misleading statistics leads to misleading conclusions Improper, incorrect or overstated conclusions While authors will discuss what they think a finding means, they may be incorrect. Perhaps they are currently missing a piece of the puzzle required to truly understand what their study means. E.g. Haemophilus influenzae bacteria was originally thought to be causative agent behind the flu Early studies showed that this bacteria could be recovered from the lungs of some flu patients (correct) which was interpreted to mean this was the likely cause (incorrect) Today we know that the cause is the influenza virus and that Haemophilus influenzae infections often follow bouts of influenza, due to weakening of the immune system by the influenza virus Statistical tools  Common Statistical tests t-test checks differences between means of 2 groups. Numerical outcomes Example: comparing the mean blood pressure between men and women. ANOVA (Analysis of variance) checks differences between means of 3 or more groups. Numerical outcomes Example: comparing the mean blood pressure between members of 3 different ethnic groups. Chi-square (χ²) checks differences between 2 or more percentages or proportions of categorical outcomes (not mean values). Example: comparing the proportion of members of 3 age groups who have essential hypertension. Pearson correlation coefficient measures the linear correlation between two continuous outcomes. Positive correlation (as one variable increases, the other variable increases) Negative correlation (as one variable increases, the other variable decreases) Hypothesis testing To determine if the difference you see between your two sample groups is real, we employ hypothesis testing Null Hypothesis: Drug X is no more effective than control for treating headaches Alternative Hypothesis: Drug X is better than control for treating headaches I set up the study to try and get evidence to disprove my null hypothesis I basically am trying to test if there is a likely a difference between the true means of Drug X and control by taking a sample P-values In an experiment, we assume that the null hypothesis is true, unless we can disprove it Once we have collected our evidence, we will have two mean values, one for each group (Drug X and Control). These are our experimental means (not the real-life values since we can’t test the entire population) We use our means and standard deviations to compute the p-value The p-value is the probability of the null hypothesis being true given the data we got P-values and Hypothesis Testing For example, let’s say the mean headache time for placebo is 28±0.5hrs and the mean headache time when you take Drug X is 2±0.5hrs. The p-value is the probability we would get this result (28 vs 2) by random chance if we do live in a world i.e., there is no real difference between Drug X and placebo’s effect. The smaller the p-value, the more likely it is that the null hypothesis is false Conventionally, p < 0.05 is referred to as statistically significant and p < 0.001 as statistically highly significant i.e., we reject the null (i.e., reject the statement that says there is no difference between intervention/event/exposed vs control) and accept the alternative hypothesis (there is a difference). If P-value is > 0.05, it is not statistically significant, so we fail to reject the null and we reject the alternative hypothesis.