Evidence Based Medicine 2 Research Design Methods (1).pptx

evidence based methods- research designs

1. Intro to Research Design & Methods Michelle Paek, Francesca Kimelman , Andrew Kerrigan

2. Primary, Secondary, and Tertiary Sources Primary sources: original materials on which other research is based Includes original journal articles, dissertations, technical reports, conference papers, etc. Secondary sources: materials that analyze, evaluate, interpret, or reorganize information from primary literature Includes review articles, textbooks, databases/data compilations Tertiary sources: summaries of information from primary and secondary sources Includes encyclopedias, almanacs, fact books, UpToDate

3. Hierarchy of Evidence https:// himmelfarb.gwu.edu /tutorials/studydesign101/ As we progress up the pyramid, the studies become more evidence-based and less numerous Systematic reviews and meta-analyses are at the top because they can only be written after much research has already been done Best Evidence Synthesized & Evaluated Literature Primary Literature May or may not be Evidence-Based

4. Primary Sources

5. Case Reports Detailed reports of symptoms, signs, diagnosis, treatment, & follow up of an individual patient Lowest level of evidence, but also first line since new issues and ideas emerge here (hence the base of the pyramid) Advantages: Identification of new trends/diseases Educational – share lessons learned Disadvantages: Cases may not be generalizable Not based on systematic studies May be emphasizing misleading elements

6. Case Reports Cont’d Things to look out for to examine quality of case reports: Information about patient’s age, sex, ethnicity, race, social situation, medical history, diagnosis, and current interventions? Carefully recorded unbiased observations? Presentation of a hypothesis and evidence for it, as opposed to declaring causality?

7. Examples of Case Reports

8. Cross-Sectional Studies Cross-sectional studies examine the relationship between a disease and other variables of interest in a defined population at a single point in time or a short period of time (provide a “snapshot”) Usually conducted to estimate the prevalence of a disease or other outcome in a population Purpose of the study is descriptive, often in the form of a survey, with no hypothesis ** ** Cross-sectional studies would fall right here

9. Cross-Sectional Studies Cont’d Advantages Inexpensive and take little time Can estimate prevalence of the disease/outcome if sample accurately represents population Many outcomes and risk factors can be assessed No loss to follow-up Disadvantages Difficult to make causal inferences (difficult to determine whether the outcome followed exposure or the other way around) Only a snapshot – the situation may differ at another time frame Unable to measure incidence

10. Examples of Cross-Sectional Studies

11. Case Control Studies Observational studies that compare patients who have a disease (cases) with patients who do not (controls) Look back retrospectively to compare exposure to a risk factor in each group to determine relationship between risk factor and disease

12. Ex. of Case Control Study Design Professor David Tybor , PhD, MS, MPH Research question: Is eating cantaloupe a risk factor for getting listeria?

13. Case Control Studies Cont’d Advantages: Less time needed since condition/disease has already occurred Good for studying rare conditions/diseases Lets you simultaneously look at multiple risk factors Useful for situations where it is not ethically permissible to assign people to certain risk factors (don ’ t want to continue to feed people contaminated cantaloupe)

14. Case Control Studies Cont’d Disadvantages: Retrospective design means data quality problems Reliance on memory L imited information about exposures or potential confounders Many potential sources of bias and more subject to bias than other types of studies i .e. selection bias, misclassification bias May be difficult for source of controls to mirror source of cases I f cases are from hospitals, want to find controls from hospital too, not from general public. However, means that your study may not be generalizable to a wider population

15. Cohort Studies Observational studies that retrospectively or prospectively follow individuals that comprise your sample (cohort) to see who develops a disease over time Aim is to select a “cohort” based on a defined characteristic/experience within a specified period and determine which exposures/risk factors are associated with the development of the disease of interest Must be free from outcome/disease being studied at start of study

16. Ex. of Cohort Study Design Research question: Is smoking a risk factor for COPD? Professor David Tybor , PhD, MS, MPH

17. Cohort Studies Cont’d Advantages: Subjects in cohorts can be matched, which limits influence of confounding variables Easier and cheaper than a randomized control trial Disadvantages: Large sample size required since only some patients will develop the disease of interest No randomization, so imbalances in patient characteristics could exist Outcome of interest could take a LONG time to occur  loss-to-follow-up, expensive, etc.

18. Review of Case Control vs. Cohort Studies Case-Control Study: studying the disease and seeing if you can associate risk factors to it. Always retrospective. Cohort Study : studying the risk factor and seeing if you can associate a disease to it. Can be prospective or retrospective (i.e. you can follow the cohort in past time by checking their hospital records). Example: A study to find the percentage of smokers among lung cancer patients versus the percentage of smokers among those who do not have lung cancer  case control A study following smokers and non-smokers in time and seeing who develops lung cancer or not  cohort study

19. Example of a Case Control Study Disease being studied is multi-drug resistant TB vs. drug susceptible TB. Risk factors for MDR TB and differences between the two groups are being analyzed.

20. Example of a Cohort Study Risk factor being studied in presence of HIV vs. absence of HIV. The outcome associated to this risk factor is adverse events in TB treatment.

21. Quiz: Studying alcohol use (risk factor) and its effects on HIV care. Thus, this is a cohort study.

22. Randomized Control Trials (RCTs) Studies that randomly assign participants to an experimental group or a control group Need to ensure proper randomization process (treatment and control groups should be identical except for treatment)

23. Ex. of RCT Study Design Research question: Does Treatment X improve Disease Y ? Professor David Tybor , PhD, MS, MPH Researcher randomly assigns patients to groups

24. RCTs Cont’d Advantages: Good randomization will remove population bias and control for all confounding variables Easier to blind than in observational studies If well-designed and well-conducted, RCTs are the gold standard, but all that gold can be heavy … Disadvantages: Expensive in terms of both time and money Volunteer bias (population that participates may not be representative of the whole) Loss to follow-up L imited applicability In the “real world” very uncommon that exposures/treatments will be limited to just one or two things, as in the studies Vulnerable to sudden changes in knowledge Does not reveal causation!

25. Examples of RCTs

26. Quick Review of Studies thus far

27. Secondary Sources

28. Practice Guidelines Statement produced by a panel of experts for clinical decisions. Produced after extensive review of literature. However, practice guidelines are not all without flaws and must still be critically reviewed Ideally, the panel should be composed of a variety of experts with assorted affiliations, not just one organization Ideally, the panel should not have any financial conflicts of interest, especially the chair Ie . payments for consulting

29. Practice Guidelines Cont’d Advantages: Created by panels of experts Based on professional published literature Practical guidance for clinicians Disadvantages Slow to change/update Not always available, especially for controversial topics Expensive and time-consuming to produce Recommendations may be biased/affected by the organization creating the guideline

30. Examples of Practice Guidelines

31. Systematic Reviews Documents that provide a comprehensive review of all relevant studies on a particular topic/question Combines information from both published and unpublished studies and summarizes the findings

32. Systematic Reviews Cont’d Advantages: Exhaustive review of current literature and other sources Less costly and timely to review prior studies than conduct new studies Results can be generalized to the general population more broadly than individual studies More reliable and accurate than individual studies Disadvantages Very time consuming May not be easy to combine studies Studies included in the systematic review may not be valid and scientifically sound themselves May be “lagging” in recommendations – recommendations not based on current best evidence May rely on expert’s knowledge rather than a systematic review and thus be missing information

33. Examples of Systematic Reviews

34. Meta-Analyses A type of systematic review in which selected studies are combined to develop a single conclusion with greater statistical power Used to: E stablish statistical significance with studies that have conflicting results Develop a more correct estimate or provide a more complex analysis Examine subgroups when individual numbers are not statistically significant themselves

35. Meta-Analyses Cont’d Advantages: Greater statistical power Confirmatory data analysis Greater ability to extrapolate to general population affected Disadvantages Very time consuming May not be easy to combine studies (logistically or due to differences in populations studied ) Not all studies provide adequate data for inclusion and analysis Requires advanced statistical techniques Heterogeneity of study populations

36. Meta-Analysis Example

37. Forest Plots ( AKA Blobbograms ) The size of the individual boxes represents how strongly the study was weighted The horizontal lines for each study indicate their confidence value (those that do not cross 1 are statistically significant) The dark vertical central line and the diamond at the bottom represents where the combined studies pooled to I-squared and the p-value tell you how homogenous or heterogeneous the individual papers are Ideally, I-squared should be below 50 and as close to 0 as possible. Ideally, the p-value should be greater than 0.05. If either of these are not true, then it means that the studies were HETEROGENEOUS, so it is difficult to combine their results to get one consistent answer

38. Conducts systematic reviews of RCTs of health-care interventions and diagnostic tests, which it publishes in The Cochrane Library. Considered amongst the highest regard of the meta-analyses. A great resource for you! Their logo is actually a mini forest plot! :)

39. Any Qs?

40. Feel free to contact us ! If you have questions: [email protected] 081 770 7328 francesca.kimelman@ tufts.edu 081 840 3749 [email protected] 081 770 7335

41. Sources https://himmelfarb.gwu.edu/tutorials/studydesign101 / http://www.nature.com/ebd/journal/v7/n1/full/ 6400375a.html Tufts professors, as cited on individual slides