Study Guides (248,019)
Canada (121,233)
Kinesiology (151)
KIN 330 (4)
Midterm

Kin 330 Week 8 post midterm.docx

15 Pages
179 Views
Unlock Document

Department
Kinesiology
Course
KIN 330
Professor
Laura Middleton
Semester
Winter

Description
Kin 330 Week 8 In case you want to watch the video in its entirety: http://www.youtube.com/watch?v=td5Ld3nURAI&feature=g-u-u&context=G242feceFUAAAA AAAEAA SYSTEMATIC REVIEWS & META-ANALYSIS Setting the stage  Sometimes there are many studies addressing the same research topic with conflicting results o Important to choose the best evidence to apply in practice  How do you know what study to trust? o Highest quality study? o Systematic review or meta-analysis?*  Summarizes results of all studies. RECALL: Evidence-based Practice: Multi-step Process  Define the problem o Formulate an answerable question  Track down best evidence to solve the problem  Critically appraise best evidence  Apply to the individual o Based on clinical expertise and context  Evaluate practice outcomes When to do a review?  There are multiple studies with results regarding a specific research question.  Those studies are relatively uniform in design.  The findings for the studies are relatively consistent with each other  There is value to be gained by understanding the actual magnitude of a statistical effect with better precision. Three Types of Reviews Narrative Reviews  Summarize findings and opinions from a numerous studies examining a topic  Summarizes large amount of data in a narrative style (tells a story) Systematic reviews.  Similar to above, but with very specific criteria set out describing how the studies were selected  More rigorous search strategy with strict inclusion/exclusion criteria Meta analyses.  Atype of systematic review  Systematic search strategy  Addition of calculation of effect sizes by totally results across studies  Results in QUANTITATIVE summary of findings across studies SYSTEMATIC REVIEW/META-ANALYSIS PROCESS: Objective & Search Strategy  State the study objective, i.e. research question o What is the P.I.C.O. of interest?  Develop the protocol o Inclusion/exclusion criteria o P.I.C.O. characteristics o Study designs Advantages for both including only highest quality versus broad range of study types  Only RCTs: o Higher internal validity o Disadvantage: May not reflect all research o Good if area there are many studies  Broad range of research designs: o Advantage: better reflects breadth of knowledge o Disadvantage: some lower quality studies may bias the results  Develop the search strategy o Databases searched o Search terms o Review of reference lists Search  Conduct search o As planned with specified inclusion/exclusion & databases  Retrieve papers  Screen & select o Further evaluate for relevance: study objective, inclusion/exclusion criteria Evaluation  Evaluate methodological quality o Quality of designs & measures evaluated against standard criteria  Have evaluated strengths/weaknesses of study designs  Will spend substantial time in the remainder of this term to discuss further evaluation of study quality • Risk of bias • Reliability & validity of measures Analyzing & Synthesizing Results First step, collect data from all papers:  Collect information relevant to PICOT  Collect information regarding study sample  Collect information regarding study quality Next Step:  Summarize results (statistics) For example, SYSTEMATIC REVIEW VS. META-ANALYSIS  Both include similar systematic search strategies  Meta-analyses include synthesis of data by analysis (effect sizes) o Systematic reviews do not  If sufficient data, then meta-analyses are generally preferred Challenges We need a statistical method to synthesize the results of existing studies BUT, the studies use slightly different methods:  Different study features o Different control groups, different measures, different time periods, different measurement of sperm quality  Different statistics o Mean differences, odds ratios, relative risks Solution: Effect Size Can be derived from:  Means  Relative Risk  Odds Ratios (basically, any summary statistic that might be described in a quantitative study...) Denoted “d”  Calculated differently, depending on the source data Means  Results fromANOVAor t-test  Presented in mean (SD) by group and p-value for difference  Effect size, d = meaexposure mean comparison SD pooled What type of studied would have use means? Experimental If there was NO difference between groups, what would d (the effect size) be? ZERO Relative risk/odds ratio  Odds ratio (OR): o Ratio of odds between I/C o Odds: #with outcome/#without outcome  Relative Risk (RR): o Ratio of risk between I/C o Risk: #without outcome/total #  Effect sized = OR or RR (or HR) What type of studied would use OR/RR/HR? Observational IF was NO diff between groups, what would D be? 1! The Process  Once all are converted into effect sizes (d), they are comparable across studies!  Now need to combine into one summary effect size  Lots of variability among studies o Should they all contribute the equally to summary statistic? NO larger designs, should contribute more Which should contribute more or less? - Larger better designed studies ****************************************************************************** For example: Smith et al. (2001) (n=320) • High vs. low physical activity (adj. age, sex, educ) • Measured with actigraphy • Risk of dementia lower in active group Taylor et al. (1998)(n=498) • High vs. low physical activity (adj. age, sex, educ) • Measured by self-reports • Risk of dementia lower in active group Bosch et al. (2008)(n=84) • High vs. low physical activity (adj. age only) • Measured with self-reports • No significant difference in dementia between groups ****************************************************************************** Weighting  Which studies should contribute more? o Studies that are Larger in size and/or better designed  Effect size is known to be a better estimate of population risk when sample size is large than small  SO...  Compute effect size (d) for each study  Also weight effect size by sample size of study  Can you think of any other considerations that we could weight studies on? Forest Plot  In addition to a text-based description of the findings, the effect size indexes and overall effect size are normally presented in something called a forest plot.  This is a highly efficient and effective way of conveying the individual study findings and cumulative findings... Study 1 Large sample Study 2 size, What do u notice about Individual Study 3 study Study 4 confidence interval? Its effectsize Study 5 smaller Study 6 Study 7 Study 8 Box Size= sample size Study 9 Study10 Horizontal line= Study 11 Study 12 confidenceinterval (arrow head means Study 13 Study 14 offthe chart) Represents overall effect Vertical line represents NO difference size (across studies) If centre = 1 observational studies combined **** Centre =
More Less

Related notes for KIN 330

Log In


OR

Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Sign up

Join to view


OR

By registering, I agree to the Terms and Privacy Policies
Already have an account?
Just a few more details

So we can recommend you notes for your school.

Reset Password

Please enter below the email address you registered with and we will send you a link to reset your password.

Add your courses

Get notes from the top students in your class.


Submit