FEBP Course Intro: Evidence-Based Practice

FEBP Course Introduction

• Lectures occur Wednesdays 12:00-13:00 in the Babbage Lecture Theatre
• Slido facilitates questions during lectures
• There are four seminar-style supervisions: two in Michaelmas, two in Lent
• The first seminar in each term focuses on reading and critically analyzing a paper
• Discussions cover study design, bias, and confounding in Michaelmas, while Lent covers analysis and interpretation
• Each lecture has multiple-choice questions that assess understanding of course materials
• Completing the questions prepares students for the seminar
• The application of the best available research to clinical care is known as evidence-based practice or medicine (EBM)
• EBM needs integration of evidence with clinical expertise and patient values
• Clinically relevant research is used to determine diagnostic tests' accuracy, precision, importance of prognostic markers, effectiveness and safety of strategies, and patient experience

Research, Evidence and Epidemiology

• Epidemiology studies the distribution and determinants of health-related states/events in specified populations, applying the study to health problem control
• Epidemiology considers person, place, and time
• John Snow's cholera outbreak mapping exemplifies scientific methods in studying disease distribution
• "Clinical epidemiology" applies epidemiology in clinical populations
• Application of epidemiological research for the care for animals and humans
• A good understanding of research methods, epidemiology, biostatistics, literature searching provides the best decisions
• Though high-quality reviews of experimental studies are more influential publication in a journal does not guarantee quality

Descriptive vs. Analytical Studies

• Analytical Studies aims to understand the association b/w two factors
• Randomized Controlled Trials, Case-Control Studies, Cohort Studies, Cross Sectional Studies, and Ecological Studies examples of analytical study designs
• Descriptive Studies describes an outcome in the population

PICO Framework

• Refining research questions involves PICO:
• Population
• Intervention (or exposure)
• Comparison ("Control" here)
• Outcome

Bias and Confounding

• Resources like quizzes on VLE (virtual learning environment) will help students recap basic concepts
• In study, the goal is to get a generalizable conclusion about relationship between exposure and outcome of interest
• Vaccination/screening programs implement primary/secondary prevention measures to implement at-risk populations
• Though it is understanding underlying causes of the outcome of focus

Chance, Bias, & Confounding

• Chance (Lecture 4), Bias, and Confounding are reasons data might not be meaningful for studying exposure-outcome relationships
• Bias is systematic error between groups and distorts the study’s association or results
• The potential source of bias explains how it might affect the results of the study

Types of bias

• Selection bias: The study population does not adequately represent the target population, so there is a systematic difference between the participants in the study and the population about which the conclusions are to be drawn
• Information bias: There is a systematic difference between groups in how information is sought, extracted, measured, or recorded
• One type of information bias: recall bias (in case-control studies, the affected group better remembers exposures unlike the control group)

Confounding

• Confounding occurs when another factor other than what you're studying causes differences between two groups
• To be a confounder:
• Must be independently associated with exposure and the outcome
• Must not be on the causal pathway between exposure and outcome

Experimental vs. Observational Studies

• In an observational study Interventions or exposures are not assigned by investigators. They look for associations to observe exposures and outcomes. Effects of confounders are difficult/impossible to control in these studies
• The intervention is assigned by investigators and there is control of which participants are in the intervention/comparison group in experimental study,

Randomized Controlled Trial (RCT)

• Randomization balances confounding factors at the study's start
• Blinding combats biases
  • Single blinding conceals info from participants
  • Double blinding conceals info from investigators

Overcoming challenges with blinding

• Placebo control will tell if the intervention was given for blinding to be aided
• Standard of care comparisons assess if a new treatment betters standard of care
• The the RAMbo mnemonic key assessment points of RCT
• Randomised (allocation, concealment, baseline similarity?)
• Attrition (follow-up, intention-to-treat analysis, equal treatment?)
• Measurement (blinded assessors, objective measures?)

Observational studies

• Observational studies used when assigning intervention isn't possible
• The intervention/exposure isn't assigned by experimenters but data about exposures and outcomes are looked for associations
• Kinds of observational studies: cohort; case control; cross-sectional; ecological

Cohort study

• Data's obtained from exposed groups and there is no exposure allocation from researcher to those groups
• This is the best way to study effect of factors of risk on an outcome
  • Advantages: ethical, subjects can be matched, timing/direction, eligibility/assessments, and cheaper than RCT
  • Disadvantages: difficult to identify controls, linked confounders, difficult blinding, no randomization and large sample sizes are needed

Case control study

• A group has particular outcome/disease is selected when info is obtained if subjects been exposed to a factor under investigation
• Advantages: quick/cheap and feasible for rare diseases or lag between exposure/outcome
• Disadvantages: relies on recall/records, confounders, selection of control groups is tough and bias

Cross sectional study

• This studies relationships of diseases in a defined population at particular time and best quantifying disease or factor from a risk

• Advantages: cheap and ethical

• Disadvantages: no causality, recall bias, group unequal sizes and confounders

Ecological studies

• This can measures exposure and outcomes at population level that can be cheap and used when one wants association
• However one measures factors on the same single timepoint and can't use understand affect or risk across groups

Appraising of papers

• Papers make up of the dissementation of research findings from looking for those with a focus clinical question and literature
• To appraise the following must be trustworthy, meaningful, relevant
• Understanding components of a paper for allocating focus
• Papers are publish in journals going through a peer review after appraisal that includes potential corrections and specialization

Components of articles

• Title: Concise, informative, reflection of content
• Abstract: Summary, purpose, methods and findings
• Introduction: Summarize context of the problem
• Methods: Details of study
• Results: Figures and Table find
• Discussion: Placed in contest with extrapolation
• Conclusion: Summarize main finding
• Abstract can provide design, result and contact

Reading the Study should trusting, enablable meaningfulness and relevancy

One-Way ANOVA

• This will see mean affect
  • Must continuous datatypes and independent datapoint in order to test

Mann-Whitney U test

• This is for distribution of data and see if they have high outcome

Pearsons' Chi Squared Test

• Used only with homogeny and proportion from category

Fishers exact test

• Testing if the tests had propotion by the outcome

The Pearson and other tests

• The Pearson has multiple measures that range linearity of data and is important relationship test to understand

Data Types for variables

• Understanding how perform analysis because there are different types to know data
• Data can be number or category data in form of
  • continuous
  • normal distribution
• time to even data
• count category

Note there are special types to test

Normal vs. Ordinal

• Test those based on what the data tells that there it's to

Understanding Risks

• Many different things must have binary test because that could have
  • Death vs. Ave
• Recover vs. Not and the list goes on
• It more is less about binary and test outcomes

Clinical Research

• Many studies look to measure common a disease and focus how know that

Study variable

• Is based on outcome of intersted

Bias reduction with variable

• Reduce with good thresholding

Incendence vs. Prevalence

• Incendence is time and disease vs Prevalence with disease

Meta-Analysis

• Meta is understanding data
• The point estimate is
• Interveial for confidence -Hypo

The points needs stats