Week 1

Question 1

What are the EBP principles?

Answer 1

• Best research evidence
  
  • Professional and clinical expertise
  • Clients Values and circumstances
    *All patient centred

Question 2

What is the only thing we are interested in? What are confounding factors?

Answer 2

Intervention

Spontaneous recovery and errors

Question 3

How do you lose the influence of confounders to get the real difference of the treatment?

Answer 3

Spontaneous recovery + errors + Device X

Spontaneous recovery + errors + Physiotherapy

Question 4

Major types of research designs

Answer 4

• Randomized controlled trials: Very strong evidence, but sometimes have ethical problems so you cant do them
  
  • Cohort studies
  • Case-control studies
  • Time-series design
  • Single case experimental studies
  • Retrospective studies
  • Find the best evidence there is with your own clinical experience and clients values
    Triangle is designed for pharmaceutical studies e.g. pills, in allied health it is hard to find placebo treatments as you can’t blind a patient. So in AH we use some designs not in this pyramid that are also valid

Question 5

What is bias?

Answer 5

• Any trend in the collection, analysis, interpretation, publication or review of data that can lead to conclusions that are systematically different from the truth
  
  • A process at any state of inference tending to produce results that depart systematically from true values
  • Systematic error in design or conduct of a study

Question 6

Why is bias a systematic error

Answer 6

• Errors can be differential (systematic) or non-differential (random)
  
  • All data is flawed in the same way, so to identify what the bias is it can compensate for the error in analysis and researchers should report on that
  • We don’t like random errors as it can be in one group more than another
  • If there is a measurement bias, the groups will have the same amount of exposure to the bias so if you subtract the bias pre and post you still have the progress
  • if you have random bias, you may have more skewed data where one is influenced more over the other, making it hard to work
    To get rid of random bias is to include large numbers of data as they have less influence as it is a small influence

Question 7

What is random and differential error?

Answer 7

Random Error
Random error is a mistake in measurement that happens by chance. It causes results to be a little bit higher or lower each time, but not in any predictable way. These errors are like “noise” and don’t favor one side over another.

Example: If you weigh yourself several times in a row and get slightly different numbers each time, that’s random error.

Differential Error
Differential error is a mistake in measurement that happens more in one group than another. It’s not random—it affects some groups differently, which can make results look better or worse than they really are.

Example: If a survey asks men and women about their height, but the measuring tape is stretched for men and not for women, the error is different for each group. This can lead to unfair or biased results.

In short:

Random error = mistakes by chance, affect everyone equally

Differential error = mistakes that affect one group more than another, can cause bias

Question 8

What is systematic/differential error?

Answer 8

a consistent bias or deviation in measurements, meaning all readings tend to be either consistently high or low compared to the true value. Unlike random errors, which fluctuate around the true value, systematic errors push the measurements in one direction. This bias can be caused by flaws in equipment, procedures, or even the observer’s interpretation.

Question 9

Chance Vs Bias?

Answer 9

• Chance is caused by random error
  
  • Bias is caused by systematic error
  • Errors from chance will cancel each other out in the long run: Large sample size
  • Errors from bias will not cancel each other out whatever the sample size
  • Chance leads to imprecise results
    Bias leads to inaccurate results

Question 10

Different types of bias: Study design

Answer 10

• Selection bias
  
  • Sampling frame bias
  • Non-random sampling bias
    Non-coverage boas

Question 11

Different types of bias: analysis

Answer 11

• Confounding bias
  • Analysis strategy bias
  • Post hoc analysis bias: A patient that improved after treatment but not as a result of the treatment

Question 12

Different types of bias: Study execution

Answer 12

• Bogus control bias
  
  • Contamination bias
  • Compliance bias

Question 13

Different types of Bias: Data collection

Answer 13

• Instrument bias
  
  • Data source bias
  • Observer bias
  • Subject bias
  • Recall bias
    Data handling boas

Question 14

Different types of bias: Interpretation of results

Answer 14

• Assumption bias
  
  • Cognitive dissonance bias
  • Correlation boas
  • Generalization bias
  • Magnitude boas
  • Significance bias
    Under-exhaustion bias

Question 15

Different types of bias: Publication

Answer 15

• All’s well literature bias: Present a study in a way that is overly positive, downplaying negative aspects
  • Positive result bias: The tendency to publish something that has good outcomes.

Hot topic bias: Being less critical when a topic is of hot conversation

Question 16

What if bias is present in a study?

Answer 16

• Incorrect measure of true association
  
  • Should be taken into account in interpretation of results
    What is magnitude: Overestimation, underestimation

Question 17

Potential effects of source of bias

Answer 17

• Positive bias: The observed measure of effect is larger than the true measure of effect
  
  • Negative: The observed measure of effect is smaller than the true measure of effect

Question 18

Controls for bias

Answer 18

• Choose study design to minimize the chance for bias
  
  • Clear case and exposure definitions: Define clear categories within groups (e.g. age groups)
    Set up strict guidelines for data collection: Good clinical practice, ISO standards, use a protocol

Question 19

What could a confounding factor be

Answer 19

Factors that influence your outcomes:
- Independent factors
- Relationship is one-on-one
Relatively easy to identify and to correct for these influences

Question 20

What is confounding?

Answer 20

• third factor which is related to both exposure and outcome and which accounts for some/all of the observed relationship between the two
  
  • Confounder not a result of the exposure
  • e.g. association between child’s birth rank (exposure) and Down syndrome (outcome); A mothers age confounder
    e.g. association between mothers age (exposure) and down syndrome (outcome); Birth rank as a confounder

Question 21

What is a confounding factor?

Answer 21

• Influences not only the primary outcome, but also other factors that are influential
  
  • Double effect/catalyst
  • Not easy to identify
  • Needs statistical correction
    In most case not an easy task ‘Mantel Haenzel’

Question 22

To be a confounding factor, what two conditions must be met

Answer 22

• Be associated with exposure: Without being the consequence of exposure
  Be associated with outcome: independently of exposure (not an intermediary)

Question 23

What is effect modification?

Answer 23

• In an association study, if the strength of the association varies over different categories of a third variable, this is called effect modification
  
  • The third variable is changing the effect of the exposure
  • The effect modifier may be sex, age, an environment exposure or a genetic effect
  Effect modification occurs when the relationship between an exposure (e.g., treatment) and an outcome differs based on a third variable (e.g., age, genetics, comorbidities)

Question 24

How to control randomization?

Answer 24

• Randomization: Assures equal distribution of confounders between study and control groups
  
  • Restriction: Subjects are restricted by the levels of a known confounder
  • Matching: Potential confounding factors are kept equal between study groups
  • Stratification: For various levels of potential confounders
    Multivariable analysis: Does not control for effect modification

Question 25

5 Questions in EBP

Answer 25

Ask, Acquire, Appraise, Apply, Assess/Audit

Question 26

Types of questions about evidence

Answer 26

- Aetiology: What causes this condition - Intervention: Does this intervention work (prevent or treat a condition) - Diagnosis: the identification of the nature of the disease or illness by examination of symptoms - Prognosis: The likely outcome or course of a disease Meaning and experience: What are peoples beliefs, experiences and concerns

Question 27

Evidence hierarchy?

Answer 27

1. Systematic review 2. Randomized control trial 3-1. Pseudorandomised controlled trial 3-2. Comparative study with concurrent controls 3-3 Comparative study without concurrent controls 4. Case study

Question 28

What study should you use and when?

Answer 28

- A well conducted systematic review that only finds poor quality randomized controlled trials or lower level evidence will not be able to answer your question with any certainty A well conducted level 3 study may be more useful than a poorly conducted level 2 study

Question 29

What is PICO

Answer 29

P: Problem or Patient I: Intervention or Issue C: Comparison O: Outcome

Question 30

What does a well-structure question do?

Answer 30

- A well-structured question makes the second step of searching for relevant evidence (ACQUIRE) more likely to be efficient and successful - If a question is too vague it will lead to an inefficient search that yields many studies that are not relevant to the problem Once you start looking for evidence (step 2), you can always refine your question to be more specific (if there is a very large amount of research) or less specific (if there is very little research)

Question 31

Current challenges in EBP

Answer 31

- Only started in the 1990's - Many practices that were once commonly used have now been discontinued as the research evidence showed they were either of no benefit or caused more harm than good There is now robust discussion about the risks and benefits of many healthcare practices

Question 32

Predatory Journals

Answer 32

- Open access journals often charge the authors a fee to publish their work, this has led to the emergence of publishers who set up bogus journals to entice researchers to submit their work and have it published for a fee - The publisher is not interested in the quality of the work and there is typically no peer-reviewed process - Most journals will not be indexed in major data basis, google scholar however will include all published articles One way to check is journal rankings, which are usually made on the journals impact factor

Question 33

Conflict of interest

Answer 33

- Research trials are much more likely to have a favourable outcome if they are funded by industry with an interest in the outcome - Risks are likely to be downplayed and benefits overstated, results that are unfavourable are less likely to be published at all Research and decision-makers with a vested interest can ensure that the conclusions and recommendations in systematic review and clinical practice guidelines are in line with their interests

Question 34

The problem of bias

Answer 34

- There is no escaping that humans are biased - Need to consider how the researchers bias may have influenced a study but how our own bias may influence the studies we choose to read Confirmation bias is a particular problem for EBP: The tendency to pay attention to and to value information that confirms our pre-existing beliefs and assumptions, while ignoring, downplaying or dismissing information that does not confirm what we already think or believe

Question 35

What does using clinical queries in Medline and CINAHL do and what are its limits

Answer 35

Allows you to limit your search to the types of studies that are best suited to the type of question you have asked Medline limits: Reviews, therapy, diagnosis, prognosis, causation-Etiology, economics, clinical prediction guides, qualitative, costs CINAHL limits: Therapy, prognosis, review, qualitative, causation (etiology)

Question 36

If your question is about intervention effectiveness, you need to select what of 3 options

Answer 36

Maximize sensitivity or high sensitivity Maximize specificity or high specificity Best balance of sensitivity and specificity

Question 37

Maximize sensitivity or high sensitivity

Answer 37

Tells databases to go wide and add a broad limit to the search. Yields the largest number of results, reduces risk of missing relevant studies, but more likely to include non-relevant studies

Question 38

Maximizes specificity or high specificity

Answer 38

Tells the database to go narrow and add a broad limit to the search. Yields the smallest number of results. Reduces number of non-relevant studies, but more likely to miss relevant studies

Question 39

Best balance of sensitivity and specificity

Answer 39

tells the database to make the best trade-off between sensitivity and specificity. Will give you results somewhere between max specificity and max sensitivity

Question 40

What is smoothing the curve?

Answer 40

Adjusting data to eliminate outliers and make results more favorable