Epidemiolgy Flashcards

1
Q

What is Equipoise?

A

It is a state of genuine uncertainty about the benefits or harms that may result from each of two or more regimens. A state of equipoise is an indication for a RCT, because there are no ethical concerns about one regimen being better for a particular patient and the do no harm to the other half.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What are the 4 phases of RCT?

A
  1. Phase 1: done healthy volunteers to determine pharmacokinetics and pharmacological effects
  2. Phase 2: done in limited number of subjects having the disease of interest, to determine drug effectiveness, short term effects
  3. Phase 3: larger number, used as it would be marketed, highly regulated, done in ideal populations, may be compared to placebo
  4. Phase 4: post-marketing, to obtain additional info such as different dosage, durations, diseases and sub populations
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What are the advantages of randomization?

A
  • Best defence against selection bias (more effective when combined with blinding)
  • Minimizes/prevents confounding by balancing the distribution of risk factors
  • Eliminates conscious bias (physician and patient)
  • Balances unknown factors among treatment groups
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

How do we maintain balance and effectiveness of randomization?

A
  • Blinding of treatment (single, double and triple)
  • High retention of study participants (compliance/adherence, positive effect in itself: Hawthorn effect)
  • Intention to treat analysis (best way, one randomized always analyzed)
    • Preserves “power” of randomization by maintaining balance in the analysis
    • Allows for protocol deviations
    • Ensures validity
    • Provides a pragmatic estimate
    • May introduce misclassification of exposure, not measure biologic efficacy and decrease study power
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What are the disadvantages of randomization?

A

Patient or physician may not care to participate in experiment involving a chance mechanism to decide treatment, may influence patient-physician relationship

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What is Stratified randomization?

A

A mini-randomization to ensure balance within characteristics

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Why do we use placebo groups?

A

If you don’t have a standard treatment for the studies disease, it is ethically appropriate to have a placebo group

  • It helps understand placebo effect
  • I helps to correctly ascribe effects of RX
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What is the best way to prevent bias?

A

Randomize and blind treatment

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What is number needed to treat (NNT) and how do we calculate it?

A

The NNT is the number of patients who need to be treated in order to prevent one additional bad outcome. It is the inverse of the Absolute Risk Reduction (ARR).

How to Calculate NNTs

_NNT = 1/ARR_
ARR = |CER - EER|

CER = control group event rate

EER = experimental group event rate

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What are the strenghts of an RCT?

A
  • Lowest susceptibility to bias if well executed
  • Best design to address questions of therapy efficacy
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What are the limitations of an RCT?

A
  • Expensive
  • Not reflective of routine practice
  • Underpowered to study uncommon but serious adverse effects
  • Infeasible in many instances because of ethical concerns
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What is essential to insure sufficent power of an RCT?

A

A SUFFICIENT SAMPLE !

  • To find a difference between the two treatments
  • Usually arbitrarily set at 80%
  • The larger the sample size, the greater the power
  • Needs to takes into account loss-to-follow-up
  • We monitor it periodically and stop if doesn’t meet criteria
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

An RCT finds an association between a treatment and an outcome while in reality there was none. What type of error is that?

A

TYPE 1 (ALPHA)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

An RCT finds NO association between a treatment and an outcome while in reality there WAS one. What type of error is that?

A

TYPE 2 (BETA)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

2 sources of error?

A

Systematic (bias)

Random (sampling variation)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

Why do we need statistics?

A

Inference of the results from the sample to a whole population - because the true result/effect of a treatment is unknown.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

What is the null hypothesis?

A

Hypothesis that there is no difference between outcomes of the group treatment VS no treatment.

In RCT’s, it’s the hypothesis that the true difference between experimental and placebo is 0.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

What does it mean when p-value<0.05?

A

It means that the result of the studies are statistically significant.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

What are the limitations of using the p-value?

A
  • Provides no info on the magnitude of the treatment difference
  • Promotes dichotomous decision-making (meaning a p-value of 0.051 and p-value of 0.5 are both classified together as non-significant, even if one is really close to be significant and the other one is really bad)
  • 5% cut-off is completely arbitrary
  • Statistical significance does not always translate into clinical significance.
20
Q

How do we calculate the p-value?

A

Calculate the test-statistic first (Test stat = treatment difference / SE)

Then find the corresponding p-value in a standard tables of probabilities.

21
Q

What type of error are we possibly making if the result of the RCTs = new treatment is not significantly different?

A

Type 2 error (probability = beta)

22
Q

What type of error are we possbily making if RCT results say that new treatment is significantly different?

A

Type 1 error (probability = alpha)

23
Q

Which error should we suspect when the results are non-significant but clinically important?

A

Type 2

24
Q

What assumption should we make when using confidence intervals?

A

A normality assumption of data

25
Q

How do we calculate the margin of error?

A

1.96 * (SD/(racine de n))

26
Q

What does the CI represent?

A

The plausible range of values for the true population value

27
Q

When the sample size increases, the CI …

A

Narrows. The more samples we have, the closer the mean is to the true value.

28
Q

If a study shows a significant statistical difference between 2 treatments, should the interval include 0?

A

oh HELL NO

29
Q

When measure is a ratio, a CI that includes 1 is considered….

A

Non-significant

30
Q

Negative trials: what should we do to determine whether or not the size is adequate?

A

Look at the upper bound: if it is more than the smallest difference considered important, the trial is not definitive - further trials are required. If the upper bound is less than the smallest difference that is clinically imporatnt, trial is definitely negative.

31
Q

What are the advantages of CI?

A
  • Not dichotomized “accept-reject” decision
  • Provides information on the size of treatment effect
  • Provides an indication of random variation or precision of the estimate
  • Provides an indication of statistical significance
  • Better addresses clinical relevance of study findings
  • Tells us whether sample size large enough to answer research question
32
Q

What gives more info: p-values or CI?

A

CI

33
Q

True or false, P-values and CI are addressing the internal validity of studies.

A

FALSE, THE PRECISION

34
Q
A
35
Q

What are the determinants of strengths of recommendation?

A
  1. Balance of desirable and undesirable effects
  2. Quality of evidence
  3. Values and preferences
  4. Costs
36
Q

What are Traditional, narrative reviews?

A

Usually written by experts in the field, are qualitative, narrative summaries of evidence on a given topic. Typically, they involve informal and subjective methods to collect and interpret information

37
Q

What is a systematic review ?

A

A review in which there is a comprehensive search for relevant studies on a specific topic, and those identified are then appraised and synthesized according to a predetermined and explicit method.

38
Q

What is a meta-analysis?

A

The statistical combination of at least 2 studies to produce a single estimate of the effect of the healthcare intervention under consideration

39
Q

What is an Individual patient data meta-analyses (pooled analyses)?

A

A meta-analysis that involves obtaining raw data on all patients from each of the trials directly and then re-analyzing them.

40
Q

What are the alements of a systematic review?

A
  • Formulate the review question & write a protocol
    • PICO Question
  • Systematically search for and include primary studies
  • Assess study quality
  • Extract data
  • Analyze data (if applicable)
  • Interpret results & write a report
41
Q

What are the types of analysis you can do with systematic review and meta-analysis?

A
  • Fixed-effects Models: Assume no between-study heterogeneity
  • Random-effects Models: Assume the presence of both within- and between-study heterogeneity
  • I2 Statistic: Estimates the proportion of the total heterogeneity (or variance) that is due to between-study heterogeneity.
42
Q

True or false: a systematic review should always lead to a meta-analysis

A

BEN NON PO TOUJOURS.

Meta-analyses should only be conducted when estimates can be interpreted in a meaningful manner. Otherwise, a systematic review is more appropriate.

43
Q

True or false: a meta-analysis should always be preceded by a systematic review

A

ÇA OUI !! Sinon, it can mislead. Apprécions mon fran-glais.

44
Q

When can a meta-analysis mislead?

A
  • When a meta-analysis is done outside of a systematic review
  • When poor quality studies are included or when quality issues are ignored
  • When inadequate attention is given to heterogeneity
  • In the presence of reporting biases
45
Q

When is it good to do a meta-analysis or a systematic review ?

A

•Limited or no RCT data are available

–No equipoise exist/not ethical

–RCTs of inadequate size and/or duration

  • Need to synthesize the available evidence
  • To better understand the heterogeneity of studies examining a given issue
46
Q

What bias can affect meta-analysis and systematic review?

A

Systematic reviews and meta-analyses are affected by the limitations and biases of the primary research in addition to their own.