Week 3

Question 1

What is a CAT

Answer 1

Critically Appraised Topic (CAT):
- A broader synthesis of evidence on a specific clinical question, incorporating multiple sources to provide a well-rounded conclusion
While CAP evaluates a single research study, a CAR combines evidence from several critically appraised papers to provide an evidence-based conclusion

Question 2

Purpose of CATS

Answer 2

• Answer a specific clinical question using multiple sources.
• Provide a broader evidence base than a single CAP.
• Summarize and synthesize findings from multiple studies.
  Offer practical recommendations for clinical practice.

Question 3

Structuring a CAT

Answer 3

1. Clinical Question (PICO format)
2. List of Included Studies (CAPs)
3. Summary of Findings (Comparison of studies, common trends, contradictions)
4. Critical Appraisal (Strengths and limitations across all studies)
5. Clinical Bottom Line (Practical implications, recommendations)
   - A CAT must ensure that only high-quality, relevant studies are included. Studies that do not directly address the clinical question should be excluded from the review

Question 4

Key different between CAT and CAP

Answer 4

CAP

Focuses on a single research article

Evaluates the validity and impact of one study

Summarizes one paper’s strengths, weaknesses, and clinical relevance

Useful for evaluating a new or pivotal study

CAT

Synthesizes multiple articles on a topic

Provides an overview of the best evidence

Offers a broader clinical conclusion based on multiple studies

Helps answer a specific clinical question

Question 5

What are case control studies

Answer 5

• The case group (disease positive) is compared to the control group on the variable/s of interest
• The control group do not have the disease but should be similar (or matched) on other characteristics e.g. gender, age, SES, ethnicity, geographical region
• The design is observational because we cannot allocate individuals to an exposed versus non-exposed group
• Case-control studies are retrospective because they rely upon identifying prior events or behaviours that might be implicated in the development of disease
• Existing databases and public records can be used e.g. lung cancer and smoking
• Survey method commonly used
• Case control studies are relatively cheap and quick to perform, compared to other methods such as cohort studies or randomized control trials, that can be expensive and labour-intensive
  Retrospective data is helpful when it is impractical or unethical to perform an experiment e.g. examining the incidence of fetal alcohol syndrome

Question 6

Example of case-control studies

Answer 6

• A retrospective case-control study was designed. 33 patients subjected to a total laryngectomy with the same protocol were included in both groups. Patients who required a complex surgical reconstruction were excluded in order to avoid biases
  We conducted a case-control study of 33 Jamaican children 7-12 years old with uncomplicated epilepsy and 33 of their classroom peers matched for age and gender whether epilepsy resulted in differences in cognitive ability and school achievement and if socioeconomic status or the environment had a moderating effect on any difference, intelligence, language, memory, attention, executive function and mathematics ability

Question 7

Matching in clinical research

Answer 7

• Matching is a statistical technique which is used to evaluate the effect of a treatment by comparing the treated and the non-treated units in an observational study or quasi-experiment (e.g. when the treatment is not randomly designed)
• The goal of matching is, for every treated unit to find one (or more) non-treated unit (s) with similar observable characteristics against whom the effect of the treatment can be assessed
  By matching treated units to similar non-treated units, matching enables a comparison of outcomes among treated and non-treated units so estimate the effect of the treatment without reduced bias due to confounding

Question 8

Example of matching in clinical research

Answer 8

• If you have a group of 15 people with Parkinson’s (7M, 8F) find similar characteristics, and the only thing that divides them is the condition.
• Why researchers love twin studies as they are genetically the same e.g. when one gets sick but the other doesn’t they can see why
• 2:1 ration: For each case/ person with a disease (e.g. 15), find 30 matched controls (two identical twins without a disease). (almost perfect set up for a matched control study)

Question 9

Advantages of case-control studies

Answer 9

• Cheap, easy and quick (existing data)
• Multiple exposures can be examined
• Rare diseases and diseases with long latency can be studies
  Suitable when randomization is unethical

Question 10

Disadvantages of case control studies

Answer 10

• Case and control selection troublesome
• Subject to bias (selection, recall, misclassification)
• Direct incidence estimation is not possible
• Temporal relationship is not clear
• Multiple outcomes cannot be studies
• If the incidence of exposure is high, it is difficult to show the difference between case and controls
  Reverse causation is a problem in interpretation

Question 11

What is a statistical test

Answer 11

• Comparing means: Comparing averages, and
• Repeated measurements: Follow people backward through time, looking back in time to see if it has been re-measured
• Correlations: e.g. no. cigarettes of smokes increases, so does chance of getting lung cancer. Calculate additional risk somebody has if they are exposed to that.
  Odds ratio: Calculate additional risk, how much more chance do I have to end up in ICU, how much more prone am I to injury

Question 12

Correlation does not mean causation

Answer 12

• Correlation between the amount of ice cream people eat at time points during the day and how many breaking and entering’s you have. At times people eat more ice cream, less burglars are active due to doing it at night when ice cream isn’t eaten
• Statistical correlation between number of ice creams being eaten (low), and high occurrences of a house being robbed
• More ice cream you eat = less burglaries (rubbish)
• But there is a confounding factor which is sunshine
  Be careful with correlations, as it doesn’t mean there is a causation

Question 13

Cause and effect relationships

Answer 13

• Temporal sequence
• Strength of association
• Consistency of association
• Biological gradient (does-response relation)
• Specificity of association
• Biological plausibility
• Coherence with existing knowledge
• Experimental evidence
• Analogy
• Did exposure precede the outcome
• How strong is the effect, measured as relative risk or odds ration
• Has this effect been seen by others
• Does increased exposure result in more of the outcome
• Does exposure lead only to the outcome
• Does the association make sense
• Is the association consistent with available evidence
• Has a randomized controlled trial been done
  Is the association similar to others

Question 14

What are case-control studies used for

Answer 14

• Investigate risk factors and protective factors for diseases and other health conditions
• Important to health practice
  Commonly used to investigate risk and protective factors, but can also be used to investigate the effects of treatments and therapies

Question 15

Characteristics of a case-control study

Answer 15

• Sample of participants is selected from the population and grouped on the basis of whether they do or do not have a particular disease or health conditions
• The groups are then compared retrospectively in terms of their history of exposure to the factor or factors of interest
  The exposure may be to one or more possible risk factors and/or protective factors for the disease or other health condition. A simple example of a case-control study is provided by a study where the disease is coronary heart disease and the exposure (risk factor) or interest is tobacco smoking.

Question 16

Design of a case control study

Answer 16

Consider: selection of participants for each group and the sources of information about exposure

Question 17

Selection of cases

Answer 17

• Clear defined inclusion criteria with respect to the disease or health condition
• Disease status may be established through a variety of means
• These include examination of medical records, scrutiny of disease registries, examination of death certificates, physical exam, use of questionnaire
• Cases may come from a variety of sources: Hospital, geographical area, registries or the general population
• Advantage of selecting cases from the general population is that it can help eliminate sources of selection bias, which can result in study participants not being typical of the population of which they are part
  Cases can include newly diagnosed, individuals affected by the disease regardless of when they were diagnosed

Question 18

Selection of controls

Answer 18

• Controls should be as similar as possible to the cases, and both cases and controls need to have had the potential for exposure to the factor of interest
• May be achieved by selecting the controls from the same population as the cases, and matching on as many characteristics as possible
• May come from a variety of sources

Question 19

Measures of exposure

Answer 19

• Exposure status may be identified through a variety of means, such as examination of medical records, scrutiny of disease registries and immunisation registries, physical examination of participants and participant questionnaires
  Two types of exposure:
• Dichotomous: Where the participants are classified into one of two categories (e.g. present/absent
• Continuous: Where the amount of exposure can have any value within a specific range (e.g. body weight, number of cigarettes smoked)
• Important to have complete and accurate information

Question 20

Analysis of data

Answer 20

• Calculate the odds ratio not risk ratio
  However where a disease is uncommon, the OR will be very close to the RR

Question 21

Odds ratio formula

Answer 21

OR= A/C divided by B/D
Where a, b, c, d are values derived from a co-called 2x2 table

e.g.
Smoker with lung cancer present/Non smoker with lung cancer present

Divided by

Smoker with lung cancer absent/non smoker with lung cancer absent

Question 22

How do i interpret an OR

Answer 22

• Can’t use relative risk in case-controls studies
• Compares odds of exposure in cases to odds of exposure in controls
• How much higher is the odds of exposure in the cases than in the controls
• If the OR equals 1.0, the exposure is the same for both the cases and the controls.
• If the OR is more than 1.0, the exposure is more common for the cases than for the controls.
• If the OR is less than 1.0, the exposure is less common for the cases than for the controls.
  It follows from the above, that if the odds ratio is greater (or less) than one, then there is a degree of association between the health condition of interest and exposure.

Question 23

Odds ratio in case-control studies and randomized trials

Answer 23

• In therapy trials where the outcome is dichotomous, the intervention groups appear on the rows and the outcomes in the columns
  Odds are calculated as the number of people in a group with the even divided by the people without the event and the OR ratio is between these two Odds.

Question 24

Bias in case control studies

Answer 24

• Selection: Difference in the way which the cases are chosen
• Misclassification: Misidentification of disease and/or exposure status. Occurs when there is an error in the way in which either disease status or exposure is determined for cases and controls
  Recall: Cases are more likely than controls to recall an exposure that did not in fact occur

Question 25

Confounding

Answer 25

- A risk factor for the disease of other health condition that is being studied and is associated with the exposure being studied but is not part of the causal pathway. There may appear to be an association between the exposure of interest and the disease of interest but this association is not a causal one

Question 26

Strengths and limitations of case-control studies

Answer 26

Strengths: - Studying rare diseases, or long induction/latent periods - Investigate multiple risk or protective factors for a single disease or health condition - When it relates to risk, protective factors for a disease or health condition, case-control studies are the quickest and least expensive Limitations: - Ineffective for investigation exposures that occur only rarely - Prone to bias Difficult to determine temporal relationship between exposure and the disease

Question 27

C

Question 28

Confidence intervals

Answer 28

- Enables the determination of a lower limit and an upper limit, within which the real or true value (odds ratio) for a population is most likely to lie - Defined as all the possible values the population value can have between these two limits - The value for a sample (OR), provides an estimate of the value for the population form which the sample was taken - e.g. 95% CI means we can be 95% sure that the population value lies somewhere within this interval

Question 29

How does CI relate to OR

Answer 29

- If the odds ratio is greater than or less than 1, there is an association between exposure and disease - Difference between the two groups (case and controls) in the terms of exposure - What needs to be determined is whether this association is real or just a chance finding for this particular set of participants in the study ( a sample of the population) - Needs to be determined if the association between exposure and disease Is also apparent in the population from which the participants were selected, rather than just being apparent for the participants in the sample Deciding if the association seen for the sample is statistically significant

Question 30

CI and 1 figure

Answer 30

If the CI does not include 1, and both the lower and upper limits of the CI more than 1, then the association between exposure and disease is statistically significant If the confidence interval does not include 1 and both the lower and upper limits of the CI are less than 1, then the association between exposure and disease are deemed statisitcally significant