SPH1904 Midterms Flashcards
Rank the following on the evidence pyramid :
- randomised controlled trials (RCTs)
- critically-appraised individual articles (article synopses)
- cohort studies
- case-controlled studies (case series/reports)
- background information/expert opinion
- systematic review
- critically-appraised topics (evidence syntheses)
From strongest to weakest evidence :
- systematic review
- critically-appraised topics (evidence syntheses)
- critically-appraised individual articles (article synopses)
- randomised controlled trials (RCTs)
- cohort studies
- case-controlled studies case series/reports
- background information/expert opinion
What does the PICO framework stand for?
P - Patient
I - Intervention
C - Comparison
O - Outcome
Rank the study designs for a Therapy/Prevention Question
RCT > Cohort > Case-control > Case series
Rank the study designs for a Etiology/Harm Question
RCT (rarely) > Cohort > Case-control > Cross-sectional
Rank the study designs for a Prognosis Question
Cohort > Case-control > Case series
Rank the study designs for a Diagnosis Question
Prospective, blind comparison to a gold standard
List the Boolean operators and how to perform a search
- – Include all forms of the same word
AND – Include words of different topics/concepts
OR – Include words of similar topics/concepts
What does the CRAAP tool stand for?
C : Currency – timeliness of information
R : Relevance – importance of information for your needs
A : Authority – source of information
A : Accuracy – reliability, truthfulness and correctness of the content
P : Purpose – the reason the information exists
For each of the following study designs [Cross-sectional study, Case-control study, Cohort-study, RCT] , answer the following questions :
1. When are the exposure and outcome measured?
2. Are participants matched on the outcome?
3. Are participants randomised to the exposure?
Case-control : Outcome measured after the exposure ; Participants matched on the outcome ; Participants not randomised to the exposure
———-
Cohort study : Outcome measured after the exposure ; Participants are not matched on outcome ; Participants are not randomised to the exposure
———
RCT : Outcome measured after the exposure ; Participants are not matched on outcome ; Participants are randomised to the exposure
List the Bradford-Hill criteria.
TSBDECCSA
1. Strength: A larger effect size makes it more likely that the association is causal.
2. Temporality: Cause must precede effect
3. Consistency: Multiple studies showing the same findings increase the credibility
4. Biological Plausibility: A rationale or theoretical basis for the finding
5. Dose response relationship: Greater the amount of exposure, the greater amount of harm
6. Experimental Evidence: Experiments make a causal association more plausible
7. Coherence: Cause-effect relationship doesn’t conflict with what is known or has other competing hypotheses
8. Specificity: Effect has only one cause
9. Analogy: A commonly accepted phenomenon in one area can be applied to another area
Define prevalence.
The frequency/number of cases in a specific population who have a specific condition during a specified time region (e.g. all current cases within a specific time frame)
Define incidence.
Incidence is the frequency/occurrence of a new outcome of interest during the specified time period being examined.
Explain the characteristics, benefits and limitations of each of the following sampling methods (Non-probability sampling, Simple Random Sampling, Stratified Sampling, Cluster Sampling, Systematic Sampling)
Non-Probability Sampling :
- Characteristics – Selects samples based on non-random criteria like convenience or judgement
- Benefits – Easy, Quick and Cost-effective
- Limitations – Biased, not representative, findings cannot be generalised
Simple Random Sampling :
- Characteristics – Every member of the population has an equal chance of being selected
- Benefits – Unbiased, easy to understand, generalisable results
- Limitations - Challenging for large populations, potential to miss segments
Stratified Sampling :
- Characteristics – Divides population into subgroups ; samples from each subgroup
- Benefits – Represents all segments, more accurate, reduces sampling error
- Limitations – Requires population knowledge, complex, higher costs
Cluster sampling :
- Characteristics – Divides populations into clusters ; selects entire clusters randomly
- Benefits – Economical for large areas, reduces costs and logistical issues
- Limitations – Higher sampling error, less precision, clusters may vary internally
Systematic Sampling :
- Characteristics – Selects members at a fixed interval from a randomly chosen starting point
- Benefits – Simple, suitable for large populations
- Limitations – Risk of periodicity bias, not truly random, ordered list required
List the data type, numerical summary, graphical display (1 variable) and graphical display (2 variables) for categorical and numerical data respectively.
Categorical data :
[Data type] Nominal : Response categories with no order and Ordinal : Response categories with order
[Numerical summary] Frequency, n ; Percentage, %
[Graphical display (1 variable)] : Bar chart
[Graphical display (2 variables)] : Grouped Bar charts, Multiple boxplots (mixed data)
Numerical data :
[Data type] Continuous / Discrete
[Numerical summary] Normally distributed, Skewed division, Mean, Standard deviation, Median, Range median, IQR
[Graphical display (1 variable)] : Histogram/Boxplot ; Mean, Mode, Median
[Graphical display (2 variable)] :
Scatter plots, Pearson/Spearman Correlation Coefficient, Multiple boxplots (mixed data)
Define sensitivity.
Probability of correcting identifying persons with the disease.
Describe Specificity.
Probability of correctly identifying persons without the disease.
Write down how sensitivity of the test is calculated and why it is important.
- Sensitivity = True positives/Disease present
- Higher sensitivity = Fewer false negatives (fewer people with disease who are missed by the test)
- Important for screening tests
Write down how specificity of the test is calculated and why it is important.
- Specificity = True negatives/Disease absent
- High specificity = Fewer false positives (Fewer people without the disease being incorrectly identified as positive by the test)
- Important for confirmatory tests
Explain the significance of ROC curves.
ROC curves are Receiver Operating Characteristic (ROC) curves which show sensitivity and specificity of a diagnostic test at various cut-off points.
Explain the significance of the area under the ROC curve.
Area under the ROC curve is an indication of the overall accuracy of the test.
Explain the significance of the positive predictive value.
The probability that a person with a positive test has the disease (post-test probability of the disease)
Explain the significance of the Negative Predictive Value.
Probability that a person with a negative test does not have the disease (post-test probability of no disease)
Describe ways to reduce systematic errors.
Conduct :
- Probability Sampling
- Calibration
- Standardisation
- Randomisation
- Statistical adjustments
Describe ways to reduce random errors.
Conduct :
- Increase sample size
What is the significance of the 95% Confidence Interval
- We are 95% confident that the true parameter lies in the interval (x,y)
- If we repeat the study 100 times with random samples of the same size, 95 of the 100 confidence intervals will contain the true parameter
- 95% CI = Mean +/- level of confidence x standard error (SE)
What is the test to be conducted if i have 2 groups of independent numerical data that is parametric?
Two-sample t-test
What is the test to be conducted if i have 2 groups of independent numerical data that is non-parametric?
Mann-Whitney U test
What is the test to be conducted if i have >2 groups of independent numerical data that is parametric?
One-way ANOVA
What is the test to be conducted if i have >2 groups of independent numerical data that is NON-parametric?
Kruskal Wallis test
What is the test to be conducted if i have 2 groups of independent categorical data?
Chi-squared test (cell count > 5) OR the Fisher’s exact test (if the cell count <5)
Explain what is a Type I error.
- Reject the null hypothesis when it is in fact true
- Related to the significance level of the test
Explain what is a Type II error.
- Fail to reject the null hypothesis when it is false
- Related to power of test