1
Q
What is a systematic sample type?
A
- An algorithm is used to select a subset
2
Q
What is a stratified sample type?
A
- Separate representations of more that one subgroup
- Example
3
Q
What is sampling bias?
A
- Occurs when a sample is selected that does not truly represent the population
4
Q
What is measurement bias?
A
- Systemic error arising from inaccurate measurement of subjects
5
Q
What is recall bias?
A
- Occurs when individuals with a disease are more prone to recalling or believing they were exposed to a possible causal factor than those without disease
6
Q
What is a confounder?
A
- A confounder is a variable that is related to both the exposure and the outcome but is not measured or is not distributed equally between groups
- Ex: having >4 children increases risk of developing trisomy 21, here advanced maternal age (as mothers with 4 kids typically older) is a confounder
7
Q
What is Sensitivity?
A
The probability that a patient with a condition will have a positive test result
- Sensitivity = True positive / True positive + False negative
8
Q
What is Specificity?
A
- The probability that a patient without a disease will have a negative test result
- Sensitivity = True negative/ True negative + False positive
9
Q
What is a likelihood ratio?
A
- The likelihood that a given test result would be expected in a patient with disease compared with the likelihood that the same result would be expected in a patient without disease
10
Q
How do you calculate a positive likelihood ratio?
A
- likelihood patient with disease gets positive test (this is sensitivity)
- Compared to likelihood that patient without disease gets false positive (FP/TN + FP) - Note this is the inverse of specificity (so 1-specificity equation gives same answer)
- Overall:
(Tp/Tp+Fn) / (Fp/(Tn+Fp))
11
Q
How do you calculate a negative likelihood ratio?
A
- Probability of a patient with disease getting negative test result (This is the inverse of sensitivity)
- Compared to the probability of a patient without disease getting a negative test result (This is specificity)
1-sensitivity/specificty
- Overall
(Fn/(Tp+Fn))/(Tn/(Tn+Fp))
12
Q
What is pre-test probability?
A
- An estimate of the likelihood a particular patient has a given disease based on known factors
- Likelihood that a person has disease of interest before the test is performed
13
Q
What is post-test probability?
A
- The estimated likelihood, after the administration of a diagnostic test, that a patient has the disease of interest
14
Q
Define efficacy
A
- The extent to which a specific intervention produces a beneficial result under ideal conditions
15
Q
Define effectiveness
A
- Measures the benefit of an intervention under usual conditions of clinical care
- Considers both the efficacy of the intervention and real world impacts (compliance, acceptance, ..)
16
Q
Define efficiency
A
- A measure of economy of an intervention with known effectiveness
- Helps in determining optimal use of resources (money, time, personnel..)
17
Q
What type of study will use an odds ratio in its analysis?
A
- Case-control
18
Q
What is an odds ratio?
A
- Odds of a particular exposure among persons with a specific disease, divided by the corresponding odds of exposure among persons without the disease of interest
19
Q
A relative risk is calculated for which type of study?
A
- Cohort study
20
Q
What is number needed to treat?
A
The number of patients who need to be treated to achieve one additional favorable outcome
21
Q
What is number needed to harm?
A
- Number of patients who, if they received treatment, would lead to one additional patient being harmed, compared with patients who received a control.
22
Q
Describe what a qualitative study design is?
A
- Qualitative data is all about quality and cannot actually be measured with numbers
- Used to generate hypothesis (why? what does it mean?)
- Employs a bottom up style (observe->look for pattern-> tentative hypothesis->theory)
- Sampling: want sample to cover your concept or idea (not necessarily representative of general)
- Typically small samples used but detailed info taken from them
23
Q
Describe what a quantitative study design is?
A
- Information about quantity, can be measured with numbers
- Used to test a hypothesis (what, how many)
- Top down style (Theory->hypothesis->observation->confirmation)
- Sampling: want to be representative of general population studied
- Large number of participants
24
Q
What is the observational study type where sampling is based on exposure?
A
- Cohort
- sampling based on presence (exposed) or absence (unexposed) of a risk factor of interest
25
What is a cohort study?
- An observational study in which subjects are sampled based on presence (exposed) or absence (unexposed) of a risk factor of interest
- Subjects are followed over time for the development of a disease outcome of interest
26
What is the observational study type where sampling is based on outcome?
- Case control
| - Information is the collected about earlier exposure to risk factors of interest
27
What is a case-control study?
- An observational study in which subjects are sampled based on the presence (cases) or absence (control) of the disease of interest. Information is then collected about earlier exposure to risk factors of interest
28
What is a cross sectional study?
An observational and analytical investigation in which subjects are sampled at a FIXED POINT OR PERIOD OF TIME and the associations between the concurrent presence or absence of risk factors are then investigated
29
What is an analytic study? What are the 3 types?
- A group of observational studies that are used to test a specific hypothesis
- Cohort
- Cross sectional
- Case control
30
What is the ecological fallacy
- An association between summary characteristics across populations without actual linkage of the characteristics within individual persons
- Ex: One study concluded that red wine consumption lowered the risk of CVS disease. When really there was a higher consumption of red wine in France and the people of France have a lower rate of CVS death than other countries
31
What is randomization?
- Procedure for assigning the treatments of patients by chance
- Done to ensure, as much as possible, equal distribution of known and unknown factors except for the experimental exposure
32
What is meant by:
1) Single blind
2) Double blind
3) Triple blind
1) The subject does not know their group assignment
2) The subject and observer both do not know the group assignment
3) The subject, observer and analyst are all unaware of the group assignment (rarely done)
33
True or False: The control group always receives a placebo treatment
- False
| - Placebo or Standard therapy if another known treatment option exists
34
What are some advantages of randomized controlled studies?
- Gold standard study provides the strongest level of evidence
- Allows assessment of effects of intervention while minimizing bias
35
What are some disadvantages of randomized controlled trials?
- Some exposures cannot be randomized (poverty, smoking) due to ethical concerns
- Difficult to study rare events, as you would require very large sample size
- Difficult to randomly allocate larger groups (communities, neighbourhoods)
- Expensive
36
What is a meta-analysis?
- A statistical combination or integration of the results of several independent research studies that are considered to be combinable
37
What are some advantages of meta-analysis?
- Attempts to overcome the problem of reduced power associated with the smaller sample size of individual studies
- Ability to control for inter-study variation
38
What are the dis-advantages of meta-analysis?
- Sources of bias may not be controlled for
- Reliance on published studies increases change of publication bias
- Decision to include of exclude particular studies is subjective
39
What is the difference between per-protocol analysis and intention to treat analysis?
- Per Protocol: Only patients who complete the entire study are counted towards the results
- Intention to treat analysis: Data is used from all patients, including those who did not complete the study and put into the group they were originally randomized to (treatment group may affect the rate of dropout which this may detect)
40
Define mean?
- The sum of all observations, divided by the total number of variables (average)
41
Define Median
- Value at the 50th percentile (line up all values in order and median is the middle)
- A better measure of the central tendency than the mean if data is skewed
42
Define Mode
The most frequently observed value in a series (the number that occurs the most)
43
Mean, median and mode are all measures of what?
- Central Tendency
44
What are the 3 measures of dispersion?
1) Range
2) Variance
3) Standard Deviation
45
Define Range
- The largest value minus the smallest value
46
Define variance
- A measure of the spread of data
| - The sum of the squared distances of each data point from the mean ie. (x-mean)^2 + (y-mean)^2 ...
47
Define standard deviation?
- The average distance of data points from the mean
| - The positive square root of the variance
48
What does a null hypothesis state?
- No relationship exists between the two stated variables
| - ie. no association between the hypothesized exposure and the outcome
49
What does the alternative hypothesis say?
- A relationship DOES exist between the two stated variables
50
What is Type I error?
- Alpha error
- States that there is an effect when in reality there is none
- (FALSE POSTIVE)
- The probability of Type I error is denoted by the p-value
- Studies tend to be designed to minimize type I error which can have a larger significance that type II
51
What is Type II error?
- Beta error
- Stating a difference is by chance when in reality it is present
- (FALSE NEGATIVE)
- Higher level of Type II error permitted in most studies
- Can be used to calculate statistical power
52
What is a P-value
The probability of obtaining a result equal to or "more extreme" than what was actually observed when the null hypothesis is true
- The probability of committing type I error
53
Define Power?
- The probability of rejecting the null hypothesis when it is in fact false (probability of a true positive)
- 1 - Type II error
- Ie: the probability of finding a specified difference to be statistically significant at a given p-value
- Power increases with sample size
54
How can you increase Power?
- Increase the sample size
55
What is a confidence interval?
- Provides a range of values withing which the TRUE population result (the mean) lies
- Ex: In a 95% confidence interval the true value will be within the interval range 95% of the time
56
Would you have more confidence in a study with a wide or narrow confidence interval?
- Narrow
- ie/ narrow 95% confidence interval would mean that 95% of the time the true mean lies withing a narrow range (if wider than there is more variability/uncertainty)
57
What is discrete data?
- Categorical date (gender, marital status)
- Ordinal (High, Medium, Low)
- Differs from continuous data (such as age)
58
What is continuous data?
- A measured number that is continuous
| - Ex: Cholesterol level, age, hemoglobin level
59
Differentiate between internal and external validity
- Internal: Degree to which findings represent findings in the study population
- External: AKA generalizability
- degree to which the results of the study can be generalized to other situations or populations
60
What are the 9 Bradford Hill criteria for causation? *note not all need to be met to establish causation*
1) Strength of association (frequency factor is found in disease and without disease)
2) Consistency (Same outcome in other pop'n or study design type)
3) Specificity (Association particular to intervention or measured outcome)
4) Temporal relationship (did exposure occur before onset of disease)
5) Biological gradient (dose-response relationship)
6) Biologic plausibility (does it make biological sense)
7) Coherence (does it make sense following what we know about science, logically)
8) Experimental evidence (strongest support)
9) Analogy (do other established associations provide a model for this type of relationship)
61
Which of the following is considered to be higher quality evidence:
- Expert opinion vs case control study?
- Case-control
62
Which of the following is considered to be higher quality evidence:
- Case control study vs Cohort study
- Cohort study
63
Which of the following is considered to be higher quality evidence:
- Cohort study vs Randomized control study?
- randomized controlled trial
64
Which of the following is considered to be higher quality evidence:
- Randomized control study vs Critically appraised individual articles (article synopses)
- Critically appraised individual articles (article synopses)
65
Which of the following is considered to be higher quality evidence:
Critically appraised individual articles (article synopses) vs Critically appraised topics (evidence synthesis)
- Critically appraised topics (evidence synthesis)
66
Which of the following is considered to be higher quality evidence:
Critically appraised topics (evidence synthesis) vs Systematic review
- Systematic review
67
In epidemiology how is risk defined? How do you calculate it?
- The likelihood that an individual will contract disease (or cumulative incidence)
- The proportion of individuals without disease who will contract disease over a specified period of time
- Calculated by:
New cases / persons at risk
* Most helpful in determining what proportion of the population will become ill in a defined period of time
68
What is the difference between risk and incidence rate?
- Incidence is a RATE (measured in person years)
| - Risk is a proportion, no units
69
What is an incidence rate? What does it measure? How is it calculated?
- Incidence rate reflects the occurrence of new cases of a disease of interest
- It is a measure of rapidity with which newly diagnosed cases of the disease of interest develop
- IR= New cases/Person time (ex: person years)
* It is a rate so must include unit of time
70
Define survival
The probability of remaining alive for a specific length of time
= number newly diagnosed - deaths observed in study time / number of newly diagnosed
71
What is a rate'?
Measures frequency with which an event occurs in a defined population
- Ex. deaths per 100 canadians in one year
72
What is a proportion?
- Number of a group that is compared to the whole
| Ex: Number of Canadians with cancer / All Canadians
73
What is a ratio?
- Obtained by dividing one quantity by another
| - Ex Males:Females
74
What is the positive predictive value? How is it calculated?
- The percentage of persons with a positive test result who actually have the disease of interest
= True positive / True positive + False positive
(True positive) / (All positive test results)
75
What is the pretest odds of disease? How is it calculated?
- Estimate of probability before testing that the patient has the disease of interest (pretest prob), divided by the probability that he patient does not have the disease (inverse of pre-test prob)
- (Pretest prob) / (1-Pretest prob)
76
What is the posttest odds of disease? How is it calculated?
- The estimate, after testing, of the probability that a patient has the disease of interest divided by the probability that the patient does not have the disease of interest
Posttest odds = pretest odds x LR+
77
What is posttest probability of disease? How is it calculated?
- The estimated likelihood, after testing, that a a patient has the disease of interest
- Posttest odds of disease / (1+ posttest odds)
78
What is lead-time bias?
An increase in survival, as measured from detection of disease to death, without lengthening of life
- If a screening test detects disease earlier but does not lengthen life compared to a patient without screening it creates the illusion that screening increases survival time
79
What is length biased sampling?
- Occurs when disease detected by screening programs are less aggressive than diseases detected without screening
Ex: Br Ca found on screening may be less aggressive
- less aggressive disease tends to grow slower, allowing for a longer period of time where screening can detect disease before clinical symptoms will appear
80
What are the 6 criteria for successful screening programs?
1) High risk pop'n must exist
- so you know who to screen
2) Effective early intervention must reduce morbid/mortality
- knowing they have disease early must improve outcome
3) Screening test acceptable in target pop'n
- ppl willing to have testing
4) Minimal risk associated with screening test
5) Screening test must be sensitive and specific
6) Work-up for positive test must have acceptable morbidity relating to number of false positive results
- Will work-up in false positive cause more harm, than good caused by catching disease early
Ex. PSA screening (not recommended for many reasons but work-up for false positives associated with morbidity)
LMCC Study - CLEO/ELOAM
flashcards
Decks in class (12)
# Cards
-
Canadian Health Care System26
-
Principals of Ethics7
-
Confidentiality10
-
Consent and Capacity22
-
Public Health21
-
Measuring Health and Disease in a Population12
-
Epidemiology80
-
Ethics in reproductive technology and end of life care19
-
Disease outbreak13
-
Environmental and Occupational Health18
-
Cohort Study and Relative risk9
-
Case-Control Studies and Odds ratio7
