Epi & biostats (p120-122 LHS)- part 1 Flashcards
what is confounding (2)
1- a variable other than the main one you’re studying is associated with both outcome and exposure
2- this distorts the true relationship between exposure and outcome
is a confounder on the causal pathway between exposure and outcome? (1)
1- NO - a confounder is NOT on the causal pathway
what is a simple example of an exposure-confounder-outcome triad? list the three (1, 2, 3) and give brief explanation of relationship (4)
1- exposure - coffee drinking
2- confounder - smoking
3- outcome - lung cancer
4- we’re trying to see whether there is a relationship between coffee drinking and lung cancer; at the outset there may seem to be, but if you stratify by smoking status, you’ll see that in smokers and in non-smokers, coffee drinking does not lead to lung cancer
how do you control for confounding through study design (3)
1- randomization
2- restriction
3- matching
what is randomization as a way of controlling for confounding (1)
1- it is a selection method that decides who is exposed and who is unexposed
what is restriction as a way of controlling for confounding (1)
1- method where you limit enrollment based on known confounders (e.g. do not include alcohol drinkers in study)
what is matching as a way of controlling for confounding (1)
1- match based on potential confounders (e.eg. match cohort based on age)
how do you control for confounding through data analysis (3)
1- stratification
2- standardization
3- multivariate analysis
what is stratification as a way of controlling for confounding (2)
1- stratification by particular variable (e.g. age, sex)
2- can tell whether or not the relationship between an exposure and outcome is due to a confounder
what is standardization as a way of controlling for confounding (2)
1- can standardize by age or sex, commonly
2- adjusting results to remove the effect of a characteristic responsible for differences in comparison
what is multivariate analysis as a way of controlling for confounding (1)
1- you can control for multiple confounders using regression models
what is effect modification (1)
1- the concept that the magnitude of effect between exposure and outcome is modified by a third variable
what is an example of effect modification (2)
1- smoking is an effect modifier for the effect of radon on lung cancer
2- smokers exposed to radon have a higher risk of lung cancer than non-smokers exposed to radon
what is a causal association (1)
1- it is a real association where change in exposure produces a change in outcome
what is a not-causal association (2)
1- a real association where change in exposure does not necessarily produce a change in outcome
2- an initial apparent association may be due to confounding factor
what is a spurious association (1)
1- a false association due to various causes of bias, or simply due to chance
what are koch’s postulates - background (AGIR)(2)
1- they are 4 criteria used to establish a causative relationship between a microbe and disease
2- the 4 criteria are AGIR = association, grown & isolated, inoculation, re-isolation
what is koch’s postulate ‘association’ (1)
1- association: microorganism must be found in abundance in those who have disease, but should not be found in healthy groups
what is koch’s postulate ‘grown and isolated’ (1)
1- isolated microorganism from a diseased host can be grown in pure culture
what is koch’s postulate ‘inoculation’ (1)
1- cultured microorganism can cause disease when introduced into a healthy host
what is koch’s postulate ‘re-isolated’ (1)
1- the microorganism of interest must be re-isolated from the inoculated, diseased experimental host
what are the 9 Bradford hill criteria of causality (C-TB-SPACES) (9)
1- consistency
2- temporality
3- biological gradient
4- specificity
5- plausibility
6- analogy
7- coherence
8- experiment
9- strength
which of the bradford hill criteria is the only necessary criteria for causality (1)
1- temporality
bradford hill causality criteria: what is strength (1)
1- larger the effect size, the more likely the association is causal
bradford hill causality criteria: what is specificity (1)
1- a single risk factor consistently relates to a single effect
bradford hill causality criteria: what is plausibility (1)
1- the effect must have biologic plausibility
bradford hill causality criteria: what is analogy (1)
1- when one causal agent is known, a second, similar agent may cause same or similar disease
bradford hill causality criteria: what is consistency (1)
1- the reproducibility of association in various populations and situations
bradford hill causality criteria: what is coherence (1)
1- any new data should not be in opposition to the current evidence
bradford hill causality criteria: what is experiment (1)
1- association can be demonstrated in experimental evidence
bradford hill causality criteria: what is temporality (1)
1- exposure must always precede outcome
bradford hill causality criteria: what is biological gradient (1)
1- a causal relationship is more likely if a dose-response gradient is demonstrated
what is a proportion (1)
1- the fraction or percentage one quantity makes up of another quantity
is the numerator included in the denominator in a proportion? (1)
1- yes - numerator is part of denominator in a proportion
what is an example of a proportion (1)
1- number of canadians with cancer divided by the total population
what is a ratio (1)
1- the comparison of one quantity to another
is the numerator included in the denominator in a ratio? (1)
1- no - numerator is not included in the denominator of a ratio
what is an example of a ratio (1)
1- male to female ratio in a class
what is a rate (1)
1- Measure of the frequency with which an event occurs in a defined population in a defined time
what is included in the denominator of ‘rate’ which is not part of the denominator for ‘ratio’ or ‘proportion’ (1)
1- time is in the denominator for rate, and not for ratio or proportion
what is an example of a rate (1)
1- number of deaths per hundred thousand Canadians in one year
what is crude rate (1)
1- overall rate for a defined population without adjustment for confounders
what is adjusted rate (1)
1- summary rate that has been statistically modified to remove the effect of one or more confounding factors
what is incidence rate (1)
1- Number of new cases of a disease that occur during a specified period of time in a population at
risk of developing the disease
how can incidence rate be measured (3)
1- frequency count
2- proportion of population at risk
3- rate per unit of time
what is prevalance (1)
1- Proportion of people with a condition divided by total number of persons at risk of the condition in
the population at that time
what are factors that influence prevalance (6)
1- incidence
2- duration that prevalence is being measured over
3- increased case fatality
4- differential migration of people with or without disease
5- changes in diagnosis/definition of disease
6- changes in reporting or case finding
what is attack rate (1)
1- number of people who become ill over the number of people exposed, or proportion of exposed persons who become infected
over what time period is an attack rate generally calculated (1)
1- The time period may not be indicated, but would typically refer to the period of the outbreak
what is the calculation for attack rate (1)
1- Calculation: AR = number of new cases / number of persons at risk
for a measure of association (RR, OR, etc.), what does the strength of association indicate? (3)
1-
< 1.0 = Exposure associated with decreased risk of outcome
2-
1.0 = No association between exposure and outcome
3-
> 1.0 = Exposure associated with increased risk of outcome
what is absolute risk (AR) (1)
1- Incidence of a disease in a population
how would you calculate absolute risk of disease in the exposed as per 2x2 table(1)
1-
AR in exposed (Ie) = a/a+b
how would you calculate absolute risk of disease in the unexposed as per 2x2 table (1)
1-
AR in unexposed (Iu) = c/c+d
what is relative risk/RR (or risk ratio) (1)
1- Ratio of the absolute risk of disease among the exposed group to the absolute risk of the disease
among the non-exposed group, i.e. RR = Ie / Iu
what does relative risk indicate (1)
1- Quantifies how much more (hazard) or less (protective) exposed persons develop an outcome
how do you calculate relative risk, as per 2x2 table (1)
1- Calculation =
1 - RR = Ie / Iu = (a/a+b) / (c/c+d)
how would you interpret, for example, a relative risk of 2.5 for smoking (exposure) and cancer (outcome) (1)
1-
if RR = 2.5 for smoking and cancer, smokers are 2.5 times more likely to develop cancer than non-smokers
what is relative risk reduction (RRR) (1)
1- Relative decrease in risk of an event in the exposed group compared to the unexposed group
how do you calculate relative risk reduction (1)
1- Calculation = (Ie - Iu) / Iu
what is odds ratio (2)
1- Probability of an event occurring relative to it not occurring
2- When a health outcome is rare, OR approximates RR
how do you calculate odds ratio (1)
1-
ad/bc
what is Attributable Risk (AR) (i.e. Risk Difference (RD), or Absolute Risk Reduction (ARR)) (1)
1- Number or incidence of cases of disease among exposed individuals that can be attributed to that exposure
does attributable risk consider exposed, unexposed or total population (1)
1- attributable risk considers exposed population only
what is the purpose of attributable risk (1)
1- it is a measure of the potential for prevention of disease if the exposure could be eliminated
how do you calculate attributable risk (1)
1-
Ie - Iu
how would you interpret an attributable risk (1)
1- if AR is X, smokers had X number of additional cases per 100 participants compared to non-
smokers
what is number needed to treat (NNT) (1)
1- Number of patients who need to be treated to prevent one additional bad outcome
how do you calculate number needed to treat (1)
1- Calculation =
1 / ARR (or attributable risk)
what is Attributable Risk Percent (AR%) (i.e. Attributable Fraction (AF), or Attributable Risk in Exposed) (1)
1- Percentage or proportion of cases of a disease among exposed individuals that can be attributed to
that exposure as a percentage/fraction
does attributable risk percent consider exposed, unexposed or total population (1)
1- 1- attributable risk percent considers exposed population only
how do you calculate attributable risk percent (1)
1-
Calculation =
(Ie - Iu) / Ie
how would you interpret attributable risk percent (1)
1- if AR% is X, among smokers, X percent of lung cancers were attributable to smoking
what is Population Attributable Risk (PAR) (1)
1- Number or incidence of cases of disease in the population that can be attributed to an exposure
does population attributable risk consider exposed, unexposed or total population (1)
1- Considers both those exposed and unexposed in
the entire population
what is the purpose of population attributable risk (1)
1- to determine the number of cases of disease that would not occur in a population if the factor were eliminated or no one was exposed
how do you calculate population attributable risk (1)
1-
Calculation: PAR = It– Iu
It = # disease total / population, i.e. (a+c/n)
Iu = # disease in unexposed / # unexposed i.e. (c/c+d)
how do you interpret population attributable risk (1)
1- if PAR = X, there were X number of additional cases of lung cancer per 100 study participants attributable to smoking
what is population attributable risk % (or population attributable fraction, PAF) (1)
1- Percentage or proportion of cases of disease in the population that can be attributed to an exposure
does population attributable fraction consider exposed, unexposed or total population (1)
1- Considers both those exposed and unexposed in
the entire population
what is the purpose of population attributable fraction (1)
1- determine proportional reduction in population disease that would occur if an exposure to a risk factor were reduced to an alternative ideal exposure scenario (e.g., no tobacco use)
how do you calculate population attributable fraction (2)
1-
Calculation =
(IT – IU) / IT
It = # disease total / population, i.e. (a+c/n)
Iu = # disease in unexposed / # unexposed i.e. (c/c+d)
2-
Alternative =
[Pexp(RRexp -1)] / [(Pexp(RRexp -1) + 1]
RR = relative risk
Pexp = probabilty of exposure (a+b)/n
how do you interpret population attributable fraction (1)
1- in the study population, X percent of lung cancer is attributable to smoking
what is sensitivity (true positive rate, SnOUT) (1)
1- Ability of a test to identify correctly those who have the disease when disease is present
how do you calculate sensitivity (1)
1- Calculation =
TP / (TP + FN)
how do you interpret sensitivity - when sensitivity is high (1)
1- If sensitivity is high, you can be confident that a negative test result can rule out disease - since you rarely miss people with the disease
how do you interpret sensitivity - when sensitivity is low (1)
1- If the sensitivity is low the test may mistakenly classify people who do actually have the condition as not having it (False Negatives) - i.e. you can’t confidently rule out disease with a negative test result
how do you calculate the false negativity rate (1)
1-
False negativity rate = 1 - sensitivity
what purpose is a highly sensitive test good for (1)
1- highly sensitive tests are good as screening tests
e.g., COVID-19 testing would benefit from high sensitivity so all potential cases can be isolated quickly, even if that means briefly isolating those who do not have the disease until follow-up test results return
what is specificity (true negative rate, SpIN) (1)
1- Ability of a test to correctly identify those without a disease when disease is absent
how do you calculate specificity (1)
1- Calculation =
TN / (FP + TN)
how do you interpret specificity - when specificity is high (1)
1- If specificity is high you can have high confidence that a positive test result can rule in a disease - Rarely misclassify people w/o disease as having the disease
how do you interpret specificity - when specificity is low (1)
1- If the specificity is low the test will mistakenly classify people who do not actually have the condition as having it (False Positive) - i.e. you cannot be confident that a positive rest result actually means you have disease
how do you calculate the false positivity rate (1)
1-
False positive rate = 1 – specificity
for what purpose is a highly specific test good for (1)
1- highly specific tests are good confirmatory tests
e.g., after a patient screens positive for HIV on a rapid test, the confirmatory test should be highly specific to ensure that the person is not given a false positive diagnosis (of a serious illness)
what is positive predictive value (PPV) (1)
1- Probability that an individual with a positive test actually has the disease
how do you calculate positive predictive value (1)
1-
Calculation =
TP / (TP + FP)
what happens to positive predictive value as the prevalence of a disease increases (1)
1- PPV increase as prevalence increases
what is Negative predictive value (NPV) (1)
1- Probability that an individual with a negative test does not have a disease
how do you calculate negative predictive value (1)
1-
Calculation =
TN / (TN + FN)
what happens to negative predictive value as the prevalence of a disease decreases (1)
1- NPV increases as prevalence decreases
what is a likelihood ratio (1)
1- Combines sensitivity and specificity to indicate “by how much having a test result will reduce the uncertainty of making a given diagnosis”
what is Positive likelihood ratio (LR+) (1)
1- expresses how much a positive test result increases the odds that a patient has the disease
how do you calculate positive likelihood ratio (1)
1-
Calculation = sensitivity / (1 – specificity)
what is Negative likelihood ratio (LR-) (1)
1- expresses how much a negative test result decreases the odds that a patient has the disease
how do you calculate negative likelihood ratio (1)
1- Calculation =
(1 – sensitivity) / specificity
