lect 4: case control Flashcards
is case control in observational or experimental
observation
which is better at determining “causality” (cannot actually determine causality, but better at telling you the relationship between two variables): Cross sectional or case control
case control
what are the differences between observational and experimental
observational: does not intervene (the intervention/exposure/risk factor and the randomization of the study subjects are NOT manipulated by the reseachers), cannot determine causality,
experimental: intervenes (the intervention/exposure/risk factor and the randomization of the study subjects are manipulated by the reseachers), can determine causality,
true or false: alllllll observational study designs cannot determine causality
false, cohort can infer causality
which of the 3 observational study designs, which provides the least evidence for causality and which provides the most
least (cross sectional)
most (cohort)
does case control provide evidence of causality and explain
no, it can identify assocaition between exposure and the outcome but we might not be 100% sure of the temporal factor (if the exposure truly occurred before the outcomes)
are crosss sectional designs done at 1 point in time or across time
one point in time
are case control designs done at 1 point in time or across time
1 point in time
what is the general purpose of the case control study
study of causal factors (identify risk factors)
=links one variable/factor to another where one is though to be the cause of the other
in a case control study, which comes first exposure or outcome
exposure precedes outcome
true or false: in case control exposure precedes outcome
true
explain the general design of case control
you have a study of “persons with the disease of interest/CASE” and a suitable control group of persons without the disease/control
=choosing populations based on the outcome
looking retrospectively to compare how the two groups differ in their exposure level to the risk factor
true or false: in a Case study design = outcome is not known and looking back to collect data to look at frequency of exposure and compare the control and case group
false, the outcome is known
using smoking and lung cancer, explain how you would conduct a case control study
outcome of interest = lung cancer
exposure/risk factor of interest= smoking
case = people who have lung cancer
control= people who do not have lung cancer
look back retrospectively to see the exposure level to smoking that the case and control groups had to determine the assocation between level of exposure to smoking and the development of lung cancer
be able to explain/draw a 2x2 case control table (know what a, b,c,d means)
in case control, is data collected in one point in time or over time
one point in time (contrast to cohort where you measure the effect over time)
explain who is the case in case control study design
those with the outcome/condition/event
those with the outcome/condition/event
case or control
case
explain who is the control in case control study design
those without the outcome./condition/event
those without the outcome./condition/event
case or control group
control
what type of bias is common in case control and why
sampling/selection (because you are not randomizing the people in the design)
information bias (recall)
since sampling/selection bias is common in case control study, what is a way to minimize that bias
matching
explain matching in case control studies
Matching is defined as making cases and controls equivalent with respect to factors other than the exposure or factor being investigated
Cases and controls need to be similar on all other variables except the exposure and the outcome
(age sex, place of residence, time of hospitalization)
to control for confounding variables incase control study, what do we do
matching
what is the general definition of odds ration
measure of association
OR describes the odds (i.e., the degree to which) the presence of one condition is associated with another condition
E.g., The degree to which having diabetes is associated with carpal tunnel syndrome
Odds Ratio of exposure
The odds of exposure in those with the outcome (cases) compared to the odds of exposure in those without the outcome (controls)
ODDs ration explnation and formula
Odds Ratio of exposure
The odds of exposure in those with the outcome (cases) compared to the odds of exposure in those without the outcome (controls)
OR = odds of exposure in cases (a/c) / odds of exposure in controls (b/d)
odds ratio is equivalent to what in this calss
relative risk
what is the relationship between CI and OR
The OR is accompanied by a confidence interval (CI), which provides a range of values within which the true OR is likely to fall
(look to CI to know if the value from the OR is signifciatnf or not)
in a 2x2 table, outcome (disease status) is the general column or row
column
in a 2x2 table, exposure status is the general column or row
row
in a 2x2 table, what does cell A mean
disease is present and exposure is present
in a 2x2 table, what does cell c mean
disease is present but no exposure
in a 2x2 table, what does cell b mean
no disease present but have exposure
in a 2x2 table, what does cell d mean
no disease present, no exposure
in a case control study, cells A and C represent to controls or the cases
cases
in a case control study, cells B and D represent to controls or the cases
controls
odds of exposure in cases is what letters
a/c
odds of exposure in controls is what letters
b/d
be ablle to construct a 2x2 table when the exposure is not binary
if OR is greater than 1 and 95% CI does not include 1, what does that mean
exposure significant increases the odds of the outcome/disease
if OR is greater less 1 and 95% CI does not include 1, what does that mean
then the exposure significant decreases the odds of the outcome/disease (protective)
if OR is = 1 and 95% CI does not include 1, what does that mean
then exposure does not affect the odds of an outcome
what would the OR and CI look like if exposure does not affect the odds of an outcome
OR =1
CI does not include 1
what would OR and CI look like if the exposure significant decreases the odds of the outcome/disease (protective)
OR less than 1 and CI does not include 1
what would OR and CI look like if exposure significantly increases the odds of the outcome/disease
OR greater than 1 and 95% C1 does not include 1
in OR, the null value (value that determines significance) is 1 or 0
1 (odds of outcome are the same in both control and cases)
in RCT, the null value (value that determines significance) is 1 or 0
0 (difference between means)
a 95% CI that includes 1 means no statistical significance or statistical significance
none
larger CI = low or high presencion of the OR
low level of precision
narrow CI = low or high presencion of the OR
higher precision of the OR
in the assocaition between home hazards and falls, interpret the OR 9.25 (CI: 1.22-70.07
Interpretation: Children with access to high furniture had 9.25 times the odds of falling compared to children without access to high furniture.
=since CI does not include one the finding it statistically significant
=there is a low precision since large CI
in the assocaition between cell phone use and cancer, interpret an OR 1.32 (CI=0.98-1.97)
The findings of this study indicate that people who use cell phones have 1.32 times higher odds of having cancer compared to people who don’t use cell phones. However, the findings are not statistically significant as the confidence interval includes 1. (CI is narrow so good precision of the OR)
Interpretation 2: The findings of this study indicate that those who use cell phones have 32% higher odds of developing cancer compared to people who don’t use cell phones. However, the findings are not statistically significant as the confidence interval includes 1
in the assocaition between exercise and intracranial aneurism, interpret an OR 0.60 (CI=0.30-0.90)
Interpretation 1: The study’s findings indicate that people who exercise have 0.60 times the odds of having an intracranial aneurism compared to people who don’t exercise. The results are statistically significant as the confidence interval does not include 1. High precision because narrow CI
Interpretation 2: The findings of this study indicate that people who exercise have 40% lower odds of having an intracranial aneurism compared to people who don’t exercise. The results are statistically significant as the confidence interval does not include 1.
what are the strengths of case control studies
Good to study rare conditions
Good to study conditions where it would be unethical to expose the participant (ie: do not have to manipulate the exposure)
Falls hazard, smoking etc
what are the weaknesses of case control studies
unclear temporal relationship b/w exposure and disease (hard to know which comes first)
information bias (recall bias)
=ascertainement of past exposures can be difficult
selection bias (use matching)
true or false: case control studies are prospective studies and explain
false, they are retrospective because they start from the outcome and look back in time to see the exposure
in case control studies, the sample is based on the outcome or the exposure
the outcome
in a case control study, is the outcome dichotomous/binary
yes because there are only 2 possibilities (case/control, disease/no disease)
OR is a measure of association: true or false
true