PHEALTH - CAUSALITY Flashcards
List 2 study designs which are best for evaluating the potential harm of an intervention (1)
Cohort studies, Systematic reviews
Outline which study design/s are better for evaluating the potential harm of a factor, for example, cell phone use and the incidence of brain tumors. In your answer include why certain study designs are not useful for evaluating potential harm (5½)
Case reports not easily applicable to others (1)
Case-control studies prone to bias (½) and confounding (½)
Trials not useful for assessing long – term effects (1)
Cohort studies with a large representative sample (½), objective measures of exposure and outcome criteria (½) and good long-term follow-up are better (½)
Systematic reviews of all the good-quality evidence are best (1)
When appraising the evidence of a journal article which concludes that there is an association between an exposure and an outcome, outline the factors in the methodology you would consider in order to decide if the conclusions are correct (5½)
Selection & measurement bias (1): Sample not representative, unequal loss to follow up (½), measures of exposure not valid, measures of outcome not valid (½)
Confounding (½): Important differences between exposure groups not adjusted for in the analysis (½)
Chance (½): Examine Results (RR, OR, 95% CI) for statistical significance (½), was the sample size large enough to detect a real association (or to exclude a false association) (½)
Causal (½): Does the association fit the Bradford hill criteria of causality? (½)
List two (2) possible explanations for this association that you would you need to consider first when assessing the validity of each of the studies (2)
Any 2 of: Bias (selection or measurement); confounding; chance (insufficient sample size)
List three factors that you should consider when evaluating a study for validity (3) Any
Was the study design appropriate to the question?
Were comparison groups clearly defined to minimise selection bias?
Were exposure and outcome consistently and independently measured in order to minimise information bias?
Is there evidence of confounding?
Could the association be due to chance? (evaluating sample size)
Was the extent of follow-up appropriate in order to minimise follow-up bias?
Give the name of the criteria that are used to assess causality in epidemiology [1]
Bradford Hill criteria
Outline five (5) factors you would consider to determine if an association is causal according to the Bradford-Hill Criteria for causation (5)/ Outline five (5) of the Bradford Hill criteria for causation (5) (Super NB
Temporal relationship: Does the cause precede the effect?
Biologic plausibility: Is the association consistent with our understanding of biologic mechanisms?
Consistency: Have similar results been shown in other studies?
Dose-response relationship: Does increasing the exposure lead to an increased effect?
Strength of association: What is the measured size of the risk?
Precision: The precision of the estimate of the risk is reflected by the width of the confidence interval (CI) – the narrower CI the greater the precision.
Cessation of exposure: Does the effect disappear if the exposure is removed
Outline four (4) distinct factors that should be appraised in a study to determine if an apparent association is real (7½)
Bias: Selection (is the sample representative) (½) & measurement (have they measure the disease properly using proper diagnostic tools, validity of measurement of exposure and outcome) (½)
Confounding (½) an independent factor which is related to the exposure and outcome. Must be adjusted for in the analysis (½)
Chance (½) - examine the OR & RR for size of the effect (½) and the associated confidence interval (½), look at p-values (½), sample size should be large enough (½)
Causality (½) - if the association is not due to chance, confounding or bias; then apply Bradford –Hill criteria of causality to determine if the association is causal, i.e.
Temporality (½) Strength of effect (½) Consistency (½) Biological plausibility of explanation of association (½) Dose-response effect (½)
You conclude that all the studies are well designed and that the observed association is therefore valid. Describe the three (3) major criteria that you would use in assessing whether the association is causal or not (3)
Temporal relationship: cause precedes effect
Biologic plausibility: association consistent with understanding of biologic mechanisms
Consistency: Different studies investigating the same hypothesis produce similar findings.
Dose-response effect
Define a necessary cause in epidemiology (1)
A factor/cause that must be present in order for the disease to occur.
Differentiate between a necessary cause, sufficient cause and component cause of a disease (3)
Necessary: Must be present in all disease cases
Component: Individual factors that work together with the necessary cause to produce disease
Sufficient: A combination of factors that is sufficient to produce disease in at least some people.
Identify five (5) component causes of diabetes mellitus type 2 (2½) (NB)
Inadequate diet, Obesity, Genetic predisposition, Physical inactivity, Stress, Drugs (corticosteroids & oral contraceptives), Pregnancy, Insulin resistance, Cushing’s syndrome, Ethnicity, PCOS, Pancreatitis, Trauma to pancreas, Increased/ older age, HLA defect
Is there a necessary cause for diabetes mellitus and explain your answer (1)
No (½) as there isn’t a factor that has to be present in order for diabetes to occur (½)
After representing a client who sued a cigarette manufacturer for negligence in failing to warn the client of the risks associated with smoking, Mr Sher recently reduced his smoking consumption from 30 cigarettes a day to less than 10 per day.
One of the experts brought to court to testify was an epidemiologist who testified that the risk of smoking-related illness exhibited a dose-response relationship to the number of cigarettes smoked per day.
Briefly explain what you understand by a dose-response relationship and how cutting back from 30 to less than 10 cigarettes per day would change Mr Sher’s risk (3)
- A dose-response relationship implies that as the exposure increases, so does the risk [1]
- This is one of the criteria for identifying a causal relationship in epidemiology. Usually, the dose-response is a linear relationship.
- If that is the case for smoking and stroke, Mr Sher would have reduced his risk three-fold by cutting his cigarette consumption to 10 cigarettes a day [1]
- However, if the risk is cumulative (as is likely), cutting down threefold will only reduce his risk from the time he cuts down, so overall, he will not enjoy a threefold reduction in his risk.
We know that smoking is a major risk factor for lung cancer. If you read a study that reported a relative risk of lung cancer from smoking 20 cigarettes a day as 6.0 (95% confidence interval 2.5 to 11.0), explain this Relative Risk to a non-epidemiologist [4]
A Relative Risk is a measure of the strength of association between a risk factor and an outcome. In this case, it indicates that a person who smokes 20 cigarettes a day is 6 times more likely to develop lung cancer. The confidence interval indicates that this is a statistically significant association (does not include 1), and that we are 95% certain that the true Relative Risk is between 2.5 and 11. A RR of 6 is a very strong association.
