PBL 4 Flashcards
Why is the case control better than cohort
Quicker to do usually – the outcome follow up in some cohort studies will be long.
More practical if is rare – (cohort would need many more patients)
Usually cheaper – fewer patients and quicker.
Why is the case control worse than cohort
May suffer from recall bias.
Cannot calculate absolute risks.
Different sorts of people may be more likely to take part, depending on whether they are a case or a control, not true of a cohort where patients are usually all well at the start
Can only study one outcome (cohort studies can look at several outcomes)
May suffer from bias in selecting patients for cases and controls.
Why are case control and cohort similar
Both can suffer from confounding – hence less accurate than randomised controlled trials – although matching deals with some confounding in case control studies
Why is a clinical trial not used
It is unethical to test something that may be potentially harmful, let alone the ethics of giving people recreational drugs!
Which study should have been used
Case control because it is more useful and practical
- outcome is rare and some of the exposures are reasonably common
What are the pros and cons for screening medical students for this type of cancer? Would you recommend all medical students get screened for this? Why or why not?
Students will be expected to talk about tailoring of information to person’s needs. Because this cancer is so rare, it may not be a good idea to screen all medical students and put them at risk for false positives.
Pros
- cancer screening may give you an indication of cancer before any symptoms develop
- cancer screening may find cancer at an early stage when treatment is more likely to be curative
- if cancer is diagnosed at an earlier stage treatment is more likely to be successful
Cons
- cancer screening can miss a number of cancers and provide false reaurrance as no cancer screening test is 100% accurate
- cancer screening can lead to unnecessary worry and investigations when there is no cancer present
- positive screening result may cause anxiety by diagnosing a slow growing tumour that may never value any harm or symptoms
How could you adapt your screening process to accommodate people who identify as nonbinary?
Depends on how person is registered with GP whether they are registered as the gender they were assigned with/to or not.
What potential discrimination and stigma could Shahid be worried about? What social determinants are intersecting and compounding here?
- Stereotyping and stigma occur in relation to some external feature such as skin colour, race, visible disease, gender or disability.
Race, gender and sexuality are the main intersecting or clustering of social influences.
We do not know if poverty is an issue although we might assume that most medical students are poor.
Likewise, we do not know if violence or abuse is an issue, if we look at this from a syndemic suffering framework perspective.
Students are asked to consider whether social determinants be seen as potential confounders.
How do you determine whether the results is causal
Bradford Hill criteria
What do you have to do before checking the bradford hill criteria
you have to know that the association is statistically significant, otherwise there could be no relationship
name the factors that make up the Bradford hills criteria
- strength of the assocaition
- dose reposnes
- reversibility
- temporality (does the exposure occur a reasonable time before the disease)
- consistency
- biological plausibility
- Coherence of evidence
- specificity
describe the parts of the Bradford hill criteria in the study
Strength of the association
- This is measured by OR or - RR, If OR >10 or <1/10 then it is strong.
- OR for Marijuana use is 2.42 for Nonseminoma / Miexed GCT so it is not that strong
Dose response
- Table 3: Frequency <1 vs ≥1 per week amongst current and former users does not show dose response. However the confidence intervals are wide so it does not preclude a dose- response effect
- Similarly <10 years vs ≥10 years doesn’t show dose response in current users, but is in right direction for former users. However the confidence intervals are wide so it does not preclude a dose- response effect
Reversibility
- If you stop the exposure does the risk change revert to 1.
- Table 3 does give current users and former users, with the risk for former users not reverting back to 1. However it might be that you need to have stopped for more than, say, 5 years for the risk to revert to 1. So this study does not have the power to detect reversibility.
Temporality
- An increased risk for current users is irrelevant as clearly there must be a time lag for the disease to develop.
- The OR is increased for > 10 years though it isn’t significant due to power of study (median age is 27 so unlikely for half of them to have 10 years exposure). Some evidence for causality
Consistency [B10]
- Has the result been repeated by different people in different places and in different circumstances and times? Yes the authors refer to two other published case-control studies
Biological Plausibility
- It helps if we think the result is biologically plausible.
The authors have given details as to why it is biologically plausible.
Coherence of Evidence
- the results should be consistent with other sorts of studies such as cohort and cross-sectional as well as perhaps animal models.
- The authors state that both TGCT and marijuana rates have been increasing in recent decades. There are no other cohort studies referred to.
Specificity
- Is the relationship with the exposure limited to specific diseases.
- Table 3 demonstrates that Marijuana use is linked to Nonseminoma rather than Seminoma tumours.
Conclusions: Causal, but not strong evidence. Other reason for the association would be confounding by other unidentified risk factors
How is the strength of association measured by in the Bradford hill criteria
- Strength of association is measured by either the odds ratio or the relative risk
- if the odds ratio is greter than 10 or less than 1/10 then it is a strong assocation
what is a casual relationship
A causal relationship is one where the cause directly affects the outcome of interest.
(There can be many causes and a causal relationship does not necessarily mean that all those with the cause will get the outcome).
What are the controls that are used in the study
The controls reflect the population from which cases have arisen and are matched for age, race, ethnicity and neighbourhood, i.e. important confounders. The study demonstrates how hard it is to find matched controls.
