Chandigarh Call Girls Service ❤️🍑 9809698092 👄🫦Independent Escort Service Cha...
Third year ebm 2 2013
1. Dr Ellie Hothersall
This version: September
2013#DundeePublicH
Evidence Based Medicine Part 2:
Observational studies
2. Dr Ellie Hothersall
This version: September 2013#DundeePublicH
Evidence Based Medicine Part 2:
Observational studies
3. Last week
The world (of evidence based medicine) is very large…
…We are very small
Randomised controlled trials, grouped together into
magnificent meta-analyses are the best way to obtain
evidence that an intervention does or does not have an
effect
Twitter in lectures is interesting but complex (especially if
not everyone can get wifi access)
#DundeePublicH
5. Twitter: a second attempt
#Dundeepublich
Switch off wifi on everything except one device – i.e.
phone, OR laptop, OR ipad.
#DundeePublicH
6. Today…
AIM: To introduce some more key terms in
Evidence Based Medicine and Research
Methodology
Objectives:
Describe cohort and case control studies and describe the
key requirements of both
Define and describe Odds Ratio, Relative Risk, and
Absolute Risk
Be able to interpret Odds Ratio, Relative Risk and
Absolute Risk
#DundeePublicH
7. #DundeePublicH
Learning Outcomes
Describe the purpose of the following types of research
study: Cohort study, Case control study
Understand the types of outcome these studies are likely
to examine and the type of evidence obtained
Describe the key requirements and limitations of these
types of studies
Define, describe and interpret Odds Ratio, Relative Risk, and
Absolute Risk
8. #DundeePublicH
The Evidence Based Medicine Theme
What? When?
Research 1: The basics Neurology
Research 2: More basics Neurology
Research 3: Understanding p
values
Psychiatry
Research 4: Critical appraisal Psychiatry
Understanding Guidelines Ophthalmology
Understanding Audit Reproduction
10. Why can’t we run a
randomised controlled
trial for everything?
http://ndarc.med.unsw.edu.au/project/randomised-controlled-trial-rct-assess-acceptability-switch-buprenorphine-naloxone-suboxone%C2%AE
#DundeePublicH
11. Money
Do you know how much
an RCT costs to run?
http://www.moneysigns.net/british-money-bills/
#DundeePublicH
18. Radiation exposure
Survivors of Hiroshima and Nagasaki bombs in 1945
Data collected since 1950
Life span study (LSS)
Mabuchi et al. http://www.irpa.net/irpa9/cdrom/VOL.1/V1_19.PDF#DundeePublicH
19. The structure of a cohort study
http://sph.bu.edu/otlt/MPH-Modules/EP/EP713_CohortStudies/#DundeePublicH
20. The structure of a cohort study
http://sph.bu.edu/otlt/MPH-Modules/EP/EP713_CohortStudies/#DundeePublicH
21. The structure of a cohort study
http://sph.bu.edu/otlt/MPH-Modules/EP/EP713_CohortStudies/
http://freeonlinenews.o
rg/weird-diseases/
http://eofdreams.com/death.html
http://joeforamerica.com/2013/02/helping-friends-with-and-without-cancer
#DundeePublicH
22. Types of cohort study
Population
Exposed
Outcome
No
outcome
Not
exposed
Outcome
No
outcome
Prospective
Retrospective
#DundeePublicH
23. What do you get from a cohort study?
Estimate of risk in the general population
Ability to pick up infrequent occurrences
Ability to find outcomes you hadn’t
expected
#DundeePublicH
28. But…
Estimate of risk in the general population
Prospective better for ensuring it really is
“general”
Ability to pick up infrequent occurrences
Needs a large (= more expensive) group
Ability to find outcomes you hadn’t expected
Depends on you collecting data (prospective)
or data quality (retrospective)
#DundeePublicH
31. What does a cohort study tell us?
Risk
Disease status
Present Absent
Present
Absent
Population
Exposed
Outcome
No
outcome
Not
exposed
Outcome
No
outcome
a
b
c
d
#DundeePublicH
34. Relative risk (RR)
Pevent when exposed
RR= Pevent when not exposed
a/a+b
= d/c+d
Also known as a RISK RATIO
Incidence in exposed
Incidence in not exposed
#DundeePublicH
36. Interpreting a Relative Risk
Risk Ratio <1 Logarithmic scale
Incidence in
exposed <
incidence in
control
#DundeePublicH
37. Interpreting a Relative Risk
Risk Ratio >1 Logarithmic scale
Incidence in
exposed >
incidence in
control
#DundeePublicH
38. “If relative risk is the
guy that drug companies
always want to have at
the party, absolute risk
is the guy who never
gets invited, the total
buzz kill, the guy who
showed up with
someone's cousin once in
a bad outfit and ended
up mumbling to himself
in the corner about how
everything would be
better if people just
listened to him.”
Heisel, 2010
http://www.reportingonhealth.org/blogs/next-
pharma-love-fest-spend-some-time-absolute-risk
#DundeePublicH
48. Do mobile phones cause
brain cancer?
http://e-healthdiary.com/category/cancer/brain-cancer#DundeePublicH
49. How do you answer the question?
Randomised controlled trial
Cohort?
#DundeePublicH
50. How a case control study works
http://library.downstate.edu/EBM2/2500.htm#DundeePublicH
51. Case-control design
Risk factor
No risk
factor
Sample of
CASES
(with disease)
Sample of Controls
(disease-free)
Risk factor
No risk
factor
Exposure Disease Status
Population
#DundeePublicH
52. The direction of time
Cases and controls identified now
Data on past events collected
Backwards in time Cases
Data
Controls
http://www.yianniscove.com/film-of-the-day-back-to-
the-future-1985-dir-robert-zemeckis/
#DundeePublicH
57. Do mobile phones cause
brain cancer?
http://e-healthdiary.com/category/cancer/brain-cancer#DundeePublicH
58. The INTERPHONE study
Cardis et al, 2010. http://ije.oxfordjournals.org/content/39/3/675.full
Regular
use in the
past ≥1
year
Cases Controls
Yes 1262 1174
No 1147 1488
#DundeePublicH
60. What does a case control study tell us?
Risk
Disease status
Present Absent
Present
Absent
Population
Exposed
Outcome
No
outcome
Not
exposed
Outcome
No
outcome
a
b
c
d
#DundeePublicH
67. Do mobile phones cause
brain cancer?
http://e-healthdiary.com/category/cancer/brain-cancer#DundeePublicH
68. The INTERPHONE study
Cardis et al, 2010. http://ije.oxfordjournals.org/content/39/3/675.full
Regular
use in the
past ≥1
year
Cases Controls
Yes 1262 1174
No 1147 1488
#DundeePublicH
69. Results
Odds ratio for regular users
compared with not-regular-users
= 0.79
20% LESS likely
#DundeePublicH
71. Case-Control Studies
Advantages
Appropriate for rare
diseases
Can examine diseases
with a long latency
Can examine many
potential risk factors
Disadvantages
No incidence /
prevalence estimates
Cannot measure disease
onset
Observer bias, Recall
bias
#DundeePublicH
72. Pitfalls
Inappropriate controls
Not from same source
Bias in data collection
Dependent on data from the past
#DundeePublicH
http://bygonebureau.com/2012
/04/13/instructions-for-viewing-
the-bottomless-pit/
75. Differences between Cohort
and Case-control studies
#DundeePublicH
Cohort Case control
Studies exposure to
outcome
Traces outcome back
to exposure
Can measure lots of
outcomes
Can measure lots of
exposures
Starts by identifying
population
Starts by identifying
cases
Relative risk Odds ratio
Prone to bias from loss
to follow up
Prone to recall bias
77. Case-Control study
Study population
With the
Outcome
Without
the outcome
Direction of the research inquiry
Un-exposed
Exposed
Exposed
Un-exposed
#DundeePublicH
Time