EBM 3

Question 2

PECO framework — what does each letter stand for?

Answer 2

P = Population | E = Exposure | C = Comparison | O = Outcome

Question 3

Best study design for disease prevalence

Answer 3

Cross-sectional (++++)— only design suitable; Case-control, Cohort, RCT = No

Question 4

Best study design for evaluation of diagnostic tests

Answer 4

Cross-sectional (++++) | Cohort (++) | RCT (+/-) | Case-control = No

Question 5

Best study design for risk factors of RARE diseases

Answer 5

Case-control (++++) — only suitable design; Cross-sectional, Cohort, RCT = No

Question 6

Best study design for risk factors of COMMON diseases

Answer 6

Cohort (++++) | Case-control (++++) | Cross-sectional (+/-) | RCT = No

Question 7

Best study design for incidence/prognosis

Answer 7

Cohort (++++) — only strong design; Cross-sectional & Case-control = No | RCT = +/-

Question 8

Best study design for therapy/prevention

Answer 8

RCT (++++) — gold standard; Case-control & Cohort = +/- | Cross-sectional = No

Question 9

Study design quick-match: prevalence → ? | rare disease RF → ? | incidence → ? | therapy → ?

Answer 9

Prevalence → Cross-sectional | Rare disease RF → Case-control | Incidence → Cohort | Therapy → RCT

Question 10

Cross-sectional study — definition

Answer 10

Measures prevalence of exposure AND outcome simultaneously (“snapshot”) — no follow-up

Question 11

Cross-sectional study — 2 uses

Answer 11

Descriptive: disease distribution (person, place, time) | Analytical: compare prevalence of exposure between groups

Question 12

Cross-sectional study — who is included?

Answer 12

Everyone, regardless of exposure or outcome status (not selected based on either)

Question 13

Cross-sectional study — strengths

Answer 13

Good for prevalence; easy, quick, hypothesis-generating

Question 14

Cross-sectional study — key limitation & why

Answer 14

Cannot establish temporal sequence → cannot prove causation (exposure & outcome measured at same time)

Question 15

Reverse causality — definition & example

Answer 15

Cannot tell which came first: exposure or disease | e.g. HTN and ESRD — did HTN cause ESRD or did ESRD cause HTN?

Question 16

Reverse causality — exception

Answer 16

Fixed exposures (e.g. genetics) are unaffected — genes clearly precede disease, so temporal sequence is not an issue

Question 17

Cross-sectional study — definition in one sentence?

Answer 17

Measures prevalence of exposure AND outcome simultaneously — “snapshot picture” with no follow-up

Question 18

Cross-sectional study — 2 uses?

Answer 18

Descriptive: disease distribution (person, place, time). Analytical: compare prevalence of exposure between groups

Question 19

Cross-sectional study — who are the participants?

Answer 19

Everyone regardless of disease or exposure status — not selected based on either

Question 20

Cross-sectional study — 2 strengths?

Answer 20

Good for measuring prevalence; easy, quick, and hypothesis-generating

Question 21

Cross-sectional study — 2 limitations?

Answer 21

Cannot establish temporal sequence → no causation. Risk of reverse causality (exposure and outcome measured at same time)

Question 22

Cross-sectional — reverse causality example and exception?

Answer 22

Example: did HTN cause ESRD or ESRD cause HTN? Exception: fixed exposures (genetics) — genes always precede disease so causation direction is clear​​​​​​​​​​​​​​​​

Question 23

Case-control study — definition and direction?

Answer 23

Retrospective study — starts with disease status, looks BACK at past exposure

Question 24

Case-control study — main use and when is it the preferred design?

Answer 24

Study risk factors, especially for RARE diseases — when disease is rare, cohort is inefficient; case-control is the design of choice

Question 25

Case-control study — 3 steps?

Answer 25

1. Select cases (disease present). 2. Select controls (disease absent). 3. Assess past exposure retrospectively

Question 26

Case-control study — measure of association and why not RR?

Answer 26

Odds Ratio (OR). Cannot calculate RR because there is no incidence, no at-risk population, and no time follow-up

Question 27

Case-control assumption 1 — what must cases represent?

Answer 27

Cases must represent the exposure distribution of ALL diseased people in the community. e.g. if 30% of all MI patients in the community smoke → ~30% of study cases should be smokers

Question 28

Case-control assumption 2 — what must controls represent?

Answer 28

Controls must represent the exposure distribution of the NON-diseased population in the community

Question 29

VENTIGUE — In a smoking/MI case-control study, controls are recruited from a COPD ward. What bias direction and why?

Answer 29

Controls from COPD ward → ↑ smokers among controls → OR artificially LOW → downward bias (makes smoking look protective)

Question 30

VENTIGUE — Same study, controls recruited from a wellness spa where only 2% smoke. What bias direction?

Answer 30

Controls underrepresent smokers → OR artificially HIGH → upward bias (exaggerates smoking as risk factor)

Question 31

Case-control — key rule for control selection bias?

Answer 31

If controls are recruited from a place that ATTRACTS or REPELS people with the exposure → results will be biased. Controls must come from the same source population as cases

Question 32

Case-control assumption 3 — equal ascertainment of exposure: what does this mean?

Answer 32

Exposure must be measured the SAME WAY in cases and controls — standardized method (same questionnaire, same labs, same definitions)

Question 33

VENTIGUE — A researcher studies risk factors for a rare liver cancer. 40 cases and 40 matched controls. Cases: 30 exposed, 10 unexposed. Controls: 20 exposed, 20 unexposed. Calculate OR?

Answer 33

OR = (30×20)/(10×20) = 600/200 = 3.0 → exposed are 3× more likely to have the disease

Question 34

VENTIGUE — OR = 0.38 in a smoking/MI study using COPD controls. What is the correct interpretation and what went wrong?

Answer 34

OR <1 suggests smoking is protective — but this is a downward bias artifact from selecting controls with high baseline smoking rate (COPD ward). Result is invalid​​​​​​​​​​​​​​​​

Question 35

Case-control advantages — 4 key points?

Answer 35

Only way to study rare disease etiology; can study multiple risk factors simultaneously; less time-consuming, less expensive, smaller sample size than cohort

Question 36

Case-control limitations — list 6?

Answer 36

Temporal relationship unclear (recall bias); cannot measure prevalence; cannot measure incidence; cannot calculate RR directly; prone to recall and selection bias; limited to one outcome; difficulty studying rare exposures

Question 37

Case-control — why can’t it measure prevalence or incidence?

Answer 37

Prevalence: not sampling whole population (use cross-sectional). Incidence: no follow-up of at-risk individuals (use cohort)

Question 38

Case-control — why use OR instead of RR, and when does OR approximate RR?

Answer 38

No incidence data available → can’t calculate RR. OR approximates RR when the disease is rare

Question 39

Case-control — recall bias: definition and why it occurs?

Answer 39

Relies on participants recalling past exposure → cases may remember differently than controls → systematic error in exposure measurement

Question 40

Case-control — matching: purpose, variables used, and limitation?

Answer 40

Controls matched to cases by age, sex, race → reduces confounding. Limitation: matching too many factors makes it difficult to find suitable controls

Question 41

VENTIGUE — Researcher uses case-control to study a common disease and wants to calculate RR. What are two problems?

Answer 41

1. Cannot calculate RR directly — no incidence data. 2. OR does not approximate RR when disease is common — use cohort instead for common diseases