Draw-it-out method: Cohort vs case-control

Cohort vs case-control questions are “free points” on Step exams once you stop trying to memorize definitions and start drawing the study flow in 10 seconds. Here’s a quick-hit, shareable method you can use on any vignette.

---

The Draw-it-out Method (10 seconds)

When you see a study design question, draw a 2×2 table and then decide: Did they start by choosing exposure or outcome?

Your universal 2×2

Now the key move:

• Cohort: investigators start with exposure (row selection) and watch disease happen.
• Case-control: investigators start with outcome (column selection) and look backward for exposure.

---

Cohort Study (Start with Exposure → Follow Forward)

One-liner

“Cohort = Chosen by exposure, then Follow to outcome.”

Draw it like this

1. Pick people by exposure status (Exposure + vs Exposure −)
2. Follow them forward to see who develops disease (Disease + vs Disease −)

Sketch:

• Start with Exposure + and Exposure − groups
• Arrow forward to Disease + / Disease −

What you can calculate (high-yield)

In a cohort, you can calculate incidence and relative risk:

• Risk in exposed: $R_E = \frac{a}{a+b}$
• Risk in unexposed: $R_U = \frac{c}{c+d}$
• Relative Risk (RR): $RR=\frac{a/(a+b)}{c/(c+d)}$
• Attributable Risk (Risk Difference): $AR = R_E - R_U$

When cohort is favored (Step-style)

• Best for rare exposures (e.g., asbestos exposure)
• Good for studying multiple outcomes from one exposure
• Often used for prognosis questions (“what happens over time?”)

Common pitfalls USMLE likes

• Loss to follow-up → threatens validity (attrition bias)
• Time and cost can be high (especially prospective cohorts)

---

Case-Control Study (Start with Disease → Look Backward)

One-liner

“Case-control = Cases first, then check prior exposure.”

Draw it like this

1. Pick people by disease status (Cases = Disease +, Controls = Disease −)
2. Look backward to see how many were exposed

Sketch:

• Start with Disease + and Disease − groups
• Arrow backward to Exposure + / Exposure −

What you can calculate (high-yield)

In a case-control, you cannot directly calculate incidence or RR (because the number of cases/controls is set by the investigator). You calculate an odds ratio:

$OR=\frac{a/c}{b/d}=\frac{ad}{bc}$

High-yield equivalence:
When the disease is rare, $OR \approx RR$ (the “rare disease assumption”).

When case-control is favored (Step-style)

• Best for rare diseases (e.g., pancreatic cancer)
• Good for diseases with long latency (e.g., mesothelioma)
• Efficient: faster and cheaper than cohort

Classic biases tested

• Recall bias: cases remember exposures differently than controls
• Selection bias: controls not representative of the source population

---

The “Start Point” Rule (Mnemonic Device)

If they start with exposure → Cohort

Think: “Expose → Enroll → Event”

If they start with disease → Case-control

Think: “Case → Control → Cause (prior exposure)”

Fast vignette tell:

• “Researchers enrolled smokers and nonsmokers and followed them for 10 years…” → Cohort
• “Researchers enrolled patients with lung cancer and matched controls, then asked about smoking…” → Case-control

---

Quick Comparison Table (Exam-Ready)

---

Mini Drill: 3 One-Sentence Prompts

1. “We identify patients with MI and compare to controls; then check prior cocaine use.”
   → Case-control (start with disease)

2. “We identify patients who took OCPs vs not, then follow for DVT.”
   → Cohort (start with exposure)

3. “We use a cancer registry to find cases/controls and assess past radiation exposure.”
   → Case-control (registry gives outcome-first sampling)

---

Step 1/2 High-Yield Takeaways

• Cohort → incidence + RR; good for rare exposures; watch loss to follow-up.
• Case-control → OR; best for rare diseases; watch recall/selection bias.
• $OR \approx RR$ when the disease is rare.