HSRT Quantitative Clinical Reasoning 2 — Questions and Answers
Question 1: A drug reduces systolic BP by 8 mmHg on average (95% CI: 2–14 mmHg, p=0.01). A colleague says: 'The effect is statistically proven and clinically definitive.' Which quantitative critique is MOST appropriate?
- The CI shows uncertainty from 2 to 14 mmHg; clinical significance of an 8 mmHg reduction is modest and depends on baseline BP and cardiovascular risk (Correct answer)
- p=0.01 confirms the finding is highly reliable, making the colleague correct
- The CI should be narrowed to 6–10 mmHg before clinical recommendations are made
- Statistical significance always implies clinical significance in hypertension research
Correct answer: The CI shows uncertainty from 2 to 14 mmHg; clinical significance of an 8 mmHg reduction is modest and depends on baseline BP and cardiovascular risk
Statistical significance (p=0.01) does not equal clinical significance. An 8 mmHg average with CI 2–14 mmHg may or may not be meaningful depending on clinical context.
The p-value of 0.01 indicates the observed effect is unlikely due to chance, but it says nothing about clinical importance. An 8 mmHg systolic reduction may be clinically relevant in a high-risk patient with pre-existing cardiovascular disease or initial SBP of 180 mmHg, but relatively modest in a low-risk patient with borderline hypertension. The 95% CI (2–14 mmHg) shows the true effect could be as small as 2 mmHg (barely perceptible clinically) or as large as 14 mmHg (substantial). Clinical significance requires interpreting effect size in patient context, not just p-values.
Question 2: A clinical trial reports the Number Needed to Treat (NNT) for a new anticoagulant to prevent one stroke is 50 over 5 years. A colleague argues this means the drug is not very effective. How should you respond quantitatively?
- NNT of 50 must be weighed against the NNH (Number Needed to Harm) for bleeding risk and the baseline risk of stroke — context determines clinical value (Correct answer)
- NNT of 50 is very high and definitively indicates the drug is ineffective
- NNT is only meaningful when compared to placebo studies, not active comparators
- NNT of 50 should be converted to absolute risk reduction before any conclusion is drawn
Correct answer: NNT of 50 must be weighed against the NNH (Number Needed to Harm) for bleeding risk and the baseline risk of stroke — context determines clinical value
NNT must be interpreted alongside NNH, baseline risk, and the severity of the outcome — an NNT of 50 for stroke prevention may be excellent given stroke's severity.
NNT (Number Needed to Treat) = 1/ARR. An NNT of 50 means 50 patients must be treated for 5 years to prevent one stroke. Whether this is 'effective' depends on: (1) What is the NNH (e.g., one major bleed per 20 patients)? — risk-benefit balance; (2) What is the baseline stroke risk? — if stroke risk is 10%/5yr, NNT of 50 represents a 2% ARR, which is substantial given stroke's disability burden; (3) Patient values regarding stroke vs. bleeding. NNT alone cannot determine clinical value — it requires context. The colleague's assertion that NNT of 50 is 'ineffective' reflects a quantitative misunderstanding.
Question 3: A clinical study reports a relative risk reduction (RRR) of 50% for a new cancer screening program. The control group event rate (cancer death) is 2%. What is the absolute risk reduction (ARR)?
- 1% (50% of 2%) (Correct answer)
- 50% (same as the RRR)
- 48% (100% minus 2% minus 50%)
- Cannot be calculated from this information alone
Correct answer: 1% (50% of 2%)
ARR = RRR × baseline rate = 50% × 2% = 1%. Relative risk reduction always needs to be converted to absolute terms to assess clinical impact.
Absolute Risk Reduction = Control Event Rate × Relative Risk Reduction = 0.02 × 0.50 = 0.01 = 1%. This is a critical quantitative literacy skill: RRR of 50% sounds impressive but when the baseline rate is low (2%), the ARR is only 1% — meaning 100 patients must be screened to prevent one additional cancer death (NNT = 1/ARR = 100). Pharmaceutical and health promotion marketing routinely emphasizes RRR over ARR because it sounds more impressive. Clinicians must convert to ARR and NNT to make meaningful patient-level risk-benefit assessments.
Question 4: A patient asks: 'My PSA test came back positive. Does that mean I have prostate cancer?' The test has 80% sensitivity and 70% specificity. The prevalence of prostate cancer in men his age is 5%. What should the clinician communicate about the positive predictive value?
- The PPV is approximately 12–15%, meaning most men with a positive PSA at this prevalence do NOT have prostate cancer (Correct answer)
- Because sensitivity is 80%, the patient almost certainly has cancer
- The PPV is 70%, the same as the specificity
- PPV cannot be estimated without the specific cutoff value used
Correct answer: The PPV is approximately 12–15%, meaning most men with a positive PSA at this prevalence do NOT have prostate cancer
At 5% prevalence with 80% sensitivity and 70% specificity, the PPV is approximately 12–15% — the majority of positives are false positives.
Using a 2×2 table with 1000 men: 50 have cancer (5% prevalence); sensitivity 80% → 40 true positives + 10 false negatives; 950 don't have cancer; specificity 70% → 665 true negatives + 285 false positives. PPV = 40/(40+285) = 40/325 ≈ 12.3%. This means only about 1 in 8 positive PSA tests at this prevalence indicates actual cancer. Patients commonly misinterpret a positive test as a near-certain diagnosis. Communicating PPV in context of prevalence is essential for informed consent and shared decision-making. This illustrates Bayes' theorem applied to clinical testing.
Question 5: A hospital compares its 30-day mortality rate (8%) for cardiac surgery against a national benchmark (6%). Before concluding performance is substandard, which quantitative adjustment is MOST essential?
- Risk-adjust for patient case mix — hospitals treating higher-acuity, comorbid patients are expected to have higher crude mortality rates (Correct answer)
- Report the finding immediately to hospital administration as a quality failure
- Exclude outlier cases to normalize the rate to the benchmark
- Increase sample size by including non-cardiac surgical cases
Correct answer: Risk-adjust for patient case mix — hospitals treating higher-acuity, comorbid patients are expected to have higher crude mortality rates
Crude mortality comparisons across institutions are only valid after risk adjustment for patient acuity, comorbidities, and case complexity.
Raw (crude) outcome rates cannot be directly compared across institutions without risk adjustment. A hospital that treats more high-risk patients (older, multi-morbid, redo surgeries, emergency cases) will have higher crude mortality rates even if its surgical quality is excellent. Risk adjustment methods (logistic regression, STS risk scores, O/E ratio) account for expected mortality based on patient characteristics. The observed-to-expected ratio (O/E) is the correct metric for comparing institutional performance. Reporting an unadjusted comparison as a quality failure violates basic quantitative reasoning principles in healthcare quality management.
Question 6: A dietitian calculates that a patient needs 1800 kcal/day. The patient reports eating an average of 1200 kcal/day. Which quantitative statement BEST characterizes the deficit?
- The patient has a 600 kcal/day deficit (33% below requirements), which if sustained over one week creates a cumulative deficit of 4200 kcal (Correct answer)
- The patient is eating 1200 kcal/day, which is within the healthy range for adults
- The 600 kcal deficit will cause immediate clinical deterioration
- The deficit equals approximately 30 grams of fat per day
Correct answer: The patient has a 600 kcal/day deficit (33% below requirements), which if sustained over one week creates a cumulative deficit of 4200 kcal
600 kcal/day deficit is 33% below calculated needs; cumulating this over 7 days (4200 kcal) contextualizes the clinical significance of the shortfall.
Quantitative clinical reasoning involves calculating not just the instantaneous deficit but its clinical trajectory. 600 kcal/day × 7 days = 4200 kcal weekly deficit. Since approximately 3500 kcal corresponds to about 0.5 kg of adipose tissue, this creates potential weight loss of about 0.5–0.6 kg/week. 1200 kcal/day is below average requirements for most adults and is NOT automatically in a healthy range — adequacy depends on the individual's calculated needs. The deficit of approximately 33% represents a significant nutritional shortfall for a patient who already has calculated needs. Stating this quantitatively is essential for clinical communication.
HSRT Quantitative Clinical Reasoning 2
A drug reduces systolic BP by 8 mmHg on average (95% CI: 2–14 mmHg, p=0.01).
A colleague says: 'The effect is statistically proven and clinically definitive.' Which quantitative critique is MOST appropriate?