Journal club biostatistics teaching — CASC/communication station · Psych CASC / OSCE

Prompt

You are the psychiatry registrar running a 10-minute journal club with a foundation doctor. You must teach calculation of NNT from response rates, correct a p-value misinterpretation, and explain why a non-significant HR CI that includes 1 does not prove equivalence.

Station brief

Format. Teaching/communication station, approximately 8–10 minutes. Facilitate understanding, not a monologue. [2]

Candidate instructions. Using the abstract numbers provided, walk the foundation doctor through ARR and NNT, correct a wrong definition of p, and interpret an HR with CI crossing 1. Use plain language and a whiteboard structure. [1][4]

Candidate scenario

Abstract summary (station material): RCT of SSRI versus placebo, N=200, 12 weeks, response 46% versus 30%, p=0.02. Authors also report time-to-discontinuation HR 0.85 (95% CI 0.60 to 1.20). Foundation doctor says: "p=0.02 means there is only a 2% chance the drug does not work, and the HR shows they are equivalent on staying in treatment." [2][3]

Marking domains

Correct ARR = 16% and NNT ≈ 7 with outcome and 12-week horizon stated [1]
Corrects p-value definition without shaming the learner [2]
Explains CI including 1 for HR: not conventionally significant; not proof of equivalence [3][5]
Prefers CI/estimation language over star-chasing [4]
Collaborative tone, checks understanding, time management
Avoids jargon pile-on; offers one take-home bottom line

Reveal assessor key

Open. "Let's write three numbers: rates, absolute difference, and what p actually means." Board: 46% vs 30% → difference 16% → NNT about 7 for one extra responder in 12 weeks. [1]

p-value correction. "If there were no real difference, data this extreme would appear about 2% of the time under our model — not '2% chance the drug is useless'." [2]

HR. "0.85 looks a bit better, but the CI from 0.60 to 1.20 still includes no difference. That is inconclusive, not equivalence — equivalence needs a designed margin and adequate precision." [3][5]

Close. Bottom line: absolute response benefit is modest and real by conventional testing; discontinuation signal is too imprecise to call equal. Offer to pull the full paper's CI for response next week. Thank the doctor. [1][4]

References

[1]Cook RJ, Sackett DL The number needed to treat: a clinically useful measure of treatment effect BMJ, 1995.PMID 7873954
[2]Greenland S, Senn SJ, Rothman KJ, et al. Statistical tests, P values, confidence intervals, and power: a guide to misinterpretations Eur J Epidemiol, 2016.PMID 27209009
[3]Altman DG, Bland JM Absence of evidence is not evidence of absence BMJ, 1995.PMID 7647644
[4]Gardner MJ, Altman DG Confidence intervals rather than P values: estimation rather than hypothesis testing Br Med J (Clin Res Ed), 1986.PMID 3082422
[5]Spruance SL, Reid JE, Grace M, Samore M Hazard ratio in clinical trials Antimicrob Agents Chemother, 2004.PMID 15273082

Station brief

Format. Teaching/communication station, approximately 8–10 minutes. Facilitate understanding, not a monologue. [2]

Candidate scenario

Marking domains

Correct ARR = 16% and NNT ≈ 7 with outcome and 12-week horizon stated [1]

Corrects p-value definition without shaming the learner [2]

Explains CI including 1 for HR: not conventionally significant; not proof of equivalence [3][5]

Prefers CI/estimation language over star-chasing [4]

Collaborative tone, checks understanding, time management

Avoids jargon pile-on; offers one take-home bottom line

Reveal assessor key

Open. "Let's write three numbers: rates, absolute difference, and what p actually means." Board: 46% vs 30% → difference 16% → NNT about 7 for one extra responder in 12 weeks. [1]

p-value correction. "If there were no real difference, data this extreme would appear about 2% of the time under our model — not '2% chance the drug is useless'." [2]

References

[1]Cook RJ, Sackett DL The number needed to treat: a clinically useful measure of treatment effect BMJ, 1995.PMID 7873954
[2]Greenland S, Senn SJ, Rothman KJ, et al. Statistical tests, P values, confidence intervals, and power: a guide to misinterpretations Eur J Epidemiol, 2016.PMID 27209009
[3]Altman DG, Bland JM Absence of evidence is not evidence of absence BMJ, 1995.PMID 7647644
[4]Gardner MJ, Altman DG Confidence intervals rather than P values: estimation rather than hypothesis testing Br Med J (Clin Res Ed), 1986.PMID 3082422
[5]Spruance SL, Reid JE, Grace M, Samore M Hazard ratio in clinical trials Antimicrob Agents Chemother, 2004.PMID 15273082