Eduovisual
Biostatistics & Population Health
Absolute vs relative risk communication framing
— Absolute risk reduction (ARR) = difference in event rates between groups (e.g., 2% → 1% = 1% ARR)
— Relative risk reduction (RRR) = proportional change (1%/2% = 50% RRR)
— Number needed to treat (NNT) = 1/ARR; the most patient-centered framing
— Same trial, same data — but RRR sounds ~10–50× more impressive than ARR when baseline risk is low
— Pharma detailing, direct-to-consumer ads, or guideline summaries quoting only RRR ("cuts events by 25%!") without baseline rates
— Patient brings a news article or social media post citing a percentage reduction
— Shared decision-making conversations about statins, anticoagulation, cancer screening, antihypertensives, SGLT2 inhibitors
— Quality metrics or P4P dashboards that highlight relative gains across small populations
— Conflict between a patient's stated risk tolerance and a "blanket" recommendation
— Risk communication is a tested outpatient, longitudinal competency under shared decision-making
— Misframing can violate informed consent (material risk disclosure) and drive overtreatment
— Step 3 stems frequently embed a 2×2 table or trial summary and ask which number is most appropriate to share with a patient
— Any time only a relative number is presented
— Low baseline risk populations (primary prevention)
— Elderly with competing mortality (lifetime benefit shrinks)
— Therapies with meaningful harm, cost, or pill burden
Board pearl: If a stem gives you a relative risk reduction without baseline event rates, that is itself the red flag — the "right answer" is usually to request or calculate the absolute risk reduction or NNT before counseling the patient.
— 55-year-old asks about starting a statin after reading "statins cut heart attacks by 30%"
— 72-year-old with AF weighing apixaban; daughter found a website citing "64% stroke reduction"
— 50-year-old woman deciding on screening mammography quoting "mortality reduced 20%"
— Patient with newly diagnosed DM2 reviewing SGLT2i "cuts CV death by 38%"
— Post-MI patient confused why "huge benefit" drugs feel underwhelming when explained in absolute terms
— Numeracy and health literacy — ask "Are you comfortable with percentages, or would you prefer numbers like 1 in 100?"
— Baseline risk drivers — age, sex, smoking, BP, lipids, family history, comorbidities (needed to personalize ARR via calculators like PCE, CHA₂DS₂-VASc, FRAX)
— Values and preferences — fear of specific outcome (stroke vs bleed, cancer vs overdiagnosis)
— Prior exposure to misleading framing — what numbers have they already heard, and from where
— Decisional conflict signs — "I keep going back and forth"
— Framing effect: same data, opposite decisions depending on gain vs loss wording
— Denominator neglect: patients fixate on numerator ("30% lower!") and ignore baseline
— Affect heuristic: dread of cancer or stroke inflates perceived benefit
— Availability bias: recent family event distorts baseline estimates
— Patients with lower numeracy benefit most from icon arrays, natural frequencies ("3 out of 1000"), and absolute terms
— Higher-numeracy patients still systematically overweight RRR — ARR framing remains preferred
Step 3 management: Before quoting any number, anchor on the patient's baseline absolute risk using a validated calculator, then express benefit as ARR and NNT over a defined time horizon (typically 5 or 10 years). This is the guideline-endorsed shared decision-making sequence.
— There is no physical exam — instead, assess the decisional and informational vitals of the encounter
— Step 3 frequently tests recognition that a numerical disclosure was inadequate or biased
— Patient repeats only the relative figure ("it cuts my risk in half") without baseline
— Patient cannot state the time horizon of the benefit (per year? over 10 years? lifetime?)
— Patient cannot articulate at least one harm in absolute terms (NNH)
— Decision driven by a single salient number rather than tradeoffs
— Use of vague qualifiers ("very effective," "major reduction") without quantification
— Comprehension: teach-back — "Can you tell me in your own words what this medication might do for you?"
— Calibration: does the patient's perceived risk match their calculated risk? Miscalibration in either direction needs correction
— Concordance: does the chosen option align with their stated values
— Confidence: decisional conflict scale clues — hesitation, regret language
— Progress notes citing RRR only ("patient counseled on 25% mortality benefit")
— Consent forms with relative numbers but no absolute risk or NNT
— Patient education handouts produced by industry without ARR
— Both benefits and harms quoted in the same metric (both ARR or both natural frequencies)
— Same denominator used throughout ("out of 1000" not switching to "out of 100")
— Same time frame for benefit and harm
Key distinction: A patient who says "50% less likely" is quoting RRR; a patient who says "1 fewer event per 100 people over 10 years" is quoting ARR. The latter signals an adequately informed consent conversation; the former signals counseling is incomplete.
— Step 3 stems often present a 2×2 table, a trial abstract, or a Kaplan–Meier curve and require quick computation
— Control event rate (CER) = events in control / total in control
— Experimental event rate (EER) = events in treated / total in treated
— ARR = CER − EER
— RRR = (CER − EER)/CER = 1 − RR
— Relative risk (RR) = EER/CER
— NNT = 1/ARR (round up to whole patients)
— NNH = 1/ARI (absolute risk increase of harm)
— Trial: 2% MI in placebo, 1% MI in drug over 5 years
— RR = 0.5; RRR = 50%; ARR = 1%; NNT = 100 over 5 years
— Same drug in secondary prevention (20% → 10%): RRR still 50%, but ARR = 10%, NNT = 10 — vastly more impactful
— Vertical gap between curves at a time point ≈ ARR at that horizon
— Hazard ratio is a relative measure — beware of impressive HRs in low-event-rate populations
— OR ≈ RR only when outcome is rare (<10%); for common outcomes ORs overstate the relative effect
— Case-control studies report ORs; cohort and RCT typically report RR or HR
— RR or HR CI crossing 1.0 = not statistically significant
— ARR CI crossing 0 = not statistically significant
— Wide CIs in small trials make impressive RRRs unreliable
Board pearl: When a question provides a 2×2 table, always compute ARR and NNT first — these are almost always the correct counseling answer even when RRR is also calculable. Step 3 rewards the patient-centered metric.
— Natural frequencies > percentages for patient comprehension: "3 out of 100" beats "3%"
— Icon arrays (pictographs) of 100 or 1000 figures with events shaded — strongest evidence base for accurate risk perception across literacy levels
— Pre/post comparison bars showing baseline risk and post-treatment risk side-by-side
— Avoid mixed metrics (e.g., benefit as RRR, harm as ARR) — this is a classic manipulation pattern
— Pooled Cohort Equations (PCE) for ASCVD 10-year risk → drives statin ARR
— CHA₂DS₂-VASc and HAS-BLED for anticoagulation net benefit
— FRAX for fracture risk → bisphosphonate ARR
— Gail / Tyrer-Cuzick for breast cancer chemoprevention
— Lung cancer (PLCOm2012, USPSTF criteria) for LDCT screening
— Always pair ARR with a horizon (1-year, 5-year, 10-year, lifetime)
— Lifetime NNT is misleading for elderly patients with competing mortality
— Restricted mean survival time and event-free years gained are increasingly used for life-expectancy framing
— A "30% RRR in MACE" may be driven by a soft component (revascularization) while hard endpoints (CV death) are unchanged
— Always disaggregate composites before counseling
— LDL reduction, A1c, BP change are surrogates — ARR must reference patient-important outcomes (MI, stroke, death, fracture)
Step 3 management: When personalizing benefit, calculate the patient's individual 10-year ASCVD (or analog) risk, apply the trial's RRR to derive patient-specific ARR and NNT, then present alongside patient-specific NNH for the major adverse effect — this is the AHA/ACC and USPSTF-endorsed approach.
— Same RRR yields very different ARR depending on baseline risk
— High baseline risk (secondary prevention, established disease): small NNT, treatment usually favored
— Intermediate risk: shared decision-making zone — ARR/NNH balance is key
— Low baseline risk (young, primary prevention): large NNT, often NNT > NNH — reconsider
— 10-year ASCVD risk 5% with 25% RRR → ARR 1.25%, NNT 80 over 10 years
— 10-year risk 20% with same 25% RRR → ARR 5%, NNT 20 — much stronger case
— USPSTF and AHA/ACC use ≥7.5–10% 10-year risk as the shared decision threshold partly for this reason
— Step 1: Identify the patient's baseline absolute risk (calculator or stem)
— Step 2: Identify the intervention's RRR from trial data
— Step 3: Compute ARR = baseline × RRR and NNT = 1/ARR
— Step 4: Identify NNH for major harm at same horizon
— Step 5: Present both in same units, same denominator, same time frame
— Step 6: Elicit values and reach a concordant decision
— Applying RCT RRRs to populations not represented in trials (elderly, multimorbid) — assume smaller absolute benefit and larger absolute harm
— Ignoring competing risks (e.g., prostate cancer screening in a man with 5-year life expectancy)
— Conflating eligibility ("meets criteria") with net benefit for that individual
Board pearl: NNT scales inversely with baseline risk; NNH does not. This is why preventive therapies look great in secondary prevention but become questionable in low-risk primary prevention — Step 3 loves stems where doing the math flips the recommendation.
— Per AHRQ, ACP, and Cochrane guidance, default to:
— Absolute risk (with and without treatment) in natural frequencies
— Same denominator throughout
— Defined time horizon
— Both benefits and harms in identical units
— "Out of 100 people like you taking a statin for 10 years: about 3 fewer will have a heart attack or stroke; about 97 will have the same outcome as without the drug; about 1–2 extra will develop diabetes; muscle aches occur in a few but usually reversible."
— Avoids RRR; ties benefit and harm to same denominator and horizon
— "Per 100 patients per year: stroke risk drops from about 4 to 1; major bleeding rises from about 1 to 2."
— Net benefit is obvious without quoting "64% RRR"
— Mismatched framing: benefit as RRR, harm as ARR ("50% lower stroke vs 1% bleeding")
— Cherry-picked endpoints: composite benefit, individual harm
— Survival framing vs mortality framing: "90% survive" feels safer than "10% die" — same data
— Gain vs loss framing: "saves lives" vs "prevents deaths"
— Low numeracy → icon arrays, "X out of 100"
— Visual learners → bar graphs of pre/post risk
— Analytical patients → NNT/NNH with CIs
— Always check comprehension with teach-back
Step 3 management: The best-answer counseling option on Step 3 is almost always the one that presents absolute numbers, in natural frequencies, with matched time horizons for benefit and harm — even if a relative-risk option sounds more "informative."
— CABG vs PCI, CEA vs medical therapy, AAA repair, cancer surgery, ICD implantation
— These decisions hinge on absolute survival or event-free survival gains, often modest
— Symptomatic 70–99% carotid stenosis: 2-year stroke risk ~26% medical vs ~9% surgical → ARR ~17%, NNT ~6
— Asymptomatic 60–99% stenosis: 5-year stroke risk ~11% medical vs ~5% surgical → ARR ~6%, NNT ~17 (and only with low perioperative complication rate <3%)
— Same RRR (~50%), wildly different clinical decisions
— EF ≤35%, NYHA II–III: ~7% ARR mortality over ~5 years → NNT ~14
— Patients often hear "reduces sudden death by 30%" but the absolute gain and procedural/shock burden tell the real story
— Mammography 50–74: ~20% RRR in breast cancer mortality but ARR ~0.1% per year → NNT ~1000 over 10 years to prevent one death; NNH for overdiagnosis ~3 per 1000
— PSA screening: ARR for prostate cancer mortality ~0.1%; NNH (overdiagnosis, biopsy harms) substantial
— Material risks in absolute terms
— Realistic benefit as ARR/NNT for this patient
— Alternatives, including no treatment, with their absolute risks
— Failure to do so is both an ethics and medicolegal liability
CCS pearl: When the CCS or written stem asks how to counsel a patient about a procedure, the highest-scoring action is typically "discuss risks, benefits, and alternatives in absolute terms" + document shared decision-making — not simply "obtain consent." The framing is the consent.
— Lifetime NNTs collapse when life expectancy is short
— A statin with 10-year NNT of 50 is meaningless for a patient with 3-year life expectancy
— Time-to-benefit concept: most preventive therapies require 2–5 years before ARR emerges (e.g., tight glycemic control ~8 years for microvascular benefit)
— Match time-to-benefit against estimated life expectancy
— Reframe ARR for continuing vs stopping: "Stopping this medication increases your 1-year risk by less than 1%, while the side effects you're experiencing are daily"
— Particularly relevant for statins, bisphosphonates, PPIs, antihypertensives in frail elders
— Trials often exclude CKD/cirrhosis → published RRRs may not apply
— Harms (bleeding on DOACs, AKI on SGLT2i, rhabdo on statins) are often higher in absolute terms
— NNH shrinks while NNT may grow → net benefit can invert
— Each added drug has its own NNT/NNH; cumulative pill burden and interactions are rarely framed numerically
— Consider composite NNT vs cumulative harm in patients on ≥5 medications
— Engage surrogate decision-makers with the same absolute framing
— Document capacity assessment when complex tradeoffs are involved
— Use simpler natural frequencies and visual aids
Board pearl: For an elderly or frail patient, the Step 3 best answer is often to estimate life expectancy, compare with time-to-benefit, and reframe ARR over the patient's realistic horizon — not to apply trial RRRs uncritically. Tools: ePrognosis, Lee Index, Charlson.
— Most therapies have observational data only; RRs come with wide CIs
— Frame baseline risk of malformation (~3% background) before adding teratogen risk
— Example: SSRI exposure and cardiac malformation — RR may sound alarming but ARR is <0.5%
— Counsel with both the absolute background risk and the absolute attributable risk
— Parents systematically overweight rare harms (e.g., vaccine adverse events) and underweight common preventable disease
— Use absolute risk of the disease vs absolute risk of the adverse event in matched denominators
— Example: MMR — measles encephalitis ~1/1000 cases; vaccine encephalitis ~1/1,000,000 doses
— Avoid percentages — use "X out of 100" or "X out of 1000"
— Use plain language: "chance" instead of "probability," "side effect" instead of "adverse event"
— Icon arrays improve comprehension across literacy levels
— Teach-back is mandatory — "What will you tell your family about this medicine?"
— Use certified medical interpreters (not family members) — required for informed consent of risk information
— Translate written ARR/NNT materials; avoid idioms ("hit or miss," "ballpark")
— Single-item screener: "How good are you with numbers?" — predicts comprehension
— Subjective Numeracy Scale for higher-stakes decisions
Key distinction: In pediatric vaccine counseling, parents quoted RRR of disease prevention are no more reassured than baseline; parents shown absolute risks of disease vs vaccine adverse events with matched denominators and icon arrays show meaningfully improved acceptance — this evidence base is testable on Step 3.
— Overtreatment — patients accept low-yield therapies based on inflated RRR perception
— Undertreatment — patients refuse high-yield therapies when harms are framed in absolute terms but benefits in relative terms
— Decisional regret — strongly correlated with framing mismatch and inadequate disclosure
— Nonadherence — patients who don't understand absolute benefit discontinue when side effects appear
— Cascade of care — screening framed only in RRR drives follow-up biopsies, imaging, and procedural harm
— Overdiagnosis (especially screening — breast, prostate, thyroid)
— Medicalization of low-risk states (prehypertension, prediabetes)
— Resource misallocation toward low-NNT interventions in low-baseline-risk groups
— Inflated public expectations from media reports of RRR
— Failure to disclose absolute risk has been cited in informed-consent litigation
— Documentation of only "discussed risks and benefits" without specifics is inadequate defense
— Consent forms quoting only relative terms are increasingly challenged
— Statins: patients told "30% reduction" but not "1% ARR over 5 years in primary prevention" often discontinue when myalgia appears
— DOACs: patients given relative stroke reduction without absolute bleed risk later present after intracranial hemorrhage with regret/litigation
— Aspirin primary prevention: USPSTF reversed recommendation as ARR shrank in modern era while ARI for bleeding remained — illustrates how guidelines themselves evolve with reframing
— Anxiety, fatalism, and decisional paralysis from poorly framed risk
Step 3 management: When a patient presents with medication nonadherence, ask whether they understood the absolute benefit of therapy — re-counseling with ARR/NNT and matched NNH often resolves adherence issues better than reinforcement of the diagnosis.
— High-stakes irreversible decisions: cancer treatment, transplant, ICD, end-of-life
— Significant decisional conflict despite adequate counseling
— Disagreement between patient values and family/surrogate
— Capacity questions during complex risk discussions
— Cross-cultural or low-literacy scenarios beyond office tools
— Decision aids — Ottawa Decision Aid library, MagicApp, Mayo Choice tools; evidence-based, often required by CMS for certain decisions (e.g., LDCT lung cancer screening)
— Shared decision-making visits — billable under CMS for lung cancer screening (G0296) and ICD primary prevention
— Patient navigators for cancer and chronic disease pathways
— Genetic counselors for inherited risk (BRCA, Lynch) — absolute lifetime risks need expert framing
— Palliative care for goals-of-care conversations where time horizons reshape ARR
— Ethics consult for capacity, surrogate disagreement, or refusal of high-NNT-benefit therapy
— LDCT for lung cancer screening in eligible smokers (a documented SDM visit with decision aid is required for reimbursement)
— Primary prevention ICD (CMS NCD requires SDM encounter)
— Left atrial appendage occlusion (Watchman)
— These are favorite Step 3 stems
— Specific numbers discussed (ARR, NNT, NNH)
— Patient's expressed values
— Decision aid used
— Teach-back completed
— Final decision and rationale
CCS pearl: For LDCT lung cancer screening in a 55-year-old with 30 pack-years, the required CCS action sequence is "counsel on smoking cessation → conduct shared decision-making visit with decision aid → order LDCT" — skipping the SDM visit forfeits both reimbursement and the highest-scoring counseling step.
— Relative risk (RR) vs odds ratio (OR): OR overstates effect when outcome common; RR preferred for cohort/RCT
— Hazard ratio (HR): instantaneous relative risk over time; like RR, a relative metric — same framing concerns
— Risk ratio vs rate ratio: ratio of cumulative incidence vs ratio of incidence rates (person-time) — interpretation differs
— ARR (absolute risk reduction) — for benefit
— ARI (absolute risk increase) — for harm
— NNT (number needed to treat) = 1/ARR
— NNH (number needed to harm) = 1/ARI
— NNS (number needed to screen) — to prevent one event via screening
— Attributable risk — same concept as ARR in observational language
— Population attributable risk (PAR) — fraction of disease in population attributable to exposure
— Attributable risk vs relative risk — different scales, different uses
— Attributable risk percent vs population attributable risk percent — individual exposed vs whole population
— Sensitivity/specificity vs PPV/NPV — test performance affected by prevalence (analogous to baseline risk affecting ARR)
— Pre-test probability + LR → post-test probability is the diagnostic mirror of baseline risk + RRR → post-treatment risk
— Both demand absolute (post-test, post-treatment) framing for patients
— Composite endpoints inflate event rates and apparent benefit
— Always disaggregate
Key distinction: RR, OR, and HR are all relative metrics and share the same framing pitfall — none should be presented to patients without conversion to absolute terms at the patient's baseline risk. The choice among them is a study-design question; the framing fix is universal.
— Shared decision-making (SDM) — process; framing is one tool within it
— Informed consent — legal/ethical doctrine requiring disclosure of material risks; framing affects whether disclosure is meaningful
— Health literacy — broader capacity; numeracy is a subset
— Motivational interviewing — behavior change technique; complementary but distinct
— Therapeutic alliance — relational; affects but doesn't replace numeric communication
— Anchoring — overreliance on first number heard (often the RRR)
— Framing effect — gain vs loss wording shifts choice
— Default bias — opt-in vs opt-out structures (organ donation)
— Status quo bias — preferring inaction
— Optimism bias — patients underestimate personal risk
— Step 3 may name a specific bias and ask for the corrective communication technique
— Nudges — choice architecture; ethically requires transparent framing
— Pre-commitment — patient agrees in advance to thresholds
— Process measures vs outcome measures — analog to surrogate vs patient-important outcomes
— Performance dashboards quoting relative improvements can mislead clinicians the same way patients are misled
— Public health messaging uses population attributable risk and PAR%
— Clinical counseling uses individual ARR/NNT
— Conflating these is a common Step 3 distractor (e.g., applying a population statistic to one patient)
Board pearl: If a Step 3 question describes a patient making a decision after hearing only a percentage reduction, the named cognitive phenomenon is the "framing effect" and the corrective action is to re-present the data in absolute, natural-frequency terms with matched harms — recognize both the bias and its antidote.
— Risk evolves with age, new diagnoses, new evidence — counseling must be revisited
— Step 3 emphasizes chronic disease management visits as natural reframing points
— New cardiovascular event → primary prevention math becomes secondary prevention math; ARR rises, NNT falls — therapy strengthens
— New diagnosis (DM, CKD) shifts baseline risk upward
— Aging into higher-risk strata
— Guideline updates (e.g., aspirin primary prevention deprecated 2022)
— Major bleeding, fall, or adverse event → reassess NNH
— Life expectancy change (cancer diagnosis, frailty) — time-to-benefit reconsidered
— Post-MI: "Now that you've had a heart attack, the benefit of these medications is much larger — out of 100 patients like you, taking a statin for 5 years prevents about 5 more deaths or repeat heart attacks (NNT ~20)."
— Post-stroke on anticoagulant: re-emphasize absolute stroke reduction vs absolute bleed risk annually
— Medications reconciled with purpose and absolute benefit stated for each
— Red flags / when to return — in absolute terms ("any chest pain lasting more than a few minutes")
— Follow-up cadence with rationale
— Written summary including ARR/NNT for major preventive therapies where feasible
— Discharge summaries that quantify benefit aid primary care continuity
— EMR templates increasingly embed ARR/NNT calculators
Step 3 management: At every annual wellness visit and post-event follow-up, recompute the patient's absolute risk and reframe ongoing therapies in ARR/NNT terms — this is what differentiates Step 3-level longitudinal management from a Step 2-level snapshot recommendation.
— Decisional regret scale at follow-up after major decisions (cancer treatment, ICD, surgery)
— Adherence patterns — drops often signal benefit was poorly understood
— Patient activation measure (PAM) — engagement and self-management
— Teach-back at each visit for high-stakes therapies
— Acute decisions (procedure consent): pre-procedure framing, immediate teach-back, post-procedure debrief
— Chronic preventive therapy (statins, antihypertensives): reframe at initiation, 3-month adherence check, annually thereafter
— Screening decisions (mammography, LDCT, colonoscopy): reframe at each eligibility window, recalculate with each new risk factor
— Anticoagulation: annual review of CHA₂DS₂-VASc, HAS-BLED, falls, renal function — re-derive ARR/ARI
— Specific numbers discussed
— Decision aid used and version
— Patient comprehension verified
— Values elicited
— Disagreements or deferrals
— Cardiac rehab, pulmonary rehab, diabetes education programs embed risk education; refer early
— Behavioral counseling (USPSTF Grade A/B services): tobacco cessation, weight, alcohol — all benefit from absolute risk framing
— Pre-visit planning with risk calculator outputs
— Patient portal access to personalized risk summaries
— Decision aids linked to EMR
— Quality measures tied to documented SDM, not just service delivery
CCS pearl: Schedule a dedicated follow-up visit within 2–4 weeks after starting a major preventive therapy or making a major screening decision — to verify comprehension, address early side effects, and re-anchor the absolute benefit. This is a high-scoring CCS action even when not explicitly prompted.
— Legal doctrine requires disclosure of material risks — those a reasonable patient would want to know
— Modern jurisprudence increasingly recognizes that relative-only framing is inadequate disclosure
— Consent for procedures and high-risk medications should document specific absolute risks of benefit and harm at relevant time horizons
— Use of standardized decision aids strengthens defensibility
— Mis-framed risk → nonadherence → preventable events (e.g., AF stroke after DOAC discontinuation due to misunderstood bleed risk)
— Mis-framed harm → preventable adverse events (e.g., unrecognized rare but severe drug reaction)
— Transitions of care are particularly hazardous — discharge instructions that quote only relative benefits leave outpatient clinicians to reconstruct rationale
— Autonomy — requires meaningful comprehension, not just signature
— Beneficence/nonmaleficence — balance demands accurate ARR/NNH disclosure
— Justice — low-literacy and LEP patients disproportionately harmed by poor framing; equity demands tailored communication
— Truthfulness — RRR-only framing, while technically accurate, can be ethically deceptive
— Pharma marketing emphasizes RRR; clinicians should independently translate to ARR/NNT before adopting
— Disclose conflicts when relevant to recommendations
— Public health communication during outbreaks (COVID-19 vaccine ARR vs RRR was a high-profile real-world example)
— Genetic risk disclosure to relatives — absolute lifetime risks
— Patient discharged on warfarin after PE; clinic note reads "counseled on bleeding risk." Patient later has GI bleed and states they understood the medication was "very safe." This is a transition-of-care framing failure — the standard requires absolute, quantified, documented disclosure.
Step 3 management: Document risk discussions with specific numbers, time horizons, decision aids used, and teach-back confirmation — this satisfies informed consent, supports continuity, and is the highest-yield documentation answer on Step 3 ethics questions.
— NNT = 1/ARR (round up); NNH = 1/ARI
— RR = 1 − RRR; ARR = baseline risk × RRR
— OR ≈ RR when outcome <10%; OR overstates effect when outcome common
— HR ≈ RR for short follow-up with constant hazards
— Statin secondary prevention: NNT ~20–40 over 5 years for major CV events
— Statin primary prevention (low risk): NNT ~100–200 over 10 years
— Aspirin primary prevention: NNT ≈ NNH in modern populations — no net benefit (USPSTF 2022)
— Apixaban for AF (CHA₂DS₂-VASc ≥2): NNT ~50/year for stroke; NNH ~100/year for major bleed
— Mammography 50–74: NNS ~1000 over 10 years to prevent one breast cancer death
— LDCT lung cancer screening: NNS ~300 over 3 years (NLST)
— Colonoscopy screening: NNS ~150–700 to prevent one CRC death
— Influenza vaccine (healthy adults): NNT ~70 to prevent one case
— RRR without baseline rate
— Different denominators for benefit and harm
— Different time horizons for benefit and harm
— Composite endpoint benefit, single-component harm
— Survival framing in one direction, mortality in the other
— "Doubles your risk" without absolute numbers
— Icon arrays > bar graphs > percentages > probabilities
— Natural frequencies (X in 100) > decimal probabilities (0.01)
— Matched denominators essential
— Most physicians also overestimate effect sizes when given RRR alone
— USPSTF, AHA/ACC, ADA, NCI, AHRQ all endorse absolute risk and decision aids
Board pearl: When in doubt on a Step 3 risk communication item, pick the choice that includes absolute numbers, natural frequencies, matched denominators, defined time horizon, and a decision aid or teach-back — at least one of those features almost always distinguishes the correct answer.
— "A 52-year-old man read that statins reduce heart attacks by 30%. He has 10-year ASCVD risk of 6%. He asks what this means for him."
— Best answer: compute ARR (~1.8%) and NNT (~55) and present in absolute terms; offer decision aid
— Distractors: agree with patient's framing; quote only RRR; recommend statin without SDM
— Trial: 80/1000 events placebo vs 60/1000 drug
— RR = 0.75; RRR = 25%; ARR = 2%; NNT = 50
— Question asks "Which best informs the patient?" → ARR/NNT
— 55-year-old with 30 pack-years asks about lung cancer screening
— Best CCS sequence: smoking cessation counseling → SDM visit with decision aid (CMS-required) → LDCT
— Distractors: order LDCT directly; defer screening; chest X-ray
— Parent refuses MMR citing "1 in 1000 autism risk" from social media
— Best answer: present absolute risks of measles complications vs absolute vaccine adverse event rates with matched denominators, ideally icon array; address values
— 85-year-old on statin for primary prevention started at age 60, now frail
— Best answer: reassess life expectancy vs time-to-benefit; reframe ARR; consider deprescribing with SDM
— Drug "reduces MACE 30%" but breakdown shows benefit driven by revascularization
— Best answer: disaggregate and counsel on hard endpoints
— Pharma rep quotes RRR for benefit, ARR for harm
— Best answer: recompute both in same metric
Key distinction: Step 2 stems often stop at "what's the diagnosis" or "first-line drug." Step 3 stems on this topic ask "how do you counsel the patient" — and the right answer is procedural: absolute terms, natural frequencies, decision aid, teach-back, documented SDM.
Always translate relative risk reduction into the patient's individualized absolute risk reduction and number needed to treat — and present benefits and harms in the same natural-frequency denominator and time horizon — because identical data framed differently produces different patient decisions, and Step 3 rewards the patient-centered, absolute, shared-decision-making framing.
— ARR = baseline risk × RRR; NNT = 1/ARR; NNH = 1/ARI
— Same RRR yields tiny ARR in low-risk populations and large ARR in high-risk populations — baseline risk is everything
— OR overstates RR when outcomes common; HR is still a relative metric
— Natural frequencies ("3 in 100") beat percentages
— Matched denominators and time horizons for benefit and harm
— Icon arrays and validated decision aids for high-stakes or low-literacy decisions
— Teach-back to verify comprehension; document specific numbers
— Elderly: match time-to-benefit against life expectancy before recommending preventive therapy
— Pediatrics/vaccines: absolute disease risk vs absolute vaccine adverse-event risk in matched denominators
— Pregnancy: anchor on baseline 3% malformation risk before adding attributable risk
— CMS-mandated SDM visits: LDCT lung screening, primary prevention ICD, LAA occlusion
— Relative-only framing can constitute inadequate informed consent
— Transitions of care are the highest-risk moments for framing failure — quantify benefits and harms in discharge documentation, and reframe at every longitudinal visit as baseline risk evolves
Board pearl: When the answer choices include one option with absolute numbers, matched denominators, a defined time horizon, and a decision aid or teach-back, pick it — this is the Step 3 signature of correct risk communication.