Biostatistics & Population Health

Risk visualization tools for shared decision-making

Clinical Overview and When to Use Risk Visualization Tools

— Preference-sensitive decisions where the "right" answer depends on patient values: PSA screening, mammography in 40–49 or ≥75, lung cancer screening with LDCT, statin initiation for primary prevention, anticoagulation in low-CHA₂DS₂-VASc atrial fibrillation, prophylactic mastectomy in BRCA carriers.

— Close trade-offs: elective AAA repair at 5.0–5.4 cm, watchful waiting vs surgery for localized prostate cancer, knee replacement timing.

— Polypharmacy/deprescribing in geriatric patients weighing fall risk vs cardiovascular benefit.

Definition: Risk visualization tools are graphical aids (icon arrays, bar graphs, pictographs, decision boxes, Kaplan-Meier curves simplified for patients) that translate probabilistic data into formats patients can interpret during shared decision-making (SDM).

Core purpose on Step 3: Help patients integrate absolute risk, absolute risk reduction (ARR), number needed to treat (NNT), and number needed to harm (NNH) when choosing between options of comparable medical reasonableness.

When to deploy in ambulatory practice:

Evidence base: Cochrane reviews of patient decision aids show improved knowledge, more accurate risk perception, reduced decisional conflict, and greater concordance between values and chosen option—without increasing anxiety or worsening outcomes.

USPSTF and ACC/AHA explicitly endorse SDM for several Grade B/C recommendations (statins for 10-year ASCVD 5–7.5%, lung cancer LDCT, aspirin primary prevention in 40–59-year-olds).

Step 3 management: When a vignette features a patient asking "Should I take this medication?" for a borderline indication, the correct next step is rarely "start the drug" or "do nothing"—it is engage in shared decision-making using a validated decision aid, then document the discussion.

Board pearl: SDM is itself a billable, quality-measured activity (CMS requires documented SDM visit for LDCT lung cancer screening and for ICD implantation in primary prevention).

Presentation Patterns and Key History — Numeracy and the Patient Encounter

— Asking "Is that a lot?" after a percentage is quoted

— Confusing relative and absolute risk ("cuts my risk in half")

— Difficulty with denominators ("1 out of how many?")

— Preferring stories/anecdotes over statistics

— Positive framing ("95% survive") encourages intervention uptake

— Negative framing ("5% die") discourages it

— Relative risk reduction (RRR) sounds larger and biases toward treatment vs the same data shown as ARR or NNT

— Loss-framed messages outperform gain-framed for screening behaviors; opposite for prevention behaviors

— Patient's stated goals ("I want to live to see my grandchild graduate")

— Prior experience with the disease (family history of breast cancer changes perceived baseline risk)

— Numeracy comfort ("Are you a numbers person, or do pictures help?")

— Cultural/religious values affecting trade-off weighting

— Time horizon the patient cares about (5-year vs lifetime risk)

Recognizing low health numeracy: Roughly one-third of US adults cannot reliably interpret "1 in 1000" vs "1 in 100," and ~36% have only basic health literacy (NAAL data). Clues in the encounter:

Framing effects to anticipate:

Key history elements before choosing a visualization:

Ask-Tell-Ask technique: Ask what they already know → Tell tailored information with a visual → Ask what they took away (teach-back).

Key distinction: Informed consent is a legal/ethical floor (disclosure of material risks); shared decision-making is a higher collaborative process that incorporates patient values. A signed consent form does NOT equal documented SDM.

Board pearl: When the stem says a patient "doesn't understand the numbers," the answer is usually an icon array (pictograph)—the format most robust to low numeracy. Avoid quoting odds ratios or hazard ratios to patients; translate to natural frequencies out of 100 or 1000.

Physical Exam Findings — Assessing Decisional Capacity and Communication Readiness

— Understanding — patient can paraphrase the condition, options, risks, benefits

— Appreciation — applies information to their own situation ("This means I could have a stroke")

— Reasoning — weighs options in a logically consistent way

— Choice — expresses and maintains a stable preference

— Visual acuity / color vision: older patients may misread red/green pie charts; use high-contrast monochrome icon arrays

— Hearing: ensure quiet room, hearing aids in; written + visual reinforces verbal

— Mini-Cog or MoCA if cognitive impairment is suspected before relying on probability discussion

— Depression screen (PHQ-2): depression skews risk perception toward pessimism

— Sit at eye level, screen turned toward patient, family member present if patient wishes

— Use a professional medical interpreter (NOT family) for limited-English-proficiency patients—family interpreters distort risk language

Although risk communication is cognitive rather than physical, Step 3 expects a structured decisional assessment before relying on any visualization.

Four-element capacity evaluation (Appelbaum criteria):

Sensory and cognitive screen that affects visualization choice:

Communication environment:

Step 3 management: A patient with fluctuating capacity (e.g., delirium, early dementia) requires either deferring nonurgent decisions until lucid or invoking a surrogate using substituted judgment (what the patient would have chosen), then best interests.

Board pearl: A patient who refuses a recommended treatment is not automatically lacking capacity—capacity is decision-specific and threshold rises with risk of the choice. Document capacity assessment explicitly when a patient declines a high-benefit, low-risk intervention.

Key distinction: Competence is a legal determination by a court; capacity is a clinical determination physicians make at the bedside.

Diagnostic Workup — Choosing and Calibrating the Right Risk Calculator

— ASCVD Pooled Cohort Equations (PCE): 10-year atherosclerotic CVD risk, ages 40–75; PREVENT (2023, AHA) is the newer successor incorporating eGFR, BMI, social determinants, and ages 30–79

— CHA₂DS₂-VASc / HAS-BLED: stroke vs bleeding in atrial fibrillation

— FRAX: 10-year fracture risk for osteoporosis treatment thresholds

— Gail / Tyrer-Cuzick / BCSC: breast cancer 5-year and lifetime risk

— Lung cancer: PLCOm2012 or Tammemägi for LDCT eligibility refinement

— Mayo / NSQIP / Revised Cardiac Risk Index: perioperative risk

— MELD-Na, Child-Pugh: hepatic decision-making

— Use non-fasting lipids if that is what was drawn (acceptable per ACC/AHA)

— BP should be average of ≥2 readings, not a single elevated office value

— Smoking status = current vs former (former smokers ≥5 years out often default to nonsmoker in calculators)

Step 1: Pick the validated calculator for the population and outcome.

Step 2: Verify calibration in the patient's group—PCE overestimates risk in some Asian-American and higher-SES populations and may underestimate in South Asian patients; PREVENT recalibrates and removes race as input.

Step 3: Compute baseline risk, then the risk with each option (e.g., 10-yr ASCVD 12% untreated vs 9% on moderate-intensity statin → ARR 3%, NNT ~33 over 10 years).

Step 4: Translate to a natural-frequency visualization the patient can see.

Inputs to double-check:

CCS pearl: Order the calculator inputs (lipid panel, A1c, BP, eGFR, urine albumin/creatinine for PREVENT) before the SDM visit so results are ready; schedule a dedicated follow-up visit specifically for the SDM conversation rather than tacking it onto an acute visit.

Board pearl: A calculator gives a point estimate with uncertainty—communicate ranges ("about 10 to 14 out of 100") rather than false precision ("11.7%").

Diagnostic Workup — Visualization Formats and Their Evidence

— Use part-to-whole display so the unaffected majority is visible

— Random scatter of affected icons is more accurate than clustering (clustering inflates perceived risk)

— Relative risk reduction without absolute context

— Numbers needed to treat without an icon array (NNT alone is poorly understood)

— 1-in-X formats with varying denominators ("1 in 8 vs 1 in 24"—patients judge by the denominator size, not the actual risk)

Icon arrays (pictographs): 100- or 1000-person grids with affected individuals shaded. Best-evidence format for low-numeracy adults; reduces denominator neglect.

Bar graphs: Effective for comparing two options side-by-side; ensure y-axis starts at 0 to avoid exaggeration.

Pie charts: Acceptable for single proportions; poor for comparisons across options.

Survival/Kaplan-Meier curves (simplified): Useful for oncology and end-of-life discussions; show median survival AND the spread (10th–90th percentile) so patients understand variability.

Decision boxes / option grids: Tabular side-by-side comparison of options across attributes (efficacy, side effects, monitoring burden, cost). Endorsed by the Mayo Clinic SDM National Resource Center.

Cates plots / "smiley face" arrays: Color-coded icons distinguishing baseline events, events prevented by treatment, and harms—superior for showing NNT and NNH simultaneously.

Avoid:

Key distinction: A decision aid is a structured tool meeting IPDAS (International Patient Decision Aid Standards) criteria—evidence-based, balanced, updated, value-clarification exercise included. A handout pulled from a pharma website is NOT a decision aid.

Board pearl: When asked the "best format for a patient with limited numeracy," choose the icon array with natural frequencies out of 100, displaying both benefits and harms on the same visual.

Risk Stratification — Matching Visualization Intensity to Decision Stakes

— Use option grid + icon array in a 15–20 minute visit

— Document risk discussed, patient values, decision reached

— Use multi-visit SDM, dedicated decision aid (Ottawa, Mayo, Healthwise), often with a decision coach (nurse, pharmacist, social worker)

— Encourage patient to bring family/values worksheet to next visit

— Emergent (STEMI, sepsis): standard informed consent, abbreviated discussion—SDM not appropriate when delay causes harm

— Elective: full SDM is the standard of care

Tier 1 — Low-stakes, strong evidence, low preference sensitivity: routine vaccines, treating symptomatic strep pharyngitis. Brief verbal risk statement suffices; formal decision aid not required.

Tier 2 — Moderate-stakes, preference-sensitive: statin for borderline ASCVD risk, bisphosphonate for osteopenia with FRAX in gray zone, low-dose aspirin in 40–59-year-olds with ≥10% ASCVD risk, anticoagulation for CHA₂DS₂-VASc = 1 (men) or 2 (women).

Tier 3 — High-stakes, irreversible, deeply value-laden: prostate cancer treatment selection, prophylactic mastectomy in BRCA carriers, LVAD/transplant candidacy, dialysis initiation in elderly, elective AAA repair, ICD for primary prevention.

Stratify by decision urgency:

Stratify by patient activation (PAM score conceptually): highly activated patients want data and options; lower-activation patients may prefer physician guidance with a clear recommendation—still ethically valid if values are elicited.

Step 3 management: For ICD primary prevention and LDCT lung cancer screening, CMS requires documented SDM using a decision aid as a condition of coverage—failure to document risks denial of reimbursement.

Board pearl: "Patient autonomy" does NOT mean dumping all data and walking out; physician recommendation remains appropriate in SDM as long as patient values are elicited and the recommendation is framed as advice, not directive.

Pharmacotherapy — Communicating Drug Benefits and Harms Visually

— Baseline 10-yr ASCVD 10% → moderate-intensity statin reduces to ~7.5% → ARR 2.5%, NNT ~40 over 10 years

— Show 100-person icon array: 90 unaffected regardless, 7–8 have MI/stroke despite statin, 2–3 prevented by statin, plus NNH icons (myalgia ~1–5%, new diabetes ~0.2%/year, hemorrhagic stroke negligible at standard doses)

— CHA₂DS₂-VASc 2 → ~2.2%/year stroke; DOAC reduces ~65% → ARR ~1.5%/year

— HAS-BLED 2 → ~1.9%/year major bleeding on anticoagulation

— Visualize as two parallel icon arrays: strokes prevented vs bleeds caused, allowing patient to weight

— Adults 40–59 with ≥10% 10-yr ASCVD: small net benefit (Grade C, individualize via SDM)

— Adults ≥60: do not initiate (Grade D—bleeding outweighs CV benefit)

— Classic SDM use case requiring visualization of GI/intracranial bleed risk vs MI prevention

— FRAX 10-yr hip fracture ≥3% or major osteoporotic ≥20% → treat; visualize fractures prevented vs rare ONJ (1 per ~10,000–100,000 patient-years) and atypical femoral fracture (~1 per 1,000 over 8+ years)

Statins for primary prevention (paradigm example):

Anticoagulation in AF:

Aspirin primary prevention (USPSTF 2022):

Bisphosphonates for osteoporosis:

Communication script: "Out of 100 people like you, in 10 years, X would have an event without treatment and Y with treatment, so the medication helps about (X−Y) people. The other (100−X) wouldn't have an event either way, and Z would have a side effect."

Board pearl: When framing NNT, always pair with NNH in the same time horizon and denominator—mismatched denominators (NNT/5 yr vs NNH/year) are a common patient misinterpretation.

Step 3 management: Document the specific numbers shown, the patient's stated preference, and the chosen plan—this is the legal and quality-measure backbone of SDM.

Procedures and Screening Decisions — Visualizing Trade-Offs

— PSA (USPSTF Grade C, 55–69): Per 1000 men screened 13 years: ~1.3 prostate cancer deaths prevented, ~3 fewer metastases, but ~100–120 false positives, ~110 biopsies, ~5 men with overtreatment-induced incontinence, ~30 with new ED. Show as side-by-side icon array of benefits and harms per 1000.

— Mammography 40–49 (Grade B, 2024 update): Per 10,000 women screened biennially × 10 yr: ~3–8 breast cancer deaths prevented vs ~1500 false positives, ~200 biopsies, ~25 overdiagnoses.

— LDCT lung cancer (50–80, ≥20 pack-years, current or quit ≤15 yr): NLST showed 3 lung cancer deaths prevented per 1000 screened over 6.5 yr, balanced against ~365 false positives per 1000 and small radiation/biopsy harms. CMS-mandated SDM visit before first LDCT.

— Colorectal screening (45–75): Multiple modality choice (colonoscopy q10 yr, FIT annual, Cologuard q3 yr, CT colonography q5 yr)—classic option-grid scenario.

— Carotid endarterectomy asymptomatic 70–99% stenosis: ARR ~1%/year stroke; perioperative stroke/death ~2–3%—net benefit highly value-sensitive in elderly

— AAA repair at 5.0–5.4 cm: annual rupture risk 1–11%; operative mortality 1–5%—surveillance is reasonable

— Knee/hip arthroplasty: function vs revision/complication trade-off; pain and lifestyle weigh heavily

Cancer screening — paradigm SDM territory:

Elective surgery decisions:

CCS pearl: In the CCS interface, "discuss screening with patient" or "counsel on risks/benefits" is the correct action before ordering the screening test for preference-sensitive screens (PSA, LDCT, mammography in borderline ages).

Key distinction: Screening (asymptomatic) vs diagnostic testing (symptomatic)—SDM expectation is far higher for screening, where the patient is well and the intervention is optional.

Special Populations — Elderly and Renal/Hepatic Impairment

— Most screening benefits accrue 5–10+ years after the test; treatment harms are immediate

— Use ePrognosis (UCSF) for 4-, 5-, 9-, 10-yr mortality in community-dwelling and nursing home elders

— When life expectancy < lag time to screening benefit, deprescribe screening

— Mammography: stop at age 75 unless life expectancy >10 yr (USPSTF I; ACS suggests continue while LE ≥10 yr)

— Colonoscopy: individualize 76–85 (Grade C); not after 85 (Grade D)

— PSA: do not screen >70 (Grade D)

— Statin primary prevention >75: insufficient evidence; SDM driven by frailty, polypharmacy, falls

— PREVENT incorporates eGFR and UACR—superior to PCE in CKD

— DOAC dosing depends on CrCl (apixaban reduce to 2.5 mg BID if 2 of 3: age ≥80, weight ≤60 kg, Cr ≥1.5)

— Avoid dabigatran if CrCl <30; warfarin preferred in CrCl <15 or dialysis (though apixaban acceptable per recent data)

— Larger font, higher contrast, avoid time horizons exceeding life expectancy

— Frame in terms patient cares about: independence, function, time at home rather than 10-year mortality

— Family caregiver often co-decision-maker—invite them with patient consent

Life expectancy as the dominant input:

Specific age-based recalibrations:

Renal impairment alters risk calculators:

Hepatic impairment: Child-Pugh B/C contraindicates rivaroxaban and apixaban; warfarin difficult due to baseline coagulopathy.

Visualization adaptation for elderly:

Board pearl: An 82-year-old with multimorbidity asking about a screening colonoscopy → do not order; engage SDM to explain that 10-yr lag time exceeds her expected benefit window, and recommend against.

Step 3 management: "Time to benefit" framing is the single most useful concept in geriatric SDM—if patient's life expectancy < time to benefit of an intervention, harms dominate.

Special Populations — Pregnancy, Pediatrics, Cultural and Linguistic Considerations

— Prenatal genetic screening (cell-free DNA, quad screen, NT) is the archetypal SDM scenario—offer all eligible patients, do not impose

— Visualize detection rate and false-positive rate per condition (T21 cfDNA sensitivity ~99%, FPR ~0.1%; quad screen sensitivity ~80%, FPR 5%)

— VBAC vs repeat C-section: uterine rupture ~0.5–1% with one prior low-transverse incision; show absolute risks of maternal/fetal complications for both paths

— GDM screening, RhoGAM, vaccination in pregnancy all benefit from visualization

— Assent ≥7 years, consent by parent; adolescents (varies by state) may consent independently for STI, contraception, mental health, substance use

— Visualizations adapted for parents AND age-appropriate child versions

— HPV vaccine SDM: show cancers prevented per 100,000 vaccinated vs minor injection-site reactions

— Some cultures emphasize family-based decisions over individual autonomy—ask "Who do you want involved in this decision?"

— Avoid assuming low numeracy from limited English; use professional interpreter and translated decision aids (Mayo, Ottawa, AHRQ libraries available in Spanish and others)

— Religious considerations (Jehovah's Witness and transfusion, end-of-life wishes) must be elicited specifically and documented

Pregnancy:

Pediatrics:

Cultural humility:

Health literacy universal precautions: Assume every patient may have limited health literacy; use plain language, teach-back, visuals as default—not only when literacy is "flagged."

Key distinction: Cultural competence (knowledge of cultural patterns) vs cultural humility (lifelong self-reflection and patient-led inquiry)—boards favor humility framing.

Board pearl: When a pregnant patient with limited English requests cfDNA but is uncertain about implications, the next step is professional interpreter + translated visual decision aid, not deferring or proceeding without comprehension.

Complications and Adverse Outcomes of Poor Risk Communication

— Decisional conflict and regret: strongly associated with poor risk understanding and inadequate values clarification

— Overdiagnosis and overtreatment when patients overestimate benefits (e.g., PSA-driven prostatectomy for indolent disease)

— Undertreatment when patients overestimate harms (statin nonadherence after social-media-driven fear of myalgia/dementia, which is not evidence-based)

— Anxiety from false-positive screening mitigated by pre-test SDM

— Relative risk hyping: "Reduces stroke by 65%" → patient assumes massive personal benefit; truth may be 2.2% → 0.8% (ARR 1.4%)

— Survival vs mortality framing inconsistency: 5-yr survival inflation due to lead-time bias is a classic Step 3 trap—disease-specific mortality is the unbiased endpoint

— Denominator neglect: "1 in 5" with no denominator clarification

— Base-rate neglect: patient with low pre-test probability misinterprets a positive screen as confirming disease (PPV depends on prevalence)

— Litigation risk when material risk was not disclosed—informed consent failure is a common malpractice allegation

— Quality-measure failure (CMS denials for unscreened LDCT/ICD SDM)

— Disparities widen when decision aids are unavailable in patients' languages

— Lower adherence (decisions made without patient buy-in)

— Lower satisfaction, worse trust, more no-shows

— Hospital readmissions when discharge plans are not co-designed

Patient-level harms:

Specific miscommunication failures:

System-level harms:

Adverse outcomes of not engaging SDM:

Step 3 management: When a patient develops a complication after a screening test they didn't fully understand (e.g., post-prostate-biopsy sepsis after PSA-driven biopsy in a borderline candidate), the systems-level fix is to implement a pre-test decision aid—not blame the individual clinician.

Board pearl: Lead-time bias and length-time bias inflate apparent screening benefit; disease-specific mortality from RCTs is the gold standard outcome to visualize.

When to Escalate Care — Decision Coaching, Ethics, and Specialty Referral

— Nurse decision coach (now standard in many oncology, cardiology, kidney centers)

— Pharmacist-led SDM for statin, anticoagulation, diabetes intensification

— Genetic counselor for cancer-syndrome decisions (BRCA, Lynch)

— Palliative care for advance care planning, goals of care, treatment-limiting decisions

— Persistent disagreement between patient and family or among family members

— Question of decisional capacity that affects a high-stakes choice

— Requests for treatment the team considers nonbeneficial (potentially inappropriate treatment)

— Surrogate appears to act against patient's known wishes

— Urology before PSA-driven biopsy decisions, with shared review of risk

— Cardiothoracic surgery for borderline AAA, valve replacement decisions

— Electrophysiology for ICD candidacy with CMS-mandated SDM documented

— Code status discussion at admission—not at the time of arrest

— Goals-of-care conversations in serious illness (cancer, advanced heart failure, ESRD, COPD GOLD 4) should occur with time-limited trials offered when prognosis is uncertain

Decision coaching referral: When time-limited primary care visits cannot accommodate a complex SDM conversation, refer to:

Ethics consultation indications:

Specialty referral:

Inpatient SDM escalation:

CCS pearl: In CCS cases involving advanced serious illness, "discuss goals of care" or "palliative care consult" is often a correct action even alongside aggressive disease-directed therapy; the two are not mutually exclusive.

Step 3 management: When a surrogate decision-maker requests an intervention the patient previously refused in a written advance directive, the advance directive governs unless the surrogate can demonstrate the patient's wishes have changed; involve ethics consultation if conflict persists.

Board pearl: Patient autonomy includes the right to refuse—a capacitated patient may refuse life-saving treatment, and clinicians must respect this even if they disagree.

Key Differentials — Same-Category Communication Approaches

— Informed consent: Legal/ethical disclosure of material risks, benefits, alternatives, and option of no treatment; patient signs. Required for procedures and significant interventions.

— Informed refusal: Same disclosure standard when patient declines; document specifically what was refused and that risks were communicated.

— Shared decision-making: Bidirectional integration of evidence and patient values; appropriate for preference-sensitive decisions.

— Paternalism (now considered inappropriate as default): Clinician decides for patient without eliciting values. Only acceptable in emergencies with no surrogate available.

— SDM is for choosing between medically reasonable options

— MI is for behavior change when there is a clear better choice but patient ambivalence (smoking, alcohol, adherence)

— Often combined: SDM for treatment selection, MI for sustaining the chosen plan

— Goals-of-care precedes SDM in serious illness—establish what the patient is trying to achieve, then SDM selects among options consistent with those goals

— Decision aid meets IPDAS criteria, includes value-clarification exercise, presents balanced harms/benefits

— Handout is informational, not interactive, and typically does not elicit values

SDM vs informed consent vs informed refusal vs paternalism:

SDM vs motivational interviewing (MI):

SDM vs goals-of-care discussions:

Decision aid vs educational handout:

Key distinction: A clinician can be both evidence-based and patient-centered; "patient-centered" does not mean abandoning recommendations—it means framing recommendations in light of patient values and accepting refusal.

Board pearl: "Patient autonomy" and "physician beneficence" are complementary, not opposing—boards reject pure paternalism and pure consumerism alike.

Key Differentials — Cognitive Biases That Distort Risk Perception

— Counter with actual frequencies in icon arrays

— Counter by comparing risks side-by-side on the same denominator

— Lead with absolute numbers; mention relative only secondarily, if at all

— Present both frames ("95 of 100 do well, 5 of 100 have a complication") to mitigate

— Use consistent denominators across options (per 100 or per 1000)

— Lead-time bias in survival statistics

— Length-time bias favoring indolent cancers detected by screening

— Selection bias in observational comparisons of treated vs untreated patients

Availability heuristic: Recent or vivid events feel more probable. A patient whose neighbor died of pancreatic cancer overestimates personal pancreatic cancer risk; one whose relative had a "statin reaction" overestimates muscle/cognitive risks.

Affect heuristic: Cancer is feared disproportionately to its actual contribution to mortality; CVD is underweighted despite being the #1 killer.

Anchoring: First number presented becomes the reference. If a patient hears "50% relative reduction" first, subsequent absolute numbers feel smaller.

Framing effect: Gain vs loss framing changes choice for identical data.

Optimism bias: Patients judge themselves at lower risk than peers, reducing prevention uptake.

Pessimism bias: Patients with depression or recent diagnoses overestimate their risk, choosing more aggressive interventions than values warrant.

Ratio bias / denominator neglect: "1286 in 10,000" feels larger than "24.14%" though they're equivalent.

Default bias: Patients tend to accept the option presented as default. Ethically, this can be leveraged (organ donation opt-out) but in clinical SDM defaults should be neutral.

Statistical traps to teach patients:

Key distinction: Statistical significance ≠ clinical significance; a p-value of 0.001 with ARR of 0.1% may not justify a therapy with substantial side effects.

Board pearl: When a vignette emphasizes how a patient "feels" about a number rather than the number itself, recognize a cognitive bias question—the answer often involves reframing with absolute risks or icon arrays.

Secondary Prevention / Discharge Planning Through Shared Decision-Making

— Each pillar adds ~20–25% RRR; combined ~73% all-cause mortality reduction vs no therapy

— Pair with NNH (hyperkalemia for MRA, hypotension for ARNI, GU infections for SGLT2)

— Medication list with picture icons for each pill (especially for low-literacy/elderly)

— Teach-back for each medication's purpose, timing, and key adverse effects

— Pill organizer demonstration; MyChart/portal walkthrough

Discharge medications as an SDM opportunity: Post-MI, post-stroke, new HF, new AF—patients return home with 5–10 new medications. Adherence at 1 year for post-MI medications is ~50%; SDM at discharge improves it.

GDMT for HFrEF (four-pillar therapy): ARNI/ACEi/ARB + beta-blocker + MRA + SGLT2 inhibitor. Visualize mortality reduction per pillar added:

Post-MI: Aspirin + P2Y12 (12 mo) + high-intensity statin + beta-blocker (if EF reduced) + ACEi (if EF reduced or HTN/DM/CKD) + cardiac rehab referral

Stroke secondary prevention: Antiplatelet (or anticoagulation if cardioembolic) + statin (LDL goal <70) + BP control (<130/80) + smoking cessation + lifestyle

Visualization tools at discharge:

Cardiac rehab and pulmonary rehab: Class I indication post-MI, post-CABG, post-PCI, stable HFrEF (cardiac); COPD GOLD B-E (pulmonary). Visualize 30–40% mortality reduction with cardiac rehab—still under-referred (~25% of eligible patients).

Step 3 management: At discharge, document medication reconciliation, follow-up appointment within 7–14 days, who to call for symptoms, when to go to ED—the four pillars of safe transition.

CCS pearl: In a post-MI CCS case, order cardiac rehab referral before ending the case; it is a frequently missed Class I order with mortality benefit.

Board pearl: Smoking cessation post-event reduces mortality by ~36%—a larger effect than any single medication. Always elicit and document a quit plan with visualized timeline.

Follow-Up, Monitoring Parameters, and Counseling Cadence

— 2–4 weeks after initiating a new chronic medication (statin, antihypertensive, anticoagulant): adherence, side effects, refine plan

— 3 months for trended labs (lipid response, A1c, BP) and values check-in—is the patient still satisfied with the chosen path?

— Annually for re-evaluation; risk trajectories change, and decisions made years ago should be revisited

— Statin: lipid panel 4–12 weeks after initiation, then 3–12 months; check ALT only if symptomatic; CK only if symptoms; A1c at baseline for those at diabetes risk

— Anticoagulation: annual CrCl for DOACs (every 6 months if CrCl 30–60 or age ≥75), INR weekly→monthly for warfarin

— Bisphosphonates: DXA every 2 yr; drug holiday discussion at 5 yr (oral) or 3 yr (IV) for those at lower fracture risk

— Screening cadence: colonoscopy q10 yr if normal, mammography biennially 40–74, LDCT annually while eligible

— Tobacco: 5 A's (Ask, Advise, Assess, Assist, Arrange) every visit for current smokers

— Alcohol: AUDIT-C annually

— Diet/exercise: USPSTF recommends intensive behavioral counseling for adults with CV risk factors (Grade B)

— Mental health: PHQ-2/9 and GAD-7 at least annually

— Clinical status changes (new diagnosis, declining function)

— Patient values shift (new caregiver role, recent loss)

— Evidence base changes (new guidelines, drug class)

Post-SDM follow-up structure:

Monitoring by intervention:

Counseling elements at every visit:

Updating decisions: Re-engage SDM when:

Step 3 management: Document "plan revisited; patient continues to prefer current approach" during annual visits—this is the longitudinal arm of SDM that Step 3 frequently tests.

Board pearl: Health behaviors (diet, exercise, smoking, sleep) dwarf single-drug effects—visualize behavioral contributions alongside pharmacologic ones to give patients an accurate sense of their levers.

Ethical, Legal, and Patient Safety Considerations

— Therapeutic privilege (withholding information believed to harm patient) is rarely justifiable in modern US practice and not a defense against disclosure failure

— Waiver of consent (patient explicitly delegates to clinician or family) must itself be documented

— Emergency exception: life-threatening situation, no surrogate available—consent implied; document the emergency

— Healthcare power of attorney → spouse → adult children (majority) → parents → adult siblings → other relatives → close friend

— Use substituted judgment (what would the patient have wanted) before best interests

— Honor written advance directives; POLST is a portable medical order across care settings

— A new surrogate cannot override a previously expressed competent wish

— Suspected child or elder abuse (mandatory, regardless of patient wishes)

— Certain communicable diseases (TB, syphilis, HIV in some states)

— Gunshot/stab wounds, impaired drivers (state-dependent)

— These trump confidentiality; document and notify the patient when safe

— Medication reconciliation at every transition (admission, transfer, discharge)

— Closed-loop communication to PCP within 24–72 hr of discharge

— Pending test follow-up assigned to a named clinician—a major source of preventable harm

Informed consent edge cases:

Capacity vs competence: Capacity is clinical, decision-specific, and may fluctuate. Competence is a legal designation. A capacitated patient may refuse care even if life-saving.

Surrogate decision-making hierarchy (varies by state but typically):

Advance directives and POLST:

Mandatory reporting (Step 3 favorite):

Confidentiality in SDM: Decision aids and discussions must remain confidential under HIPAA; family involvement requires patient permission (except for surrogate decision-making for incapacitated patients)

Transition-of-care safety (Step 3 critical):

Step 3 management: When a patient with capacity refuses a recommended intervention, the answer is document informed refusal with specific risks discussed, offer alternatives, and maintain the therapeutic relationship—not discharge from practice.

Board pearl: Disclosure of medical errors is ethically and (in many states) legally required; apology laws in most US states protect expressions of empathy from being used as admission of liability.

High-Yield Associations and Rapid-Fire Clinical Facts

Icon array = best format for low-numeracy patients—use 100- or 1000-person grids with random scatter

ARR, NNT, NNH > RRR for patient communication; always paired with the same time horizon

CMS-mandated SDM with decision aid: LDCT lung cancer screening (first scan only) and primary-prevention ICD

USPSTF SDM-flagged decisions: PSA 55–69 (Grade C), aspirin primary prevention 40–59 (Grade C), tobacco-treatment intensity choice, BRCA testing referral criteria

Time-to-benefit framing: screening colonoscopy ~10 yr, mammography ~10 yr, statin primary prevention ~2–5 yr, bisphosphonate ~12–18 mo—match to life expectancy

PREVENT (AHA 2023) replaces Pooled Cohort Equations: ages 30–79, includes eGFR/UACR/BMI/social deprivation, removes race

FRAX thresholds for treatment (in absence of fragility fracture or T-score ≤−2.5): 10-yr hip ≥3% OR major osteoporotic ≥20%

CHA₂DS₂-VASc ≥2 (men) or ≥3 (women): anticoagulation indicated; =1 or =2 respectively: SDM-driven choice

Lead-time bias and length-time bias inflate apparent screening benefit—use disease-specific mortality from RCTs, not 5-yr survival

Number needed to screen (NNS) for mammography 50–74: ~1300 over 10 yr to prevent one breast cancer death; PSA: ~570 over 13 yr; LDCT: ~320 over 6.5 yr

Cardiac rehab: Class I post-MI, post-CABG, post-PCI, HFrEF; 30–40% mortality reduction; under-referred—always order on discharge

Smoking cessation post-CV event: ~36% mortality reduction—larger than statin or beta-blocker effect

Health literacy universal precautions: apply plain language and teach-back to every patient

Apology laws (most US states): expressions of empathy/regret after adverse events are inadmissible as liability admissions

Decision aid standards: IPDAS; major libraries: Ottawa Decision Aid Inventory, Mayo SDM tools, AHRQ Effective Health Care

Step 3 management: When stem mentions a "borderline" decision and asks for next step, default to shared decision-making using a decision aid

Board pearl: "Statistical significance ≠ clinical significance"—always evaluate ARR magnitude, not just p-value, when discussing benefit with patients.

Board Question Stem Patterns

— Correct answer: Engage in shared decision-making using an icon array showing ARR and NNT, considering risk enhancers (family history, hsCRP, Lp(a), CAC score)

— Distractors: "Start high-intensity statin," "No statin indicated," "Recheck lipids in 1 year"

— Correct: SDM visit using decision aid before ordering LDCT (CMS requirement); discuss benefits/harms

— Distractor: Order LDCT immediately

— Correct: Discuss anticoagulation risks/benefits with the patient (CHA₂DS₂-VASc 1 in men is the SDM gray zone)

— Distractor: Start DOAC reflexively; start aspirin (no longer recommended)

— Correct: Discuss that screening benefit accrues over ~10 years; individualize based on life expectancy and values

— Distractor: Continue annual mammograms; absolutely stop

— Correct: Explain that "in half" is relative reduction; absolute reduction over 10 years is from 4% to 2%—help them weigh personally

— Correct: Explore values, ensure understanding, document informed refusal, maintain relationship, offer alternatives

— Distractor: Pursue court order; involve ethics for forced treatment

— Correct: Follow the advance directive; involve ethics if family conflict persists

Stem 1 — Borderline ASCVD risk: 52-year-old with 10-yr ASCVD risk of 8.5%, LDL 130, no diabetes, BP 128/80. Asks whether to start a statin.

Stem 2 — Lung cancer screening: 62-year-old with 30 pack-year smoking history (quit 8 years ago) presents for annual visit.

Stem 3 — AF anticoagulation at CHA₂DS₂-VASc = 1: 58-year-old man with HTN, new paroxysmal AF.

Stem 4 — Elderly screening: 81-year-old healthy woman asks about continuing mammograms.

Stem 5 — Risk miscommunication: Patient says "the medication cuts my risk in half."

Stem 6 — Refusal of recommended care: Capacitated patient with stage I treatable cancer refuses surgery citing personal values.

Stem 7 — Surrogate disagreement with advance directive: Daughter requests intubation for father whose AD specifies no mechanical ventilation.

Step 3 management: Pattern recognition—if the stem contains "preference-sensitive," "borderline," "asks about risks and benefits," or describes a screening with both meaningful benefits and harms, the answer is almost always SDM with a validated visual decision aid.

Board pearl: Watch for distractors framed as "more testing" or "definitive intervention"—boards reward the SDM pathway over reflexive ordering.

One-Line Recap

Risk visualization tools turn abstract probabilities into patient-interpretable images, enabling shared decision-making that aligns evidence-based options with individual values—and on Step 3, "engage in SDM using a validated decision aid" is the right answer whenever a decision is preference-sensitive, the benefits and harms are close in magnitude, or guidelines explicitly call for individualization.

Format hierarchy: Icon array (100- or 1000-person pictograph) with natural frequencies beats RRR, beats raw NNT, beats verbal probability—especially for low-numeracy patients.

Communication scaffold: Absolute risk → Absolute risk reduction → NNT paired with NNH on the same time horizon and denominator → Patient values elicitation → Documented joint decision.

Mandatory SDM scenarios to memorize: LDCT lung cancer screening (CMS), primary-prevention ICD (CMS), PSA screening 55–69 (USPSTF C), aspirin primary prevention 40–59 (USPSTF C), borderline ASCVD risk for statin, AF anticoagulation at low CHA₂DS₂-VASc, elderly cancer screening continuation, advance care planning in serious illness.

Time-to-benefit + life-expectancy alignment is the geriatric SDM keystone—deprescribe screening and preventive medications when lag time exceeds remaining life expectancy.

Ethics floor: Informed consent is mandatory; SDM is the higher standard for preference-sensitive choices; capacity is decision-specific; capacitated refusal must be respected; advance directives govern when conflicts arise.

Documentation triad that satisfies Step 3 quality measures and medicolegal expectations: risks/benefits discussed with specific numbers, patient values elicited, decision and follow-up plan recorded.

Final board pearl: When in doubt between "order the test/drug" and "discuss with patient," the Step 3 answer is discuss with patient using a decision aid—then revisit at follow-up.