Eduovisual
Biostatistics & Population Health
Diagnostic vs screening test design criteria
— Examples: mammography, colonoscopy/FIT, low-dose chest CT for lung cancer, HbA1c for diabetes, lipid panel, BP screening, HIV/HCV one-time adult screen.
— Examples: colonoscopy after positive FIT, breast biopsy after suspicious mammogram, CT-PA for suspected PE, troponin in chest pain.
— A stem orders a "screening" CT-PA in a tachycardic, dyspneic patient (that's diagnostic).
— A stem uses a highly specific confirmatory test (e.g., Western blot historically) as first-pass screen.
— A stem screens for a disease lacking effective treatment or with no mortality benefit (e.g., screening healthy adults for pancreatic cancer — USPSTF Grade D).
— Screening favors high sensitivity (minimize false negatives), low cost, safe, acceptable, applied to a defined target population.
— Diagnostic favors high specificity and PPV (minimize false positives that trigger invasive workup), often costlier or more invasive.
Board pearl: If the question gives an asymptomatic, average-risk adult and asks "next step," you are in screening logic — apply USPSTF age/interval rules, not diagnostic thresholds.
— Cue words: asymptomatic, routine, well visit, preventive care, health maintenance, no complaints.
— The test of choice depends entirely on age, sex, risk factors, and USPSTF grade (A or B = offer).
— Family history of CRC in a first-degree relative <60 → colonoscopy at age 40 or 10 years before the affected relative's age.
— 20+ pack-year smoker, age 50–80, quit <15 years ago → annual low-dose chest CT.
— Men/women with abdominal aortic risk (men 65–75 ever-smokers) → one-time AAA ultrasound.
— BRCA, Lynch, or familial syndromes → earlier, more frequent, sometimes MRI-augmented surveillance.
— Any symptom referable to the organ (hematochezia, breast lump, hemoptysis, dysphagia).
— Abnormal prior screen (positive FIT, BI-RADS 4 mammogram, AUDIT-C ≥4 with concern).
— Known precursor lesion (Barrett esophagus, adenomatous polyp) → now surveillance, a third category with its own intervals.
Key distinction: A symptomatic patient never undergoes "screening." Re-label the encounter as diagnostic and pursue the workup appropriate to the complaint, even if a USPSTF screen happens to be due.
— Confirm absence of red-flag findings that would shift the encounter to diagnostic.
— Capture incidental findings (thyroid nodule, skin lesion) that themselves trigger diagnostic pathways.
— Document BP, BMI, and waist circumference, which are validated screens.
— Low pretest probability + you need to rule out → choose a sensitive test (SnNout: a Sensitive test, when Negative, rules Out).
— High pretest probability + you need to rule in → choose a specific test (SpPin: a Specific test, when Positive, rules In).
— Same patient asymptomatic on routine visit: no D-dimer, no CT — neither test is a screen for PE.
— Post-test odds = pretest odds × LR.
— LR+ >10 or LR− <0.1 are large effects; near 1 is uninformative.
— A "great" test in a low-prevalence population still yields low PPV — the screening trap.
Board pearl: When a stem provides a Wells, PERC, HEART, or CURB-65 score, it is forcing you to choose between a sensitive screening/rule-out tool and a specific confirmatory test. Match test characteristic to clinical task.
— Sensitivity = TP / (TP+FN) — among diseased, fraction detected. Property of the test, independent of prevalence.
— Specificity = TN / (TN+FP) — among non-diseased, fraction correctly cleared.
— PPV = TP / (TP+FP) — among test-positives, fraction truly diseased. Depends on prevalence.
— NPV = TN / (TN+FN) — among test-negatives, fraction truly non-diseased. Depends on prevalence.
— Maximize sensitivity so few cases are missed; accept some false positives that will be sorted by confirmatory testing.
— In low-prevalence populations (screening), even highly specific tests produce many false positives, so PPV is low — anticipate confirmatory follow-up.
— Maximize specificity and PPV; a false positive triggers biopsy, anticoagulation, surgery, or labeling.
— Often applied in higher-prevalence (enriched) populations, so PPV rises naturally.
— LR+ = sensitivity / (1 − specificity).
— LR− = (1 − sensitivity) / specificity.
— Use the Fagan nomogram conceptually: combine pretest probability with LR to get post-test probability.
— Moving the cutoff to maximize sensitivity (screening) increases false positives.
— Moving it to maximize specificity (confirmatory) increases false negatives.
— AUC summarizes discrimination; 0.5 is chance, 1.0 is perfect.
Step 3 management: When asked which test to add after a positive screen, the answer is almost always the more specific confirmatory test — not repeating the screen. Example: positive HIV antigen/antibody → HIV-1/2 differentiation immunoassay, not repeat 4th-gen screen.
— Step A: sensitive screen casts a wide net.
— Step B: specific confirmatory test sorts true from false positives.
— Net effect: combined specificity rises sharply, sensitivity falls slightly. Used when false positives are costly.
— HIV: 4th-generation Ag/Ab immunoassay (sensitive) → HIV-1/2 antibody differentiation (specific) → HIV RNA if discordant.
— Syphilis: treponemal EIA or RPR screen → confirmatory treponemal/non-treponemal opposite test; reverse-sequence algorithm common.
— Colorectal cancer: FIT or Cologuard → colonoscopy (diagnostic + therapeutic).
— Cervical cancer: Pap ± hrHPV co-test → colposcopy with biopsy.
— Lung cancer: LDCT → tissue biopsy for Lung-RADS 4 lesions.
— Diabetes: HbA1c or FPG abnormal → repeat on a separate day (or confirm with a second test type) before diagnosis, unless symptomatic with random glucose ≥200.
— Increases sensitivity, decreases specificity.
— Used when rapid rule-out is essential (e.g., trauma panels, ACS rule-out with ECG + troponin + risk score).
CCS pearl: In a CCS-style case after a positive screen, advance the clock and order the specific confirmatory test and arrange follow-up at the appropriate interval — do not loop on duplicate screens; the grader rewards efficient diagnostic sequencing.
— Grade A/B: offer or provide (e.g., BP, colorectal cancer 45–75, lung cancer LDCT in eligible smokers, HIV one-time 15–65, HCV one-time ≥18).
— Grade C: offer selectively based on individual circumstances.
— Grade D: do not screen — harm equals or exceeds benefit (e.g., PSA in men ≥70, ovarian cancer screening in average-risk women, pancreatic cancer in average-risk asymptomatic adults, carotid stenosis in asymptomatic adults).
— Grade I: insufficient evidence; document shared decision-making.
— Lynch syndrome: colonoscopy every 1–2 years starting age 20–25.
— BRCA1/2: annual MRI + mammogram starting age 25–30.
— Cirrhosis or chronic HBV: liver US ± AFP every 6 months for HCC.
— HIV: more frequent cervical cancer screening; anal cytology in select patients.
— Lead-time bias: earlier diagnosis lengthens apparent survival without changing date of death.
— Length-time bias: screening preferentially detects slow-growing indolent disease, inflating apparent survival.
— Overdiagnosis is the extreme of length-time bias — detecting disease that would never have caused harm (DCIS, indolent prostate cancer, papillary thyroid microcarcinoma).
Board pearl: If a stem cites "5-year survival" as evidence that screening works, suspect lead-time bias; the correct answer references disease-specific mortality from RCTs.
— Will the result change management? If no → don't order (low-value care, Choosing Wisely).
— Has the patient consented to potential downstream evaluations triggered by a positive result?
— Is the patient within the eligible screening window and life expectancy >10 years (the typical screening benefit horizon)?
— Limited life expectancy (frail elderly, advanced dementia, metastatic disease) — stop cancer screening.
— Pregnancy contraindicates several modalities (CT with contrast relatively, fluoroscopy when avoidable).
— Patient values opposing intervention even if test positive → screening is ethically inappropriate.
— Mammography: biennial 50–74 (USPSTF 2024 expanded to start at 40, biennial through 74).
— Cervical cytology: every 3 years 21–29; co-test every 5 years 30–65; stop at 65 if adequate prior screening.
— Colonoscopy: every 10 years 45–75 if normal; consider 76–85 individually; stop >85.
— LDCT: annual until 15 years since cessation or age >80.
Step 3 management: Before ordering any screening test, document life expectancy, patient preference, and a plan for a positive result — this triad is the test-ordering equivalent of informed prescribing and is graded on CCS-style stems.
— Bowel prep adequacy determines sensitivity; inadequate prep → repeat within 1 year.
— Risks: perforation (~0.05–0.1%), bleeding (especially with polypectomy), sedation complications.
— Surveillance interval depends on polyp pathology: tubular adenoma <10 mm → 7–10 years; advanced adenoma or ≥3 adenomas → 3 years; sessile serrated lesions with dysplasia → 3 years.
— Stereotactic core needle biopsy preferred; surgical excision reserved for discordant or specific pathology (e.g., ADH, radial scar).
— BI-RADS 3 → short-interval (6-month) follow-up imaging, not biopsy.
— Lung-RADS 3 → repeat LDCT in 6 months.
— Lung-RADS 4A → 3-month LDCT or PET; 4B/4X → biopsy or surgical referral.
— Sampling error reduces sensitivity (prostate biopsy, endometrial biopsy) — negative result does not fully exclude disease in high pretest probability cases.
— Operator-dependent specificity (US, colposcopy).
— Verification bias: when only test-positives undergo gold-standard confirmation, sensitivity is overestimated.
CCS pearl: After ordering an invasive confirmatory test, advance the simulated clock to obtain pathology, schedule a results-discussion visit, and document informed consent for the procedure — the grader credits the full diagnostic loop, not just the order.
— Cervical cancer screening: stop at 65 if adequate prior negative screening and no high-grade history.
— Colorectal cancer screening: routine through 75; individualized 76–85; stop >85.
— Mammography: USPSTF supports biennial through 74; 75+ individualized.
— PSA: do not screen ≥70 (Grade D).
— Lung cancer LDCT: stop at 81 or once 15 years since cessation.
— Use validated tools (e.g., ePrognosis) to estimate 5- and 10-year mortality.
— In advanced dementia, metastatic cancer, dialysis-dependent CKD with limited transplant candidacy, screening for new cancers is generally inappropriate.
— Contrast-enhanced confirmatory imaging (CT-PA, CT urography) requires eGFR assessment; consider V/Q scan or non-contrast MRI alternatives.
— Gadolinium and NSF risk in eGFR <30 — use group II agents cautiously.
— Bowel prep choice: avoid sodium phosphate prep in CKD (phosphate nephropathy); use PEG.
— Cirrhosis converts a screening question into surveillance: US ± AFP every 6 months for HCC regardless of other comorbidities, as long as transplant or curative therapy remains feasible.
— Coagulopathy increases biopsy risk; consider transjugular approach.
Board pearl: "Stop screening" is the right answer when estimated life expectancy is under the time-to-benefit (~10 years for most cancer screens, ~5 years for AAA repair benefit). Document the conversation rather than reflexively ordering tests because they are "due."
— Defer: mammography (unless symptomatic), routine LDCT, elective colonoscopy.
— Continue: BP, gestational diabetes (24–28 weeks, 1-hour 50 g GCT → 3-hour GTT if abnormal), depression (PHQ-9 each trimester and postpartum), IPV.
— First trimester: HIV, syphilis, HBV surface antigen, rubella, varicella history, urine culture, blood type/Rh and antibody screen, hemoglobinopathy as indicated.
— Aneuploidy: cell-free DNA (high sensitivity/specificity for trisomy 21 across risk groups) or combined first-trimester screen → confirm with diagnostic CVS or amniocentesis if positive — classic screen-then-confirm sequence.
— Group B Strep rectovaginal culture at 36–37 weeks 6 days (universal screening).
— Anatomy ultrasound 18–22 weeks.
— Newborn metabolic screen (state-mandated panels); critical congenital heart disease pulse oximetry.
— Lead screening at 12 and 24 months if Medicaid-enrolled or risk factors.
— Vision and hearing per AAP Bright Futures.
— Lipid screening once between ages 9–11 and again 17–21.
— Depression annually starting age 12; HIV one-time 15–18.
— Autism-specific screening at 18 and 24 months.
Key distinction: cfDNA is a screening test no matter how good its sensitivity sounds; CVS/amniocentesis is the diagnostic test. Step 3 questions exploit this confusion specifically.
— Thyroid microcarcinomas (incidental on neck imaging).
— DCIS detected only by mammography.
— Indolent prostate adenocarcinoma (Gleason 6).
— Small renal masses on abdominal imaging.
— Pulmonary subsolid nodules.
Management drift toward active surveillance reflects recognition of this harm.
— Inadequate bowel prep colonoscopy.
— Dense breast tissue on mammography (consider supplemental US/MRI).
— Early HIV (window period for antibody-only tests).
— Sampling error in prostate, endometrial, or liver biopsy.
Board pearl: When a stem describes anxiety, repeat testing, and biopsy after a screening test, the answer often involves shared decision-making before the next screen — not aggressive workup. Quantify residual risk before recommending procedures.
— Positive confirmatory test requires staging/treatment (oncology, surgery, GI, pulmonary, cardiology).
— Indeterminate result persists after appropriate workup (Lung-RADS 3 stable but patient anxious; equivocal cfDNA results).
— Hereditary cancer syndrome suspected (genetic counseling referral): early-onset, multiple primaries, syndromic patterns, Ashkenazi ancestry with breast/ovarian/pancreatic family history.
— Breast cancer ≤45, triple-negative ≤60, male breast cancer, ovarian cancer at any age.
— Colorectal cancer <50, multiple Lynch-spectrum cancers, abnormal MMR/MSI testing on tumor.
— ≥3 first-degree relatives with related cancers.
— Reportable diseases from positive screens: HIV, syphilis, TB, certain STIs, hepatitis B/C — to local health department.
— Cancer registry reporting is institutional but flows through pathology.
— Track screening rates as HEDIS/MIPS quality measures.
— Closed-loop result management is a Joint Commission patient safety priority; missed abnormal results are a frequent malpractice driver.
— ACA mandates first-dollar coverage for USPSTF Grade A/B screens; diagnostic follow-up may carry cost-sharing — a known barrier and exam-relevant inequity.
Step 3 management: When a stem hands you a positive screen, do not stop at the order — explicitly arrange specialist referral, communicate the result to the patient, and document a follow-up plan with a defined timeline. This closed-loop process is the CCS-graded gold standard for diagnostic safety.
— Screening = population-based, asymptomatic, structured intervals.
— Case-finding = opportunistic identification during a clinical encounter prompted by another concern (e.g., asking about smoking during a visit for back pain).
— Screening = looking for new disease in those without it.
— Surveillance = monitoring those with a known precursor or prior disease (Barrett esophagus EGD intervals, post-polypectomy colonoscopy, post-curative-cancer follow-up imaging).
— Diagnostic test applied to a symptomatic patient or after positive screen; confirmatory specifically follows a screen and is selected for high specificity.
— Prognostic predicts disease course independent of therapy (Oncotype DX recurrence score).
— Predictive identifies who will respond to a specific therapy (HER2 for trastuzumab, EGFR for osimertinib).
— Validity = accuracy (does it measure what it claims; assessed by sensitivity, specificity, AUC).
— Reliability = reproducibility (Cohen kappa for categorical, ICC for continuous).
— Internal: study design rigor (randomization, blinding, intention-to-treat).
— External: generalizability to your patient population.
— Efficacy under trial conditions; effectiveness under real-world conditions — screening uptake, adherence to follow-up, and care continuity all degrade real-world performance.
Key distinction: A "good test" in isolation means nothing — clinical utility depends on the right test for the right patient at the right pretest probability with a downstream plan that improves outcomes.
Mitigation strategies on the boards:
— Demand RCT evidence with mortality endpoints for screening recommendations.
— Recognize that "stage shift" alone is insufficient evidence of benefit.
— Interpret 5-year survival cautiously; population-based mortality is the credible metric.
Board pearl: If a question shows that screening "improves 5-year survival from 30% to 70%" but mortality rates are unchanged, the answer is lead-time and overdiagnosis bias — not screening efficacy.
— Update problem list, medications, immunizations.
— Review age- and sex-appropriate USPSTF Grade A/B screens.
— Verify intervals; avoid both under- and over-screening.
— Document shared decision-making for Grade C/I or patient-preference-sensitive items (PSA, mammography 40–49, lung cancer LDCT, AAA in non-smoking men).
— EHR registries and population health dashboards identify overdue screens.
— Closed-loop result management protocols ensure no abnormal screen is lost to follow-up.
— Patient portal messages and outreach calls escalate when patients miss follow-up.
— Define confirmatory test, who orders it, and timeline (typically within weeks for cancer screens).
— Communicate results in person or via secure messaging with clear next steps.
— Connect to specialty care; warm handoffs reduce no-show rates.
— Tobacco cessation, BP control, statin therapy for primary prevention (USPSTF Grade B in ASCVD risk ≥10%), influenza and pneumococcal vaccines, HPV vaccination as primary prevention reducing future screening burden.
CCS pearl: When closing a CCS case, schedule the next health maintenance visit with the appropriate interval and explicitly document deferrals or stops — the grader rewards completed longitudinal planning, not just acute issue resolution.
— Negative routine screen → next scheduled interval per guideline.
— Borderline/indeterminate → short-interval repeat (e.g., 6-month LDCT, 6-month mammogram for BI-RADS 3).
— Positive screen → confirmatory test within defined window (typically 2–6 weeks).
— Confirmed disease → handoff to disease-specific surveillance/treatment pathway.
— Purpose and target disease.
— Sensitivity and specificity in plain language ("about 1 in 10 positives turns out to be a true cancer").
— Possible outcomes: negative, false positive, true positive, indeterminate.
— Downstream testing required if positive, including risks.
— Patient's right to decline; documentation of decision.
— Use natural frequencies, not probabilities ("Out of 1000 women like you, …").
— Avoid relative risk in isolation; pair with absolute risk and number needed to screen.
— Number needed to screen examples: ~250 for breast cancer mortality reduction over 10 years (age 50–74); ~320 for colorectal cancer over 10 years.
— Address language, literacy, transportation, and cost-sharing barriers.
— Patient navigators improve follow-through after abnormal screens, especially in underserved populations.
Step 3 management: When the stem describes a confused patient about a "positive" cfDNA result for trisomy 21, the correct counseling reframes the result as high pretest probability needing diagnostic amniocentesis, not a definitive diagnosis — and offers genetic counseling referral.
— Discussion of benefits, harms, alternatives, and option to decline.
— Use of decision aids when available.
— GINA protects against employment and health insurance discrimination but not life, disability, or long-term care insurance.
— Cascade testing of family members raises consent and disclosure dilemmas.
— Direct-to-consumer test results often lack clinical validity; confirm with clinical-grade testing before action.
— HIV, syphilis, gonorrhea, chlamydia, TB, certain hepatitis cases reported to local health departments.
— Child abuse, elder abuse, dependent adult abuse identified on psychosocial screening — mandatory reporter obligations apply across jurisdictions.
— Lost-to-follow-up after abnormal Pap, mammogram, or FIT is a top malpractice category.
— Implement closed-loop tracking: result acknowledgment, patient notification, and confirmatory test scheduling all documented.
— On hospital discharge, abnormal incidental findings (e.g., pulmonary nodule on CT for pneumonia) must be communicated to outpatient PCP with explicit follow-up timeline.
— Screening guidelines should reduce, not exacerbate, disparities; updated CRC screening at 45 partially addresses earlier CRC incidence in Black Americans.
— Insurance status–driven gaps in confirmatory testing represent an ethical, not just logistical, problem.
Board pearl: A CT lung nodule discovered during ED workup that is not communicated to outpatient follow-up is a patient safety event — the correct answer involves explicit communication to the PCP and patient, with documented follow-up plan, not "no action needed."
Key distinction: Memorize which tests are screens, which are confirmatory, and the typical sensitivity–specificity trade-off that justifies each role.
Board pearl: When two answer choices are both reasonable, choose the one that matches test characteristic to clinical task (sensitive for screening/rule-out, specific for confirmation/rule-in).
High-yield recap bullets:
Step 3 management: Approach every test order like a prescription — confirm indication, contraindications, dosing (interval), adverse-effect counseling, and a plan for managing the result before you click "order."