United States Medical Licensing Examination (USMLE) Behavioral Science, Psychiatry, and Population Health Flashcards

24 sample cards from the Behavioral Science, Psychiatry, and Population Health deck

Sampled from the end of the deck, so these are different cards from the ones shown on the syllabus page.

1. Which study design is best for a rare disease, and which is best for a rare exposure?

   Rare DISEASE: case-control study (diseased vs. non-diseased, look back at exposure; yields odds ratio). Rare EXPOSURE: cohort study (exposed vs. unexposed, follow forward; yields relative risk).

2. How does a cohort study differ from a cross-sectional study in temporality and the measure produced?

   Cohort: longitudinal follow-up; establishes temporality and computes incidence/relative risk. Cross-sectional: single point in time (snapshot); yields prevalence but cannot establish temporality or causality.

3. What feature of a randomized controlled trial minimizes confounding, and what does double-blinding control for?

   Randomization distributes known and unknown confounders equally between groups, minimizing confounding. Double-blinding (patient and assessor unaware of allocation) minimizes observer/ascertainment bias and placebo effects.

4. Define sensitivity and specificity, and give the SnNout and SpPin mnemonics.

   Sensitivity = TP/(TP+FN), true positive rate. Specificity = TN/(TN+FP), true negative rate. SnNout: a highly Sensitive test, when Negative, rules OUT disease. SpPin: a highly Specific test, when Positive, rules IN disease.

5. Define PPV and NPV and state how disease prevalence affects each.

   PPV = TP/(TP+FP) = probability of disease given a positive test. NPV = TN/(TN+FN) = probability of no disease given a negative test. As prevalence rises, PPV increases and NPV decreases. Sensitivity/specificity are independent of prevalence.

6. Define positive and negative likelihood ratios and what values indicate a clinically useful test.

   LR+ = sensitivity/(1 - specificity); LR- = (1 - sensitivity)/specificity. LR+ >10 and LR- <0.1 strongly shift disease probability. Likelihood ratios are prevalence-independent and convert pre-test to post-test odds.

7. How does lowering a test's cutoff to increase sensitivity affect specificity, and what does the ROC curve show?

   Increasing sensitivity (lower threshold) generally decreases specificity (a trade-off). The ROC curve plots sensitivity vs. 1 - specificity across cutoffs; larger area under the curve (AUC closer to 1) indicates a better test.

8. Write the formulas for relative risk (RR) and odds ratio (OR), and state when OR approximates RR.

   RR = [a/(a+b)] / [c/(c+d)] (incidence in exposed / unexposed; cohort/RCT). OR = (a*d)/(b*c) (case-control). The OR approximates the RR when the disease is rare.

9. Define attributable risk, relative risk reduction, absolute risk reduction, and number needed to treat.

   AR = risk in exposed - risk in unexposed. RRR = 1 - RR. ARR = control event rate - treatment event rate. NNT = 1/ARR (number treated to prevent one outcome); NNH = 1/(absolute risk increase).

10. Define incidence and prevalence and state their relationship.

    Incidence = new cases / population at risk over time. Prevalence = existing cases / total population at a point in time. Prevalence ~= incidence x average disease duration; it rises with longer survival or duration.

11. Define type I error, type II error, alpha, beta, and statistical power.

    Type I error (alpha): rejecting a true null hypothesis (false positive). Type II error (beta): failing to reject a false null (false negative). Power = 1 - beta, the probability of detecting a true effect; increased by larger sample and effect size.

12. What is a p-value, and when does a 95% confidence interval indicate significance for a difference of means versus a ratio (RR/OR)?

    A p-value is the probability of results as extreme as observed if the null were true; p<0.05 is conventionally significant. A 95% CI for a difference of means is significant if it excludes 0; a 95% CI for an RR/OR is significant if it excludes 1.

13. Match the appropriate statistical test: two means, three or more means, and two categorical variables.

    Two means: t-test. Three or more means: ANOVA. Two or more categorical variables/proportions: chi-square test. (Correlation/regression assess relationships between continuous variables.)

14. For a normal distribution, what percentage of values fall within 1, 2, and 3 standard deviations of the mean?

    About 68% within +/-1 SD, 95% within +/-2 SD (precisely 1.96), and 99.7% within +/-3 SD. The standard error of the mean = SD/sqrt(n) and decreases with larger sample size.

15. What is the hierarchy of evidence in evidence-based medicine, from strongest to weakest?

    Systematic reviews/meta-analyses of RCTs > individual RCTs > cohort studies > case-control studies > case series/reports > expert opinion. Meta-analyses pool data but are limited by publication bias and heterogeneity.

16. What characteristics make a disease appropriate for screening (Wilson-Jungner criteria)?

    The disease should be common/serious with a detectable asymptomatic (preclinical) phase; an effective, acceptable treatment must exist; a suitable, accurate, acceptable test must be available; and early treatment must improve outcomes.

17. Define lead-time bias and length-time bias in screening.

    Lead-time bias: earlier detection makes survival APPEAR longer though death timing is unchanged. Length-time bias: screening preferentially detects slowly progressive disease, falsely improving apparent survival. Mortality-based randomized trials avoid both.

18. Define primary, secondary, and tertiary disease prevention with an example of each.

    Primary: prevent disease onset (vaccination, seatbelts). Secondary: detect/treat early or asymptomatic disease (Pap smear, mammography). Tertiary: reduce complications/disability in established disease (stroke rehab, diabetic foot care).

19. What are the general screening recommendations for colorectal and cervical cancer?

    Colorectal cancer: begin at age 45 (e.g., colonoscopy every 10 years) for average-risk adults through 75. Cervical cancer: start at 21 with cytology every 3 years; ages 30-65 cytology q3y or HPV/co-testing q5y.

20. What are the recommended screening parameters for breast cancer and abdominal aortic aneurysm?

    Breast cancer: biennial mammography for women 50-74 (offered starting at 40 per updated guidance). AAA: one-time ultrasound for men aged 65-75 who have ever smoked.

21. What is the Plan-Do-Study-Act (PDSA) cycle, and how does quality improvement differ from research?

    PDSA is an iterative QI cycle: Plan a change, Do (implement on a small scale), Study results, Act (adopt/adapt/abandon). QI improves local processes via rapid cycles; research generates generalizable knowledge with fixed protocols.

22. In the Swiss cheese model, how do active errors differ from latent errors, and what is a sentinel event?

    Active errors occur at the sharp end (point of care) with immediate effect (e.g., wrong drug given). Latent errors are system/organizational flaws (blunt end), e.g., understaffing. A sentinel event causes death or serious harm and triggers root cause analysis.

23. Name the six Institute of Medicine aims for quality healthcare (STEEEP).

    Safe, Timely, Effective, Efficient, Equitable, and Patient-centered. (Donabedian separately evaluates quality by Structure, Process, and Outcome.)

24. What are the leading global causes of death and child mortality, and what does a DALY measure?

    Cardiovascular disease (ischemic heart disease, stroke) leads adult mortality; lower respiratory infections, diarrheal disease, and neonatal conditions dominate child mortality. DALY = years of life lost (YLL) + years lived with disability (YLD), quantifying total disease burden.

What this deck covers

Behavioral Science, Psychiatry, and Population Health flashcards FAQ