• i: 4 (Cohort study)
• ii: 5 (Experiment / RCT)
• iii: 1 (Cross-sectional study)
• iv: 2 (Case-control study)
• v: 3 (Observational study)

• i. Cross-sectional survey
• ii. Prospective cohort study
• iii. Double-blind Randomized Controlled Trial (RCT)
• iv. Diagnostic accuracy / descriptive evaluation study (Cross-sectional diagnostic study)
• v. Case series (or descriptive study)

• Study Designs:
  • 1. Experimental Studies: Randomized Controlled Trials (RCTs), Non-randomized clinical trials.
  • 2. Observational Studies:
    • A. Analytic Studies:
      • Prospective (Cohort) studies
      • Retrospective (Case-control) studies
    • B. Descriptive Studies: Cross-sectional surveys, Case series, Case reports.

• a. Probability sampling technique
• b. Systematic Random Sampling (SyRS)
• c. Sampling interval ($k=N/n=1000/250=4$)
• d. Easy, fast, and convenient to implement once the sampling frame and interval are established.
• e. Periodic or cyclic patterns in the sampling frame can introduce substantial bias; fails to give proportional representation if hidden periodicity exists.

• a. Receiver Operating Characteristic (ROC) curve
• b. Evaluates diagnostic test accuracy across continuous thresholds, identifies the optimal cut-off value, and compares diagnostic performance between different diagnostic tests.
• c. $1-\text{Specificity}$ (False Positive Rate)
• d. Different cut-off points of the diagnostic test plotting their respective Sensitivity vs. ($1-\text{Specificity}$).
• e. The point closest to the upper left corner represents the optimal cut-off (maximizing Youden index: $\text{Sensitivity}+\text{Specificity}-1$). The Area Under the Curve (AUC) quantifies overall accuracy:
  • $\text{AUC}=0.5$: No discrimination (diagonal reference line)
  • $\text{AUC}0.8\text{–}0.9$: Good accuracy
  • $\text{AUC}>0.9$: Excellent accuracy

• a. Fagan's Nomogram
• b. Provides a rapid graphical method to calculate post-test probability from pre-test probability and likelihood ratio (LR).
• c.
  • Left column: Pre-test probability (%)
  • Middle column: Likelihood ratio (LR)
  • Right column: Post-test probability (%)
• d. A straight line connecting the known pre-test probability through the test's Likelihood Ratio is extended to intersect the right axis, giving the post-test probability.
• e. Post-test probability of approximately 59%.

• a. Inter-rater agreement / Cohen's Kappa ($\kappa$) coefficient
• b. Crude / Observed Agreement ($P_o$):
  $P_o=\frac{46+32}{100}=\frac{78}{100}=78\mathrm{\%}\text{ (0.78)}$
• c. Expected Chance Agreement ($P_e$):
  • Clinician I positive: $46+10=56$; Clinician II positive: $46+12=58⟹\text{Chance positive}=\frac{56\times 58}{100}=32.48$
  • Clinician I negative: $12+32=44$; Clinician II negative: $10+32=42⟹\text{Chance negative}=\frac{44\times 42}{100}=18.48$
  • Total $P_e=32.48\mathrm{\%}+18.48\mathrm{\%}=50.96\mathrm{\%}\approx 51\mathrm{\%}$
• d. Actual agreement beyond chance:
  $P_o-P_e=78\mathrm{\%}-51\mathrm{\%}=27\mathrm{\%}\text{ (0.27)}$
• e. Potential agreement beyond chance:
  $100\mathrm{\%}-P_e=100\mathrm{\%}-51\mathrm{\%}=49\mathrm{\%}\text{ (0.49)}$
• f. Kappa coefficient ($\kappa$):
  $\kappa =\frac{P_o-P_e}{1-P_e}=\frac{27}{49}=0.55\text{ (Moderate agreement)}$

• a. Bland–Altman Plot (Difference plot)
• b. Assesses the agreement between two clinical measurement methods / instruments to determine if they can be used interchangeably.
• c. X-axis: Mean of the two measurements: $\frac{A+B}{2}$
• d. Y-axis: Difference between the two measurements: $(A-B)$
• e.
  • Solid horizontal line: Mean difference (systematic bias).
  • Dotted horizontal lines: 95% Limits of Agreement ($\text{Mean difference}\pm 1.96\times \text{SD}$).

• Systematic reviews and meta-analysis: PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses), QUORUM, AMSTAR-2, CASP checklist
• Meta-analysis of non-RCTs: MOOSE (Meta-analysis Of Observational Studies in Epidemiology)
• Randomized controlled trials (RCTs): CONSORT (Consolidated Standards of Reporting Trials), Cochrane Risk of Bias tool (RoB 2)
• Observational studies: STROBE (Strengthening the Reporting of Observational Studies in Epidemiology)
• Studies on diagnostic test evaluation: STARD (Standards for Reporting of Diagnostic Accuracy Studies), QUADAS-2

• a.
  • Image A: Cochrane Risk of Bias Graph (overall percentages across domains).
  • Image B: Cochrane Risk of Bias Summary (study-level judgments).
• b. Evaluates internal validity and potential biases in randomized controlled trials included in systematic reviews.
• c.
  • Green: Low risk of bias
  • Yellow: Unclear risk of bias / Some concerns
  • Red: High risk of bias
• d.
  • "+": Low risk of bias
  • "?": Unclear risk / insufficient information
  • "-": High risk of bias
• e. Domains evaluated: Random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other sources of bias.

• a. Decision algorithm for assessing Risk of Bias in selection of the reported results (Domain 5 of RoB 2).
• b. Used during quality appraisal of clinical trials in systematic reviews.
• c.
  • Y / PY: Yes / Probably Yes
  • N / PN: No / Probably No
  • NI: No Information
• d. Standardized signaling questions for selective outcome reporting.
• e. Overall risk of bias classification for that domain.

• a. Funnel Plot
• b.
  • Symmetric plot: Absence of publication bias (studies evenly distributed around the pooled effect estimate).
  • Asymmetric plot: Presence of publication bias or small-study effects.
• c. Individual studies included in the meta-analysis (X-axis: Effect size; Y-axis: Study precision / standard error / sample size).
• d. Missing small studies with negative, null, or unfavorable results that were not published.
• e. Missing small studies with larger positive effect size.

• Continuous quantitative: Blood pressure, height, weight, age
• Discrete quantitative: Number of children, number of asthma exacerbations per week
• Binary/Dichotomous: Yes/No, Dead/Alive, Present/Absent
• Ordinal categorical: Cancer grades (I–IV), Likert scale (Agree/Neutral/Disagree)
• Nominal categorical: Blood groups (A, B, AB, O), Eye color, Marital status

• a. Box-and-Whisker Plot
• b. Non-parametric / non-Gaussian (skewed) continuous data.
• c. Captures central tendency (median), dispersion (interquartile range), skewness, and outliers.
• d. Median (50th percentile)
• e. Box ends represent the 25th percentile ($Q_1$) and 75th percentile ($Q_3$) (Interquartile Range, $\text{IQR}=Q_3-Q_1$). The outer whiskers represent minimum and maximum non-outlier values ($\pm 1.5\times \text{IQR}$).

2x2 Contingency Table:

Salt Intake	Hypertension (+)	Hypertension (−)	Total
High Salt (Exposed)	34	380	414
Normal Salt (Unexposed)	28	524	552
Total	62	904	966

• Incidence in exposed ($I_e$): $\frac{34}{414}=8.21\mathrm{\%}$
• Incidence in unexposed ($I_u$): $\frac{28}{552}=5.07\mathrm{\%}$
• Incidence in total population ($I_p$): $\frac{62}{966}=6.41\mathrm{\%}$

---

• a. Attributable Risk (Risk Difference, $AR$):
  $AR=I_e-I_u=8.21\mathrm{\%}-5.07\mathrm{\%}=3.14\mathrm{\%}\text{ (3.14 per 100)}$
• b. Attributable Risk Fraction ($AR\mathrm{\%}$):
  $AR\mathrm{\%}=\frac{I_e-I_u}{I_e}\times 100=\frac{3.14}{8.21}\times 100=38.2\mathrm{\%}$
• c. Population Attributable Risk ($PAR$):
  $PAR=I_p-I_u=6.41\mathrm{\%}-5.07\mathrm{\%}=1.34\mathrm{\%}\text{ (1.34 per 100)}$
• d. Population Attributable Risk Fraction ($PAR\mathrm{\%}$):
  $PAR\mathrm{\%}=\frac{I_p-I_u}{I_p}\times 100=\frac{1.34}{6.41}\times 100=20.9\mathrm{\%}$
• e.
  • i. Relative Risk ($RR$):
    $RR=\frac{I_e}{I_u}=\frac{34/414}{28/552}=\frac{0.0821}{0.0507}=1.62$
  • ii. Odds Ratio ($OR$):
    $OR=\frac{a\times d}{b\times c}=\frac{34\times 524}{380\times 28}=\frac{17816}{10640}=1.67$

• a. Kaplan–Meier Survival Curves
• b. Occurrence of the event of interest (e.g., mortality, treatment failure, relapse).
• c. Magnitude of decrease in cumulative survival probability at that specific time point.
• d. Instantaneous drop in survival probability at the time of an event.
• e. Censored observations (patients lost to follow-up, withdrew, or completed the study period without experiencing the event).

• a. Standard Error of the Mean ($SE$):
  $SE=\frac{SD}{\sqrt{n}}=\frac{3.0}{\sqrt{250}}=\frac{3.0}{15.81}=0.19\text{ mg/dL}$
• b. 95% Confidence Interval ($95\mathrm{\%}\text{ CI}$):
  $\text{Mean}\pm (1.96\times SE)\approx 94.0\pm (2\times 0.19)=94.0\pm 0.38=[93.62\text{ to }94.38]\text{ mg/dL}$

• a. Sample size calculation for prevalence:
  $n=\frac{Z_{\alpha /2}^2\times p\times (1-p)}{d^2}$
  • $Z_{\alpha /2}=1.96$ (for 95% confidence level)
  • $p=0.041$ (predicted prevalence = 4.1%)
  • $d=0.01$ (absolute precision = $\pm 1\mathrm{\%}$)
    $n=\frac{(1.96{)}^2\times 0.041\times (1-0.041)}{(0.01{)}^2}=\frac{3.8416\times 0.041\times 0.959}{0.0001}=\frac{0.15105}{0.0001}=1511\text{ children}$

![[images/image_rsrc1ZU2.jpg]]

• a. Sample size for comparing two independent proportions:
  $n=\frac{Z_{\alpha /2}^2\times [p_1(1-p_1)+p_2(1-p_2)]}{d^2}$
  • $Z_{\alpha /2}=1.96$
  • $p_1=0.95$, $1-p_1=0.05$
  • $p_2=0.75$, $1-p_2=0.25$
  • $d=0.05$ (precision = 5%)
    $n=\frac{3.8416\times [(0.95\times 0.05)+(0.75\times 0.25)]}{(0.05{)}^2}=\frac{3.8416\times [0.0475+0.1875]}{0.0025}=\frac{3.8416\times 0.235}{0.0025}=361\text{ per group}$
    (Or $n\approx 36$ per group depending on formula precision notation).

2x2 Contingency Table:

Head Nodding	Hypoxia Present (+)	Hypoxia Absent (−)	Total
Present (+)	29 (TP)	2 (FP)	31
Absent (−)	28 (FN)	53 (TN)	81
Total	57	55	112

• a. Sensitivity: $\frac{\text{TP}}{\text{TP}+\text{FN}}=\frac{29}{57}=50.9\mathrm{\%}$
• b. Specificity: $\frac{\text{TN}}{\text{TN}+\text{FP}}=\frac{53}{55}=96.4\mathrm{\%}$
• c. Positive Predictive Value (PPV): $\frac{\text{TP}}{\text{TP}+\text{FP}}=\frac{29}{31}=93.5\mathrm{\%}$
• d. Negative Predictive Value (NPV): $\frac{\text{TN}}{\text{TN}+\text{FN}}=\frac{53}{81}=65.4\mathrm{\%}$
• e.
  • i. Positive Likelihood Ratio ($LR+$):
    $LR+=\frac{\text{Sensitivity}}{1-\text{Specificity}}=\frac{0.509}{1-0.964}=\frac{0.509}{0.036}=14.1$
  • ii. Negative Likelihood Ratio ($LR-$):
    $LR-=\frac{1-\text{Sensitivity}}{\text{Specificity}}=\frac{1-0.509}{0.964}=\frac{0.491}{0.964}=0.51$

• a. Forest Plot (Blobbogram) in a meta-analysis.
• b. Names of individual primary studies (author/year) included in the systematic review.
• c. Sample sizes, event counts, or continuous variable means/SDs for intervention and control groups.
• d. Quantitative effect sizes (Odds Ratio / Relative Risk / Mean Difference) with their 95% Confidence Intervals and statistical study weights (%).
• e. Line of Null Effect ($OR=1$ / $RR=1$ / $\text{Difference}=0$).

• 1. 9 (0.6–12.0): Not significant (95% CI includes 1.0)
• 2. 9 (0.5–12.0): Not significant (95% CI includes 1.0)
• 3. 9 (0.8–10.0): Not significant (95% CI includes 1.0)
• 4. 9 (5.0–12.0): Statistically Significant (95% CI does not include 1.0; entirely above 1.0)

• 1: c (Target group of investigation)
• 2: a (Measures obtained for entire target population)
• 3: e (Uncertainties in medical care)
• 4: b (Fraction of the population that is selected)
• 5: d (Measures obtained for sample population)

• 1: c (No relationship / difference exists)
• 2: e (No specific direction is predicted)
• 3: b (There is a relationship / difference)
• 4: a (A specific direction is predicted)
• 5: d (Specific testable statement of prediction)

• 1: b (Student's t-test - Unpaired)
• 2: a (Wilcoxon rank-sum / Mann–Whitney U test)
• 3: e (Analysis of Variance - ANOVA)
• 4: c (Student's t-test - Paired)
• 5: d (Wilcoxon signed-rank test)

• 1 (Level I): c (Systematic review of multiple well-designed RCTs)
• 2 (Level II): a (At least one properly designed large RCT)
• 3 (Level III): e (Well-designed trials: cohort, case-control, nonrandomized trials)
• 4 (Level IV): b (Well-designed nonexperimental/observational studies from >1 center)
• 5 (Level V): d (Opinions of expert authorities, clinical reports, committee consensus)

• 1. False (IEC must function independently of institutional administration and financial influence).
• 2. True (All human subject research requires prior ethical clearance).
• 3. False (IEC can and should evaluate both scientific validity and ethical rigor).
• 4. False (Large institutions can have multiple specialized IRBs, e.g., biomedical, socio-behavioral).
• 5. True (Small institutions without their own IEC can utilize accredited external or institutional ethics committees).