---

WBUHS Community Medicine - Comprehensive Exam Answers (Park's Textbook)

---

CHAPTER 2: EPIDEMIOLOGICAL STUDY DESIGNS

---

Q1. Cross-Sectional Study for Prevalence of Obesity Among Medical Students (Barasat GMC, 2+13=15)

• The objective is to determine prevalence (not incidence or causation)
• The time frame is short - only 3 months, which precludes cohort studies
• The study population is well-defined and accessible (medical students)
• Cross-sectional studies measure exposure and outcome simultaneously at one point in time

---

• Title: "Prevalence of obesity among medical students of [Institution] — a cross-sectional study"
• Primary objective: To estimate the prevalence of obesity
• Secondary objectives: To identify associated socio-demographic factors (dietary habits, physical activity, screen time)

• Obesity: BMI ≥ 30 kg/m² (WHO criteria) or BMI ≥ 27.5 kg/m² for Asian Indians (ICMR criteria)
• Study population: All MBBS students of the institution (Phase 1, 2, 3)

• Sampling frame: Roll-call register of all enrolled medical students
• Sample size calculation: Using the formula n = Z²pq/d² where Z=1.96, p = estimated prevalence (use 20% from literature), d = allowable error 5%
• Sampling technique: Stratified random sampling - stratify by year of MBBS, then proportionate random sampling from each stratum to ensure all years are represented
• Alternatively: Systematic random sampling from the complete student roll

• Pre-tested, semi-structured questionnaire covering:
  • Demographic data: age, sex, year of study
  • Dietary habits (24-hour dietary recall / food frequency questionnaire)
  • Physical activity (IPAQ - International Physical Activity Questionnaire)
  • Family history of obesity, diabetes, hypertension
  • Sleep duration, screen time
• Anthropometric measurements: height (stadiometer), weight (calibrated weighing scale), waist circumference, hip circumference

• Obtain approval from Institutional Ethics Committee (IEC)
• Obtain written informed consent from all participants
• Maintain confidentiality of data
• Voluntary participation - no coercion

• Conduct pilot study on ~10% sample to pretest questionnaire, check feasibility, and refine data collection procedures

• Trained data collectors
• Standardized measurement techniques
• Blinding of anthropometric measurers to questionnaire responses
• Duration: 2-3 weeks during a designated period

• 10% re-interview for verification
• Daily review of filled forms
• Spot checks by supervisor

• Data entry in SPSS/Epi Info
• Descriptive statistics: frequency, proportion, mean ± SD
• Calculate prevalence of obesity with 95% CI
• Chi-square test for association with categorical variables
• Logistic regression for independent predictors of obesity

• Prepare report with STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines
• Submit to institution and publish findings

---

Q2/Q5. Village Reported 85 Cases of Acute Diarrhoeal Disease After Community Feast (Midnapore, 2+8+3+2=15)

• 85 cases in 3 days is a sudden, sharp increase — clearly in excess of expected (endemic) level
• The occurrence is time-limited (3 days) and place-limited (one village after a community feast)
• It has a common source (the community feast) — suggesting a point-source outbreak
• The temporal clustering (all within 3 days = within one incubation period of common food-borne pathogens) confirms a point-source outbreak

---

• Visit the affected area immediately
• Clinically examine a sample of cases
• Collect stool/rectal swab samples for culture and sensitivity
• Send samples to laboratory (E. coli, Vibrio cholerae, Salmonella, etc.)
• Do not wait for lab results — proceed with epidemiological investigation simultaneously

• Compare current case frequency with baseline (previous weeks/months)
• 85 cases in 3 days = clearly in excess of expected → outbreak confirmed
• Often obvious in common-source epidemics (cholera, food poisoning) — no complex comparison needed

• Obtain a map of the village
• Conduct a rapid house-to-house survey
• Count the total population (denominator for attack rate calculation)
• Identify all attendees of the community feast (true population at risk)

• Use an Epidemiological Case Sheet (case interview form) containing:
  • Name, age, sex, address
  • Date and time of symptom onset
  • Symptoms (vomiting, diarrhoea, fever, abdominal pain)
  • Foods eaten at the feast (food history — food-specific attack rates)
  • Water source
  • Whether they attended the feast or not
• House-to-house survey for unreported cases
• Check local hospital admissions and sub-centre records

• Time: Plot an epidemic curve (date/time of onset on x-axis, number of cases on y-axis)
  • A sharp unimodal peak = point-source outbreak
  • The peak will fall within one incubation period of the source exposure
• Place: Draw a spot map to show geographic clustering of cases
• Person: Analyze by age, sex, food eaten
  • Calculate food-specific attack rates: (persons who ate food X and got ill / total who ate food X) × 100

• Based on time-place-person data: "The outbreak is likely caused by contaminated food served at the community feast"
• Identify the most suspicious food item based on highest food-specific attack rate

• Compare attack rates among those who ate vs. those who did not eat the suspected food
• A Case-Control Study design can be used at this stage (see Part C)
• Microbiological testing of suspected food items, water samples, and hand swabs from food handlers

• Inspect the food preparation area
• Check sanitation of the feast venue
• Interview food handlers (check for illness among them)
• Inspect water supply (well, overhead tank, piped supply)
• Environmental samples from cooking vessels, water source

• Document all findings
• Submit to BMOH, CMO(H), and State surveillance system (IDSP)
• Make recommendations for future prevention

---

• Cases: persons who developed diarrhoea after the feast
• Controls: persons who attended the same feast but did NOT develop diarrhoea (matched by age/sex)
• Exposure of interest: specific foods eaten
• Calculate Odds Ratio for each food item
• The food with the highest OR and statistical significance is the likely vehicle

---

1. Safe Water Supply: Ensure access to safe drinking water — distribute ORS (Oral Rehydration Solution) and chlorination tablets; chlorinate the village water supply (0.5 ppm residual chlorine)
2. Food Safety and Disposal: Condemn and destroy all leftover food from the feast; prevent consumption of the suspected food vehicle; temporary closure/inspection of food establishments
3. Treatment: Set up treatment facilities — ORS, IV fluids for severe dehydration; hospitalize critical cases
4. Health Education: Advise community on hand hygiene (WASH) and safe food practices
5. Notification: Report to BMOH/CMO(H)/IDSP under Integrated Disease Surveillance Programme

---

Q3. Acute Watery Diarrhoea Outbreak - 45 Admitted, 2 Deaths (Barasat GMC, 6+4+5=15)

1. Verification of Diagnosis - Clinical examination + stool samples → likely Vibrio cholerae (rice-water stools, severe dehydration); confirm by dark-field microscopy and culture on TCBS agar
2. Confirmation of Outbreak - 45 cases in 48 hours with 2 deaths = clearly exceeds endemic level; Cholera must be reported to WHO as it is a Notifiable disease under IHR 2005
3. Notify Authorities - Immediate report to BMOH → CMO(H) → State Health Dept → IDSP
4. Define Population at Risk - Total village population = 1,500; identify high-risk groups (age, location, water source used)
5. Case Finding - House-to-house survey; epidemiological case sheets; check PHC and hospital admissions
6. Data Analysis: Time (epidemic curve), Place (spot map near water sources), Person (age/sex distribution, water source used)
7. Formulate Hypothesis - Most likely common source (contaminated water/food); if cases cluster around one water source, suspect waterborne transmission
8. Test Hypothesis - Cohort study of population; water samples for culture; compare attack rates by water source
9. Ecological Investigation - Inspect water supply (well depth, distance from latrine, structural integrity); check latrines and open defecation areas; assess sanitation
10. Report Writing - Document, submit, recommend

---

Attack Rate = (Number of new cases of disease / Population at risk during the specified period) × 100

• Total population of the village = 1,500
• Number of cases admitted in 48 hours = 45 (cases identified so far)
• Attack Rate = (45/1,500) × 100 = 3%

---

1. Safe Water: Emergency chlorination of all water sources (wells, overhead tanks) — target 0.5 ppm residual chlorine at point of use
2. ORS Distribution: Set up ORS corners at PHC and sub-centres
3. Isolation: Identify and isolate cholera cases; treat with ORS/IV fluids + Tetracycline/Doxycycline
4. Disinfection of Environment: Disinfect latrines, defecation sites with bleaching powder (lime)
5. Food Safety: Condemn contaminated food; advise cooking food thoroughly and eating within 2 hours
6. Waste Disposal: Prevent fecal contamination of water and food — ensure disposal of night-soil away from water sources
7. Health Education: WASH (Water, Sanitation and Hygiene) campaign in the village

1. Safe Piped Water Supply: Establish chlorinated piped water connection to all households under Jal Jeevan Mission
2. Construction of Sanitary Latrines: Under Swachh Bharat Mission - total sanitation coverage; eliminate open defecation
3. Solid Waste Management: Systematic garbage collection and disposal
4. Sewage Treatment: Proper drainage system to prevent contamination of groundwater
5. Food Safety Legislation: Enforce FSSAI norms; train and license food vendors and handlers
6. Health Education: Sustained community education on personal hygiene, handwashing with soap, food hygiene
7. Surveillance: Strengthen IDSP (Integrated Disease Surveillance Programme) for early detection of future outbreaks
8. Vector Control: Fly control measures (screens, disposal of organic waste)

---

Q4. Block 'X' - Children with Fever, Jaundice after Village Fair with Fuchka/Velpuri (College of Medicine & Sagore Dutta, 2+5+3=10)

• Children aged 5-10 years (most common affected group for Hepatitis A)
• Incubation period of Hepatitis A = 2-6 weeks; 2-week clustering after fair is consistent
• Symptoms: fever, loss of appetite (anorexia), nausea, jaundice (yellowish discoloration of eyes/skin) = classic presentation
• Route of transmission: Fecal-oral route via contaminated food (fuchka/puchka/jhalmuri prepared with contaminated water / unclean hands)
• Street food vendors at village fairs are a well-known common vehicle for Hepatitis A outbreaks in India

---

1. Verify Diagnosis - Liver function tests (elevated bilirubin, SGOT/SGPT), serology for anti-HAV IgM (confirmatory for acute Hepatitis A infection); clinically examine sample of cases
2. Confirm Outbreak Exists - Compare with baseline incidence of jaundice in the block; community awareness that multiple children are affected confirms outbreak
3. Define Cases - "A case is defined as any child aged 5-10 years residing in Block X who developed fever, anorexia, and jaundice within 2-6 weeks of attending the village fair"
4. Case Finding - House-to-house survey; questionnaire on food eaten at fair; school health check-ups; sub-centre reporting
5. Data Analysis:
   • Time: Epidemic curve - when did cases start? Uni-modal peak after fair
   • Place: Spot map - are cases clustered around specific stalls at the fair?
   • Person: Food-specific attack rates by type of food eaten (fuchka vs. jhalmuri vs. velpuri)
6. Hypothesis: Fuchka/puchka made with contaminated water used as tamarind water, served at specific stalls at the fair = likely common vehicle
7. Test Hypothesis - Case-control study: cases = jaundiced children, controls = fair-attending children without jaundice; Odds Ratio for each food item
8. Ecological Investigation - Inspect fair grounds, identify vendors, test water used in food preparation, hand hygiene practices of vendors
9. Control Measures (see Part C)
10. Report - Submit to BMOH, IDSP, CMO(H)

• Identify and close the implicated food stall
• Hepatitis A immunization (if available) for close contacts and high-risk individuals
• Passive immunoprophylaxis: Normal Human Immunoglobulin (0.02-0.06 mL/kg IM) for contacts within 2 weeks of exposure
• Safe water supply; hand hygiene; food hygiene education
• Isolate jaundiced cases (not strictly necessary as already infectious before symptoms, but to prevent further transmission)

---

1. Food Safety at Fairs/Public Gatherings:
   • Mandatory registration and licensing of food vendors under FSSAI
   • Ensure food vendors use safe (potable) water for all food preparation
   • Ban use of contaminated/surface water for street food
   • Regular inspection of food stalls during fairs by sanitary inspectors
2. Hepatitis A Vaccination: Include Hepatitis A vaccine in immunization schedule for high-risk groups; consider mass vaccination in endemic areas
3. Health Education: Community awareness about fecal-oral transmission, handwashing before eating, avoiding raw/undercooked street food, especially at public gatherings
4. Water Safety: Ensure only treated, chlorinated water is supplied at fairgrounds; provide handwashing facilities for vendors
5. Surveillance: IDSP-based sentinel surveillance for jaundice in schools/communities; early detection and response

---

Q6. Radiation Effects Among Nuclear Plant Workers - 5 Years (JNM Kalyani, 2+8+5=15)

• The exposure is defined at the start (nuclear plant workers vs. non-exposed workers)
• The disease has not yet occurred — workers will be followed forward in time
• Duration = 5 years → follows cohort forward = prospective
• Allows calculation of Incidence Rate and Relative Risk (RR)
• Best design for occupational exposure studies where the exposed group is identifiable
• Establishes temporal relationship (cause precedes effect) → strong causal inference

---

• Exposed cohort: All workers currently employed at the nuclear plant who are regularly exposed to radiation (classified by dose levels using dosimeters)
• Unexposed (comparison) cohort: Workers in the same plant's administrative section (no radiation exposure) OR workers in an unrelated industry — matched for age, sex, smoking status, socioeconomic status
• Exclusion criteria: workers with pre-existing radiation-related illness; those who will not be available for follow-up

• Complete medical history and physical examination
• Baseline investigations: CBC (to detect baseline haematological changes), thyroid function, serum markers
• Radiation exposure assessment: cumulative dose measurement (TLD badges/film badges)
• Demographic data: age, sex, duration of employment, type of work

• Categorize workers by radiation dose (low, medium, high) using personal dosimetry records
• Occupational Health Records maintained by the plant

• Active follow-up: Annual medical check-ups, CBC every 6 months
• Monitor for outcomes: leukaemia, lymphoma, thyroid cancer, aplastic anaemia, cataracts, genetic effects
• Track and minimize losses to follow-up (attrition)
• Maintain updated contact information; designate field workers

• Standardized case definitions for each outcome
• Blinding: the physician assessing outcomes should be blinded to exposure status (prevent observer bias)

• Calculate Incidence Rate in exposed and unexposed groups
• Relative Risk (RR) = Incidence in exposed / Incidence in unexposed
  • RR > 1 = positive association (radiation increases risk)
  • RR = 1 = no association
  • RR < 1 = protective
• Attributable Risk (AR) = Incidence in exposed − Incidence in unexposed
• Population Attributable Risk (PAR)
• Control for confounders using stratified analysis or multivariate regression (Cox proportional hazards model for time-to-event data)

• IEC approval; written informed consent from all participants

• Submit findings to regulatory authority (Atomic Energy Regulatory Board — AERB)
• Recommend dose reduction measures if RR is elevated

---

1. Time-consuming and expensive: Requires following a large cohort for several years; high cost for follow-up, investigations, and data management
2. Loss to follow-up (attrition): Workers may change jobs, retire, die from unrelated causes, or emigrate → selection bias and reduced statistical power; reduces validity
3. Not suitable for rare diseases: If the outcome (e.g., a specific cancer) is rare, very large numbers are needed, making it impractical
4. Change in exposure status: Workers may change their radiation exposure level over 5 years (job rotation, improved shielding), complicating exposure classification
5. Healthy worker effect (occupational cohort specific): Workers are healthier than the general population at baseline (pre-employment fitness screening), leading to underestimation of risk
6. Secular changes: Over 5 years, diagnostic criteria, treatment modalities, and environmental conditions may change, creating confounding
7. Not suitable for very long latency diseases: Some radiation-induced cancers have latency > 10 years; a 5-year study may miss outcomes
8. Ethical concerns: Cannot randomize workers to radiation exposure; observational nature limits causal inference

---

Q7. Case-Control Study for Obesity as Risk Factor for Knee Osteoarthritis (PC Sen, 8+4+3=15)

• Cases: Persons aged 35-65 years, diagnosed with osteoarthritis of the knee joint by a rheumatologist or orthopedic surgeon, using ACR (American College of Rheumatology) criteria (knee pain + at least 3 of: age >50, morning stiffness <30 min, crepitus, bony tenderness, bony enlargement, no warmth — with X-ray changes: osteophytes)
• Cases recruited from orthopedic OPD/rheumatology clinics

• Hospital-based: Orthopedic and rheumatology departments of teaching hospitals
• Alternatively community-based (population-based case-control)

• Controls: Persons aged 35-65 years, from the same hospital/community, WITHOUT knee osteoarthritis
• Matching: Match controls to cases by age (± 5 years) and sex (1:1 or 1:2 matching)
• Controls should be free from any arthritis or connective tissue disease
• Source: other OPD departments (e.g., ophthalmology, dermatology) or neighborhood controls

• Measure BMI (Body Mass Index = weight kg/height m²)
• Obese: BMI ≥ 30 kg/m² (WHO) or ≥ 27.5 kg/m² (Asian)
• Also measure waist circumference and waist-hip ratio (central obesity)

• Structured questionnaire: age, sex, BMI, occupation, dietary history, physical activity
• Weight and height measured by trained personnel with calibrated equipment
• Blinded data collection: interviewer should not know who is a case vs. control

• Use standard formula; assuming prevalence of obesity in controls = 20%, OR to detect = 2.5, power = 80%, α = 0.05

• 2×2 contingency table:

• Odds Ratio (OR) = (a × d) / (b × c)
• If OR > 1 and statistically significant (95% CI excludes 1), obesity is a risk factor
• Chi-square test for significance
• Multivariate logistic regression to adjust for confounders (age, sex, occupation, previous knee injury)

---

1. Relatively easy to conduct — quick and inexpensive
2. Suitable for rare diseases (e.g., osteoarthritis in specific age group)
3. Requires few subjects compared to cohort studies
4. Can study multiple aetiological factors simultaneously (obesity, prior injury, occupation, etc.)
5. No attrition problem — no follow-up needed
6. Ethical issues minimal — no experimental exposure
7. Rapid results — ideal for testing causal hypotheses

1. Recall bias (memory bias): Cases may better remember past obesity than controls
2. Selection bias: Cases and controls may not be representative of the target population
3. Cannot calculate incidence rates or relative risk directly — only Odds Ratio
4. Berkesonian bias: Hospital-based controls may differ systematically from the source population
5. Cannot study rare exposures efficiently
6. Temporal relationship may be difficult to establish (was obesity present before OA?)
7. Only one outcome (disease) can be studied at a time
8. Dependent on availability and accuracy of past records

---

1. Recall Bias (Memory Bias): Cases (with OA) tend to recall and report their past obesity/weight gain more accurately or more frequently than controls (who are healthy and less motivated to remember). This leads to overestimation of OR. Control: Use objective records like past medical records, standardized anthropometric data.
2. Selection Bias: Cases selected from tertiary care hospitals may represent severe disease; controls from different OPD departments may have different characteristics than controls in the general population. Control: Use population-based controls; ensure comparable source populations.
3. Berkesonian Bias: If both cases and controls are hospital patients, those with two conditions are over-represented in hospitals (admission rate bias). Control: Use community-based case-control design.
4. Interviewer Bias: If the interviewer knows who is a case, they may probe cases more thoroughly for obesity history than controls. Control: Blind the interviewer to case/control status (double-blinding).
5. Confounding: Variables like age, sex, occupation (heavy manual labor), previous knee injury all confound the obesity-OA relationship. Control: Matching, stratified analysis, multivariate logistic regression.

---

Q8. Radiation Study / Rare Fatal Disease Associated with Smoking (JNM Kalyani / Bankura, 2+8+5=15)

• For rare diseases, cohort studies are impractical (need enormous sample sizes and long follow-up)
• For fatal diseases, by the time enough events occur in a cohort study, many subjects would have died, making follow-up impossible
• Case-control studies start with people who already have the rare/fatal disease → efficient
• Cost-effective, rapid
• Park states: "Case control studies are particularly suitable to investigate rare diseases or diseases about which little is known"

• Cases: persons with the rare fatal disease (e.g., mesothelioma, laryngeal cancer)
• Controls: persons without the disease, matched by age/sex
• Exposure: smoking history (pack-years, type of tobacco, duration)
• Calculate Odds Ratio

---

Q9. OCP and Breast Cancer — Causal Association Study (ESI Joka, 1+2+6+3+3=15)

• To establish causal association and temporal relationship (exposure precedes disease)
• Allows calculation of Relative Risk (direct measure of risk)
• Can measure dose-response relationship (duration, dosage of OCP use)
• Unlike case-control, cohort minimizes recall bias as exposure is documented prospectively
• Park: "Cohort studies provide the most direct method of estimating the risk of disease associated with a suspected exposure"

1. Selection of exposed cohort: Women aged 20-45 years currently using OCPs (no prior breast cancer diagnosis)
2. Selection of unexposed cohort: Age-matched women NOT using OCPs, from same community
3. Baseline assessment: Physical exam, clinical breast exam, mammography, family history, parity, age at menarche
4. Follow-up: Annual breast examination, mammography; track for breast cancer diagnosis (confirm by biopsy)
5. Duration: Minimum 10-15 years (long latency of breast cancer)
6. Outcome: New cases of breast cancer in both groups
7. Analysis: Calculate incidence rate, Relative Risk (RR), Attributable Risk

1. Directly estimates incidence and Relative Risk
2. Establishes temporal sequence (cause before effect)
3. Can study multiple outcomes from the same exposure (OCP → breast cancer, cervical cancer, thromboembolism)
4. Minimizes recall bias (exposure documented at baseline)
5. Allows dose-response analysis

1. Very expensive and time-consuming
2. Loss to follow-up (women may stop OCP, move, or be lost)
3. Not suitable for rare diseases/outcomes
4. Healthy cohort effect — those who continue OCP are healthier
5. Confounding (women on OCP may have other breast cancer risk factors — nulliparity, family history)

---

Q10. Screen Time and Mental Disorders — Most Appropriate Study (Sagore Dutta, 2+8+2+3=15)

• Exposure = screen time in early childhood → outcome = mental disorders in adolescence
• Long lag time between exposure and outcome requires prospective follow-up
• Allows direct measurement of temporal sequence
• Can measure dose-response (hours of screen time per day)

1. Define cohort: Children aged 2-5 years in selected schools/communities
2. Measure screen time at baseline (structured questionnaire to parents; validated screen time diaries)
3. Record confounders: family structure, parental education, socioeconomic status, physical activity
4. Follow-up annually until adolescence (10-16 years)
5. Outcome assessment: Standardized screening tools for mental disorders — SDQ (Strengths and Difficulties Questionnaire), CBCL (Child Behavior Checklist), or ICD-10/DSM-5 criteria for diagnosis
6. Analyze by screen time categories (<1 hr/day, 1-2 hr/day, >2 hr/day per WHO guidelines)
7. Calculate incidence rates and RR for each mental disorder category

1. Information bias (recall bias): Parents may not accurately recall or report screen time — use objective tools (device usage logs)
2. Confounding bias: Socioeconomic status, parental mental health, quality of parent-child interaction all confound the screen time-mental disorder relationship — control by multivariate analysis
3. Attrition bias: Loss to follow-up (families move, children change schools) — minimize by active follow-up, incentives, multiple contact points

• Use standardized, validated tools for both exposure and outcome measurement
• Blinded outcome assessors
• Stratification and multivariate regression to control confounders
• Multiple imputation for missing data

---

Q11. Epidemiology — Definition, Classification, Cohort Study Steps, Bias (MCK, 2+5+5+3=15)

• Distribution: Who gets the disease? (Person, place, time)
• Determinants: Why/how do they get it? (Risk factors)
• Specified populations: Groups, not just individuals
• Application: To control health problems (action-oriented)

EPIDEMIOLOGICAL STUDIES
├── OBSERVATIONAL (no intervention)
│   ├── DESCRIPTIVE
│   │   ├── Case Report / Case Series
│   │   ├── Cross-Sectional Study (Prevalence Study)
│   │   └── Ecological Study (Correlational Study)
│   └── ANALYTICAL
│       ├── Case-Control Study (Retrospective)
│       └── Cohort Study (Prospective / Longitudinal)
│           ├── Prospective Cohort
│           ├── Retrospective Cohort
│           └── Ambispective Cohort
└── EXPERIMENTAL (intervention)
    ├── Randomized Controlled Trial (RCT)
    │   ├── Clinical Trial
    │   └── Field Trial
    └── Community Trial (Quasi-experimental)

1. Selection bias: The exposed and unexposed cohorts may differ in baseline characteristics (confounding)
2. Loss to follow-up (attrition bias): Differential loss between exposed and unexposed groups distorts results — those who drop out may be sicker (or healthier)
3. Healthy worker effect: Exposed workers (in occupational studies) are healthier at baseline than general population
4. Information bias: Differential measurement of outcome in exposed vs. unexposed (observer bias); surveillance bias — exposed group more closely monitored → more outcomes detected
5. Confounding: Residual confounding from unmeasured variables

---

Q12. Compare Two Drugs for Hypertension (IQ City, 1+6+3=10)

1. Define the research question: Is Drug A superior/equivalent/non-inferior to Drug B in reducing BP in hypertensive patients?
2. Eligibility criteria: Include newly diagnosed or uncontrolled hypertension (SBP ≥ 140 or DBP ≥ 90 mmHg); Exclude: renal failure, secondary hypertension, pregnancy, other antihypertensives
3. Sample size calculation: Based on expected difference in mean BP reduction, SD, power (80%), α (0.05)
4. Randomization: Randomly allocate eligible consenting participants to Drug A or Drug B using computer-generated random number sequence; maintain allocation concealment (sealed envelopes or central randomization)
5. Double-blinding: Both participants and outcome assessors are blinded to which drug is received (to prevent performance bias and detection bias); identical-looking capsules
6. Administration: Administer Drug A to Group 1, Drug B to Group 2, for defined duration (e.g., 6-12 weeks)
7. Follow-up: BP measurement at baseline, 4 weeks, 8 weeks, 12 weeks; standardized measurement protocol (same time, same arm, calibrated sphygmomanometer)
8. Outcome Measures: Primary: Reduction in SBP/DBP; Secondary: adverse effects, tolerability, adherence, MACE (major adverse cardiovascular events)
9. Analysis: Intention-to-treat analysis; compare mean BP reduction using t-test/ANOVA; adverse events comparison

---

Q13. Cohort Study for Obesity and Hypertension (Deben Mahata GMC, 1+6+2+2+4=15)

• 6,000 adult males, 25-30 years
• 2,000 obese, 4,000 non-obese
• 200/2,000 obese developed hypertension
• 100/4,000 non-obese developed hypertension
• Follow-up: 15 years

1. Establishes temporal sequence (obesity precedes hypertension)
2. Direct measurement of incidence and Relative Risk

1. Long duration (15 years) → expensive, high attrition
2. Healthy worker/cohort effect possible

---

Q14. Alcohol and Dyslipidemia — Nested Case-Control Study (MJN Coochbehar, 10 marks)

• Recruit participants from a defined block in Cooch Behar district
• Cases: Adults diagnosed with dyslipidemia (LDL >130, total cholesterol >200 mg/dL, or on lipid-lowering drugs)
• Controls: Age- and sex-matched adults from same community without dyslipidemia
• Exposure: Alcohol consumption (AUDIT questionnaire — quantity/frequency/duration)
• Confounders: BMI, diet, physical activity, smoking, diabetes, family history
• Calculate Odds Ratio for alcohol consumption

• Recall bias: Past alcohol consumption may not be accurately recalled
• Social desirability bias: Underreporting of alcohol (stigmatized behavior) — use validated AUDIT tool; anonymous questionnaire
• Selection bias: Community-based controls may refuse participation if alcohol users (MNAR — Missing Not at Random)
• Confounding: Diet, smoking, BMI all confound alcohol-dyslipidemia relationship — control by matching and multivariate logistic regression

• The exposure data is collected prospectively (at cohort enrollment) → eliminates recall bias
• Controls are drawn from the same population that gave rise to the cases (eliminates selection bias)
• More efficient than a full cohort study for rare outcomes
• Can compute Odds Ratio which approximates Relative Risk

---

SHORT NOTES

---

"Bias and Confounding are NOT synonymous" (Calcutta National MC)

• A systematic error in study design, data collection, or analysis that leads to an incorrect estimate of the association between exposure and disease
• It is a flaw in the study — it cannot be adjusted for statistically once it has occurred
• Types: Selection bias, information bias (recall, observer), Berkesonian bias
• Example: If obese people remember their diet history better than non-obese (recall bias), the association between diet and obesity is distorted

• A third variable that is associated with both the exposure and the outcome, and distorts the apparent relationship between them
• A confounder is NOT a bias — it is a real phenomenon in the data
• It CAN be controlled — by matching (design stage), restriction, stratification, or multivariate analysis (analysis stage)
• Example: In studying smoking → lung cancer, age is a confounder (older people smoke more AND are more likely to get cancer)

---

"Cohort Study is gold standard for temporal association, but inappropriate for rare disease" (Sagore Dutta)

• In a cohort study, the exposure is documented BEFORE the disease occurs
• This establishes the temporal sequence (cause precedes effect) — the most important criterion for causal inference (Bradford Hill's criteria)
• Directly calculates Incidence Rate and Relative Risk
• Therefore it is the gold standard for establishing that an exposure causes a disease

• If a disease is rare (e.g., incidence = 1 in 10,000/year), you would need to follow hundreds of thousands of people for many years to observe enough cases
• This makes the study impractical, prohibitively expensive, and time-consuming
• Solution for rare diseases: Use case-control study — start with existing cases of the rare disease and look backward at exposure

---

Relative Risk (RR) Calculation — Silicosis Data (Diamond Harbour GMC&H)

1. Directly measures incidence and RR (true measure of risk)
2. Establishes temporal relationship (exposure before disease)

• Incidence in exposed = 120/800 = 0.150 = 15%
• Incidence in unexposed = 30/1200 = 0.025 = 2.5%
• RR = 15% / 2.5% = 6.0

---

"Randomization and Blinding are NOT used for the same purpose" (Deben Mahata GMC&H)

• Purpose: To ensure equal distribution of known and unknown confounders between study groups at baseline
• It eliminates selection bias and ensures comparability of groups before intervention
• Example: In a drug trial, randomization ensures both groups have similar age, sex, co-morbidities
• Randomization is done at the allocation/design stage

• Purpose: To prevent performance bias and detection/measurement bias during the study
• It ensures that knowledge of treatment assignment does not influence behavior of participants (performance bias) or assessment of outcomes (detection bias)
• Single-blind: Only participant is blinded; Double-blind: Both participant and assessor are blinded
• Blinding is done at the conduct/measurement stage

---

"RR and AR are Not Synonymous" (Midnapore)

• Ratio of incidence in exposed to incidence in unexposed
• RR = I₁ / I₀
• Measures strength of association (how much more risk does exposure add?)
• Used for etiological purposes — is exposure causally related to disease?
• Example: RR = 4 → exposed persons have 4 times the risk

• Difference in incidence between exposed and unexposed: AR = I₁ - I₀
• Measures the absolute excess risk due to exposure
• Relevant for public health action — if we eliminate the exposure, how many cases are prevented?
• Example: AR = 7.5% → 7.5 extra cases per 100 exposed persons are due to the exposure

• An exposure may have a high RR but low AR (if the disease is rare even in exposed) — e.g., RR = 10 but AR = 0.001% (rare disease)
• An exposure may have a moderate RR but very high AR (if the disease is common) — e.g., RR = 1.5 for smoking-cardiovascular disease, but AR is huge due to high baseline incidence

---

"Incidence is Preferred over Prevalence in Studying Disease Causation" (JNM Kalyani)

• Incidence: Number of NEW cases arising in a defined population during a specified period
• Prevalence: Total number of existing cases (old + new) at a point in time

1. Temporal clarity: Incidence measures NEW cases, ensuring that exposure precedes disease (establishing cause-effect temporality). With prevalence, we cannot tell if exposure occurred before or after disease.
2. Unaffected by disease duration: Prevalence = Incidence × Duration. A disease with long duration inflates prevalence without representing new risk. Incidence directly reflects the risk of getting the disease.
3. Direct measure of risk: Relative Risk (RR) = ratio of incidences. Cohort studies (which measure incidence) provide the most reliable measure of causation.
4. Prevalence is influenced by survival: Factors that affect survival (treatment, fatality rate) alter prevalence but not the true risk (incidence). This confounds any causal analysis using prevalence.

---

Sentinel Surveillance is Useful for Early Outbreak Detection (MCK, 4 marks)

• Involves selected "sentinel" sites (e.g., medical colleges, district hospitals, PHCs) strategically located
• Data collected on specific pre-identified diseases
• Regular, timely, and systematic reporting
• Sites are chosen to be representative of the larger population

1. Timeliness: Sentinel sites report regularly (weekly/monthly) → anomalies in case counts are quickly detected
2. Quality data: Trained staff at sentinel sites ensure better case ascertainment and laboratory confirmation
3. Early warning signal: Sudden increase in sentinel cases triggers alert and investigation before an outbreak becomes widespread
4. Specific diseases: Particularly useful for vaccine-preventable diseases (sentinel surveillance for AFP for polio), influenza-like illness, dengue, etc.
5. Resource efficient: Does not require universal reporting from all facilities — feasible even in resource-limited settings
6. Example in India: IDSP (Integrated Disease Surveillance Programme) uses sentinel sites for influenza (ISFN — Indian Sentinel Surveillance Network); Acute Respiratory Illness surveillance

---

Population Attributable Risk (PAR) (KPC Medical College)

• PAR = Incidence in total population (IT) - Incidence in unexposed (I₀)
• OR using Relative Risk: PAR = [P(RR-1)] / [P(RR-1) + 1] (Population Attributable Risk Fraction) where P = proportion exposed in the population

• PAR measures the public health impact of an exposure
• It estimates how many cases in the total population can be attributed to (and therefore prevented by eliminating) the exposure
• It helps prioritize public health interventions — high PAR exposures (even with moderate RR) are most important to target

• PAR fraction = [0.30(10-1)] / [0.30(10-1)+1] = 2.7/3.7 = 73%
• Meaning: 73% of all lung cancers in the population could be prevented if smoking were eliminated

---

"Monitoring and Surveillance are Not Synonymous" (Jhargram GMC&H)

---

---

Summary of Questions Answered

Long Answer Questions (15 marks each)

Short Notes

• Bias vs. Confounding (NOT synonymous)
• Cohort = gold standard for temporal but bad for rare disease
• RR calculation for silicosis data (RR = 6.0)
• Randomization vs. Blinding (NOT same purpose)
• RR vs. AR (NOT synonymous)
• Incidence preferred over prevalence for causation
• Sentinel surveillance for early outbreak detection
• Population Attributable Risk (PAR)
• Surveillance vs. Monitoring (NOT synonymous)