Public Health & Epidemiology

# Public Health & Epidemiology Knowledge Pack

## Core Epidemiological Concepts

### Disease Frequency Measures
Understanding disease burden requires precise measurement:

**Prevalence**: Proportion of a population with a condition at a specific point (point prevalence) or during a period (period prevalence). Formula: existing cases / population at risk. Prevalence is influenced by both incidence and disease duration — chronic conditions (diabetes, HIV on Tx) have high prevalence relative to incidence because patients live for years with the condition.

**Incidence rate (person-time)**: New cases per person-time at risk. Denominator accounts for varying follow-up times — someone followed for 6 months contributes 0.5 person-years. Expressed as cases per 1,000 person-years (or 100,000 person-years for rare diseases). Essential for etiologic studies because it measures the rate at which new disease occurs.

**Cumulative incidence (risk)**: Proportion of at-risk population developing disease over a defined period. Requires a closed cohort (no one enters/leaves during observation) or statistical adjustment for losses. Example: 5-year cumulative incidence of 8% means 8 out of 100 disease-free people developed the condition over 5 years.

**Attack rate**: Cumulative incidence during a defined outbreak period. Used in outbreak investigations — the proportion of exposed people who became ill. Comparing attack rates between exposed and unexposed groups identifies the source.

### Measures of Association
**RR (relative risk)**: Incidence in exposed / incidence in unexposed. RR = 2.0 means exposed individuals are twice as likely to develop the disease. Only calculable from cohort studies or RCTs where you can directly measure incidence in both groups.

**OR (odds ratio)**: (Odds of exposure among cases) / (Odds of exposure among controls). Used in case-control studies because you cannot calculate incidence (you start with cases and controls, not exposed and unexposed cohorts). OR approximates RR when the disease is rare (<10% incidence) — the rare disease assumption.

**HR (hazard ratio)**: From survival analysis (Cox proportional hazards model). Accounts for time-to-event data and censoring. HR = 1.5 means the exposed group has a 50% higher instantaneous rate of the event at any given time, assuming proportional hazards hold.

**Attributable risk (AR)**: Incidence in exposed minus incidence in unexposed. Represents the excess risk due to exposure. Population attributable risk (PAR) extends this to the population level: what proportion of disease in the population is attributable to the exposure, accounting for exposure prevalence. PAR = prevalence of exposure × (RR - 1) / [prevalence of exposure × (RR - 1) + 1].

### Study Designs
**RCT**: Gold standard for establishing causation. Random allocation eliminates confounding (known and unknown). Key features: allocation concealment (prevent selection bias), blinding (single, double, or triple-blind to prevent information bias), intention-to-treat analysis (analyze by assigned group regardless of compliance). Limitations: expense, ethical constraints (cannot randomize harmful exposures), volunteer bias, limited generalizability.

**Cohort study**: Follow exposed and unexposed groups forward in time. Prospective cohorts collect data as events occur (expensive but high-quality data). Retrospective cohorts use existing records to reconstruct cohorts (faster, cheaper, but limited to available data). Strengths: directly measures incidence and RR, can study multiple outcomes from one exposure. Weakness: inefficient for rare diseases.

**Case-control study**: Start with cases (disease) and controls (no disease), look backward at exposures. Efficient for rare diseases — the only practical design when disease incidence is 1 per 100,000. Calculate OR as the measure of association. Selection of controls is critical — they must represent the population that generated the cases. Hospital-based controls may have different exposure patterns than population-based controls.

**Cross-sectional study**: Measure exposure and outcome simultaneously. Cannot establish temporal sequence (did exposure precede disease?). Useful for prevalence estimation and hypothesis generation. Neyman bias: prevalent cases over-represent diseases with long duration and under-represent rapidly fatal conditions.

## Biostatistics for Public Health

### Screening and Diagnostic Tests
**Se (sensitivity)**: Probability the test is positive given disease is present = true positives / (true positives + false negatives). High Se means few missed cases — important for serious conditions where missing a case is dangerous. Rule of thumb: a sensitive test, when negative, rules OUT disease (SnNout).

**Sp (specificity)**: Probability the test is negative given no disease = true negatives / (true negatives + false positives). High Sp means few false alarms — important when false positives cause harm (unnecessary surgery, psychological distress, toxic Tx). Rule: a specific test, when positive, rules IN disease (SpPin).

**PPV and NPV**: PPV = probability of disease given positive test. NPV = probability of no disease given negative test. Critically, PPV and NPV depend on prevalence. A test with 99% Se and 99% Sp has PPV of only 50% when prevalence is 1% (half of positive tests are false positives). This is why mass screening of low-prevalence conditions generates many false positives — screen only populations with sufficient pre-test probability.

**NNT**: 1 / absolute risk reduction. If Tx reduces mortality from 8% to 5% (ARR = 3%), NNT = 1/0.03 = 34. You need to treat 34 patients for one additional patient to benefit. NNH = 1 / absolute risk increase from adverse effects. Compare NNT and NNH to assess risk-benefit: NNT of 34 with NNH of 500 is favorable; NNT of 34 with NNH of 40 requires careful consideration.

## Disease Surveillance

### Surveillance Systems
**Passive surveillance**: Healthcare providers and laboratories report notifiable conditions to public health authorities. Advantages: low cost, broad coverage. Limitations: under-reporting (estimated 10-50% of cases go unreported for most conditions), reporting delays, incomplete data. The CDC's National Notifiable Diseases Surveillance System (NNDSS) collects data on ~120 conditions.

**Active surveillance**: Public health officials regularly contact healthcare facilities, laboratories, and other sources to identify cases. Higher completeness than passive but far more resource-intensive. Used for high-priority conditions (polio eradication, Ebola response) or in defined geographic areas.

**Syndromic surveillance**: Monitors pre-Dx health indicators — emergency department chief complaints, OTC medication sales, school absenteeism, 911 call volumes — to detect outbreaks before confirmed Dx. Advantage: timeliness (captures signal days before laboratory confirmation). Limitation: low Sp — many non-outbreak events trigger alerts (Monday effect on ED visits, seasonal allergy medication purchases).

**Sentinel surveillance**: Selected reporting sites provide high-quality data on a specific condition. WHO's Global Influenza Surveillance and Response System uses ~150 national influenza centers worldwide to monitor circulating strains and inform annual vaccine composition decisions.

### Outbreak Investigation
The classic 10-step approach:

1. **Confirm the outbreak**: Is the observed number of cases above the expected baseline? Compare to historical data for the same time period. Account for reporting artifacts (new test available, changed case definition, media-driven testing increase).

2. **Define the case**: Clinical case definition specifies person, place, time, and clinical criteria. Confirmed (laboratory-verified), probable (clinical criteria met without lab confirmation), suspected (some criteria met). Cast a wide net initially, narrow later.

3. **Count cases and characterize by person, place, time**: The epi curve (cases plotted by onset date) reveals the epidemic pattern. Point source: single peak, cases cluster within one incubation period. Continuous common source: prolonged plateau. Propagated (person-to-person): successive peaks separated by one incubation period.

4. **Generate hypotheses**: Based on descriptive epi (who is getting sick, where, when), clinical features (incubation period narrows the pathogen list), and environmental assessment (inspect food facilities, water systems). Interview cases with standardized questionnaire about exposures during the relevant exposure window.

5. **Test hypotheses**: Analytical study — typically case-control during outbreaks (cases are already identified, recruit controls from the same population). Calculate OR for each suspected exposure. Dose-response relationship strengthens causal inference (those who ate more of the implicated food had higher attack rates).

6. **Implement control measures**: Don't wait for definitive results when the evidence is strong enough and the disease is serious. Remove contaminated product, close the exposure source, isolate cases, prophylax contacts, issue public communications.

## Infectious Disease Dynamics

### Transmission and R0
R0 is the average number of secondary cases generated by one primary case in a fully susceptible population. R0 > 1: epidemic grows. R0 < 1: epidemic dies out. R0 = 1: endemic steady state.

R0 values for reference: measles 12-18 (highest known for respiratory pathogens), chickenpox 10-12, COVID-19 original strain 2.5-3.5, influenza 1.5-2.0, Ebola 1.5-2.5. R0 is NOT a fixed biological constant — it depends on contact patterns, population density, and environmental conditions.

Rt (effective reproduction number) accounts for immunity in the population. Rt = R0 × (proportion susceptible). Herd immunity threshold = 1 - 1/R0. For measles (R0 = 15), herd immunity requires 93% immunity — explaining why even small drops in vaccination coverage trigger outbreaks.

### Infection Control
**Chain of infection**: Pathogen → Reservoir → Portal of exit → Mode of transmission → Portal of entry → Susceptible host. Breaking any link interrupts transmission.

**Standard precautions**: Applied to ALL patient care regardless of Dx. Hand hygiene (single most effective measure — WHO's 5 moments: before patient contact, before aseptic procedure, after body fluid exposure, after patient contact, after touching patient surroundings), PPE based on anticipated exposure, safe injection practices, respiratory hygiene.

**Transmission-based precautions**: Contact (MRSA, C. difficile — gown and gloves, dedicated equipment), Droplet (influenza, meningococcus — surgical mask within 6 feet), Airborne (TB, measles, varicella — N95 respirator, negative-pressure room, minimum 6 air changes per hour in existing facilities, 12 in new construction).

**Healthcare-associated infections (HAI)**: Central line-associated bloodstream infections (CLABSI), catheter-associated urinary tract infections (CAUTI), surgical site infections (SSI), ventilator-associated pneumonia (VAP), and C. difficile infections. Evidence-based bundles reduce HAI rates by 50-70%: CLABSI bundle (hand hygiene, maximal barrier precautions, chlorhexidine skin prep, optimal site selection — avoid femoral, daily line necessity review). CDC estimates 1 in 31 hospital patients has at least one HAI on any given day.

## Global Health

### Health Systems Frameworks
WHO's six building blocks: service delivery, health workforce, health information systems, access to essential medicines, financing, and leadership/governance. Strengthening primary healthcare is the most cost-effective investment — every $1 spent on primary care in low-income countries returns $3-11 in reduced hospitalizations, improved productivity, and economic growth.

### IHR and Pandemic Preparedness
The IHR (2005) requires 196 countries to detect, assess, report, and respond to public health events of international concern. Core capacities include surveillance, laboratory, rapid response teams, risk communication, and points of entry (ports, airports, ground crossings).

Joint External Evaluation (JEE) assesses country preparedness across 19 technical areas. Global Health Security Index scores reveal that most countries — including high-income nations — have significant gaps in preparedness. The COVID-19 pandemic demonstrated that high index scores did not reliably predict performance, highlighting the importance of political will, public trust, and social infrastructure alongside technical capacity.