15.2 Observational & Experimental Epidemiological Study Designs

Observational & Experimental Epidemiological Study Designs

Epidemiological studies are broadly classified into Observational (where the investigator observes natural exposures without intervention) and Experimental (where the investigator actively manipulates exposure status via randomization).

+---------------------------------------------------------------------------------------------------+
|                                 EPIDEMIOLOGICAL STUDY DESIGNS                                     |
+---------------------------------------------+-----------------------------------------------------+ 
|            OBSERVATIONAL STUDIES            |                 EXPERIMENTAL STUDIES                |
+----------------------+----------------------+-----------------------------------------------------+
|     Descriptive      |      Analytical      |            Randomized Controlled Trials (RCT)       |
+----------------------+----------------------+-----------------------------------------------------+
| • Case Report        | • Case-Control       | • Clinical Trials (Patients)                        |
| • Case Series        | • Cohort Study       | • Field Trials (Healthy Individuals)                |
| • Cross-Sectional    |                      | • Community Trials (Communities/Groups)             |
| • Ecological Study   |                      |                                                     |
+----------------------+----------------------+-----------------------------------------------------+

---

1. Descriptive Studies & Cross-Sectional Design

Cross-Sectional Study (Prevalence Study)

In a cross-sectional study, exposure status and disease outcome are measured simultaneously in a defined population at a single point in time ("snapshot").

• Advantages: Fast, inexpensive, excellent for assessing public health burden and disease prevalence.
• Disadvantages: Cannot establish temporal sequence (whether exposure preceded outcome); subject to survivorship bias (selects for cases of long duration).
• Key Formula: $P = I \times D$ (Prevalence = Incidence $\times$ Mean Duration of disease).

Ecological Study (Correlational Study)

The unit of observation in an ecological study is a group or population (e.g., countries, states) rather than individual human beings.

• Ecological Fallacy: The logical error of inferring that an association observed between variables at the aggregate group level necessarily holds true for individuals within those groups.

---

2. Case-Control Studies (Trohoc / Retrospective Design)

A Case-Control study begins with diseased individuals (Cases) and non-diseased individuals (Controls), looking backward in time to compare the frequency of past exposure to potential risk factors.

               Backward Direction of Inquiry (Effect -> Cause)
                      <===================================
TIME:               Past Exposure                      Present Status
                +-------------------+               +-------------------+
                | Exposed (a)       | <------------ | CASES (Disease +) |
                | Unexposed (c)     |               +-------------------+
                +-------------------+               +-------------------+
                | Exposed (b)       | <------------ | CONTROLS (Disease-|
                | Unexposed (d)     |               +-------------------+
                +-------------------+

2x2 Contingency Table for Case-Control Studies

Measure of Association: Odds Ratio (OR)

Because case-control studies select participants based on disease status (fixing the number of cases and controls), incidence rates cannot be directly calculated. The measure of association is the Odds Ratio (OR), also called the Cross-Product Ratio:

$\text{Odds of exposure among Cases} = \frac{a}{c}$ $\text{Odds of exposure among Controls} = \frac{b}{d}$ $\text{Odds Ratio (OR)} = \frac{a/c}{b/d} = \frac{a \times d}{b \times c}$

• Interpretation:
  • $OR = 1.0$: Exposure is not associated with disease.
  • $OR > 1.0$: Positive association (Exposure is a potential risk factor).
  • $OR < 1.0$: Negative association (Exposure is a protective factor).

Matching & Overmatching

• Matching: Process of selecting controls so that they are similar to cases in terms of specific confounding variables (e.g., age, sex, social class).
• Overmatching: Occurs when matching is performed on variables that are closely tied to the exposure of interest, or when matching is done on intermediate variables in the causal pathway. Overmatching reduces study efficiency and obscures true exposure-disease relationships.

Major Biases in Case-Control Studies

1. Recall Bias: Cases (having suffered a disease) remember past exposures more vividly or differentially than healthy controls.
2. Selection Bias: Non-representative selection of cases or controls (e.g., Berkson's Bias / admission rate bias when controls are selected from hospitalized populations).

---

3. Cohort Studies (Prospective / Follow-Up / Longitudinal Design)

A Cohort study begins with a group of exposed and unexposed disease-free individuals and follows them forward over time to measure and compare the incidence of disease development.

               Forward Direction of Inquiry (Cause -> Effect)
                      ===================================>
TIME:               Present Status                     Future Outcome
                +-------------------+               +-------------------+
                | EXPOSED GROUP     | ------------> | Disease + (a)     |
                | (Disease-Free)    |               | Disease - (b)     |
                +-------------------+               +-------------------+
                | UNEXPOSED GROUP   | ------------> | Disease + (c)     |
                | (Disease-Free)    |               | Disease - (d)     |
                +-------------------+               +-------------------+

2x2 Contingency Table for Cohort Studies

Key Quantitative Metrics in Cohort Studies

1. Incidence Rates:

   • Incidence in Exposed ($I_e$) = $\frac{a}{a+b}$
   • Incidence in Unexposed ($I_u$) = $\frac{c}{c+d}$
   • Total Population Incidence ($I_t$) = $\frac{a+c}{a+b+c+d}$

2. Relative Risk (RR) / Risk Ratio: $RR = \frac{I_e}{I_u} = \frac{a / (a+b)}{c / (c+d)}$

   • Direct measure of the strength of causal association between exposure and disease.

3. Attributable Risk (AR) / Etiological Fraction: $AR = \frac{I_e - I_u}{I_e} \times 100$

   • Quantifies the proportion of disease in the exposed population that can be attributed directly to the exposure (and eliminated if exposure is removed).

4. Population Attributable Risk (PAR): $PAR = \frac{I_t - I_u}{I_t} \times 100$

   • Quantifies the proportion of disease incidence in the entire target population attributable to the risk factor. Crucial for prioritizing public health interventions.

---

4. Head-to-Head Comparison: Case-Control vs. Cohort Study

---

5. Experimental Studies & Randomized Controlled Trials (RCT)

An RCT is an experimental design where human participants are randomly assigned to receive an intervention (experimental arm) or standard treatment/placebo (control arm).

+---------------------------------------------------------------------------------------------------+
|                                 STAGES OF A CLASSIC PARALLEL RCT                                  |
+---------------------------------------------------------------------------------------------------+
| Target Population  ---> Study Sample (Informed Consent) ---> RANDOMIZATION                        |
|                                                                 |                                 |
|               +-------------------------------------------------+---------------------------------+
|               |                                                                                  |
|               v                                                                                  v
|   [ Experimental Group ]                                                             [ Control Group ]
|   (Receives New Drug / Intervention)                                                 (Receives Placebo / Standard)
|               |                                                                                  |
|               v                                                                                  v
|     Follow-Up Assessment                                                               Follow-Up Assessment
|               |                                                                                  |
|               v                                                                                  v
|   Outcome: Improved / Unimproved                                                    Outcome: Improved / Unimproved
+---------------------------------------------------------------------------------------------------+

Key Methodological Elements of RCTs

1. Randomization: The central mechanism that allocates subjects to treatment arms purely by chance. Its vital function is to eliminate selection bias and ensure equal, balanced distribution of both known and unknown confounding variables between arms.
2. Blinding (Masking):
   • Single-blind: Subject is unaware of treatment assignment.
   • Double-blind: Subject and evaluating clinician/investigator are unaware. Eliminates observer/investigator bias.
   • Triple-blind: Subject, investigator, and biostatistician analyzing data are all unaware.
3. Crossover Design: Subjects receive Treatment A, undergo a washout period (to eliminate residual pharmacological effect), and then switch to receive Treatment B. Each subject acts as their own control, controlling for inter-individual variability.