Study Designs Cheatsheet

Cheatsheet Content

### Observational Studies Studies where researchers observe subjects and measure variables without intervening. #### Types of Observational Studies #### Cohort Studies - **Definition:** Follows a group (cohort) over time to see who develops a disease/outcome. - **Prospective:** Exposure measured at baseline, then followed into future. - **Retrospective:** Exposure measured from past records, then followed to present/future. - **Advantages:** - Can establish incidence and natural history of disease. - Can examine multiple outcomes for a single exposure. - Good for rare exposures. - **Disadvantages:** - Expensive and time-consuming, especially prospective. - Loss to follow-up can bias results. - Not suitable for rare diseases. - Prone to confounding. #### Case-Control Studies - **Definition:** Compares a group with a disease/outcome (cases) to a similar group without the disease/outcome (controls), looking retrospectively for differences in exposure. - **Advantages:** - Efficient for rare diseases. - Less time-consuming and expensive than cohort studies. - Can examine multiple exposures for a single outcome. - **Disadvantages:** - Prone to recall bias (cases may remember exposures differently). - Difficult to establish temporal sequence (cause and effect). - Selection of controls can be challenging. - Cannot directly calculate incidence or prevalence. #### Cross-Sectional Studies - **Definition:** Measures exposure and outcome simultaneously at a single point in time. - **Advantages:** - Quick and inexpensive. - Can estimate prevalence of exposure and outcome. - Useful for hypothesis generation. - **Disadvantages:** - Cannot establish temporality (cause and effect). - Cannot measure incidence. - Prone to selection bias. ### Randomized Controlled Trials (RCTs) Experimental studies where participants are randomly assigned to an intervention group or a control group. - **Definition:** Gold standard for establishing causality. Participants are randomly allocated to receive an intervention (e.g., new drug) or a control (e.g., placebo, standard treatment). - **Key Features:** - **Randomization:** Minimizes confounding by distributing unknown and known confounders evenly between groups. - **Blinding:** - **Single-blind:** Participant unaware of assignment. - **Double-blind:** Participant and researchers/assessors unaware. - **Triple-blind:** Participant, researchers/assessors, and data analysts unaware. Reduces bias. - **Control Group:** Provides a baseline for comparison. - **Advantages:** - Strongest evidence for causality due to randomization. - Minimizes bias and confounding. - Can measure incidence and efficacy. - **Disadvantages:** - Expensive and time-consuming. - Ethical concerns (e.g., withholding treatment from control group). - May not be generalizable to real-world settings (highly controlled environment). - Not suitable for rare outcomes or long-term effects. - Can be practically impossible or unethical for certain exposures (e.g., smoking). ### Study Design Comparison | Feature | Observational Studies | Randomized Controlled Trials (RCTs) | | :----------------- | :-------------------------------------------------- | :-------------------------------------------- | | **Intervention** | No researcher intervention; observe natural course | Researcher intervenes (assigns treatment) | | **Randomization** | No | Yes | | **Causality** | Cannot definitively establish (association only) | Strongest evidence for causality | | **Bias/Confounding** | High risk (especially confounding) | Minimized by randomization | | **Cost/Time** | Generally less expensive and time-consuming (except prospective cohort) | More expensive and time-consuming | | **Ethics** | Fewer ethical issues related to intervention | Significant ethical considerations | | **Generalizability** | Often high (real-world settings) | Can be lower (highly controlled settings) | | **Examples** | Cohort, Case-Control, Cross-Sectional | Clinical trials, intervention studies |