Study design: Overview
- ANALYTIC: attempts to quantify the effect of an exposure on an outcome.
- Experimental: the exposure is manipulated, prospective.
- Observational analytic: an exposure and its effect on the outcome are measured, can be retrospective or prospective.
- SYSTEMATIC REVIEW: examines existing evidence from multiple sources to answer a specific research question.
- Systematic review: rigorous identification and examination of published studies with goal of summarizing best available research.
- Meta-analysis: builds on systematic review, adding a statistical re-analysis of pooled data from all applicable valid studies to obtain a summary estimate of the effect of an intervention on an outcome.
- DESCRIPTIVE (non-analytic): attempts to describe what is happening. It does not quantify any relationships.
Analytic Studies: Design Characteristics
Analytic studies: quantifies the effect of an exposure on the outcome
|
Design of study |
Description |
Pros |
Cons |
Typical outcomes |
Type of question |
Examples |
|---|---|---|---|---|---|---|
|
Experimental: the exposure is manipulated |
||||||
|
RCT - parallel - crossover |
Prospective, groups randomized to exposure |
Minimize bias and confounding |
Expensive, time consuming, need for follow-up, ethics |
Odds ratio Relative risk |
Therapy |
Dabigatran versus Warfarin, early goal directed therapy |
|
Non-randomized controlled trial |
Prospective, groups not randomized |
Detects association between intervention and outcome |
No randomization, lack of control of selective bias, additional cons as above |
Odds ratio Relative risk |
Therapy |
|
|
Observational: measures an exposure and its effect on the outcome |
||||||
|
Cohort |
Prospective or retrospective, examines associations of exposures and outcomes over time |
Stronger evidence for causality, yields incidence, relative risk |
Confounding, time consuming, need for follow-up, condition must be sufficiently prevalent |
Incidence Relative risk Hazards ratio |
Prognosis Etiology Therapy |
Framingham Heart Study |
|
Cross-sectional |
Surveys sample population at one point in time |
Determines prevalence and associations, analysis of diagnostic tests |
Single point in time, unable to determine causality, bias, confounding |
Prevalence |
Diagnosis Etiology |
NHANES (National Health and Nutrition Examination Survey) |
|
Case-control |
Compares group with and without disease |
Cheap, efficient, rare diseases |
Bias, chance, confounding |
Odds ratio |
Therapy Prognosis Harm Etiology |
Food-borne outbreak investigation, diethylstilbestrol and vaginal cancer
|
Systematic Review: Design Characteristics
Systematic review: Examines existing evidence from multiple sources
|
Design of study |
Description |
Pros |
Cons |
Type of question |
Examples |
|---|---|---|---|---|---|
|
Systematic review |
Rigorous examination of existing data from multiple sources using rigorous criteria and search strategy |
Cheap |
Limited longevity of relevance depending on quality of studies |
Diagnosis Therapy Prognosis |
Cochrane reviews |
|
Meta-analysis |
Statistical analysis of pooled data found in studies obtained through rigorous criteria and search strategy |
Produce summary statistics, cheap |
Depends on quality of included data (“garbage in, garbage out”) |
Diagnosis Therapy Prognosis |
Safety/efficacy of bare metal vs drug eluting stents |
|
Guidelines |
Combination of systemic reviews, meta-analyses, and expert consensus |
Expert analysis |
May be influenced by sponsoring organization (i.e. government analysis usually more cost-conscious versus societies more aggressive with testing/treatment)
|
Diagnosis Therapy Prognosis Harm |
US Preventative Services Taskforce (USPSTF) |
Descriptive Studies: Design Characteristics
Descriptive studies: describes what is happening, does not quantify any relationships
|
Design of study |
Description |
Pros |
Cons |
|---|---|---|---|
|
Case series or report |
Report on treatment or outcome of individual patient(s) |
Describes unique case |
Little statistical validity |
|
Survey |
Assess opinions, experiences |
Cheap, far-reaching |
|
|
Qualitative |
Explore, describe, and explain-health related phenomena |
Gives voice, adds human experience |