20th Sept Flashcards
What is PECOS?
Patient Exposure Comparison Outcome Study design
3 options for locating studies with specific designs?
Free text searching using appropriate search terms
Using controlled vocabulary PECOS search terms
Using methodological PECOS search filters
What are free text terms?
Study design classification eg audit, service evaluation, research
Design nomenclature eg cross-sectional, case control, cohort
Design components eg prospective, retrospective, random, matched, double blind
How do you use controlled vocabularies?
Allows search for studies whose PECOS has been classified
Medline has controlled vocab thesaurus called MeSH (Medical Subject Heading)
How do you use methodological PECOS search filters?
combination of free text and controlled text search terms to create specialist methodological filters
MedLine uses these in ‘Publication Type’ and ‘Clinical Queries’ filters
What are the 3 competing priorities of clinical studies?
Provide correct answer (addressing potential biases)
Maximised efficient use of resources
Comply with ethical, legal and professional regulations
How do we conduct a study?
1) Identify problem and review previous studies
2) Formulate a question to address a gap in evidence
3) select a study design
4) sample/ select participants/ data sources
5) collect data/ measure variables
6) analyse and interpret analysis findings
7) disseminate results
What are the potential sources of bias with sampling/ selection?
external validity: non-representative samples
confounding: selection influences exposure and outcome
What are the potential sources of bias with measurement?
Information bias: extent of info varies among participants
Observer bias: influenced by the researcher’s beliefs
Recall/ response/ prestige bias: influenced by prior knowledge of belief
What are the potential sources of bias with analysis?
Loss to follow up: specific participants excluded
Omitted variable bias: imprecise adjustment for confounding
Attributional bias: interpretation as causality
What are the potential sources of bias with dissemination?
Publication bias: eventful results more likely to be published
What are the analytical study designs that have been developed to address issue of attributional bias?
Cross sectional: provides evidence of association within a sample
Case-control: provides evidence of association between samples
Cohort: provides evidence of directionality of associations
Trial: provides evidence of causality
Meta-analysis: provides evidence of reproducibility
^^ this is order of hierarchy from bottom to top
What are the 2 approaches to study design?
Descriptive (inductive) and analytical (deductive)
What are the types of study design within analytical?
Observational (selective sampling)
Experimental (selective exposure)
What are the 6 types of primary study design and what do they come under?
Descriptive: case study, cross sectional
Observational: cross sectional, case control, cohort
Experimental: trial
