Lecture 3 Flashcards
3 main paper designs
- Primary (empirical studies) Observe, collect, and record data.
- Secondary (integrative studies) Examine data collected by someone else.
- Quantitative studies. Broad fields of research
What is a primary, empirical design study? Examples
Observe–> collect data –> Record data
Lab, clinical trials, surveys, organizational case studies.
What is a secondary, integrative design study? examples
Examine data collected by someone else.
Ex: Non systematic reviews, systematic reviews, meta-analysis, guidelines, databases, economic analysis
Non systematic review (secondary study design)
Most common. Summarizing primary studies.
Systematic reviews. (secondary study design)
Apply rigorous, transparent, and auditable methods. Can be done with or without meta-analysis.
Meta-analysis (secondary study design)
Integrates numerical data from multiple studies
Guidelines (secondary study design)
Draw conclusions from primary studies about clinical behavior
Databases (secondary study design)
Evaluate existing databases.
Economic analysis (secondary study design)
Determine course of action. Gov usually
6 types of quantitative studies
Therapy Diagnosis Screening Prognosis Causation Psychometric studies
Best study design for a therapy quantitative study
Randomized controlled trial.
Best study design for a diagnosis quantitative study
Cross sectional
Best study design for a screening quantitative study
Cross sectional
Best study design for a prognosis quantitative study
Longitudinal
Best study design for a causation quantitative study
Cohort, case control
Best study design for a psychometric quantitative study
Liekert scale
Case report (1) or case series (multiple)
- What design
- Strong or weak evidence?
- Other characteristics
Retrospective (what happened)
Understandable, straight forward, weak evidence
Written about treatment, adverse rxn
Rapid way to alert clinicians
Case control
- What design?
- Great for?
- Other
Usually retrospective (did previous risk factors influence patients with disease?)
Great for rare conditional. Looks at associations.
May have bias.
Subjects are matched in either “disease present” or “disease absent” groups. Each group is studied to see if they had a risk factor present or absent.
Cohort studies.
- What design?
- Longitudinal or cross sectional
- What is the endpoint?
Usually prospective. Enroll patient and follow into future.
CAN Be retro.
Longitudinal. Will follow patients from years to decades.
Endpoint is prevalence.
Inception cohort
Looks at prognosis. Enrolled automatically. Could be at the beginning of disease, or birth (millennials)
Cross sectional survey
- Measures what?
- What surveys are used?
- What specific clinical question is asked?
- Measures 1 point in time with 1 population.
- Surveys include questionnaire, interviews, examinations, retrospective records.
- Prevalence. How many people have disease at 1 time?
Randomized controlled trials. What are the advantages?
Gold standard Random assignment Reduced bias Facilitates meta-analysis Evaluates single variable Hypotheticodeductive reasoning
Randomized controlled trials. What are the disadvantages?
Expensive
Requires large number of subjects
research agendas are dictated
Investigators are not blinded and it may have imperfect randomization (hidden biases)
Hypotheticodeductive reasoning
Karl popper.
Cannot prove something is true or false, but you can identify support for your hypothesis.
Meta-Analysis review
what is it?
pros?
Limitations?
Analysis that combines results of multiple studies. Higher statistical power due to large sample number.
A “study of studies”
Limitations: variability of study design and study selection bias
Two examples of psychometric studies
Questionnair: Quantifies options, characteristics.
Often has words. Researcher uses numerical value.
Leikert scale: Widely used. Subject selects words or numbers on a scale. Never, sometimes, half time, most time, always.
Leikert scale psychometric study
Widely used. Subject selects words or numbers on a scale. Never, sometimes, half time, most time, always.
Questionnair psychometric study
Quantifies options, characteristics.
Often has words. Researcher uses numerical value.
Are there alternatives to RCT- other controlled clinical trials ?
No. Impossible and impractical and unethical.
