BIOSTATS FML #2
Card Set Information
BIOSTATS FML #2
What is meta-analysis?
Combine literature about a topic to arrive at a single summary estimate
Why is meta analysis used?
Realize and quantify undetected treatment effect
Steps in a meta analysis?
Clearly specify all databases
Identify all studies on the topic even unpublished data
Systematic and transparent process to select studies for inclusion
Use structured format to extract outcomes from study
Combine the data
Have explicit inclusion and exclusion criteria
What are the strengths of the meta analysis?
Quality of methods
Finding relationships across studies
Imposes process of summing up research findings
T/F The quality of the methods is central to the quality of the results?
Limitations of meta anaylsis?
If study is comparable based on individual opinion
Includes some less than optimal studies
Selection and publication bias
When to use a meta analysis?
Studies lack substantial power
Available results support different conclusions
More precise estimate of treatment effect
T/F - Do all hypothesis refer to the population?
What does a parameter describe?
Describes the characteristic of a population.
Always in GREEK letters
What does a statistic describe?
A chracteristic of the sample
Always in ROMAN letters
What are the 2 statistical methods?
What is descriptive statistics?
Collecting, presenting, describing data using formulas, graphs, surveys
What is inferential statistics?
Draw conclusions/make decisions about a population based on sample data only
What are the 4 levels of measurement?
Nominal - unordered categories no true meaning of value
Ordinal - rank ordered categories
Interval - no true zero point
Ratio - true zero point
What is a cohort study?
Can be prospective or retrospective design
Nonrandomized study grps are assigned
Observational - researchers only observe
Longitudinal - takes a long time to do
What are the 6 advantages of cohort study?
Study rare diseases
Avoid recall and survival biases
Assess multiple outcomes
Measure exposure over time
Measure outcome incidence
What are the disadvantages to cohort study?
Large sample for rare/slow disease
Loss to follow up
Exposure status changes over time
What is prospective cohort study?
Exposed at beginning of study follow up in the future
What is retrospective cohort study?
Exposed in the past (historical) outcome assessed as historical records
What are equal measurements?
Equal testing and examining in the same way
Same diagnostic tests
What are homogeneous grps?
How to calculate relative risk? pg.18
Relative risk = (a/a+c)/(b/b+d)
>1 = no association
<1 = exposed ppl have less risk
=1 = equal risk
checklist of what needs to be included in cohort studies. ex. intro methods...
What is case control?
Determine the association between risk factors and outcome of interest
Retrospective data collection by choosing subjects in present time already with outcome and checking for risk factors
What are the advantages to case-control studies?
Study rare diseases/ long development type things
Easy to perform and collect data
What are the disadvantages to case control studies?
Correlation no causation
No time link of outcome and risk
Can't calculate outcome incidence
Inefficient for rare risk factors
Hard to select representatives for cases and controls
What bias occurs in case-control studies?
Recall bias - based off of memory
What is a randomized controll trial?
Random assignment of participants into experimental grp and control grp
Only difference between the grps is the outcome variable studied
T/F Are all clinical trials interrelated?
What is the format for how information for a trial should be?
IMRAD - intro, methods, results, and discussion
What are the advantages of parallel trials?
Robust for most conditions
Fits most experimental situations
What are the disadvantages of parallel trials?
Only one treatment
Need to match subjects for
What are the 4 types of parallel tests?
3 grp parallel
2 part parallel
Multiple dose parallel
What are characteristics of crossover trials compared to parallel trials?
Each subject gets all treatments
Subject is its own control
Less data varability
What are the 3 order effects?
Carry-over = drug effect lasts from treatment period to treatment period, reduced with washout period
Period = change in pt condition between treatment
What is an N-of-1 trial?
Controlled cross-over study of 1 person comparing their treatment effects to their placebo effects to see if effective
Can't apply results to population only to pt
How many phases in a random control trial?
What happens in phase 1?
Healthy human volunteers
Assess PK, toxicity, pharm effects
What happens in phase 2?
Safe and effective
Have disease condion of interest
What happends in phase 3?
Safe and effective
Large grp 1000's
Have disesase and comorbid conditions
What happens in phase 4?
Post market studies
Use, effects, and safety in real world conditions
What are cross sectional studies?
Conducted at single or more points in time
Simultaneous selecton, assignment, and asessment of subject
What are the 2 types of cross sectional studies?
Descriptive - have or no have, point prevalence (1 point in time) or period prevalence
Analytical - compare
What are case reports?
More knowledge of treatment/condition
Safer treatment practices
Start new research
What are case series?
Grp of single cases in one report
Subjects have common characteristic
Quick and cheap way to report info
Start new research
What are the limitations of case series?
Can't determine causal inferences
No control grps
No circumstantial evidence
Fail to compensate for influences of external forces
What are the 3 types of review articles?
What is a narrative review article?
Editorial type - short and narrow focused
Commentary type - biased, express a particular opinion
Narrative Overview - author's summary of literature (bias)
What is systemic articles better than (narrative articles)?
Focused questions (broad)
Comprehensive lit search (not specified)
Criterion based selection (not specified)
Rigorous appraisal (variable)
EBM inferences (sometimes)
Why is meta analysis good?
Combine independent results of trials using statistical analysis to draw conclusions about therapeutic effectiveness or plan new studies
Small samples used in original trials
Large trial too expensive to perform
Solves disagreements on magnitude or direction of effect
List the tests in order that provide the most to least evidence?