Research Flashcards
3 pillars of EBP
best scientific evidence
clinical experience
patient values
5 orders for appraisal
- ask question
- search evidence
- critically appraise (pros and cons of each article)
- implement - determine clinical relevance
- evaluate - during clinical application of intervention
levels of evidence
- meta-analysis
- systematic review
- RCT
- cohort
- case control
- cross sectional
- case series/case reports
meta-analysis
multiple articles with statistical analysis
systematic review
gives summary of findings from many articles
less statistics/no statistics
RCT
control group doesnt get intervention
experimental group gets intervention
cohort studies
-observational study that compares COHORT who share COMMON CHARATERISTIC; with and without the exposure
- type of LONGITUDINAL study
- ex framingham study - ex residents of framingham is whats in common, and follows to see heart data
- prospective: over period of time in future
- retrospective: use info thats already happened in past
case control studies
Compare group of individuals with specific condition with group of people without same condition
- observational
- ex group of older women: age matched 15 with chronic LBP and 15 w/o chronic LBP to see if it affects pain
cross sectional studies
- observation study
- data is collected from ppl at single point in time (snapshot) of info at specific moment
- ex look at balance 4 wks after surgery
longitudinal studies
- observational study
- repeated observations or measurementsof same subjects across extended time periods to understand patterns and factors influencing those changes
case series/case reports
document clinical case of single patient or series of patients
nominal data
mutually exclusive
- qualitative
- ex blood types (no overlap)
ordinal data
- qualitative
- order matters
- ex MMT
discrete data
whole numbers
- ex counting people
- quantitative
- can be ratio or interval
continuous data
- quantitative
- has decimals, ex measure of weight
- can be interval or ratio
interval data
- quantitative
- NO true 0, can have negatives
- ex. temp
ratio data
- quantitative
- true 0
- highest level of measurement
- ex height, money
Reliability
Types of reliabiliy
Consistency of measure, extent of research instrument to consistently have same results with used on multiple occasions
- intra-rater: test performed by 1 person several times, use equipment that already know is reliable
- inter-rater: test performed by 2+ ppl on testing 1 variable
- test-retest reliability: same test to same subjects on 2 occasions, testing new equipment that dont know is reliable
Validity
-consistency of instrument or measure
- extent to which instrument used is measuring what you want it to measure
concurrent validity
- strongest form
- test performed and compared to GOLD stanard and test results are matched
content validity
- test measure specifically what patient problem is
- ex for balance/fall risk - use TUG
construct validity
- test should measure what its supposed to measure
- ex using specific tool: goni
face validity
- weakest form
- outcome measure should measure what it looks like it will based on patient problem
- what it appears to measure
type I error
falsely reject null
false +
type II error
falsely accept null
false -
Sensitivity
- ability to ID trule disease WITHOUT LEAVING ANYONE OUT
- few false negatives
- SNOUT
specificity
ability to be CORRECTLY NEGATIVE in ABSENCE OF DISEASE without mislabeling
- few false positives
-SPIN
sensitivity equation
Sn=TP/(TP+FN)
specificity
specificity=TN/(TN+FP)
parametric data
- bell shaped curve
- quantitative data
- randomization of sample
- equal distribution (=sample size in each group)
non-parametric data
- unequal distribution (unequal sample size)
- no randomization of sample
- skewed curve
independent T test
- parametric
- 2 independent groups
- ex athletes vs non-athletes
paired T-test
- 2 matched/same groups
- ex athletes before and after training
1 tailed T test
know which way hyp will go/directional hyp
- ex: inc or dec pain after PT
2 tailed T test
- non-directional hypothesis
- ex trying out new intervention
1 way anova
- 3+ independent groups compared on 1 intervention
2 way anova
- 3+ independent groups compared on 2 interventions
repeated measures anova
- individuals measured over time
- ex: 3 diets measured at different times
ANCOVA
anova + controlling for cofounding factor/ covariant
chi squared test
use nominal data to find difs btw groups
Mann Whitny U test
- compare 2 independent groups using ordinal data
- ex MAS
- similar to independent T test
Kruskal Wallis test
- compare 3+ groups with ordinal data
- similar to 1 way anova
pearson correlation
for parametric data
rho/spearman correlation
for non-parametric data
correlation scale
r=correlation
0-.25=low
.26-.5=fair
.51-.75=moderate
.76-1=high
regression
- explains how change in 1 variable can have change in another variable
- ex: height and weight
- gives in form of equation that relates x to y so that if given x - y can be predicted
regression values
R^2=regression
>.5 is strong
<.5 is weak