research Flashcards
1
Q
PICO
A
- patient/problem - target of interest
- intervention
- comparison
- outcome
2
Q
levels of evidence
A
3
Q
systematic review
A
- comprehensive review of med lit that uses explicit methods to systematically search, ID, appraise, and summarize all literature about a specific issue
- Cochrane
4
Q
meta-analysis
A
- systematic review that uses a statistical technique to derive an estimate of effect size by combining results of several randomized control trials to determine overall effectiveness of a treatment
- minimze problem of small sample size from individual studies - pooling of trials increases overall sample size
5
Q
randomized control trials (RCT)
A
- experimental research to assess relative effect of a specific intervention compared to a control condition
- pts randomized into control group and at least 1 experimental group
- control goup - no treatment or standard default treatment
- random assignment reduces bias and increases probability that differences btw groups are d/t intervention
6
Q
cohort study
A
- longitudinal, observational study
- individuals w/ a risk factor or exposure are followed over time to compare disease occurrence in the exposed group to that in a group not exposed
- measure of association is relative risk - ratio of incidence rate of exposed to that of controls
- prospectively and retrospectively
- limitations: excessive length of time, other variables can impact
7
Q
case control study
A
- retrospective, observational study
- individuals w/ a disease are matched with a comparison group of individuals without disease
- history of exposure and characteristics recorded through interview
- control group provides estimate of frequency and amount of exposure in subjects in population without disease
- measure of association btw exposure and occurrence of disease is odds ratio - ratio of oddds of exposure in diseased subjects to odds of exposure in non-diseased
8
Q
cross-sectional study
A
- observational study of data or observations made at only one point in time and all subjects tested at same time
- aims to describe relationship btw disease and factors of interest in a specified population at a given time
- describe prevalence of disease and demonstrate association
- cannot distinguish between newly occurring and long-established conditions
- cannot ID causal relationships
9
Q
case report or case series
A
- in-depth description of a condition or response to a treatment
- can be used to generate theories and hypotheses for future research
- cannot test hypotheses or establish cause-effect relationships
10
Q
institutional review board (IRB)
A
- scientists and non-scientists charged with protecting rights and welfare of participants in research
- review and approve research invoving human subjects
11
Q
informed consent for research
A
- statement that study involves research
- explanation of purpose of research
- description of procedures
- description of foreseeable risks
- description of benefits to subject
- disclosure of alternative procedures or treatments that might be advantageous
- description of who will have access to records that ID subjects
- explanation of compoensation if involves > minimal risk
- who to contact for answers about rights ro issues d/t study
- participation is voluntary
12
Q
types of data
A
- continuous: height, weight, distance, ROM; covers range with no gaps
- discrete: measured in whole units; HR, pts diagnosed w/ cancer, number of visits to PT
- dichotomous: yes or no, pass or fail, smoking or nonsmoking
- qualitative: categorical, non-numeric; eye color, blood type, hand dominance
- quantitative: numbers that represent counts of measurements
13
Q
scales of measurement
A
- nominal: qualitative, you can only be in one category; blood type, type of breath sounds, type of arthritis
- ordinal: ranking; MMT grades, level of assist, pain
- interval: temeprature, functional status tests
- ratio: distance walked, time to complete an activity
14
Q
internal consistency
A
- extent to which items or elements that contribute to measurement reflext one phenomenon or dimension
- ex: a functional assessment scale should only include items that relate to patients’ physical function
15
Q
intrarater reliability
A
- consistency of repeated measurements by one person over time
16
Q
interrater reliability
A
- consistency of measurements made by more than 1 person
17
Q
test-retest reliability
A
- consistency of repeated measurements of same person on different occasions
- can be affected by interval between tests, fatigue, learning
18
Q
face validity
A
- degree to which a measurement tests what it is supposed to
19
Q
content validity
A
- degree a measurement reflects meaningful elements of a construct and items in a test reflect domain of interest and not other extra things
- ex: McGill pain questionnaire is good bc it addresses analogue pain scale and ALSO intensity, location, quality, duration
20
Q
construct validity
A
- degree to which a theoretical construct is measured by a test
- ex: MMT scores would have construct validity as indicators of innervation status of muscle if there was a relationship between MMT scores and results of EMG testing
21
Q
criterion-related validity
A
- validity of measurement estabilished by comparing it to either different measurement often considered “gold standard”
22
Q
concurent validity
criterion related
A
- interpretation is justified by comparing to a gold standard at the same time
- ex: HR by palpation and ECG
23
Q
predictive validity
criterion related
A
- measurement is valid because it is predictive of a future event
- ex: use of GPA or GRE as admission criteria for grad school based on presumptive ability to predict future academic success
24
Q
prescriptive validity
criterion related
A
- measurement suggests form of treatment person should receive - based on successful outcome of treatment
- ex: asystole on ECG has prescriptive validity if patients with this arrhythmia are successfully revived by CPR
25
Q
probability sampling
A
- method of sampling that uses some form of random selection
- every member of population must have same probability of being selected for sample
- sample should be free of bias and representative of the population
26
Q
simple random sampling
probability sampling
A
- subjects have an equal chance of being selected for sample
- relies on table of random numbers or random number generator
- not most statistically efficient
- might not be most representative sample
27
Q
systematic sampling
probability sampling
A
- subjects are selected by taking every nth subject from population
- size of interval based on size of populatin and desired sample size
- simple
28
Q
stratified random sampling
probability sampling
A
- proportional or quota random sampling
- population divided into homogenous subgroups (strata) and then simple random sample drawn from each
- assures sample will be representative of key subgroups of population in addition to overall population
29
Q
cluster sampling
probability sampling
A
- population divided into clusters or areas and a random sample of clusters is selected
- less costly, more efficient than simple random sampling
30
Q
non-probability sampling
A
- any method of sampling that does not involve random selection of subjects
31
Q
convenience sampling
non-probability sampling
A
- selected from subjects who are convenient or readily available to researcher
32
Q
purposive sampling
non-probability sampling
A
- subjects deliberately selected based on predefined criteria chosen by investigators
33
Q
quota sampling
non-probability sampling
A
- subjects not selected randomly
- after ID subgroups of interest, researcher uses convenience sampling to select required number of subjects
34
Q
snowball sampling
non-probability sampling
A
- subjects IDed by asking existing subjects to ID names of other potential participants
- when characteristic to be studied is rare and would be difficult and costly to ID individuals with this characteristic
35
Q
external validity
A
- degree to which results of research are generalizable to populations or circumstances beyond those included in study
- threats to EV: interaction of treatment with specific types of subjects tested and place (setting) and time (history) of the experiment
36
Q
internal validity
A
- degree to which an intervention being evaluated (independent variable) is the cause of the outcome measured in the study (dependent variable) and not the result of extraneous factors
- threats to IV: hisotry, maturation, attrition, testing, instrumentation, regression toward mean
37
Q
hawthorne effect
A
- an untreated subject experiences change simply from participating in the research study
- tendency for people to change behaviors just because they know they are being watched
38
Q
independent variable
A
- variable that is presumed to have caused or influenced dependent variable
- what is controlled or manipulated by researcher
39
Q
dependent variable
A
- response or outcome assumed to be caused by effect of independent variable
- what the researchers are looking for change in
40
Q
A
41
Q
p-value
A
- probability that a particular statistical result could have happened by change
- when smaller than alpha, null hypothesis is rejected
- when larger than alpha, null hypothesis is not rejected
- alpha: significance level, probability of rejecting null hypothesis when it is true, usually 0.05 or 0.01
42
Q
steps in testing a statistical hypothesis
A
- state null hypothesis and alternate hypothesis
- select appropriate test statistic
- select level of significance for statistical test (0.05)
- calculate test statistic from sample data
- interpret result
43
Q
types of errors
A
- type I error (alpha error): wrongly deciding to reject null hypothesis, concluding that there is significant different when there is not; false positive; if signif is 0.01, 1% chance of Type I error
- type II error (beta error): error the researcher makes when wrongly deciding not to reject null hypothesis, concluding there is no relationship when there is; false negative
44
Q
statistical power
A
- chance that statistical test will lead to rejection of false hypothesis
45
Q
MCID vs MDD
A
- minimal clinically important difference (MCID): smallest difference in patient’s condition that pt or clinician considers worthwhile for the costs and side effects
- minimal detectable difference (MDD): smallest difference or change that would be statistically significant, standard error of measurement
46
Q
what each graph is best for
A
- bar graph: magnitude or frequency of categories
- box and whisker plot: distribution of values within a group with minimum, lower quartile, median, upper quartile, max
- forest plot: in meta-analysis, results of individual studies and cumulative summary of all studies
- histogram: display of frequency distribution
- line graphs: demonstrate relationship between two or more quantitative variables
- scatter plot: graphical display that illustrates relationship between two quantitative variables
- stem and leaf plots: graphical display which enables reader to observe entire distribution of data without losing info
47
Q
bell curve
A
- kurtosis: peakedness of a distribution
- in a normal distribution, majority of data are clustered around mean - mean, median, and mode are all the same
- skewness: asymmetry of a distribution for (+) or (-) skey; whatever the skew is has the longer tail and the mean, then the median, then the mode
48
Q
standard deviation
A
- descriptive measure of the spread or dispersion of data
- positive square root the variance
49
Q
variance
A
- sum of squared deviations of each data point from the mean
50
Q
inferential statistics
A
- branch of statistics that use sample data to make inferences about a population
- parametric and nonparametric
51
Q
parametric statistics
A
- assume that samples come from populations that are normally distributed and there is homogeneity of variance
52
Q
analysis of variance
A
- ANOVA
- inferential statistic procedure used to test the equality of means between two or more populations
53
Q
confidence interval
A
- range of values used to estimate a population parameter
54
Q
intraclass correlation (ICC)
A
- form of correlation coefficient that assesses both degree of correspondence and agreement among scores
- ranges from 0.0 to 1.0 and is calculated from variance estimates derived from an analysis of variance
55
Q
nonparametric statistics
A
- do not assume that samples come from populations that are normally distributed and do not assume homogenity of variance
- uses chi-square test: for nominal data, evaluates difference between observed and expected frequencies to examine association or independence between categorical variables
56
Q
sensitivity
A
- percentage of people who test positive for a disease among a group of people who had the disease
- SnNout: with high sensitivity, a negative rules out the diagnosis
57
Q
specificity
A
- percentage of people who test negative for a specific disease among a group of people who do not have the disease
- SpPin: with high specificity, a positive diagnostic test rules in diagnosis
58
Q
incidence vs prevalence
A
- I: number of new cases of a disease during a time, new cases per 100,000 people at risk
- P: number of existing cases, including new and pre-existing, number of existing cases per 100,000 at risk
