Home
Subjects
Explanations
Create
Study sets, textbooks, questions
Log in
Sign up
Upgrade to remove ads
Only $2.99/month
Science
Medicine
Public Health
EBDM
STUDY
Flashcards
Learn
Write
Spell
Test
PLAY
Match
Gravity
Terms in this set (66)
Evidence Pyramid Hierarchy
Systematic Reviews, high quality RCTs
Observational studies, prospective
Observational studies, retrospective
Case series
Textbooks and Expert Opinion without critical appraisal
Critical appraisal
Systematic process of assessing and interpreting evidence, considering its validity, results, and
relevance
4 steps of EBDM
Ask
Acquire
Appraise (Internal construct,
External validity of evidence)
Apply
Systematic Reviews
(Published, exhaustive) Critical appraisals of scientific literature, ideally randomized control trials.
(Cochrane database is a comprehensive set)
PICO
The PICO model has become a standard for stating a searchable question
Patient, intervention, control outcome
Observational study, prospective
A research study design planned and conducted with outcome data acquired in a prospective fashion, i.e., after the study has started. Example: prospective cohort study. A strength of this design is that incidence rates can be compared in groups that differ by levels of exposure.
Observational study, retrospective
A research study design planned and conducted with outcome data that has already occurred in the past, i.e., before the study has started. Example: case control study.
Randomized clinical trial (or randomized controlled trial)
An interventional study in which the patients are randomly selected or assigned either to a group which gets the intervention or to a control group. Essentially an experiment in which individuals are randomly allocated to receive or not receive an experimental preventive, therapeutic, or diagnostic procedure and then followed to determine the effect of the intervention
Systematic Review
*Strongest study design
A formal review of a focused clinical question based on a comprehensive search strategy and structured critical appraisal of all relevant studies.
Case series
Descriptive study design
A study design that reports data on a series of consecutive patients with the same diagnosis. It is crucial that a rigorous case definition be used to insure all actually are members of the same group. There is NO control group.
Clinical epidemiology
A discipline that describes quantifies and postulates causal mechanisms for health phenomena in populations.
Type of evidence that would lead researchers to presume a causal association
1. Strength of association - magnitude of measure of association
2. Consistency upon repetition - different studies find consistent results
3. Specificity - specific risk factors associated with specific diseases
4. Time sequence - shorter the duration of time the stronger the evidence
5. Dose response - increase the risk factor dose is associated with an increase in rate of outcome
6. Biological plausibility - basic science offers a sound explanation
7. Coherence of explanation - explanation isn't contradictory to what is already known about the disease or risk factor
8. Experimental data
Case report
Descriptive study design
Describes a single individual case
Descriptive studies
case report, case series, cross-sectional study
Observational - retrospective
Case-control, outcomes and effectiveness registries
Cross-sectional study
Descriptive
at one point in time, the individuals with the same or similar disease from a certain population are studied (such as their potential risk exposures, treatment side-effects, etc.). Capable of estimating prevalence if done properly; not capable of calculating incidence.
Case-control
Observational - Retrospective
Used to calculate the ODDS of exposure to a risk factor in individuals with and without a disease; subjects for case-control studies are chosen because they have (cases) or do not have (controls) a certain disease. Relatively inexpensive and quick to perform. Subject to selection bias, recall bias, and reporting bias. Cannot, and should not, be used to estimate prevalence or calculate incidence.
Well done case-control studies have the following characteristics
a. Controls are representative of the target population
b. Cases are representative of all cases of the disease
c. The disease is rare (allows OR to estimate RR)
Outcome and effectiveness registries
Observational - retrospective
Existing databases are retrospectively studied searching for associations between specific outcomes and risk exposures or therapies. Examples include large insurance databases (Medicare - CMS, Kaiser Permanente, Trauma Center databases). Data on individuals entered for non-research reasons then "mined" for the secondary research purposes.
Cohort
Observational - prospective
Patients are chosen for the study if they are documented to NOT HAVE the disease outcome in question. The subjects are then carefully screened on the basis of exposure to a risk factor under investigation or no exposure. The disease outcome (onset of disease) is then determined over time. Incidence of disease (and other outcomes evaluated) can be accurately calculated. Relative Risk can also be calculated. This is the gold-standard study design for estimating prognosis and measuring the association between risk factors and disease outcomes. Very expensive and time consuming to conduct
Incidence
Number of new cases of disease occurring during a specified period of time (usually one year); expressed as a percentage of the number of people at risk. This data should come from prospective Cohort studies or RCTs - these are the ONLY studies that should be used to calculate incidence.
Prevalence
Proportion of people affected with a particular disease at a specific point in time; an accurate prevalence rate can inform a clinician's pre-test probability estimate
Mortality rate
the incidence or probability of death; the number of people who die (during a specific period of time) divided by the number of people at risk of death.
Risk
Probability
Quantitative estimate of an occurrence
Frequentist theory
Chance or frequency of a random event occurring
Bayesean
Conditional probability that depends on certain data; a pre-test and post-test probability exist that depend on whether the data or test result in known
Absolute risk
Probability that an individual with a risk factor exposure has the outcome.
a/(a+b)
Relative risk
(RR) - incidence of disease in exposed/incidence of disease in non-exposed
RR > 1 ; risk exists
ratio of risk of an outcome among individuals exposed to the risk factor being considered to the risk of the outcome in people NOT exposed; a measure of the strength of the association between risk factor and outcome. This data should come from prospective Cohort studies or RCTs.
[a/(a+b)]/[c/c+d)]
Odds ratio
(OR) - probability exposure
OR > 1 ; association exists
Ratio of the ODDS of an outcome among individuals exposed to the ODDS of the outcome in people NOT exposed to the risk factor; used as an estimate of RR in case-control studies. This data usually comes from Case Control studies. Incidence CANNOT therefore be calculated (the subjects are selected because they already have the disease).
(a/b)/(c/d) = ad/bc
RR vs OR
Relative risk is a BETTER estimate of risk. The odds ratio is an ESTIMATE of the relative risk and does this reasonably well when a disease is very rare. In those situations, RR is approximately equal to OR.
Attributable risk
Estimate of the amount of risk of an outcome in individuals exposed to the risk factor is actually due to, or attributable to, the risk factor; Two major ways to express this (AAR and ARI)
Absolute attributable risk
(AAR)
simple difference between absolute risks with and without the risk factor. Also known as the Absolute Risk Reduction because this is the estimated amount of disease burden that could be reduced if we could just control for this risk factor. Number needed to Harm (NNH) is the number of individuals that need to be exposed to the risk factor to result in ONE disease outcome.
AAR = a/(a+b) - c/(c+d)
NNH = 1/AAR
Relative attributable risk or attributable risk percent
(ARP)
is the attributable risk relative to the risk in non-exposed
individuals.
ARP = [[a/(a+b) - c/(c+d)]]/[c/(c + d)]
Two-by-two table
Way to compare two dichotomous variables - such as the status of a risk factor and the status of a disease outcome.
(Other names include: contingency table, 4-fold)
Skewed
not symmetric, shifted to one side or the other
Negative skew
fewer scores at the low, peak
shifted to the right
Positive skew
fewer scores at the high end, peak
shifted to the left
Avoiding bias
-Simple random sample
-Random assignment
Inferential statistics
estimate parameters population from sample
• can rarely study entire population so need to study samples
• sampling & predicting outcomes on basis of samples is fundamental to research
Alternative hypothesis (H1)
a difference is expected
» two-tailed = simply expect a difference to exist (group A & B will differ)
» one-tailed = expect a difference and state in which direction (group A will do better than group B)
Null hypothesis (H0)
there will be no difference
significance level (alpha)
criteria decide accept/reject H0
a = 0.05 = minimum established by scientific community - when you find sufficient evidence to reject null can be 95% certain that doing so because truly is a difference in data because of experimental manipulation; but accept that 5% of time results occurred by chance alone
P-value
The likelihood that the result observed
is due to chance if H0 is correct (a is the value of p at which you are willing to reject H0 even if it is correct) - p < 0.05 indicates statistical significance
Type I (α) error
reject H0 when it's true (more serious of two errors)
Type II (β) error
accept H0 when it's false (less serious)
• errors can be minimized by good design & sufficient power but cannot be eliminated
Error trade-off
As you decrease the probability of making a Type I Error you increase the probability of making a Type II Error - as Type I is more serious most people set the Type I Error (typically at 0.05)
Confidence limits
attempt to capture population parameters
Power
(1-Β)
Probability correctly reject H0 when H1 true
• typically set at 80%
- power of 80% means that when H0 is truly false and there is a true treatment/experimental effect, a significant difference will be detected 80% of the time
• increasing power reduces the probability of making
Type II error
Ways to increase power
1. lower a to 0.1
2. 1-tailed instead of 2-tailed H1
3. increase effect size (difference between means) - can't control
4. decrease variance/standard deviation - can't control
5. increase sample size = best & most effective method
Unpaired t-test
used for comparing means from 2 samples
- unpaired t-test = typically have control & treatment/experimental groups
» control = get standard of care or placebo (no intervention)
» treatment/experimental = get new treatment, drug or some type of intervention
Paired t-test
typically have same subjects in both
groups, for example in a pre-post (before & after) design (e.g., single group of hypertensive patients has
blood pressure measured before going on any drug, go on the drug for 6 weeks then measure blood pressure again and compare pre and post levels)
» takes into account that same subjects measured twice and thus there are correlations or common relationships between the two sets of data
H0 for T-Test
No significant difference between the means
H1 for T-Test
can be either 1-tailed or 2-tailed
• 2-tailed H1 = there is a significant difference between the means
• 1-tailed H1 = the treatment group will have a higher mean score than the control group
Accepting/rejecting H0 for a T-Test
• need a-value, 1-tailed or 2-tailed, degrees of freedom
• df = number independent scores go into estimate of population parameter minus number of parameters estimated as intermediate steps in the estimation of the parameter itself
• t > table reject H0; t < table accept H0
T-statistic
t-test calculation basically gets the difference between 2 means and divides by standard error of the difference (square root of the average standard deviation of the two groups) - this takes into account the central tendency of the 2 groups and an estimate of the average dispersion of the data .
-formula yields single value called t-statistic
-compare calculated t-statistic with theoretical sampling distribution for the t-distribution (tables are found in
statistics books or on-line) to decide if accept/reject H0
ANOVA
comparing 3+ means
Compares variability each treatment/experimental group across all subjects (between variance) to variability individual subjects across all treatment conditions (within variance)
assumes data normally distributed & similar variances
• F-statistic examined for significance using F-distribution
• H0 = 3+ means do not differ; H1 = 3+ means differ
if reject null all you can say is the means differ - cannot say exactly which ones differ
Chi-Squared
Goodness of fit
Do observed (collected) data fit expected pattern (chance) or are trends observed in distribution
One-variable Chi-square (X2)
Used with typical survey questions whether subject picks from set of pre-set categorical answers (e.g., how much do you agree with the statement "compared to 5 years ago I take better care of myself" strongly disagree, disagree, neutral, agree, strongly agree)
Multi-variable X2
proportions/frequencies/percents of observed categorical values for 2+ groups (minimum = 2 x 2) in 2+ conditions
• is there a difference in the proportion of males vs females benefitting from low dose aspirin in terms?
• calculating expected values more complicated than single-variable situation but still finding "goodness of fit" between expected and observed
Fisher's exact test
replacement for Chi-square test when outcome of interest occurs infrequently and thus data are "lopsided" and one variable has two few counts.
Correlation
examines strength & direction of relationship between 2 variables (can use 3+ in multi correlation)
0 = no correlation (or no linear relationship)
correlation coefficient (r)
measure used to express extent or strength of relationship; often referred to as Pearson r
Positive correlation
0 < r < 1; score high on 1 variable &
score high on the other; score low on 1 variable score & score low on the other
Slope cannot exceed 1.0, 1.0 being perfect correlation
Negative correlation
-1 < r < 0; score high on 1 variable & score low on the other; negative slope when plot the data;
Slope cannot exceed -1.0; -1.0 being perfect correlation
Spearman R
Ranked data
regression (r)
using correlation in models of prediction
• if linear relationship exists between 2 variables can use that to calculate equation of line that best represents relationship, then use to predict what one variable (weight) would be if know value for other (height)
• can use multiple regression techniques with 3+ variables
coefficient of determination (r2)
gives the proportion of 1 variable explained by the other
Sets with similar terms
Step 2 CK Biostats
58 terms
Biostats teacher
83 terms
Epidemiology
55 terms
PAS II Final - Biostatistics (SL)
48 terms
Other sets by this creator
Cardio Autonomic SuperDeck
312 terms
CV Autonomics (external)
66 terms
Autonomics Pharmacology (external)
54 terms
CARDIO ANS (external)&Cardiology Therapeutics
56 terms
Other Quizlet sets
Chapter 6 Study
18 terms
NLN Science Section
33 terms
Sr. Dog questions
46 terms
Audit Test 3
20 terms