Exam 11 Flashcards
Define informed consent.
Duty to disclose risks, benefits, and alternatives to treatment
Patient must have capacity, cannot be coerced
Reasonable physician standard, reasonable patient standard
An adult with capacity has the right to make health care decisions for him or herself. What about a minor? An adult that lacks capacity?
Minor - typically require a parent or legal guardian to make healthcare decisions
Adult that lacks capacity - surrogate decision maker
What are some limits to parental authority?
Generally cannot refuse a lifesaving medical treatment
5 exceptions to confidentiality
- Continued treatment
- Consent
- Communication of a threat
- Compliance with the law
- Court order
Measures the total number of people with a condition of interest at a given point in time
Prevalence
Measures the number of new cases that arise in a population without the condition of interest over a given period of time
Incidence
What does a standardized mortality ratio of 1.2 mean?
There are 20% more deaths in this population than would be expected relative to other populations
What are the 2 types of epidemiological studies?
- Descriptive
2. Analytic
What are the 2 types of descriptive epidemiological studies?
- Case report
2. Case series
What are the two categories of analytical epidemiological studies?
- Observational
2. Interventional
What are the 5 types of observational studies?
- Community surveys
- Ecological
- Cross-sectional
- Case-control
- Cohort
What are the 2 types of interventional studies?
- Clinical trial (randomized control, non-inferiority, equivalence)
- Community intervention
Studies done looking at variables at the group or population level
Ecological
Advantages - can use when measurement of individual level variables is not possible or ethical, quick, cheap, quick assessment of a health status in a community
Disadvatnages - bias, ecological fallacy
Type of clinical trial exploring if a new treatment is “no worse” than a current treatment
Non-inferiority
One sided trial vs. two sided trial
1 - non-inferiority
2 - equivalence
Type of clinical trial that shows a new treatment should not be inferior to NOR superior to current treatment (not different - good for generic medications)
Equivalence trials
Study that randomizes communities instead of individuals
Community interventions
Comprehensive review of many studies to summarize findings on a particular health topic
Systematic review
Advantages - more studies/sources provide more information, cheaper than repeating studies, results more generalizable than individual studies, strong evidence-based resource
Disadvantages - can be difficult to combine studies, summary only as good as the individual source
Statistical analysis of multiple studies to develop a single conclusion
Meta-analysis
Advantages - more studies = greater statistical power, more complex analysis, can do subgroup analysis
Disadvantages - advanced statistical techniques needed, heterogeneity of study population/study can make extraction difficult, possible publication bias, only as good as source
Quantitative measure of difference between groups
Effect size
Assumption - if all studies included were very large, they would all yield the same results; studies results can only vary by chance
Fixed-effect model
very restrictive, over-estimates precision, does not allow for study heterogeneity
Assumption - studies included are a random sample from all studies that addressed the question; study questions might be different but are related enough to be combined
Random-effect model
more conservative, will slow less statistical significance, more realistic, allows for study heterogeneity
Describe features of a forest plot - square, horizontal line, diamond
Each study is a square, size proportional to weight of study
Confidence interval is the horizontal line
Summary effect is diamond
Variability in study outcomes between studies
Heterogeneity
How can heterogeneity be assessed?
Forest plots Statistical tests (Cochran Q test, I^2 statistic)
Features of a Cochran Q test
Statistical significance - yes/no
Assumption - underlying effect in each study is the same
Low p value indicates significant heterogeneity
Recognized as a poor test
Measures the magnitude of variability between studies
I^2 statistic
How can you assess the cause of heterogeneity?
- Subgroup analyses - assess separate criterion in each study to see if at a subgroup level there is still significant heterogeneity (categorical variables)
- Meta-regression - assess separate criterion in each study to see if at a subgroup level there is still significant heterogeneity (continuous variables)
Studies with significant results published earlier than those with non-significant findings indicate ___ bias.
Time-lag
Define publication bias
Only positives studies published
How can bias be assessed?
Kappa statistic (interrater reliability, >0.75 excellent, <0.40 poor)
Bias risk scale
Funnel plots
Maternal mortality rate = ?
Number of deaths/100,000 live births
Neonatal mortality rate = ?
Number of deaths in first 28 days/1,000 live births
Infant mortality rate = ?
Number of deaths in children <1 y/o divided by 1,000 live births
Mortality rate = ?
Total deaths in a time interval or cause-specific deaths/1,000 or 100,000
Standardized mortality rate = ?
Observed deaths in a population/expected deaths in a standard population
Case fatality rate = ?
of deaths in a population of people who have a disease/total number of people who have a disease (in a given period of time)
Morbidity rate = ?
of non-fatal cases of a disease in a population
Attack rate = ?
of cases of disease ina population during a specific time period
Secondary attack rate = ?
new cases of a specified disease among contacts of known cases/size of contact population at risk
Prospective studies are what?
Cohort studies
Retrospective studies are what?
Case-control studies
Absolute risk = ?
New cases in a period of time/population at risk = incidence
Relative risk = ?
Probability of getting disease in exposed/probability of getting disease in non-exposed = (A/(A+B))/(C/(C+D))
Attributable risk = ?
(A/(A+B)) - (C/(C+D))
Proportion of risk attributable to exposure
RR > 1?
Increased risk of outcome in exposed group
RR = 1?
No difference in risk between exposed and unexposed
RR < 1?
Reduced risk in exposed group
Determines how risk decreases with treatment
Absolute risk reduction
ARR = ?
(C/(C+D)) - (A/(A+B))
+ value - % prevented
- value - increases risk
Number of people you need to treat in order to prevent disease in one person = NNT = ?
1/ARR
Number of people needed to treat for one person to suffer a harmful event = NNH = ?
1/Attributable risk
Population attributable risk = ?
Incidence (population) - Incidence (unexposed)
Proportion of the disease incidence in the population due to exposure, aka disease incidence that would be eliminated if exposure were eliminated
Population attributable risk percent = ?
PAR/Incidence (population)
% of the disease incidence in the population that is due to the exposure = disease incidence that would be eliminated if exposure were eliminated
When are odds ratios used?
Case-control (retrospective) studies
Probability is used to calculate ___ vs. odds is used to calculate ___.
Both describe the likelihood of an event occurring
RR; odds ratios
Ratio of outcome, value from 0 to infinity
Odds
Fraction of an outcome when every outcome is considered, (0-1)
Probability
0 - impossible
1 - inevitable
OR = ?
Odds of having disease in exposed/odds of having disease in unexposed = AD/BC
Hazard ratio = ?
Hazard in the intervention group/hazard in the control group
Hazard = probability that an event will occur at any given moment in time
HR = 0.5?
Half as many patients in treatment group are experiencing the event compared to controls
HR = 1?
Event rates same between groups
HR = 2?
2x as many patients in treatment group are experiencing event compared to controls
Tells us how many times more (or less) likely a test result is found in a diseased vs. non-diseased group; depends on odds, not probability
Likelihood ratio
Odds = ?
Probability of an event/(1- probability of event)
Probability = ?
Odds/(1+ odds)
LR + = ?
Diseased people with + test/non-diseased people with + test = sensitivity/(1-specificity)
LR - = ?
Diseased people with test/non-diseased people with test = 1 - sensitivity/(Specificity)
If LR + is > 1…
Increased likelihood of disease if test result is +
If LR - is <1…
Decreased likelihood disease if test is negative
Validity - aka ?
Accuracy
Reliability - aka?
Precision (reproducibility)
Causal criteria include…
Temporality (cause precedes effect)
Strength (strong association)
Dose-response (larger exposures lead to highe rrates of disease)
Reversibility (reduced exposure leads to lower rates of disease)
Consistency
Biologic plausibility (known MOA)
Specificity (singular cause for disease)
Analogy (other similar exposures cause same disease)
Type of bias - if researchers know which treatment a study participant is getting, they may treat the participant differently
Ascertainment bias
Prevent with allocation and double blinding
Type of bias - occurs when participants are selected in such a way that they differ in variables that affect the outcome other than the variable being studied or the sampled population does not reflect the general population
Selection/sampling bias
Hawthorne effect?
Participants know they are being studied so they act differently
Early detection of a disease leads to an apparent increase in overall survival (screening does not affect the natural history of disease)
Lead-time bias
Occurs when a screening test is mroe likely to detect a slowly developing cancer but misses rapidly progressive cancer - cancers picked up by screening tend to be less aggressive, these patients do better
Length-time bias
___ is an assumption about a population of interest; what are the 2 types?
Hypothesis
Null - hypothesis of no difference (no association)
Alternative - some difference (some association)
Alpha value?
Level of significance
(Probability of a false positive RE significance = Type I error)
Standard is 0.05
Alpha < 0.05 ?
Results occurred by chance 5% of the time
P value?
Calculated probability from the study data; helps us determine significance
Compare p and alpha value -
If p value is less than or equal to alpha value, reject the null, statistically significant
Type II or beta error
Probability of a false negative
Probability of rejecting the null hypothesis when it is false
Study power (probability of a true positive)
Power = ?
1 - beta
___ reduces both types of error; ___ reduces type II error.
Increase sample size; increase power
Range f values that are likely to include the true effect size
Confidence interval
Rule - Gaussian distribution
68 (1 SD)
95 (2 SD)
99 (3 SD)
Negative skew
Shifts to the right
Positive skew
Shifts to the left
Compares means between 2 different groups
T-test
Used to see if there is an association between categorical or nominal variables
Chi-square goodness of fit test
Compares proportions, determines if the observed proportions between data categories are significantly different than what would be expected by theory
Used when there is 1 categorical variable with 2+ values
(Sex - M/F), age, types of cars
Chi squared = ?
Sum of [(observed - expected)^2]/expected
Degrees of freedom = ?
k - 1
k = # of levels of categorical value
a
Test to compare means of two distinct sample groups
Independent t-test
Test to compare one sample mean with a known population mean (?)
Single sample t-test
Test to compare two categorical variables/if there is a relationship between them
Chi-square test of independence
Categorical = hour of day, # produced
Test with two independent variables and 1 dependent variable
ANOVA
___ validity tells us if the results are believable and trustworthy for the sample of people in the study vs. ___ validity tells us if the results could be applied to total population
Internal; external
CI indicating results are not statistically significant
If CI includes 0 or 1
3 things that increase the power of a study
Increase sample size, increase effect size, Decrease type 2 error
