Biostatistics Flashcards
What is continuous data?
numerical data in which the magnitude from one value to the next is equal (HR, age, height, degrees C and F)
What is discrete/categorical data?
- Nominal- arbitrary order (gender, mortality, ethnicity, marital status)
- Ordinal- logical order but magnitude from one value to the next is not equal (1-10 pain scale, NYHA functional class
What are measures of central tendency?
a typical value that describes all the possible values and likelihoods that a random variable can take in a given range:
1. mean
2. median
3. mode
What is mean and when is it preferred?
- average value
2.continuous data that is normally distributed
What is median and when is it preferred?
- middle value
- ordinal or continuous data that is skewed
What is mode and when is it preferred?
- most frequent value
- nominal data
How is the variability of data (spread) described?
- range
- standard deviation
What is the range?
difference between the highest and lowest values
What is standard deviation?
indicates how spread out the data is and to what degree the data is dispersed away from the mean (highly dispersed data have a larger SD)
What are the characteristics of Gaussian (normal) distribution?
- large sets of continuous data
- normal/symmetrical bell-shaped curve
- mean, median, and mode are the same value at the center point of the curve
- 68% of data fall within 1 SD of the mean; 95% of data fall within 2 SD of the mean
- half of the values are on the right and the other half on the left with a small number of data in the tails
What are the characteristics of a skewed distribution?
- not symmetrical
- data is skewed toward outliers (extreme high value is skewed right and extreme lows are skewed left)
- 68% of values do not fall within 1 SD of the mean, median, and mode
- mean, median and mode are not the same value
- usually occurs when the number of values (sample size) is small and/or there are outliers
What are the characteristics of outliers?
- in a small population outliers have a large effect on the mean
- median is a better measure of central tendency
- distortion from outlier is decreased by increasing the population
What is a variable?
any data point that can be measured or counted
What is an independent variable?
variable that is changed (manipulated) by the researcher to determine its effect on the dependent variable (the outcome)
What are examples of dependent variables?
- A1c
- cholesterol values
- mortality
the outcome of an independent variable
What are examples of independent variables?
- drug/dose regimen
- patients included (age,gender, comorbid conditions)
What is a null hypothesis?
Ho: hypothesis that the researcher is trying to disprove or reject (drug X does not treat HF better than drug y); not statistically significant difference
What is an alternative hypothesis?
Ha: hypothesis researcher is trying to prove or accept ( drug x treats HF better than drug y); statistically significant
What is the alpha level?
standard for maximum permissible error margin and threshold for rejecting the null hypothesis; usually 0.05 (5%)
What is a p-value?
calculated based on statistical test data and compared to alpha; when the p-value is < 0.05 the results are statistically significant and the null hypothesis is rejected
What is a confidence interval?
- includes data on statistical significance and precision of results
- CI= 1- alpha
- when alpha is 0.05 the study reports a 95% confidence interval
- when alpha is 0.01 the study reports a 99% CI
What type of confidence interval has high precision?
narrow CI;
ARR 0.12 (0.95 CI 0.06-0.15, 6%-15%) we are 95% confident that the true value of the ARR for the general population lies somewhere within the range of 6-15%
What type of confidence interval has poor precision?
Wide CI;
ARR 0.12 (0.95 CI 0.06-0.35, 6%-35%) we are 95% confident that the true value of the ARR for the general population lies somewhere within the range of 6-35%
What is a type 1 error?
False positive: null hypothesis (no change in drug x and drug y) was rejected in error and alternative (drug x is better that drug y) was accepted incorrectly;
95% CI= 1- alpha (0.05) the probability of making a type 1 error is <5% and we are 95% confident that the result is correct and not due to chance
What is a type II error?
Probability of a false negative (Beta): null hypothesis (no change in drug x and drug y) is accepted in error and the alternative hypothesis (drug x is better that drug y) is rejected incorrectly; usually set at 0.1 or 0.2 (the risk of type II error is 10% or 20%); risk of type II error is increased with small sample size
What is study power?
the probability that a test will reject the null hypothesis correctly (the power to avoid a type II error;
Power= 1-Beta if Beta is set at 20% the study has 80% power; as power increases chance of type 2 error decreases
How is study power determined?
- number of outcomes collected
- difference in outcome rates between groups
- significance (alpha) level
What is risk?
the probability that an event (how likely it is to occur) when an intervention is given;
Risk= # subjects with unfavorable event/ total # of subjects
What is relative risk?
ratio of risk in the in the exposed group (treatment) divided by risk in the control group;
RR= risk in tx group/ risk in control group
How is RR interpreted?
RR=1 (100%) equal risk between intervention and control groups, intervention had no effect
RR<1 (<100%) lower risk (reduced risk) of the outcome in the treatment group, patients treated with metoprolol were 57% as likely to have progression to disease as placebo treated patients
RR>1 increased risk of the outcome in the treatment group
What is relative risk reduction?
indicates how much the risk is reduced in the treatment group compared to the control group, calculated after RR;
RRR= (% risk in control group-% risk)/ % risk in the control) or 1-RR;
(RR+RRR=100%) 1-0.57= 0.43 metoprolol patients were 43% less likely to have HF progression than placebo patients
What is absolute risk reduction?
includes the reduction in risk and the incidence rate, most useful for clinicians
ARR= (% risk in control group)-(% risk in treatment group)
How is absolute risk reduction interpreted when ARR=12%?
- 12 out of 100 patients benefit from the treatment
- for every 100 patients treated with metoprolol 12 fewer patients will have HF progression
What test is used for normally distributed continuous data from a single sample group compared with the general population?
one sample T-test
What test is used for normally distributed continuous data from a single sample group compared with before and after measurements (serves as its own control)?
Paired t-test (dependent)
What test is used for normally distributed continuous data from 2 independent groups (treatment and control)?
Unpaired t-test (independent)
What test is used for normally distributed continuous data from ≥3 groups?
ANOVA (F test)
What test is used for skewed continuous data from a single sample group?
Sign test
What test is used for skewed continuous data from a single sample group with before and after measurements (own control)?
Wilcoxon Signed Rank test
What test is used for skewed continuous data from 2 sample groups (treatment and control)?
Mann-Whitney test (Wilcoxon Rank-Sum)
What test is used for skewed continuous data from ≥3 sample groups?
Kruskal-Wallis test
What test is used for nominal/ordinal data from a single sample group?
Chi-squared test
What test is used for nominal/ordinal data from a single sample group with before and after measurements (own control)?
Wilcoxon Signed-Rank test
What test is used for nominal/ordinal data from 2 groups (treatment and control)?
Chi-squared test or Fisher’s exact
(Mann-Whitney preferred for ordinal data)
What test is used for nominal/ordinal data from ≥3 groups?
Kruskal Wallis test
What is correlation?
used to determine if a variable (days hospitalized) changes or is related to another variable (incidence of nosocomial infection); DOES NOT prove causal relationship
1. when independent variable (days hospitalized) causes the dependent variable to increase= positive correlation
2. when independent variable (days hospitalized) causes the dependent variable to decrease= negative correlation
What is regression?
used to describe the relationship between dependent and 1 or more independent variables or need to control for confounding factors
What is a case-control study?
Retrospectively compares patients with a disease (cases) to those without a disease (control) the outcome is already known
What is a cohort study?
Prosective or retrospective comparison of outcomes of a group of patients exposed and not exposed to a treatment to see if they develop an outcome
What is a cross-sectional study?
estimates the relationship between variables and outcomes at one particular time in a defined population
Which type of study may be influenced by confounders?
cohort study
What is a meta-analysis?
Combines results from several studies to develop a conclusion with higher statistical power
What is a systematic review?
Summary of clinical liturature focusing on a specific topic/question
What is a factorial design?
randomizes more than 2 groups
What is pharmacoeconomics?
identifies/measures/compares costs (direct/indirect/intageble) and consequences (clinical/ecomonic/humanistic) of pharmaceutical products/services
What is an incremental cost-effectivness ratio?
represents the change in costs and outcomes when 2 different treatments are compared
What is a cost minimization analysis?
used when 2 or more interventions have demonstrated equivalence in outcomes and costs of each intervention are compared
What is a cost benefit anaylsis?
Compares benefits and costs of an intervention in monetary units
What is a cost effectivness analysis?
compare clinical effects of 2+ interventions to the respective costs
What is a cost utility analysis?
included quality of life component of morbidity assessments (quality-adjusted life years or disability adjusted life years)