Biostats Flashcards
If a sample has a normal distribution, what percent of the sample is included in…
- 1 SD
- 2 SD
- 3 SD
- 1 SD = 68%
- 2 SD = 95%
- 3 SD = 99.7%
How do the mean, median, and mode relate if a data set is positively skewed? Negatively skewed?
- positive: mean > median > mode (long tail to right)
- negative: mode > median > mean (long tail to left)
What are type I and type II errors?
- type I is rejecting the null when it is actually true
- type II is accepting the null when it is actually false
What are alpha and beta values? How do they relate to error?
- alpha is the probability of a type I error and is equal to the p value
- beta is the probability of a type II error and 1 - B equals the power of the study
What is the power of a study?
- it equals 1 - B
- it increases with sample size as the probability of a type II error decreases
What are the following tests used for?
- unpaired t-test
- paired t-test
- ANOVA
- Chi squared
- unpaired T-test: compares numerical measurements taken from two different groups
- paired T-test: compares two different numerical measurements taken from a single group
- ANOVA: compares numerical means of three or more groups
- Chi squared: compares categorical outcomes between two or more groups
What is the best statistics test to compare the BMI in patients before versus after getting a sleeve gastrectomy?
paired T-test
What is a Kaplan-Meier analysis good for?
identifying differences in survivorship over time between two or more groups
What are multivariate logistic and multivariate linear regressions used for?
- multivariate logistic regression: to identify and adjust for multiple potential factors contributing to a categorical outcome
- multivariate linear regression: to identify and adjust for multiple potential factors contributing to a numerical outcome
What are case reports and case series?
- a case report is one of a single event or patient
- a case series is of a small number of similar events or patients
What is the difference between the following:
- cross-sectional study
- cohort study
- case-control study
- cross-sectional study: analyze a population at a particular moment in time to determine the prevalence of factors and disease
- cohort: a population of subjects are analyzed to associate certain factors with an outcome (prospective or retrospective), determining the relative risk
- case-control: patients who had an outcome happen are compared to those who did not (always retrospective and can only determine an odds ratio)
What is a propensity-score-matched study?
- an observational study that attempts to reduce the presence of confounding variables and replicate a RCT
- cases are selected so that the subjects in both groups are similar across multiple factors
What is a crossover study?
one in which each subject receives both interventions, serving as their own control
What occurs in the four phases of a clinical trial?
- 1: a small number of healthy subjects are used to test the pharmacology and side-effects of the drug
- 2: a small number of diseased subjects are used to test the efficacy and dosing of the drug
- 3: RCT to compare the drug to an existing therapy
- 4: ongoing surveillance after the drug is approved to identify long-term side-effects and efficacy
What are absolute and relative risk?
- absolute is the overall probability of the outcome (percent it occurred in both groups)
- relative is the probability of the outcome occurring in the exposed group divided by the probability of it occurring in the unexposed group
Give an equation for relative risk and for the odds ratio.
- RR = (a/(a + b)) / (c/(c+d))
- OR = (a/b) / (c/d)
How is relative risk reduction calculated?
- it equals 1 - RR
- it is the proportion of decreased risk due to not being exposed
What is an odds ratio and how is it calculated?
- the odds of the outcome in the exposed group compared to the unexposed group
- (a/b) / (c/d)
How is absolute risk reduction calculated?
% of outcome in unexposed group - % of outcome in exposed group
How do PPV and NPV differ from sensitivity and specificity?
predictive values take into account the prevalence of the disease in the study population
How are PPV, NPV, sensitivity, and specificity calculated?
- PPV = TP / (TP + FP)
- NPV = TN / (TN + FN)
- sensitivity = TP / (TP + FN)
- specificity = TN / (TN + FP)
What are the following types of bias:
- lead-time
- length-time
- surveillance
- allocation
- Hawthorne
- lead-time: falsely concluding that patients live longer because they are diagnosed earlier
- length-time: falsely concluding that screening improves survival when it only detects more benign disease courses
- surveillance: falsely concluding that a disease is becoming more prevalent when we are testing for it more frequently
- allocation: a confounding factor affects the likelihood that a subject is assigned to a particular study group
- Hawthorne: subjects behave differently when knowingly observed
