Biostats and epidemiology Flashcards
Definition of type 1 error (alpha error)
Rejecting the null hypothesis when it is really true
Definition of a type 2 error (beta error)
Not rejecting the null hypothesis when it is really false
Ways to increase statistical power
- Increase the sample size (most common)
- Increase the expected effect size
- Increase precision of measurement
Use of the t-test
Comparing the means of two groups from a single nominal variable, using means from an Interval variable to see whether the groups are different
Use of one-way ANOVA
Compares means of many groups (2 or more) of a single nominal variable using an interval variable
Use of chi squared
Tests to see whether two nominal variables (not mean) are independent
Infectious diseases that require mandatory reporting to the local Public Health Department, who will inform the CDC
“Salas siente comezón, grita y chilla al horinar”
- Chlamydia
- Gonorrhea
- Chicken pox
- AIDS
- Syphilis
- Salmonella
- Hepatitis A and B
- Tuberculosis
- Lyme disease
- Other: pertussis, legionnaires, mumps, rubella and measles
Coverage of Medicare
- Ambulance transport
- Dialysis
- Speech and occupational therapy
Data analysis test used in cross-sectional studies
Chi-squared
Data analysis test used in case-control studies
Odds ratio
Data analysis test used in cohort studies
Relative risk
Describe the sample size, data evaluated, and duration of phase 1 clinical trials
- Sample size: 20 to 100 (healthy)
- Data evaluated: safety, toxicity, pharmacokinetics/pharmacodynamics, adverse effects
- Duration: 1 to 30 days
“Is it SAFE?”
Describe the sample size, data evaluated, and duration of phase 2 clinical trials
- Sample size: 100 to 300 (disease)
- Data evaluated: dose, tolerability, efficacy, adverse effects
- Duration: months
“Does it WORK?”
Describe the sample size, data evaluated, and duration of phase 3 clinical trials
- Sample size: 100s to 1000s
- Data evaluated: compares the new treatment to the current standard of care
- Duration: months to years
“Any IMPROVEMENT?”
Describe the sample size, data evaluated, and duration of phase 4 clinical trials
- Sample size: 1000s
- Data evaluated: epidemiology, postmarketing surveillance, common and rare side effects
- Duration: years
“MARKET”
Define sensitivity
Proportion of truly diseased persons in the screened population who are identified as diseased by the screening test (“true positive rate”)
Define specificity
Proportion of truly disease-free persons who are identified as non-diseased by the screening test (“true negative rate”)
Define positive predictive value
Probability that a person with a positive test is a true positive
Define negative predictive value
Probability of no disease in a person with a negative test
Define accuracy
Degree to which a measurement, or an estimate based on measurements, represents the true value of the attribute that is being measured
Formula for odds ratio
OR = AD/BC
*La probabilidad de que los que estuvieron expuestos desarrollen la enfermedad sobre la probabilidad de que los que no estuvieron expuestos desarrollen la enfermedad ((A/B)/(C/D))
Formula for relative risk
RR = (A/A+B)/(C/C+D)
*La incidencia de la enfermedad en los que estuvieron expuestos sobre la incidencia de la enfermedad sobre los que no estuvieron expuestos
Define the attributable risk
The difference in risk between exposed and unexposed groups, or the proportion of disease occurrences that are attributable to the exposure
Formula of attributable risk
AR = (A/A+B) - (C/C+D)
*La incidencia de la enfermedad en los que estuvieron expuestos menos la incidencia de la enfermedad en los que no estuvieron expuestos
Define a selection bias
When the sample is not representative (nonrandom smapling or treatment allocation of subjects)
Examples of selection biases
- Berkson bias
- Non-response bias
- Healthy worker effect
Define the berkson bias
Study population selecten from hospital is less healthy than general population
Define the non-response bias
Participating subjects differ from nonrespondents in meaningful ways
Define the healthy worker effect
Study population is healthier than the general population
Solutions to a selection bias
- Randomization
* Ensure the choice of the right comparison/reference group
Define a measurement bias
Information is gathered in a systemically distorted manner
Define the Hawthorne effect
Measurement bias in which participants change their behavior in response to their awareness of being observed
Solution to a measurement bias
- Using objective, standardized and previously tested methods of data collection
- Using control/placebo groups (Hawthorne)
Define an expectancy bias
Researcher’s beliefs affect outcome (aka Pygmalion effect)
Solution to an expectancy bias
Double-blind design
Solution to a lead-time bias
Measure “back-end” survival (adjust survival according to the severity of disease at the time of diagnosis)
Solution to a recall bias
- Decrease time from exposure to follow-up
* Multiple sources to confirm information
Define a confounding bias
Unanticipated factors obscure results (a factor is related to both the exposure and the outcome)
Solution to a confounding bias
- Multiple/repeated studies
- Crossover studies
- Matching
- Restriction
- Randomization
Formula for standard error
SE = SD/square root of the samples
Define the nested case-control design
Study starts with cohort study, and those who develop an outcome of interest become cases for a case-control study
Black box warnings are added during which phase of the clinical trial
Phase 4
Most important characteristic of an ecological study
Unit of analysis is populations, not individuals
What is an ecollogical fallacy
Making conclusions regarding individuals within populations
Describe a crossover study
Subjects are randomly allocated to a sequence of 2 or more treatments given consecutively
Advantage of a crossover study
It allows patients to serve as their own controls
Disadvantage of a crossover study
Effects of one treatment may carry over and alter response to a subsequent treatment
Solution to the drawback of a crossover study
Have a washout phase
Formula for relative risk reduction
RRR = 1 - RR
Formula for absolute risk reduction
ARR = (C/C+D) - (A/A+B)
Formula for number needed to treat
NNT = 1/ARR
Formula for number needed to harm
NNH = 1/AR
Define precision
The consistency and reproducibility of a test
Methods to increase precision
- Decrease standard deviation
* Increase statistical power
Formula of attack rate
Number of individuals who become ill divided by the number of individuals who are at risk of contracting the illness
Define an attrition bias
Lose to follow up occurs disproportionately between exposed and unexposed groups
Difference between effect modification and confounding bias
Effect modification refers to when the effect of an exposure on an outcome is modified by another variable
What should you do to differentiate between effect modification and confounding bias
Stratified analysis (stratification is based on the confounder, if association disappears, it is a confounding bias)
Formula for the CI for a population mean
CI = mean +/- Z(SE)
Z values for a 95% CI and 99% CI
- 95% = 1.96
* 99% = 2.58
Define a late-look bias
Survey doesn´t uncover patients with severe disease because they die first
Define a proficiency bias
When the different skill levels of physicians delivering treatment might affect patient outcomes more than the treatment selection itself
Disorders related to a high socioeconomic status (SES)
- Anxiety disorders
- Breast cancer
- Bipolar disorders
Definition of socioeconomic status
Weighted combination of education and occupation status
Formula for accuracy
(TP + TN)/Everything
Leading causes of death overall in the US (top 3)
- Heart disease
- Cancer
- Unintentional injuries
Age group in which suicide is the number 2 cause of death
10 - 34 years old
