Biostatistics Flashcards
Continuous Data
equal distance between values, with or without a true meaningful zero (nominal data or interval data respectively)
ratio data- age, height, weight, BP, time
interval data- Celsius and Fahrenheit
Discrete (Categorical) Data
nominal data- yes or no; male or female
ordinal data - pain score, NYHA Class I-IV
Mean
average
Median
Middle number
Mode
value that occurs most often
Null hypothesis
There is no difference observed
example: metoprolol equals carvedilol
Metoprolol DOES NOT EQUAL Carvedilol (p-value: < 0.05)
Reject the Null Hypothesis or Accept the Alternative Hypothesis
Metoprolol DOES EQUAL Carvedilol (p-value: > or equal to 0.05)
Accept the Null Hypothesis or Fail to accept the Alternative Hypothesis
Confidence Interval
1 - alpha
alpha determines p-value
example: alpha = 0.05
1-alpha= 95% confident that the conclusion is correct
Confidence interval for MEAN data. Significant vs. Non-significant
Significant- NO ZERO
Non-significant- Includes ZERO
Confidence interval for RR, OR and HR. Significant vs. Non-significant
Significant- DOES NOT include ONE
Non-significant- Includes ONE
Type I error (false positive)
Null H0 is rejected in error
alpha= risk of type one error
Type II error
Null H0 is accepted in error
Power= 1- beta (probability of avoiding a type II error)
example: Power 80% means the probability of avoiding a type II error equals 20%
Risk
= # of subjects with the unfavorable event / # of subjects in that arm of the study
RR (relative risk)-specific to tx group
Risk Tx group/ Risk Control group
equal to 1= no difference
> 1: higher risk in tx group
< 1: lower risk in tx group
RR > 1
less likely
RRR
RRR is how much the risk is reduced (less likely)
formula: 1-RR (express as decimal)
RR + RRR must equal?
ONE
ARR (absolute risk reduction)
absolute difference between groups
formula- risk (control group)- risk (tx group)
example: 0.12 or 12% means for every 100 patients treated with metoprolol, 12 fewer patients will have HF progression
NNT/NNH
number of people who need to be treated for a certain period of time in order for 1 patient to benefit
formula: 1/ARR (decimal form)
NNT: 8.3- round up to 9
NNH: 41.9- round down to 41
Odds Ratio
AD/BC
When to use an unpaired student t-test?
CONTINUOUS data that is normally distributed with 2 GROUPS
When to use an ANOVA test?
CONTINUOUS data that is normally distributed with 3 GROUPS
When to use Chi-square test?
Discrete (categorical) data w/ one to two groups.
OR Fisher’s exact test
Sensitivity
true positive
- test will be positive in all patients that have the condition
Specificity
true negative
-test will be negative in all patients that have condition
Sensitivity/Specificity of 87%?
Sensitivity: 87% of patients with the condition are diagnosed while 13% are not.
Specificity: 87% without the condition are diagnosed correctly while 13% of patients w/o the condition are diagnosed w/ the condition or test positive
Case-control study
compares patients with DISEASE to those WITHOUT DISEASE (retrospectively)
Cohort Study
compares EXPOSED patients to NON-EXPOSED (retrospective or prospective)
example: patients taking statins vs patients not taking statins
Top 4 medical studies
- Systematic Review/ Meta-analyses
- RCT
- Cohort Studies
- Case-controlled
Cost-effectiveness analysis (CEA)
most common
- costs are monetary but outcomes are clinical
(MUST BE ONE OUTCOME)
cost- utility analysis
outcomes based on QALY
Cost-minimization analysis
cost of interventions that are EQUIVALENT
cost-benefit analysis
compares benefits and costs in monetary unites ($$$ values)
Having a p-value of that is less than alpha (p<0.05) mans what?
The probability that a conclusion is incorrect is LESS THAN 5% (1 in 20)
