Biostatistics Flashcards
two types of continuous data
Ratio and interval data
Ratio data
equal difference between values, with a true meaningful zero
0 = none
example: age, height, weight, time, BP
Interval data
equal difference between values, but without a meaningful zero
0 doesnt equal none
example: temp
two types of Discrete (Categorical) data
nominal and ordinal
Nominal data
sorted into arbitrary categories
males/females
known as yes/no data
ordinal data
ranked and has a logical order
ie. pain scale, 1-10
Which data is mean preferred
continuous data that is normally distributed
Which data is median preferred
continuous data that is no normal distributed
Which data is mode preferred
nominal data
Percent of values within 1 SD? 2 SD?
1 SD = 65%
2 SD = 95%
When do you often see skewed distribution
sample size is small
outliers in data
by collecting more values, effect of outliers is decreased
Independent vs dependent variable
Independent = changed (manipulated) by researcher to determine if it as affect on the dependent variable (outcome)
null hypothesis
states that there is no statistically significant deference between groups
researchers want to reject it to show that their drug/product is statistically different
Alternative hypothesis
states that there is a statistically significant difference between the groups
what the researcher hopes to prove or accept
Alpha
maximum permissible error margin
Alpha is the threshold for rejecting the null hypothesis
commonly set to 0.05 or 5%
Comparing p-value to alpha
if alpha set to 0.05, and p-value is less than then null hypothesis is rejected and result is statistically significant
How to tell if something is statistically significant with CI and without p-value
if crosses zero = not statistically significant
if doesnt cross zero = statistically significant
How to tell if something is statistically significant if it has ratio data
if crosses 1 = not statistically significant
if doesnt cross 1 = statistically significant
Narrow vs Wide Confidence interval
Narrow = high precision
Wide = lower precision
narrow is preferred
Type 1 error
False positive
The probability of making a type 1 error relates to….
the alpha
if alpha is 0.05 and p < 0.05, then probability of error is < 5%, 95% confident that result is correct
Type 2 error
False negative
The probability of making a type 2 error relates to….
beta
usually set to 0.1 or 0.2, meaning 10%-20%
type 2 error increases with smaller sample sizes
Study power is….
the probability that a test will reject the null hypothesis correctly
ie. power to avoid type 2 error
power = 1- beta
Power is determined by….
number of outcome values collected, difference in outcome rates between groups and significance (alpha) lvl
larger sample size = increases study power
Relative risk (Risk ratio) is….
ratio of risk in exposed group (txm) divided by risk in control group
Risk is….
number of subjects in group with an unfavorable event / total number of subjects in group
Risk Ratio is…
risk in txm group / risk in control group
Risk Ratio interpretation
RR =1 = no difference in risk of outcome between groups
RR > 1 = greater risk of outcome in txm group
RR < 1 = lower risk (reduced risk) of outcome in txm group
Relative risk reduction….
indicates how much the risk is reduced in the txm group compared to the control group
1 - Risk Ratio = RRR
RRR interpretation of 43%
XXXXX patients were 43% less likely to have YYYYY than placebo-treated patients
Absolute risk reduction
includes the reduction in risk and incidence rate of the outcome
ARR = % risk in control group - % risk in txm group
ARR interpretation of 12%
12 out of every 100 patients will benefit from txm
or
For every 100 patients tx with XXX, 12 fever will have YYYYY
Number needed to treat
number of patients who need to be treated for a certain period of time for one patient to benefit
1/ ARR
Number needed to treat interpretation, NNT = 9
For eery 9 pts who receive XXX, YYYY is prevented in 1 patient
Number needed to harm
number of patients need to be treated for one patient to experience harm
same formula, 1/ ARR
Number needed to harm interpretation, NNH = 90
1 patient will be harmed for every 90 patients treated with XXX instead of placebo
Which studies are not suitable for relative risk calculations
Case control studies
use Odds ratio
Odds Ratio formula
Present | A | B. |
——————————————————————————–
Absent. |. C. |. D. |
Formula = AD/ BC
Interpreting Odds ratio of 1.23
XXXXXX is associated with 25% increased risk of YYYYYY
Hazard ratio
used in survival analysis instead of using “risk”
Hazard rate is rate a which an unfavorable even occurs within a short period of time
Hazard Ratio formula
Hazard rate of txm drop / HR rate of control group
Odds Ratio and Hazard Ratio interoperation
OR/HR = 1 = event rate is same in txm/control arms, no advantage to txm
OR/HR < 1 = event rate is lower in txm group than control
OR/HR > 1 = event rate is higher in txm group than control
When is T test used
used when endpoint has continuous data and normal distributed
when single sample group = one sample T test
when single sample group used for pre/post measurement (pt serves as own control) = paired T test
When study has 2 independent variables = student T-test
When is ANOVA or F-test used
Used for continuous data
when using continuous data with 3 or more samples or groups
What test is used for nominal or ordinal data
Chi square test
ie. Assessing difference in mortality between 2 groups or pain scores based on pain scale
Correlation is….
technique used to determine if one variable changes or is related to another variable
when independent causes dependent to increase = positive correlation
when independent causes dependent to decrease = negative
correlation doesn’t prove causal relationships
Regression is….
used to describe relationship between a depends variable and one or more independent
regression is common in observational studies where researchers need to assess multiple independent variables or need to control for many confounding factors
Sensitivity vs Specificity
Sensitivity = how effectively test identifies pts with condition
100% sensitivity = test will be positive in all pts with condition
Specificity = how effectively test identifies pts without condition
100% specificity = test will be negative in all pts without condition
Sensitivity formula……
(A / A + C) X 100
A = # that have condition w/ positive test result
C = # that have condition w/ negative test result
Specificity formula…..
( D / B + D) X 100
D = # without condition with a negative test result
B = # without condition with a positive test result
Intention to treat analysis
include data for all pts original allocated to each txm group, even if the patient did not complete the trial according to study protocol
Per Protocol analysis
conducted for subset of trial population who completed the study according toe the protocol
Equivalence trials
attempt to demonstrate new txm has roughly same effect as old txm
Non-inferiority trials
attempt to demonstrate new txm is no worse than current standard based on predefined non-inferiority (delta) margin
delta margin is minimal difference in effect btw 2 groups that is considered clinically acceptable based on previous research
When are forest plots used
often in meta-analysis, when multiple studies results are pooled into a single study
Case control study Benefits & Limitations
Info: compares pts with disease vs without, retrospective
Benefits: data easy to get, less $$ than RCT, good for looking at outcomes in unethical interventions
Limitations: cause and effect cant reliably by determined
Cohort Study Benefits & Limitations
Info: compares outcomes of group of pts exposed and not exposed to txm, follows prospectively
Benefits: good for looking at outcomes when intervention would be unethical
Limitations: more time consuming and $$$ than retrospective, can be influence by confounders (other factors that affect outcome)
Case Report/ Case Series Benefits & Limitations
Info: signe patient = report, few patients = series
Benefit: Identify new diseases, drug SE or potential uses, can generate hypothesis for other studies
Limitations: conclusions cant be drawn from single or few cases
Randomized control trial Benefits & Limitations
Info: pts randomized and sometimes blinded to txm groups
Benefit: Less potential bias, preferred study to determine cause/effect/ superity
Limitations: $$$ and time consuming, hard to reflect real life scenarios
Crossover RCT info
pts receive treatment A first, then switch to txm B and second group does opposite
Benefit: act as own control, minimize confounders
Limit: washout period req
Meta-analysis Benefits & Limitations
Info: combines results from multiple studies to come to conclusion that has more statistical power
Benefit: smaller studies can be pooled, instead of making large study
Limit: studies may no be uniform in size/inclusion/exclusion etc
Systemic review article info
summary of clinical literature that focuses on specific topic or question
Benefit: cheap, studies already exist
Direct Medical costs
Drug prep, admin, etc
Inpatient direct costs = hospital bed, staff, procedures etc
Outpatient direct costs = office/clinic visits
Direct Non medical costs
traveling/lodging to hospital/clinic
household costs like childcare, etc
home health aides, etc
Indirect costs
Lost work time
Low work productivity
morbidity, costs from having disease
mortality = death
Intangible costs
pain, suffering, anxiety etc
Incremental cost-effectiveness ratios
represents change in costs and outcomes when 2 txm alternatives are compared
ex. Drug A $200, 5 txm success vs Drub B $300, 7txm success
$300- $200 / 7 - 5 = $100/2 = $50
Drug B costs $50 more relative to Drug A for each additional txm success
Cost minimization analysis
used when 2 or more interventions have demonstrated equivalence in outcoems and the costs of each intervention are being compared
Cost benefit analysis
systemic process for calculating and comparing benefits of an intervention in terms of monetary units (dollars)
Cost effectiveness analysis
used to compare clinical effects of two or more interventions to the respective costs
advantages = outcomes easier to quantify
disadvantages = inability to directly compare different types of outcomes
Cost utility analysis
specialized form of CEA that includes quality of life competent of morbidity assessments, usually use QALY and DALYs
