Biostatistics

Question 1

What are the steps that research needs to take to get published?

Answer 1

• Begin with research question
• Design the study
• Enroll the subjects
• Collect the data
• Analyze the data
• PUBLISH

Question 2

What is Continuous Data? What are the 2 types?

Answer 2

• Data that has a logical order with the values “continuously” increasing or decreasing by the same amount
• Ratio & Interval data

Question 3

What is the difference between interval & ratio data?

Answer 3

• Interval: NO meningful zeros (i.e.; Celsius or Fahrenheit)
• Ratio: HAS meningful zeros (i.e.; age, height, weight, time, BP…)

Question 4

What is discrete (categorical) data? What are the 2 types?

Answer 4

• Data fitting into categories
• Nominal & Ordinal data

Question 5

What is the difference between nominal & ordinal data?

Answer 5

• Nominal: subject are sorted by arbitary categories (i.e.; gender, ethnicity, marital status…)
• Ordinal: ranked in logical order BUT does not increase my same amount (i.e.; pain scales, HF classes…)

Question 6

What are the statistics that measure the central tendecy? What about the spread (variability) of data?

Answer 6

• Tendency: Mean (average), Median (middle), Mode (most common)
• Spread: Range (highest - lowest), Standard Deviation (how spread the values are)

Question 7

What is the gaussian (normal) distribution?

Answer 7

• “Bell Curve” - normally seen in larger data sets

Question 8

What are some of the characterisitics of a guassian distribution?

Answer 8

• Normal data = symmetrical (half of the values are on the left & half on the right)
• Normal data has the same mean, median, mode. 68% fall within 1SD of the mean & 95% fall within 2SD

Question 9

What is a skewed distribution and how does it differ from a normal one?

Answer 9

• NOT symmetrical
• 68% do not fall within 1SD of the mean
• Mean, Median, Mode are NOT the same value
• Normally occurs when there are outliers

Question 10

What is a variable in a data set? What are the 2 types?

Answer 10

• ANY data point or characterisitic that can be measured or counted (i.e.; age, gender, BP, pain)
• Independent & Dependent Variable

Question 11

What is the difference between independent & dependent variables?

Answer 11

• Independent: Changed or manipulated by the researcher to make sure it effects the outcome
• Dependent: the outcome

Question 12

What does the null hypothesis mean?

Answer 12

• That there is NO statistically significant difference between groups (Drug vs Placebo –> trying to state that there is NO differnce between them)
• “what the researcher TRIES to DISPROVE”

Alternative Hypothesis STATES statistically significant difference and is what the researcher is trying to PROVE

Question 13

What is the Alpha Level

Answer 13

• Maximum error margin (threshold for rejecting the null hypothesis)
• Alpha is commonly set to 5% or 0.05

The smaller the choosen alpha level; the more data is needed

Question 14

What is the way that we compare the P-value to Alpha?

Answer 14

• If Alpha IS 0.05 and P-Value is less than 0.05 = REJECT the null (statisitcally significant)
• if Alpha IS 0.05 and P-Value is more than 0.05 = FAILED TO REJECT the null (NOT statistically significant)

Question 15

What is a confidence interval? What is the equation to find it?

-

Answer 15

• Gives the same information about significance as the P-Value plus precision
• Cl = 1 -a
• Alpha is 0.05 then Cl is 95%

A narrow Cl range = high precision & a wide Cl range = low precision

Question 16

What is a Type I error?

“False Positive”

Answer 16

• Basically accpeting the alternative hypothesis when the null should have been been (was rejected with an error)
• When alpha is 0.05 and result is p < 0.05 = it is statistically significant and the probability of error is 5%

Question 17

What is a Type II error?

“False Negative”

Answer 17

• “beta” error; whent eh null should have been rejected but was accepted
• Normally set to 10 - 20% and increases with smaller sample sizes

Question 18

What is study power and how is it determined?

Answer 18

• The probability that the test will reject the null correctly (trying to avoid Type II error)
• Larger sample sizes increase the power and decrease the risk of type II

Question 19

What is relative risk (or risk ratio) and what is the way to interpreate it?

Answer 19

• Ratio of risk in the exposed group divided by the risk in the control group
• RR = 1: no difference between groups
• RR > 1: greater risk in treatment group
• RR<1: lower risk in treatment group

Question 20

What is Relative Risk Reduction?

Answer 20

• After RR is found; its looking at how much risk is reduced in the treatment group compare to control
• 1 - RR

Question 22

What does the Absolute Risk Reduction show? What is the way that it is interpreated?

Answer 22

• Includes the reduction in risk and the incidence rate of the outcome
• Basically says that X out of 100 patients will benefit from the treatment

ARR = % of risk in control group - % of rick in treatment group

Question 23

What is number needed to treat?

Round Up

NNT = ( 1 / ([Risk of control group] - [risk of treatment group]) )
or 1 / ARR

Answer 23

• The number of patients who need to be treated for a certain period of time in order for one patient to BENEFIT

Question 24

What is number needed to harm?

Round Dwn

NNH = ( 1 / ([Risk of control group] - [risk of treatment group]) )
or 1 / ARR

Answer 24

• The number of patients who need to be treated for a certain period of time in order for one patient to have HARM

Question 25

What is the **odds ratio** trying to find out? ## Footnote OR = AB / BC

Answer 25

- The **estimate** of the **risk of unfavorable events** with the treatment (Basically looking to find out the **odds of the outcome** occuring **WITH OR WITHOUT** exposure

Question 26

What is the **Hazard Ratio**/

Answer 26

- The rate in which an **unfavorable event** occurs **within a short period of time** - HR in treatment group / HR in control group

Question 27

What is the way that we **interpreate** both **odds ratio (OR) & hazard ratio (HR)**?

Answer 27

- **OR** or **HR =** 1: event rate is the same (no advantage to treatment) - **OR** or **HR >** 1: event rate in **treatment group** is **higher** than control group (twice as many issues) - **OR** or **HR <** 1: event rate in **treatment group** is **lower** than control group (half as many issues)

Question 28

What do the **primary** & **composite** endpoints mean?

Answer 28

- Primary: **main** result that is measured to see if the **treatment** had a **benefit** - Composite: combines **multiple primary** endpoints to get **one measurement** ## Footnote Composite Endpoints **MUST** be similar in magnitude

Question 29

What is a **T-Test**

Answer 29

- Parametric method used with **continuous data** - **One-Sample T-Test**: single smaple group is compared with gen pop - **Paired T-Test**: when the patient is their control too)

Question 30

What is **ANOVA**?

Answer 30

- Analysis of Variance (F-Test) - helps find statistical significacne with **continuous data of 3 or more** samples

Question 31

What is a **Chi-Square Test**?

Answer 31

- For **nominal or ordinal data**; helps find the statistical significance betwen treatment groups

Question 32

what is **correlation**?

Answer 32

- Statistical technique used to determine if **one variable changes** compared to **another variable** ## Footnote **DOES NOT** prove causal relationships

Question 33

What is **sensitivity**? ## Footnote "True Positive"

Answer 33

- Describes how effectively a test **identifies** patients **WITH** the condition - Higher the better; **100%** sensitivity is **+** in those with the condition ## Footnote A / (A + C)

Question 34

What is **specificity**? ## Footnote "True Negative"

Answer 34

- Describes how effectively a test **identifies** patients **WITHOUT** the condition - Higher the better; **100%** specificity is **-** in those without the condition ## Footnote D / (D + B)

Question 35

What is **intention-to-treat**?

Answer 35

- Data from **ALL** patients from **each treatment group** (active; control) **even if** the patient **DO NOT** complete the trail (provides a conservative estimate of the treatment effect)

Question 36

What is **Per Protocol**?

Answer 36

- Only for those that **completed** the trail

Question 37

What is an **equivalence** trail?

Answer 37

- Trying to show that the "new" treatment is **roughly the same** as the "old" treatment

Question 38

What is a **noninferiority** trail?

Answer 38

- Trying to show that the "new" treatmenti is **NO worse** than the current standard of care treatment

Question 39

What is a **forest plot** and what are the **different components** of them?

Answer 39

- Able to show the results from **multiple studies** (pooling them) - **meta analysis** - **Boxes**: effect estimate & **Diamonds**: pooled results - **Horizontal Lines**: length of **CI** for that endpoint - **Vertical Lines**: line of **"no effect"** - **Zero** for difference data & **One** for ratio data ## Footnote If the **CI** crosses the **vertical line** = **NO** significance

Question 40

Wht are the **common types** of **Medical studies**?

Answer 40

- **case-control**: Retrospective comparisons of cases and controls - **Cohort Studies**: Retrospective or Prospective comparisons of those with exposure to those without - **Randomized**: prospecitve comparison to patients in random groups - **Meta-Analyses**: Results from multiple studies