PNN Biostatistics

What are the 3 measures of central tendency?

Mean, median, and mode

What do measures of central tendencies measure?

The central point of collected data

What does the mean represent?

The average of all the numbers in the data

What type of data is mean generally used for?

Continuous data

What is the most commonly used to measure central tendency?

Mean

Mean is a reliable measure of central tendency unless there is a what?

Extreme outlier

What does the median represent?

The middle value in the set of data

What type of data is median generally used for?

Ordinal data

What does the mode represent?

The entry that occurs the most frequently

What does the range represent?

The interval between the lowest and highest numbers in the data

What does the variance represent?

The measure of data spread or distribution

What does the standard deviation represent?

The square root of the variance

With a normal distribution curve, the area from 1 SD below to 1 SD above mean covers ______% of the population.

68%

From 1 SD to mean, it would be 34%

With a normal distribution curve, the area from 2 SD below to 2 SD above mean covers ______% of the population.

95%

From 2 SD to mean, it would be 47.5%

What is the equation to calculate ARR?

CER - EER OR EER - CER

Difference between the event rates

What does ARR not show?

The degree of difference relative to the baseline risk, so relative risk solves that problem

What is the equation to calculate RR?

EER/CER

What does RR express?

The correlation in relation to the baseline risk

What does a RR of 1 mean?

No difference between the 2 groups

What does a RR of < 1 mean?

The treatment reduced the event rate compared to the control

What does a RR of > 1 mean?

The treatment increases the event rate compared to the control

What is the equation to calculate RRR?

1 - RR

What is the equation to calculate NNT/NNH?

1/ARR

How is NNT rounded?

Rounded UP

How is NNH rounded?

Rounded DOWN

What is the equation to calculate OR?

(EER/ENR) / (CER/CNR)

OR

(EER x CNR) / (ENR x CER)

EER = experimental event rate

ENR = experimental non-event rate

What does an OR of 1 mean?

No correlation

No benefits seen

What does an OR of < 1 mean?

A decrease in likelihood

What does an OR of > 1 mean?

An increase in likelihood

Increased likelihood of the event

What is the difference between HR and OR?

RR is calculated at the end of the study, and HR can be calculated at any point or time interval in the study to measure the strength of association before waiting for the investigation to end

What is discrete data?

Countable whole numbers with no decimals (cannot be divided) - typically number of people

What is continuous data?

Where the use of decimals allows for an infinite number of possible values (ex: lab values, height, weight, drug concentration). Can be expressed in decimals and not whole numbers

What are the 2 types of discrete (non-parametric) data?

Nominal (categorical) and ordinal (ranked)

What is nominal data?

Assigning data to different categories based on the occurrence of outcome (ex: people who had MI and people who did not, race, and gender)

"Name-only"

What is ordinal data?

Data that comes in a certain order or ranking, but the intervals between the values are not necessarily equal (ex: pain scale from 1-10, NYHA HF classification 1-4)

"ordered-categories)

What are the 2 types of continuous (parametric) data?

Interval and ratio

What is interval data?

Measurable data with equal intervals between values (ex: temperature). The data type does NOT have an absolute zero (0 temperature does not mean no temperature)

What is ratio data?

Type of interval data that has an absolute zero (ex: height, weight, hemoglobin level) (zero weight does mean no weight)

What are the 3 things a type of test used in a study depends on?

1. Type of data (continuous, nominal, or ordinal)

2. Number of groups tested

3. Whether the groups are independent or paired

What are the types of continuous data tests?

- T-test

- Student's t-test

- Paired t-test

- Analysis of Variance (ANOVA) test

What are the types of nominal data tests?

- Chi-Square test

- McNemar's test

- Cochran's Q Test

What are the types of ordinal data tests?

- Wilcoxon Rank Sum test

- Wilcoxon Signed Rank test

- Kruskal-Wallis test

- Friedman test

What type of data and groups is Student's t-test used for?

Continuous data

2 independent and separate groups

What type of data and groups is Paired t-test used for?

Continuous data

2 paired groups

What type of data and groups is ANOVA used for?

Continuous data

3 or more independent or paired groups

What type of data and groups is Chi-Square test used for?

Nominal data

2 or more independent groups

What type of data and groups is McNemar's test used for?

Nominal data

2 paired groups

What type of data and groups is Cochran's Q test used for?

Nominal data

3 or more paired groups

What type of data and groups is Wilcoxon Rank Sum test used for?

Ordinal data

2 independent groups

What type of data and groups is Wilcoxon Signed Rank test used for?

Ordinal data

2 paired groups

What type of data and groups is Kruskal-Wallis test used for?

Ordinal data

3 or more independent groups

What type of data and groups is Friedman test used for?

Ordinal data

3 or more paired groups

What is a randomized controlled trial?

A clinical study comparing an experimental group to a control group where the members of the two groups are chosen randomly without bias and respect to their characteristics, so the groups are similar

What is the gold standard for clinical trials?

Randomized control trials

What is a cohort study?

A group of people who share a common trait, where exposed and non-exposed individuals of the cohort are observed over time and an outcome of interest is measured

Are cohort studies retrospective or prospective?

Prospective

Who is considered the control group in a cohort study?

The individuals who are NOT exposed to a particular risk or treatment

What type of studies are done for clinical treatment trials of new medications?

Cohort studies

What is a case-control study?

Individuals with a particular outcome that has already happened are compared with people who have never had the outcome (control) and their prior risk factor exposure is compared

Who is considered the control group in a case-control study?

The individuals who have never had the outcome or event of interest regardless of their exposure status

Are case-control studies retrospective or prospective?

Retrospective

Which is less time consuming and less expensive, cohort studies or case-control studies?

Case-control studies since they are retrospective (data just needs to be collected - no following over time)

What is a cross-sectional study?

Risk factors and health status of a group s studied at one specific point in time. It helps determine the prevalence of a disease at a specific point in time

What is a cross-over study?

Study participants serve as their own control. Subjects who were initially exposed to placebo for a certain amount of time will "cross-over" to the experimental group and vice versa

What is an intent to treat analysis?

Uses data from ALL PATIENTS, including the ones who started the study but did not complete the study as instructed

What is a per protocol analysis?

Counts the data only from the patients who completed the entire study. Provides more accurate information about the true effectiveness of treatment if patients are fully compliant

What is a meta-analysis/systematic review?

The results of many relevant studies are reviewed objectively, quantitatively, and systematically to come up with a clinically useful conclusion

What type of trials are preferred for meta-analysis?

Randomized controlled trials

How are end results of a systematic review meta-analysis reported?

Odds ratio of each individual study with 95% CI and a final composite odds ratio of all the studies reviewed

What is a null hypothesis (HO)?

Shows that there is NO difference or association between groups being tested

What is an alternative hypothesis (HA)?

Shows that there is a correlation between events or a difference between groups being tested

What is a type 1 error?

False rejection of the null hypothesis

"False positive"

You imagined it, it doesn't exist

What is the probability of a type 1 error represented by?

alpha (a = 0.05), which corresponds to a p-value < 0.05 (statistical significance)

Means a type I error will be made less than 5 times out of 100 times

What is a type 2 error?

False acceptance of the null hypothesis

"False negative"

Type 1 error is the false ____________ of the null hypothesis.

Rejection

Type 2 error is the false ____________ of the null hypothesis.

Acceptance

What is the probability of a type 2 error represented by?

Beta (B) = 0.20, which corresponds to a power of 0.8 (power of study = 1 - B)

What does a p-value represent?

The probability of incorrectly finding an association or difference when it does not exist (type 1 error)

A p-value of < _____ is generally considered to be statistically significant.

0.05

Sometimes, in a study with a large sample size, we use a p-value of < _____ to be considered statistically significant.

0.01

What is clinical significance?

Finding a difference between the treatment group and the control group the matters clinically and has an impact on patients' lives

What is the equation to determine power?

Power = 1 - B

What does power represent?

The ability of a test to find an association or a difference when one exists

Power is the likelihood of not making a type ______ error.

2

When does power increase?

When the difference of interest is more noticeable and/or when the sample size increases

What can an underpowered study lead to?

A type 2 error

What does a 95% confidence interval represent?

The interval or range that will include the true mean of your population 95 out of 100 times

In any study, the calculated values of RR, RRR, and OR are estimates from only _____ sample of the population.

One (if you repeat the study, the values will be slightly different each time)

What does a RR with 95% CI range < 1 mean?

Treatment decreases event rate compared to control

What does a RR with 95% CI range > 1 mean?

Treatment increases event rate compared to control

What does a RR with 95% CI range that crosses 1 mean?

The change in risk can be in either direction or non-existent which translates to inconclusive results (NOT STATISTICALLY SIGNIFICANT)

Since RRR is not a ratio, a ______ value means no reduction.

Zero

What does a positive RRR (> 0) mean?

Treatment REDUCES risk

What does a negative RRR (< 0) mean?

Treatment INCREASES risk

What does a RRR of 0 mean?

No association of change in risk

What does a RRR with a positive 95% CI range (> 0) mean?

Treatment REDUCES risk

What does a RRR with a negative 95% CI range (< 0) mean?

Treatment INCREASES risk

What does a RRR with a 95% CI range that spans from a negative value to a positive value (crosses 0) mean?

No conclusion can be drawn - NOT STATISTICALLY SIGNIFICANT

A 95% CI range for RRR that does NOT cross 0 means the study is statistically significant with a p-value < _______.

0

A 95% CI range for RR that does NOT cross ______ means the study is statistically significant.

1