Biostatistics Ch.14 Naplex 2026

Before studies of a drug on human subjects can begin, what application is needed with the FDA?

Investigational New Drug (IND)

Outline phase 1 of clinical trials

20-100 healthy volunteers or people with the disease state:

- safety

- dosage

Outline phase 2 of clinical trials

100-300 ppl w/ the disease:

- efficacy

- ADRs

Outline phase 3 of clinical trials

300-3000 ppl w/ disease:

- efficacy compared to other tx/placebo

- ADRs

Outline phase 4 of clinical trials

Post-marketing surveillance after drug is approved:

-long term and real world efficacy and safety is assessed

After phase 3 of clinical trials is complete, what needs to be done before phase 4?

A new drug application (NDA) with the FDA

What is a supplemental new drug application (sNDA)

when a manufacturer makes any change in an approved drug or in its production (eg new indication or dose)

T/F: It's optional for researchers in drug trials to disclose any conflicts of interest such as financial incentives

F

It's required

What is sampling bias

A bias that occurs when the sample is not representative of the whole population.

What is selection bias

refers to bias in how people are chosen to participate; due to lack of randomization or strict study exclusion criteria

What is performance bias

When study groups are not treated equally, such as if the study is not blinded then the tx group may receive a diff lvl of care and attention

What is detection bias?

Biased assessment of outcomes, can be avoided by blinding the researchers

What is attrition bias?

when participants with specific characteristics, such as less severe disease, have a higher drop out rate, skewing results

What is reporting bias?

selective reporting of data

What is publication bias?

Inclination for journals to publish positive outcomes more frequently than negative outcomes

What does continuous data mean

Data in a logical order with values that continuously increase or decrease by the same amt

Name two types of continuous data and describe them

1) Interval data:

has NO meaningful zero

eg) 0 degrees in temperature does not mean "none" degrees

2) Ratio data:

has meaningful zero

eg) A heart rate of zero means no heart is beating

What does discrete/categorical data mean

Data that fits into a limited number of categories

Name two types of discrete/categorical data and describe them

1) Nominal data:

categories are in an arbitrary order

eg) gender, marital status, ethnicity

2) Ordinal data:

categories are ranked in logical order

eg) NYHA class I-IV

Mean, median and mode are all categorized as ...

Measures of central tendency

What is mean

arithmetic average

What is mode

most frequently occurring score

What is median

Middle number when placed in order of lowest to highest

What is range

Difference between highest and lowest values.

What is standard deviation?

Indicates how spread out the data is away from the mean

Data that is highly dispersed has a higher SD = wider histogram value

Large sample sets of continuous data tend to form a bell-curve aka ___ or ___

Gaussian; "normal"

In a Gaussian (normal) distribution, what are the mean, median and mode values?

They are all the same values, at the center point of the curve

In a Gaussian (normal) distribution, __% of values fall within 1 standard deviation, and __% of values fall within 2 standard deviations

In a Gaussian (normal) distribution,

68% of values fall within 1 standard deviation

95% of values fall within 2 standard deviations

What are two reasons that cause a bell-curve to have skewed distributions?

1) Small sample size/small # of values

2) Outliers in data

Draw a negative skew

Draw a positive skew

What is a null hypothesis

null means zero

aka no statistically significant difference between two groups.

Researchers want to prove the null hypothesis is not true, reject it.

What is an alternative hypothesis

the alternative to the null hypothesis

Researches want to prove that the alternative hypothesis is true, that there is a statistically significant difference between groups, to accept it.

What is the maximum permissible error margin in a study aka

What is the common % and (decimal) number

alpha (α)

Set at 5% or (0.05); It determines the risk of Type I errors

What is a p-value?

probability that the null hypothesis is true

It's compared with the alpha value.

If alpha is set at 0.05, and the p-value is less than that, then it rejects the null hypothesis, meaning the results are statistically significant

If a p-value is greater than alpha, what does that tell you about the results?

Not statistically significant, fails to reject the null hypothesis

What is a confidence interval?

Another way of expressing statistical significance (p-value)

If alpha is 0.05 then the CI is 95%

If alpha is 0.01 then the CI is 99%

When is a confidence interval statistically significant?

For difference data (means):

Data points should not include 0

For ratio data (relative risk, odds ratio, hazard ratio):

Data points should not include 1

---

way to remember:

1-1 = 0

1/1 = 1

If a p-value is less than alpha, what does that tell you about the results?

Statistically significant, it rejects the null hypothesis

Independent and dependent variables

IV- whats being changed

DV- effect of whats being changed

What is a type-I error

False positive; rejecting the null hypothesis when it is true

How can a type-I error be avoided?

1) Increasing the study power

What is a type-II error

What is another name for type-II error?

False negative; accepting the null hypothesis when it should have been rejected

aka Beta

How can a type-II error be avoided?

1) Large sample size

2) Using a composite endpoint

What is study power?

probability that a test will reject the null hypothesis. Used to avoid type 2 error.

The higher the study power, the smaller the likelihood of error (aka Beta)

Study power = 1 - beta

What is relative risk (RR)

ratio of risk in the exposed (tx group) divided by risk in control group

A RR = 1 means ...

A OR or HR = 1 means...

No difference between groups

1/1 = 1

A RR > 1 means ...

---

A OR or HR > 1 means ...

Greater risk of the outcome in the treatment group

---

Higher event rate in the treatment group

A RR < 1 means ...

---

A OR or HR < 1 means ...

Lower risk of the outcome in the treatment group

---

Lower event rate in the treatment group

What is relative risk reduction (RRR)

How much the risk is reduced in the treatment group compared to the control

RRR = 1 - RR*

*must use decimal form of RR

What is absolute risk reduction (ARR)

Reduction in risk and the incidence rate of the outcome

ARR = (% risk in control grp) - (% risk in tx grp)

Answers questions like "How many fewer patients out of 100 treated with abatacept instead of placebo will experience

worsening of disease?"

What is number needed to treat (NNT)

The # of pts who need to be treated for a certain period of time in order for 1 pt to benefit

1/ARR*

*expressed as decimals

NNT answers will always be rounded UP

eg) 8.3 => 9

What is number needed to harm (NNH)

The # of pts who need to be treated for a certain period of time in order for 1 pt to experience harm.

1/ARR*

*expressed as decimals

NNH answers will always be rounded DOWN

eg) 8.3 => 8

What is odds-ratio (OR)

Used in case-control studies to estimate the risk of unfavorable events

(A) (D) / (B) (C)

For example: There is an OR of 12 for pts that will get pulmonary fibrosis when taking amiodarone compared to placebo. What does this mean?

=> It means pts exposed to amiodarone are 12 times more likely to develop pulmonary fibrosis than

those not exposed.

What are case-control studies

Retrospective look on subjects that have the outcome of interest. Compares pts with and without disease

Use the odds-ratio to compare data

What is hazard-ratio (HR)

In a survival analysis (analysis of death or disease progression), instead of using "risk", a HR is used

It's the rate of an unfavorable event occurring within a short period of time

Rate in tx group/ rate in control group

Similar to relative risk

What is a t-test

A statistical measure used when an endpoint has continuous data and the data is normally distributed

What is a one-sample t-test

When data from a single sample group is compared with known data from the general population

What is a paired t-test

Before and after comparison from a single sample group

What is a student t-test

Used when a study has two independent samples: the treatment and the control groups

What is ANOVA

Analysis of variance

Used for continuous data with 3 or more samples to test for statistical significance

What is chi-square test

Used for nominal or ordinal data to determine statistical significance

Can be done for one sample group or for 2 (treatment and control)

What is Fisher's exact test

Used to determine statistical significance among treatment and control group using nominal or ordinal data

T/F: Correlation of variables equals causation

F

What is regression analysis?

Regression analysis is a widely used procedure in biostatistics to relate an outcome, or dependent, variable (denoted Y) to multiple independent variables (denoted X1, X2, etc), or need to control for many confounding factors

There are three types

1) Linear

2) logistic

3) Cox regression

What is sensitivity

Probability that a patient WITH a disease will have a POSITIVE result.

If a test has 100% sensitivity, it will have a positive result in all patients with the condition

"True-positive"

-----

SeNsitivity Negative→ rules OUT (SNOUT)→ If a highly sensitive test is Negative, you can rule OUT disease

What is specificity

Probability that a patient WITHOUT disease will have a NEGATIVE result

If a test has 100% specificity, it will have a negative result in all patients with the condition

"True-negative"

----

Specificity Positive→ rules IN (SPIN)→ If a highly specific test is Positive, you can rule IN disease

What is intention-to-treat analysis

Analysis of a study including all patients in each treatment group, even if they did not complete the trail according to the study protocol

What is per protocol analysis

Analyzing outcomes only for those who received the full treatment as assigned.

Label the parts of a forest plot:

Boxes

Diamonds

Horizontal lines

Vertical solid lines

Boxes = effect estimate

Diamonds = pooled results from multiple studies

Horizontal lines = length of the confidence interval for that endpoint or study. If it crosses the vertical solid line, the data is NOT statistically significant

Vertical solid lines = no effect. Line is set at 0 for difference data and 1 for ratio data

What is a cohort study

Comparisons of pts with an exposure to those without an exposure, usually prospectively.

Can be influenced by confounders (eg smoking, lipid lvls)

eg) The British Doctors Study: Initiated in 1951, this study followed over 40,000 male doctors and tracked their smoking habits. It was instrumental in establishing the causal link between smoking and lung cancer.

What is single vs double-blinded RCT

What is open label

Single blind: patient is unaware

Double blind: pt and investigator unaware

Open label: all parties know which tx is being given to the pt

What is a parallel RCT

Subjects are randomized to the treatment or control arm for the entire study

What is a crossover RCT

Group 1 gets treatment A first, then treatment B

Group 2 gets treatment B first, then treatment A

What is a factorial RCT

randomized to more than 2 groups to test multiple experimental conditions

What is the ECHO model of pharmacoeconomics?

A theoretical framework used in pharmacoeconomics to comprehensively evaluate the value of pharmaceutical treatments and healthcare services

E: Economic outcomes - direct, indirect and intangible costs of the drug compared to a medical intervention

C: Clinical outcomes - medical events that occur as a result of the tx or intervention

HO: Humanistic Outcomes: consequences of the disease/tx as reported by the pt or caregiver (eq QoL)

Incremental Cost Effectiveness Ratio (ICER)

Cost-minimization analysis (CMA)

Compares costs of interventions with equivalent outcomes

Cost Measurement Unit = $

Outcome Unit = Equivalent

eg) two drugs have equal efficacy, but one costs more than the other

Cost-benefit analysis (CBA)

Compares benefit and costs of an intervention in terms of monetary units $

Cost-effectiveness analysis (CEA)

Compares the clinical effects of two or more interventions to the respective costs

Units: natural units such mmHg blood pressure, % at tx goal

Pros:

Outcomes are easy to quantify

Cons:

Inability to compare different types of outcomes

Cost-unit analysis (CUA)

A type of cost-effectiveness analysis (CEA) that uses variables such as quality-adjusted life years (QALYs) and disability-adjusted life years (DALYs).

What does it mean if a drug is noninferior to another drug?

Not quite "equal": It doesn't necessarily mean the new drug is better or even exactly as good as the old one. It just means it is NOT WORSE by more than a tiny, predefined, and clinically acceptable limit.