BIOSTATS

Question 1

What is another name for sensitivity?

Answer 1

True positive rate

Question 2

Front

Answer 2

Back

Question 3

What is length-time bias in screening?

Answer 3

Length-time bias occurs when screening disproportionately detects slower-growing or less aggressive (benign) diseases because these remain detectable for longer periods. As a result, screened populations may appear to have better survival, even if the screening doesn’t truly improve outcomes.

Question 4

What is lead-time bias in screening?

Answer 4

Lead-time bias occurs when earlier detection of a disease (through screening) falsely appears to increase survival time, even though the actual time of death is unchanged. The disease is diagnosed earlier, so the patient appears to live longer from diagnosis, but the course of the disease is not altered.

Question 5

What is the difference between lead-time bias and length-time bias?

Answer 5

• Lead-time bias: Screening detects disease earlier, making survival time appear longer without actually improving outcome.
• Length-time bias: Screening is more likely to detect slow-growing, less aggressive diseases, making outcomes seem better because aggressive cases are missed or present later.

Both create the illusion of improved survival due to earlier detection, but for different reasons.

Question 6

What is Hawthorne bias?

Answer 6

Hawthorne bias occurs when individuals alter their behavior because they know they are being observed in a study. This can lead to misleading results, as the observed effects may not reflect real-world behavior.

Question 7

What is another name for specificity?

Answer 7

True negative rate

Question 8

Draw out the table for sensitivity, spec., PPV and NPV

Answer 8

Question 9

What is the difference between case-control and cohort studies?

Answer 9

Case-control studies: Start with outcome status (cases vs. controls) and look back in time to assess exposures. Good for studying rare diseases.

Cohort studies: Start with exposure status (exposed vs. unexposed) and follow subjects forward to assess outcomes. Good for studying rare exposures.

Question 10

Cohort studies are related to odds ratio or relative risk?

Answer 10

Odds ratio

Question 11

Case control studies are related to odds ratio or relative risk?

Answer 11

Relative risk

Question 12

What is the difference between odds ratio and RR (relative risk)?

Answer 12

Odds ratio: 1 number / 1 number
RR: 1 number / 2 numbers

Question 13

What is the biggest risk of bias in a cross-sectional study?

Answer 13

Risk of establishing a relationship between outcome and exposure without a clear timeline because patients are not followed over time

Question 14

Formula for ARR

Answer 14

Absolute risk reduction = event rate in the controls - event rate in the exposed

(patients with disease in controls/all controls) - (patients with disease in exposed/all exposed)

Question 15

Formula for NNT

Answer 15

NNT = 1/ARR

Question 16

What is a cross-sectional study?

Answer 16

A cross-sectional study measures both exposure and outcome at a single point in time. It provides a snapshot of a population, allowing for assessment of prevalence but not causality.

Question 17

What is the difference between incidence and prevalence?

Answer 17

Incidence: Measures new cases of a disease in a population over a specific period of time.

Prevalence: Measures all existing cases (new + old) of a disease at a given point or period in time.

Incidence = risk, Prevalence = burden of disease.

Question 18

Does incidence or prevalence affect sensitivity and specificity?

Answer 18

Neither incidence nor prevalence affects sensitivity or specificity.
Sensitivity and specificity are inherent properties of a diagnostic test and remain constant regardless of disease frequency in the population.

However, prevalence does affect predictive values (PPV and NPV).

Question 19

What are the formulas for + vs - likelihood ratio?

Answer 19

Question 20

What is the attributable risk percentage (AR%)? What is the formula?

Answer 20

Attributable Risk Percentage (AR%) estimates the proportion of disease incidence in the exposed group that is due to the exposure.

ARP = (RR - 1) /RR

Question 21

Draw out the null hypothesis table

Answer 21

Question 22

What is the difference between Type I and Type II errors?

Answer 22

Type I error (α): rejecting the null hypothesis when it is actually true.
→ “You think there’s an effect when there isn’t.”

Type II error (β): failing to reject the null hypothesis when it is actually false.
→ “You miss a real effect.”

Question 23

What are the two Hardy-Weinberg equations?

Answer 23

Allele frequency equation: p + q = 1
 - p = frequency of dominant allele
 - q = frequency of recessive allele

Genotype frequency equation: p² + 2pq + q² = 1
 - p² = frequency of homozygous dominant (AA)
 - 2pq = frequency of heterozygotes (Aa)
 - q² = frequency of homozygous recessive (aa)

Question 24

n a normal distribution, what percentage of data falls within each standard deviation?

Answer 24

• 68% of data falls within ±1 SD
• 95% of data falls within ±2 SD
• 99.7% of data falls within ±3 SD

Question 25

If the recombination fraction is 0.1, what are the odds that a linked allele is inherited, given that the child has the allele?

Answer 25

If the recombination fraction (θ) = 0.1, then there is a 90% chance (1 − 0.1) that the linked allele is inherited with the marker allele (non-recombinant). 90% chance of co-inheritance 10% chance recombination occurred, so the allele was not inherited with the marker This is used in linkage analysis to infer gene inheritance patterns.

Question 26

What is the difference between Intention-to-Treat (ITT) and Per-Protocol Analysis (PPA)?

Answer 26

Intention-to-Treat (ITT): Includes all randomized participants in the group to which they were assigned, regardless of whether they completed the intervention. Preserves randomization; reflects real-world effectiveness. Per-Protocol Analysis (PPA): Includes only participants who fully adhered to the protocol. Reflects ideal efficacy but may introduce bias due to loss of randomization.

Question 27

What is the difference between confounding and effect modification?

Answer 27

Confounding: A third variable distorts the true association between exposure and outcome. Eliminated via randomization, matching, or stratification. Effect modification: The effect of the exposure on the outcome changes depending on the level of a third variable. Not a bias — it’s a real interaction, best reported (e.g., stratified results).

Question 28

How does sample size (n) affect confidence intervals?

Answer 28

As sample size increases, confidence intervals (CIs) become narrower. This is because the standard error decreases, leading to more precise estimates. CIs reflect both the estimate and its precision — larger n = more precision.

Question 29

What is standard error and how is it calculated?

Answer 29

Standard Error (SE) measures the variability of a sample mean from the true population mean. Formula: SE = SD / √n As sample size increases, SE decreases Used to construct confidence intervals and assess precision

Question 30

Accuracy vs precision

Answer 30

Question 31

How to approach problems with ROC cut-off values

Answer 31

1. Label everything 2. Find the cut-off value and analyze the impact of its movement on FN and FP