HE300: Epi Test 2

Sampling allows us to make _______ about the population

Inferences

What is internal validity?

Ensuring the sample is representative of the population

What is external validity?

Ensuring that the results of a study can be applied to the population

What are the 3 types of sampling? What do they mean?

1. Random - every person has an EQUAL chance of being selected

2. Probability - every person has a KNOWN chance of being selected

3. Convenience - non-random sampling

Observational vs. Experimental: Which has more potential bias?

Observational W

Observational vs. Experimental:

Observational vs. Experimental: h

Observational vs. Experimental: Which has better controlling of confounders?

Experimental

Observational vs. Experimental:

Observational vs. Experimental: Which can study causes of disease AS WELL AS prevention and treatment?

Observational

Observational vs. Experimental:

Observational vs. Experimental: Which is better at making causal associations?

Experimental

Observational vs. Experimental:

Observational vs. Experimental: Which do you choose?

• question involves prevention or treatment

• small effect expected

• ethical, feasible, have money available

Experimental trial

Observational vs. Experimental:

Observational vs. Experimental: Which do you choose?

• the question involves the cause of disease

• moderate or large effect expected

Observational alidiV

Validity: the absence of _____ and _______

Bias; confounders reP

Precision: the absence of _____ ______

Random error

What is the placebo effect?

Sometimes taking placebo can relieve severe, unpleasant symptoms Hat is the HawthorneW

What is the Hawthorne effect?

People have the tendency to change their behaviour when they know they are being observed

Define the study population

Those being considered for enrolment

What are some differences between historical controls and non-random controls?

Historical

• chosen from a point in the past

• records patients with the same disease

Non-random

• controls are chosen at the same time as treatment grouphW

• assignment is not random

What is randomization? Why do we do it?

Randomization is the process of randomly assigning study participants into the treatment of control group

1. makes sure of unbiased assignment

2. makes sure the groups are comparable, confounders will be evenly distributed between groups HaW

What are the 3 levels of masking?

1. Single masked: participants do not know which group they are in

2. Double-masking: participants and those giving treatment are unaware

3. triple-masking: Participants, those giving treatment, and those monitoring effects are masked

What is the primary outcome of an experimental trial?

The main outcome the trial is designed to evaluate hW

What is ‘blocking’? What is stratification?

Blocking: randomizing in set blocks (e.g. 10) to ensure group sizes stay equal

Stratification: separating participants based on a key confounder, then randomizing within the stratum

In a crossover experimental trial, a _______ period must happen to ensure results are valid

Washout period

What type of experimental design inv solves two or more treatments being tested in the same group, using two or more rounds of randomization?

Factorial design

Unplanned crossovers are a form of ___-_________

Non-compliance ntentionI

Intenion-to-treat analysis compares everyone in treatment 1 group with those of treatment 2 group, regardless of what?

Regardless of who actually took the treatment

Why do we use ITT analysis?

Keeps randomization and power, shows effectivenesshW

What does an efficacy analysis compare?

Those who actually took treatment A vs. those who actually took treatment B hW

What are the consequences of non-compliance?

Shrinks the observed difference between groups and reduces statistical power FI

If you anticipate a smaller difference between the treatment and comparison groups, what sample size should you use?

LARGE sample sizeW

_____ ____ depends on the acceptable probability of statistical errors (alpha and beta)

Sample size

Too small of a sample can produce null results because ______ is too low

POWER HaW

What is a type I error (p = alpha)

Concluding that the treatments do differ when they do not

What is a type II error (p = beta)

Concluding that the treatments do not differ from each other when they do P

Power = 1 - beta. What does power represent?

Probability of correctly concluding treatments differ

What does percent reduction in risk tell us?

Tells us the extent of the reduction in disease by the use of the vaccine in the experiment

What does effectiveness tell us?

How well a treatment works in “real-life” situations

What does NNT represent? How do you round NNT?

Number of patients who would be needed to be treated in order to prevent one adverse outcome (e.g. death)

• round UP to nearest integer

What is NNH? How do you round NNH?

Number of patients needed to be treated in order for one adverse outcome to occur

• round DOWN to the nearest integer

New Drug Trials: Phase 1 tests for what?

Safety and pharmacology

New Drug Trials:

New Drug Trials: Phase 1 find the what?

Maximally tolerated dose

New Drug Trials: What does Phase 2 test for?

Preliminary efficacy; safety and side-effect data

New Drug Trials: Phase 3 compares what?

Compares new treatment with standard of care

New Drug Trials: Phase 4 surveils what?

Post-market surveillance; finds rare and long-term effects

New Drug Trials: Randomization is only ethical when we have ________

Equipoise

What common purpose do placebos and masking serve?

Unbiased outcome assessment

ITT analysis preserves ________

Randomization

What is absolute risk?

Incidence in a group on its own

What is relative risk?

Risk in one group compared to another group

What type of studies calculate RR?

• Cohort studies

• Trials

OR=RR when…

• When disease is rare

• When the cases represent all cases in the population

• When the controls represent all without the disease who have been exposed in the population

What types of studies calculate OR?

Case-control (major), cohort (minor)

What is a discordant pair?

One where the case and control differ in exposure

What is a concordant pair?

One where the case and control share exposure history

When completing matched analysis, (concordant/discordant) pairs are excluded from OR calculation

Concordant

Additive measures ask how much disease could be _______ by _______ an exposure

Prevented; removing

What is attributable risk?

Amount of proportion of disease incidence that be attributed to a specific exposure

RR and OR measure _______, AR measures _______ _______

• Strength

• Excess burden

What is bias?

Systematic error by the investigator in design and/or conduct

What is confounding?

Natural mixing of effects with a third variable

What is random error?

Error by chance

Which direction does random error push the result? Which direction does systematic error push the result?

• Can be either direction

• Pushes result in one direction

What is selection bias?

That which results from differences between those who take part in a study and those who do not

What is control selection bias?

Controls don’t reflect the source population’s exposure

What is self-selection bias?

Participation or refusal related to both exposure and disease

What is differential surveillance diagnosis or referral?

Cases found or referred differentially by exposure

What is differential loss to follow up?

Drop-out related to other exposure and outcome

Healthy worker effect

Working population compared with a general population that includes the ill, working population performs better

What is recall bias?

Cases and controls recall past exposures differently

• common in retrospective studies

What is interviewer bias?

Information is solicited, recorded, or interpreted differently by disease or exposure status

What is misclassification?

When cases are identified as controls and vice versa

What are the types of information bias?

• Recall bias

• Interviewer bias

• Misclassification

What are the two types of misclassification?

Differential: rate of misclassification differs between groups

Non-differential: same degree of misclassification between group

What direction does differential misclassification pull the results?

Can pull in either direction, away, or towards the null

What is the counterfactual theory?

The idea that attributing causation requires the counterfactual; if the exposed group were unexposed their risk would be different

What are the three criteria for a confounder?

• must NOT be an intermediate step

• related to the exposure

• an independent cause or predictor of the outcome

What is the difference between a mediator and a confounder?

A mediator is a step in the causal pathway

What is confounding by education?

Treatment is given because of a condition, and the condition itself affects the outcome

What is confounding by severity?

Sicker patients are more likely to get the treatment

How do you control confounding in the design?

• Randomization

• Restriction

• Matching

How do you control confounding in the analysis?

• Stratification

• Standardization

• Multivariable analysis

What is random error?

A false association arising from chance

What are the two types of random error?

• measurement error (inexact assessment of exposure or disease

• sampling variability (unrepresentative sampling by the luck of the draw)

Confidence interval: (narrow/wide) interval = precise

Narrow

What is effect measure modification?

The strength (or direction) of the association between an exposure and disease differs according to the level of the third variable, the effect modifier

Confounding asks if the ____ _____ is distorted?

Crude-estimate

Confounding asks if the ___-specific and ______ estimates differ from the crude?

• Strata

• Adjusted

Effect modification asks if the ________ differs across levels of the third variable

Association

Effect modifiers: What is synergy?

The third variable increases the effect of the exposure

Effect modifiers: What is antagonism?

The effect modifier diminishes or eliminates the effect of the exposure