Memorize the table below for easy calculation of sensitivity, specificity, and positive and negative predictive values.

Know the differences between different types of statistical tests.

Proportion of truly diseased persons in a screened population who are identified as being diseased by the test. It is a measure of the probability of correctly diagnosing a condition.

True positive/(true positive + false negative)

The proportion of truly nondiseased persons who are so identified by the screening test

True negative/(false positive + true negative)

1 – specificity

1 – sensitivity

The probability that a person with a positive test result has the disease

True positive/(true positive + false positive)

The probability that a patient with a negative test result really is free of the disease

True negative/(false negative + true negative)

The total number of cases of a given disease in a specified population at a designated time

The number of new cases of a given disease during a given period in a specified population

The absolute arithmetic difference in outcome rates between control and experimental patients in a trial

The proportional reduction in outcome rates between control and experimental patients in a trial

Confidence intervals

### A method of studying a drug or procedure in which both the subjects and investigators are kept unaware of who is actually getting which specific treatment:

Double-blind method

Number Needed To Treat

Risk

Sample size

Mortality

Survival

Denoted by H

_{o}; it is a proposal that there is no difference in a comparison.

Rejecting the null hypothesis tested when it is true (α)

Failing to reject the null hypothesis when a given alternative hypothesis was true (β)

### The probability that the test will reject the hypothesis tested when a specific alternative hypothesis is true:

Power (1 – β)

Mean

Median

Mode

Validity

Reliability

Sample mean

For a large enough sample size n, the distribution of the sample mean will approach a normal distribution

It is a predictable measure of dispersion from the mean in a Gaussian normal distribution

SD

Observational retrospective study to study risk factors and causation for desired/predefined cases (outcome). Good for rare diseases.

For example, comparing smoking history of acute myocardial infarction (MI) (cases) patient versus smoking history of patient without MI (control)

Observational study of treatment outcome for patient groups that could not be randomized (cohorts) for ethical reasons.

For example, following long-term weight loss outcome of patients who received roux-en-y gastric bypass versus sleeve gastrectomy

Differ from prospective cohort studies in that the exposure in question being studied are collected retrospectively

For example, studying if certain preprocedural comorbidities will affect the long-term weight loss outcome of patients who received roux-en-y gastric bypass versus sleeve gastrectomy

Randomly selecting subjects to be allocated into treatment versus control group. Best design, minimizes biases

Level I: Evidence from at least 1 randomized controlled trial

Level II-1: Evidence from well-designed controlled trial without randomization

Level II-2: Evidence from well-designed cohort or case control studies

Level II-3: Evidence from multiple time series with or without the treatment, or dramatic result in uncontrolled trials

Level III: Expert opinion, clinical experience

To assess statistical difference between sample means of the testing group versus known population mean (1-sample

*t*test) or sample means of 2 independent or dependent groups (2-sample*t*test) for continuous variables

A specific

*t*test to compare 2 sample groups that are not randomly selected.For example, the second sample group is the first sample group after treatment.

Similar to

*t*test, however, is used when sample size is greater than 30 (*t*test is for a limited sample size, ie, less than 30) and when SDs of the population are known

Analysis of variance. It is used to determine whether there is a difference between the means of several different groups (as opposed to

*t*tests, which only compare 2 groups).

It tests whether the distribution of multiple categorical variables in the experimented population differs from the control.

For example, proportion of surviving and nonsurviving cancer patients in 5 years after chemotherapy versus proportion of surviving and nonsurviving cancer patients without treatment.

Similar to

*χ*^{2}test, but specifically only compares 2 categorical variables