Ch. 18 & 19 Statistics and Epidemiology + Registry Research

Statistics

Describes the characteristics of a sample of a population to draw conclusions about the population from which the sample was taken

Statistical Test

Evaluates whether the characteristics of two populations differ

Categorical/Qualitative Data

Mutually exclusive data that is grouped

Categorical/Qualitative Data

eg. Race, primary site, sex, etc.

Nominal data

Categorical data with no numerical significance

Nominal data

eg. Sex, zip code

Ordinal data

Values can be ranked in a meaningful order

Ordinal data

eg. Stage of disease

Dichotomous data

Containing two possible values

Polychotomous data

Containing three or more possible values

Continuous/Quantitative Data

Numerical data that are not innately grouped

Continuous/Quantitative Data

eg. Age, tumor size

Interval data

Beginning at an arbitrary point

Interval data

eg. Temperature

Ratio data

A true zero value indicates “none”

Ratio data

eg. Tumor size, height

Measures of central tendency

Mean, median, mode

Measures of Variation

Range, variance, standard deviation

Variance

Average of squared difference from the mean

Standard deviation

Square root of the variance

Median

Middle value

Mode

Most frequently occuring number

Descriptive statistics

Used to describe the distribution of data

Statistical tests/Inferential statistics

Make conclusions about a population using information from a specific sample

Epidemiology

The study of the determinants, distribution, and frequency of disease in human populations

Epidemiology

Provides a basis for developing disease control and prevention measures for groups at risk

Prevalence

Proportion of population affected with a disease at a specific time

Prevalence

Measure to understand the total burden of disease on a population

Incidence

Number of new cases of a specific disease in a specified population during a specific time period

Incidence

Number of new cases that occur in an at-risk population and the rate at which new cases develop

Cumulative incidence

The proportion of at-risk people who develop a disease during a specific time period

Prevalence rate

Includes all new and existing cases in the population

Mortality

Number of deaths due to a specific cause in a specific population

Screening

Provides early detection of a disease before symptoms appear

Screening

Fundamental aspect of public health

Sensitivity

Proportion of persons who test positive among those who truly have the diasease

Specificity

The proportion of persons who test negative among those who truly do not have the disease

Survival

Length of time from a specified start date to occurrence of an event

Median survival

Time at which 50% of the population being studied has experienced the event of interest

Observed/All-cause survival

Estimate of the probability of survival all causes of death

Net survival

Hypothetical probability of surviving cancer in the absence of other causes of death

Relative survival

Observed survival : expected survival in a comparable set of cancer-free individuals

Cause-specific survival

Net survival measure representing survival of a specified cause of death in the absence of other causes of death

Case-control study

Based on the absence or presence of a disease or condition

Cases

Those with the disease or condition

Control

Those without the disease or condition

Case-control study

Evaluates the differences in risk factors among the study participants with and without the disease/condition of interest

Case-control study

eg. Cases represent patients dx with colorectal cancer after screening colonoscopy. Researcher would generate hypotheses

Cohort study

Participants selected based on exposure or risk factor, not a disease or condition

Retrospective cohort study

Looking back in time

Prospective cohort study

Looking forward in time

Relative risk

Used to compare the exposure of those with and without the disease

Cohort study

eg. All selected participants use e-cigarettes. Can either look back in time and follow from that point onward. Or the participants can be followed over time

Cross-sectional study

Exposure and outcome for each subject is determined at the same time

Cross-sectional study

No follow-up time between exposure and outcome

Cross-sectional study

Provides a snapshot of what is going on at the time the cases are pulled, extracted, or exported

Cross-sectional study

eg. Describes the number of analytic colorectal cancer patients that were diagnosed/treated at Hospital A from 2010-2018. Then stratify count by he number of analytic female versus male colorectal patients

Bias

Impacts internal and external generalizability of the study results

Selection bias

Occurs when there is a difference in the characteristics of study participants compared to those who are not in the study

Selection bias

Occurs in observational studies

Selection bias

eg. Only premium users are surveyed. Insights don’t reflect opinions of low-tier users

Misclassification bias

Occurs in a case-control study

Misclassification bias

Study participants are incorrectly classified as a case when they should be a control

Misclassification bias

Occurs when study participants have recall bias

Confounding bias

When a factor associated with both the exposure and the outcome interferes with measuring true association

Confounding bias

eg. Incidence rates of lung cancer are greater in coffee drinkers, however, Coffee drinkers tend to smoke more than non-coffee drinkers

Suppression

The redaction of data from presentations and publications when the number of cases in the category are very low

Blocking

Initial linkage step that reduces the number of record comparisons between data sources when matching takes place

Blocking

Exact matching between the two data sources on key data items, creating smaller blocks of data that are most likely to match

Blocking

Removes very unlikely matches before the actual matching process

Matching

Performed within identified blocks of data

Types of automatic record linkage

Deterministic and probabilistic

Deterministic linkage

Compares data items between two records and identifies a match only if the values on the matching items are identical

Probabilistic linkage

Calculates a linkage score that indicates how likely it is that a pair of records are a match

Probabilistic linkage

Recommended method of automatic linkage

Probabilistic linkage

A method of handling uncertainty