Glossary

These are terms that I use in this overall Analytics discussion and that I wanted to establish a definition for. You may also find this list in some of my other work.

Absolute Risk
Absolute risk is the statistical risk of a certain diagnosis/occurrence within a certain specified population over a certain defined time. For instance, an absolute risk for prevalence of Coronary Heart Disease is the statistical risk that an individual within that population has Coronary Heart Disease. To obtain absolute risk for a population, find out the total number of (usually estimated) persons who express the issue at risk (i.e. who have CHD) and divide that number by the total population studied at the same instance of time (i.e. if you are looking at the total number of estimated U.S. citizens with CHD in 2001, find the total population of the U.S. in 2001 and divide the first by the second), and express the decimal number you get as a percentage (multiply by 100 and append a % sign after it). E.g. Absolute Prevalence Risk of CHD for the U.S. with numbers so far collected in this spreadsheet appears to be about 2.5% – 5.5%, depending on whose statistics you use.

CHD
Coronary Heart Disease – “Heart disease is caused by narrowing of the coronary arteries that feed the heart. Like any muscle, the heart needs a constant supply of oxygen and nutrients, which are carried to it by the blood in the coronary arteries. When the coronary arteries become narrowed or clogged by fat and cholesterol deposits and cannot supply enough blood to the heart, the result is coronary heart disease (CHD). If not enough oxygen-carrying blood reaches the heart, you may experience chest pain called angina. If the blood supply to a portion of the heart is completely cut off by total blockage of a coronary artery, the result is a heart attack. This is usually due to a sudden closure from a blood clot forming on top of a previous narrowing.” (http://www.nhlbi.nih.gov/chd/chdexp.htm)

Confidence Interval
Definition to be provided.

CVD
Cardiovascular Disease – “Cardiovascular disease is a broad term used to describe a range of diseases that affect your heart or blood vessels. The various diseases that fall under the umbrella of cardiovascular disease include coronary artery disease, heart attack, heart failure, high blood pressure and stroke.” (http://www.mayoclinic.com/health/cardiovascular-disease/HB00032)

Incidence
The overall number of new cases of whatever the disease/disorder is in the overall population, per year.

Mortality
The overall number of deaths attributable to whatever the disease/disorder is out of the total number of deaths, per year.

Prevalence
The overall number of cases of whatever the disease/disorder is in the overall population.

Relative Risk
Often found in scientific papers, mostly because the change in absolute risk is so slight, or used to rhetorical advantage. Relative Risk is usually quite greater in pure magnitude than absolute risk. Relative risk represents the percentage change between absolute risk and risk observed after a certain treatment protocol change or change in lifestyle or other change is made and/or observed in an experimental population. Often in small populations, a control population is created which is only observed and not intervened with and another population, the experimental population, is both intervened with and observed. The difference in absolute risk is then compared and expressed as a relative risk (usually in percent). For example, relative risk is often expressed as a 50% change in risk between those lowering cholesterol and those not.

Statistical Rhetoric
It’s always possible to use rhetorical methods to make statistics indicate what we want them to indicate. Relative risk, for instance, is a statistic that is particularly prone to statistical rhetoric. It can be artificially inflated by changing the way it is calculated (is it a fraction of control to experimental, or the reverse?). It can be used to make a very small change in absolute risk look gigantic and colossal instead. Other statistical methods can be used to either minimize and ignore difference or to inflate and exaggerate difference. This is why finding a standard approach and using it consistently matters strongly in the production and consumption of statistical data, and why it is important to be math-literate enough to be able to double-check researchers’ numbers and self-reported statistics.