The chance that something will happen. Synonyms include chance, probability, and risk. For example, a complete absolute risk statement might read:

• A typical 50-year-old American woman has a 4 in 1,000 chance of dying from breast cancer in the next 10 years.

An absolute comparison of risks: it tells you how much lower the modified risk is than the starting risk in absolute terms.

• Absolute risk reduction = starting risk−modified risk

For example, in a randomized trial, women take drug × or a placebo. After 10 years:

• 3 out of 1,000 women in the placebo group die of breast cancer (starting risk).
• 2 out of 1,000 women in the drug × group die of breast cancer (modified risk).

• Absolute risk reduction	= risk of breast cancer death (placebo group) − risk of breast cancer death (drug × group)
  	= 0.003−0.002 = 0.001 = 0.1%

Here are two ways to express this absolute risk reduction:

• Drug × lowers the 10-year risk of breast cancer death by 0.1 percentage points.
• For every 1,000 women who take drug × for 10 years (instead of a placebo), there will be 1 less breast cancer death.

Risk information should be derived from studies of people like you. The more similar you are to the people on whom the statistics are based, the more confident you can be that the statistics apply to you. Ideally, this means that the participants in a study were at least people of your age and sex and, preferably, were people whose health is very similar to yours—that is, they have the same diseases you do, or they are healthy like you.

An observational controlled research study in which scientists compare two groups of people, one with a disease or condition and the other without it. The scientists then analyze the two groups to look for clues that would explain the difference (diet, lifestyle, or medical history, for instance). If one group has heart disease and the other does not, for example, the researchers might ask about behaviors such as drinking coffee. If the people with heart disease are more likely to drink coffee, this suggests that coffee may have something to do with heart disease. Because the people in the two groups might differ in many other ways, however, you need to be cautious in interpreting the results. The hypothetical study we just described shows that coffee drinking is associated with heart disease, but it does not prove that coffee drinking causes heart disease.

The likelihood that something will happen. Synonyms include absolute risk, risk, and probability. In health statistics, chance is referred to as an absolute risk, and a complete statement would include the outcome and the time frame.

A research study in which scientists compare groups of people who differ in some important way. For example, people in one group may drink a lot of coffee, while people in the other group don’t drink any. The scientists then observe what happens to the people in each group over time. For example, they might measure what proportion of each group dies from heart disease. Since people in such cohorts can differ in many ways, you need to be cautious in interpreting the results. In this example, people who drink coffee may have other behaviors (such as smoking) that affect their chance of dying from heart disease. This kind of study might show that coffee is associated with heart disease, but it does not prove that coffee causes heart disease.

In a study, the control group (also called the comparison group) does not receive the therapy being studied (a test or a treatment, for example). The control group typically receives a placebo or conventional medical care, while the intervention group receives the new therapy. Investigators compare the outcomes for the two groups to determine whether the new therapy is better or worse than the current approach.

The rate of death in a group or population (also called the mortality rate); often calculated for a specific illness. For example, the 1-year death rate for lung cancer is the number of people in a group who died of lung cancer over the past year divided by the total number of people in the group at the start of the year.

The bottom number in a fraction. For example, in the following fraction, 250 is the denominator: $\frac{10}{250}$

The bad things that can happen if you take an action, including the side effects of drugs or treatments, the inconvenience, and the costs.

The perspective in which information is presented. Different emotional responses are elicited when the same information is cast in a positive light and in a negative light. For example, the following messages give the same information, but many people find the first message scarier:

• 9 out of 1,000 fifty-year-old men will die of cancer in the next 10 years.
• 991 out of 1,000 fifty-year-old men will not die of cancer in the next 10 years.

Your chance of experiencing some outcome with an intervention. In a randomized trial, the modified risk is the chance that someone in the intervention group experiences the outcome.

The top number in a fraction. For example, in the following fraction, 10 is the numerator: $\frac{10}{250}$

The event under consideration. Outcomes can include death, both death from all causes combined and death from a specific cause, such as breast cancer or a heart attack (referred to as disease-specific mortality). Types of outcomes include those that people experience directly (patient outcomes), such as needing an operation or being hospitalized, as well as those that are measured in blood tests or X-rays (surrogate outcomes).

Comparative information that can help you make a judgment about the magnitude of a risk. For example, knowing that at age 65 a woman who has never smoked has a 5 in 1,000 chance of dying from lung cancer in the next 10 years is much more meaningful if you know how that risk compares to other risks (a 25 in 1,000 chance of dying from a heart attack, a 3 in 1,000 chance of accidental death).

An inert substance, sometimes called a “sugar pill” (although it isn’t necessarily made from sugar). Placebos are often used in randomized trials to test an intervention. For example, if researchers want to test whether drug × reduces the risk of catching a cold, they can randomly assign participants to either a group who will take drug × or a group who will take an identical-looking but ineffective placebo. At the end of the study, the researchers compare how often people in the two groups caught colds.

The purpose of a placebo is to help ensure that patients in each study group are treated in exactly the same way. Without a placebo group, everyone would know which patients were getting the drug under investigation. Those patients might be treated differently—or might report their information differently—and this could bias the results.

An experiment in which study participants are assigned to study groups solely on the basis of chance—essentially by flipping a coin. This method is the best way to ensure that participants in one group are very similar to those in the other group. The findings from randomized trials are the results that doctors (and you) should trust the most.

A relative comparison of risks: it tells you how much lower the modified risk is relative to the starting risk.

Here’s a way to express this relative risk reduction:

• Drug × lowers the 10-year risk of dying from breast cancer by 33 percent.

The chance that something (good or bad) will happen. Synonyms include absolute risk, chance, and probability.

Screening means testing people who have no symptoms of a disease to look for hidden, early evidence of the disease. Many tests can be used for screening, including mammography for breast cancer, colonoscopy for colon cancer, and PSA testing for prostate cancer.

How a test is used determines whether it is a screening test. When a woman with no signs or symptoms of breast cancer gets an annual mammogram, she is getting a screening test. But when a woman feels a lump in her breast and gets a mammogram, she is getting a diagnostic test in response to a symptom, not a screening test.

Your chance of experiencing some outcome without an intervention. In a randomized trial, the starting risk is the chance that someone in the control (or placebo) group experiences the outcome.

Statistics are numbers that summarize observations about groups of people. For example, they might summarize typical age or weight by taking the average among a group of people. In this book, statistics summarize the probability of different outcomes by looking at the experience of groups of people. Statistics are useful in predicting what is likely to happen in the future. Most of the numbers in this book are statistics.

The proportion of patients diagnosed with a disease who are alive at some fixed time (typically 5 or 10 years) after diagnosis. Although this statistic can tell you your prognosis and is a good outcome measure of how well treatments work in a randomized trial, it is misleading as a measure of how well screening works. Survival rates will increase whenever cancers are diagnosed earlier, even if the time of death is not postponed.