Cancer Studies and Statistics: What Do the Terms Mean?

As you learn all you can about treatments for your type of cancer, making sense of the results of cancer studies can be daunting. But understanding a few basic terms can help.

Statistical Significance
When researchers conduct studies on potential treatments, they must determine if the results they observe are in fact due to the treatment, or are simply the result of chance. They start with the assumption that there's no difference in the effectiveness of treatments until they gather reliable evidence to the contrary.

With study results in hand, researchers must then calculate the statistical significance of their findings. Since studies only include a portion of all those in the population (e.g. all breast cancer patients), researchers can't be 100 percent sure the results are due to the treatment, so they assign a value (called p-value) to represent the likelihood their observations are the result of the treatment and not of chance. P-values range from 0 (little likelihood results are due to chance) to 1 (very likely).

In medical studies, scientists generally select a p-value of .05 to indicate statistical significance. They use the p-value to state their confidence the findings would reflect the true effect of treatment if they could include everyone. For example, if the study shows the overall survival rate of patients receiving the cancer treatment averages 81 percent, with a p-value of .05, the researchers are 95 percent confident

(1 - .05 = .95, or 95%) that 78 to 83 percent of the population would survive following this treatment.

Statistical significance demonstrates an association between treatment and outcomes, but doesn't necessarily confirm cause and effect. Furthermore, a treatment deemed scientifically significant may not be meaningful to you. For example, you may decide the side effects are not worth the possible benefits of this treatment.

Number Needed to Treat (NNT)
A given treatment will not help every cancer patient; some may be helped, while others may not experience any difference, or even be harmed. We don't know which patients will fall into each category, so NNT tells us how many people we must treat to help just one person. Ideally, NNT would be one: if you treat one person, you help one person. Researchers also use NNT to determine how many people they must screen for one to benefit from cancer screening. The lower the NNT, the better.

Evaluating statistical significance and NNT will help you gauge how likely a treatment is to benefit you.


Sources: "Research Statistics." Web . 26 January 2011. "The Significance of Statistical Significance." Web. 21 January 2001.

Vachani, Carolyn. "Interpreting a Cancer Research Study." Oncolink. Web. 8 June 2009.

The NNT. "The NNT Explained." Web.