Positive and negative predictive value

Predictive value in screening

Cancer screening can be confusing and overwhelming, and there is an abundance of data available on the various screening methods, often without clarity on what that data means. Sensitivity and specificity is one important metric of any screening test, but the likelihood that an individual test result–whether positive or negative–is accurate is more complex than those values. In order to determine that likelihood, we use positive and negative predictive values.

In technical terms, the positive predictive value (PPV) is the proportion of positive results in diagnostic tests that are true positives. If an individual were to test positive using a cancer screening test that has a 75% PPV, there is a 75% chance that the individual does in fact have cancer. Conversely, negative predictive value (NPV) is used to measure confidence in negative results, i.e. true negatives.

It’s clear why this is an important concept to keep in mind when undertaking cancer screening, as it adds vital context to test results. A screening exam with just a 30% PPV can give valuable information, but 70% of individuals who receive a positive result do not actually have cancer. 

PPV and NPV combine information about test sensitivity and specificity with the overall prevalence of a specific disease in the population. Mathematically, they’re calculated as: 

‍

Let’s take an example from breast cancer.  The most basic form of breast cancer screening for individuals at average risk is a mammogram. This test has 72% sensitivity and 98% specificity. The average incidence of breast cancer at age 45 is approximately 0.2% (i.e., 2 out of a thousand  45-year olds in the United States can expect to be accurately diagnosed with breast cancer). That means the NPV and PPV of a mammogram for someone at average risk are: 

PPV = 72% x 0.2% / [ (72% x 0.2%) + (1-98%) x (1-0.2%) ]= 6.7%

NPV = 98% x (1-0.2%) / [ 98% x (1-0.2%) + (1-72%) x 0.2%] = 99.9%

This means that if a 45-year old woman at average risk for breast cancer receives a positive result on a mammogram, there is only a 6.7% chance they in fact have breast cancer. If they receive a negative result, there is a 99.9% chance they do not in fact have the disease–overwhelming odds would say this was a true negative.

However, PPV and NPV change as disease prevalence changes, and disease prevalence is based on personalized risk factors. That’s where the Catch risk assessment comes into play.

For a 45-year-old woman who works as a flight attendant and drinks an average of 2 drinks per day, personalized disease risk may be higher than the population average of 0.2%; if it were to jump to 0.4%, the PPV and NPV of a mammogram shift to 12.4% and 99.8% respectively. A positive result would be a bit more concerning, and a negative result should not connote quite as much confidence. 

At Catch, we provide this vital context for all the screening tests we recommend, which can help you not only interpret your own test results, it can help you make informed decisions about frequency of testing, which tests to pursue, and your own comfort level with the results you receive. Providing the information you need to make informed decisions, and using your personalized risk assessment to recommend a screening plan, will give you the best chance of catching cancer early.