Survival & Mortality: Measuring the Burden of Cancer
Survival statistics and mortality statistics may seem like opposite sides of the same coin, but they provide distinct types of information. One is tied to death, and the other to a diagnosis of cancer. Mortality is a count of cancer deaths in the population during a given calendar period. Survival measures how long a person is alive after diagnosis.
Clinicians and patients tend to be interested in survival statistics because they contain information relevant to prognosis and treatment. NCI's Surveillance, Epidemiology, and End Results (SEER) program has collected data on survival by cancer type and extent of disease for decades. Today, the information can guide both clinicians and patients.
"Survival statistics are being used to help inform patients about their disease," says Dr. Brenda K. Edwards, associate director of the Surveillance Research Program in NCI's DCCPS.
"We know more today about identifying, diagnosing, and characterizing cancer, and this complicates our comparisons of survival statistics for some cancers over time," she continues. "But knowing more about the disease and how to treat it will lead to better outcomes for patients."
For instance, a diagnosis of early-stage breast cancer today may not be the same as one a decade ago because of advances such as the discovery of the HER2 gene's role in some breast cancers and the development of trastuzumab to treat HER2-positive disease.
Clinical trials often use survival to measure the effects of an intervention on individuals randomly assigned to one group or the other.
"Survival is a metric we use in treatment studies because everyone is the same at the beginning of the study, and then you give them drug A or drug B," says Dr. Peter Bach of Memorial Sloan-Kettering Cancer Center. "This is reasonable because you assume no biases at the beginning."
While survival data have many applications, they have to be considered carefully. Survival is keyed to the year of diagnosis, and anything that advances the time of diagnosis, such as screening programs and improved access to care, may make survival look better than it really is.
Detecting cancers that would never have caused any harm, known as overdiagnosis, can inflate survival rates. For example, the widespread introduction of prostate-specific antigen testing for prostate cancer has likely led to overdiagnosis and artificially boosted prostate cancer survival rates.
"Survival statistics have limitations and should be used with caution," notes Dr. Edwards. "You need to know what survival tells us and what it doesn't."
Some experts caution against using survival as a measure of progress. Comparing survival across time or between geographic locations can be problematic because patterns of diagnosis are likely to differ.
These differences change the timing of diagnosis and may make the comparisons invalid, notes Dr. Steven Woloshin of the Dartmouth Medical School, who has written about survival and mortality rates.
"Five-year survival rates can provide useful information about prognosis," says Dr. Woloshin. "But 5-five-year survival is not a reliable metric for showing whether there has been progress against cancer."
As a panel of experts convened by NCI concluded in 1990, mortality rates are the most important measure of progress against cancer. But like survival rates, they can be problematic. The accuracy of mortality rates depends on how death is determined, and this may be complicated if a patient has other health conditions such as diabetes or dies from another cause.
Nonetheless, says Dr. Woloshin, "there's no question that of all the measures we have of measuring success in cancer, mortality is the one to hang your hat on."
—Edward R. Winstead