STUDY: New way of calculating odds finds older estimates pessimistic

JOURNAL: Lancet

AUTHORS: Hermann Brenner

ABSTRACT: The chances of surviving many types of cancer are better than statisticians thought, according to a new way of calculating the odds that takes into account improvements in treatment.

---

COMMENTARY: The new approach, proposed by German epidemiologist Hermann Brenner, is commonly used in other areas of medicine, such as predicting life expectancy.

In the Lancet study, Brenner analyzed more than 1.7 million patients recorded in the U.S. National Cancer Institute database.

He found that the new method estimates American breast cancer patients had a 71 percent chance of surviving 15 years, while the conventional approach put the chance at 58 percent.

Similarly, the 15-year survival rate for American men with testicular cancer was 91 percent, compared to 86 percent with the old method.

The survival rate for ovarian cancer five years after diagnosis was 55 percent with the new method, compared with 49 percent previously.

The conventional method of estimating cancer-patient survival, called the cohort approach, estimates the chances of surviving a particular cancer for, say 10 years, by looking at what has happened to patients diagnosed between 1990 and 2000.

“Cohort estimates are generally not appropriate for predicting the survival of newly diagnosed patients since the estimates are heavily weighted toward the survival experience of patients diagnosed many years in the past,” said Paul Dickman, a professor of biostatistics at the Karolinska Institute in Stockholm, Sweden, who is a proponent of the new method but was not connected with the latest study.

The new approach, called period analysis, is based only on recent years — for example, on patients who were alive and under follow-up during the year 2000.