Findings from a new cancer study may seem obvious — a patient’s overall health is a critical factor in assessing prognosis. But the study’s authors say data about patients’ non-cancer ailments, called comorbidities, are not currently incorporated into cancer statistics.
The study will appear in the May 26 issue of the Journal of the American Medical Association.
Physicians intuitively use a patient’s overall health when determining the best course of treatment, but records used to track and study cancer data focus on tumor size and typically ignore comorbidity statistics, according to principal investigator Jay F. Piccirillo, M.D. Those omissions put both clinical trials and cancer physicians at a disadvantage.
“Only 5 percent of adults with cancer are treated in clinical trials, so many treatments are assessed from the study of routine patient care information recorded in cancer registries rather than through controlled clinical trials,” says Piccirillo, who is an associate professor of otolaryngology and of medicine at Washington University School of Medicine and Siteman Cancer Center in St. Louis. “We therefore need to incorporate comorbidity statistics into cancer registry records to make our data more accurate and, by extension, to help clinicians determine the best treatments for cancer patients.”
In 1995, Piccirillo and his colleagues trained cancer registrars to record comorbidity information during the normal process of documenting cancer cases at the School of Medicine’s clinical affiliate, Barnes-Jewish Hospital. The process incorporates the 27 most common comorbid ailments, including the severity level of these conditions on a four-point scale. Registrars combine this information with data they already collect, including tumor size and type.
Their latest study presents findings based on the more than 17,700 patients in the Barnes-Jewish Hospital cancer registry whose comorbidity information was recorded between 1995 and 2001. Patients had one of several types of cancer, including prostate, lung, breast, gynecological, colon or head and neck cancers.
The data revealed that the more severe an individual’s comorbidity, the worse his or her chances were for survival. Overall, patients with severe comorbidities were 24 percent less likely to be alive after three years than patients with no comorbidities. Their overall survival rate was 33 percent less after five years.
Comorbidity also was correlated with the likelihood that a patient’s cancer would return. For example, patients with severe comorbidities were 15 percent more likely to have a cancer recurrence than those with no comorbidities.
The researchers found that comorbidity information was most useful in more treatable types of tumors, like in prostate and breast cancer, and less helpful in sites that have very poor chances of survival, like lung cancer.
“The bottom line is that comorbidity data provides critical additional information for determining a patient’s prognosis,” Piccirillo says.
He believes recording comorbidities will have two key benefits. First, it will improve the accuracy of computer-based programs that estimate survival so patients receive more accurate information on their chances for survival.
Second, it will improve the validity of cancer research. For example, it may not be appropriate to include all cancer patients in a given drug trial, because comorbidities may influence drug effectiveness. Similarly, because some hospitals tend to see sicker patients, comorbidities can cloud comparisons of quality-of-care assessments across different institutions.
“Our ultimate goal is to be able to individualize cancer therapy,” Piccirillo explains. “To do so, we need to keep in mind the big picture, and comorbidities are an important part of that.”
Recognizing the importance of comorbidity information in determining treatment and prognosis, the American College of Surgeons’ Commission on Cancer recently mandated the collection of this information based on the ICD-9 coding system used for medical bills. This approach relies on billing procedures already in use and does not require additional training, but Piccirillo says it has several key flaws.
First, it does not include information about disease severity. An individual with mild diabetes controlled by dietary restrictions receives the same code as an individual with severe diabetes taking daily insulin shots. Among other potential drawbacks, medical bills also may omit certain diseases for socio-political reasons (for example, mental illness or AIDS).
In 1996, Piccirillo’s team received a grant from the National Cancer Institute to develop a training program based on the comorbidity data-collection procedure implemented at Barnes-Jewish Hospital. The Comorbidity Education Program includes a training manual, video and data collection forms.
After improvement and refinement during the first several years of its use, the program now takes only 10 hours to learn. In a previous study, the team showed that registrars at Barnes-Jewish Hospital and four other centers around the country successfully learned and implemented the program with relative ease.
The team currently is finalizing a web-based training program based on their experiences with in-person seminars. The United Kingdom’s National Health Service is planning to incorporate this program into its current cancer registry system.
“The Commission on Cancer based their approach to comorbidity on the theory that training registrars on a new coding system is time-consuming and impractical,” Piccirillo explains. “But we’ve taken one example of a new method and we’ve studied it, we’ve documented it and we’ve published it. Cancer registrars can learn comorbidity coding. And, as our latest study shows, that information is an important factor in understanding prognosis.”