The standards are so high among the Hutchinson Center's Cancer Surveillance System (CSS) staff, encountering a mistake in their work is like finding the proverbial needle in a haystack. The cancer-registry employees face the daunting task of tracking down the cancer cases embedded in more than one million pathology reports every year. As they cull through the reports to find the 200,000 they're searching for, they may miss one or two — in a million.
These expectations of perfection are what keep CSS at the top of their game tracking cancer incidence and survival data. Like Olympians, they want to be the best in the world, and they're achieving that enviable goal year after year through the accuracy of their work, their contributions to countless research projects and by staying at the forefront of technological changes that allow them to do more with less.
Dr. Noel Weiss, of the Public Health Sciences Division, established CSS in 1974. In 1975, CSS became a reporting registry for the newly formed Surveillance, Epidemiology and End Results (SEER) Program of the National Cancer Institute (NCI), the first national cancer-registration system in the United States. Dr. David Thomas, former CSS principal investigator, worked with state officials to make cancer a reportable disease in Washington state in 1992. CSS had already established almost 20 years of collegial, cooperative relationships with the hospitals, radiation centers and pathology departments in their 13-county western Washington coverage area.
SEER collects and publishes data on cancer incidence and survival from 14 population-based cancer registries around the country. The information includes the primary site where the cancer develops, the type of tumor, tumor size and extent of disease at diagnosis and the first course of treatment. These facts and figures provide critical information for cancer researchers and help inform government health care-funding decisions.
"SEER is recognized nationally and internationally as having a very, very high quality cancer-registry system," said Brenda Edwards, associate director of the Surveillance Research Program at NCI. "The Seattle program has always been among the top registries. They're leaders in a registry that's quite distinguished from most registries around the world — they're the best of the best."
Edwards' praise is based on concrete measures of quality in the data its registries collect. Each year, the SEER member programs are judged on the number of cases collected and the consistency and completeness of the data, as well as on performance criteria unknown ahead of time by the registries — somewhat of an invisible, moving target. CSS is the only registry to be twice ranked first and to place in the top three every year since the inception of the Data Quality Profile award.
The 'lab of PHS'
"SEER has 14 evaluation criteria, and we were the first registry ever to meet all of them," said Mary Potts, director of the registry. "We have a fabulous staff and a terrific partnership with the registrar community. Our registry wouldn't be as good as it is without all of these people pulling together."
Led by principal investigators Drs. Tom Vaughan and Steve Schwartz, the registry fosters research in a variety of ways. In addition to tracking cancer trends, CSS is largely responsible for making many Hutchinson Center population-based studies possible. Such studies look at cancer causes, treatment and factors affecting survival. Dubbed the "lab of PHS" by division researcher Dr. Janet Daling, CSS data fuel case-control studies in which cancer patients are compared to the general population to try to identify causes of disease, as well as cohort studies, in which people with certain exposures are monitored over time to see if they get cancer in numbers greater than normal.
Some studies are based on CSS data only, while others use the entire SEER program. Because of the CSS staff's noted efficiency and timeliness, researchers can access regional pathology information within two to six weeks of diagnosis. National data take longer to be released; as a result, CSS can provide data through 2004 while other SEER registries only have complete data from 2002.
Since its inception, CSS has aided hundreds of cancer epidemiology and prevention studies at the Center. More than 500 publications by Center investigators in the past three decades have been based on CSS data. Investigators who utilize the fruits of the registry's labor voice grateful appreciation. "We simply could not do our studies without the complete and very high-quality work that CSS does throughout the region," said Dr. Polly Newcomb, head of the Cancer Prevention Program, who is currently working with CSS on a complicated study of colon cancer in families. The study has been the basis for many other genetic, environmental, economic and behavioral studies related to colon cancer.
"We are very, very fortunate to have this outstanding resource here at the Center," she said. "I have worked with registries in many other states, and without exception, CSS surpasses all others. Because of them, we are able to conduct the very highest quality epidemiological research."
Abstractors in the field
It takes a lot of detailed steps to track a cancer case from diagnosis to its inclusion in the CSS database. When a patient is discharged from a hospital, the institution assigns a code to every disease the patient had treated at their facility. Any cancer-related code is checked by the CSS case-finders. They investigate whether the cancer has ever been reported before and whether the data in the registry is complete. SEER expects its registries to maintain 99 percent accuracy in case identification; CSS exceeds this standard by averaging one to three missed out of 30,000 to 40,000 cases.
The abstractors, or "field staff," spend most of their time away from the Center, gathering all of the medical information available for each cancer patient from the 98 facilities in the CSS reporting area and summarizing the patient's demographic data, cancer details and treatment. "The abstractors take a medical record — no matter how big it is — and give us a one- or two-page version," Potts said. "Data-reporting requirements have become more complex over time while medical records are getting skimpier, which requires more follow-up."
After the record is distilled into key elements, it is assigned specific cancer codes based on SEER-coding schemes that can change each year, depending on the cancer site. While most new jobs come with a learning curve, for CSS coders, gaining competence is a long and winding road. It takes up to a year for new coding staff to learn the basics of the complex work and a couple more years to hone those skills.
The final step for the registry is the entry of the cancer information into the database by the dataflow staff. Along the way, they ensure that each individual and cancer case is only reported once.
SEER also requires its registries to track all cancer patients in order to glean survival data. Many people survive or live longer with cancer today and relocate more often, which has made the work grow exponentially. CSS currently has more than 200,000 cases on active follow-up.
Through the years, CSS has been forced to keep up with an ever-increasing workload without additional staffing. In the early days, there were less than 9,000 cases; they now record about 22,000 cases annually. Technology has been key to meeting the demands.
To that end, CSS relies on the PHS Epidemiology/Information Technology team. One example of their partnership is E-Path, a program created by software developer Franz Hernandez. It uses keyword searches to help the case-finders find cancer-related pathology reports, decreasing the number of reports needing online visual review by 80 percent. "We were pioneers of the keyword-search concept with E-Path," said Technical Operations supervisor Jennifer Hafterson, who adds that the software is also a valuable training tool.
The increase of cancers being treated on an outpatient basis in doctors' offices drove the demand for a way to generate physician queries asking about details for each type of cancer. Potts worked with now-retired software developer Paul Maier to develop a program that merged NCI's physician data-query (PDQ) system with needed treatment questions. The software greatly improved treatment data collection by generating site-specific treatment questions for reporting physicians. SEER plans to incorporate PDQ's logic into its new database management system.
The high-achieving CSS staff is often tapped to give talks at national cancer-registry meetings, train other national and international registry personnel and even write textbook chapters on their work. "The Seattle group has long been recognized as leaders," Edwards said. "Excellent performance and productivity is a hallmark of their team, and it pays off in their consistent, high quality of data and in getting things done on time and right."