ayden Mullen fought with cancer for ten years, undergoing surgery and chemotherapy treatments multiple times before he died in the intensive care unit of a medium-sized hospital in 1982. The cause of death li sted on Mullen's death certificate read "cardiac arrest."

Historically, epidemiologists have relied on the mortality information recorded on death certificates to determine the pervasiveness of cancer in our society. However, according to Jonathan Liff, professor of epidemiology, there are great limitations t o using mortality data for that purpose. In some cases, such as Mullen's, cancer may not be listed as cause of death, although it may be a contributing factor or even the underlying cause. In other cases, death certificates may mislead researchers, causin g them to dramatically underestimate the problem. For example, over 80% of women diagnosed with breast cancer are still alive five years later.

Since the 1970s, cancer epidemiologists have turned away from using only mortality data to track cancer, instead creating registries to examine the extent of the disease problem. These registries monitor the incidence of cancer in a defined population, identify populations with unusually high risks of cancer, and determine the survival rates for people diagnosed with cancer.

In 1973, the National Cancer Institute established the SEER registry program. SEER stands for surveillance (monitoring the rates and types of cancer), epidemiology (further investigation of those cancers), and end results (survival rates of individuals after they are diagnosed). Currently SEER operates throughout five states - Connecticut, Hawaii, Iowa, New Mexico, Utah - and six major metropolitan areas, including Detroit, Los Angeles, San Francisco/Oakland, San Jose, Seattle, and Atlanta. The Georgia Center for Cancer Statistics, directed by Dr. Liff, is the operations hub for the Atlanta SEER registry as well as a smaller SEER registry in rural Georgia.

The geographic areas included in the national program offer a balance of settings in which to study cancer prevalence. Iowa, for instance, has predominantly a rural population, whereas Utah is home to a large number of Mormons, whose prohibitions on sm oking and drinking result in low rates of cancer. By contrast, Atlanta is the only SEER registry in the Southeast and one of the few registries that includes a large African-American population as well as groups of people whose socioeconomic status is wid e-ranging.

The national SEER registry covers about 14% of the US population, reporting between 97% and 99% of all cases in the designated geographical areas. Those numbers translate to approximately 2 million cancer cases recorded through 1991. The figure will be much higher when data from Los Angeles, whose registry is one third the size of the other sites combined, are added retroactively.

In addition to overseeing the Atlanta and rural Georgia SEER registries, the Georgia Center for Cancer Statistics operates two additional registries: the Savannah River Region Health Information System (SRRHIS) and a statewide registry of Georgia, spon sored by the state's Department of Human Resources. The SRRHIS registry, funded by the Department of Energy, is a collaboration between the Center and the Medical University of South Carolina and includes counties in both South Carolina and Georgia. The r egistry is a response by the Department of Energy to citizen concerns about possible harmful effects of a nuclear plant in Aiken, South Carolina. "This collaboration between states in a cancer registry is unprecedented," says Dr. Liff, "and overcomes many of the problems state registries encounter."

For example, many South Carolina residents come to Georgia for cancer diagnosis and treatment. But South Carolina epidemiologists do not have authority to retrieve data from Georgia hospitals. Therefore, they have no way to track cancers among South Ca rolina residents that are diagnosed outside their state. Because of the collaboration, Emory has the ability to identify cancers diagnosed in South Carolina hospitals, as the Medical University of South Carolina does in Georgia. Both the SEER and Savannah River registries are what Dr. Liff calls "comprehensive" lists: their goal is to identify and track every person diagnosed with cancer in those areas. Because of current funding, the statewide registry, however, is not comprehen sive at the present time. That effort gathers information from individual hospital registries throughout the state and consolidates the data.

Each year, the Georgia Center for Cancer Statistics processes more than 21,000 cancer reports among Georgia residents, including 10,000 for SEER, 2,500 for the SRRHIS, and some 8,600 cases reported to the statewide registry from other Georgia counties. If recent funding from the Centers for Disease Control and Prevention is matched by funds from the Georgia legislature, identification of cases in all 159 counties will be possible. Despite the large numbers, the staff reviews the cases for the comprehen sive registries one at a time. "In the comprehensive registries, we must process each case individually," Dr. Liff says.

Field workers leave here to collect information from hospitals and pathology laboratories and bring the data back for analysis. In examining medical records, these abstractors look at the type of tumor diagnosed, its location in the body, the advanceme nt of disease at the time of diagnosis, the cancer's behavior - whether it is aggressive or slow-growing - and information on the first course of treatment. Using sophisticated software developed at the Center, they enter those data onto laptop computers and complete what Dr. Liff refers to as "the first edit." During this editing, abstractors review the collected information for completeness, reliability, and consistency. If any problems appear, they can review records immediately instead of waiting unti l they return to their offices.

Back at home base, an editing staff takes over, reviewing each case individually and comparing the new data to information from the registry database. During this step, editors often discover case duplication or potential errors. For complicated or unu sual cases, they call in Dr. Bill Eley, an experienced medical oncologist who can help them make decisions about ambiguous data - for example, whether a cancer should be categorized as a new or recurrent cancer.

After the staff downloads the information to the database, they subject the data to yet another series of examinations, trying to produce the most accurate registry possible. They also check annually to see if cases on the registries match with records of the Georgia Death Certificate Data File, county voter registration files, hospital tumor registries, physicians' office records, or other sources. By following all patients until death, epidemiologists can determine survival rates for different kinds of cancer.

Each year, the Center provides its compiled data on a computer tape to the National Cancer Institute, which combines this information with SEER figures from other geographic areas to produce a book of statistics examining 20 to 30 types of cancer. Phys icians, cancer researchers, and epidemiologists use the guide as a primary source of information about cancer in the United States. Politicians and their staff as well as National Cancer Institute personnel examine it to measure progress in the fight agai nst cancer and to determine governmental priorities.

This study is only one example of how researchers can tap into registry data to reveal startling trends. Those trends, however, may be misleading, says Dr. Liff. For example, rates of prostate cancer seem to have increased dramatically in the 1990s. Th ese higher statistics may reflect earlier diagnosis due to a more accurate screening test, introduced in 1987, rather than a true increase in the occurrence of prostate cancer.

"Part of the registry's job is to facilitate research in cancer epidemiology," Dr. Liff says. Currently, faculty in the Department of Epidemiology and staff at the Georgia Center for Cancer Statistics conduct three principal types of investigations in addition to surveillance: those of cancer causes, prognosis, and control. For example, they are examining whether certain treatments for breast cancer put women at risk for uterine cancer. They have investigated the barriers to early detection of cervical cancer - in other words, why some women do not get Pap smears more often. And they are comparing differences in the stage of disease at diagnosis between residents of urban and rural areas in Georgia.

Sometimes the results of such research may take years to complete. To evaluate the effect of mammography screening on breast cancer mortality rates, for instance, may involve many years of research. On the other hand, results from the study of early de tection of cervical cancer can be available in as few as three years.

Whether investigations into cancer can be completed in months or expand to fill decades, the registries coordinated by the Georgia Center for Cancer Statistics will remain a source not only for counting cancer cases but also for understanding the far-r eaching scope of the disease.