Mapping the links between statistics and health

For Lance Waller, finding unusual connections in science isn't the end of the story—it's just the beginning.

By Mike King


Caring for Haiti

Stories from Haiti

Certificate program prepares students for disasters like the Haitian earthquake

Gary Miller's mice have Parkinson's. Can he cure them?

A weighty matter

The maps dominating Lance Waller's office represent more than just interesting artwork. For the chairman of the Department of Biostatistics and Bioinformatics, the dozens of maps provide a daily reminder of early work in a field of study that may only now be reaching its true place in preventing disease and improving health.

The largest, commanding much of the wall behind his desk, is the self-described “Map of the Hydrographical Basin of the Upper Mississippi” by geographer and mathematician Joseph Nicollet.

Printed in 1843, the map represents the first detailed rendering of the hydrology system for a landmass in North America that is half the size of Europe. The French scientist’s remarkably accurate journals, among other things, were used to chart Minnesota’s famed 10,000 lakes. But it doesn’t include one of the largest, Lake Minnetonka—an omission that was likely the result of Native American tribal guides not wanting the geographer to know about their sacred burial grounds nearby.

“You can’t account for what you don’t know exists,” Waller says of the significance of Nicollet’s map.

On a shelf opposite the map, Waller has earmarked a page from a book on John Snow, one of the pioneers of epidemiology and the field now known as “spatial statistics.” It depicts a miniature version of the 1854 map Snow had made pinpointing the proximity of the dead to a Broad Street pump in London. Snow’s discovery linking the source of contaminated water to the spread of cholera fell on deaf ears, and more than 600 residents died.

“Finding an unusual connection isn’t the end of the story, it’s just the beginning,” says Waller of the science that forms his life’s work.

The rudimentary statistical analysis that Snow applied more than 150 years ago has evolved into use of Geographic Information Systems (GIS), data mining studies, and other sophisticated tools to count, analyze, map, and predict health threats that may—like Lake Minnetonka to Joseph Nicollet—otherwise never have been noticed.

Waller has been at the forefront of many of these changes. The youthful-looking Rollins Professor is among a new generation of biostatisticians who have pushed the once behind-the-scenes use of bioinformatics front and center in epidemiology and, increasingly, as a major component of biomedical research and health care reform.

As associate director of the Center for Comprehensive Informatics (CCI), based in the Woodruff Health Sciences Center, he provides an important link with other scientists. The CCI fosters collaboration among software and biostatistics experts and basic science and clinical researchers at Emory and Georgia Tech.

“We are in a new age of biomedical research where technological breakthroughs allow the collection of data on vast scales, such as medical imaging, gene expression arrays, and sequencing,” Waller explains. “The CCI represents a focused effort to bring together these types of experts with research teams to make the most of newly available data.”

Mining data, drawing conclusions

The son of a statistician—his father worked at the Los Alamos National Laboratory studying the reliability and life span of mechanical components—Waller has always been drawn to the field. He majored in mathematics and minored in ­computer science as an undergraduate but got hooked on spatial statistics when an academic advisor led a project that looked at disease clustering.

In the years since, much of Waller’s work has been directed at linking spatial statistics, GIS information, and statistical assessments of environmental justice and creating new and more sophisticated models for small-area health statistics. More recently, Waller has concentrated on spatial point process methods for alcohol epidemiology, conservation biology, and disease ecology.

From the study of sea turtle nesting patterns in the Southeast to looking at geographic links between alcohol sales, crime, and drug use in Houston, Waller’s research has helped create policies and procedures that, among other goals, promote conservation and reduce the risk of crime.

“What we can do with mapping is look at how things line up and draw some conclusions about, for instance, zoning ordinances regarding where alcohol licenses might be problematic,” he explains.

Similarly, data can be mined from traffic flow patterns; emergency room visits for allergy, asthma, and other respiratory diseases; and proximity to highways or clogged arterial streets. That information can lead to better decisions about the impact of new highways or traffic corridors on public health in urban areas—the way environmental impact statements have informed public officials about land use practices in rural regions.

One of his studies involved the spread of rabies among raccoons from Florida to Georgia and up the eastern seaboard to Virginia and West Virginia. Since rabies is a reportable disease, researchers were able to map the spread and even slow the spread by predicting where it might go next. (They baited the woods with feed that contained a vaccine.) “It was like fighting a forest fire. They put down a line of defense that worked,” says Waller. He currently is developing statistical methods to analyze spatial data in disease ecology, using raccoon rabies in the northeastern United States and Buruli ulcer in Ghana as case studies.

Such practical applications of bioinformatics have been important in the fight against the spread among humans of animal pathogens like monkey pox, West Nile virus, SARS, and swine flu. The success thus far at keeping H1N1 in check can be directly linked, at least in part, to timely, accurate assessments of how the virus has spread, especially among children and college students, Waller believes.

With 30 faculty members, the department he leads is similar in size to those in public health schools around the country, and like the others, it is growing. Faculty collaborate with Emory researchers on projects such as developing new ways to monitor cancer cell growth and new methods of analyzing survival patterns in cancer patients. Some of their most hopeful lines of study involve the Center for Biomedical Imaging Statistics, the Emory Center for AIDS Research, and the Atlanta Clinical and Translational Science Institute.

Waller took over the department last July, succeeding Michael Kutner, who joined Emory in 1971, when the department was part of Emory’s School of Medicine. Kutner, who continues to teach and conduct research, aided in recruiting Waller from the University of Minnesota in 1998.

The careers of both men represent succeeding generations of biostatisticians. Their profession grew in stature during the 1960s and 1970s when the NIH greatly expanded training programs and provided scholarships and tuition assistance for students interested in the field at medical schools across the country. That stimulus helped the image of biostatisticians evolve from technicians and analysts to their more appropriate role as full-fledged researchers in clinical medicine and public health.

An explosion of new data

As clinical trials of new drugs for cancer, HIV, and other diseases exploded in the 1980s, it became increasingly clear that accurate, meaningful statistical analysis is essential to the successful outcome of large-scale studies. The quest to find drugs that could combat the death and disability associated with HIV/AIDS in particular marked a turning point for biostatistics, Waller believes.

With the death toll from AIDS rising monthly and an increasing demand to enroll patients in trials testing dozens of anti-viral drugs, researchers needed to know quickly which drugs were harmful, which showed potential, and which needed more study. Collecting and analyzing data from experiments around the country became paramount.

Harvard researchers were working with large hospitals that were treating AIDS patients, while the University of Minnesota—where Waller was a faculty member at the time—worked with physicians who were monitoring AIDS patients in their family practices.

Using what Waller calls “a very detailed design” for studying the data collected, and other tools of the trade, biostatisticians could quickly measure the effectiveness of the “cocktail” of anti-retroviral drugs that became the gold standard for people infected with HIV.

One goal of biostatistics, Waller says, is to help design clinical trials “where you don’t have to wait for all the results to be in before you know whether something works or not.”

In recent years, researchers have shut down experiments with investigational drugs that were likely to cause harm based on preliminary data analyzed by biostatisticians. But they also have determined quickly that some experimental drugs are both safe and effective and shortened the length of time needed to move them from the first tests in humans to FDA approval. Such trials may still be rare, he says, but without timely, well-designed data gathering, the information gleaned from them could take years to accumulate, leading to unnecessary risks for some patients and missed treatment opportunities for others.

As their role in clinical studies grew, biostatistics experts were thrust into evaluating the cost-effectiveness of drugs, medical devices, and other therapies. The field, now known as comparative effectiveness research, has been touted as one of the best ways to bend the curve on health care spending.

Waller agrees that he and his colleagues will play an essential role in such research. He also knows that such research can cause controversy. A recent example: The value of screening mammography for women was based, in part, on comparative effectiveness analysis that was accurate, even though the conclusions derived from it, such as what age women should be screened, varied.

“Unfortunately, we can’t guarantee an average outcome for everybody,” Waller says. The goal should be to determine what to measure, how to measure it, and then test all conclusions as thoroughly as possible. It’s a challenge he relishes as RSPH biostaticians analyze and link data in a variety of applications across Emory.

That may be why Waller feels such kinship for his predecessor’s work in London and why the little map in the John Snow book speaks so eloquently more than 150 years later.

Mike King is a former medical writer and editor with The Atlanta-Journal Constitution.

Table of Contents

Bookmark and Share

spring 2010 cover