Public Health, Fall 1997



Trials of Life
Biostatisticians end their particular talents to evaluations of a newly refined treatment called pallidotomy that brings hope to those with Parkinson's disease

by Jeanie Davis

When researchers develop new procedures to counteract devastating disease, how do they determine if their experimental ideas are helping or harming patients? How do they really know if the new treatment is mor e effective or safer than the old? How do clinicians and surgeons exploring these new options judge the procedures scientifically and objectively? How do they ensure that seemingly promising results are not skewed by unrecognized factors? The randomized c linical trial is the answer to these questions.

A case in point: In the surgical suites of Emory University Hospital, neurosurgeons are perfecting a technique that may change how some of the nearly 1 million Americans afflicted with Parkinson's disease are treated. Called pallidotomy, it involves cr eating a lesion deep within the brain in difficult-to-locate areas in the globus pallidus. A team--led by Emory Department of Neurology Chairman Mahlon DeLong and including neurologist and neurophysiologist Jerrold Vitek and neurosurgeon Roy Bakay--is pio neering a brain-mapping system guided by a microelectrode, which allows surgeons to precisely locate certain areas within the globus pallidus. Once they pinpoint the location, individualized for each patient, surgeons can remove overactive cells, thereby decreasing the tremors, stiffness, and other movement problems characteristic of Parkinson's. But as they proceed, it is vital that they assess their efforts.

That's where the Department of Biostatistics at the Rollins School of Public Health comes in. In 1994, with the help of the department's biostatisticians, the School of Medicine researchers received a $2.5 million grant from the National Institutes of Health (NIH) to conduct a long-term (four-year) study of the pallidotomy procedure.

Such clinical trials bring hope to many for whom standard treatments have failed, according to Vicki Hertzberg, chair of the Department of Biostatistics at the school. "They are crucial in evaluating the therapeutic efficacy of a treatment," Hertzberg says. "To these studies, biostatisticians bring the ability to work through scientific questions in a systematic manner so they can be answered quantifiably and with potential sources of bias being removed to the highest possible degree."

While physician investigators bring their own research, clinical, and surgical expertise to clinical trials, biostatisticians are the scientists responsible for determining if results from those trials are statistically accurate. They develop the resea rch design and analytic models necessary to accomplish that goal.

Two faculty members from the Department of Biostatistics, Huiman X. Barnhart and Michael Haber, have collaborated with the neurosurgery research team since the pallidotomy study's inception. To ensure good statistical results, they assisted in the plan ning and implementation phases of the study. They also developed and applied complex statistical and mathematical models to determine the efficacy of the experimental treatment.




Huiman X. Barnhart

Going into the clinical trial's design, Barnhart and Haber, along with their neurology colleagues, analyzed results from the pilot study, a longitudinal investigation in which all patients received treatment and monitoring of their progress. Noting a need for comparison, the team recommended a randomized trial with a control group composed of patients who would not receive surgery immediately. "While it is possible to perform a surgical procedure and simply m onitor its effect on patients, you won't know if the improvement you see is due to the procedure itself or to another factor that is not immediately evident," says Barnhart.

In this study, control group patients are permitted to cross over into the "procedure group" six months after entering the trial. While the control group is necessary for comparative purposes, "for ethical reasons, it was important that we give the tre atment as soon as possible if it might be beneficial to the patient," Barnhart adds.

Considerable data on each patient's progress--nearly 700 variables regarding overall physical, neuropsychological, and psychiatric status--are collected during each biannual visit to Emory University Hospital's Clinical Research Center. These data make a thick stack of forms, but to researchers they represent a valuable opportunity to collect a large body of information that will enhance existing knowledge. To further ensure accuracy of results, a data manager uses a double-entry system that effectivel y weeds out entry errors.

When evaluating data, the team focuses primarily on these issues: When patients have experienced statistical and clinical improvement, is the improvement due to pallidotomy? Is the improvement still evident on follow-up visits? How do the patient's age , onset of symptoms, or other factors affect outcome? The answers they have uncovered so far have opened doors for many more patients to receive the groundbreaking surgery.

Anatomy of a clinical trial



Michael Haber

Haber brings up a critical issue in data analysis--the confounding of data. "Anyone can run numbers," he says, "but it's another thing to understand what's going on and to interpret them." A data entry error could produce a strange answer, or other variables may skew the analysis. "For instance," Haber says, "if the surgery patients happened to be younger than patients in the control group, we would need to see if that biased the results. Age may be related t o the patient's outcome or overall improvement. We would have to run a regression model--controlling for age--so it wouldn't make a difference."

Quality control over each aspect of the clinical trial is another challenge, says Barnhart. Every six months, the data safety and monitoring committee--composed of NIH officers, a neurologist, surgeon, and statistician--meets to review the study's prog ress and take an interim look at data. "When we look at data, clinical investigators have to leave the room," Barnhart says. "It's an integral part of the clinical trial. If they see how patients are progressing, it could influence their treatment of that patient." For example, physicians might want to add to or remove a patient from the study based on these discussions.

"When you finish the study," Barnhart says, "when you publish your results and others want to use this procedure as a guide, you won't know whether the pallidotomy was truly successful or whether your results come from removing patients from the study. By defining and following your target population, you make the study applicable in the future."

As statisticians in the health sciences, both Haber and Barnhart bring a wealth of experience to the pallidotomy clinical trial. Since coming to Emory in 1992, Barnhart has worked with Emory's Angioplasty vs. Surgery Trial (EAST), which compares corona ry angioplasty with coronary bypass surgery to determine outcomes of these treatments in patients referred for revascularization. She also has worked with the Centers for Disease Control and Prevention (CDC) in analyzing the progression of HIV disease fro m birth to death among a large number of perinatally infected children, a study which underscored these children's need for clinical care before the onset of full-blown AIDS.

Recently, Barnhart consulted on another Emory study of the effectiveness of two exercise programs--Tai Chi and a more expensive computerized balance training program--to reduce frailty and injuries in the geriatric population. The study, which found Ta i Chi to be the more beneficial program, won an Excellence in Research award from the American Physical Therapy Association.

As with most biostatisticians, Barnhart believes her work affords an opportunity to apply her mathematical aptitude and statistical expertise to daily life problems. "I wanted to make a contribution to society, not just work on abstract mathematical pr oblems," she says.

For Haber, the quest to understand data and investigate relationships among variables has spanned a 30-year career. For the past several years, he has explored vaccine efficacy, analyzing data on a 1991 measles outbreak in Burundi. In 1995, he won a Fu lbright Award to conduct research on statistical methods for infectious disease, working with a British research group to design and analyze data from vaccine trials in East Africa.

Currently, he is working with the CDC to evaluate the effectiveness of flu vaccines in day care centers--whether they actually reduce incidence of flu among the children or help prevent spread of the disease to other family members. He is also analyzin g data for a School of Medicine Department of Radiology study evaluating magnetic resonance angiography as a method of noninvasive screening for atherosclerotic carotid artery stenosis.

Voice of experience



A brain-mapping system, guided by a microelectrode, helps surgeons precisely locate the affected area of the globus pallidus, which varies from patient to patient.

Concerning the pallidotomy study, "We decided early on to conduct an interim analysis," says Haber. "If we had enough evidence that pallidotomy was a success, then we would stop the study. We don't want to wi thhold treatment from someone."

After one year, the committee had their evidence, and they stopped the randomized aspect of the trial. They now feel "it is important to continue examining the long-term effects of the procedure," Haber says, "to follow patients for at least two years. "

A second study is already being discussed, as the neurology team seeks funding to explore use of a microelectrode-guided deep stimulation procedure. They have found that "stimulating" brain cells at various sites in the brain, rather than destroying th em, can be effective in further reducing Parkinson's symptoms.

With many unanswered questions remaining about the benefits of pallidotomy and the new deep stimulation procedure, researchers will certainly examine outcomes for many years, further refining their surgical techniques. As the second pallidotomy study t akes shape, biostatisticians Haber and Barnhart undoubtedly will play a role in bringing hope to those whom traditional treatments have failed.

Pallidotomy study, phase II



For some, the pallidotomy procedure can decrease the tremors, stiffness, and other movement problems characteristic of Parkinson's disease.


Fall 1997 Issue | At the Pinnacle | Ambassadors of Hope | A Toxic Legacy | Trials of Life
School Sampler | Alumni Sampler
WHSC | RSPH

Copyright © Emory University, 1998. All Rights Reserved.
Send comments to hsnews@emory.edu.
Web version by Jaime Henriquez.