Subscribe to eTOC



In an analysis of a Medicare data, investigators found significant differences in Parkinson disease incidence among Hispanics, blacks, whites and Asians.

The first nationwide epidemiologic study of Parkinson disease (PD) confirms that there are significant racial differences in disease incidence, and suggests that disease rates are highest in urban areas. The study, which was published in February in Neuroepidemiology, identified over a half million PD cases from 36 million Medicare records.

“As long as Parkinson disease has been of interest, we have never had disease rates based on large-scale studies in the United States,” said Allison Wright Willis, MD, assistant professor of neurology at Washington University School of Medicine in St. Louis, MO, who led the study. “Previously, the largest study looked only at about a thousand cases. It's quite difficult to really understand the burden of disease with so few cases.”

Furthermore, she said, for a disease such as PD, in which the environment is presumed to play an important etiologic role, “it is important to know who gets the disease, and where the disease is most common.”


Dr. Willis obtained Medicare files from the US Department of Health and Human Services, containing demographic, diagnostic, and treatment data on all Medicare-eligible individuals, a group that comprises 98 percent of all Americans over the age of 65. The analysis concentrated on the years 2000 to 2005.

Altogether, the study identified approximately 480,000 cases of PD across the United States. She found that annual incidence increased with age, as expected, but without a plateau in older groups, as has sometimes been described. Incidence grew from approximately 123 per 100,000 in 65-to-69 year olds, to approximately 960 per 100,000 in those over age 85.

PD was more common in men than in women, also as expected. In whites, for instance, mean incidence over the study period was 560 per 100,000 in males, but only 380 per 100,000 in females. When adjusted for age, incident PD affected 146 men for every 100 women.

Racial differences were even more striking. The mean age-adjusted incidence per 100,000 was 452 for whites, 476 for Hispanics, 362 for blacks, and 339 for Asians.


DR. CAROLINE M. TANNER said that one of the caveats of working with Medicare data is that it would be difficult to know if the diagnosis of Parkinson disease is right, since many diagnoses are not confirmed by neurologists.

“With such large numerators and denominators in the study, you would hope to eliminate sample bias,” she said. “You really couldn't hide in this study.”

Differential access to care is always a concern, but because Medicare covers everyone over age 65, and because a lower rate was seen in both blacks and Asians, but not Hispanics, she discounts this as the most significant factor. “I would really hesitate to presume that health care disparities are always the answer.”

Whether the explanation lies in differences in genetics, environmental exposure, or presentation and thus diagnosis between groups, remains a question for further study.

One other racial difference emerged from the data. The ratio of incidence in blacks versus whites was 0.74, while the ratio of prevalence in the two groups was 0.58, indicating survival in blacks with diagnosed PD was lower than in whites. This disparity was not seen for Asians versus whites. “That makes sense, when you realize that blacks have higher mortality rates for all causes than any other race or ethnic group in the United States,” Dr. Willis said. Whether the explanation lies in severity of disease or difference in care is a question she is investigating now.

Dr. Willis also hopes to look at environmental correlates of disease risk. Because the data included zip codes, she could locate each case using a Geographical Information System, allowing her to map PD cases and environmental data. Such systems are widely used in environmental studies, since multiple kinds of data, from watershed boundaries to toxic chemical sites, can be overlaid to create a multidimensional picture of any region in the country.

Her analysis indicates that the highest incidence of disease is in the midwestern and northeastern parts of the country. The data suggest that, contrary to previous work, PD is more common in urban, rather than rural, areas. Differences in definition of “rural” in previous studies may be part of the explanation. “The word ‘rural’ means many things to many people,” Dr. Willis said. “I think what my study shows more than anything is that rurality is not a good way to measure PD burden.”


Commenting on the study, PD epidemiologist Caroline M. Tanner, MD, director of clinical research at the Parkinson's Institute in Sunnyvale, CA, said working with Medicare files is a huge challenge, and full of potential pitfalls. “It's hard work working with these files, and in that, they did a good job.”

Dr. Tanner, who was not involved in the study, said: “The study is useful for saying there are a lot of people in this country with PD, and that's an important statement,” especially given the coming rise in all kinds of diseases of the elderly as the population ages.

In addition, the results confirm previous work on racial differences in PD done in smaller populations, she said. But there are some important caveats in interpreting the results, she said. “With Medicare, you don't know if the diagnosis is right,” since many diagnoses are not confirmed by neurologists. The big concern with really large population studies like this is essential tremor, she pointed out, since as many as 10 percent of cases diagnosed as PD may in fact be essential tremor, according to studies by movement disorder experts.

Conclusions about the geography of PD must also be taken cautiously, she said. “Current residence isn't a very good indicator etiologically, because we think the disease starts years before diagnosis.” This, combined with the high mobility of the US population, makes it difficult to assess risk geographically without further study, she added.

In addition, widespread urbanization of formerly rural areas makes it harder to assign relative risk based on urban versus rural environment. “I wouldn't over-read the geographic conclusion,” Dr. Tanner said. Follow-up studies would need to account for each of these confounders.

Nonetheless, she said, “I think this study is a very good starting point for hypothesis generation.”

Walter A. Rocca, MD, professor of epidemiology at the Mayo Clinic in Rochester, MN, commended the authors “for conducting an innovative and useful study.” He noted that the emphasis on the environmental basis of PD is important. “In recent years, under the pressure of industry, too much funding was consumed by research on genetics and biomarkers. It is time to go back to the real causes of PD, and ecological studies may be a great starting point.”


AW Willis, BA Evanoff, SR Criswell, BA Racette. Geographic and ethnic variation in Parkinson disease: A population-based study of Medicare benefits. Neuroepidemiology 2010;34:143–151.