V. Cautions Regarding Use of the Survey Data
A number of cautions need to be kept in mind when using the results
of the earlier studies in making current estimates.
1. Some of the studies (e.g., Baltimore) were not done for the
primary purpose of determining the disease incidence and prevalence,
but rather for the purpose of identifying causal factors.
2. The proportion of older people in the overall population has
increased. Also, the introduction of levodopa increased longevity by
a few years for people with Parkinson's disease. Therefore more
recent studies should reflect correspondingly higher incidence and
prevalence rates than the older ones did.
The following figures from the U.S Census web pages (17) indicate the
population growth in the age groups most affected by Parkinson's
disease:
Table 2. Growth of U.S. Population
and Proportion of Older Adults (In thousands)
ages % of ages % of
_Year___Population___45 & up___total___65 & up___total_
| |
| 1950 150,697 42,908 28.5 12,195 8.1 |
| |
| 1960 179,323 52,616 29.3 16,559 9.2 |
| |
| 1970 203,212 61,875 30.4 20,065 9.9 |
| |
| 1980 226,546 70,053 30.9 25,550 11.3 |
| |
| 1990 248,710 77,248 31.1 31,078 12.5 |
| |
| 2000 274,634 99,977 36.4 38,986 14.2 |
|_______________________________________________________|
Notes: The totals for ages 45 and up include the totals of
65 and up. Figures for year 2000 are projections as of 1996.
3. Different definitions of Parkinson's disease, some allowing
various forms of Parkinsonism, some not, have been used in different
studies. In some, different forms of Parkinsonism have been
included; in others only idiopathic Parkinson's disease was counted.
Diagnostic categories themselves have changed over time.
4. Some people with Parkinson's disease were not included in the
surveys, as in the case of those based on medical records data which
is absent for undiagnosed or untreated people.
5. Limited regions of the country were studied. No attempt was made
to create samples representative of the U.S. as a whole. The
occurrence of Parkinson's disease is not geographically uniform, so
no assumptions can be made regarding areas which have not been
surveyed.
In reports of earlier studies, extrapolations were made to the entire
U.S population. For example, a report of the Baltimore study
contains the following statements: "The estimated prevalence of
Parkinson's disease among whites in the present survey, when adjusted
to [the ethnic proportions in the 1950 census], is 112.2. . . .
Extrapolation to the United States population of approximately
200,000,000 in 1972 yields a minimum estimate of 176,200 and a
maximum of 314,000 persons afflicted with Parkinson's disease" (12).
The "extrapolation" procedure and error margins are not mentioned.
Later studies are cautious about the general applicability of the
results. In a report of the Copiah County study the authors state,
"We do not make claims of 'generalizability' concerning our results.
It is risky to assume simply that a finding from Copiah County will
hold in another community" (13).
VI. Other Interesting but Incomplete Data
We found other studies which provide figures which seemed relevant to
our purpose, but there were missing pieces.
1. Visits to neurologists. A report based on the National
Ambulatory Medical Care Survey conducted by the Center for Disease
Control determined that for 1991 and 1992 an estimated average of
7.253 million visits annually were made to neurologists in the U.S.
Of these visits, 331,000, or 4.6 percent, had a diagnosis of
Parkinson's disease. Parkinson's disease was also the most
frequently reported diagnosis for patients 65 years of age and over,
and it was the seventh leading diagnosis for people aged 45-64 years
(18). This data would be useful for our purpose if numbers of
patients with Parkinson's disease diagnoses were extracted and
totaled for all types of physicians.
Although neurologist visit data cannot tell us the total number of
people with Parkinson's disease in the U.S., it does provide some
evidence of the impact of Parkinson's disease, especially on those
over 65, in terms of health care usage, and one could infer, health
care expenditures. Thus it deserves our attention.
2. Prevalence among nursing home residence. In the National Nursing
Home Survey (NNHS) of 1985-86, conducted by the National Center for
Health Statistics, Parkinson's disease was found in 47.4 persons per
1000 (19). The total nursing home population for the U.S. was
estimated to have been 1.491 million at that time. In a more recent
study of a sample of 5020 nursing home residents older than age 55,
6.8% had either a primary or a secondary diagnosis of Parkinson's
disease (20).
The 1995 estimated total of nursing home residents aged 65 and over
based on U.S. Census data is about 1.4 million (21). The NNHS of
1995 put the number at about 1.5 million (22). The survey figures
suggest that the total number of nursing home residents with
Parkinson's disease is in the range 70,000 to 100,000. Nursing home
survey data would supply an important piece of the puzzle if we knew
either what proportion of all people with Parkinson's disease are in
nursing homes, or else the prevalence among people who are not in
nursing homes.
VII. U.S. National Estimates
Recognizing, or perhaps in spite of, the limitations of the survey
data, it has been necessary to estimate the incidence and prevalence
of Parkinson's disease in the U.S. Incidence, as we noted, seems to
be based on the Rochester studies, and the number is generally not
questioned. With prevalence, however, it is a different story.
Total U.S. prevalence estimates are wide-ranging, and go to roughly
one million and over. Total U.S. estimates are not found in
professional journals, but do occur in books and articles on
Parkinson's disease intended for the general public. Such writings do
not usually cite survey results. They give estimates made by authors
who are presumably aware of the shortcomings of the surveys, such as
people not included, and who often use other, unidentified criteria,
apparently to compensate for these shortcomings.
At the low end are numbers cited by the NIH (1) and in the in the 2nd
and 3rd editions (1984 and 1991) of Dr. Duvoisin's "Parkinson's
Disease: A Guide for Patient and Family." This book contains the
statement "Parkinson's disease is believed to affect about a half-
million persons in the United States, or approximately 1% of the
population over age 50" (23). Half a million is about 1% of the
population over age 50 in 1970, and it is roughly in the same
proportion as the results of the Rochester study.
At the high end, the earliest reference we found to the number one
million occurs in the book "Caring for the Parkinson Patient" by
Hutton and Dippel. "In the United States alone, approximately one
million persons are estimated to suffer some form of Parkinsonism,
the most common form of which is Parkinson's disease . . ." (24).
Dr. Lieberman wrote in his book "Parkinson's Disease, the Complete
Guide for Patients and Caregivers" (Abraham N. Lieberman M.D. and
Frank L. Williams, New York: Simon & Schuster, 1993) ." . .
Parkinson's disease, a degenerative brain disease suffered by about
one million Americans . . ." (25).
Dr. Koller gives a somewhat larger number in 1993. On the
Caregiver's web page (2) it says "It is estimated that Parkinson's
disease affects 1 to 1.5 million people in the United States, most
commonly striking people over the age of 50. However, Parkinson's can
affect people even before age 40. Approximately 10 percent of
Parkinson's cases are estimated to be young-onset." There is a
reference there to the source: Koller, W.C., 1993, Epidemiology of
Parkinson's Disease, American Parkinson's Disease Association
Newsletter, Fall 1993.
A few years earlier Dr. Koller and several co-authors published
articles containing speculations regarding the number of people in
the pre-clinical, or pre-symptomatic stage of Parkinson's disease
(26, 27). A recent statement by Dr. Koller contained in a
transcript of an interview on the occasion of the introduction of
Tasmar, does not refer to an upper limit. "It's estimated that over a
million patients in the United States have suffered with this disease
and that one in a hundred people over age 64 will get it.... Some
studies have suggested 20 or 30 percent of patients early in the
disease go undiagnosed" (28).
The attention given to the pre-symptomatic stage of the disease at
the time the number one million started being cited along with the
size of the total U.S. numbers relative to the results of the surveys
suggest that the total U.S. estimates include people with symptoms
who were not yet diagnosed and people in the pre-symptomatic stage as
well.
These inclusions are justified, although they make the results very
imprecise. The usual criteria for diagnosis (that two of the three
basic symptoms, resting tremor, stiffness and bradykinesia, are
present) were used in surveys. But about 80% of the dopamine-
producing cells in the substantia nigra have ceased functioning
before outward indications are prominent enough to permit diagnosis.
If there is neurodegeneration in progress, one certainly can be said
to have the disease just as surely as an undetected malignancy is
cancer.
VIII. The Scope of a Complete U.S. Survey
No survey of the U.S. as a whole for Parkinson's disease has been
undertaken. Door-to-door medical examinations by which diagnoses
could be made would give the most accurate results. How much of an
effort would this be? It depends on the size of the sample group
which is studied (this in turn determined by the required precision
of the results) and on the uniformity of the occurrence of
Parkinson's disease geographically.
In statistics there is a formula for determining the size of a single
randomly selected sample group for a given percent of probability
that the occurrence within the large population which the sample
represents falls within a certain range (29). Anticipating a
prevalence range centered about the number one million (1 in 270)
with a 95% certainty that the true number is within a range plus or
minus 100,000 people, the number of people who need to be surveyed in
a single sample is slightly over 100,000. Including people from only
the older half of the population roughly halves the sample size, but
it misses younger people with Parkinson's disease.
Using this formula to calculate sample sizes with various error
ranges gives these results:
Table 3. Approximate Random Sample Group Sizes
For Various Error Ranges
Error range
___+/- people____ Sample size___
| |
| 100,000 103,339 |
| 200,000 25,834 |
| 300,000 11,482 |
| 400,000 6,458 |
| 500,000 4,134 |
|________________________________|
But the formula applies only to random occurrences. Parkinson's
disease does not occur entirely at random. In order to offset
geographical variations, the sample group would have to be larger
and would have to be widely dispersed. How much larger will be
better determined as the extent of regional variations becomes
better known.
