вход по аккаунту


Evaluating the quality of quantitative data.

код для вставкиСкачать
Evaluating the Q d t y of Quantitative Data
“Medicalstatistics are like a bikini.
What they reveal is interesting but what they
conceal is vital.”
Anonymous 12)
In this issue of Annals of Neurology, Rosati and colleagues describe the results of a study of the prevalence and incidence of multiple sclerosis among a welldefined population in northwestern Sardinia. As with
several other investigations carried out in Sicily and
Sardinia in recent years 13-5, 8-10], the morbidity
rates appear to be much higher than those reported by
previous studies [5]. Of particular interest in this present paper is that the incidence rate of multiple sclerosis
was over twice as great in the last five years of the
investigation as compared to the first five years. As
Rosati and his colleagues point out, “Changes of such
magnitude in so short an interval may indicate either
serious differences in case finding over time, a rapidly
increasing occurrence of the disease, or both.” How
can we decide which of these explanations is the most
When confronted with such a situation, we must
carefully consider (a) the different strategies employed
to identify cases and (b) how likely it is that factors
influencing the level of case ascertainment have
changed over time. There are two broad approaches to
detect those in the population with neurological disease: door-to-door survey and review of medical records 1111. With a door-to-door survey, one need not
rely on the existing medical care system to identify
those in the population likely to have multiple sclerosis. Provided one has a sensitive screening instrument
and a cooperative population, one can detect all those
with signs and symptoms suggestive of multiple sclerosis. Individuals identified in the screening phase of the
survey can then be carefully examined by neurologists
who rigorously apply diagnostic criteria. This strategy
is generally very costly and poses particular difficulties
with multiple sclerosis, since it may be necessary to
follow a person with recently developed symptoms for
some time before the diagnosis can be established with
reasonable clinical certainty.
O n the other hand, if one chooses to utilize the
existing medical care system for the identification of
potential cases, there are a number of conditions that
must be fulfilled:
1. Those with complaints of neurological dysfunction
must have sufficient motivation and resources to seek
medical care. If symptoms are mild or transitory, there
may be no strong incentive to obtain medical advice.
Similarly, if the cost of such advice is high, one may
reserve professional medical consultation for very special circumstances.
2. Neurological expertise must be readily available to
members of the community. If people have to travel
great distances to see a neurologist or if such a consultation is very costly (both in terms of fees and lost
wages or time), there may be considerable reluctance
to seek such care. If primary care physicians tend to
avoid referrals to their neurological colleagues, potential cases of multiple sclerosis will be missed.
3. Uniform criteria must be utilized to establish the
diagnosis. Those making the diagnosis must use the
same set of rules and diagnostic procedures.
4. There must be sufficient detail in the medical record to allow for retrospective review. Otherwise, all
suspected cases will need to be examined. For those
potential cases in which the person died before such an
examination, one is restricted to data that can be obtained through interviews or available records.
If any of the factors have changed over time, it may
be difficult to decide whether the figures reported represent true changes or whether they are the result of
variations in one or more of the above circumstances.
In general, the more intensely one is able to search for
cases, the more potential cases one will discover. Such
intensive searches are difficult or impossible to implement in studies of large populations. As pointed out by
Granieri and Rosati [ 5 ] , many of the early Italian studies based on populations in excess of 200,000 reported
prevalence ratios for multiple sclerosis of less than 30
cases per 100,000. More recent investigations using
intensive case-finding approaches in Italian populations
of less than 70,000 reported prevalence figures of between 3 0 to 5 3 per 100,000 151.
These difficulties are not limited to studies of multiple sclerosis. Increases in incidence rates, particularly among the elderly, have been reported for amyotrophic lateral sclerosis in Rochester, Minnesota 161,
in Israel [7], and in an area of northeastern Italy
111. These changes may represent better case ascertainment.
A similar situation occurred in comparisons of incidence rates for primary brain tumors between Roches-
ter, Minnesota, and the state of Connecticut C12). The
age-specific curve for Connecticut showed a small
childhood peak followed by a taller peak in adult life.
In Rochester, there was a sustained increase in incidence rates with increasing age, together with higher
age-specific incidence rates than those for Connecticut.
Careful comparisons revealed that the larger percentage of cases diagnosed at autopsy in Rochester accounts in large part for these discrepancies.
In the epidemiological literature, one is often confronted by massive tables of age-adjusted morbidity
rates for a given disease. Such data can reveal interesting patterns and have led to etiological hypotheses.
However, we must look beyond the actual numbers
given in such tables. We must critically examine how
these rates were generated. Are they truly comparable
in terms of case ascertainment and diagnostic accuracy?
Similar questions arise in looking at changes in rates
over time. Rosati and his colleagues provide us with
arguments as to why they believe the increase in reported rates may be real. We must weigh these data
critically to decide whether variables known to influence the level of case ascertainment have been relatively stable during the ten-year period of the Alghero
study. Whatever this decision may be, the careful investigation by Rosati and his associates provides further evidence that Sardinia cannot be regarded as a
low-risk area for multiple sclerosis.
Bruce S. Schoenberg, M D , Dr P H , FACP
Neuroepidemiology Branch
lntramural Research Program
National Institute of Neurological and Communicative
Disorders and Stroke
Bethesda, M D
Georgetown University School of Medicine
Washington, DC
196 Annals of Neurology
Vol 21
No 2
February 1987
1. Angelini C, Armani M, Bresolin N: Incidence and risk factors
of motor neuron disease in the Venice and Padua districts of
Italy, 1972-1979. Neuroepidemiology 2:236-242, 1983
2. Anonymous: Quotation. In Strauss MB (ed): Familiar Medical
Quotations. Boston, Little, Brown, 1968, p 569
3. Dean G , Grimaldi G, Kelly R, et al: Multiple sclerosis in southern Europe: I. Prevalence in Sicily in 1975. J Epidemiol Community Health 33:107-110, 1979
4. Dean G, Savettieri G, Giordano D, et al: The prevalence of
multiple sclerosis in Sicily: 11. Agrigento city. J Epidemiol Community Health 35:118-122, 1981
5. Granieri E, Rosati G: Italy: a medium- or high-risk area for
multiple sclerosis? An epidemiologic study in Barbagia, Sardinia, southern Italy. Neurology 32:466-472, 1982
6. Juergens SM, Kurland LT, Okazaki H, et al: ALS in Rochester,
Minnesota, 1925-1977. Neurology 30:463-470, 1980
7. Kahana E, Zilber N : Changes in the incidence of amyotrophic
lateral sclerosis in Israel. Arch Neurol 4 1:157-1 60, 1984
8. Rosati G, Aiello I, Granieri E, et al: Incidence of multiple sclerosis in Macomer, Sardinia, 1912-1981: Onset of the disease
after 1950. Neurology 36:14-19, 1986
9. Savettieri G, Daricello B, Giordano D, et al: The prevalence of
multiple sclerosis in Sicily. J Epidemiol Community Health
35~114-117, 1981
10. Saverrieri G, Elian M, Giordano D, et al: A further study on the
prevalence of multiple sclerosis in Sicily: Caltanissetta city. Acta
Neurol Scand 73:71-75, 1986
11. Schoenberg BS: Clinical neuroepidemiology in developing
countries: neurology with few neurologists. Neuroepidemiology
11137-142, 1982
12. Schoenberg BS, Christine BW, Whisnant JP: The resolution of
discrepancies in the reported incidence of primary brain tumors.
Neurology 28:817-823, 1978
Без категории
Размер файла
191 Кб
data, evaluation, quality, quantitative
Пожаловаться на содержимое документа