Ethnic variation in the incidence of ALS

Data extraction.

A Medline literature search was performed to identify all published studies on ALS incidence, prevalence, or mortality from 1966 to March 2006, using the MeSH terms “ALS,” “motor neuron(e) disease,” “Lou Gehrig’s disease,” “Charcot’s disease,” “incidence,” “prevalence,” “mortality,” and “epidemiology.” Additional references were sought from article citations. Data were extracted by a single investigator (S.C.) and reviewed by two ALS neurologists (O.H., B.J.T.). Where groups reported data from a single population on more than one occasion, the most recent study was included.^7–10 Reports from high-incidence clusters (such as the Guam/Kii Peninsula of Japan) were excluded. Included studies were then grouped according to the type of data presented: 1) studies providing crude incidence or prevalence rates; among these, studies providing age- and sex-adjusted incidence rates allowing standardized comparisons with other countries were identified; 2) studies that reported data stratified by ethnic group within a single population; 3) migration studies; and 4) other mortality studies. A study may appear in more than one category. To facilitate meaningful comparison between studies, a key to the sources of case ascertainment and diagnostic criteria¹¹ for inclusion is provided in table E-1 (available on the Neurology Web site at www.neurology.org).

Data analysis.

Incidence and prevalence rates are crude and unadjusted unless otherwise stated. Where published data allowed, incidence rates for the 45- to 74-year-old age bracket were calculated and age- and sex-adjusted to the US 2000 standard population by the direct method.¹² This age band reflects the peak ages of presentation where ascertainment is likely to be complete and has been used by a number of other authors.^2,13 CIs were calculated assuming Poisson distribution. Within individual studies, race-specific confidence limits have been calculated according to the size of the ethnic population during the time period studied (information from the US Census Bureau). Ethnic and racial categories were defined according to NIH guidelines (1997 Office of Management and Budget Directive 15). Significance indicates p < 0.05.

Results of data extraction.

Sources of case ascertainment.

Among studies of ALS incidence and prevalence (table E-1), 17 had a single source of case ascertainment^{7,10,14,16,18,26,27,33,37,39,40,43,44,46,48,50,53} and six had two.^{17,20,24,25,30,35} Only seven register-based studies of ALS incidence have been published to date, five of which were based in the Caucasian populations of Europe and North America.^{9,13,19,31,32} The other prospective population-based studies described the epidemiology of ALS in Japan and Uruguay.^8,15 The majority of studies in non-Caucasian populations tended to be smaller and to use less reliable data sources.

Inclusion criteria of cases.

Heterogeneity in the definition of MND was observed among studies. More studies included progressive spinal muscular atrophy in the calculation of incidence rates than excluded it (table E-1). Three studies included juvenile spinal muscular atrophy and consequently may have overestimated the occurrence of MND.^26,27,36

Limitations of the data.

Further comments and critique of the studies discussed below are provided in table E-1.

Studies that provide incidence or prevalence rates by region (table E-1).

Within the Caucasian populations of Europe and North America, the highest reported crude incidence of ALS was 2.4 per 100,000 person-years in Middle Finland, and the lowest reported was 0.6 in Italy. Most studies in European and American populations observed a crude incidence above 1 per 100,000 person-years. Outside of Europe and North America, high incidence was noted only from a single South American study¹⁵ and in the mixed population of Hawaii.²⁴ The remainder of studies outside Europe and North America document a crude incidence below 1 per 100,000 person-years. However, there was a wide disparity between the epidemiologic methods employed in these regions compared to Europe and North America (table E-1).

To facilitate meaningful comparisons, age- and sex-adjusted incidence rates from 22 of the epidemiologic studies were standardized to the 2000 US population (table 1). A number of observations can be made. First, prospective designs with multiple sources of case ascertainment provide higher estimates than retrospective studies performed in the same populations. Second, the incidence of ALS across Europe is remarkably uniform, as demonstrated by four prospective, population-based studies from Ireland, Scotland, and Italy.^9,13,31,32 Finally, only a limited number of the adjustable studies examine incidence of ALS among non-Caucasian populations.^8,22,25

Only three epidemiologic studies of ALS have been performed in black African populations^25–27 (table E-1). Among these, a single incidence study,²⁵ from Benghazi in Libya, was available. This reports an apparently low crude incidence rate of 0.87 per 100,000 person-years. However, Libya has a young population, with 81% younger than the age of 40,²⁵ and thus the standardized incidence for the 45 to 74 age band of 4.6 per 100,000 person-years (95% CI: 2.9 to 6.8) does not differ from rates seen across Southern Europe (table 1). Conversely, marked differences in ALS incidence are not to be expected between Libya and Southern Europe because the Libyan population encompasses mixed descendants of Arabs, Europeans, and Africans, and <1% are of black African origin.²⁵ There have been no well-conducted ALS incidence studies elsewhere in Africa.

There are two published studies concerning the incidence of ALS in Asian populations (table E-1 and table 1). A large, prospective incidence study of ALS on Hokkaido Island, Japan⁸ reported a standardized incidence for the 45 to 74 age band of 2.0 per 100,000 person-years (95% CI: 1.6 to 2.4). The lowest incidence observed by prospective studies in Caucasian populations is 3.8 per 100,000 person-years (95% CI: 3.2 to 4.4; Puglia, Italy), indicating a significant difference. This apparently lower ALS incidence among Asian populations is supported by a retrospective study of ALS among Hong Kong Chinese,²² in which the standardized ALS incidence for the 45 to 74 age band was 0.8 (95% CI: 0.5 to 1.0).

Regarding Central and South American populations, there has been speculation that a “resistance” to ALS exists among Mexicans, based on a study performed in Mexico City reporting an incidence of 0.4 per 100,000.¹⁷ Low incidence rates have also been reported from other Hispanic white populations in South America,^14,16,18 particularly a population-based prospective study from Uruguay (1.4 per 100,000 person-years¹⁵). However, these studies had several methodologic flaws that may explain their observations (table E-1). Furthermore, it may be difficult to assign a single rate under the term Hispanic, as individual populations in Central and South America comprise different admixtures of Spanish, European, Amarinidian, Carthagian, and Arab peoples.⁶⁹ Thus, the term Hispanic is likely to be a less homogeneous and useful epidemiologic category than Asian or African.⁶⁹

ALS epidemiology studies that stratified their findings by ethnicity.

Differences in case finding methodology and population demographics hinder comparison of ALS incidence between different continents. An alternative approach is to search for ethnic variation within a single study employing a uniform method of case ascertainment within a racially diverse population. Six such studies, mainly from the United States, are summarized in table 2.

A prospective, population-based study of ALS in Washington State reported lower age-adjusted incidence rates among nonwhites (mostly African American) compared to the white population, although the 95% CIs markedly overlapped.¹⁹ In contrast, a retrospective incidence study in Harris County, TX, found similar incidence rates between African Americans and non-Hispanic whites, but was underpowered to exclude a decreased incidence among blacks.²¹

Three population-wide ALS mortality studies have been published on the United States (table 2). Mortality studies have the advantage of providing a larger population for observation than localized epidemiologic studies. All these have reported a lower mortality rate among the Hispanic and African American populations when compared to the white population. It must be remembered, however, that limited access to health care among African Americans may have impeded case ascertainment.^70,71

Finally, an incidence study from the Hawaiian Islands stratified its findings by Filipino, Japanese, and Caucasian ethnicity.²⁴ In contrast to the lower incidence of ALS observed in other Asian studies,^8,22 there was no significant difference in ALS incidence between Japanese and white ethnic subgroups.

Migration studies.

Migration studies compare the incidence or mortality from ALS among ethnic subgroups that have migrated to a single location. A study of immigrants to London during the 10-year period 1979 to 1988 showed ALS mortality rates to be lower among Asian (standardized mortality ratio [SMR] = 37.6%), East African (SMR = 55.5%), Caribbean (SMR = 73.6%), and West African (SMR = 76.9%) immigrants to London than among native English.⁵⁸ In contrast, white immigrants from the Indian subcontinent had the expected ALS mortality rate (SMR = 113%). However, the denominators of ethnic populations were not known for that time period and ethnicity was assigned based on the individual’s name.

A further migration study reported a lower ALS incidence among Afro-Asians who had migrated to Israel (standardized incidence ratio = 83%) when compared to the native Israeli Caucasian population.⁵⁹

A migration study in northwestern Italy showed a higher incidence of ALS for Southern Italian and other European immigrants to northern Italy than the native population.⁶⁰ Prospective studies have since demonstrated a uniform incidence across the Caucasian populations of Europe, highlighting the difficulties in interpreting migration study data. Immigrant populations may not be representative of the nonimmigrant population of the same ethnic background¹ and migration studies do not rule out the existence of environmental factors in the home country.

Mortality studies.

Crude ALS mortality rates reported by 20 studies from 22 ethnically diverse populations are presented in table 3. Of these, five examined rates in the Hispanic Caucasian populations of Central and South America,^{16–18,66,68} and three examined rates in Asian populations.^22,65,67 Lower mortality rates are consistently observed in Hispanic and Asian countries compared to Caucasian populations. However, mortality studies are by definition based on a single source of data. Local variation exists in death certificate coding, further complicated by changes to the International Classification of Diseases in 1979 that have seen separate codes for ALS, progressive bulbar palsy, and progressive muscular atrophy amalgamated as a single code for MND.² Furthermore, some mortality studies only include cases where MND is listed as the underlying cause of death, whereas others accept any cases where MND appeared anywhere on the death certificate. Overall, mortality studies are more suited to determining the trend of a disease over time within the same country, rather than international comparisons.³

Proposals for the design of an epidemiologic study to address ethnic variation in ALS frequency.

1. Study design: Prospective study designs provide higher disease rate estimates than comparable retrospective studies in the same populations (table 1), and are the preferred method.

2. Population size: The relative rarity and rapidly fatal course of ALS mean that a study with minimum of 3 to 5 years of follow-up of a population no smaller than one million is required to generate meaningful incidence data within a single ethnic group. If the population is racially heterogeneous, then a larger population cohort will be required to provide differential race-specific incidence data with sufficient precision. For example, in a population composed of two subgroups at an 80:20 ratio, approximately 10 million individuals would require observation for 5 years to detect a 25% difference with certainty (see appendix E-1 for statistical derivation).

3. Complete case ascertainment: ALS incidence studies without multiple sources of case ascertainment are likely to underestimate disease frequency, as has been demonstrated in Italy (table E-1). In populations of mixed ethnicities, particular attention should be given to case finding among ethnic minorities who may not have access to health insurance or specialist care through liaison with ethnic minority organizations and community-based medical services.

4. Representation of older age groups: It is desirable that such a study be performed in an area of long-established ethnic diversity so that older age strata are well represented across all ethnic groups.

5. Uniform diagnostic criteria must be applied. The gold standard El Escorial criteria are suggested, although access to neurophysiologic confirmation may be limited in economically underserved regions.

6. Reporting of age bands and ethnic subgroups: We suggest that all future studies of ALS incidence report their data stratified by ethnicity and also provide age- and sex-specific incidence rates, so that results can be standardized and meaningful comparisons can be made between different populations.