Temporal and geographic variation in United States motor neuron disease mortality, 1969–1998

Motor neuron disease (MND) includes a group of neurodegenerative disorders that selectively affect motor neurons. ALS accounts for 85% or more of all MND cases. In most individuals who are initially diagnosed with other MNDs, the illness will progress to include both upper and lower motor neurons and thus will ultimately be diagnosed as ALS. The annual incidence rate from studies in the United States ranges from 1 to 1.8 per 100,000 population and prevalence ranges from 2.5 to 6.5 per 100,000 population. The onset of ALS is age-related with the highest rate of onset occurring between ages 55 and 75 years. This debilitating neurodegenerative disease is nearly always fatal with expected survival of approximately 3 years, and only limited gains in survival (i.e., approximately 2 months) have been achieved with new treatment strategies.^1,2

The understanding of ALS etiology is limited. Five to ten percent of ALS cases are familial, some of which have been linked to mutations in the enzyme superoxide dismutase.³ Chromosomal rearrangements have been identified in a small proportion of apparently sporadic forms of the disease as well.⁴ Exposure to certain environmental agents may also play a role in ALS disease etiology. Several environmental risk factors, including heavy metals, organic solvents, and agricultural chemicals, have been investigated, with variable and inconclusive results.^5,6

Variations in MND mortality over time and by geography, sex, and race/ethnicity may reveal etiologic clues. Several studies have demonstrated changes in ALS or MND incidence and mortality over time.⁷ ALS is more common in men than women by a ratio of 1.5–2 to 1, but recent observations in European countries suggest that this sex difference is decreasing over time.^8,9 ALS frequency was considered to be geographically evenly distributed with rare exceptions (i.e., foci of ALS and parkinsonism-dementia complex in the western Pacific),¹⁰ but several studies in the United States and elsewhere have found modest geographic variations.^11–17 We evaluated United States MND mortality data to determine if previously reported trends in MND frequency and sex distribution have continued in recent years and to characterize any regional differences in mortality rates.

Methods.

The National Center for Health Statistics multiple-cause mortality files for the years 1969 through 1998 were searched for all death records in the United States containing codes that correspond to MND. Multiple-cause mortality files maintain four-digit International Classification of Diseases (ICD) coding. For ICD-8, in use for 1969 through 1978, the three-digit code for MND was 348, with subcodes for ALS and primary lateral sclerosis (348.0), progressive bulbar palsy (348.1), other progressive muscular atrophy (348.2), and other/unspecified manifestations (348.9). ICD-9, in use for 1979 through 1998, lost this specificity. The category code was anterior horn cell disease (335) and all contents of ICD-8 code 348 were placed in code 335.2, excluding other/unspecified. All mortality rates were calculated using records with a coding for MND listed anywhere on the death certificate. Variables obtained for each record included age at death, sex, race, and state and county of residence.

Mortality rates (per 100,000) were direct age-adjusted using the year 2000 US standard population. Average annual mortality rates and 95% CIs were calculated by strata of sex and race for 5-year intervals using the average number of deaths over the interval and the middle year population estimates from the US Census Bureau. CIs were calculated assuming a Poisson distribution. Differences in age-adjusted mortality rates between time periods were evaluated using the two-tailed Z test. Trends over time for 5-year average mortality rates by sex, age, and race were assessed using the Cochran-Armitage trend test.

To compare geographic differences for the years 1989 through 1998, the coterminous, or continental, United States was divided a priori into 12 areas that have been used previously for a study of multiple sclerosis.¹⁸ These groupings allow for a general assessment of how average annual mortality rates vary by latitude and longitude. The northern tier of states generally comprises states located above the 41st to 42nd north latitudes. The southern tier of states includes states below the 37th north latitude. The remaining states are included in the middle tier. California was divided by including those counties north of Fresno County in the middle tier and including the remaining counties in the southern tier. Further, the United States was divided longitudinally into West, Mountain, Central, and Eastern areas. Age-adjusted rates by the 12 areas for the years 1992 through 1998 also were calculated for Hispanics, non-Hispanic whites, and non-Hispanic African Americans. New Hampshire and Oklahoma were not included in the geographical comparisons by race and ethnicity because these states did not collect Hispanic origin data on death certificates during the entire period studied. Differences in age-adjusted mortality rates between geographical areas were evaluated using the two-tailed Z test.

Results.

From 1969 through 1998, 105,318 deaths were reported in the United States for which the death certificate had codes corresponding to MND. The number of these death certificates by age, sex, race, and time period are presented in table 1. MND was listed as the underlying cause of death on 79.3% of these death certificates during the years 1969 through 1978, 83.5% during the years 1979 through 1988, and 87.5% during the years 1989 through 1998. For the period 1979 through 1998 (e.g., the ICD-9 period), the most common underlying causes of death among death certificates where MND was elsewhere listed were ischemic heart disease (3.5%), cerebrovascular disease (1.9%), and pneumonia (1.4%). Among deaths occurring during 1994 through 1998 with MND listed anywhere on the death certificate, the mean age was 69 years for men and 71 years for women. These ages were slightly higher than the mean ages at death during the earliest period evaluated, 1969 through 1973 (64 years for men and 66 years for women). When restricting this evaluation to only those death certificates with MND listed as an underlying cause of death a similar pattern was observed, but the age at death was approximately 2 to 3 years younger for each category.

Average annual age-adjusted mortality rates for 5-year periods are presented in table 2. Overall, MND mortality increased during the 30-year period with a 46% increase in the point estimates between the first 5-year period and the most recent 5-year period (p < 0.001). Mortality was higher for men than for women over the entire period, but the ratio of MND mortality rates among men to the rates among women declined from 1.71 during 1969 through 1973 to 1.47 during 1994 through 1998. MND mortality rates increased steadily for women, particularly white women, during the 30-year period, but no increases were evident for men in the most recent decade (table 2 and figure 1). The mortality rate increased by 60% for white women (p < 0.001) and increased by 37% for white men (p < 0.001). Despite overall increases in mortality among African Americans during the 30-year period (p < 0.001), the increase in mortality rates over time were not as strong or consistent for African American men and women when compared with white men and women (see figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Average annual age-adjusted motor neuron disease mortality for 5-year periods by sex and race, United States, 1969 to 1998 (solid lines: ▴ = African American women, ▵ = white women; dashed lines: ▪ = African American men, □ = white men).

The recording of Hispanic origin on death certificates was available only for the years 1992 through 1998. During this period, the average annual age-adjusted MND mortality rate per 100,000 (and 95% CI) for Hispanics was 0.92 (0.75 to 1.08). Hispanic men had a rate of 1.11 (0.83 to 1.39), and Hispanic women had a rate of 0.78 (0.58 to 0.98). For the same period, the rate per 100,000 (and 95% CI) for non-Hispanic whites was 1.96 (1.89 to 2.02). Non-Hispanic white men had a rate of 2.34 (2.23 to 2.44) and non-Hispanic white women had a rate of 1.59 (1.52 to 1.66). The mortality rate per 100,000 (and 95% CI) for non-Hispanic African Americans was 1.06 (0.94 to 1.18). Non-Hispanic African American men had a rate of 1.36 (1.15 to 1.58) and non-Hispanic African American women had a rate of 0.85 (0.71 to 1.00).

Trends in mortality rates over time varied by age group and birth cohort. Mortality rates rose steadily by age group up to the age of 85 years and then declined for those age 85 years and older in each quinquennium. The highest mortality rates and the largest increases over time occurred in the age group 75 to 84 years (see table 2). The largest increases in age-specific mortality for both men and women occurred among those born after 1905 (see figures E-1 and E-2 on the Neurology Web site at www.neurology.org).

Average annual age-adjusted MND mortality rates during the years 1989 through 1998 demonstrated a southeast to northwest gradient when deaths were grouped by the 12 geographic areas (table 3 and figure 2). The Middle West area (i.e., northern California) was the only place with a mortality rate point estimate that did not follow this pattern. The mortality rate for the North West area (2.22 per 100,000) was higher than the mortality rates of all South areas, all East areas, and the Middle Central area (p < 0.05 for each comparison). The mortality rates for the North Mountain (2.05 per 100,000) and North Central areas (2.02 per 100,000) were higher than the South Central and South East areas (p < 0.05 for each comparison). A similar geographic pattern was observed for previous decades, 1969 through 1978 and 1979 through 1988 (see table 3). The geographic pattern of MND mortality rates was still evident when restricting the analysis to non-Hispanic whites. The rates per 100,000 (and 95% CI) for non-Hispanic whites for the years 1992 through 1998 were 2.33 (1.99, 2.68) in the North West and 1.75 (1.59, 1.90) in the South East areas (p < 0.05 for difference between rates). A similar geographic pattern in rates was not observed among Hispanics or non-Hispanic African Americans, but these rates were based on few observations and thus were more unstable (see table E-1 on the Neurology Web site). Crude average annual MND rates by state are presented as a supplemental table, but these state rates should be viewed with caution as they are less precise and not adjusted for age distribution (see table E-2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Average annual age-adjusted motor neuron disease mortality rates per 100,000 for 12 US areas, 1989 to 1998 (dark shade: ≥2.0 per 100,000; light shade: 1.70 to 1.99 per 100,000; no shade: <1.7 per 100,000).

Discussion.

This analysis of United States MND mortality data revealed that the overall MND rate rose considerably from 1968 to 1998. We observed inconsistent increases in MND mortality among African American men and African American women, and MND rates among African Americans were about half those observed for whites. Similarly, Hispanics had rates approximately 53% lower than non-Hispanic whites. Age at MND death rose approximately 5 years over the 30-year period. These increases in MND age at death are higher than the increase in mean age for the US population during the same time period, approximately 0.9 years for men aged 45 years and older and 1.4 years for women aged 45 years and older.¹⁹

We observed stronger and more recent increases in MND mortality rates among women (60%) during the 30-year period than among men (35%). Indeed, rates among men remained stable in the most recent decade evaluated while rates among women continued to rise. Studies conducted in other countries also demonstrate this reduction in the ratio of male to female mortality rates. In Norway, the sex ratio changed from 1.58 in the 1960s to 1.32 in the early 1990s.¹⁶ During a 32-year period, the MND mortality rate in England and Wales increased 117% in women and 83% in men.²⁰ For a similar time period, the rates in Italy increased by 60% for women and 24% for men.²¹ A change in sex differential in ALS mortality over time was not observed in Finland, but, unlike most other countries, an approximately equal number of men and women died on average from ALS in Finland during the years 1986 through 1995.⁹ No overall change in sex ratio for ALS mortality was observed in Sweden during the years 1961 through 1992.¹³ However, the change in sex ratio in the United States was largely driven by a plateau in MND mortality rates among men during the years 1989 through 1998, so this may be a more recent phenomenon.

These data also revealed marked geographic differences in MND mortality. We observed an increasing southeast to northwest gradient when states were grouped into 12 areas. These geographic differences were still evident when the analysis was restricted to non-Hispanic whites; therefore, the geographic pattern cannot be attributed to differences in the racial/ethnic population distribution in the United States. A United States proportionate mortality analysis looking at state of birth for MND deaths in 1981 demonstrated a similar geographic gradient.¹¹ In the 1970s, an evaluation of county-level US data found the highest MND mortality rates to be in the northwest region of the country.²² Thus, the geographic differences observed in the more recent US mortality data, 1989 through 1998, were consistent with previous observations. Some have argued that MND mortality is relatively evenly distributed both between countries and within countries.¹⁰ However, lower mortality rates have been observed among certain populations, including the Chinese in Hong Kong (0.33/100,000 compared to approximately 1.9/100,000 worldwide) and whites in South Africa (almost 50% lower than expected when compared to England and Wales).^7,20,23 Heterogeneity of MND mortality rates by geography also was observed within several countries including Sweden, Finland, Italy, Norway, England and Wales, and Spain.^12–17

In the United States and in several other countries overall MND mortality rates have continued to increase.⁷ While US mortality data do not demonstrate a rise in rates among men since 1988, rates have continued to rise among women. Some investigators have argued that increases in MND mortality can be explained by decreasing mortality from other causes, resulting in a growing proportion of the susceptible subpopulation that survives long enough to develop the disease.²⁴ We observed that the absolute number of MND deaths among the elderly increased substantially over the 30-year period examined. Some of this increase would be expected, given the increase in survival over time due to a reduction in competing causes of death as well as the growth in the population. To adjust for differences in the proportional age distribution of the population over time, we examined age-standardized rates, using the 2000 US population as the standard. We observed a significant increase in age-specific mortality rates over time for each age group except age 45 to 54 years. Changes in survival over time due to reductions in competing causes of death are addressed by the direct age-standardization of rates and therefore cannot explain the observed temporal increase in MND mortality rates for specific age groups.

Improvements in diagnosis or access to specialty care are possible contributing factors to increased case ascertainment that could result in the observed increases in MND mortality. Compared to previous decades, there are a growing number of clinical research centers, training opportunities, and publications focused on ALS diagnosis, patient care, and research. These developments likely translate to more awareness among the medical community and the possibility of improved or earlier diagnosis for ALS patients. It is unclear whether this would also result in greater sensitivity for MND reporting on death certificates, but it could contribute to apparent increases in MND mortality.

Greater access to specialty neurology care over time could also result in improved ALS diagnosis. In recent decades, the number of neurology specialists practicing in the United States and the rate of neurology patient visits have both increased.²⁵ However, the geographical distribution of neurologists in the United States, if that can be used as a surrogate for access to specialty care, is not consistent with the geographical variation in MND mortality rates. On the basis of 1998 American Academy of Neurology (AAN) membership data,²⁶ the area with the greatest concentration of neurologists was the North East (6.06 per 100,000), but this area had one of the lowest MND mortality rates (1.71 per 100,000). The area with the lowest concentration of neurologists was the North Mountain (2.59 per 100,000), but this area had one of the highest MND mortality rates (2.05 per 100,000). This inverse correlation between the concentration of neurologists and MND mortality rates suggests that access to specialty neurologist care is not a large determinant in explaining the geographic variation in MND mortality rates. Note that these AAN data are limited to members of a professional organization, and availability of neurologists is only one of the factors that accounts for access to specialty neurology care.

Cause of death appearing on death certificates underestimates true frequency for cause of death, and changes in disease coding or reporting accuracy are additional factors that could impact temporal trends in mortality rates. In 1979, the ICD-9 coding replaced ICD-8 coding. Despite the shift from more specific classification of the different MNDs (348.0–348.9 for ICD-8) to a more general coding for these diseases (335.2 for ICD-9), this did not reflect any change in diagnostic criteria for these conditions, and all four-digit codes that fell under MND were captured for the 30-year period. The 5-year period immediately following the ICD coding change, 1979 through 1983, did represent the largest increase in overall age-adjusted MND mortality among the six 5-year periods. MND mortality rates increased 17% in these years compared to the preceding 5-year period, while increases during other periods ranged from 2% to 10%. However, MND mortality rates rose steadily well after the change in ICD coding, particularly among women.

This study relied on mortality as the measure of disease occurrence. Inherent inaccuracies exist when identifying cases from death certificate data, particularly when considering neurologic disorders with complex case definitions and a high potential for misdiagnosis. Mortality data provide a better reflection of incidence for MND than for most other neurologic diseases because the condition is well recognized clinically, and it typically is progressive and fatal within a few years.²⁷ Follow-up of ALS patients from specialty neurology clinics in the United States revealed that the disease was reported in 72% to 91% of the death certificates.^28,29 In a review of medical records for a sample of the 504 veterans of World War II whose deaths were coded to ALS, 36 of 37 were confirmed as ALS; the 37th was considered primary myelopathic muscular atrophy.³⁰ The data from these studies are based on special populations from previous decades, and no information is available on the reporting accuracy of MND in the United States during the 1990s. It is also possible that death certificate accuracy for MND could vary by geography. A recent study in Italy found the accuracy of death certificate reporting for MND to vary by geography.³¹ Again, there is a lack of reliable data on how such reporting could vary by geography, sex, or race/ethnicity in the United States.

ALS is a complex disease of unknown etiology. The general categories of environmental risk factors that have been investigated for an association with ALS are heavy metals, trace elements, solvents and other volatile organic chemicals, ionizing and non-ionizing radiation, and agricultural chemicals.⁵ Several other exogenous risk factors have been evaluated in the scientific literature including infectious agents, nutritional intake, physical activity, smoking, and trauma.⁶ A recent review article concluded that smoking was a probable risk factor for ALS.⁶ The variations we observed in mortality rates over time and by race/ethnicity, sex, and geography could be indicative of differential exposures to relevant environmental factors with respect to time, person, or place. The greater increase in MND mortality rates among women vs men in the past 30 years may suggest increased opportunities for an unidentified environmental, or possibly occupational, exposure among women. A southeast to northwest geographic gradient in United States MND mortality rates, rather than a more diffuse variation, could be consistent with a similarly distributed unidentified environmental risk factor. Racial and ethnic disparities in MND mortality rates could be related to exposures to etiologic agents which are more common among persons of higher socioeconomic status. The data presented in this article are descriptive and insufficient for providing strong evidence for or against a specific environmental or behavioral risk factor for MND. Nevertheless, these descriptive data continue to support the need for rigorous hypothesis-driven research to examine environmental or behavioral risk factors for MND in combination with factors of genetic susceptibility.

Acknowledgment

The authors thank Dr. Harland Austin for his review of this manuscript.