Investigation of a cluster of multiple sclerosis in two elementary school cohorts.

Introduction

Multiple sclerosis (MS) is a demyelinating disease of the central nervous system and one of the most common causes of neurological disability in young adults (Institute of Medicine, 2001). Although the number of people with MS in the United States is unknown, it is estimated that 250,000 to 350,000 people are affected by this disease (Anderson et al., 1992). MS differentially affects women, people 30-60 years of age, and Caucasians. Although MS can be diagnosed at any age, the diagnosis is unusual before adolescence, rises steadily from the teens to age 35, and then declines gradually thereafter (Hauser, 1994). Disease progression varies but typically involves a series of remissions and relapses that become more severe over time (Williamson & Henry, 2004).

In December 1994, the Texas Department of Health (TDH) was asked to investigate an apparent "cluster" of 15 individuals with multiple sclerosis who had grown up in the Kern Place-Mission Hills neighborhood in El Paso, Texas, from the 1940s through the 1960s. If a national prevalence estimate of 102 per 100,000 was applied to the estimated 3,100 residents in the area, approximately three cases of MS would have been expected (Minden, Marder, Harrold, & Dor, 1993). Although most investigations of non-infectious-disease clusters are resource intensive and rarely contribute to an understanding of the etiology of the disease--particularly in the case of cluster concerns that center on past environmental exposures--the El Paso cluster request had a number of unique features that prompted TDH to pursue the cluster investigation (Williamson & Henry, 2004). These features included the magnitude of the apparent excess of MS in the community, the possibility that historical environmental and biologic sampling data could be located and linked to individual members of the study cohort, and a unique opportunity to examine a retrospective cohort in which the purported exposure preceded diagnosis of the disease.

Although the etiology of MS is unknown, a number of environmental exposures have been investigated as possible risk factors, including infectious agents, solvents, trauma, pets, and diet, but the findings have not been consistent (Kahana, 2000; Marrie et al., 2000; Minden et al., 1993; Sorensen & Ransohoff, 1998; Weinshenker, 1996). One current theory suggests that environmental exposures to a genetically susceptible individual may trigger the development of MS, with some studies suggesting that the critical time period for environmental exposures is before puberty (Hogancamp, Rodriquez, & Weinshenker, 1997; Kahana, 2000; Kurtzke, 2000; Kurtzke & Heltberg, 2001; Marrie et al., 2000; Poser, 1994; Sadovnick, Dyment, & Ebers, 1997). Exposure to metals has also been considered as an etiologic factor in several studies, including cluster investigations, although no definitive conclusions or associations have been found (Ingalls, 1986; Ingalls, 1989; Irvine, Schiefer, & Hader, 1988; Schiffer, Weitkamp, Ford, & Hall, 1994; Stein, Schiffer, Hall, & Young, 1987).

Concerns raised by community residents in the initial stages of the El Paso cluster investigation included the possible impact of a local metals smelter, particularly impacts from past operations at the facility and neighborhood environmental contamination. The smelter operated in El Paso from 1887 to 1999, processing primarily lead, copper, cadmium, and zinc, and is located approximately one mile from the Kern Place-Mission Hills neighborhood. It is also located adjacent to a second neighborhood of concern, Smeltertown. Unlike the Kern Place-Mission Hills neighborhood, which was predominantly non-Hispanic white and middle to upper middle class during the 1950s and 1960s, Smeltertown was predominantly Hispanic and primarily comprised families of the smelter workers (Agency for Toxic Substances and Disease Registry [ATSDR], 2002).

Extensive environmental and biological sampling was conducted in El Paso in the early 1970s by local, state, and federal environmental and public health agencies. The sampling demonstrated the potential for study cohort members to have been exposed to heavy metals during their pre-adolescent years. Historical sampling efforts documented high levels of metals from the facility's air emissions (ATSDR, 1996), as well as elevated levels of lead, zinc, arsenic, and cadmium in the soil. Blood lead surveys conducted by the local health department and the Centers for Disease Control and Prevention (CDC) in the early 1970s showed that many children living in the study neighborhoods were exposed to harmful levels of lead. In one study, 53 percent of children one to nine years of age living within one mile of the smelter had blood lead levels of 40 micrograms per deciliter of blood ([micro]g/dL) or greater. The authors concluded that particulate lead in dust and air accounted for most of the lead exposure in El Paso children and that the smelter was the principal source of the lead (Landrigan et al., 1975). Smeltertown was razed in the early 1970s because of the heavy-metals contamination in the soil (ATSDR, 2002).

In 1997, TDH received funding from ATSDR, a federal public health agency, to examine the prevalence of MS in El Paso. Early in the initial stages of the cluster investigation, three women who had been diagnosed with MS provided childhood hair samples for metals analysis. The hair samples consisted of ponytails collected in 1956 and 1962 from two women when they were 12 years old and a braid taken from a third woman when she was six years old. Analyses showed elevated levels of lead and mercury in all three samples and elevated levels of zinc and copper in the hair taken from both of the 12-year-olds (ATSDR, 2002). These early findings led the investigators to search for the original blood lead studies and environmental sampling data collected in the early 1970s.

The goals of our study were to 1) determine the number of individuals with MS in the Kern Place-Mission Hills cohort, 2) evaluate the feasibility of determining the number of individuals with MS in the Smeltertown cohort, 3) compare the prevalence of MS in the two cohorts to national estimates to determine if there was an excess, and 4) evaluate the feasibility of obtaining historical environmental and biological sampling data and matching the data with study cohort members.

Methods

A complete census of the children residing in El Paso neighborhoods from the 1940s through the 1960s was not available. Attendance at the local public elementary schools that served the two neighborhoods of concern was used as a surrogate for residence. The Mesita Elementary School served the Kern Place-Mission Hills neighborhood, and Smeltertown children attended the E.B. Jones Elementary School. A retrospective study design was used to investigate the prevalence of MS among students who had attended the elementary schools at any time from 1948 through 1970. This time frame was selected because of the expected availability of school records, the feasibility of tracing the cohort, and the natural history of the disease. The CDC Institutional Review Board approved the study protocol.

The cohort was identified primarily through school district records for the local elementary schools, with school record information used to trace cohort members and obtain current addresses. Multiple tracing techniques were used to locate current addresses for the eligible cohort, including Web-based commercial databases, review of state drivers' license records, inquiries made to El Paso high school alumni associations, inquiries made to the TDH Bureau of Vital Statistics, public meetings, and referrals from classmates. A one-page, self-administered questionnaire that solicited basic demographic data, information on school attendance, and MS status was mailed to each former student for whom an address could be located. TDH staff re-contacted individuals reporting a diagnosis of MS to obtain authorization to review medical records related to their MS diagnosis and a signed release-of-information form before requesting medical records. An independent board-certified neurologist reviewed copies of medical records and classified individuals according to case status using the Poser criteria (Poser et al., 1983). Individuals were classified into one of four diagnostic categories: definite, probable, possible, or not MS.

Standardized morbidity ratios (SMRs) were used to determine if an excess of MS was present in the cohort. An SMR is the ratio of observed number of cases to expected number of cases. The observed number of cases is the number of individuals with the disease under consideration, and the expected number of cases is calculated on the basis of the disease occurrence in a reference population, usually a registry or surveillance system (Greenland & Rothman, 1998). The expected number of cases is the number of cases that would be present in the study cohort if those individuals developed the disease at the same rate as the people in the reference population.

For our study, the observed value used in the SMR calculation was the number of neurologist-confirmed cases of definite and probable MS in the study cohort. Because no population-based registries or ongoing surveillance efforts exist for MS in the United States, prevalence estimates from four sources were used to calculate the expected number of cases: 1) the Mayo Clinic in Olmstead County, Minnesota (Wynn, Rodriguez, O'Fallen, & Kurland, 1990); 2) the 1989-1994 National Health Interview Survey (Noonan, Kathman, & White, 2002); 3) a 1976 survey of physicians and hospitals by the National Institute of Neurological Disorders and Stroke National Survey (Baum & Rothschild, 1981); and 4) a recently completed pilot MS surveillance project conducted by TDH in 19 Texas counties (ATSDR, 2004).

To calculate the SMR, one divides the observed number of cases by the expected number of cases. A ratio of 1.0 indicates that the number of cases observed in the population being evaluated is equal to the number of cases expected on the basis of the rate of disease in the reference population. A ratio greater than 1.0 indicates that more cases occurred than expected, and a ratio less than 1.0 indicates that fewer cases occurred than expected. The SMRs were tested for significant deviation from 1.00 with Fisher's exact test and exact confidence intervals for the Poisson variate.

An attempt was also made to locate the original environmental and biological sampling data from the early 1970s. The authors of the published articles that summarized the studies were contacted (CDC, 1973; Landrigan et al., 1975; Morse, Landrigan, Rosenblum, Hubert, & Housworth, 1979; Rosenblum, Shoults, & Candelaria, 1976). The investigators also examined hard copies of files from the El Paso City-County Health and Environmental District dating back to the late 1960s and early 1970s.

Results

A total of 5,272 students were identified as having attended the two public elementary schools from 1948 through 1970. The Mesita cohort had 3,891 students, and the E.B. Jones cohort had 1,384 students. Three students attended both schools during this time period. Table 1 provides basic demographic information about the elementary school cohorts and the survey respondents. The proportion of women was slightly higher among the survey respondents than in the two school cohorts, and the survey respondents also spent more years at the elementary schools. The racial and ethnic distribution of the survey respondents could not be compared with that of the entire cohort because neither race nor ethnicity was recorded on the original school records. A current address or confirmation of death was obtained for one-third of the cohort (n = 1,680, or 32 percent). Of the Mesita elementary school students for whom a current address was available (n = 1,248), 54 percent returned a completed questionnaire (n = 677). Of the E.B. Jones elementary school students for whom a current address was available (n = 432), 29 percent returned a completed questionnaire (n = 125).

No cases of MS were reported from the E.B. Jones Elementary School cohort. Twenty-two people in the Mesita cohort self-reported a diagnosis of MS (Table 2). Two people self-reported MS but declined to participate in the case confirmation portion of the study. Definite or probable MS was confirmed by an independent neurologist in a total of 16 individuals. Based on the 16 confirmed definite and probable cases, the crude MS prevalence estimate for the Mesita school cohort was 411 per 100,000 (95 percent CI = 197-603 per 100,000). All 16 individuals classified their race/ethnicity as non-Hispanic white, and all were between 46 and 59 years of age.

Table 3 illustrates the wide range of risk estimates for the Mesita cohort that follow from use of the four comparison prevalence estimates. At the extremes, if prevalence estimates from Olmstead County, Minnesota, are used, the 16 cases of MS reported for the study cohort would represent a deficit of cases (SMR = 0.9; 95 percent CI = 0.51-1.44). If recently obtained Texas MS surveillance data are used as the baseline comparison prevalence, however, students in the El Paso Mesita elementary school cohort have four times the risk of developing MS (SMR = 4.0; 95 percent CI = 2.29-6.5).

Historical environmental and blood lead data were located in the course of the study. The samples had been collected in 1972, however, after the end of the study period. The environmental data, primarily from analyses of heavy metals in soil and dust, did indicate the potential for residences surrounding the elementary school to have been exposed to heavy metals. Blood lead surveys indicated that children living in the study neighborhood in the early 1970s were exposed to harmful levels of lead. The environmental and biological data could not, however, be matched to cohort individuals because the identifying information was limited.

Three women with confirmed diagnoses of MS who had attended the elementary school during the study period provided TDH with childhood hair samples that were tested for heavy metals as a part of the study reported here. The samples were hair of a 12-year-old girl from 1956, the braid of a six-year-old girl from 1959, and the ponytail of a 12-year-old girl from 1962. All three samples showed elevated levels of lead (146.1 [micro]g/g, 42 [micro]g/g, and 49.3 [micro]g/g respectively; normal range = 5-29 [micro]g/g) and mercury (27 [micro]g/g, 1.6 [micro]g/g, and 2.1 [micro]g/g respectively; normal range = 0.4-1.2 [micro]g/g). The samples from both 12-year-old girls showed elevated concentrations of zinc (330 [micro]g/g and 250 [micro]g/g; normal range = 130-220 [micro]g/g) and copper (320 [micro]g/g and 240 [micro]g/g; normal range = 5-46 [micro]g/g), and the sample from the six-year-old showed an elevated concentration of arsenic (8.46 [micro]g/g; normal range < 4 [micro]g/g) (ATSDR, 2002). The source of the metals found in the hair samples, however, could not be determined with certainty (Landrigan et al., 1975).

Discussion

This paper illustrates a number of the challenges faced by public health agencies tasked with addressing reported clusters of MS or other neurologic diseases for which little or no epidemiologic information is available. The most basic challenge was the lack of MS data available for Texas during the initial stages of the investigation. In most cluster investigations, a population-based registry or surveillance system would have been able to quickly identify cases in the geographic area and time period of interest. An existing Texas surveillance system would have been able to provide comparison prevalence estimates based on a uniform case definition that were timely and specific to the state, as well as sex-and age-specific estimates.

The prevalence estimates for the Mesita cohort presented in this paper are likely underestimates since there may be additional MS cases among the 67 percent of the Mesita school cohort who could not be located and surveyed during the course of the study.

The excess of MS reported for the Mesita cohort versus no cases reported for the E.B. Jones cohort may represent a true difference in the disease experience of the two cohorts, or it could be due to the lack of participation by the former E.B. Jones school students. A lack of trust in government and perceived lack of benefit may have been contributing factors in the low participation rate. Many members of the E.B. Jones cohort were either part of the original childhood lead studies conducted in the 1970s or had family or friends who participated. Several study participants raised concerns about being used as study subjects again and a perceived lack of government response to the environmental concerns raised in the 1970s. Issues related to availability and access to medical care for the E.B. Jones cohort may also have played a role in the different MS prevalences in the two cohorts.

Conclusions

Although the El Paso MS cluster investigation had some unique study features, it shares many attributes with the host of MS cluster studies conducted over the past several decades. These studies often involve resource-intensive case-finding efforts, lack of region-specific comparison estimates, a wide range of etiologic hypotheses, and a frustrated community at the end of the investigation. Most of the cluster investigations generally leave more questions than answers, particularly those that demonstrate an excess of MS in a community. The number of cases in typical community cluster investigations, including this one, is too small to allow for a meaningful examination of potential etiologies.

Local, state, and federal public health agencies around the United States are receiving ever-increasing numbers of MS inquiries, and expectation is growing among the public that these concerns will be taken seriously and addressed in a timely manner. Lessons learned over the past decades from cancer and birth defect cluster concerns and disease surveillance for these conditions can be applied to MS and other neurological diseases. The opportunity exists now for the public health community to begin to plan and implement MS surveillance strategies that are able not only to assist states with cluster investigations, but also to provide the much-needed descriptive epidemiological basis from which we can pursue additional etiologic studies to define specific risk factors.

In addition to surveillance, innovative strategies are needed to address ongoing community concerns about MS clusters. ATSDR has taken the lead in developing a large-scale, multisite case control study. This study will be able to examine a number of risk factors for MS, including heavy metals and the role of gene-environment interactions. Texas, Missouri, and Ohio will be participating in the case control study.

Acknowledgements: The authors would like to thank the study participants who have been diagnosed with multiple sclerosis and who shared their experiences. We also acknowledge the participation and support of the El Paso City-County Health and Environmental District and the Region 10 director and staff who contributed to the study. The investigation was funded with a grant from the Agency for Toxic Substances and Disease Registry' (Grant Number U50/ATU60289812) and conducted with support from the Texas Tech Health Sciences Center Department of Neuropsychiatry.

Corresponding Author: Dhelia M. Williamson, Epidemiologist, Agency for Toxic Substances and Disease Registry, Division of Health Studies, 1600 Clifton Road MS E-31, Atlanta, GA 30333. E-mail: [e-mail]

REFERENCES

Agency for Toxic Substances and Disease Registry. (1996). Public health consultation multiple sclerosis cluster in El Paso. El Paso, TX: Author.

Agency for Toxic Substances and Disease Registry. (2002). El Paso multiple sclerosis cluster investigation. El Paso, TX: Author.

Agency for Toxic Substances and Disease Registry. (2004). Multiple sclerosis pilot surveillance in 19 Texas Counties. Lubbock, TX: Author.

Anderson, D.W, Ellenberg, J.H., Leventhal, C.M., Reingold, S.C, Rodriguez, M., & Silberberg, D.H. (1992). Revised estimate of the prevalence of multiple sclerosis in the United States. Annals of Neurology, 31 (3), 333-336.

Baum, H.M., & Rothschild, B.B. (1981). The incidence and prevalence of reported multiple sclerosis. Annals of Neurology, 10, 420-428.

Centers for Disease Control and Prevention. (1973). Epidemiologic notes and reports: Human lead absorption--Texas. Morbidity and Mortality Weekly Review, 22, 405-407.

Greenland, S., & Rothman, K.J. (1998). Introduction to categorical statistics. In K.J. Rothman & S. Greenland (Eds.), Modern Epidemiology (pp. 234-235). Philadelphia: Lippincott-Raven Publishers.

Hauser, S. (1994). Multiple sclerosis and other demyelinating diseases. In K.J. Isselbacher & J.B. Martin (Eds.), Harrison's Principles of Internal Medicine (pp. 2289-2295). New York: McGraw-Hill.

Hogancamp, W.E., Rodriguez, M., & Weinshenker, B.G. (1997). The epidemiology of multiple sclerosis. Mayo Clinic Proceedings, 72, 871-878.

Ingalls, T.H. (1986). Endemic clustering of multiple sclerosis in time and place, 1934-1984. Confirmation of a hypothesis. American Journal of Forensic Medicine & Pathology, 7(1), 3-8.

Ingalls, T (1989) Clustering of multiple sclerosis in Galion, Ohio, 1982-1985. American Journal of Forensic Medicine & Pathology, 10(3), 213-215.

Institute of Medicine. (2001). Multiple sclerosis: Current status and strategies for the future. Washington, DC: Author.

Irvine, D., Schiefer, H., & Hader, W.J. (1988). Geotoxicology of multiple sclerosis: The Henribourg, Saskatchewan, cluster focus. II. The soil. Science of the Total Environment, 77(2-3), 175-188.

Kahana, E. (2000). Epidemiologic studies of multiple sclerosis: A review. Biomedicine and Phamacotherapy, 54(2), 100-102.

Kurtzke, J.F. (2000) Multiple sclerosis in time and space--Geographic clues to cause. Journal of Neurovirology, 6(S2), S134-S140.

Kurtzke, J.F., & Heltberg, A. (2001) Multiple sclerosis in the Faroe Islands: An epitome. Journal of Clinical Epidemiology, 54(1), 1-22.

Landrigan, P.J., Gehlbach, S.H., Rosenblum, B.F., Shoults, J.M., Candelaria, R.M., Barthel, WF., Liddle, J.A., Smrek, A.L., Staehling, N.W., & Sanders, J.F. (1975). Epidemic lead absorption near an ore smelter. The role of particulate lead. New England Journal of Medicine, 292(3), 123-129.

Marrie, R.A., Wolfson, C, Sturkenboom, M.C., Gout, O., Heinzlef, O., Roullet, E., & Abenhaim, L. (2000). Multiple sclerosis and antecedent infections. Neurology, 54, 2307-2310.

Minden, S.L., Marder, W.D., Harrold, L.N., & Dor, A. (1993). Multiple sclerosis: A statistical portrait. New York: National Multiple Sclerosis Society.

Morse, D.L., Landrigan, P.J., Rosenblum, B.E, Hubert, J.S., & Housworth, J. (1979). El Paso revisited: Epidemiologic follow-up of an environmental lead problem. Journal of the American Medical Association, 242, 739-741.

Noonan, C.W., Kathman, S.J., & White, M.C. (2002). Prevalence estimates for MS in the United States and evidence of an increasing trend for women. Neurology, 58(1), 136-138.

Poser, C.M., Paty, D.W., Scheinberg, L., McDonald, W.I., Davis, EA., Ebers, G.C., Johnson, K.P., Sibley, W.A., Silberberg, D.H., & Tourtellotte, W.W. (1983). New diagnostic criteria for multiple sclerosis: Guidelines for research protocols. Annals of Neurology, 13(3), 227-231.

Poser, C.M. (1994). The epidemiology of multiple sclerosis: A general overview. Annals of Neurology, 36(S2), S180-S193.

Rosenblum, B.F., Shoults, J.M., & Candelaria, R.M. (1976). Lead health hazards from smelter emissions. Texas Medicine, 72(1), 44-56.

Sadovnick, A.D., Dyment, D., & Ebers, G.C. (1997). Genetic epidemiology of multiple sclerosis. Epidemiologic Reviews, 19(1), 99-106.

Schiffer, R.B., Weitkamp, L.R., Ford, C., & Hall, W.J. (1994). A genetic marker and family history study of the upstate New York multiple sclerosis cluster. Neurology, 44, 329-333.

Sorensen, T.L., & Ransohoff, R.M. (1988). Etiology and pathogenesis of multiple sclerosis. Seminars in Neurology, 18(3), 287-294.

Stein, E.C., Schiffer, R.B., Hall, W.J., & Young, N. (1987) Multiple sclerosis and the workplace: Report of an industry-based cluster. Neurology, 37, 1672-1677.

Weinshenker, B.G. (1996). Epidemiology of multiple sclerosis. Neurological Clinics, 14(2), 291-308.

Williamson, D.M., & Henry, J.P. (2004). Challenges in addressing community concerns regarding clusters of multiple sclerosis and potential environmental exposures. Neuroepidemiology, 23(5), 211-216.

Wynn, D.R., Rodriguez, M., O'Fallen, M., & Kurland, L.T. (1990). A reappraisal of the epidemiology of multiple sclerosis in Olmsted County, Minnesota. Neurology, 40, 780-786.

Judy P. Henry, Ph.D.

Dhelia M. Williamson, Ph.D.

Randolph Schiffer, M.D.

Laurie Wagner

Jeffrey Shire, M.S.

Matthew Garabedian, M.D., M.P.H.

TABLE 1 Demographic Characteristics of the Elementary School Cohorts
and Survey Respondents

                                                          E.B. Jones
                           Mesita Elementary School         School
Characteristic        School Cohort  Survey Respondents  School Cohort

Gender
  Male                1896           302                 714
  Female              1995           375
Age
  Range               35-69          36-69               35-67
  Mean (SD)*          50.4 (6.3)     51.0 (6.6)          51.1 (6.2)
Years in school
  Range               1-9            1-9                 1-10
  Mean (SD)           3.6 (2.5)      5.1 (2.4)           3.9 (2.4)
Race/Ethnicity**
  Non-Hispanic white  n/a            584                 n/a
  Hispanic            n/a            67                  n/a
  Other/unknown       n/a            26
Total                 3891           677                 1384

                       E.B. Jones
                         School
Characteristic        Survey Respondents

Gender
  Male                61
  Female              64
Age
  Range               35-67
  Mean (SD)*          52.9 (7.5)
Years in school
  Range               1-9
  Mean (SD)           4.5 (2.6)
Race/Ethnicity**
  Non-Hispanic white  3
  Hispanic            116
  Other/unknown       6
Total                 125

* SD = standard deviation.
** Race and ethnicity were not recorded on the school records.

TABLE 2 Status of Self-Reported MS Cases from the Mesita Elementary
School Cohort

Case Status              Male  Female  Total

Definite                 5      9      14
Probable                 0      2       2
At risk                  0      0       1
Not MS*                  0      2       2
Other**                  0      0       1
Declined to participate  1      1       2
Not eligible***          0      2       2
Total                    6     16      22

* Did not meet the case definition for MS.
** MS diagnosis could not be confirmed because supporting diagnostic
information was lacking.
*** Did not attend elementary school during the study period.

TABLE 3 Age-Adjusted Standardized Morbidity Ratios (SMRs) for El Paso
Derived from Comparisons with Four Other U.S. Studies

                   Cases Expected  Cases Observed
Comparison Study   in El Paso      In El Paso      SMR  95% CI

Olmsted County     18.0            16              0.9  0.51-1.44
NHIS*               9.7            16              1.7  0.94-2.68
NINDS**             6.8            16              2.4  1.35-3.82
19 Texas counties   4              16              4    2.29-6.5

* National Health Interview Survey.
** National Institute of Neurological Disorders and Stroke.