Prevalence of multiple sclerosis in 19 Texas Counties, 1998-2000.

Introduction

Multiple sclerosis (MS) is the most common demyelinating disorder of the central nervous system (Institute of Medicine, 2001). Estimates of the number of people affected in the United States range from 250,000 to 350,000 (Anderson et al., 1992). National prevalence estimates vary from 57 to 160 per 100,000 population (Anderson et al.; Baum & Rothschild, 1981; Noonan, Kathman, & White, 2002; Wynn, Rodriguez, O'Fallon, & Kurland, 1990). Many of these estimates are not current, however, and depend on the methodology and population sample used for the estimate. Although MS is not usually fatal, it usually results in severe disability and decreased quality of life, with enormous economic and emotional costs for the affected individuals, families, and society. The etiology of MS is unknown, but it differentially affects women, people between the ages of 30 and 60 years, and Caucasians (Hauser, 1994). Diagnosis of MS is often difficult because no single clinical sign or diagnostic test is unique to this disease and because the presentation of signs and symptoms varies widely (Williamson & Henry, 2004).

Several communities in Texas have contacted the Texas Department of State Health Services (DSHS) to express concern about elevated numbers of individuals with multiple sclerosis (MS) in their areas. Public health officials have been unable to address these concerns because of a lack of basic epidemiologic data on the disease, particularly a lack of current background prevalence estimates for many geographic regions and ethnic groups in the United States. To address this deficiency the Agency for Toxic Substances and Disease Registry (ATSDR), a federal public health agency, provided funding to DSHS to conduct a pilot MS surveillance project in 19 counties centered on the city of Lubbock, Texas (ATSDR, 2004). This 19-county study area offered a unique opportunity to conduct a pilot surveillance project because of the relatively isolated geographic location, the defined number of neurologists, and the race/ethnicity distribution of the population. The goal of the study was to determine sex-, age-, race-, and ethnic-specific MS prevalence for the 19-county study area. The study was approved by the Institutional Review Boards (IRBs) of the DSHS and the Centers for Disease Control and Prevention (CDC).

Methods

Study Area

Nineteen counties in north Texas were included in the surveillance project: Bailey, Borden, Cochran, Crosby, Dawson, Dickens, Floyd, Gaines, Garza, Hale, Hockley, Kent, Lamb, Lubbock, Lynn, Motley, Scurry, Terry, and Yoakum. The city of Lubbock, located in Lubbock County, is the only metropolitan area in the 19-county area.

Case Ascertainment

The primary data source used to ascertain cases was medical records from the offices of neurologists practicing in the study area (neurologists in the Texas Tech University Medical System and eight private neurologists). Records were considered for inclusion in the case count if 1) the patient resided in the 19-county study area; 2) the patient made an office visit between January 1, 1998, and December 31, 2000; and, 3) the condition had the following International Classification of Disease, 9th Revision (ICD-9), codes or corresponding conditions: MS (340), other demyelinating diseases (341.8, .9), transverse myelitis (323.9), or optic neuritis (377.3). Residence was determined by the address indicated on the patient's medical record. ICD-9 codes for conditions similar to MS or symptomatic of MS were also included to ensure an accurate case count. The study used a three-year time period to allow sufficient time for patients to have visited their physician.

Additional data sources, such as self-reported cases or cases that were identified through MS advocacy groups, nursing homes, hospital discharge records, and death certificates were also explored to ensure complete case ascertainment. Death certificates were available and were used as a secondary source of case ascertainment. Electronic hospital discharge records with identifying information were not available for the study time period. Other potential sources for identification of cases might have included nursing facilities and primary care physicians, particularly in the rural part of the study area. These data sources could not be accessed for this project, however.

Case Verification

A standard form was used to abstract data from medical records that met the inclusion criteria listed above. Clinical-examination information and patient histories were collected, as well as laboratory and magnetic resonance imaging (MRI) results. Information on the following descriptive variables was collected: sex, race/ethnicity, occupation, family history of MS, country/state of birth, MS diagnosis by the treating physician, criteria used to determine diagnosis, date of diagnosis, and date of symptom onset. Individual identifiers (i.e., name, address, and social security number) were recorded to ensure accurate counts of cases and to prevent duplicate counting from other sources. The abstracted records of all potential cases were evaluated by a board-certified neurologist (R.S.). The neurologist rated each case according to the Poser criteria of 1983 and classified cases as definite, probable, possible, or not MS (Poser et al., 1983).

Data Analysis

Age-, sex- and race/ethnicity-specific prevalence estimates were calculated for the 19-county study area. Definite and probable MS cases ascertained from the surveillance activities were used as the numerator, and the year 2000 census counts for the study areas were used as the denominator.

Results

A total of 480 medical records were screened for eligibility. Thirty-four records (7 percent) were found to be duplicates. Two hundred and forty-seven records met the criteria for the study: eligible ICD code, residence in study area, and a documented physician visit during the study period. Of the 247 records that met the study criteria, 224 were submitted to the consulting neurologist for review. The 23 records that were not submitted for review were records for which abstractors were not able to find any indication that MS had ever been considered as a diagnosis or records for which little or no medical information was included in the chart. For those records, an abbreviated abstract form that collected demographic data was completed.

Table 1 gives a summary of the diagnostic categories assigned to each medical record abstract by the consulting neurologist using the 1983 Poser criteria. The majority of abstracted records (n = 182, or 81 percent) were classified as definite or probable MS by the reviewing neurologist. Medical records coded for multiple sclerosis (ICD-9 340) accounted for 98 percent (n = 221) of all abstracts that were classified as definite and probable MS according to the Poser criteria. One case was identified under ICD-9 377.3 (optic neuritis), and two cases were identified under ICD-9 341.9 (demyelinating disease of central nervous system, unspecified).

Table 2 gives the potential sources for identification of MS cases. Fourteen death certificates were identified through a search of vital records that listed multiple sclerosis as either the immediate, the underlying, or the contributing cause of death. Five decedents were identified through medical-record abstraction as well as vital-records review. One decedent, however, had not visited a neurologist during the study time period. Nine other individuals were not identified through the Texas Tech system or private neurologists' offices. Seven of nine individuals who were identified exclusively through vital records died in 1998 or 1999. Four of five individuals who were identified both through medical abstraction and through vital statistics died in 2000.

The distribution of cases by sex, age, race/ethnicity, and residence is presented in Table 3. The overall prevalence for the 19-county study area, based on the Poser criteria, was 42.8 per 100,000 (95 percent CI = 36.8-49.5). The prevalence estimate was 68.6 per 100,000 for females (95 percent CI = 58.0-80.6) and 16.6 per 100,000 (95 percent CI = 11.6-23.1) for males. The two age groups with the highest prevalence were the group 40-49 years of age (103.1/100,000; 95 percent CI = 78.5132.9) and the group 50-59 years of age (119.9/100,000; 95 percent CI = 88.7-158.5).

Information on race and ethnicity was missing in 30 percent (n = 55) of the medical records of the MS cases. Race/ethnicity information from the medical records was supplemented with information from public records and vital-statistics data. Individuals for whom no additional information could be located were listed under the "unknown" category The highest prevalence estimate was for non-Hispanic whites at 56.0 per 100,000 (95 percent CI = 47.1-66.1), followed by non-Hispanic blacks at 22.1 per 100,000 (95 percent CI = 8.1-48.1), and Hispanics at 11.2 per 100,000 (95 percent CI = 6.4-18.2).

Discussion

The results of the study support previous research findings showing a regional difference in the prevalence of MS across the United States. The prevalence in Lubbock, Texas, was 42.8 per 100,000, approximately half the prevalence of 85 per 100,000 reported in the National Health Interview Survey (NHIS) data for the entire United States (Noonan, Kathman, & White, 2002). Sex-specific prevalence estimates from the study area were substantially lower for both males and females (16.6 per 100,000 for males and 68.6 per 100,000 for females) than were the estimates reported for the nation as a whole (48 per 100,000 for males and 123 per 100,000 for females) as well as for the Southern region of the country (36 per 100,000 for males and 91 per 100,000 for females). The female-tomale prevalence ratio from the Lubbock area was substantially higher at approximately 4 to 1, than the typically reported national ratio of 2 to 1 and the ratio reported for the Southern region of the United States (2.5 to 1) as reported in the NHIS survey. This difference in the sex-specific prevalence estimates could be the result of under-ascertainment of male cases, under-diagnosis of the disease in males in the Lubbock area, or an actual difference in the prevalence estimates.

With respect to age-specific prevalence, however, the pattern in the 19-county study area was similar to that reported in the NHIS data. Prevalence was estimated to be highest for the groups 40-49 and 50-59 years of age, and lowest for the groups less than 30 and over 70 years of age. The Texas data, however, indicate a considerably lower prevalence in each of the age groups.

Race and ethnicity proved to be the most challenging variable to collect. Thirty percent of the medical records reviewed had no information on race or ethnicity. Additional information was retrieved from other sources, but no source of information provided objective criteria for determination of race and ethnicity. Even with supplemental information, the authors still were not able to classify 10 percent (n = 20) of the cases with respect to race and ethnicity. Because of the varying definitions used to classify ethnicity among the sources that listed ethnicity, the estimates provided for Hispanics should be considered approximate.

The relative proportion of MS cases among blacks and non-Hispanic whites in Lubbock, Texas, was similar to the proportion reported nationally, with the prevalence among non-Hispanic whites approximately twice that among blacks. The actual prevalence estimates for both non-Hispanic whites and blacks in Lubbock were, however, approximately half that reported nationally: 48 per 100,000 versus 22 per 100,000 for blacks and 96 per 100,000 versus 56 per 100,000 for whites (Noonan, Kathman, & White, 2002). No estimates for Hispanics were included in the NHIS data that would allow comparison of the Lubbock data with national data.

The lack of information on Hispanic prevalence, not only in the NHIS data, but also in the published scientific literature, represents an important data gap. One Mexican study published in 2000 estimated MS prevalence in Mexico at 1.5 per 100,000--approximately 10 percent of the prevalence reported for Hispanics in the Texas study (11.2 per 100,000) (de la Maza, Garcia, Bernal, & Fuentes, 2000). There are also indications in the Mexican literature that MS is on the rise in Mexico (Gonzalez & Sotelo, 1995). Although the results from the study reported here are an important contribution, additional work in Hispanic populations is needed for an understanding of the basic epidemiology of MS in Hispanics and the underlying etiology of the disease. Additional work is also needed to clarify issues related to the definition and reporting of ethnicity

In the authors' study, the prevalence estimate in Lubbock County, the only county with a major metropolitan area--the city of Lubbock--was twice that of the 18 rural counties. This finding could represent the effect of differential migration of people diagnosed with MS to Lubbock to be closer to specialty neurology care, or it could represent an under-ascertainment of cases, particularly less severe cases of MS, in the rural counties.

The geographic location of the 19-county study area and the cooperation of the local neurology community were critical to the success of the surveillance efforts. Because of the distances that would have been involved if patients had used a neurologist located outside the study area, there was essentially a circumscribed population in the study area. This circumstance, coupled with the cooperation of all of the practicing neurologists and the Texas Tech Medical system, increases our confidence that the case ascertainment efforts and subsequent counts were as complete as surveillance allows. One neurologist did retire from practice at the start of the study period, but those patients were seen at other area practices.

The primary limitation of the study was the lack of case information from other sources, specifically family and general practitioners in the 19-county area who may have provided care to individuals with less progressive forms of MS. In future surveillance efforts, active surveying of family and general practitioners in the study area will help to fill that gap. There is also concern about individuals who may not have had access to the medical system, lacked sufficient insurance resources to pursue a diagnosis, chose to manage their MS with alternative therapies, or had a mild or remitting form of the disease and required no care during the study period. The DSHS Institutional Review Board did not allow direct patient contact, so there was no formal involvement of MS advocacy groups or contact with individuals with MS in the study area. Active participation of the local MS community and the ability of DSHS to collect limited case information from individuals who volunteer to participate would also help to fill this data gap in future efforts. The limitations discussed here are important and may account for the low prevalence of MS in this area. Without additional surveillance efforts and the involvement of both the medical community and the MS community, it will not be possible to address these limitations adequately in the future.

Conclusions

The project described here has generated the first Texas-specific population-based MS prevalence estimates, including the first MS prevalence estimates for Hispanics and blacks in Texas. While prevalence estimates cannot reveal the trend in newly diagnosed cases of MS, they are useful in that they provide a snapshot of the burden of disease in a geographic area. In addition, the prevalence estimates help researchers understand how the 19-county study area compares with the rest of the United States. This comparison can provide important clues for additional studies. The results apparently support previous research findings of regional differences in disease prevalence across the United States. These results also underscore the need for additional epidemiologic information on the distribution of MS in other areas of Texas and the United States, as well as information on the underlying etiology of the disease.

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. (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, 333-336.

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

de la Maza, M., Garcia, J., Bernal, J., & Fuentes, M. (2000) A review of the epidemiology of multiple sclerosis in Mexico. Revista de Neurologia, 31(5), 494-495.

Gonzalez, O., & Sotelo, J. (1995). Is the frequency of multiple sclerosis increasing in Mexico? Journal of Neurology, Neurosurgery and Psychiatry, 59(5), 528-530.

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.

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

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, 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.

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'Fallon, W.M., & Kurland, L, (1990). A reappraisal of the epidemiology of multiple sclerosis in Olmsted County, Minnesota. Neurology, 40, 780-786.

Dhelia M. Williamson, Ph.D.

Judy P. Henry, Ph.D.

Randolph Schiffer, M.D.

Laurie Wagner

TABLE 1 Outcome of Reviewed Medical Records According to MS
Diagnostic Criteria

Poser 1983 Criteria     Number of Records Reviewed

Definite                170
Probable                 12
Possible                  2
Not MS                   40
Total records reviewed  224

TABLE 2 Sources by Which Cases Were Ascertained (a)

Source                                   Number of MS Cases

Neurologist offices                      224 (b)
General practitioner or other physician  n/a
MS patient advocacy group                n/a
Death certificates                       9 (c)
Hospital discharge data                  n/a
Self-identified                          n/a
Other                                    n/a
Total                                    233

(a) Cases identified through both neurologist-office records and another
source (or other sources) were counted only under the neurologist-office
category.
(b) Includes four cases identified through both death certificate and
medical-record review.
(c) Death certificates identified multiple sclerosis as either an
immediate, an underlying, or a contributing cause of death. Medical
records for these patients were not abstracted or reviewed.

TABLE 3 Descriptive Characteristics and Respective Strata-Specific
Prevalence of Definite and Probable MS Cases

Characteristic                   MS Cases (a)  Population

Overall prevalence, 19 counties  182           424,916
Sex
  Female                         147           214,235
  Male                            35           210,681
Age
  <30                             19           201,420
  30-39                           33            57,282
  40-49                           59            57,239
  50-59                           49            40,869
  60-69                           17            30,676
  70+                              4            37,430
  Unknown                          1                --
Race/ethnicity (b)
  Hispanic                        16           142,448
  Non-Hispanic white             140           249,882
  Non-Hispanic black               6            27,173
  Unknown                         20                --
Residence (c)
  Lubbock County                 132           242,628
  18 rural counties               50           182,288

                                 Strata-Specific Prevalence per
Characteristic                   100,000 Population (95% CI) (c)

Overall prevalence, 19 counties   42.8 (36.8-49.5)
Sex
  Female                          68.6 (58.0-80.6)
  Male                            16.6 (11.6-23.1)
Age
  <30                              9.4 (5.7-14.7)
   30-3                           57.6 (39.7-80.9)
   40-49                         103.1 (78.5-132.9)
   50-59                         119.9 (88.7-158.5)
   60-69                          55.4 (32.3-88.7)
   70+                            10.7 (2.9-27.4)
  Unknown                         --
Race/ethnicity (b)
  Hispanic                        11.2 (6.4-18.2)
  Non-Hispanic white              56.0 (47.1-66.1)
  Non-Hispanic black              22.1 (8.1-48.1)
  Unknown                         --
Residence(c)
  Lubbock County                  54.4 (45.5-64.5)
  18 rural counties               27.4 (20.4-36.2)

(a) Cases include those with a diagnosis of definite or probable MS
according to the Poser 1983 criteria.
(b) Race/ethnicity was determined through information available on the
medical records and public and vital records.
(c) Prevalence = cases per 100,000 population (Fisher's exact 95%
confidence intervals).