Assessment of Guillain-Barre Syndrome Mortality and Morbidity in the United States: Implications for Acute Flaccid Paralysis Surveillance

Transcription

1 SI51 Assessment of Guillain-Barre Syndrome Mortality and Morbidity in the United States: Implications for Acute Flaccid Paralysis Surveillance D. Rebecca Prevots and Roland W. Sutter National Immunization Program, Centers for Disease Control and Prevention. Atlanta, Georgia To estimate age-specific incidences and assess the national morbidity and mortality burden for Guillain-Barre syndrome (GBS) in the United States, a national hospital discharge database compiled by the Commission on Professional and Hospital Activities(CPHA) and national death certificate data reported to the National Vital Statistics System were reviewed. During , 1,453 patients with GBS were discharged from CPHA-participating hospitals (estimated annual incidence, 3.11, population). The age-specific incidence of GBS increased with age from 1.5/1, in persons < 15 years old to 8.611, in persons 7-79 years old. The total estimated number of GBS-related deaths from 1985 through 199 was 377 (95% confidence interval, ), for an average of 628 GBS deaths per year. These rates suggest that the proposed national surveillance system for acute flaccid paralysis should capture at a minimum the 796 GBS cases in persons < 15 years old. GBS remains a significant health burden among older adults in the United States, with a marked increase in risk after age 4. Since the control of poliomyelitis, Guillain-Barre syndrome (GBS) has emerged as the most important cause ofacute flaccid paralysis (AFP) in the United States and many other developed countries [1,2]. GBS, first described in 1916 [3], is considered to be an immune-mediated disease that targets peripheral nerves. About half of GBS patients report antecedent acute infectious gastrointestinal or respiratory illnesses occurring 1 4 weeks before onset [4]. However, despite extensive investigation over the last several decades, no single cause for GBS has been identified [1]. Instead, the disease appears to be triggered by a variety of infectious agents, including Campylobacter jejuni, Mycoplasma pneumoniae, cytomegalovirus, Epstein Barr virus, vaccinia virus, and human immunodeficiency virus [1,5, 6], and certain noninfectious antigens, including influenza vaccine [7, 8]. Specific criteria have been developed to establish the diagnosis of GBS [9]. No national studies have estimated the rates of GBS in the pediatric population in the United States. The Institute ofmedicine report on the Safety of Childhood Vaccines recently emphasized the need for age-specific estimates of GBS incidence by year of life, particularly among infants and preschool-age children [1]. Moreover, the feasibility ofimplementing AFP surveillance among children in the United States will depend in part on the background rate of GBS. The availability of vital statistics records and a large national sample ofhospital discharges provideda unique opportunityto obtain precise age- Presented in part: 32nd Interscience Conference on Antimicrobial Agents and Chemotherapy, Anaheim, California, October Reprints or correspondence: Dr. D. Rebecca Prevots, National Immunization Program (E61), Centers for Disease Control and Prevention, Atlanta, GA The Journal of Infectious Diseases 1997; 175(Suppl 1):SI ; by The Universityof Chicago. All rights reserved /97/ $1. specific estimates of GBS. Here we describe the current epidemiology of GBS in the United States, including the current burden of GBS-related morbidity and mortality across all age groups. Methods Morbidity data. A large hospital discharge database is maintained by the Commission on Professional and Hospital Activities (CPHA) Professional Activities Survey. The CPHA database represents over 1 nonfederal short-term, self-selected hospitals in the United States. These data include a mean of21 % ofhospitalizations in acute-care facilities that occurred among hospitals participating in the CPHA system from 1985 through 1991 (3% in 1985, 26.4% in 1986, 21.5% in 1987, 19.6% in 1988, 16.7% in 1989, 16.1% in 199, and 16.6% in 1991). Discharge diagnoses are coded using standard International Classification of Diseases, 9th revision (ICD-9) codes. Any person with at least one discharge between 1985 and 1991 for which GBS (ICD-9 code 357) was a primary or secondary (subsequent) diagnosis was included in the CPHA database. Second and subsequent admissions were excluded from the analyses using as matching criteria state of residence and date of birth. Only the first admission for each match was used to calculate incidence. In this data set, cases in which death was coded as the outcome were identified to estimate efficiency of mortality reporting (see below). Mortality data. National vital statistics data are available from the National Center for Health Statistics (NCHS); these data include deaths for which GBS (ICD-9 code 357) was coded as an underlying or contributing cause of death. The NCHS data set comprises information recorded on the death certificate. Any death that occurred in the United States between 1985 and 199 and for which GBS was listed as an underlying or contributing cause on the death certificate was included in our analyses. Incidence. The incidence of GBS morbidity and mortality was calculated using 1988 United States census data. For national GBS morbidity estimated from CPHA data, the total number of hospital-

2 S152 Prevots and Sutter JID 1997; 175 (Suppl 1) ized incident cases in the United States was projected based on the average coverage of participating CPHA hospitals during (21%). Estimated total GBS mortality and morbidity. Lincoln-Peterson capture-recapture methods were used to estimate the total number of GBS deaths occurring in the United States between 1985 and 199 [11]. In addition, the completeness of reporting to NCHS and CPHA was assessed. This capturerecapture method compares events reported to the two systems, assuming that one system (in this case, the CPHA system) surveys a subset of the second system (in this case, NCHS system). This method further assumes that the two systems are independent. That is, events recorded in the first system have a probability of being recorded that is independent of the probability of being reported to the second system. Completeness of reporting is expressed as the proportion of all deaths (estimated by the Lincoln-Peterson method) that were reported to NCHS or CPHA, respectively [11]. Confidence limits around point estimates were calculated according to Chandra Sekar and Deming [11]; the modification proposed by Chapman was applied to strata with small sample sizes [12]. To determine the number of deaths reported to both systems, records from NCHS and CPHA were matched on state of residence, year and month of death, sex, and age ± 1 year. Duplicate matcheswere identifiedthroughcomparison of remainingvariables for multiple matched records and were excluded from the subsequent analysis. Results GBS hospitalizations. Between 1985 and 1991, 1,453 persons with at least one hospitalization for GBS were discharged from hospitals represented in the CPHA. The projected average annual incidence in the United States during this time period was 3./1,. The estimated yearly incidence ofgbs hospitalizations in the United States ranged from 2.4 (1991) to 3.1 (1987) per 1,. No secular trends over this time period were observed. The annual mean rate ofgbs hospitalizations increased with age, from 1.5/1, in persons < 15 years of age to a peak of 8.6/1, in persons 7-79 years old. The rate of GBS hospitalizations in children < 15 years old ranged from 1.4/ 1, in children <5, 1.4/1, in children 5-9, and 1.7/ 1, in children 1-14 (table 1). The rate ofgbs hospitalizations among preschool-aged children ranged from.38/ 1, in infants < 1 year of age to 1.9/1, in children 2 and 4 years of age (figure 1). The relative risk for GBS according to sex varied with age (tabl~. 1). Among persons < 15 years of age, the incidence of GBS hospitalizations was similar for both boys and girls. In the 15- to 19-year-old age group, females had a 33% increased risk compared with males. In persons >4 years of age, men had a consistently higher risk of GBS than women. GBS was associated with increased mortality among older age groups (figure 2). The overall case-fatality ratio was 4.4%. The case-fatality ratio ranged from.7% among persons <15 years old to 8.6% among persons >65. Overall, 12% of patients required hospitalization and intensive care unit treatment. The length of hospital stay increased with age, from a median of 7 days (range, -434) for persons < 15 years old, 9 days (range, -52) for persons 15-44, 11 days (range, -835) for persons 45-64, and 12 days (range, -537) for persons >65. No overall seasonal trends were apparent. In addition, no seasonal trends within years or by age groups were detected. GBS mortality. From 1985 through 199, a total of 2344 GBS-related deaths were reported to NCHS, for an average annual mortality rate of.16/1,. No significant year-toyear variation in mortality incidence was observed, with a range of.15/1, (1989) to.18/1, (1987). No seasonal trends in mortality were noted. The GBS-associated mortality increased with age (figure 3), from.5/1, in persons < 15 years ofage to 1.3/1, among persons >8. Persons >6 years old had a 6-foldincreased risk of death compared with persons 4-59 years old and a 157-fold-increased risk compared with persons <15 years of age. The death rate increased with age among both males and females but, after age 4, was 1.3 times greater among men than among women. Estimated total GBS mortality. A total of 423 GBS-related deaths were identified in the CPHA system for , and 2344 deaths were reported to NCHS during the same time period. Based on deaths reported to the two systems and the 263 GBS deaths reported to both systems (matched pairs), the total GBS mortality burden for was estimated to be 377 (95% confidence interval, ), for an average of628 GBSdeaths per year. The completeness ofgbs mortality reporting was estimated to be 62% to NCHS and 11% to CPHA (compared to capture of 21% ofhospitalizations). The age and sex distributions were similar for deaths reported to CPHA, to NCHS, and to both systems (table 2). The completeness of death reporting to NCHS was relatively uniform for the age groups 15-44, 45-64, and >65 years and ranged from 57% to 67%. However, the completeness of GBS death reporting to NCHS was substantially lower (25%) for those 15 years old (table 2). The age and sex distributions were similar for deaths reported to both systems (table 2). Males made up 52% of cases reported to both systems compared with 54% of cases reported to CPHA and 52% of cases reported to NCHS. Sixty-two percent of deaths reported to both systems occurred among persons >65 years ofage compared with 68% of deaths reported to NCHS and 61% of deaths reported to CPHA. Discussion Our study provides the most current and precise age-specific estimates ofgbs morbidity and mortality in the United States and suggests that >62 GBS-related deaths and >7 GBSrelated hospitalizations occur each year in the United States.

3 JID 1997;175 (Suppll) US Guillain-Barre Syndrome and AFP S153 Table 1. Annual incidence and relative risk ofguillain-barre syndrome by age and sex, United States, Projected annual cases Rate (/1,) Projected Rate Relative risk annual cases (/1,) M F M F (MIF) < < ;; All NOTE. M, male; F, female. The incidence of GBS is relatively uniform during the first four decades of life, ranging from 1.3 to 2.3/1,/year. Infants appear to be at lowest risk for GBS. After age 4, the risk of incident GBS and severe GBS among incident cases shows a marked linear increase with age. These national estimates of GBS are key to evaluating the feasibility of AFP surveillance in the United States, which in turn will allow monitoring of progress toward the goal of worldwide poliomyelitis eradication. Although> 15 years have passed since the last outbreak of paralytic poliomyelitis in the United States [13], wild poliovirus continues to cross country and continental boundaries [14], and poliomyelitis cases continue to be imported into the United States [15]. Development of surveillance for AFP among children < 15 years old has been proposed to document the absence of wild poliovirus transmission in the United States. The rate of GBS estimated here provides a minimum estimate of the rate of AFP in this age group. The incidence of GBS in children < 15 years old 1 8 Figure 1. Average annual agespecific incidence of Guillain-Barre syndrome, United States, o < D-39 4D

4 S154 Prevots and Sutter JIO 1997;175 (Suppll) 1 hospitalized patients [16]. Second, the GBS hospitalization discharge information obtains information from discharges 8 only among hospitals participating in the CPHA system. Third, the specificity of classifying deaths due to GBS is 6 unknown. Our data suggest that this bias may be applicable to GBS deaths among children and adolescents that appear 4 to be greatly underreported to NCHS. Although estimates for the potential magnitude ofthese biases are not available, 2 there is no evidence to suggest that any of these biases would have influenced our findings substantially. Regardless ofthe exact estimates, GBS appears to occur frequently >65 in childhood, including among preschool-aged children. Our study may have implications for the potential underlying Figure 2. Case-fatality ratio for hospitalized Guillain-Barre syndrome cases, United States, pathophysiologic mechanisms ofgbs. The stable incidence of GBS in children and young and middle-aged adults may suggest that exposure to infectious and other conditions that trigger this illness may also be relatively constant and that exposure to these causes may greatly increase with age among persons ~5 years old. Alternatively, the low prevalence of GBS among infants and the high prevalence among the elderly could possibly be explained by an immature immune system in infants and deterioration of immune system functions in the elderly. GBS remains a significant cause ofmortality and morbidity among persons >4 years of age in the United States. The health burden posed by GBS, with an estimated annual cost of $2-3 billion [19], underscores the need for increased efforts to further elucidate the causes of GBS. Although mortality reporting to CPHA and NCHS was suboptimal, little difference was seen in characteristics of fatal cases reported to the two systems, suggesting that these systems may be adequate to monitormortality trends. The estimates ofgbs in the pediatric population and the finding ofstable rates throughout childhood and adolescence provide background rates for more accurate evaluation ofvaccine safety. Finally, the estimated number of 26 cases ofgbs in children <5 years old supports the feasibility ofafp surveillance in this group. Until global poliomyelitis eradication is achieved, sensitive surveillance systems are (1.5/1,) estimated from our study is most consistent with the rate (1.2/1,) reported from two provinces of Canada [16] and slightly higher than the AFP rate of 111, used by the Pan American Health Organization (PAHO) to monitor the quality of poliomyelitis surveillance. These rates translate into a projected annual average of 796 cases in persons < 15 years ofage and 26 cases in children < 5 years old that would need to be investigated yearly if AFP surveillance were adopted in the United States. Although the rate of GBS from our study was similar to the overall AFP rate for the Americas, it is not clear how much higher the overall AFP rate would be for the United States if other causes of flaccid paralysis could have been evaluated. GBS may account for a higher proportion of AFP cases in the United States than in developing, particularly tropical, countries included in the AFP surveillance network in Latin America. This may be due to a lower circulation in the United States of enteroviruses [17] that may cause non GBS polio-like paralytic syndromes. Although diagnostic methods are not standardized across the Americas within the PAHO system, a diagnosis of GBS is a final classification based on extensive physician review, and therefore GBS in the PAHO system should be relatively specific. Although data are not available on the age-specific positive predictive value of a GBS diagnosis, overall positive predictive values of 67%-79% for GBS-coded hospital discharge diagnoses have been found in the United States and Canada [16, 18]. These values applied to the rate of GBS calculated in this study would yield a "true rate" of /1, children < 15 years of age. Our study had several limitations. First, although the specificity of our case definition was unknown, estimated positive predictive values of GBS-coded hospital discharge diagnoses of 67%-79% [16, 18] suggest that we may have overestimated the GBS hospitalization and mortality burden. Alternatively, true GBS cases may have been misclassified by physicians as non-gbs, although available evidence suggests a low rate of false-negative cases among T"" Q).8 a. ~.6 s: 1a.4 (I).2 ol-...e~~== < Figure 3. Average annual age-specific incidence of Guillain-Barre syndrome related mortality, United States,

5 JID 1997; 175 (Suppl 1) US Guillain-Barre Syndrome and AFP Sl55 Table 2. Estimated total national Guillain-Barre syndrome (GBS) mortality by period, age group, and sex, United States, No. (%) GBS deaths GBS deaths Completeness reported to reported to Matched Lincoln-Petersen estimate of reporting Attribute CPHA NCHS pairs (95% confidence interval) to NCHS Period: (1) 2344 (1) 263 (1) 377 ( ) 62% (2) 15 (1) 2 (1) 6 (25-95) 25% (14) 238 (1) 41 (16) 354 (298-41) 67% (23) 497 (21) 56 (21) 87 (73-11) 57% ~ (61) 1594 (68) 164 (62) 2488 ( ) 64% Sex Male 228 (54) 1229 (52) 137 (52) 245 ( ) 6% Female 195 (46) 1115 (48) 126 (48) 1726 ( ) 65% NOTE. NCHS: National Center for Health Statistics; CPHA: Commission on Professional and Hospital Activities. needed to document absence ofwild poliovirus circulation and to ensure detection of all suspected paralytic poliomyelitis cases. References 1. Ropper AH. The Guillain-Barre syndrome. N Engl J Med 1992; 326: Andrus JK, de Quadros CA, Olive lm. The surveillance challenge: final stages of eradication of poliomyelitis in the Americas. MMWR CDC Surveill Summ 1992;41: GuillainG, Barre JA, Strohl A. Sur un syndrome de radiculonevrite avec hyperalbuminose du liquide cephalo-rachidien sans reaction cellulaire: remarques sur les caracteres cliniques et graphique des reflexes tendineux. Bull Soc Med Hop Paris 1916;4: Hurwitz ES, Holman RC, Nelson DB, Schoenberger LB. National surveillance for Guillain-Barre syndrome: January 1978-March Neurology 1984; 33: Winer J. Guillain-Barre syndrome revisited. Pathogenesis still unknown. BMJ 1991; 34: Mishu B, Blaser B. Role of infection due to Campylobacter jejuni in the initiation of Guillain-Barre syndrome. Clin Infect Dis 1993; 17: Schonberger LB, Bregman DJ, Sullivan-Bolyai JZ, et a1. Guillain-Barre syndrome following vaccination in the National Influenza immunization program, United States, Am J Epidemiol1979; 11: Chen R, Kent J, Rhodes P, et a1. Investigation of a possible association between influenza vaccination and Guillain-Barre syndrome in the United States, [abstract]. Post Marketing Surveillance 1992; 5: Asbury AK, Comblath DR. Assessment of current diagnostic criteria for Guillain-Barre syndrome. Ann Neurol 199;27(suppl):S Stratton KR, Howe CJ, Johnston RB, eds. Adverse events associated with childhood vaccines. Evidence bearing on causality. Washington, DC: National Academy Press, Chandra Sekar C, Deming WE. On a method of estimating birth and death rates and the extent of registration. J Am Stat Assoc 1949;44: Chapman DG. Some properties of the hypergeometric distribution with applications to zoological sample census. University ofcalifornia Publications in Statistics (Berkeley) 1951; 1: Centers for Disease Control. Poliomyelitis-United States, MMWR 1986;35: Centers for Disease Control and Prevention. Isolation of wild poliovirus type 3 among members of a religious community objecting to vaccination-alberta, Canada, MMWR Morb Mortal Wkly Rep 1993; 42: Strebel PM, Sutter RW, Cochi SL, et a1. Epidemiology of poliomyelitis in the United States one decade after the last reported case of indigenous wild virus-associated disease. Clin Infect Dis 1992; 14: McLean M, Duclos P, Jacob P, Humphreys P. Incidence of Guillain-Barre syndrome in Ontario and Quebec, Epidemiology 1994;5: Morens DM, Pallansch MA, Moore M. Polioviruses and other enteroviruses. In: Belshe RB, ed. Textbook of human virology. St. Louis: Mosby Year Book, 1991: Koobarian TJ, Birkhead GS, Schramm MM, Vogt RL. The use of hospital discharge data for public health surveillance of Guillain-Barre syndrome. Ann NeuroI1991;3: Buzby JC, Allos-Mishu Ban, Roberts T. Annual costs of Guillain-Barre syndrome preceded by food-borne Campylobacter jejuni [abstract T248]. In: Proceedings of the 12th annual meeting of the American Neurological Association (Washington, DC). Washington, DC: American Neurological Association, 1995.