Population Incidence of Guillain-Barré Syndrome: A Systematic Review and Meta-Analysis

Population incidence of Guillain-Barré syndrome (GBS) is required to assess changes in GBS epidemiology, but published estimates of GBS incidence vary greatly depending on case ascertainment, definitions, and sample size. We performed a meta-analysis of articles on GBS incidence by searching Medline (1966–2009), Embase (1988–2009), Cinahl (1981–2009) and CABI (1973–2009) as well as article bibliographies. We included studies from North America and Europe with at least 20 cases, and used population-based data, subject matter experts to confirm GBS diagnosis, and an accepted GBS case definition. With these data, we fitted a random-effects negative binomial regression model to estimate age-specific GBS incidence. Of 1,683 nonduplicate citations, 16 met the inclusion criteria, which produced 1,643 cases and 152.7 million person-years of follow-up. GBS incidence increased by 20% for every 10-year increase in age; the risk of GBS was higher for males than females. The regression equation for calculating the average GBS rate per 100,000 person-years as a function of age in years was Received: December 7, 2010 Accepted: January 27, 2011 Published online: March 21, 2011 Dr. James Sejvar, Division of High-Consequence Pathogens and Pathology and Division of Vector-Borne Infectious Diseases Centers for Disease Control and Prevention (CDC) 1600 Clifton Road, Mailstop A-39, Atlanta, GA 30333 (USA) Tel. +1 404 639 4657, E-Mail zea3 @ cdc.gov © 2011 S. Karger AG, Basel 0251–5350/11/0362–0123$38.00/0 Accessible online at: www.karger.com/ned Sejvar/Baughman/Wise/Morgan Neuroepidemiology 2011;36:123–133 124 tigenic stimulus for which potential associations with GBS have been reported, including formulations of Semple rabies vaccine, tetanus toxoid vaccine, and some formulations of influenza vaccine [6–8] . With rare exceptions, the biological or epidemiological evidence for a causal association between GBS and antecedent infections or vaccination is equivocal. A firm measure of the incidence of GBS is increasingly important. GBS appears to be the most frequent cause of nonpoliovirus acute flaccid paralysis worldwide; however, accurate estimates of GBS incidence are unknown for many countries. Additionally, the rare association of various vaccines with GBS has made this syndrome an important focus of vaccine safety monitoring [9] . Assessing the presence, magnitude, and attributable risk of vaccine-associated GBS requires reliable age-specific incidence estimates. However, reported estimates of GBS incidence for all ages combined vary from 0.16 to 3.0 per 100,000 person-years [10] . Some of the variability may be due to true differences in GBS incidence; for example, GBS incidence is thought to be higher in parts of Asia [11] . However, even in Europe and North America where most studies have been conducted, reported GBS incidence varies considerably [10] . Some variability is likely artifactual resulting from different case ascertainment methods, case definitions, and case inclusion criteria. A recent comprehensive systematic literature review summarized data from articles worldwide describing the epidemiology of GBS, including trends in incidence [10] . However, the expansive nature of this review included all articles irrespective of methodology, precluding direct comparisons of incidence estimates. Here we present findings of a systematic review and meta-analysis of published studies reporting GBS incidence to obtain the most reliable estimates of populationbased age-specific incidence of GBS in North America and Europe. Materials and Methods Search Strategy We searched for published work in any language recorded in Medline (January 1, 1966 to December 28, 2009), Embase (1988 to December 28, 2009), Cinahl (1981 to December 28, 2009) and CABI (1973 to December 28, 2009). For searching databases, we used the following key words: ‘Polyradiculoneuropathy’, ‘Incidence’, ‘Epidemiology’, ‘Guillain-Barré Syndrome’, ‘Immunization’, ‘Vaccination’, ‘Campylobacter’, and ‘Respiratory Tract Infections’ (Appendix). We also searched the reference lists of articles selected for full-text review for additional references. Selection Criteria We selected studies based upon the following criteria: population based (cases were identified from a well-defined enumerated population); case finding was either prospective, retrospective, or a combination of both; at least 20 cases were identified; GBS cases were confirmed by subject matter experts (neurologists) from prospective patient evaluation, medical chart review, or both, and a clear and widely accepted case definition for GBS was used [e.g. National Institute of Neurological and Communicative Disorders and Stroke definition (NINCDS) [12] , the case definition developed by Asbury and Cornblath [13] , or the Brighton Collaboration [14] criteria]. We excluded studies that were not population based, or which depended upon administrative medical codes only (e.g. International Classification of Diseases codes) to identify cases. We limited the assessment to studies conducted in North America and Europe, because incidence of GBS in many parts of the world is not known, and some evidence suggests that the epidemiology of GBS may be substantially different in other regions. Study Selection and Data Collection Two investigators (J.J.S., O.W.M.) independently reviewed the title and abstract of all citations identified by the initial search strategy and excluded citations that clearly did not meet the inclusion criteria. We retrieved the full text of the remaining studies and both investigators reviewed each study to assess whether it met the inclusion criteria. When reviewers disagreed or were uncertain about the suitability of a study, a third investigator (M.W.) reviewed the paper and all investigators arrived at a consensus by discussion. One investigator, a board-certified neurologist (J.J.S.), extracted the following data from studies that met the inclusion criteria: study design, case ascertainment method, case definition used, study period, number of GBS cases identified (crude and age-specific), denominators (crude and age-specific), reported GBS incidence, and perceived study limitations. These data were verified by a second investigator (M.W.). When papers did not report the numerator and denominator used to calculate rates or only presented age-specific rates graphically, we attempted to contact the study authors for this information. Statistical Analysis For each study that reported age-specific incidence rates of GBS, we plotted the rate versus the midpoint of the reported age group and superimposed the plots on one graph for comparison. Because the oldest age group was open-ended in all of the studies, we assigned the median age for these groups using publicly available vital statistics data from the country in which the assessment was performed. For these assignments, we used data from the geographic area and time period that most closely matched each study population. We fit random-effects Poisson and negative binomial regression models to the age-specific data. Models that included age as a continuous variable with a random effect for the intercept, slope, or both were explored [15] . Six of the 13 studies reported information to calculate age-specific rates of GBS by sex. For these 6 studies, we fit the same regression model used for the 13 studies overall, with the addition of the effect of sex. All regression models were fit using the NLMIXED procedure in SAS version 9.2. We used the results from the negative binomial regression model to derive an average rate of GBS for 9 suc-