Abstract
Objective To assess the association between Guillain-Barré syndrome (GBS) and recent surgery based on French nationwide data.
Methods Data were extracted from the French health administrative databases (SNIIRAM/PMSI). All patients hospitalized for GBS between 2009 and 2014 were identified by ICD-10 code G61.0 as main diagnosis. Patients previously hospitalized for GBS in 2006, 2007, and 2008 were excluded. Surgical procedures were identified from the hospital database. Hospitalizations for surgery with no infection diagnosis code entered during the hospital stay were also identified. The association between GBS and a recent surgical procedure was estimated using a case-crossover design. Case and referent windows were defined as 1–60 days and 366–425 days before GBS hospitalization, respectively. Analyses were adjusted for previous episodes of gastroenteritis and respiratory tract infection, identified by drug dispensing data.
Results Of the 8,364 GBS cases included, 175 and 257 patients had undergone a surgical procedure in the referent and case windows, respectively (adjusted odds ratio [OR] = 1.53, 95% confidence interval [CI]: 1.25–1.88). A slightly weaker association was observed for surgical procedures with no identified infection during the hospitalization (OR = 1.40, 95% CI: 1.12–1.73). Regarding the type of surgery, only surgical procedures on bones and digestive organs were significantly associated with GBS (OR and 95% CI = 2.78 [1.68–4.60] and 2.36 [1.32–4.21], respectively).
Conclusion In this large nationwide epidemiologic study, GBS was moderately associated with any type of recent surgery and was more strongly associated with bone and digestive organ surgery.
Glossary
- CI=
- confidence interval;
- GBS=
- Guillain-Barré syndrome;
- ICD-10=
- International Classification of Diseases, Tenth Revision;
- OR=
- odds ratio;
- PMSI=
- Programme de Médicalisation des Systèmes d'Information;
- SNIIRAM=
- Système National d'Information Interrégimes de l'Assurance Maladie
Guillain-Barré syndrome (GBS) is potentially life-threatening and typically occurs after gastroenteritis and respiratory tract infection.1 Surgery has also been suspected to trigger GBS on the basis of a large number of published case reports.2,–,32 According to the few published retrospective case series on this subject, between 5% and 19% of patients with GBS had undergone a surgical procedure during the 633,–,35 or 836 weeks preceding the onset of symptoms, which corresponds to a much higher surgery rate than in the general population.33,34 Several biological mechanisms have been proposed to explain the onset of GBS after surgery. It has been hypothesized that surgery may alter the balance of the immune system, resulting in transient immunosuppression,30 which would allow autoantibodies to promote an attack on peripheral nerves.33 This immunosuppression has also been hypothesized to promote subclinical infections that could predispose to the development of GBS.33 Intraoperative release of antigens and subsequent autoimmunization to these antigens has also been proposed as a possible sensitizing mechanism contributing to the development of GBS.31,37
The aim of this population-based epidemiologic study was to assess the association between GBS and recent surgery based on French nationwide health administrative databases.
Methods
A case-crossover design (main analysis) was performed using French health administrative databases.
Data sources
This study was based on the French National Health Insurance database (Système National d'Information Interrégimes de l'Assurance Maladie [SNIIRAM]) linked to the national hospital discharge database (Programme de Médicalisation des Systèmes d'Information [PMSI]) by means of a unique anonymous identifier allocated to each individual,38,39 providing detailed information on real-life management of all of the population living in France (approximately 65,000,000 people).
The SNIIRAM contains comprehensive individualized and anonymous data on all outpatient services reimbursed by French National Health Insurance, including drugs, physician visits, and laboratory or imaging investigations. This database now includes outpatient data from the entire population living in France but initially only included data from the 86% of the population covered by the general scheme (before 2010).39 The PMSI database describes all hospital stays in both public and private hospitals. It contains a main diagnosis and associated diagnoses coded according to the ICD-10. The main diagnosis corresponds to the disease justifying admission to the medical unit. The PMSI database also contains classifying procedures performed during the hospital stay, such as surgical procedures. This information is necessary to classify the patient in a given disease-related group. The PMSI database also contains information on expensive drugs administered in hospital, such as immunoglobulins.
Study population
All patients hospitalized in any medical unit for GBS between 2009 and 2014 in metropolitan France were identified from the PMSI database by the ICD-10 code G61.0 as the main diagnosis, as defined in a previous study by the same team.40 To be eligible for inclusion in this study, these patients had not been hospitalized with a GBS code (G61.0) (main or associated diagnoses) between 2006 and 2008. The date of GBS diagnosis (index date) was the date of admission of the first hospitalization for GBS over the study period.40 At least one health care claim between 36 and 13 months before the index date was also required to ensure that patient data were present in the SNIIRAM database throughout the observation period.39
Design
The association between GBS and recent surgery was estimated using a case-crossover design. This design only includes cases, which automatically controls for all time-invariant confounders41 such as genetic background. Each GBS case contributed to one case window (the “at risk” window) and one referent window, which were defined as 1–60 days and 366–425 days before the index date, respectively.
Exposure assessment
Surgical procedures during the case and referent windows were identified from the PMSI hospital discharge database and were considered globally and grouped according to the organs and tissues targeted by the procedure: eye (intraocular); heart, arteries, and veins; digestive organs; urinary system and reproductive organs; bones (vertebrae and limbs); periarticular anatomical structures and herniated disk; soft tissues; and other organs. The standard duration of the procedure, as determined by French National Health Insurance expert committees and used for billing of surgical procedures,42 was used. When several surgical procedures were performed on the same day, the maximum duration was considered. Hospitalizations for surgery with no infection diagnosis code entered during the hospital stay (main and associated diagnoses) were also distinguished from hospitalizations associated with infection codes (data available from Dryad, table e-1, doi.org/10.5061/dryad.h9t1675), as some infections, such as cholecystitis, may require surgery, and other infections, such as surgical site infections, can complicate the procedure.
Covariate assessment
Clinically diagnosed gastroenteritis and acute respiratory tract infections occurring during the case and referent windows were identified by the dispensing of drugs (retail pharmacies) used for their treatment43,44 and covered by French National Health Insurance (some antidiarrheals, intestinal adsorbents, antipropulsives and dehydration salts for gastroenteritis, and extended-spectrum penicillins, cephalosporins, macrolides, pristinamycin, and certain cough and cold preparations for respiratory tract infections). Additional infections during the case and referent windows were identified from the PMSI hospital discharge database (main diagnosis as intestinal infectious diseases [A01–A09] or acute respiratory tract infections [J00–J22, J85.0–J85.2, J86]), which added very few infection events.
Statistical analysis
Conditional logistic regression was used to estimate matched odds ratios (ORs) and 95% confidence intervals (CIs) of the association between GBS and recent surgery. Analyses were adjusted for gastroenteritis and respiratory tract infection exposure in the case and referent windows.
Sensitivity analyses
Analyses were also performed by considering 30-day and 90-day case and referent windows, and after restriction to cases treated with immunoglobulins or plasmapheresis. A case-time control design was performed to take potential exposure time trends into account.45 For this purpose, 10 controls for each case were randomly sampled from the SNIIRAM database, from all beneficiaries with at least one health care claim during the calendar year of diagnosis of the index case. Controls were matched to cases by year and month of birth, and department of residence. They were also required to have at least one health care claim during the calendar year preceding the year of diagnosis of the matched GBS case. The case and referent windows of the controls were the same as those of their matched case (i.e., 1–60 days and 366–425 days before the index date, respectively).
Standard protocol approvals, registrations, and patient consents
This study was approved by the French data protection agency (Commission Nationale de l'Informatique et des Libertés) (regulatory decision DE-2011-078).
Data availability
The present data cannot be shared, for legal reasons. Applications to access the French health insurance claims data must be submitted to the Institut National des Données de Santé (www.indsante.fr/).
Results
Of the 8,944 GBS incident cases identified in metropolitan France between 2009 and 2014, 8,364 had at least one health care claim 36 to 13 months before the date of diagnosis of GBS, and constituted the study population. The mean age of this study population was 51 years, and 57% were men. Twenty-nine percent of patients were hospitalized in intensive care units and 13% required mechanical ventilation (table 1). Immunoglobulins were administered in 76% of patients and plasmapheresis was performed in 3.5% of patients. Three percent of patients died during the 3 months following the first admission to hospital for GBS. A seasonal variation was observed with a peak in winter. As expected, GBS was associated with a recent history of gastroenteritis and respiratory tract infection (table 2).
Study population and characteristics of GBS cases
Association between Guillain-Barré syndrome and infections during the previous 60 days (case-crossover design)
One hundred seventy-five (2.1%) and 257 (3.1%) GBS cases had undergone a surgical procedure in the referent and case windows, respectively (table 3). Of interest, the winter peak was no longer observed among the 257 GBS cases with a history of recent surgery (case distribution: winter: 26.1% [n = 67 cases]; spring: 26.1% [n = 67 cases]; summer: 22.2% [n = 57 cases]; and autumn: 25.7% [n = 66 cases]). Data available from Dryad (table e-2, doi.org/10.5061/dryad.h9t1675) indicate the procedures performed during the referent and case windows. The crude OR of the association between GBS and surgery during the previous 60 days was 1.48 (1.22–1.80) (table 4). The OR was slightly higher after adjustment for gastroenteritis and respiratory tract infection (OR = 1.53 [1.25–1.88]; table 4) and slightly lower when only considering surgical procedures with no infection coded during the hospitalization (adjusted OR = 1.40 [1.12–1.73]). Surgical procedures on digestive organs and bones were significantly associated with GBS (OR = 2.36 [1.32–4.21] and OR = 2.78 [1.68–4.60], respectively) (table 5). These associations were not limited to a group of organs (limbs and vertebrae for bone surgery, and digestive tract, biliary tract, liver, and other organs for digestive surgery) and also remained significant for procedures with no infection coded during the hospitalization (data available from Dryad, table e-3, doi.org/10.5061/dryad.h9t1675). Other categories of surgery (eye, heart/arteries/veins, urinary system and reproductive organs, periarticular anatomical structures, soft tissues, and other) were not associated with an increased risk of GBS (table 5), although a positive association was observed with female genital organ surgery (OR = 3.34 [1.12–9.92], p = 0.03; data available from Dryad, table e-3, doi.org/10.5061/dryad.h9t1675). Overall, the OR of the association with surgery increased with the duration of the procedure, reaching 2.22 (1.48–3.34) for procedures lasting more than 120 minutes (table 6). However, the duration of the procedure was strongly correlated with the type of surgery. In particular, 52% of patients in whom the surgical procedure lasted more than 120 minutes in the case window underwent a procedure on bones or digestive organs (42 of 80 patients), whereas this percentage was only 12% for procedures lasting less than 60 minutes (11 of 93 patients) (data available from Dryad, table e-4, doi.org/10.5061/dryad.h9t1675). Of note, the “other organs” surgery subgroup was significantly associated with GBS for procedures lasting more than 120 minutes (OR = 2.94 [1.15–7.53], p = 0.02) (data available from Dryad, table e-4, doi.org/10.5061/dryad.h9t1675). Long surgical procedures include lung surgery and transplantations, which were more frequent in the case window than in the referent window (data available from Dryad, table e-3, doi.org/10.5061/dryad.h9t1675).
Number of cases of Guillain-Barré syndrome according to the presence of surgery in case and referent windows
Association between Guillain-Barré syndrome and surgery during the previous 60 days (case-crossover design)
Association between Guillain-Barré syndrome and surgery during the previous 60 days (case-crossover design) for each main group of surgery
Association between surgery during the previous 60 days and Guillain-Barré syndrome (case-crossover design), according to the standard duration of the procedure
No significant heterogeneity for the association between GBS and surgery was observed by age group (p for interaction = 0.64), considering groups younger than 40 years (OR = 1.23 [0.75–2.01]), 40–59 years (OR = 1.62 [1.04–2.52]), and 60 years and older (OR = 1.60 [1.24–2.06]), although the association was slightly less marked among the younger patients.
Sensitivity analysis
The choice of 30-day and 90-day case and referent windows resulted in ORs close to those observed with a 60-day window (any surgery: adjusted ORs equal to 1.66 [1.25–2.19] for 30 days and 1.54 [1.30–1.83] for 90 days). Besides, the association between GBS and surgery was still positive and significant among patients treated with immunoglobulins or plasmapheresis, with ORs of the association equal to 1.80 (1.42–2.27) for any surgery and 3.86 (1.84–8.09) and 3.12 (1.79–5.43) for surgery on digestive organs and bone surgery, respectively (data available from Dryad, table e-5, doi.org/10.5061/dryad.h9t1675). The other 6 main groups of surgery were not significantly associated with GBS.
Only minor variations in the prevalence of exposure were observed between the referent window and the case window among control subjects (2.8% increase of surgical procedures over the 12-month period). The case-time control ORs for the association between GBS and any surgery was therefore only slightly modified (crude OR = 1.44 [1.16–1.77], adjusted OR = 1.49 [1.20–1.86]).
Discussion
In this large nationwide study including more than 8,000 cases, GBS was moderately associated with any type of recent surgery. The association remained significant for surgical procedures with no identified infection during the hospitalization. GBS was more strongly associated with bone and digestive organ surgery.
Population-based epidemiologic studies investigating the relationships between GBS and recent surgery are rare. Based on case series data, Gensick et al.33 and Sipilä and Soilu-Hänninen34 estimated a high relative risk of GBS after surgery compared to the general population (13.1 [95% CI = 5.68–30.3]33 and 6.3 [95% CI = 4.2–9.5],34 respectively). In the present study, the association between GBS and any type of surgery was much lower than that reported by Gensick et al.33 and Sipilä and Soilu-Hänninen.34
The main strength of this study is that it was conducted on a prospectively collected comprehensive database covering more than 65 million inhabitants.38,39 Cases were therefore included on a nationwide basis, excluding the possibility of any selection bias attributable to inclusion of GBS cases from one or only a few hospitals (recruitment that could be directly or indirectly related to a recent history of surgery). Furthermore, the large sample size of the study allowed investigation of the duration of the procedure and the type of surgery. Surgical procedures are classifying procedures, meaning that they must be precisely recorded for the hospital to receive full National Health Insurance payment. Data were also adjusted for proxies for common gastroenteritis and respiratory tract infections based on drug dispensing data from the SNIIRAM database. Adjustment for these cofactors slightly increased the OR, which is not surprising, as onset of a common infection would be expected to postpone elective surgery. A self-controlled design was used, allowing time-invariant confounders to be considered over the study period. The case-crossover design is particularly relevant for studying transient exposure, such as surgery, and rare events.41 The case-time control design was also used to take into account a potential time trend in the population exposure prevalence.45 A 1-year lag between the referent and case windows was chosen to take into account the effect of seasonality of GBS and surgery.
A main limitation of this study is the impossibility to confirm GBS cases based on chart review, because the database is irreversibly anonymized. Some validation studies from other countries have shown that the positive predictive value was not perfect for GBS algorithms based on discharge diagnosis codes, although it was improved by considering only the main or primary diagnosis,46,47 as in the present study. Moreover, GBS epidemiologic patterns and characteristics of the hospitalization for GBS derived from the French health administrative data40 were in line with the data reported in several studies from Europe and North America.40 Another limitation is that the identification of infectious events was based on drug dispensing and hospitalization diagnosis codes, without the support of clinical charts or patient interviews. Some infections may therefore not have been identified. However, the present study still confirmed the well-known association between GBS and recent gastroenteritis or respiratory tract infection. We observed a highly significant association for digestive organs and bone surgery (p values of 0.0037 and <0.0001, respectively). Significant associations were also observed for female genital organ surgery, and long surgical procedures in the “other organs” subgroup. However, because of multiple testing, these 2 results could also be explained by chance, as the p values were equal to only 0.03 and 0.02. The results derived from this case-crossover design could be biased when the exposure of interest increases mortality rates, as this would lead to an apparent higher prevalence of exposure in the case window compared to the referent window among selected individuals, because only survivors reach the index date and are therefore included in the study. However, postoperative mortality is low in France,48,49 and is therefore unable to explain the magnitude of the significant associations.
GBS is typically caused by an autoimmune attack on peripheral nerves. However, the mechanisms linking GBS to surgery remain unclear. The present study supports a mechanism that is not specific of a particular surgical procedure, in line with previously formulated hypotheses. These include the role of a transient immunosuppression following surgery,33 altering the balance of the immune system and promoting subclinical infections, or the intraoperative release of antigens and subsequent autoimmunization to these antigens.31,37
In this first population-based epidemiologic study on the relationship between GBS and recent surgery, we report a moderate association between GBS and any type of surgery. Stronger associations were observed with bone and digestive organ surgery. Further research is needed to elucidate the mechanisms by which surgery may increase the risk of GBS.
Author contributions
J.R., A.D., Y.M., F.B., J.C., and A.W. made substantial contributions to the conception and design of the study. A.D. and J.R. performed data management and statistical analyses. J.R. wrote the first draft of the paper. J.R., A.D., Y.M., F.B., J.C., and A.W. contributed to interpretation of the data, revised the manuscript, and approved the final manuscript.
Study funding
This research was funded by the Caisse Nationale de l'Assurance Maladie (French National Health Insurance Fund).
Disclosure
The authors report no disclosures relevant to the manuscript. Go to Neurology.org/N for full disclosures.
Footnotes
↵* These authors contributed equally to this work.
Go to Neurology.org/N for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article.
CME Course: NPub.org/cmelist
- Received February 19, 2018.
- Accepted in final form June 28, 2018.
- © 2018 American Academy of Neurology
References
Disputes & Debates: Rapid online correspondence
- Author response to Dr. Steiner
- Guillain-Barré syndrome following surgery
NOTE: All authors' disclosures must be entered and current in our database before comments can be posted. Enter and update disclosures at http://submit.neurology.org. Exception: replies to comments concerning an article you originally authored do not require updated disclosures.
- Stay timely. Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- 200 words maximum.
- 5 references maximum. Reference 1 must be the article on which you are commenting.
- 5 authors maximum. Exception: replies can include all original authors of the article.
- Submitted comments are subject to editing and editor review prior to posting.