This study consisted of two parts, namely: a prospective GBS surveillance system related to influenza vaccination, and a retrospective assessment of GBS incidence in Spain. In both parts, we attempted to detect incidence rates that were higher than previously reported.
A(H1N1)pdm09 vaccine campaigns
In Spain, influenza vaccination is offered free of charge each year to people in high-risk groups, those over 6 months old with chronic conditions, elderly people over the age of 60 or 65 years depending on the region, healthcare workers, workers in essential public services, and caregivers. The A(H1N1)pdm09 vaccine was not recommended for elderly people without chronic disease.
2009 seasonal trivalent influenza vaccine
The vaccine included A/Brisbane/59/2007(H1N1), A/Brisbane/10/2007(H3N2) and B/Brisbane/60/2008 strains. In the 2009 campaign, 10.2 million doses were administered country-wide between weeks 40 to 48.
2009–2010 pandemic vaccine
The vaccine included the A/California/07/2009 (H1N1) virus strain. The vaccine brands were Focetria® (Novartis), adjuvanted with MF59, recommended to children and the elderly, Pandemrix® (Glaxo SmithKline), adjuvanted with ASO3, recommended for adults, and Panenza® (Sanofi Pasteur), unadjuvanted, recommended for pregnancy. Actual exposure involved administration of 970,468 doses in the second half of November 2009, 641,829 in December 2009, 114,220 in January 2010, coming to an end in late February 2010, making a total of 1.7 million doses, less than 5 % of which were administered to persons under 18 years of age. In all, 12 % of health care workers, 9 % of workers in essential public services, 15 % of people from 18 to 60 years old with chronic conditions, and 28 % of people 60 years old and over received the vaccine (Division of Vaccination Programmes, Spain’s Ministry of Health, Social Services, and Equality, unpublished report).
2010 and 2011 campaigns
The vaccines included the A/California/07/2009 (H1N1), A/Perth/16/2009 (H3N2) and B/Brisbane/60/2008 strains [17, 18], a total of 8.1 million doses in each campaign.
In terms of vaccine effectiveness, estimates in Spain suggest that the 2009-TIV had no protective effect, A(H1N1)pdm09 vaccine had a good effect (66–78 %) and 2010-2011TIVs had a lower effect (50–55 %) [17–19].
Neurologist network
An 11-hospital neurologist network, conceived as a special surveillance system [20], was established in 1996 to set alarm thresholds in GBS monthly incidence, so that incidence monitoring could be undertaken in situations where risk was perceived. The National Centre for Epidemiology at the Carlos III Institute of Health in Madrid co-ordinated the network, which then covered an adult population of 3.9 million. People under 20 years were not covered. The network incorporated epidemiological features of GBS in adults from 1985 to 1997 to estimate upper limits in monthly incidences, and then conducted a 2-year pilot prospective study to update upper thresholds in 1998–1999. As a result, the network estimated curves with expected monthly incidences for all and certain age groups [21, 22].
Surveillance design
At the request of the Spanish Agency of Medicines and Medical Devices (Agencia Española de Medicamentos y Productos Sanitarios/AEMPS), the National Centre for Epidemiology, acting as a Central Unit, contacted network members in August 2009 and re-established the network to detect a potential GBS outbreak related to A(H1N1)pdm09 vaccination. The eleven hospitals constituting the original network remained functionally appropriate for surveillance with minor changes in referral, staff and catchment population but one hospital which had difficulties in notifying cases was excluded. In Spain, a national health service covers the majority of population. To estimate incidence rates, each participant hospital updated the number of people it covered. Of a total of 36.9 million Spanish residents aged ≥20 years, the population under surveillance numbered 4.68 million, approximately 1/8 of the country population of that age. Prospective notification to the Central Unit of all patients suspected of presenting with GBS started in September 2009, approximately one month before onset of the A(H1N1) monovalent vaccine campaign, and ended on 31 December 2011. Retrospective reporting dated back to 1 January 2009. The Central Unit was tasked with identifying potential outbreaks on the basis of the information supplied by local neurologists, and, when applicable, notifying these to the AEMPS.
Study subjects
A GBS case was defined by reference to the National Institutes of Neurological Disorders and Stroke (NINDS) criteria [23]. Two months after notification of a suspected case of GBS, local neurologists confirmed or excluded the diagnosis on the basis of clinical, neurophysiological or cerebrospinal fluid parameters. Miller-Fisher syndrome (MFS) was not compulsorily reported. Each GBS or MFS case was re-classified at the end of the 36-month study by one local neurologist and JdP-C, at the Central Unit, using the recent and prevailing Brighton Collaboration case-definition categories [23, 24].
The AEMPS checked the presence of other possible GBS cases reported as suspected adverse drug reactions in the population covered by network hospitals. For data-completion purposes, in 2012 we requested all hospital pharmacies to provide information on registered IVGG deliveries to wards for treatment of GBS patients admitted during the period 2009–2012. These requests did not add any new patient.
Clinical antecedents
We collected information on immunisations, including influenza vaccines, administered during the 42-day period preceding clinical onset of suspected GBS. For other clinical antecedents, we only investigated those occurring during the 30-day period prior to clinical onset of suspected GBS, namely, respiratory infections (fever with cough or expectoration and other respiratory symptoms), gastrointestinal infections (nausea, vomiting, diarrhoea), other infections, and other antecedents. We collected antecedents through interviews with patients. The AEMPS, working in collaboration with the public health authorities, would certify the vaccine received (brand, health centre where the prescription was issued, and administration date) in A(H1N1)pdm09 post-vaccination cases.
Database of hospital discharges
At the end of 2013, we requested the National Hospital In-patient Registry to supply us with data on all patients aged ≥20 years admitted in the period 2009–2011 and having International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 357.0 (acute infective polyneuritis) as their principal or other diagnosis at discharge. This registry constitutes a database on discharges generated at all hospitals serving the Spanish National Health Service, which provides services for the large majority of the resident population country-wide and encompasses all hospitals in the surveillance network [25].
Ethical approval
This study was approved by the Research Ethics Committee at the Carlos III Institute of Health. All patients, with no exceptions, gave their informed consent.
Statistical analysis
We computed exact Poisson confidence intervals (CI) for observed incidence rates for all, age groups, and genders [26]. We accepted as valid the background incidence and threshold values obtained from the previous study in 1998–1999, despite a possible change in hospital admissions due to the widespread use of IVGG as early treatment for GBS, including mild cases. We defined alarm signals when post A(H1N1)pdm09 immunisation cases were notified by the network or the total monthly incidence exceeded the estimated monthly upper 95 % CI limit for background incidence [22]. Monthly upper 95 % CI limit for background incidence had been calculated with ARIMA and Poisson models [21]. Crude monthly incidences were calculated from notifications and confirmed cases, and plotted on a graph showing predicted values.
We checked the positive predictive value of ICD-9-CM diagnostic code 357.0 in the Hospital In-patient Registry against the judgment of a neurologist (JAG-M) at one of the hospitals, the Puerta de Hierro University Hospital in Majadahonda. For each centre, we estimated a sensitivity value as the complementary of the false-negative rate, defined as the ratio between the number of notified and confirmed GBS cases not registered at the National Hospital In-patient Registry over the total number of notified and confirmed GBS cases. We calculated monthly incidences of hospital-admitted GBS country-wide and plotted these on the epidemic curve defined by the 95 % CI limits of expected incidences depicted in the above-mentioned graph.