We included in the study all patients who both were referred to the 16-bed ICU of the Roger Salengro University Hospital (Lille, France) between January 1996 and March 2009 and fulfilled standard clinical diagnostic criteria for GBS. The diagnosis was always confirmed by electromyography (EMG), which was systematically performed during the first week after hospital admission. Patients were excluded if they had nonidiopathic GBS. Patients were transferred to our ICU from either an emergency department or a neurological ward. Reasons for admission were usually either a suspected or established acute respiratory failure due to neurological impairment of respiratory muscles.
Patients entering the ICU in acute respiratory distress or hypercapnia were intubated without delay. In the other cases, tracheal intubation was decided when VC measured at the bedside was less than 20% of the predicted value. Patients were ventilated in assist-controlled mode with repeated sequences of assisted spontaneous breathing under physiotherapist supervision. All patients were treated by immunotherapy with either polyvalent high-dose IVIgs or PE (four to six procedures). Each PE procedure allowed the exchange of 1.5 plasma mass, replaced by albumin or plasma substitutes. The IVIg treatment regimen was always 0.4 g/kg of body weight daily for 5 consecutive days. These treatments were completed within 8 days after admission. Additionally, during the ICU stay, patients received symptomatic treatments and nursing care, including systematic anticoagulation to prevent thrombosis, early enteral nutrition, and parenteral vitamins. Tight glucose control was not used.
Throughout the study, the same physicians assessed patients' neurological condition and the same physiologist (JFH) interpreted EMG data. Respiratory function was assessed by repeated spirometry performed at the bedside. Spirometric measurements were done by the same well-trained respiratory physiotherapists and included minute ventilation, respiratory frequency, VC, and maximal minute ventilation [12].
Weaning from MV was decided on the basis of the results of clinical assessment and sequential spirometric measurements. Weaning procedures followed the criteria and recommendations of the 8th and 21st French Consensus Conferences in ICU and Emergency Medicine [13, 14]. Weaning and extubation were usually decided when VC was greater than 20% of the predicted value and maximal minute ventilation was twice the minute ventilation, after a 3- to 6-hour trial of T-tube ventilation. The physicians in charge made the decision to perform tracheotomy on the basis of the patient's status, usually when no functional and respiratory improvement was observed within 10 days after completion of immunotherapy.
The following characteristics were prospectively collected for all patients: age, sex, severity of illness at admission as assessed by the Simplified Acute Physiology Score II (SAPS II), delays from GBS onset to hospital admission and ICU admission, and ICU length of stay. Neurological testing used the standardized data collection elaborated by the French Cooperative Group on Plasma Exchange in GBS [4]. The following data were noted: (a) weakness of the limbs, (b) deep tendon reflex abolition in the upper and lower limbs, (c) cranial nerve impairment, (d) weakness of trunk and respiratory muscles (VC level and ability to lift and hold one's head above the bed), (e) presence of paresthesia or impaired vibratory sensitivity in the limbs, and (f) cardiovascular autonomic dysfunction defined as an increase or decrease (40 mm Hg) in systolic blood pressure, spontaneous or induced bradycardia (heart rate decrease of greater than 20 beats per minute), or spontaneous tachycardia (increase to greater than 120 beats per minute without fever). Cerebrospinal fluid analysis and the following biochemical tests were performed in all patients at ICU admission: natremia, glycemia, creatine phosphokinase, liver enzyme, platelet count, prothrombin time, and C-reactive protein levels. Standard viral serology, including cytomegalovirus and Campylobacter jejuni, was realized. Clinical and laboratory data were collected daily until the plateau phase was recognized and thereafter when significant neurological changes were observed. To search for significant neurological improvement or worsening, changes in motor scores were analyzed for all patients between admission and completion of immune therapy. Neurological improvement or worsening was defined as the reappearance or the disappearance of a spontaneous perceptible movement in one of the items of the standardized neurological functional testing. The definite functional outcome was assessed by walking ability (independent ambulation, requiring assistance to walk, or bed-bound) at a follow-up medical consultation 4 to 6 months after discharge from the ICU. Data from electrophysiological testing were analyzed to classify GBS as acute motor axonal neuropathy, acute motor and sensory axonal neuropathy, acute inflammatory demyelinating polyradiculoneuropathy (AIDP), or AIDP with axonal involvement. Motor conduction block for sciatic nerves was defined as an at least 50% decrease in amplitude of the compound muscle action potential (CMAP) [15].
For the purpose of this study, we compared the patients according to the need of MV to verify whether our population was representative of most GBS cases and exhibited previously known prognostic markers. Then the group of ventilated patients (MV+ group) was specifically studied, and the data collected at ICU admission, at the end of immunotherapy, and 8 days after completion of immunotherapy were extracted for each patient. The following time periods were calculated: time from ICU admission to tracheal intubation, time from the end of specific treatment to MV weaning, duration of MV, and timing of tracheotomy. MV weaning was defined as definite extubation or, in tracheotomized patients, when spontaneous ventilation was obtained 24 hours daily. Finally, two subgroups were defined according to the duration of MV (≤ or >15 days) and all of their clinical and biological data were compared to seek for significant predictors of prolonged MV. Prospective collection of patients' data was approved by the ethics committee of our university hospital in 1996. Since this was a fully retrospective study with all patients' data rendered anonymous and according to French law, informed consent was not requested.
Statistical methods
Descriptive analysis (frequencies for categorical data, mean ± standard deviation, median with ranges for numerical variables) was performed in the total population and compared according to the need of MV and to its duration (≤ or >15 days). Categorical variables were compared by the chi-square test or the Fisher test when appropriate. Continuous variables were compared by the Mann-Witney Whitney test. Differences between groups tested by univariate analysis were considered to be significant for variables yielding a P value of less than 0.05. Variables attaining a 0.05 α value were included in a multiple logistic regression analysis model with stepwise selection. All statistical analyses were performed with the SAS Software, version 8.2 (SAS Institute Inc., Cary, NC, USA).