The presence of any cardiovascular comorbidity was adjusted for, because it may have modified survival outcomes. This was operationalized as a binary variable. A dichotomous variable was generated to categorize cases according to their initial symptoms and disease history, which were classified as either bulbar onset (difficulties in facial function or swallowing as an initial symptom) or spinal onset (voluntary muscle fatigue as initial symptoms). Patients presenting with both bulbar and spinal symptoms were classified as bulbar Inhibitors,research,lifescience,medical onset cases. Ventilatory support

All patients received information regarding hypoventilation treatment and the possibility of participating in the NIV trial. When hypoventilation occurred, suitability for NIV was assessed

by a pulmonologist and an anaesthesiologist. The primary criteria for recommending for NIV were Inhibitors,research,lifescience,medical an increase in the partial pressure of carbon dioxide (pCO2) to over 5.5 kPa, or a decrease in the partial pressure of oxygen pO2 to below 10 kPa, measured by a morning Inhibitors,research,lifescience,medical arterial blood gas sample. Additional measurements included dyspnoea at rest, forced vital capacity (FVC), peak cough flow (PCF), maximum inspiratory mouth pressure (MIP), maximum expiratory mouth pressure (MEP), and sniff nasal pressure (SNP); all of which are considered secondary criteria for NIV diagnosis. These additional measurements were not always taken at the time Inhibitors,research,lifescience,medical of NIV initiation. Therefore, only pCO2 and pO2 measurements were reported, which were available for all patients. The final decision

was based on each patient’s willingness to undergo NIV treatment, regardless of observed dyspnoea or an elevated morning pCO2. NIV was given using a pressure-assisted ventilator (VPAP III ST®, ResMed, Bella Vista, Inhibitors,research,lifescience,medical Australia). The average weekly duration of NIV use was collected using the device’s in-built counter, normally at 3-month intervals. Patients undergoing NIV less than 4 hours per day at the last control visit, timed one week to 3 months prior to death, were considered NIV-intolerant and were allocated to the Conventional Group. Statistical almost analyses The results are given as mean with 95% confidence intervals if not otherwise stated. Chi-square tests were used to compare discrete variables between the groups. Time (in months) from the onset of the symptoms until diagnosis was analysed using a Mann–Whitney U test. Comparison of the mean arterial pCO2 and pO2 at the moment of NIV initiation and the mean daily use of NIV was performed using a Student’s t-test. Survival time was measured in months from diagnosis until death or June 2012, when the follow-up ended. The interactions of age and NIV use with survival were assessed using a Cox regression. Survival curves were analysed using the Kaplan-Meier MG-132 molecular weight method and the Log-Rank Test. Proportional hazard assumptions were evaluated using Kaplan-Meier plots, with p<0.