Mechanical ventilation for amyotrophic lateral sclerosis/motor neuron disease
Abstract
This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:
The primary objective of the review is to examine the efficacy of mechanical ventilation (tracheostomy and non‐invasive ventilation) in improving survival in ALS. The secondary objective is to examine the effect of mechanical ventilation on functional measures of disease progression and on quality of life of people with ALS and their carers.
Background
Amyotrophic lateral sclerosis (ALS), also known as motor neuron disease (MND), is a progressive neurodegenerative disorder characterised by loss of corticospinal tract motor neurons (upper motor neurons), and bulbar and spinal motor neurons (lower motor neurons) (Brooks 1994; Brooks 2000). The incidence of ALS is 1 to 2 per 100, 000 of the population and the age specific incidence and mortality rates peak at 55 to 75 years (Worms 2001).
Without mechanical ventilation, death usually follows within two to five years of the onset of symptoms, although a few patients may survive for 10 years or more (Haverkamp 1995; Turner 2003). Death is usually due to respiratory failure as a consequence of respiratory muscle weakness. Measures of respiratory muscle strength are useful in monitoring the progression of respiratory muscle weakness and some measures are predictive of respiratory failure or survival (Bach 1995; Haverkamp 1995; Vitacca 1997; Stambler 1998; Fitting 1999; Chaudri 2000; Lyall 2001 a; Varrato 2001; Lechtzin 2002).
Tracheostomy with mechanical ventilation (TV; 'invasive ventilation') can prolong survival for many years (Bach 1993; Cazzolli 1996) but some people become unable to communicate, and home ventilation is costly and exacts a significant emotional toll on people with ALS and carers (Moss 1993; Cazzolli 1996; Moss 1996). Nevertheless, people affected by ALS are increasingly aware of this option. In Japan, tracheostomy ventilation is undertaken more frequently than in Europe and North America (Hayashi 1997; Borasio 1998; Yamaguchi 2001). In Europe and North America 4 to 10% of people with ALS taking part in drug trials have a tracheostomy or use non‐invasive ventilation (Lacomblez 1996; Cedarbaum 2001).
Non‐invasive ventilation (NIV) is another option for treating respiratory failure in people with ALS. Non‐invasive ventilation utilises a face or nasal mask and a small portable volume or bilevel ventilator to provide an intermittent positive inspiratory pressure ventilator to support ventilation in people with clinically significant (symptomatic) respiratory muscle weakness (Miller 1999). No single test of respiratory function or respiratory muscle weakness can be used to predict reliably the onset of respiratory failure or the most appropriate timing for initiating NIV (Miller 1999; Lyall 2001 a). Measures of respiratory muscle weakness are poorly predictive of respiratory failure in people with bulbar symptoms (Lyall 2001 a).
The use of NIV varies greatly across North America and Europe (Melo 1999; Bradley 2001; Borasio 2001; Cedarbaum 2001; Chio 2001; Bourke 2002). Evidence from several retrospective and some prospective studies indicates that NIV may be associated with a gain in survival (Pinto 1995; Aboussouan 1997; Kleopa 1999; Bach 2002), improved quality of life (QL) (Hein 1997; Hein 1999; Aboussouan 2001; Bourke 2001; Jackson 2001; Lyall 2001 b) and improved cognitive function (Newsom‐Davis 2001). People with ALS who have significant bulbar involvement may tolerate NIV better than those with little or no bulbar muscle weakness (Cazzolli 1996; Aboussouan 1997). Although NIV is less costly than TV, it nevertheless requires significant health resources and commitment from carers.
A Cochrane review of nocturnal mechanical ventilation for people with chronic hypoventilation (Annane 2003) identified four randomised trials, only one of which included people with ALS. This review concluded that nocturnal ventilation may relieve chronic hypoventilation related symptoms and prolong survival, but that the quality of the studies was poor and the benefit of long‐term mechanical ventilation should be confirmed in further trials.
Continuing uncertainties over the benefits and unwanted effects of TV and NIV suggest that a systematic review of all randomised clinical trials of these interventions is required. The results will need to be considered in the context of non‐randomised studies in the 'Discussion' section of this review.
Objectives
The primary objective of the review is to examine the efficacy of mechanical ventilation (tracheostomy and non‐invasive ventilation) in improving survival in ALS. The secondary objective is to examine the effect of mechanical ventilation on functional measures of disease progression and on quality of life of people with ALS and their carers.
Methods
Criteria for considering studies for this review
Types of studies
All randomised and quasi‐randomised controlled trials, involving non‐invasive or tracheostomy assisted ventilation. Quasi‐randomised trials are those where treatment allocation was intended to be random but may have been biased (for example alternate allocation or allocation according to the day of the week).
Types of participants
All those with a clinical diagnosis of ALS/MND (pure mixed upper motor neuron and lower motor neuron degeneration with supportive electromyogram) according to the El Escorial criteria (Brooks 1994; Brooks 2000), at any state of disease and with any clinical pattern of the condition (e.g., bulbar and limb onset). Subgroups of interest will be participants with or without significant bulbar symptoms as categorised by the authors of the papers reviewed.
Types of interventions
All forms of non‐invasive ventilation (using a nasal or facial mask or mouth piece) and tracheostomy assisted ventilation.
Types of outcome measures
Primary outcomes
The primary outcome will be overall survival after initiation of assisted ventilation as assessed by a pooled hazards ratio using life table/Cox regression methods to combine disparate periods of observation from all studies. This will be supplemented where possible by pooled estimates of the 75%, 50% (median) survival times and confidence intervals if available as appropriate. This is to allow for the situation where the proportional hazards assumption, necessary for Cox regression, has not been met.
Secondary outcomes
The secondary outcome measures will be:
-
survival at one month and six months or longer;
-
quality of life assessed using validated health status questionnaires (for example SF36 (Lyall 2001 b)) at one month and six months or longer;
-
any validated functional rating scale such as the ALS Functional Rating Scale (ALSFRS 1996) or the ALSFRS‐Revised (Cedarbaum 1999), Norris (Norris 1974) or Appel (Haverkamp 1995) scales at one month and six months or longer;
-
the proportion of people experiencing adverse events related to mechanical ventilation. Adverse events will be considered in two categories. The first category will include the proportion of participants experiencing any adverse event attributed to ventilation (for example: fistulae, pneumothorax, bleeding, local infection, hospitalisation or death) and the second category will include participants experiencing severe complications of mechanical ventilation, including life‐threatening episodes, prolonged hospitalisation, and death.
Search methods for identification of studies
(1) We will search the Cochrane Neuromuscular Disease group specialised register for randomised or quasi‐randomised controlled trials using 'amyotrophic lateral sclerosis' or 'ALS' or 'motor neuron disease' or 'motor neurone disease' 'motorneurone disease' or 'motorneuron disease' or 'motoneuron disease' or 'motoneurone disease' or 'MND' combined using AND with 'mechanical ventilation' or 'artificial ventilation' or 'assisted ventilation' or 'artificial respiration' or 'respiratory failure' or 'intubation, intracheal' or 'tracheotomy' or 'tracheostomy' or 'BiPAP' or 'positive pressure ventilation' or 'positivepressure ventilation' or 'non invasive ventilation' or 'noninvasive ventilation'.
We will adapt this strategy to search MEDLINE (January 1966 to the present), EMBASE (January 1980 to the present), CINAHL (January 1982 to the present), AMED (January 1985 to the present) and LILACS (January 1982 to the present). We will check the bibliographies of trials identified and contact the authors and other disease experts seeking published and unpublished trials.
(2) In addition, we will search MEDLINE (January 1966 to the present) and EMBASE (January 1980 to the present) for well‐designed published observational studies of mechanical ventilation in ALS with controls where the diagnosis is clearly stated, the interventions are clearly described and the outcomes are clearly reported for all patients. For example we will collect prospective cohort studies or case control studies with concurrent controls. We will also consider retrospective case series of more than five participants where patients were treated consecutively and experience of ALS researchers at clinic centres. This non‐randomised evidence will be reviewed in the 'Discussion'. This strategy is deemed necessary because we suspect there are no or few randomised controlled trials or quasi‐randomised controlled trials on this particular topic.
Data collection and analysis
Selection of studies
All titles and abstracts identified by the searches for randomised or quasi‐randomised trials will be checked by two reviewers (DA, KJ). All titles and abstracts identified by the searches for non‐randomised evidence for use in the 'Discussion' will be checked by two reviewers (PNL, NM). The reviewers will obtain the full text of all potentially relevant studies and will assess them independently.
Data extraction and management
Two reviewers will extract data independently onto a specially designed form (NL, NM), one reviewer will enter the data into the software (KJ) and one will check the data entered (DA). We will obtain missing or additional data from the authors wherever possible.
Assessment of risk of bias in included studies
All four reviewers will decide which trials fit the inclusion criteria for the review and will then assess the methodological quality of the studies under review. Disagreement about inclusion will be resolved by discussion and consensus between the reviewers. The assessment of methodological quality will take into account: secure randomisation, allocation concealment, blinding of subject and observer, completeness of follow‐up and blinding of outcome assessors. We will grade methodological quality A‐adequate, B‐unclear, C‐inadequate, D‐not done. Where methodological quality is graded C‐unclear, we will attempt to obtain further information from the trial investigators. We will describe the individual study quality criteria scores in an additional table, and discuss the overall reliability of the evidence in the light of these individual quality criteria scores in the text. If possible, we plan to undertake sensitivity analyses to demonstrate the effect of downweighting or ignoring those studies that have received low scores in the individual aspects of quality criteria on the meta‐analyses.
Measures of treatment effect
For the primary outcome measure we will calculate an overall measure of treatment efficacy combining survival results at different time points. This measure is based on estimating a pooled hazard ratio (i.e. at any given time point the risk of death for the survivors in the treated group divided by risk of death for the survivors in the control group) as described by Parmar (Parmar 1998). We will use this measure rather than the summary relative risk (RR) calculated by the Review Manager (RevMan) program, because the Parmar method uses all the data on survival from the whole observation period and the RevMan program needs the survival rates at a fixed point in time since the start of observation to be the same for all the studies if they are to be combined.
For the secondary outcome measures to combine trial results for appropriate pairings of treatments, we will calculate a weighted mean of the difference between their effects using the Cochrane statistical package RevMan 4.2. If we find trials using dichotomous outcome measures such as death rates after a fixed time, e.g. three months, we will obtain relative risks (RR) with 95% confidence intervals (CI). We will express results as weighted mean differences (WMD) with 95% CI for continuous outcomes.
If we find trials where the studies measure continuous outcomes that are conceptually the same but are measured in different ways (such as different assessment scales), we will combine the results and express them as standardised mean differences (SMD ‐using standard deviation units) with 95% CI.
Assessment of heterogeneity
We will test for heterogeneity across trials and if heterogeneity is found we will undertake sensitivity analyses by repeating the calculation omitting the trials which have low scores on individual quality items. If heterogeneity is not explained by variations in trial quality we will use a random effects approach to obtain the pooled estimates from the group of trials.
We will consider non‐randomised evidence concerning adverse events and the cost and cost‐effectiveness of different forms of mechanical ventilation in the 'Discussion' of the review.
