The cost-effectiveness of domiciliary non-invasive ventilation in patients with end-stage chronic obstructive pulmonary disease: a systematic review and economic evaluation

Background: Chronic obstructive pulmonary disease (COPD) is a chronic progressive lung disease characterised by non-reversible airflow obstruction. Exacerbations are a key cause of morbidity and mortality and place a considerable burden on health-care systems. While there is evidence that patients benefit from non-invasive ventilation (NIV) in hospital during an acute exacerbation, evidence supporting home use for more stable COPD patients is limited. In the UK, domiciliary NIV is considered on health economic grounds in patients after three hospital admissions for acute hypercapnic respiratory failure. Objective: To assess the clinical effectiveness and cost-effectiveness of domiciliary NIV by systematic review and economic evaluation. Data sources: Bibliographic databases, conference proceedings and ongoing trial registries up to September 2014. Methods: Standard systematic review methods were used for identifying relevant clinical effectiveness and cost-effectiveness studies assessing NIV compared with usual care or comparing different types of NIV. Risk of bias was assessed using Cochrane guidelines and relevant economic checklists. Results for primary effectiveness outcomes (mortality, hospitalisations, exacerbations and quality of life) were presented, where possible, in forest plots. A speculative Markov decision model was developed to compare the cost-effectiveness of domiciliary NIV with usual care from a UK perspective for post-hospital and more stable populations separately. Results: Thirty-one controlled effectiveness studies were identified, which report a variety of outcomes. For stable patients, a modest volume of evidence found no benefit from domiciliary NIV for survival and some non-significant beneficial trends for hospitalisations and quality of life. For post-hospital patients, no benefit from NIV could be shown in terms of survival (from randomised controlled trials) and findings for hospital admissions were inconsistent and based on limited evidence. No conclusions could be drawn regarding potential benefit from different types of NIV. No cost-effectiveness studies of domiciliary NIV were identified. Economic modelling suggested that NIV may be cost-effective in a stable population at a threshold of (sic)30,000 per quality-adjusted life-year (QALY) gained (incremental cost-effectiveness ratio (sic)28,162), but this is associated with uncertainty. In the case of the post-hospital population, results for three separate base cases ranged from usual care dominating to NIV being cost-effective, with an incremental cost-effectiveness ratio of less than (sic)10,000 per QALY gained. All estimates were sensitive to effectiveness estimates, length of benefit from NIV (currently unknown) and some costs. Modelling suggested that reductions in the rate of hospital admissions per patient per year of 24% and 15% in the stable and post-hospital populations, respectively, are required for NIV to be cost-effective. Limitations: Evidence on key clinical outcomes remains limited, particularly quality-of-life and long-term (> 2 years) effects. Economic modelling should be viewed as speculative because of uncertainty around effect estimates, baseline risks, length of benefit of NIV and limited quality-of-life/utility data. Conclusions: The cost-effectiveness of domiciliary NIV remains uncertain and the findings in this report are sensitive to emergent data. Further evidence is required to identify patients most likely to benefit from domiciliary NIV and to establish optimum time points for starting NIV and equipment settings. Future work recommendations: The results from this report will need to be re-examined in the light of any new trial results, particularly in terms of reducing the uncertainty in the economic model. Any new randomised controlled trials should consider including a sham non-invasive ventilation arm and/or a higher-and lower-pressure arm. Individual participant data analyses may help to determine whether or not there are any patient characteristics or equipment settings that are predictive of a benefit of NIV and to establish optimum time points for starting (and potentially discounting) NIV.