Article Text

Original research
Cost-effectiveness of outpatient versus inpatient non-invasive ventilation setup in obesity hypoventilation syndrome: the OPIP trial
  1. Patrick Brian Murphy1,2,
  2. Maxime Patout3,4,
  3. Gill Arbane1,
  4. Swapna Mandal5,
  5. Georgios Kaltsakas1,2,
  6. Michael I Polkey6,
  7. Mark Elliott7,
  8. Jean-François Muir8,9,
  9. Abdel Douiri10,
  10. David Parkin11,
  11. Jean-Paul Janssens12,
  12. Jean Louis Pépin13,14,
  13. Antoine Cuvelier15,
  14. Clare Flach10,
  15. Nicholas Hart1,2
  1. 1 Lane Fox Respiratory Service, Guy's and St Thomas' Hospitals NHS Trust, London, UK
  2. 2 Centre for Human & Applied Physiological Sciences (CHAPS), King’s College London, London, UK
  3. 3 Service des Pathologies du Sommeil (Département R3S), Groupe Hospitalier Universitaire APHP-Sorbonne Université, Site Pitié-Salpêtrière, Paris, France
  4. 4 UMRS1158 Neurophysiologie Respiratoire Expérimentale et Clinique, Sorbonne Université, INSERM, Paris, France
  5. 5 Thoracic Medicine, Royal Free London NHS Foundation Trust, London, UK
  6. 6 NIHR Respiratory BRU, Royal Brompton Hospital and National Heart and Lung Institute, London, UK
  7. 7 Respiratory Medicine, St James' University Hospital, Leeds, UK
  8. 8 Institute for Research and Innovation in Biomedicine (IRIB), Normandie Univ, UNIRouen, Rouen, France
  9. 9 ADIR Assistance, Fédération ANTADIR, Paris, France
  10. 10 School of Population Health & Environmental Sciences, King’s College London, London, UK
  11. 11 OHE, Office of Health Economics, London, UK
  12. 12 Division of Pulmonary Diseases, Geneva University Hospital, Geneva, Switzerland
  13. 13 HP2 laboratory, INSERM U1042, Universite Grenoble Alpes, Saint-Martin-d'Heres, France
  14. 14 Pôle Locomoteur, Rééducation et Physiologie, CHU de Grenoble, Grenoble, France
  15. 15 Service de Pneumologie, CHU Rouen Normandie, Rouen, France
  1. Correspondence to Dr Patrick Brian Murphy, Guy's and St Thomas' Hospitals NHS Trust, London, London, UK; patrick.murphy{at}


Background Current guidelines recommend that patients with obesity hypoventilation syndrome (OHS) are electively admitted for inpatient initiation of home non-invasive ventilation (NIV). We hypothesised that outpatient NIV setup would be more cost-effective.

Methods Patients with stable OHS referred to six participating European centres for home NIV setup were recruited to an open-labelled clinical trial. Patients were randomised via web-based system using stratification to inpatient setup, with standard fixed level NIV and titrated during an attended overnight respiratory study or outpatient setup using an autotitrating NIV device and a set protocol, including home oximetry. The primary outcome was cost-effectiveness at 3 months with daytime carbon dioxide (PaCO2) as a non-inferiority safety outcome; non-inferiority margin 0.5 kPa. Data were analysed on an intention-to-treat basis. Health-related quality of life (HRQL) was measured using EQ-5D-5L (5 level EQ-5D tool) and costs were converted using purchasing power parities to £(GBP).

Results Between May 2015 and March 2018, 82 patients were randomised. Age 59±14 years, body mass index 47±10 kg/m2 and PaCO2 6.8±0.6 kPa. Safety analysis demonstrated no difference in ∆PaCO2 (difference −0.27 kPa, 95% CI −0.70 to 0.17 kPa). Efficacy analysis showed similar total per-patient costs (inpatient £2962±£580, outpatient £3169±£525; difference £188.20, 95% CI −£61.61 to £438.01) and similar improvement in HRQL (EQ-5D-5L difference −0.006, 95% CI −0.05 to 0.04). There were no differences in secondary outcomes.

Discussion There was no difference in medium-term cost-effectiveness, with similar clinical effectiveness, between outpatient and inpatient NIV setup. The home NIV setup strategy can be led by local resource demand and patient and clinician preference.

Trial registration numbers NCT02342899 and ISRCTN51420481.

  • Non invasive ventilation

Data availability statement

Data are available on reasonable request. Requests for deidentified individual participant data for individual patient data meta-analysis will be considered by the trial steering committee. Applications may be submitted to the corresponding author any time after publication.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • Current guidelines recommend inpatient admission and overnight sleep studies to setup home non-invasive ventilation in patients with chronic respiratory failure. There is evidence to support outpatient setup in patients with obstructive airways or neuromuscular disease but few data in obesity hypoventilation syndrome which is now one of the most common indications for home non-invasive ventilation.


  • This clinical trial demonstrated the safety of an outpatient setup pathway, compared to the guideline recommended inpatient pathway, for home non-invasive ventilation in patients with obesity hypoventilation syndrome. Additionally, there was no difference in cost-effectiveness between setup strategies.


  • The results of the study allow for service models to adopt an inpatient or outpatient setup service dependent on local need.


Sleep-disordered breathing (SDB) is a common comorbid condition in obesity1 and patients with obesity and SDB who develop chronic respiratory failure, in the absence of another cause, are diagnosed with obesity hypoventilation syndrome (OHS).2 3 OHS patients have higher rates of morbidity and healthcare utilisation than patients with simple obstructive sleep apnoea (OSA).4 with rates of OHS increasing in line with rising obesity making the evaluation and management of OHS an important issue for all healthcare systems worldwide.5

The mainstay respiratory therapy for obesity-related SDB is positive airway pressure (PAP) therapy, delivered as continuous PAP (CPAP) or bilevel PAP, also referred to as non-invasive ventilation (NIV). Both therapies reverse chronic respiratory failure,6 improve quality of life7 and reduce healthcare utilisation.8 NIV settings are set by clinicians during initial evaluation to manage SDB. The device settings are then periodically reviewed alongside inpatient or outpatient clinical evaluation.9

Recent advances in NIV technology, permitting autotitration of the NIV settings within prespecified parameters, have been shown to manage hypoventilation and upper airway obstruction.10–12 To date, clinical trials, including those of automated modes of NIV, have all utilised inpatient attended sleep studies to establish NIV settings.8 10 13 The use of an outpatient pathway would facilitate a greater volume of patients as the infrastructure requirement is reduced and thus enabling NIV services to meet the increasing demand for home NIV in OHS patients. This is of particular relevance as we emerge from the COVID-19 pandemic. NIV is an aerosol generating procedure and therefore its use requires the provision of personal protective equipment for medical staff and specialist admission pathways to reduce patient and staff risk. The use of an outpatient pathway would reduce staff exposure by the use of single outpatient rooms for NIV and avoiding multiple staff exposure in a ward environment.

In contrast to OSA safe and effective outpatient diagnostic and therapeutic algorithms, the data to support outpatient NIV setup in OHS patients are limited.14–17 We report an international clinical trial to assess the cost-effectiveness and safety (change in daytime arterial carbon dioxide) of outpatient NIV setup, using an autotitrating NIV device, in patients with OHS compared with standard clinician delivered NIV setup using inpatient attended overnight respiratory monitoring.


A trial steering and independent data monitoring and safety committee were established for trial oversight.

Trial design and patients

The outpatient vs inpatient trial was a phase 3 international multicentre open labelled parallel randomised clinical trial. The full initial trial protocol and amendment have been published previously18 but is reprised in brief. Patients referred to participating home NIV centres for the evaluation of obesity related respiratory failure were screened for trial participation. Eligibility criteria included obesity (body mass index >35 kg/m2), stable chronic hypercapnic respiratory failure (arterial partial pressure carbon dioxide; PaCO2 >6 kPa (criteria altered from initial 6.5 kPa following trial steering committee meeting in July 2016 to assist trial recruitment and align with international definition of OHS); pH ≥7.3, clinical stability off NIV for >2 weeks), SDB (4%ODI>10/hour and/or >30% of total analysis time with an SpO2 <90%). Patients with severe respiratory failure (arterial partial pressure oxygen; PaO2 <7.0 kPa and or PaCO2 >9 kPa), a previous episode of decompensated respiratory failure requiring invasive mechanical ventilation, another identified cause of respiratory failure (eg, COPD or neuromuscular disease), those unable to provide informed consent or comply with the trial protocol were excluded. Patients who had a previous episode of decompensated respiratory failure treated with acute NIV were eligible for trial inclusion if they had only received NIV as an inpatient with discharge from hospital either without therapy or with CPAP and demonstrated clinical stability at trial assessment for over 2 weeks.


Suitable patients were randomised using a web-based randomisation system. Stratification criteria were gender (male or female), randomising centre and previous use of PAP therapy (hospital NIV or hospital or home CPAP in previous 3 months vs PAP naive).


All patients had standard NIV education and interface assessment as per usual care. NIV was delivered to patients in both arms using a home NIV device (A40; Philips, Murrysville, Pennsylvania, USA). Oxygen therapy during daytime was established in patients with daytime hypoxia (PaO2 <7.3 kPa or <8 kPa with evidence of organ dysfunction secondary to chronic hypoxia). Patients requiring oxygen therapy had oxygen entrained into the NIV circuit at the same flow rate as the daytime prescription.

Detailed description of the inpatient and outpatient setup can be found in the eMethods section of the online supplement (OLS). In brief inpatients were admitted and underwent overnight respiratory polygraphy, including transcutaneous carbon dioxide (CO2), and titration of NIV (online supplemental eFigure E1) until adequate SDB control or maximal tolerated pressures were reached. Outpatient setup was conducted by an experienced NIV clinician (doctor/nurse/physiotherapist) using a predefined protocol (online supplemental eFigure E2) and an autotitrating NIV mode (AVAPS-AE). Initial acclimatisation and setting adjustment occurred during daytime with home overnight oximetry performed in the first 2 weeks of NIV use to assess SDB control and settings manipulated as required. Participants were followed up by the clinical centre as per usual clinical practice during the 3-month trial period including, but not limited to, a face-to-face assessment at 6 weeks.

Supplemental material

Supplemental material

Primary outcome and clinical safety analysis

The primary outcome was defined as the difference in treatment costs and health-related quality of life (HRQOL) between inpatient and outpatient NIV setup over the trial period (3 months). An a priori safety analysis was embedded to ensure clinical efficacy was equivalent. The safety analysis was defined as non-inferiority of outpatient versus inpatient NIV setup in terms of PaCO2 reduction. We hypothesised that outpatient NIV setup would be more cost-effective than inpatient NIV setup (superiority) with similar clinical efficacy (non-inferiority).

Outcomes and assessments

All outcomes were assessed at trial completion (3 months). NIV setup costs (fixed costs) were calculated based on costing of NIV service delivery for the largest NIV unit in each country as described in online supplemental eMethods. Healthcare utilisation (variable costs) was collected at each study assessment, including telephone consultations and home visits from the study and clinical team which were costed at the hourly rate attributable of the appropriate grade and profession delivering care and costed using national representative costings for 2017–2018. In order to equalise the purchasing power of the currencies costs were converted into £(GBP) from €(euro) using the purchasing power parities methodology at an exchange rate of €1.10–£1.00.19

Supplemental material

Details on methods for measurements of secondary outcomes can be found in online supplemental eMethods.

Statistical analysis

The sample was described by arm in terms of baseline characteristics using means and SDs or frequencies and per cent as appropriate. All analyses were conducted under a modified intention to treat analysis to include all patients with available data. The treatment effect was reported with 95% CI and corresponding non-inferiority p value.

The primary outcome of the trial was defined as the difference in treatment costs and HRQOL between NIV setup strategies. Using previously published data, the change in HRQOL, measured by the Severe Respiratory Insufficiency questionnaire (SRI), was 20% in the group receiving autotitrating NIV and 12% in the standard fixed level NIV group.10 A sample size of 74 randomised 1:1 provides a power of 90% at a two-sided significance level of 0.05 using an analysis of covariance (ANCOVA) analysis. Accounting for a drop-out rate of 10% required a total sample size of 82.

The value of patient reported HRQOL was estimated using EQ-5D-5L data, converted using crosswalk calculator into EQ-5D-3L values according to published value sets for the UK and France.20 These were converted into quality-adjusted life months (QALMs) for each patient by the area under the curve method assuming a linear interpolation between follow-up points. Considering the correlation between each outcome due to the data from the same subjects, seemingly unrelated regression with bootstrapped confidence intervals was used to estimate the incremental mean cost, QALMs and SRI between treatment groups after adjusting for country (other stratification variables were not included due to low sample numbers). Baseline SRI score was included as an adjustment in the SRI regression analysis. Analyses were conducted on a complete case basis. The appropriate EQ-5D value sets and costs were applied to individuals in each country. Analysis was carried out for each country separately and combined.

Secondary outcomes were analysed using linear regression models produced of outcomes at month three adjusting for baseline values and country.

An a priori safety analysis was embedded to ensure clinical efficacy was equivalent. The safety analysis was defined as non-inferiority of outpatient NIV setup in term of the decrease in PaCO2. Non-inferiority was assessed using an ANCOVA model adjusting for baseline PaCO2. A non-inferiority margin of 0.5 kPa was prespecified based on clinical consensus. Data from previous work demonstrated a SD of PaCO2 at 3 months of 0.7 kPa.10 Therefore, 74 patients randomised 1:1 provides a power of 86% using an ANCOVA analysis with a single sided level of significance set at 0.025.


Seven centres with established treatment programmes for chronic respiratory failure were selected to participate in the study; four from the UK, two from France and one from Switzerland. The first patient was recruited on 13 May 2015 and the last study assessment was performed on 7 August 2018. Following a trial steering committee meeting in January 2017, the Swiss centre withdrew from the trial due to trial infeasibility having screened 71 patients without randomisation of a patient. The subsequent data reported are from the four UK and two French centres. There were 385 patients screened at the 6 centres which randomised patients for trial participation of which 82 were randomised (UK 53 and France 29) and 76 (inpatient 37, outpatient 39) completed the final trial assessment (figure 1). The most common reason for screening failure was absence of chronic respiratory failure, that is, the patients did not meet the definition for OHS. A total of six participants withdrew from the study (two from each group following randomisation and before the 6-week trial visit and a further two participants in the inpatient group withdrew between the 6-week and 3-month visit).

Figure 1

Trial profile and patient participation diagram. BMI, body mass index; NIV, non-invasive ventilation.

Baseline demographic data are reported by allocated group (table 1) and country of randomisation (online supplemental OLS table E1).

Table 1

Baseline demographics by allocated non-invasive ventilation (NIV) setup strategy

Severity of SDB, based on baseline sleep study, was similar between groups (table 1). Supplementary oxygen was required by 18 patients (inpatient 8, outpatient 10; mean flow rate 0.4±0.9 and 0.4±0.7 Lpm, respectively). Humidification was added as per patient preference in 53 patients (inpatient 27, 68%; outpatient 26, 65%). Inpatient setup required a median 1.0 range (1.0–2.0) night stay. Discharge NIV settings for the inpatient group, set during overnight sleep study, were inspiratory PAP (IPAP) of 22.8±6.3 cmH2O, expiratory PAP of 9.8±3.3 cmH2O and back up rate 14.4±2.9. The outpatient group had a target tidal volume of 575±83 mL set to be delivered by variable pressure support mode. Full device settings can be found in the online supplemental OLS eTable 9.

The initial safety analysis conducted demonstrated improvement in PaCO2 in both groups (inpatient −0.44±1.06 kPa; outpatient −0.85±1.04 kPa) with the 95% CI of the change in PaCO2 within the prespecified non-inferiority margin allowing for subsequent health economic analysis (figure 2).

Figure 2

Comparison of change in PaCO2 between NIV setup strategies. PaCO2, arterial partial pressure of carbon dioxide, The blue dashed line labelled Δ represents the non-inferiority threshold or the maximum allowable excess of PaCO2 difference arising from the Out-patient NIV compared with the inpatient NIV. The tinted area represents the non-inferiority zone. NIV, non-invasive ventilation.

Total per patient cost (fixed, OHS-related and non-OHS-related healthcare utilisation) for inpatient setup (£2962±£580) were similar to outpatient cost (£3169±£525) with no significant difference in cost of NIV setup between groups (£188.20, 95% CI £−61.61 to £438.01). Healthcare utilisation and costs differed between groups with higher healthcare utilisation in the outpatient arm (table 2).

Table 2

Cost of NIV setup (fixed and variable costs) collected over trial period (3 months)

There were similar HRQL improvements at month three in both groups when measured by EQ-5D-5L (inpatient 0.11±0.09, outpatient 0.11±0.10; difference −0.006, 95% CI −0.05 to 0.04) and SRI (inpatient 56.4±20.1, outpatient 62.9±21.2; difference −1.96, 95% CI −8.8 to 4.9). QALMs were similar between groups (inpatient 1.35±1.06, outpatient 1.28±1.19; mean difference −0.07, 95% CI −0.57 to 0.43 months). When applying country specific costs to own country data there was no difference in cost-effectiveness whether inpatient or outpatient NIV setup was used (table 3).

Table 3

Cost-effectiveness analysis of inpatient and outpatient non-invasive ventilation setup in obesity hypoventilation syndrome

Health economic sensitivity analysis extrapolating country specific costs across the dataset (French costs used for patients recruited in both UK and France and conversely UK costs used for all patients) demonstrated a cost saving in the French health system when using inpatient NIV setup (mean difference £700.18, 95% CI £449.04 to £951.32) with the cost difference attributable to reduced cost of inpatient titration (online supplemental OLS eTables 2–3).

Clinical secondary outcomes

There was no difference in clinically relevant secondary outcomes between inpatient and outpatient setup at 3 months, including change dyspnoea, lung function, exercise capacity, body composition and anthropometrics. Furthermore, there was no change in quality of life, with the exception of the physical function domain of the SRI score which favoured patients established on NIV using the inpatient strategy (∆−8.90 95% CI −17.4 to −0.37; p=0.041) (online supplemental OLS eTables 4–6).

Management of SDB, change in sleep quality and NIV settings at 3 months

There was no difference between inpatient and outpatient setup in terms of SDB management, subjective and objective sleep quality, NIV adherence and NIV settings at 3 months (online supplemental OLS eTables 7–9).


This current clinical trial demonstrated that there was no difference in medium-term cost-effectiveness between home NIV setup employing an outpatient pathway and an autotitrating NIV device, compared with inpatient NIV setup incorporating clinician-led attended overnight titration, in clinically stable ambulatory patients with OHS. Furthermore, these current data confirm the safety of an outpatient pathway with a similar improvement in daytime carbon dioxide level at 3 months compared with inpatient home NIV setup.

Clinical outcomes

At 3 months, there was similar overnight control of SDB when using either an inpatient attended nurse-led manual NIV titration or an outpatient clinician-led auto-NIV titration approach. It is important to acknowledge that there are important parameters that require close attention to optimally set an auto-NIV device to ensure that there is appropriate acclimatisation and delivery of therapeutic NIV. However, these data confirm that this approach produces not only similar control of SDB but also important physiological and clinical parameters including daytime symptoms, sleep quality, HRQOL, lung function and exercise performance. Although a statistical difference was detected in the physical function domain of the SRI this must be viewed in the context of a secondary outcome without correction for multiplicity. Furthermore, the analysis of objective physical activity on actigraphy, patient reported activity measured on visual analogue scale and exercise capacity indicated no difference between groups and is therefore not supportive of a clinically relevant difference.

Healthcare utilisation

Of importance is that the outpatient arm required more healthcare contacts (outpatient hospital visits, telephone calls, hospital stays and emergency home visits) and more frequent modification of ventilator settings (ventilator setting changes post initial NIV setup: inpatient 56% vs outpatient 62%). The outpatient setup model used the autotitrating device, daytime acclimatisation and home oximetry which may have failed to control nocturnal hypoventilation in the initial period, although control of SDB was similar at 3 months once the additional modifications to NIV settings had been completed. The difference in healthcare utilisation required to alter NIV settings after the initial setup impacted on the overall cost of the setup strategy. We acknowledge the use of advanced home monitoring with transcutaneous carbon dioxide monitoring may have altered the need for additional NIV setting manipulation, but this would have required more patient training and specialist equipment, which is not routinely available. Indeed, this would have further added to healthcare costs.

The study protocol used both telephone and face to face visits to modify ventilator settings in response to patient symptoms and overnight oximetry results. This process may have been simplified with an impact on health practitioner time, and thus cost, if remote monitoring system capable of NIV device prescription changes capacity were used in the study. However, this approach has not been tested robustly and at the start of this study this type of technology was not available.

Comparison with previous studies

Previous data investigating the economic advantage of outpatient NIV setup in chronic respiratory failure has indicated cost reductions and similar efficacy of therapy. The cost reductions reported with outpatient NIV setup are largely delivered through the reduction in inpatient bed day cost. In the study by Luján et al,17 the cost saving was demonstrated to be £1019 (inpatient £2301 vs outpatient £1755), however, this was based on a median length of stay of 6 days to establish home NIV.16 If the median length of stay were reduced to 2 days (one night), similar to that observed in the current clinical trial, the cost of the outpatient setup would be similar to inpatient setup. Interestingly, the data of Luján et al showed excess costs in the outpatient arm that were driven by unscheduled emergency hospitalisation events (inpatient £216 vs outpatient £5289).16 The reasons for this significant cost difference are speculative but could include the lower contact time in the outpatient group reducing the time for education for NIV troubleshooting and education. This rationale is supported by data from other studies assessing inpatient and home setup of NIV where direct contact time with the nurse specialist was shown to be lower in the home setup group compared with inpatient group.21 Furthermore, the time to acclimatisation to home NIV was longer with home setup (14 days vs 7 days; p<0.001).22

Unscheduled care

In the current trial, the excess treatment costs in the outpatient NIV setup group were driven by the cost of the autotitrating ventilator, outpatient hospital visits, telephone calls, unscheduled hospital stays and emergency home visits. Similar to the study by Lujan et al, the current trial demonstrated a predominance of unscheduled admissions to hospital in the outpatient NIV setup group.16 It would be possible to modify the outpatient strategy to increase contact time in an attempt to reduce unplanned admissions but this would require further work to demonstrate that this is effective. The imbalance of emergency admissions provides an argument to use the inpatient approach for those patients with clinical instability, significant comorbidities and when unplanned admissions are important to avoid from a patient and healthcare perspective. A home ventilation service should offer both inpatient and outpatient approach with a personalised assessment of patient preference, risk of emergency admission, clinical stability and predicted adherence to select the setup strategy that will best deliver care for the individual.

Critique of the method

Although the authors acknowledge that the data from this trial are limited as there was no blinding of the participants or supervising clinicians to the allocation group, blinded assessors were used for outcome measures whenever possible. In addition, and unlike other health economic evaluations which use healthcare tariffs (‘top down’ costings provided by healthcare provider price schedules),16 22 23 this present trial comprehensively costed inpatient and outpatient NIV setup using the ‘bottom up’ costing for each of the home ventilation centres involved. The actual cost of service delivery was based on unit expenditure and activity. However, it must be acknowledged that for the healthcare utilisation outside of NIV setup, generic costs were used in-line with other published data. It is important to understand that the cost-effectiveness data presented cannot necessarily be extrapolated beyond the time horizon of the trial as patterns of healthcare utilisation cannot be assumed to remain stable, although healthcare utilisation is known to be highest in the year before and after diagnosis of OHS and so healthcare utilisation is likely to fall over time.24

The use of a multicentre international study design allowed for evaluation of country specific costs. We demonstrated lower cost for inpatient NIV setup in France compared with the UK, which was reflected in the sensitivity analysis (online supplemental OLS eTable 2) as a cost saving compared with the outpatient model. This was as a consequence of the lower daily inpatient bed cost to establish NIV in the French healthcare system. The NIV setup in the French centres was delivered using inpatient sleep laboratory services rather than through respiratory critical care services as in the UK as we acknowledge that the design of the service, both inpatient and outpatient, will influence the relative cost-effectiveness of the chosen NIV setup model. The results of previous single centre and single country studies15 16 22 must be considered with caution outside of the healthcare system where the assessment were made, whereas the data presented in the current study will be generalisable to both the UK and French home ventilation services.

The study utilised an auto-NIV device which has been shown to be clinically effective compared to fixed bilevel NIV in patients with OHS titrate in an inpatient setting.13 Whilst CPAP has been shown to deliver similar medium and long-term clinical outcomes to NIV in the management of stable OHS this has been in the context of inpatient polysomnography titrated CPAP6 8 whereas auto-titrating CPAP has not been evaluated in the management of OHS and therefore was not felt suitable for the study comparator. Future studies could examine this area to further simplify the treatment pathway for patient with OHS and severe OSA.

While a full randomisation process was used with stratification there were differences in the baseline data between groups. A numerical difference in Epworth sleepiness score (13.1 vs 9.9) above the accepted minimum clinically important difference was noted as was a potentially clinically important difference in lung function (FVC 2.2 L vs 1.8 L). However, our planned statistical analysis accounted for baseline values and thus these baseline differences would ne be expected to impact on the primary outcome or the safety outcome. It must be appreciated that the cohort recruited for the study were stable ambulatory patients with OHS with moderate to high levels of obesity, moderate restrictive ventilatory defect and mild to moderate hypercapnia and that the results cannot be extrapolated to patients in the postacute phase or with more severe respiratory failure.

Future home NIV services

These data provide strong evidence to underpin the planning of home NIV provision for obese patients with chronic respiratory failure. Development and delivery of chronic respiratory failure pathways to manage the increasing number of obese patients requiring NIV by employing autotitrating NIV devices in an outpatient setting which will reduce the demand on inpatient capacity.17 However, clinicians must be cognisant that there is still a requirement for adequate contact time if adopting an outpatient setup pathway for stable OHS patients using autotitrating NIV as these patients have a higher incidence of acute deterioration as evidenced by the current data. The ability to flex the setup strategy for home NIV may have particular relevance during the recovery stage of the COVID-19 pandemic. NIV is an aerosol generating procedure and, therefore, it is recommended that it be conducted in side rooms with staff using personal protective equipment. This will further increase costs and reduce availability of inpatient beds where current inpatient delivery occurs on opens wards and may be more easily contained in single room, well-ventilated outpatient setting.


There was no difference in medium-term cost-effectiveness based on costings from the UK or French health systems, and similar clinical efficacy, between outpatient NIV setup using an autotitrating device and inpatient NIV setup guided by nurse-led overnight titration in patients with stable OHS referred to ambulatory services.

Data availability statement

Data are available on reasonable request. Requests for deidentified individual participant data for individual patient data meta-analysis will be considered by the trial steering committee. Applications may be submitted to the corresponding author any time after publication.

Ethics statements

Patient consent for publication

Ethics approval

The trial was registered prospectively ( NCT02342899) and approved by relevant ethics committees (UK: 11/LO/0414, France: 2015-A00058-41) local research and development committees at participating centers. All recruited patients provided written informed consent and all trial procedures conformed to local policies.


Supplementary materials


  • Twitter @maximepatout, @DrPanik, @NickHartThorax

  • Contributors Conception and design: NH, PBM, GA, MIP, ME, JF-M, AD, J-PJ, JLP, AC and SM. Data collection: PBM, MP, GA, GK and J-PJ. Data interpretation: NH, PBM, AD, CF, DP. Manuscript drafting: NH, PBM and CF. Manuscript review, critical appraisal and final approval: NH, PBM, CF. MP, GK, GA, MIP, ME, JF-M, AD, J-PJ, JLP, AC, SM, DP.

  • Funding The study was supported by an unrestricted educational grant from Philips-Respironics. The study was supported by Guy’s and St Thomas’ NHS Foundation Trust and King’s College London, National Institute of Health Research Comprehensive Biomedical Research Centre, London, UK and the NIHR Respiratory Disease Biomedical Research Unit at the Royal Brompton and Harefield NHS Foundation Trust and Imperial College London.

  • Disclaimer The funders were not involved in design and conduct of the study; collection, management, analysis and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Philips-Respironics provided the A40 devices and Actiwatch spectrum devices used in the study.

  • Competing interests Detailed conflicts of interest forms are supplied for each author.Role of funderThe study was supported by an unrestricted educational grant from Philips-Respironics. The funders were not involved in design and conduct of the study; collection, management, analysis and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Philips-Respironics provided the A40 devices and Actiwatch spectrum devices used in the study. The study was supported by Guy’s and St Thomas’ NHS Foundation Trust and King’s College London, National Institute of Health Research Comprehensive Biomedical Research Centre, London, UK and the NIHR Respiratory Disease Biomedical Research Unit at the Royal Brompton and Harefield NHS Foundation Trust and Imperial College London.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Linked Articles