Annals of the American Thoracic Society

Rationale: Hospitalized patients with acute-on-chronic hypercapnic respiratory failure due to obesity hypoventilation syndrome (OHS) have increased short-term mortality. It is unknown whether prescribing empiric positive airway pressure (PAP) at the time of hospital discharge reduces mortality compared with waiting for an outpatient evaluation (i.e., outpatient sleep study and outpatient PAP titration).

Objectives: An international, multidisciplinary panel of experts developed clinical practice guidelines on OHS for the American Thoracic Society. The guideline panel asked whether hospitalized adult patients with acute-on-chronic hypercapnic respiratory failure suspected of having OHS, in whom the diagnosis has not yet been made, should be discharged from the hospital with or without empiric PAP treatment until the diagnosis of OHS is either confirmed or ruled out.

Methods: A systematic review with individual patient data meta-analyses was performed to inform the guideline panel’s recommendation. Grading of Recommendations, Assessment, Development, and Evaluation was used to summarize evidence and appraise quality.

Results: The literature search identified 2,994 articles. There were no randomized trials. Ten studies met a priori study selection criteria, including two nonrandomized comparative studies and eight nonrandomized noncomparative studies. Individual patient data on hospitalized patients who survived to hospital discharge were obtained from nine of the studies and included a total of 1,162 patients (1,043 discharged with PAP and 119 discharged without PAP). Empiric noninvasive ventilation was prescribed in 91.5% of patients discharged on PAP, and the remainder received empiric continuous PAP. Discharge with PAP reduced mortality at 3 months (relative risk 0.12, 95% confidence interval 0.05–0.30, risk difference −14.5%). Certainty in the estimated effects was very low.

Conclusions: Hospital discharge with PAP reduces mortality following acute-on-chronic hypercapnic respiratory failure in patients with OHS or suspected of having OHS. Well-designed clinical trials are needed to confirm this finding.

Obesity hypoventilation syndrome (OHS) is defined by obesity (body mass index or BMI ≥30 kg/m2), sleep-disordered breathing, and daytime hypercapnia during wakefulness (partial pressure of arterial CO2 or PaCO2 ≥ 45 mm Hg at sea level) that is not due to other conditions associated with hypercapnia (1). OHS is both common and consequential. Its prevalence is estimated between 17% and 30% among high-risk groups of individuals such as patients with obesity and obstructive sleep apnea (OSA) (2) and is likely to increase because the prevalence of severe obesity is increasing (3, 4). Patients with untreated OHS are at increased risk for respiratory and cardiovascular morbidity and mortality (5, 6).

Positive airway pressure (PAP) therapy has become the primary management option for controlling sleep-disordered breathing and reversing awake hypoventilation in patients with OHS. The most commonly prescribed PAP treatment modalities are noninvasive ventilation (NIV) or continuous positive airway pressure (CPAP). NIV consists of the application of positive-pressure ventilation, usually with bilevel PAP settings (with or without a backup respiratory rate) or volume-targeted pressure support, an autotitrating pressure support mode that delivers to a preset target volume and includes a backup respiratory rate.

Hospitalized patients with OHS or suspected of having OHS experiencing acute-on-chronic hypercapnic respiratory failure have higher short-term mortality compared with ambulatory patients with OHS (5, 7, 8). It is unknown whether prescribing empiric PAP at the time of hospital discharge reduces mortality compared with awaiting outpatient workup such as a sleep study and outpatient PAP titration. Thus far, there have been no clinical trials addressing this clinically relevant question.

The American Thoracic Society recently developed clinical practice guidelines on the evaluation and management of OHS (9). The guideline panel asked, “Should hospitalized adults suspected of having OHS, in whom the diagnosis has not yet been made, be discharged from the hospital with or without PAP treatment until the diagnosis of OHS is either confirmed or excluded?” This systematic review and individual patient data meta-analyses were conducted to answer the question.

The methods used to conduct this systematic review were similar to what we described in our prior study (10) and adhered to standards described in detail in the Cochrane Handbook for Systematic Reviews of Interventions (11). The systematic review was not registered because it was performed as a component of guideline development rather than as standalone research. The guideline, systematic review, and individual patient data meta-analyses were funded by the American Thoracic Society.

Research Question

We used the population, intervention, comparator, outcome format to frame the guideline panel’s question, “Should hospitalized adults suspected of having OHS in whom the diagnosis has not yet been made, be discharged from the hospital with or without PAP treatment until the diagnosis of OHS is either confirmed or excluded?” The population was defined as hospitalized patients with confirmed or suspected OHS, the intervention as hospital discharge on any mode of PAP, and the comparator as hospital discharge with no PAP. The panel of experts that participated in the guideline development identified and prioritized patient-important outcomes using a 9-point Likert rating scale. Critical outcomes included all-cause mortality, resolution of OHS, quality of life, daytime sleepiness, cardiovascular events, hospitalizations, and motor vehicle accidents. A patient advocate actively participated in assessing the importance of the clinical outcomes from a patient’s perspective.

Literature Search

Our methods for study selection have been described in detail in the published clinical practice guidelines (9). We first searched the Cochrane Library and Medline for recent relevant systematic reviews. Given that there were no systematic reviews that addressed our question, we developed a search strategy to identify studies that evaluated PAP as a treatment for OHS. Our search was broad to identify both direct evidence and indirect evidence. Medline (using the Ovid interface) and Embase were searched for articles from January 1946 to March 2019. ClinicalTrials.gov was also searched for ongoing trials. Last, guideline panelists were asked to identify any additional studies that may be relevant and the search had not identified (see Table E1 in the online supplement).

Study Selection

Studies were sought that enrolled hospitalized patients with confirmed or suspected OHS and evaluated the effects of PAP. We planned to include randomized trials, nonrandomized comparative studies (i.e., prospective and retrospective cohort, case–control, before–after studies), and nonrandomized studies without a comparator (i.e., single-arm cohorts, case series). In contrast, we planned to exclude case reports and studies that enrolled patients who did not have OHS, had a tracheostomy, or had diseases that could also lead to hypercapnia such as COPD, interstitial lung disease, neuromuscular disease, and chest wall disease. All PAP modalities (i.e., CPAP or any type of NIV) were allowed. Outcomes were not criteria for selecting studies. Study selection was performed in duplicate by two authors; disagreements were resolved through discussion and consensus.

Data Acquisition

For each selected study, the guideline cochairs (B.M. and J.F.M.) contacted the study authors and requested individual patient data on age, sex, BMI, baseline arterial blood gases (pH, PaCO2, and PaO2), arterial blood gases upon hospital discharge, forced expiratory volume in one second (FEV1), forced vital capacity (FVC), FEV1/FVC, whether the patient was discharged on PAP, type of PAP (NIV or CPAP), and survival status at 3, 6, 9, and 12 months after hospital discharge. Details regarding PAP titration and setting adjustments were not available. The individual patient data were recorded in a spreadsheet developed for the systematic review, along with information about the study setting, participant demographics, details of the intervention, study design, follow-up, and criteria for risk of bias according to the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) approach. Missing BMI and baseline PaCO2 data were replaced by the study’s mean BMI and baseline PaCO2, respectively.

Data Synthesis

Two types of individual patient data meta-analyses were performed (IPD-MA). First, using individual patient data from nonrandomized comparative studies (studies comparing PAP to no PAP within a population), absolute and relative effects were calculated for each study, and those effects were pooled across studies using a Mantel-Haenzsel random effects model. The advantage of this approach is that because PAP and no PAP are compared within populations, the pooled effect is likely due to PAP rather than population differences. The disadvantage is that the total number of patients analyzed is relatively small.

Second, to overcome the disadvantage of the previous approach, individual patient data from both nonrandomized comparative and nonrandomized noncomparative studies (studies reporting outcomes of PAP, without comparing PAP to no PAP) were used to create groups of patients with similar baseline characteristics in which PAP and no PAP could be compared, like a controlled study. To create the groups, propensity scores were estimated for each patient using logistic regression that incorporated age, BMI, and severity of OHS (defined by the baseline PaCO2) as pertinent baseline characteristics. Other variables such as spirometry values or PaO2 could not be incorporated into the propensity score calculations due to the large percentage of missing data. Patients were initially aggregated into quintiles on the basis of their propensity score; however, some quintiles had very few patients who did not receive PAP, so the patients were aggregated into tertiles instead. The tertiles became the groups in which PAP and no PAP were compared. The absolute and relative effects were calculated for each group, and then those effects were pooled across studies using a Mantel-Haenzsel random effects model.

Severity of OHS (defined by the baseline PaCO2), age, and BMI were determined a priori to be subgroups of interest. For each subgroup, patients with values above or equal to the median were compared with patients with values below the median. For both types of IPD-MA, relative effects were reported as a relative risk (RR), and absolute effects were reported as a risk difference (RD). The 95% confidence interval (CI) was provided for all estimates. Heterogeneity was assessed using the I2 statistic and, when encountered, sensitivity analyses. Commercial software (Review Manager, version 5.1.1) was used to conduct the meta-analyses.

Quality of Evidence Appraisal

Quality of evidence indicates certainty in the estimated effects. It was appraised using the GRADE approach, which assesses five domains: risk of bias (internal validity), indirectness (external validity), inconsistency (increased I2 statistic), imprecision (wide confidence intervals), and likelihood of publication bias (1216). To assess the risk of bias domain, the Newcastle-Ottawa scale for nonrandomized studies was used. A flaw in any domain is a reason to downgrade the quality of evidence. The GRADE approach also assesses reasons to upgrade quality (17).

Evidence Profile

To summarize the estimated effects, an evidence profile table was developed for each outcome and included the judgments made during the appraisal of the quality of evidence.

Through our search we identified 2,994 relevant articles. Only 10 articles were identified that included hospitalized patients (5, 6, 1825) (Figure E1). The cochairs (B.M. and J.F.M.) requested and obtained limited individual patient data for patients who survived hospitalization from the authors of nine studies. Data could not be obtained for one study which included 47 patients (5). There were no randomized trials. The studies included two nonrandomized comparative studies (i.e., observational studies) (18, 25) and seven nonrandomized noncomparative studies (i.e., case series) (6, 1924). In four studies, OHS was defined as BMI ≥30 kg/m2, PaCO2 ≥45 mm Hg, and FEV1/FVC >70% (18, 2022). Three studies used slight variations of the above-mentioned criteria to define OHS: same BMI and PaCO2 criteria, but FEV1/FVC ≥60% (19); same BMI and FEV1/FVC criteria, but with PaCO2 >50 mm Hg and PaO2 <60 mm Hg (6); same PaCO2 and FEV1/FVC criteria, but BMI >40 kg/m2 (23). A registry study defined OHS as BMI >30 kg/m2, a primary diagnosis of hypoventilation due to obesity or OSA, and exclusion of possible overlap syndrome (FEV1/FVC <70%). However, nearly 44% of the 573 patients in this registry did not have spirometry data available (24). Another study defined OHS as BMI ≥30 kg/m2, PaCO2 ≥45 mm Hg, but 39% of the patients in this study lacked data on spirometry (25). We considered patients from these two studies as “suspected of having OHS,” given that a large proportion were missing spirometry data (24, 25). However, despite the missing information on spirometry, all nine studies explicitly stated that patients with other conditions that could explain the presence of hypercapnia (i.e., asthma/COPD, pneumonia, interstitial lung disease, neuromuscular disorders, kyphoscoliosis, and hypothyroidism) were excluded. The studies collectively enrolled 1,275 hospitalized patients with acute-on-chronic hypercapnic respiratory failure due to OHS. In three studies, the entire cohort consisted of hospitalized patients (18, 19, 25). The remaining six studies had a subgroup of patients who were hospitalized (6, 2024). Individual patient data were obtained for 1,162 (91%) patients who survived hospitalization and were discharged. In aggregate, 90% of patients (n = 1,043) were discharged on some form of PAP therapy. The remaining 10% of patients (n = 119) who survived hospitalization were discharged from the hospital without any PAP therapy. NIV was prescribed in all patients from seven studies (n = 955) (6, 1921, 2325). In two studies (18, 22), patients were discharged on either CPAP or NIV. PAP settings upon discharge and outpatient PAP adherence data were not available.

The only outcome that was reported by all studies was mortality. Other outcomes deemed critical by the guideline panel were not reported. Among the individual patient data, cause of death was not available, and reliable data on spirometry (either FEV1, FVC, or both) and arterial blood gases upon discharge were missing from 528 (45%) and 834 (72%) of patients, respectively.

Two nonrandomized comparative studies enrolled patients with acute-on-chronic hypercapnic respiratory failure due to OHS and compared discharge with PAP to discharge without PAP (18, 25). Mean patient age was 73 and 74 years, mean BMI was 41 and 42 kg/m2, and mean baseline PaCO2 was 76 and 84 mm Hg (Table 1). The seven nonrandomized, noncomparative studies included one study that enrolled patients with acute-on-chronic hypercapnic respiratory failure due to OHS exclusively (19). The other studies enrolled both ambulatory patients with stable chronic OHS and hospitalized patients with acute-on-chronic hypercapnic respiratory failure due to OHS (6, 2024); for these studies, we requested individual patient data on the subgroup of hospitalized patients only. In the acute-on-chronic hypercapnic respiratory failure subgroup, mean patient age was between 54 and 64 years, mean BMI was between 40 and 54 kg/m2, and mean baseline PaCO2 was between 54 and 70 mm Hg (Table 1). Most studies had low to moderate risk of bias due to possible selection bias.

Table 1. Characteristics of selected studies

AuthorYearDurationPatients, including NSetting/DesignAge (yr)BMI (kg/m2)Baseline PaCO2 (mm Hg)Type of PAPRisk of Bias
Randomized Trials
Nonrandomized Studies with a no-PAP Control Group
 Carillo et al. (18)20121 yr163 consecutively enrolled patients with acute-on-chronic hypercapnic respiratory failure due to OHS survived hospital discharge. IPD was obtained for all 163 patients. Survival status could be verified in 159 patients (87 discharged on PAP and 72 discharged with no PAP).18-bed ICU in a general academic hospital in Spain73.7 ± 10.941.8 ± 5.784.1 ± 17.9Bilevel PAP ST (n = 16) or CPAP (n = 71)Low risk of bias
 Romero et al. (25)20141.32 ± 0.94 yr178 consecutively enrolled obese patients with acute-on-chronic hypercapnic respiratory failure due to OHS were discharged from the hospital. IPD was obtained for all 178 patients. Survival status could be verified in 157 patients (110 discharged on bilevel PAP and 47 discharged with no PAP).General wards, respiratory wards, and emergency department in a general academic hospital in Spain73.0 ± 12.241.4 ± 3.776.4 ± 15.8Bilevel PAPLow risk of bias
Nonrandomized Studies without a no-PAP Control Group
 Borel et al. (20)201343 ± 14 mo107 patients with OHS treated with NIV. Among these, NIV was initiated during hospitalization due to acute-on-chronic hypercapnic respiratory failure in 38 patients who survived to hospital discharge. IPD was obtained for all 38 patients.One tertiary care hospital, one general hospital, and three private practices in France63.6 ± 9.240.5 ± 7.054 ± 13.3Bilevel PAP STModerate risk of bias due to possible selection bias
 Budweiser et al. (19)200741.3 ± 27.6 mo126 hospitalized patients with OHS who were discharged with NIV. IPD obtained for all 126 patients.One university-affiliated hospital in Germany55.6 ± 10.644.6 ± 7.855.5 ± 7.7Bilevel PAPModerate risk of bias due to possible selection bias
 Castro-Añón et al. (6)20157 ± 4 yr110 patients with OHS. Among these, 29 were started on NIV during hospitalization due to acute-on-chronic hypercapnic respiratory failure and survived hospital discharge. IPD was obtained for all 29 patients.823-bed university hospital in Spain61.6 ± 11.242.3 ± 8.668.6 ± 10.4NIV mostly in form of bilevel PAPModerate risk of bias due to possible selection bias
 Howard et al. (22)20173 mo60 patients with OHS, 25 hospitalized patients following an episode of acute-on-chronic hypercapnic respiratory failure. IPD was obtained for all 25 patients.Ventilatory failure services from two academic medical centers in Australia54.8 ± 11.953.2 ± 10.667.3 ± 17CPAP (n = 11) and Bilevel PAP ST (n = 14)Moderate risk of bias due to possible selection bias
 Murphy et al. (23)20123 mo50 consecutive patients with OHS referred for elective assessment of stable disease or following treatment of acute hypercapnic respiratory failure. Among these, 17 were hospitalized patients who had acute-on-chronic hypercapnic respiratory failure. IPD was obtained for all 17 patients.Respiratory units in two academic medical centers in the United Kingdom58.5 ± 11.751.1 ± 8.454.8 ± 6.3Bilevel PAP ST or AVAPSLow risk of bias
 Palm et al. (24)20163 yr1,527 patients recorded in an NIV database as “hypoventilation due to obesity and/or OSA” and having a BMI >30 kg/m2. Among these, 585 hospitalized patients were started on NIV for acute-on-chronic hypercapnic respiratory failure, and IPD was obtained for 573 patients.National database from Sweden; patients initiated on NIV in any type of healthcare setting (academic, community, etc.)61.8 ± 11.941.7 ± 8.456.5 ± 7.8Type of NIV not describedSevere risk of bias due to missing data and possible selection bias
 Priou et al. (21)20104.1 ± 2.9 yr130 consecutive patients with OHS discharged on NIV. Among these, 38 hospitalized patients were started on NIV for acute-on-chronic hypercapnic respiratory failure, and IPD was obtained for all 38 patients.Single university hospital in France60.7 ± 16.343.8 ± 7.969.5 ± 11.8Bilevel PAPLow risk of bias

Definition of abbreviations: AVAPS = average volume-assured pressure support; Bilevel PAP = bilevel positive airway pressure; BMI = body mass index; CPAP = continuous positive airway pressure; ICU = intensive care unit; IPD = individual patient data; NIV = noninvasive ventilation; OHS = obesity hypoventilation syndrome; OSA = obstructive sleep apnea; PAP = positive airway pressure; ST = spontaneous-timed.

IPD-MA of the two nonrandomized comparative studies included 316 patients (197 discharged with PAP and 119 discharged without PAP) (18, 25). When patients discharged with PAP were compared with those discharged without PAP, there were nonstatistically significant mortality reductions at 3 months (RR 0.39, 95% CI 0.05–2.74, RD −11.7%), 6 months (RR 0.56, 95% CI 0.16–1.93, RD −10.5%), 9 months (RR 0.67, 95% CI 0.37–1.21, RD −5.7%), or 1 year (RR 0.75, 95% CI 0.48–1.18, RD −3.0%). These effects would be clinically important if real, but there were too few events to either confirm or exclude the effect. Certainty in the estimates was very low because they derived from nonrandomized studies with imprecision.

IPD-MA of all nine studies (i.e., nonrandomized comparative and nonrandomized, noncomparative) included a total of 1,162 patients (1,043 discharged with PAP and 119 discharged without PAP) (6, 1825). Patients discharged with PAP were younger than patients discharged without PAP (Table 2). To mitigate confounding, patients were divided into three groups (tertiles) on the basis of propensity scores. Each group was similar with respect to age, BMI, and severity of OHS. Discharge with PAP reduced mortality at 3 months (RR 0.14, 95% CI 0.05–0.35, RD −14.5%), 6 months (RR 0.24, 95% CI 0.11–0.52, RD −17.8%), 9 months (RR 0.37, 95% CI 0.21–0.64, RD −18.6%), and 1 year (RR 0.34, 95% CI 0.14–0.81, RD −21.8%) (Figure 1). The meta-analyses were limited by moderate heterogeneity, which sensitivity analyses confirmed was attributable to a larger effect within group or tertile 1 than the other groups. The only significant difference between the groups was a lower baseline PaCO2 (64.9 ± 15.4 mm Hg [group 1] vs. 67.0 ± 67.0 mm Hg [other groups combined], P = 0.05). Certainty in the estimates was very low because they derived from nonrandomized studies with a risk of bias due to nonconsecutive enrollment (Table 3).

Table 2. Patient characteristics for the entire cohort and based on tertiles or groups calculated using propensity scores

 PAPNo PAPP Value
All patients, N1,043119 
 Age, yr65.9 ± 13.968.8 ± 13.70.03
 BMI, kg/m243.0 ± 7.242.8 ± 7.10.77
 Baseline PaCO2, mm Hg66.9 ± 16.769.6 ± 17.10.10
Group 13757 
 Age, yr65.1 ± 13.866.7 ± 13.80.76
 BMI, kg/m243.2 ± 7.543.4 ± 8.00.94
 Baseline PaCO2, mm Hg64.9 ± 15.467.3 ± 17.00.68
Group 237124 
 Age, yr65.6 ± 14.068.0 ± 13.70.42
 BMI, kg/m243.1 ± 7.443.0 ± 7.40.95
 Baseline PaCO2, mm Hg65.5 ± 15.667.4 ± 16.00.56
Group 329788 
 Age, yr66.0 ± 13.969.1 ± 13.60.07
 BMI, kg/m243.0 ± 7.242.7 ± 7.10.73
 Baseline PaCO2, mm Hg67.0 ± 16.769.9 ± 17.20.17

Definition of abbreviations: BMI = body mass index; PAP = positive airway pressure.

Data presented as mean (±standard deviation).

Table 3. Evidence profile: Should hospitalized adults suspected of having OHS, in whom the diagnosis has not yet been confirmed, be discharged from the hospital with or without PAP treatment while awaiting confirmation of the diagnosis?

Assessment of Certainty in EffectsEffectsCertaintyImportance
Number of StudiesStudy DesignRisk of BiasInconsistencyIndirectnessImprecisionOther ConsiderationsDischarge with PAPDischarge without PAPRisk Ratio (95% CI)Risk DifferenceNumber Needed to Treat (NNT)
3-mo Mortality, All Patients
9 (6, 18–25)IPD-MAserious*seriousnot seriousnot seriouslarge magnitude of effect§24/1043 (2.3%)20/119 (16.8%)0.14 (0.05–0.35)−14.5%6.9⨁◯◯◯ VERY LOWCRITICAL
6-mo Mortality, All Patients
9 (6, 18–25)IPD-MAserious*seriousnot seriousnot seriouslarge magnitude of effect§49/1001 (4.9%)27/119 (22.7%)0.24 (0.11–0.52)−17.8%5.6⨁◯◯◯ VERY LOWIMPORTANT
9-mo Mortality, All Patients
9 (6, 18–25)IPD-MAserious*not seriousnot seriousnot seriouslarge magnitude of effect§75/1001 (7.5%)31/119 (26.1%)0.37 (0.21–0.64)−18.6%5.4⨁◯◯◯ VERY LOWIMPORTANT
1-yr Mortality, All Patients
9 (6, 18–25)IPD-MAserious*seriousnot seriousnot seriouslarge magnitude of effect§101/1001 (10.1%)38/119 (31.9%)0.34 (0.14–0.81)−21.8%4.6⨁◯◯◯ VERY LOWIMPORTANT

Definition of abbreviations: CI = confidence interval; I2 = statistical heterogeneity; IPD-MA = individual patient data meta-analysis; OHS = obesity hypoventilation syndrome; PAP = positive airway pressure.

*One study had missing data when obtained data were compared to reported data (Palm [24]).

Five studies did not report enrolling consecutive patients (Borel [20], Budweiser [19], Castro-Anon [6], Howard [22], and Palm [24]).

I2 statistic >50%.

§Large magnitude of effect defined as relative risk >2.0 or <0.5.

Subgroup analyses evaluated the effect of being discharged with PAP on 3-month mortality in patients with median baseline PaCO2 of 61 mm Hg or higher (compared with <61 mm Hg), median BMI of 41.4 kg/m2 or higher (compared with <41.4 kg/m2), and median age of 67.3 years or greater (compared with <67.3 yr). The mortality benefit was seen entirely in patients with a BMI of 41.4 kg/m2 or higher (RR 0.09, 95% CI 0.04–0.17, RD −23.8%), with no effect in patients with a BMI of <41.4 kg/m2 (RR 1.09, 95% CI 0.14–8.28, RD +2%). Being above or below the median of baseline PaCO2 and age did not alter the benefits of being discharged with PAP (Figure 2).

This study is the only systematic review comparing the effect on mortality of discharge with PAP to discharge without PAP among patients with OHS or suspected of having OHS hospitalized for acute-on-chronic hypercapnic respiratory failure. An IPD-MA using data from >1,100 patients found that discharge with PAP was associated with markedly lower mortality at 3 months, 6 months, 9 months, and 1 year. Most patients (∼92%) were discharged on NIV. The beneficial effect is noteworthy for its large magnitude, which is particularly prominent among patients whose BMI is 41.4 kg/m2 or greater. On the basis of this evidence, the guideline panel recommended that patients with OHS or suspected of having OHS hospitalized for acute-on-chronic hypercapnic respiratory failure be discharged from the hospital on empiric NIV to be used during sleep at home.

It is important to acknowledge that NIV was the predominant form of PAP therapy prescribed upon hospital discharge without patients undergoing sleep studies or PAP titration studies. As such, it remains unknown at the time of hospital discharge whether OHS will be as responsive to CPAP (i.e., OHS phenotype without severe OSA). Therefore, clinicians should consider empiric NIV as the PAP modality of choice while awaiting outpatient evaluation with sleep studies and PAP titration studies in the sleep laboratory, preferably during the first 3 months after hospital discharge (9). On the basis of the present data and given the lack of data on NIV settings at the time of discharge, we cannot recommend settings for NIV. However, clinicians may consider using the same NIV settings used during hospitalization or consider autoadjustable NIV. Regardless of what PAP settings are chosen, empiric NIV settings should not be a substitute for an outpatient sleep study to appropriately select the PAP mode and titrate it accurately.

This systematic review has several strengths. First, it was performed within the context of clinical practice guideline development. International experts selected outcomes important to clinical decision-making, helped identify important studies, and obtained individual patient data from the authors of the primary studies. A patient panelist ensured that the outcomes were important to patients. Second, the search criteria applied to multiple databases were extremely broad; thus, it is unlikely that important studies were missed. Finally, using individual patient data, as opposed to aggregate data, improved the precision of estimated effects and enabled a few subgroup analyses.

The primary limitation of the systematic review is the quality of the studies from which the individual patient data were obtained. There were no randomized trials and only two nonrandomized comparative studies; thus, there were relatively few patients who were discharged without PAP compared with those discharged with PAP. Only some studies enrolled patients consecutively, and, therefore, selection bias cannot be eliminated. The studies had variable inclusion and exclusion criteria, did not specify the decision-making process to discharge with or without PAP, and did not report other relevant outcomes. Moreover, there were no data on PAP settings upon discharge or data on PAP adherence after hospital discharge. Another important limitation is that we could not ascertain whether PAP therapy was initiated at some point within the year after hospital discharge. However, we speculate the likelihood of PAP therapy being initiated within the first 3 months after hospital discharge is low. We also could not ascertain the cause of mortality. Given the observational nature of the studies and the lack of information on various socioeconomic factors and comorbidities, it remains unclear whether the measured outcome of mortality is only related to being discharged without PAP therapy or whether other factors may have contributed. Last, there are many other important contributors to mortality that cannot be accounted for in our study such as adherence to PAP therapy after hospital discharge, smoking, and weight loss. On the basis of these limitations, the reported data are at serious risk of bias, and the level of certainty regarding the reported outcome is very low.

Notwithstanding these limitations, the importance and the large magnitude of the outcome (decreased mortality), coupled with the low risk and burden of the intervention, support the guideline panel’s recommendation that patients with OHS or suspected of having OHS who are hospitalized for acute-on-chronic hypercapnic respiratory failure be discharged with empiric NIV. We believe that most patients discharged from the hospital without PAP would likely require PAP after an outpatient sleep study. However, further studies will be necessary to confirm the cost-effectiveness of discharging patients on empiric NIV while awaiting outpatient testing. Moreover, there are important regional variations in the cost and availability of PAP which may impact willingness of healthcare providers to prescribe PAP at discharge and third party payers to cover its cost.

In summary, well-designed clinical trials are necessary to address the important limitations of our meta-analysis and to more accurately assess whether hospitalized patients with acute-on-chronic hypercapnic respiratory failure due to OHS should be discharged on PAP (NIV or CPAP) or not. These studies should focus not only on mortality, but also on other patient-centered outcomes such as healthcare resource utilization, cost effectiveness, and symptom resolution. Furthermore, studies are needed to ascertain the optimal timing of outpatient sleep study after discharge. Whether autotitrating NIV will allow clinicians to forego in-laboratory PAP titration requires further investigation.

1 . Mokhlesi B. Obesity hypoventilation syndrome: a state-of-the-art review. Respir Care 2010;55:13471362, discussion 1363–1365.
2 . Balachandran JS, Masa JF, Mokhlesi B. Obesity hypoventilation syndrome: epidemiology and diagnosis. Sleep Med Clin 2014;9:341347.
3 . Flegal KM, Kruszon-Moran D, Carroll MD, Fryar CD, Ogden CL. Trends in obesity among adults in the United States, 2005 to 2014. JAMA 2016;315:22842291.
4 . Hales CM, Fryar CD, Carroll MD, Freedman DS, Ogden CL. Trends in obesity and severe obesity prevalence in us youth and adults by sex and age, 2007–2008 to 2015–2016. JAMA 2018;319:17231725.
5 . Nowbar S, Burkart KM, Gonzales R, Fedorowicz A, Gozansky WS, Gaudio JC, et al. Obesity-associated hypoventilation in hospitalized patients: prevalence, effects, and outcome. Am J Med 2004;116:17.
6 . Castro-Añón O, Pérez de Llano LA, De la Fuente Sánchez S, Golpe R, Méndez Marote L, Castro-Castro J, et al. Obesity-hypoventilation syndrome: increased risk of death over sleep apnea syndrome. PLoS One 2015;10:e0117808.112.
7 . Marik PE, Chen C. The clinical characteristics and hospital and post-hospital survival of patients with the obesity hypoventilation syndrome: analysis of a large cohort. Obes Sci Pract 2016;2:4047.
8 . Masa JF, Pépin JL, Borel JC, Mokhlesi B, Murphy PB, Sánchez-Quiroga MA. Obesity hypoventilation syndrome. Eur Respir Rev 2019;28:180097.114.
9 . Mokhlesi B, Masa JF, Brozek JL, Gurubhagavatula I, Murphy PB, Piper AJ, et al. Evaluation and management of obesity hypoventilation syndrome. An official American Thoracic Society clinical practice guideline. Am J Respir Crit Care Med 2019;200:e6e24.
10 . Soghier I, Brożek JL, Afshar M, Tamae Kakazu M, Wilson KC, Masa JF, et al. Noninvasive ventilation versus CPAP as initial treatment of obesity hypoventilation syndrome. Ann Am Thorac Soc 2019;16:12951303.
11 . Higgins JPT, Green S. Cochrane handbook for systematic reviews of interventions version 5.1.0. The Cochrane Collaboration; 2011 [updated 2011 Mar; accessed 2020 Jan 19]. Available from www.handbook.cochrane.org.
12 . Guyatt GH, Oxman AD, Vist G, Kunz R, Brozek J, Alonso-Coello P, et al. GRADE guidelines: 4. Rating the quality of evidence—study limitations (risk of bias). J Clin Epidemiol 2011;64:407415.
13 . Guyatt GH, Oxman AD, Montori V, Vist G, Kunz R, Brozek J, et al. GRADE guidelines: 5. Rating the quality of evidence—publication bias. J Clin Epidemiol 2011;64:12771282.
14 . Guyatt GH, Oxman AD, Kunz R, Brozek J, Alonso-Coello P, Rind D, et al. GRADE guidelines: 6. Rating the quality of evidence—imprecision. J Clin Epidemiol 2011;64:12831293.
15 . Guyatt GH, Oxman AD, Kunz R, Woodcock J, Brozek J, Helfand M, et al.; GRADE Working Group. GRADE guidelines: 7. Rating the quality of evidence—inconsistency. J Clin Epidemiol 2011;64:12941302.
16 . Guyatt GH, Oxman AD, Kunz R, Woodcock J, Brozek J, Helfand M, et al.; GRADE Working Group. GRADE guidelines: 8. Rating the quality of evidence—indirectness. J Clin Epidemiol 2011;64:13031310.
17 . Guyatt GH, Oxman AD, Sultan S, Glasziou P, Akl EA, Alonso-Coello P, et al.; GRADE Working Group. GRADE guidelines: 9. Rating up the quality of evidence. J Clin Epidemiol 2011;64:13111316.
18 . Carrillo A, Ferrer M, Gonzalez-Diaz G, Lopez-Martinez A, Llamas N, Alcazar M, et al. Noninvasive ventilation in acute hypercapnic respiratory failure caused by obesity hypoventilation syndrome and chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2012;186:12791285.
19 . Budweiser S, Riedl SG, Jörres RA, Heinemann F, Pfeifer M. Mortality and prognostic factors in patients with obesity-hypoventilation syndrome undergoing noninvasive ventilation. J Intern Med 2007;261:375383.
20 . Borel JC, Burel B, Tamisier R, Dias-Domingos S, Baguet JP, Levy P, et al. Comorbidities and mortality in hypercapnic obese under domiciliary noninvasive ventilation. PLoS One 2013;8:e52006.18.
21 . Priou P, Hamel JF, Person C, Meslier N, Racineux JL, Urban T, et al. Long-term outcome of noninvasive positive pressure ventilation for obesity hypoventilation syndrome. Chest 2010;138:8490.
22 . Howard ME, Piper AJ, Stevens B, Holland AE, Yee BJ, Dabscheck E, et al. A randomised controlled trial of CPAP versus non-invasive ventilation for initial treatment of obesity hypoventilation syndrome. Thorax 2017;72:437444.
23 . Murphy PB, Davidson C, Hind MD, Simonds A, Williams AJ, Hopkinson NS, et al. Volume targeted versus pressure support non-invasive ventilation in patients with super obesity and chronic respiratory failure: a randomised controlled trial. Thorax 2012;67:727734.
24 . Palm A, Midgren B, Janson C, Lindberg E. Gender differences in patients starting long-term home mechanical ventilation due to obesity hypoventilation syndrome. Respir Med 2016;110:7378.
25 . Romero C, Sánchez J, Almadana V, Gómez-Bastero A, Guerrero P, Valido A, et al. Results of noninvasive ventilation in obese patients with acute respiratory failure [abstract]. Chest 2014;145:544A.
Correspondence and requests for reprints should be addressed to Babak Mokhlesi, M.D., M.Sc., Section of Pulmonary and Critical Care, Sleep Disorders Center, University of Chicago, 5841 S. Maryland Avenue, MC6076/Room M630, Chicago, IL 60637. E-mail: .

Supported by the American Thoracic Society.

Author Contributions: B.M. and J.F.M.: Conception and design of the project, acquisition of data, analysis of data, and writing the manuscript. M.A., I.S., and M.T.K.: Preparation and validation of search strategy, search of the bibliographic databases, title abstract screening, full text screening, and data extraction. V.A.P., D.J.B., J.-C.B., S.B., A.C., O.C.-A., M.F., F.G., R.G., N.H., M.E.H., P.B.M., A.P., L.A.P.d.L., A.J.P., J.L.P., P.P., and J.F.S.-G.: Provided individual patient data, reviewed and edited the manuscript. K.C.W.: Analysis of data and writing the manuscript.

This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org.

Author disclosures are available with the text of this article at www.atsjournals.org.

Related

No related items
Comments Post a Comment




New User Registration

Not Yet Registered?
Benefits of Registration Include:
 •  A Unique User Profile that will allow you to manage your current subscriptions (including online access)
 •  The ability to create favorites lists down to the article level
 •  The ability to customize email alerts to receive specific notifications about the topics you care most about and special offers
Annals of the American Thoracic Society
17
5

Click to see any corrections or updates and to confirm this is the authentic version of record