Relationship Between CPAP Termination and All-Cause Mortality

• DOI: 10.1016/j.chest.2022.02.013
• Publication Date: 2022-02-15T02:57:10Z
• Authors: Jean-Louis Pépin; Sébastien Bailly; Pierre Rinder; Dan Adler; Adam V. Benjafield; Florent Lavergne; Anne Josseran; Paul Sinel-Boucher; Renaud Tamisier; Peter A. Cistulli; Atul Malhotra; Pierre Hornus
• Keywords: OSA; 03 medical and health sciences; 0302 clinical medicine; CPAP; [SDV]Life Sciences [q-bio]; 616; 610; adherence; mortality

BackgroundRandomized controlled trials have failed to demonstrate an effect of CPAP therapy on mortality. However, these studies a number important limitations, including low adherence, patient selection, and small mortality events.Research QuestionWhat are the effects termination in first year all-cause patients with OSA from Nationwide Claims Data Lake for Sleep Apnea study?Study Design MethodsData Système National des Données de Santé (SNDS) database, French national health insurance reimbursement system, all new users ≥ 18 years age were analyzed. The SNDS contains comprehensive, individualized, anonymized data spending reimbursements > 99% individuals living France. diagnosis was based specific disease codes, whereas prescription identified using treatment method codes. defined as cessation triggered by respiratory physician or sleep specialist charge follow-up. Patients who terminated propensity score matched those continued use CPAP. primary outcome Three-year survival visualized Kaplan-Meier curves. Contributors also determined.ResultsData two groups each 88,007 included (mean age, 60 years; 64% men). Continuation associated significantly lower risk death compared (hazard ratio [HR], 0.61; 95% CI, 0.57-0.65; P < .01, log-rank test). Incident heart failure less common vs (HR, 0.77; 0.71-0.82; .01).InterpretationThese real-world unbiased database highlight potential ongoing reduce OSA. Randomized events. What study? determined. .01). These FOR EDITORIAL COMMENT, SEE PAGE 1444Take-home PointsStudy Question: study?Results: In groups, continuation termination. addition, incidence year.Interpretation: is characterized repeated upper airway collapse during sleep. episodes several consequences, sympathetic activation, marked negative intrathoracic pressure swings, intermittent oxygen desaturation, hypercapnia, arousal turn, thought contribute comorbidities OSA, hypertension, cardiovascular cerebrovascular disease, metabolic abnormalities.1Bonsignore M.R. Suarez Giron M.C. Marrone O. Castrogiovanni A. Montserrat J.M. Personalised medicine disorders: focus obstructive apnoea treatment.Eur Respir Rev. 2017; 26: 170069Crossref PubMed Scopus (44) Google Scholar, 2Knauert M. Naik S. Gillespie M.B. Kryger Clinical consequences economic costs untreated apnea syndrome.World J Otorhinolaryngol Head Neck Surg. 2015; 1: 17-27Crossref 3Lévy P. Kohler McNicholas W.T. et al.Obstructive syndrome.Nat Rev Dis Primers. 15015Crossref (478) 4Pepin J.L. Borel A.L. Tamisier R. Baguet J.P. Levy Dauvilliers Y. Hypertension sleep: overview tight relationship.Sleep Med 2014; 18: 509-519Crossref (144) Scholar could be responsible increased that has been reported OSA.5Lavie Lavie L. Herer All-cause males syndrome: declining rates age.Eur J. 2005; 25: 514-520Crossref (243) 6Yaggi H.K. Concato Kernan W.N. Lichtman J.H. Brass L.M. Mohsenin V. Obstructive factor stroke death.N Engl Med. 353: 2034-2041Crossref (2354) 7Marin Carrizo S.J. Vicente E. Agusti A.G. Long-term outcomes men apnoea-hypopnoea without continuous positive pressure: observational study.Lancet. 365: 1046-1053Abstract Full Text PDF (2688) 8Young T. Finn Peppard P.E. al.Sleep disordered breathing mortality: eighteen-year follow-up Wisconsin cohort.Sleep. 2008; 31: 1071-1078PubMed 9Marshall N.S. Wong K.K. Cullen S.R. Knuiman M.W. Grunstein R.R. 20-year mortality, stroke, cancer Busselton Health Study cohort.J Clin 10: 355-362Crossref (313) 10Punjabi N.M. Caffo B.S. Goodwin al.Sleep-disordered prospective cohort study.PLoS 2009; 6e1000132Crossref (977) Despite association between randomized clinical evaluating treating events not demonstrated any beneficial therapy, gold standard moderate severe OSA.11McEvoy R.D. Antic N.A. Heeley al.CPAP prevention apnea.N 2016; 375: 919-931Crossref (1150) 12Peker Glantz H. Eulenburg C. Wegscheider K. Herlitz Thunström Effect coronary artery nonsleepy apnea. RICCADSA trial.Am Crit Care 194: 613-620Crossref (358) 13Sánchez-de-la-Torre Sánchez-de-la-Torre Bertran al.Effect its prevalence acute syndrome (ISAACC study): randomised trial.Lancet 2020; 8: 359-367Abstract (149) ability influence hard end points may limited factors, adherence selection. total trials, limiting statistical power detect between-group differences perhaps representative what happens real world. Thus, although provide high level evidence, able more accurate generalizable picture routine study uses system. This analysis investigated long-term therapy. which individualized project approved Commission Nationale Informatique Liberté, information technology personal protection authority. Specific approval obtained Liberté perform this (Identifier: DR-2019-78, no. 919194). Eligible adults had previously used initiated January 2015 December 2016. codes.14Pépin Bailly Rinder al.on behalf medXcloud GroupCPAP phenotype: nationwide analysis.J 2021; 936Crossref (20) 1 matching eliminate biases confounding factors affecting both groups. Propensity following factors: demographics (age sex), coverage, socioeconomic status, (stroke, failure, peripheral arterial occlusive diabetes mellitus, other diseases, COPD, bariatric surgery, neurotic disorders, psychotropic medication, kidney diseases). To account possible selection bias, sensitivity performed untruncated initiation being starting date Cox model time-dependent covariate. Variables adjustments same variables analysis. One after initiation, applied select population nonusers; then followed up additional 3 (e-Fig 1). recommendations device 4 h/night. Reimbursement progressively decrease when very occurs, delivery can continue 2 h/night, requirement education coaching. A mandatory visit occurs at months every thereafter determine renewal. For analysis, it assumed linked nonadherence. Individuals valid documented reason stopping (ie, cure otorhinolaryngology switch oral appliances, death) censored registered death, but cause available. expressed median (interquartile range) quantitative (percentage) qualitative data. Comparisons (CPAP continuation) Student t test χ Mortality cumulative rate curves, comparisons test. analyses separately women. objective assessed First, compute probability year. nonparsimonious multivariate regression created major (list results e-Table 1:1 greedy caliper 0.1%. Standardized ensure quality matching. standardized difference reduced 2). Finally, semiparametric assess impact (mortality, incident (CAD), hospitalization diabetes, arrhythmias, hypertension); evaluated because comparatively short time competing event outcomes, considering only alive. Hazard proportionality assumption checked, hazard (HR) values must interpreted average HR, rather than instantaneous HR.15Stensrud M.J. Hernán M.A. Why proportional hazards?.JAMA. 323: 1401-1402Crossref (93) Analyses Python version 3.6.7 software libraries Numpy 1.18.1 Pandas 0.24.2 management Statsmodel 0.12.1 logistic regression, Lifelines 0.14.1 curves models. value .05 considered statistically significant. includes 480,000 patients, whom 365,301 undergone least did (4,882 year) (e-Table Of 76% (n = 277,242) 24% 88,059) After matching, group (total 176,014 patients) (Fig As expected, generated well baseline characteristics (Table 1).Table 1Baseline Characteristics Matched PopulationVariableCPAP 88,007)CPAP Termination 88,007)Age, y60.0 (70.0-50.0)59.0 (69.0-49.0)Female sex32,227 (36.6)31,666 (36.0)Comorbidity Chronic psychiatric disorders4,621 (5.2)4,606 (5.2) Stroke2,735 (3.1)2,684 (3.1) Heart failure2,306 (2.6)2,046 (2.3) Coronary disease8,023 (9.1)8,037 (9.1) Hypertension42,568 (48.4)43,231 (49.1) Diabetes mellitus18,610 (21.1)18,304 (20.8) COPD7,156 (8.1)7,387 (8.4)Data presented No. (%) range). Open table tab Over 3-year observation period, occurred 3,204 (3.6%) 2,053 (2.3%) 3). test) similar women 0.63 [95% 0.57-0.68] 0.54 0.47-0.62]; .01 both) showed significant reduction continuation, HR 0.73 (95% 0.70-0.76; [10,795 336,415 3.2%]). (based codes entered database) 3, Fig During follow-up, hypertension frequently 4). trend toward hospitalizations found (P .06) Sensitivity excluded period yielded 4).Figure 4Forest plot showing potentially contributing death. indicates continuation. ratio.View Large Image Figure ViewerDownload Hi-res image Download (PPT) dataset mechanism underlying seen our main survived long-enough discontinue included. such, either died, discontinued avoid bias resulting arbitrarily splitting into baseline, we covariate overall survival. confirmed strengthened findings there 27% Our contrast Cardiovascular Endpoints (SAVE) study, Impact Syndrome Evolution Acute Syndrome—Effect Intervention With Continuous Positive Airway Pressure CAD composite point nonfatal events.11McEvoy All no usual care respect alone secondary point.11McEvoy (3.3 ± 2.3 h/night SAVE 2.78 2.73 study),11McEvoy Scholar,13Sánchez-de-la-Torre levels do seem reflect broader settings.16Cistulli P.A. Armitstead Pepin al.Short-term apnea: big world data.Sleep 2019; 59: 114-116Crossref (80) Scholar,17Pack A.I. Magalang U.J. Singh B. Kuna S.T. Keenan B.T. Maislin G. RCT RCT? Depends question. response McEvoy al.Sleep. 44: zsab042Crossref (3) Device needed benefits realized.12Peker Scholar,18Barbé F. Durán-Cantolla trial.JAMA. 2012; 307: 2161-2168Crossref (583) example, preplanned subgroups events, (adjusted 0.29; 0.10-0.86; .026).12Peker Furthermore, adherent cerebral group.11McEvoy Second, highly selected populations, namely existing disease. particular, noninclusion excessive daytime sleepiness ethical reasons might exclude likely adhere respond therapy.19Mazzotti D.R. Lim D.C. Gottlieb D.J. Kim Pack Symptom subtypes predict outcomes.Am 200: 493-506Crossref (181) Recently, comparison made consecutive clinic 3,965) participants prominent recent examining adverse OSA.20Reynor McArdle N. Shenoy al.Continuous patients?.Sleep. 2022; 45: zsab264Crossref (7) Less 20% eligibility criteria younger, sleepier, women.20Reynor (25 20 12 14 study), point. contrast, current France indication making applicable general large deaths provides adequate analyses. differentiating clearer use. highest better effectiveness populations encountered practice. Others recently suggested design features enrolled limit identify treatment.21Labarca Dreyse Drake Jorquera Barbe Efficacy (CPAP) systematic review meta-analysis.Sleep 52: 101312Crossref (48) Contrary classical exposed unexposed discontinuation homogeneous initial further limits unmeasured bias. Specifically, scores overcome limitations around inclusion experience trials.22Pack understanding overcoming bias.Sleep. zsaa229Crossref (35) US Food Drug Administration indicated methods appropriate support medical devices such CPAP.23Food Center Devices Radiological HealthDesign considerations pivotal investigations devices: guidance industry, investigators, institutional boards staff. 2013.https://www.fda.gov/regulatory-information/search-fda-guidance-documents/design-considerations-pivotal-clinical-investigations-medical-devicesDate accessed: March 23, 2021Google Two studi