SLEEP DISORDERS IN CYSTIC FIBROSIS: A SYSTEMATIC REVIEW AND META-ANALYSIS

SUMMARY

Cystic fibrosis (CF) is a genetic disorder that leads to airway mucus accumulation, chronic inflammation, and recurrent respiratory infections – all likely impacting sleep. However, controlled studies of sleep in CF patients are limited, and have shown mixed results. We reviewed all publications on CF and sleep indexed in PubMed, CINAHL, and Scopus through April 2019. In the meta-analysis, we calculated pooled weighted mean differences for sleep quality, sleepiness, oximetry, and polysomnographic (PSG) parameters, using fixed or random-effects models as appropriate. A total of 87 manuscripts were reviewed. Compared to controls, children with CF had lower nighttime oxygen saturation nadirs, decreased sleep efficiency and a higher respiratory event index, with no differences in the percentage of REM sleep. Adults with CF had lower oxygen saturation nadirs, with a trend towards reduced sleep efficiency and no differences in REM sleep. In addition, patients with CF cough more during sleep and experience painful events that interfere with sleep. Actigraphy and questionnaires suggest disturbed sleep and daytime sleepiness. Noninvasive ventilation appears to improve gas exchange and symptoms.

We conclude that when sleep is evaluated objectively or subjectively in patients with CF, perturbations are common, emphasizing the importance of their identification and treatment and inclusion as part of routine care. Additional research, with larger sample sizes and standardized outcomes, are necessary.

INTRODUCTION

Cystic fibrosis (CF) is an autosomal recessive disorder caused by mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) gene [1]. While it is a multi-systemic disease, the most significant manifestations involve the respiratory and gastrointestinal tracts [2], where there is accumulation of thick mucus, impaired mucociliary clearance, chronic inflammation, recurrent upper and lower respiratory infections, as well as pancreatic insufficiency that may lead to malabsorption and thus a deficient nutritional status [3].

Sleep disruption, insufficient sleep, and sleep-disordered breathing (SDB) in children, adolescents and young adults impose deleterious psychological and neurocognitive effects, including anxiety, depression, impaired quality-of-life, daytime sleepiness, attention deficits, poor academic performance, and increased risk for cognitive dysfunction and motor vehicle accidents [4–15]. They have also been associated with immune dysfunction, obesity [16], metabolic dysregulation [17], inflammation [18–23], cardiovascular disease [16, 24], and all-cause mortality [25].

Sleep may be adversely impacted in CF by multiple factors, including chronic cough, chronic airway inflammation and infections, abdominal pain, gastroesophageal reflux, frequent stooling, and medications side effects [26, 27]. However, controlled studies focusing on sleep in CF are scarce and the results conflicting. Here, our primary objective was a systematic review of the literature to assess the current knowledge on sleep disorders in patients with CF, focusing on the differences with healthy controls when available, and also on the correlation between CF disease severity and sleep disruption. Our secondary objective was to perform, where feasible, a meta-analysis of studies assessing the differences in sleep between children and adults with CF and healthy individuals. We aimed to compare both objective findings such as sleep study parameters and subjective differences such as those seen on validated questionnaires.

METHODS

Study selection

All studies indexed in PubMed, Scopus, and CINAHL through April 2019 were reviewed for “Sleep” AND “Cystic Fibrosis”, and using Medical Subject Heading (MeSH) terms. In addition, we reviewed the citations of the identified studies for additional studies. Two authors (JR and AGH) independently screened references according to the selection criteria, including only human studies, in English or containing sufficient data in English to abstract results, starting January 1980. References were excluded if they were case reports, reviews, or studies of mixed patient populations (i.e. cystic fibrosis with other chronic diseases); a third author (EF) resolved any discrepancies. For the meta-analysis, we included all studies comparing sleep in patients with CF and normal controls wherein data could be separated by age groups. To assess the risk of bias, study methods were analyzed for subject and control recruitment and blinding. For the correlation analysis we included all studies with data on the association between lung disease and sleep characteristics. The protocol for this systematic review was prospectively registered in the international prospective register of systematic reviews, PROSPERO (ID CRD42019137977), and the review was conducted following PRISMA statement guidelines [28].

Data extraction

Using a standardized form, JR and AGH independently excerpted data on reference information (first author, publication year, journal, title); sample characteristics; polysomnography (PSG) data [total sleep time (TST), sleep latency, rapid eye movement (REM) latency, sleep efficiency, sleep stage percentage, arousal index, oxygen saturation means and nadir, respiratory rate, heart rate, apnea-hypopnea index (AHI), respiratory-disturbance index (RDI)]; forced expiratory volume in one second (FEV₁); and questionnaire total scores [Pittsburgh sleep quality index (PSQI) and Epworth sleepiness scale (ESS)].

Data analysis

Where feasible, we performed a meta-analysis to calculate pooled summary estimates. We calculated weighed or standardized mean differences (WMD or SMD, respectively): SMD were used when studies utilized different units or scales for the outcome. Each study was weighted by its inverse effect size variance [29]. Heterogeneity was quantified using I² [30]. Fixed-effects models were used when heterogeneity between studies was nonsignificant, and random-effects models for analyses with significant heterogeneity [31]. All analyses were performed in STATA v14 (STATA Corporation, College Station, TX), and a P-value <0.05 was considered statistically significant.

RESULTS

After removal of duplicates, 916 non-overlapping manuscripts were identified (Figure 1) and, after screening of titles and abstracts, 87 manuscripts were included in the systematic review. There were 32 PSG studies, of which 12 were in pediatric patients, 12 in adults, and eight in a mix of both children and adults. Twenty studies evaluated the association between CF disease severity and objective sleep disorders as measured by PSG or oximetry. Eleven studies focused on noninvasive ventilation (NIV), and 39 studies utilized sleep questionnaires or actigraphy. Eleven studies explored how cough interfered with sleep, and three studies examined how pain interacted with sleep in CF. In the next sections we describe the results of these studies providing a bullet summary at the end of each section.

Figure 1 – Flow diagram of study selection.

Adapted from PRISMA statement guidelines[28]. ESS: Epworth sleepiness scale; PSG: Polysomnogram; PSQI: Pittsburgh sleep quality index.

Polysomnography studies – Sleep architecture and sleep apnea

Pediatric studies:

Of twelve pediatric PSG studies (Table 1A), five compared children with CF to controls [32–36]. The meta-analysis of controlled studies, of which four were prospective [32–35] and one noted blinding of the scorers [33], showed lower nighttime oxygen saturation (S_aO₂) nadirs in children with CF [CF=78 subjects, controls=69 subjects; WMD −3.1% (95%CI −5.1%, −1.0%), p=0.004; I²=71.1%] (Figure 2A). Children with CF also had reduced sleep efficiency [CF=99, controls=67; WMD −7.9% (95%CI −12.1%, −3.7%), p<0.001; I²=0%] (Figure 2B) and a higher respiratory event index (AHI/RDI) [CF=93, controls=62; SMD 0.44 (95%CI 0.11–0.78), p=0.009; I²=40.3%] (Figure 2C). There were no significant differences in REM as a percentage of TST [p=0.47] (Figure 2D). Three studies reported specifically on slow-wave sleep (SWS): two studies reported stage 3 and 4 of TST [33, 34] and two stage 3 of TST [35, 36], none with significant differences. Assessment of bias for all pooled analyses are also shown in the Online Supplement.

Table 1A –

Polysomnography studies in children with CF

Reference	Controlled study	Sample size		Age (years)		Summary of main findings
		Cases	Controls	Cases	Controls
Tepper 1983 [105]	No	6	n/a	10–16	n/a	Hypoxia, decreased minute ventilation, tidal volume during sleep compared with awake state; most significant during REM sleep.
Avital 1991 [106]	No	12	n/a	7–17	n/a	Theophylline showed lower heart rate, better SaO2, but disrupted sleep, lower SE, higher nocturnal wake time. No effect on AHI or periodic leg movements.
Villa 2001 [32]	Yes	19	20	13.1mos (3–36)	Matched	Lower mean SaO2 (95.6% vs 96.9%), SaO2 nadir (85.9% vs 89.1%); no differences in %REM. Differences more significant in children with airway inflammation.
Naqvi 2008 [33]	Yes	24	14	14.2 (3.8)*	10.7 (4.4)	Lower SaO2 nadir (90.3% vs 95.6%), SE (75.2% vs. 86.2%), %REM (12.7% vs 18.3%).
Ramos 2009 [107]	No	63	n/a	2–14	n/a	Upper airway bone and soft tissue structural changes, chronic rhinosinusitis associated with OSA in patients with CF.
Suratwala 2011 [35]	Yes	25	25	8–20	7–20	Lower mean SaO2 (96.6% vs 97.5%), SaO2 nadir (92.5% vs 93.8%); no differences in SE or %REM. Lower SaO2 correlated with worse glucose regulation.
Spicuzza 2012 [34]	Yes	40	18	0.5–11	Matched	Lower mean SaO2 (94.7% vs 97%), SE (80.4% vs. 87.8%), %REM (11.7% vs 13.1%); higher AHI (7.3/hour vs 0.5/hour).
Ramos 2013 [38]	No	67	n/a	2–14	n/a	Disease severity correlation with sleep variables.
Paranjape 2015 [36]	Yes	10	10	9.6 (3.6)†	9.6 (3.6)	CF (vs snoring controls) had lower mean SaO2, SaO2 nadir (90% vs 93%); no differences in %REM, SE.
Silva 2016 [37]	No	33	n/a	6–18	n/a	Disease severity correlation with sleep variables.
Waters 2016 [40]	No	46	n/a	8–12	n/a	Disease severity correlation with sleep variables.
Isaiah 2019 [39]	No	35	n/a	11.6 (9.5–13.1)**	n/a	FEV1 <53% predicted was the best predictor for nocturnal hypoxemia.

^*mean (SE).

^†median (SD).

^**mean (95% confidence interval).

n/a: not applicable. AHI: apnea hypopnea index; NREM: non rapid eye movement; OSA: obstructive sleep apnea; REM: rapid eye movement; %REM: REM as percent of total sleep time; S_aO₂: oxygen saturation; SE: sleep efficiency as percent of total sleep time.

Figure 2 – Meta-analysis of polysomnography parameters in patients with CF vs. controls.

WMD: Weighted mean difference.

Since data were available on respiratory event norms in the general population, we also calculated the weighted mean event index for all studies reporting these indices. Eight studies reported either the AHI [32, 34, 37–39] or the RDI [33, 36, 40]. Since the RDI may lead to an overestimation, whenever available the AHI was used. In a total of 274 children, the weighted average was 3.2 respiratory events per hour, which is at the upper limit of the 90^th percentile for healthy children (2.1–3.2 events per hour) [41].

Studies in adults:

Of twelve PSG-based studies in adults (Table 1B), three compared patients with CF to healthy controls prospectively [42–44], and none included blinding of the scorers. Compared to controls, adults with CF had lower nighttime S_aO₂ nadirs [CF=84, controls=43; WMD −8.1% (95%CI - 10.2%, −6.0%), p<0.001; I²=31.3%] (Figure 2A) and sleep efficiency [CF=84, controls=43; WMD −10.7% (95%CI −22.8%, −1.3%), p<0.001; I²=86%] (Figure 2B). There were no differences in REM as a percentage of TST [p=0.26] (Figure 2C) with small none significant differences in stage 3 as percent of TST (data not shown). One study in adults with hypercapnia reported a mean AHI of 10/hour TST, consistent with mild sleep apnea [45]; the remaining studies reported respiratory-event index means (AHI or RDI) of 1.5–4.3 per hour [46–50], considered to be within the normal range in adults [51].

Table 1B –

Polysomnography studies in adults with CF

Reference	Controlled study	Sample size		Age (years)		Summary of main findings
		Cases	Controls	Cases	Controls
Spier 1984 [108]	Yes	10	7	23 (4)*	16	Study of low flow oxygen in CF w/stable severe obstructive lung disease – treatment improved hypoxemia and did not lead to hypercapnia.
Ballard 1996 [109]	No	5	n/a	29.6 (3.6)*	n/a	Tidal volume and respiratory neuromuscular output decreased during NREM, airway resistance and lung volumes unchanged.
Bradley 1999 [42]	Yes	14	8	25.9 (3.5)*	27.6 (3.2)*	Hypoxemia during sleep. Hypoxemia and hypercapnia correlated with gas trapping (RV, RV/TLC).
Milross 2001** [48]	No	32	n/a	27 (8)*	n/a	Evening PaO2 contributed significantly to prediction of sleep-related desaturations, rise in transcutaneous CO2 from NREM to REM.
Dancey 2002 [43]	Yes	19	10	30 (6)*	27 (5)*	Reduced SE, more frequent awakenings. Short sleep latency (6.7+/−3 min), lower levels of activation, happiness, greater fatigue - correlated with sleep loss. Adult patients with severe lung disease have impaired neurocognitive function, daytime sleepiness, related to chronic sleep loss and nocturnal hypoxemia.
Milross 2002 [50]	No	33	n/a	27 (8)*	n/a	Subjective sleep quality associated with CF disease severity. Better SE, %REM in patients with low PSQI.
Milross 2002 [49]	No	31	n/a	27 (8)*	n/a	Single-night PSG yields reliable information on nocturnal oxygenation and respiratory disturbance, despite first-night effects on SE, REM latency, wake after sleep onset.
Young 2008 [45]	No	8	n/a	37 (8)*	n/a	Study of NIV for hypercapnia in CF.
Perin 2012 [44]	Yes	51	25	25.1 (6.7)*	25.5 (7.3)*	Impaired subjective sleep quality, higher AI. No difference in AHI.
Fauroux 2012 [46]	No	11	n/a	29 (4)*	n/a	Poor sleep quality did not predict nocturnal gas exchange. Nocturnal pulse oximetry and transcutaneous CO2 identified REM hypoventilation.
Forte 2015 [83]	No	51	n/a	25.1 (8.8)*	n/a	AHI and AI were independent predictors of CF QOL.
Íscar-Urrutia 2018 [47]	No	18	n/a	32 (18)*	n/a	Low SE due to periods of intra-sleep wakefulness and microarousals; normal NREM and REM sleep latencies, normal ratio of different sleep phases.

^*mean (SD). n/a: not applicable.

^**Three studies with overlapping populations but different outcomes reported.

AHI: apnea hypopnea index; AI: arousal index; COPD: chronic obstructive pulmonary disease; NIV: noninvasive ventilation; NREM: non rapid eye movement; OSA: obstructive sleep apnea; PaO2: partial arterial pressure of oxygen; PSG: polysomnography; PSQI: Pittsburgh sleep quality index; QOL: quality of life; REM: rapid eye movement; %REM: REM as percent of total sleep time; RV: residual volume; S_aO₂: oxygen saturation; SE: sleep efficiency as percent of total sleep time; TLC: total lung capacity.

Combined adult and pediatric populations:

Eight studies described PSG results in patients of varying ages (Table 1C) and were therefore not included in the meta-analyses. One study included measurements in patients during and after a pulmonary exacerbation, showing disrupted sleep with increased wake time, reduced REM sleep, and increased nocturnal hypoxia during exacerbations [52]. A recent study showed a correlation between pulmonary artery systolic pressures and oxygen saturations during sleep; the authors suggested that these desaturations may contribute to the development of subclinical pulmonary hypertension [53].

Table 1C –

Polysomnography studies combining pediatric and adult CF populations

Reference	Controlled study	Sample size		Age (years)		Summary of main findings
		Cases	Controls	Cases	Controls
Muller 1980 [110]	Yes	20	5	9–29	22–30	REM desaturations mainly due to a decrease in FRC, leading to dependent lung region airway closure.
Stokes 1980 [111]	No	9	n/a	17–26	n/a	REM desaturations. Cough associated with desaturations and disruption of sleep cycles.
Francis 1980 [112]	Yes	20	5	18.2*		Same patients as Muller [110] Mean 7.4% desaturation, maximal during REM sleep in patients with CF compared with 2% in healthy controls.
Regnis 1994 [60]	No	7	n/a	14–39	n/a	Study of nCPAP comparing night on RA/O2 via nasal cannula with CPAP. see table 3.
Dobbin 2005 [52]	Yes	22	22	26 (9)*	30 (8)*	Study of the effect of an exacerbation and its treatment. During the exacerbation patients had more WASO, less REM, more hypoxemia. With treatment SE, REM, and hypoxia improved.
de Castro Silva 2009 [91]	Yes	40	20	6–28	8–25	Study compared patients with and without clinical lung disease with no differences in sleep architecture. Sleep desaturations predicted by FEV1.
Veronezi 2015 [92]	No	34	n/a	6–33 15.9 (7)	n/a	Determinants of sleep apnea were nutritional status, SaO2, and daytime sleepiness – model to predict AHI - explained 51% of the variation in the AHI.
Ziegler 2018 [53]	No	51	n/a	25.1 (8.8)*	n/a	Study of the association between clinical, lung function, sleep quality, and polysomnographic variables with PH showing a strong correlation between PH and sleep SaO2.

^*mean (SD).

n/a: not applicable. AHI: apnea hypopnea index; CF: cystic fibrosis; FEV₁: forced expiratory volume in one second; FVC: forced vital capacity; nCPAP: nasal continuous positive airway pressure; PH: pulmonary hypertension; REM: rapid eye movement; S_aO₂: oxygen saturation; SE: sleep efficiency as percent of total sleep time; WASO: wake after sleep onset.

Summary points:

• Patients with CF, both adults and children, have lower nocturnal saturation nadirs and poor sleep efficiency compared to healthy controls, but show no significant differences in REM as percent of TST.

• Children with CF have higher respiratory event indices compared with healthy children.

Correlation between lung disease and nocturnal findings

Pediatric studies:

In children with CF, five PSG and two nocturnal oximetry studies (Table 2) reported a correlation between the severity of lung disease, as assessed by FEV₁, and S_aO₂. Five studies showed a direct correlation between FEV₁ and mean S_aO₂ [37, 54] or S_aO₂ nadir [37, 40], and one reported that nocturnal hypoxia (S_aO₂<90% more than 5% of TST) correlated with lower FEV₁ [38]. Isaiah and colleagues reported that FEV₁ was the best predictor of nocturnal hypoxemia in children referred for PSG [39]. Conversely, Uyan and Spicuzza reported no correlation between FEV₁ and mean S_aO₂ [34, 55]. One additional study in preschool children found abnormal lung clearance index (LCI) in CF, but no significant associations with nocturnal saturation levels.[56]. Finally, Silva found a correlation between FEV₁ and SDB [37].

Table 2 –

Studies showing the correlation of lung disease and nocturnal findings

Reference	Type	N (cases)	Age (years)	Age group	Summary of main findings
Montgomery 1989 [113]	Oximetry	14	14 (9–34)**	Combined	FVC, but not FEV1 correlated with the severity of nocturnal desaturation.
Versteegh 1990 [114]	Oximetry	24	16 (10–22)†	Combined	FEV1 correlated with the lowest hourly mean (mean values of SaO2 for periods of one hour) SaO2.
Coffey 1991 [115]	Oximetry	21	21.9 (6.9)†	Combined	Disease severity correlated with nocturnal SaO2 mean and nadir.
Braggion 1992 [116]	Oximetry	31	15.2 (7.6–33.6)**	Combined	Pulmonary function parameters, except FVC, showed a poor correlation with overnight saturations.
Bradley 1999 [42]	PSG	14	25.9 (3.5)†	Adults	Air trapping (RV and RV/TLC) but not FEV1 correlated with nocturnal SaO2 nadir, peak transcutaneous CO2.
Milross 2001 [48]	PSG	32	27 (8)†	Adults	Disease severity correlated with minimum average SaO2 and REM related hypoventilation
Frangolias 2001 [57]	Oximetry	70	27.3 (8.7)†	Adults	Disease severity correlated with mean nocturnal SaO2 and 5% sleep time with SaO2 <90%.
Dobbin 2005 [52]	PSG	22	26 (9)†	Combined	When cases presented with an exacerbation, there was a strong correlation between percent-predicted FEV1 and TST minimum average SaO2, which persisted after treatment.
Uyan 2007 [55]	Oximetry	24	9.5 (8–12.5)*	Pediatrics	No correlation between FEV1 and mean nocturnal SaO2.
de Castro Silva 2009 [91]	PSG	40	6–28	Combined	Disease severity correlated with nocturnal SaO2 mean and nadir.
Bakker 2012 [56]	Oximetry	20	2.4 (1.3)†	Pediatrics	No correlation between LCI and mean nocturnal SaO2.
van der Giessen 2012 [54]	Oximetry	22	13 (6.3–18.7)†	Pediatrics	Disease severity correlated with mean nocturnal SaO2.
Fauroux 2012 [46]	Oximetry	80	24 (10)†	Combined	Disease severity correlated with nocturnal SaO2 mean and nadir.
Spicuzza 2012 [34]	PSG	40	0.5–11	Pediatrics	No correlation between FEV1 and mean nocturnal SaO2.
Perin 2012 [44]	PSG	51	25.1 (6.7)†	Adults	Disease severity correlated with mean saturation, % of sleep time <90% and peak end-tidal CO2.
Ramos 2013 [38]	PSG	67	2–14	Pediatrics	Disease severity correlated with nocturnal hypoxia defined as >5% sleep time with SaO2 <90%.
Silva 2016 [37]	PSG	33	6–18	Pediatrics	Disease severity correlated with nocturnal SaO2 mean and nadir but not trans-cutaneous CO2.
Waters 2016 [40]	PSG	46	8–12	Pediatrics	Disease severity correlated with nocturnal SaO2 nadir, the change in trans-cutaneous CO2 between NREM and REM.
Isaiah 2019 [39]	PSG	35	11.6 (9.5–13.1)††	Pediatrics	FEV1 <53% predicted was the best predictor of nocturnal hypoxemia, hypoxia defined as 5% sleep time with SaO2 <90%.
Gómez Punter 2019 [58]	Oximetry	57	27.5 (7.7)†	Adults	No correlation between 6-minute walk test or FEV1 and nocturnal SaO2.

^*median (IQR).

^**median (range).

^†mean (range or SD).

^††mean (95% confidence interval).

FEV₁: forced expiratory volume in one second; FVC: forced vital capacity; LCI: lung clearance index; NREM: non-rapid eye movement; PSG: polysomnography; REM: rapid eye movement; RV: residual volume; S_aO₂: oxygen saturation; TLC: total lung capacity.

Studies in adults:

Four studies reported a correlation between lung disease severity and nocturnal oxygen saturations: Bradley showed that air trapping (RV and RV/TLC), but not FEV₁, correlated with nocturnal S_aO₂ nadir and trans-cutaneous peak CO₂ [42], while Milross associated disease severity with mean minimum S_aO₂ from each 30-second epoch and REM hypoventilation reflected by the change in trans-cutaneous CO₂ between NREM and REM sleep [48]. Perin showed that disease severity correlated with mean saturations, the percent TST <90% and peak nocturnal end-tidal CO₂. Frangolias reported that disease severity correlated with nocturnal desaturations at home (mean and over 5% of sleep <90%), but could not determine thresholds that would reliably exclude those who did not desaturate [57]. Most recently, Gómez-Punter concluded that desaturations during the six minute walk test could not predict nocturnal desaturations, nor did the FEV₁ [58].

Combined adult and pediatric populations:

Several studies of PSG, oximetry, and CF disease severity combined both pediatric and adult populations, and are also summarized in Table 2.

Summary points:

• The majority of studies to date, in both children and adults, report a direct correlation between disease severity, mostly expressed by FEV₁, and nocturnal hypoxia.

Nocturnal noninvasive ventilation (NIV)

Studies of NIV in CF have mostly been conducted in adults with advanced lung disease and respiratory failure (Table 3), and most consist of small case series or retrospective reports of nocturnal NIV. Madden described their single center experience with 113 patients on NIV started in the setting of an acute deterioration. Some of the patients used NIV only at night. Patients were on NIV for up to 600 d, and showed improved oxygenation but did not exhibit improved hypercapnia [59]. The authors concluded that in CF patients awaiting lung transplantation, NIV does not merely prolong the time to inevitable death, but rather serves as “a bridge to transplantation”. Several studies documented improved oxygenation [60–63], alveolar ventilation [45, 61, 63–66], sleep [64–68] and daytime activity and symptoms [45, 64–68]. Sleep architecture in the PSG, total sleep time, and sleep efficiency were essentially unchanged [45, 60, 61, 63]. Flight compared the year prior to starting NIV to the following years, and showed a decelerated or reversed decline in lung function [69]. However, studies that reported shorter periods of follow-up mostly failed to show improved pulmonary function [45, 68]. In their Cochrane report of NIV in cystic fibrosis, which was recently updated in 2017, Moran concluded that NIV in addition to oxygen is superior to oxygen alone in improving gas exchange. However, the impact on pulmonary exacerbations and disease progression remains unclear, along with a need for long term randomized controlled trials [70].

Table 3 –

Studies of noninvasive ventilation (NIV) in CF

Reference	N (cases)	Age (years)	Age group	Summary of main findings
Piper 1992 [67]	4	25–30	Adult	Patients with respiratory failure, improved sleep and daytime activity with nocturnal NIPPV.
Regnis 1994 [60]	7	14–39 25 (8)*	Adult + pediatrics	Patients with severe lung disease. nCPAP improved RDI and oxygenation. Did not improve ventilation (CO2), sleep efficiency or total sleep time.
Padman 1994 [66]	7	9–30	Adult + pediatrics	Patients with clinically important derangements in gas exchange. BiPAP improved respiratory acidosis and subjective dyspnea, was associated with decline in respiratory rates and improved quality of sleep and capacity to perform activities of daily living.
Gozal 1997 [61]	6	13–28 22.3 (4.7)*	Adult + pediatrics	Patients with moderate to severe lung disease. NIPPV compared with low flow oxygen. Both improved oxygenation, CO2 increased with oxygen alone but improved with NIPPV. Sleep architect & arousal index unchanged.
Caronia 1998 [62]	9	19–43	Adult	End stage patients awaiting transplantation. Oxygenation, and respiratory rate improved with nocturnal BIPAP.
Hill 1998 [64]	10	26 (1.4)*	Adult	Patients awaiting transplantation. patients had a subjective improvement in headache and quality of sleep. At three months, there was significant improvement in FVC, and ventilation (PaCO2), and there was a reduction in the number of hospital inpatient days.
Milross 2001 [63]	13	26 (5.9)*	Adult	Patients with moderate to severe lung disease. Bilevel compared with low flow oxygen. Both improved oxygenation, only bilevel improved ventilation. Neither affected sleep architecture.
Granton 2002 [68]	8	Unavailable	Unavailable	Patients with respiratory failure and severe lung disease, awaiting transplantation. After two months on NIPPV, subjective but not objective improvement (sleep quality, daytime blood gases, PFT, exercise tolerance).
Efrati 2004 [65]	9	13–40	Adult + pediatrics	Patients with end stage lung disease. Long term NIPPV improved ventilation, blood gases, BMI and subjective sleep quality, morning headaches, and daily activity level.
Young 2008 [45]	8	37 (8)*	Adult	Compared with oxygen and room air, NIV for six weeks improved chest symptoms, exertional dyspnea, nocturnal hypoventilation and peak exercise capacity. Sleep architecture and PFT were unchanged.
Flight 2011 [69]	47	17–45	Adult	Retrospective review of patients started on NIV. Comparing the year prior to NIV with the following year NIV slowed or reversed the decline in lung function.

^*mean (SD).

n/a: not applicable. BIPAP: bilevel positive airway pressure; CF: cystic fibrosis; FVC: forced vital capacity; nCPAP: nasal continuous positive airway pressure; NIPPV: noninvasive positive pressure ventilation; NIV: noninvasive ventilation; PaCO₂: partial arterial pressure of CO₂; PFT: pulmonary function tests; RDI: respiratory disturbance index.

There were no specific reports of NIV in pediatric patients. Efrati [65] reported on a series of 9 patients, two of whom were children ages six and 17 y. The group showed improvement in physiologic parameters and subjective symptoms, but there appeared to be no changes in the physiologic parameters reported for the children.

Sleep questionnaires and actigraphy

A total of fourteen studies utilized different sleep questionnaires, actigraphy, and other outcomes measurements in children [2, 33, 37, 54, 71–80]. Similarly, fifteen studies of adults incorporated sleep questionnaires [27, 43, 44, 47, 48, 50, 81–90]. Finally, ten studies combined results from both children and adults [46, 53, 91–98]. These are all reviewed in the Online Supplement and in Supplementary Tables 1A-1C.

Cough, pain, and other symptoms affecting sleep in CF

Several studies have investigated cough and pain in CF, and how it affects sleep. These and several additional studies exploring unique questions are reviewed and summarized in the Online Supplement.

Assessment of bias

Supplementary Table 2 shows the results of the quality assessment for the studies included in the diverse meta-analyses performed in this review. Supplementary Figure 2 shows the funnel plots and Egger tests; we found no evidence of publication bias for any of the outcomes, although there was a non-significant trend for S_aO₂ nadir (Egger p=0.065; p>0.10 for all outcomes).

DISCUSSION

In this review, we have found that patients with CF have lower S_aO₂ nadir and that children with CF exhibit lower sleep efficiency than their peers without CF. While adults have an AHI/RDI in the normal range, this remains unclear in children, and the possibility exists that children with CF may be at higher risk of developing SDB. There appears to be a direct relationship between CF lung disease severity and lower overall S_aO₂ values as well as reduced S_aO₂ nadir. NIV improves gas exchange and symptoms with end-stage lung disease, but the long-term effects on pulmonary function and the effect in children remain unclear. Patients have more nocturnal cough than their peers, and experience painful events that interfere with their sleep. Actigraphy shows evidence of poor and fragmented sleep and questionnaires reveal a high frequency of disturbed sleep and of increased daytime sleepiness.

There are multiple reasons for poor sleep in CF. Most of the studies assessed sleep during stable conditions, suggesting that the sleep disruption identified was not solely the result of gas-exchange or other abnormalities during acute pulmonary exacerbations, but also the result of the chronic lung disease, nocturnal cough, and gastrointestinal symptoms such as gastroesophageal reflux and abdominal pain [2, 27]. In addition, TST may be limited by time-consuming treatments, such as chest physiotherapy and inhaled medications [71]. Our analysis indicates that sleep alterations begin early in the course of the disease. As sleep disturbances have implications on physical, psychological and cognitive health, early diagnosis and treatment would to be a prudent and worthwhile endeavor.

Polysomnographic data in children with CF demonstrate differences in sleep architecture and quality relative to healthy controls, as evidenced by the lower sleep efficiency. The effect of CF disease on OSA remains unclear, however. CF affects the entire respiratory system, causing sinopulmonary disease and impaired mucociliary clearance, which may play a role in the pathophysiology of OSA [99, 100]. This may explain the high prevalence of snoring and mouth breathing on questionnaires. The importance of this issue is emphasized by two case reports describing a gradual deterioration that improved after adenotonsillectomy [101, 102]. Furthermore, Suratwala reported an association between low S_aO₂ in CF and abnormal glucose regulation [35], and Simon a correlation between poor sleep on actigraphy and reduced insulin sensitivity [77]. The pathophysiology of OSA in adults is different from children, with obesity playing a major role, which likely explains the low prevalence of OSA in adult with CF.

Questionnaire studies have shown poor sleep quality in CF, however, the majority of studies reviewed did not directly explore the reason for sleep disturbances in CF. The exceptions to this are the studies on pain and cough in CF. Although the effect of pain on sleep quality is not entirely understood, it is clear that poor sleep leads to a higher perception of pain symptoms [103]. It remains to be seen whether improving the quality of sleep in these patients might also improve their pain. Even though patients cough less during the night this disrupts sleep causing both arousals and awakenings.

There are several limitations in the studies analyzed and the review itself. Though the majority of studies included in the meta-analysis were prospective, only a limited number noted blinding of polysomnography scorers [33] or researchers [27, 44]. Studies included in the systematic review were heterogenous, with different methods and outcomes. As is the case with any systematic review, available studies were limited in number, possibly a limitation resulting from looking at peer-reviewed, published manuscripts alone; however, we did not find evidence of publication bias.

From this review, it is clear that there are areas in which further studies are needed. In children, the question of the role of OSA and its treatment remains unclear. Similarly, the use and long-term effects of NIV in CF require further investigation with appropriately powered randomized controlled studies. With recent advances in CF and mutation specific therapies, new and highly sensitive methods to assess treatment response are required, with limited options to date [104]. A single study of both children and adults with the G551D mutation documented improved QOL on a sleep domain in patients started on Ivacaftor [96]. We suggest that future studies include sleep assessment by both sleep questionnaires, complementing the QOL questionnaires often used, and objective data by PSG and actigraphy.

In summary, there is both objective and subjective evidence of frequent sleep disorders in patients with CF, emphasizing the importance of addressing these concerns during routine patient visits. Additional research is needed, with larger sample sizes and standardized methods and outcomes, in order to better define these sleep abnormalities and their relationship with disease severity.

Practice Points

• Both subjective and objective sleep disorders are common in patients with cystic fibrosis and may have implications for quality of life and overall outcomes.

• Sleep assessments should be included in the routine care of cystic fibrosis.

Research Agenda

Several questions remain unanswered:

• In children with cystic fibrosis the role of obstructive sleep apnea and its treatment remain unclear.

• What is the relationship between sleep disorders and disease severity.

• What role does nocturnal non-invasive ventilation have in the management of patients and what are its long-term effects.

• How do the new modulator therapies affect sleep in cystic fibrosis, sleep measures should be included as an outcome measure in future studies of these medications.

• Do sleep disorders adversely affect overall longevity and clinical outcomes in patients with cystic fibrosis?

Abbreviation list (alphabetical):

AHI

apnea-hypopnea inde

AI

arousal index

CF

cystic fibrosis

CFTR

cystic fibrosis transmembrane conductance regulator

COPD

chronic obstructive pulmonary disease

ESS

Epworth sleepiness scale

FEV₁

forced expiratory volume in one second

FVC

forced vital capacity

LCI

lung clearance index

NIV

non-invasive ventilation

OSA

obstructive sleep apnea

PaO2

partial arterial pressure of oxygen

PSG

polysomnography

PSQI

Pittsburgh sleep quality index

QOL

quality of life

REM

rapid eye movement

RDI

respiratory disturbance index

RV

residual volume

S_aO₂

arterial oxygen saturation

SDB

sleep disordered breathing

SDSC

sleep disturbance scale for children

SE

sleep efficiency

SMD

standardized mean difference

SVAS

sleep visual analogu scale

TLC

total lung capacity

TST

total sleep time

WASO

wake after sleep onset

WMD

weighed mean difference