Intracerebral Hemorrhage and Sleep-Disordered Breathing

Abstract

Objective/Background

Limited data are available on sleep-disordered breathing (SDB) following intracerebral hemorrhage (ICH). Our aim was to characterize objective measures of post-ICH SDB, and questionnaire-reported pre-ICH sleep characteristics, overall and by ethnicity.

Patients/Methods

Participants with ICH enrolled in the population-based Brain Attack Surveillance in Corpus Christi project (2010–2016) reported their pre-ICH sleep duration and completed the Berlin Questionnaire to characterize pre-ICH risk of SDB. A sub-sample was screened for SDB (respiratory event index ≥10) using ApneaLink Plus portable monitoring. Ethnic differences in post-ICH SDB or reported pre-ICH sleep were assessed using log binomial or linear regression models or a Fisher’s Exact test.

Results

ICH cases (n=298) were enrolled (median age 68 years, 67% Mexican American). Among 62 cases with complete ApneaLink data, median time to post-ICH SDB screening was 11 days (IQR: 6, 19). Post-ICH SDB prevalence was 46.8% (95% CI: 34.4–59.2) and did not differ by ethnicity (p=1.0). Berlin Questionnaires for 109 of the 298 ICH cases (36.6% (95% CI: 31.1–42.0)) suggested high risk for pre-ICH SDB, and median pre-ICH sleep duration was 8 hours (IQR: 6, 8). No ethnic differences emerged in high risk for SDB or sleep duration pre-ICH, after adjustment for confounders.

Conclusions

Nearly half of patients had objective confirmation of SDB after ICH, and more than one-third had questionnaire evidence of high risk for pre-ICH SDB. Opportunities to address SDB may be common both before and after ICH.

1.1 Introduction

Sleep-disordered breathing (SDB) is prevalent following stroke.¹ Little high quality data are available more specifically on SDB following intracerebral hemorrhage (ICH). In a meta-analysis, in which 4 studies of ICH were included, prevalence of post-ICH SDB (defined as an apnea-hypopnea index ≥ 10) was 71% and did not vary by stroke type.¹ More recently, a single center Japanese study (N=92) found that 94% of ICH patients had SDB defined as a respiratory event index ≥ 5.² Data from population-based studies are lacking. ICH is a severe disease with high mortality and poor functional outcome.³ Understanding SDB burden in ICH may highlight new prevention and recovery strategies for this population that often faces a poor prognosis. Our objective was to determine prevalence of post-ICH SDB in a bi-ethnic population-based study. Given the higher prevalence of post-ischemic stroke SDB in Mexican Americans (MAs) compared with non-Hispanic whites (NHWs), we also sought to compare post-ICH SDB by ethnicity.⁴ Secondarily, we estimated prevalence of questionnaire-defined pre-ICH SDB risk and sleep duration, and compared results by ethnicity.

1.2 Methods

Methods for the Brain Attack Surveillance in Corpus Christi (BASIC) Project have been described.^{5, 6} BASIC is a population-based stroke surveillance study conducted in Nueces County, Texas. Stroke cases from all acute care hospitals in the study area are ascertained through active and passive surveillance and validated by study physicians. The present analysis included subjects with primary ICH who consented to participate in BASIC (July 2010–May 2016) and underwent a baseline interview, which included the Berlin questionnaire to assess pre-ICH SDB risk⁷ and a question on sleep duration (hours/night). Eligible patients were asked to participate in the SDB sub-study where an ApneaLink Plus portable cardiopulmonary monitor was used to screen for SDB defined as respiratory event index (REI) ≥ 10 (see Online Supplement for additional details regarding the SDB sub-study eligibility and scoring of the ApneaLink Plus data).⁸ Data were obtained from interviews or medical records. Informed consent was obtained from subjects or a surrogate. The study was approved by the Institutional Review Boards of the University of Michigan and the two local hospital systems.

Ethnic differences were assessed using a Fisher’s Exact test (post-ICH SDB), log binomial (pre-ICH SBD), or linear regression models (sleep duration). Given the larger sample for interview data, models were adjusted for pre-specified potential confounders (age, sex, alcohol intake, diabetes, Glasgow Coma Scale (GCS), body mass index (BMI)).

1.3 Results

Among 411 ICH cases, 298 (72.5%) had a complete baseline interview. Those who completed the interview (Online Supplement Table 1) were similar to those who did not participate (Online Supplemental Table 2). Seventy-five of the 298 cases participated in the SDB sub-study. Participants versus non-participants of the SDB sub-study were younger, more likely male, had higher BMI, and were less likely to have atrial fibrillation or coronary disease; they also had milder ICHs (Online Supplemental Table 3).

Among 75 sub-study participants, 62 had complete ApneaLink data. Median time to SDB screening was 11 days (IQR: 6, 19). Data on the prevalence of post-ICH SDB and high risk for pre-ICH SDB overall and by ethnicity are included in Table 1. Among those with post-ICH SDB, 71.4% were at high risk for pre-ICH SDB. REI was not correlated with ICH severity (Online Supplemental Figure 1).

Table 1.

Pre- and Post-ICH Sleep Characteristics of Intracerebral Hemorrhage Patients

	Overall % (N) or median (Q1,Q3)	MA % (N) or median (Q1,Q3)	NHW % (N) or median (Q1,Q3)
Pre-ICH (N=298)
High risk for SDB	36.6 (109)	42.7 (85)	22.4 (19)
Sleep duration (hours)	8 (6, 8)	8 (6, 8)	8 (7, 8)
Post-ICH (N=62)
REI	9.5 (5, 19)	9.5 (6, 19)	8.5 (3, 18)
CAI	0 (0, 1)	0 (0, 1)	0 (0, 2)
REI≥10	46.8 (29)	46.3 (25)	50 (3)
REI≥5	83.9 (52)	87 (47)	66.7 (4)
REI≥15	35.5 (22)	35.2 (19)	33.3 (2)
REI≥30	19.4 (12)	20.8 (11)	16.7 (1)

ICH = intracerebral hemorrhage, SDB = sleep disordered breathing, REI = respriatory event index, CAI = central apnea index

Prevalence of post-ICH SDB did not differ by ethnicity (p=1.0) but prevalence of high risk for pre-ICH SDB was significantly higher among MAs than NHWs (Prevalence Ratio (PR)=1.91; 95% CI:1.25, 2.93). Baseline characteristics by ethnicity are included in Online Supplemental Table 4. After adjustment, no ethnic difference remained (PR=0.99; 95% CI: 0.65, 1.43). Pre-stroke sleep duration (Table 1) did not differ by ethnicity (unadjusted mean change=−0.44; 95% CI: −1.77, 0.89; adjusted mean change=0.58; 95% CI: −0.95, 2.11).

1.4 Discussion

Nearly half of ICH patients had objective evidence of post-ICH SDB. Among those with post-ICH SDB, more than two-thirds were at high risk for pre-ICH SDB. This supports the idea that SDB often precedes ICH, as has also been suggested in ischemic stroke.⁹ Though still high, our prevalence of post-ICH SDB is lower than previous reports.^{1, 10} Sample size ranged from only 6 to 15 in the studies included in the meta-analysis, suggesting these populations were highly select. In a more recent single-center study (n=32) with a reported prevalence of 67% using the same criteria for defining SDB, polysomnography was completed on the first night of hospitalization in contrast to our screening, which occurred a median of 11 days after ICH onset. If, as hypothesized, the acute phase of ICH and related edema are associated with SDB,¹¹ the later timing of our screening could partially account for our lower estimates. Our lower prevalence could also be due to the exclusion of ICH patients at higher risk for SDB. Arguing against this, participants in the SDB sub-study had a higher prevalence of some post-stroke SDB risk factors, including male sex, hypertension, diabetes, and higher BMI, than non-participants.^{9, 12, 13} In contrast, participants were younger, had a lower prevalence of coronary artery disease and atrial fibrillation, and milder ICHs, though ICH clinical severity did not correlate with the REI in our sample. Our results are bolstered by the population-based study design which adds to the generalizability of our findings to the broader ICH population.

Prevalence of post-ICH SDB was lower than our reported prevalence of 62% in post-ischemic stroke patients from the same community.⁴ Others similarly have observed a higher prevalence of SDB in ischemic versus hemorrhagic stroke patients,¹¹ while the meta-analysis did not find a difference.¹ Possible hypotheses for differences include 1) SDB is less permanent after ICH than after ischemic stroke,¹¹ or 2) SDB is already more common before ischemic stroke than it is before ICH, although our results suggest that many patients with post-ICH SDB were likely to have had SDB before their stroke. Additional research is needed to understand differences in SDB by stroke type.

In contrast to the finding of a higher prevalence of SDB in MAs compared with NHWs following ischemic stroke,⁴ we found no ethnic difference in post-ICH SDB. Our results should be interpreted with caution given small numbers. Prevalence of high-risk for pre-ICH SDB was considerable at 37%, but lower than our ischemic stroke population (59.6%). Similar to our ischemic stroke population, ethnic differences were evident. Although this difference was explained by a higher prevalence of SDB risk factors in MAs, the ethnic difference suggests an opportunity to identify and treat SDB in MAs which could lessen stroke disparities. Although pre-stroke sleep duration did not differ by ethnicity, 50% of the population reported having a long sleep duration which has been shown to be associated with ICH risk.¹⁴

Limitations include use of an overnight cardiopulmonary monitoring device. However, the device has been well validated for the identification of SDB, and full polysomnography is not well tolerated in the acute stroke period. The Berlin Questionnaire was administered post-stroke in reference to the pre-stroke period although the interview was done soon after stroke to minimize measurement error. The Berlin Questionnaire has a sensitivity of 86% and specificity of 77%; thus, some false positives and false negatives are possible.⁷ Due to the observational nature of the study, standardized imaging was not performed on all subjects; thus, MRI and blood vessel imaging were not collected and ICH location (e,g. deep vs lobar) was not available.

Conclusions

In our population-based study, post-ICH SDB was prevalent with nearly half of patients showing SDB by objective assessment. High risk of pre-ICH SDB was suggested by pre-ICH symptoms in over a third of ICH patients. Though additional outcomes research is required, these observations highlight potential clinical opportunities to address SDB both before and after ICH.

Highlights.

• Nearly half of patients had objective confirmation of sleep-disordered breathing after intracerebral hemorrhage.

• More than one-third of patients had questionnaire evidence of high risk for pre-intracerebral hemorrhage sleep disordered breathing.

• Opportunities to address sleep disordered breathing may be common both before and after intracerebral hemorrhage, a severe disease with poor prognosis