Demographic and clinical characteristics between OSA and no OSA patients were reported in Table 1. The mean age of patients was 69.39 ± 10.13 years with an expected male predominance in the OSA group (86.1% versus 61.1%, p = 0.038). All 27 patients (27.8% of the cohort) with definite OSA based on polysomnography had a mNOSAS score ≥ 8 and so belonged to the OSA group (34.2% of the OSA group).
Table 1
Demographic characteristics.
variables | Total (n = 97) | OSA (n = 79, 81.4%) | no OSA (n = 18, 18.6%) | p-value |
age (at admission) | 69.4 (± 10.1) | 70.3 (± 9.7) | 65.3 (± 11.3) | 0.0962 |
male | 79 (81.4%) | 68 (86.1%) | 11 (61.1%) | 0.038¹ |
education degree* | | | | 0.774¹ |
1 | 14 (17.9%) | 11 (16.9%) | 3 (23.1%) | |
2 | 36 (46.2%) | 31 (47.7%) | 5 (38.5%) | |
3 | 28 (35.9%) | 23 (35.4%) | 5 (38.5%) | |
≥ 1 vascular risk factors | 74 (76.3%) | 63 (79.7%) | 11 (61.1%) | 0.124¹ |
Body Mass Index (BMI) | 27.97 (± 5.57) | 28.99 (± 5.62) | 23.5 (± 2.22) | < 0.0012 |
smoking | 16 (16.8%) | 14 (17.9%) | 2 (11.8%) | 0.728¹ |
blood pressure hypertension | 63 (64.9%) | 53 (67.1%) | 10 (55.6%) | 0.416¹ |
diabete | 36 (37.1%) | 31 (39.2%) | 5 (27.8%) | 0.427¹ |
dyslipidemia | 34 (35.1%) | 29 (36.7%) | 5 (27.8%) | 0.589¹ |
Definite OSA** | 27 (27.8%) | 27 (34.2%) | 0 | 0.003¹ |
apnea hypopnea index (/h)** | 53.68 (± 24.92) | 53.68 (± 24.92) | - | - |
night breath device for OSA** | 13 (50.0%) | 13 (16.5%) | 0 | - |
modified NOSAS score** | 11.12 (± 3.62) | 12.49 (± 2.28) | 5.11 (± 1.81) | < 0.0012 |
preexisting cognitive disorder*** | 17 (17.7%) | 14 (17.9%) | 3 (16.7%) | 0.999¹ |
preexisting heart disease*** | 18 (28.1%) | 16 (30.2%) | 2 (18.2%) | 0.714¹ |
preexisting respiratory disease*** | 9 (14.1%) | 8 (15.1%) | 1 (9.1%) | 0.999¹ |
toxic use*** | 10 (10.3%) | 9 (11.4%) | 1 (5.6%) | 0.683¹ |
¹ Fisher's exact test. Table results were given in number of patients (percentage of total number of patients per group).
2 t-test. Table results were given in median (± interquartile ratio).
* Education degree was defined as followed: 1=primary education, 2=lower secondary education, 3=upper secondary education.
** Definite OSA was assessed by polysomnography (gold standard).12,15 The Apnea Hypopnea Index was used to indicate the severity of definite OSA. The Apnea Hypopnea Index is the number of apneas or hypopneas recorded per hour of sleep (number of events per hour). Based on the Apnea Hypopnea Index, the severity of OSA is classified as follows: None/Minimal: < 5 per hour; Mild: ≥ 5, but < 15 per hour; Moderate: ≥ 15, but < 30 per hour; Severe: ≥ 30 per hour.12,15 Among patients with definite OSA, some of them usually used night devises such as continuous positive airway pressure (CPAP) or oral appliances like the mandibular advancement device.12 The NOSAS score classified patients at high risk for significant OSA with the following items: neck circumference, obesity, snoring, age and sex (NOSAS).18 Snoring information was recorded for only 19 patients (missing data: 78), so we used a modified NOSAS score that did not include snoring information.
*** Preexisting cognitive disorder was defined as any cognitive disorder impairment by a neurologist before the beginning of COVID-19 acute encephalopathy. Preexisting heart and respiratory disease were defined as any heart or respiratory disease diagnosed by a cardiologist, a pulmonologist or a general practitioner prior to the onset of COVID-19 acute encephalopathy. Toxic use was reported by the physician in charge at the time of hospital admission (anamnesis/heteroanamnesis).
Abbreviation: OSA= obstructive sleep apnea.
Table 2 shows patients’ characteristics at the time of COVID-19 onset (general examination data and COVID-19 pulmonary imaging status) and at COVID-19 AE paroxysm (neurological signs, encephalopathy features, epidemiological features, biological data from blood and CSF, brain MRI data and EEG results). At the time of COVID-19 onset, common general symptoms were dyspnea (55.7%), cough (63.2%) and fever (82.5%). Oxygen needs at admission was FiO2 32.31% (± 17.30), 27.84% (± 11.38) in the no OSA group compared to 33.35%(± 18.31) in the OSA group (p = 0.113). The percentage of COVID-19 related pulmonary lesions determined with pulmonary CT scan was similar between the two groups (p = 0.963).
Table 2
Patient characteristics at the time of COVID-19 onset and COVID-19 acute encephalopathy.
variables | Total (n = 97) | OSA (n = 79, 81.4%) | no OSA (n = 18, 18.6%) | p-value |
General examination and parameter at COVID-19 onset |
dyspnea | 54 (55.7%) | 42 (53.2%) | 12 (66.7%) | 0.4311 |
cough | 60 (63.2%) | 50 (64.9%) | 10 (55.6%) | 0.5881 |
fever | 80 (82.5%) | 67 (84.8%) | 13 (72.2%) | 0.299¹ |
FiO2 | 32.31 (± 17.30) | 33.35 (± 18.31) | 27.84 (± 11.38) | 0.1132 |
Percentage of lung parenchyma with COVID-19 lesions (pulmonary CT scan) | 0.5832 |
0–30% | 13 (20.0%) | 11 (20.4%) | 2 (18.2%) | |
30–50% | 22 (33.8%) | 19 (35.2%) | 3 (27.3%) | |
50–70% | 17 (26.2%) | 14 (25.9%) | 3 (27.3%) | |
> 70% | 13 (20.0%) | 10 (18.5%) | 3 (27.3%) | |
Neurological signs at COVID-19 acute encephalopathy |
fluctuation | 87 (89.7%) | 73 (92.4%) | 14 (77.8%) | 0.085¹ |
inattention | 86 (88.7%) | 74 (93.7%) | 12 (66.7%) | 0.005¹ |
thought disturbance | 68 (74.7%) | 61 (82.4%) | 7 (41.2%) | 0.001¹ |
alertness trouble | 48 (48.5%) | 44 (54.4%) | 4 (22.2%) | 0.026¹ |
drowsiness | 51 (52.6%) | 42 (53.2%) | 9 (50.0%) | 0.999¹ |
agitation | 33 (34.0%) | 29 (36.7%) | 4 (22.2%) | 0.283¹ |
psychomotor slowdown | 63 (65.6%) | 55 (70.5%) | 8 (44.4%) | 0.053¹ |
obnubilation | 32 (34.4%) | 28 (36.8%) | 4 (23.5%) | 0.401¹ |
perseveration | 53 (58.2%) | 47 (64.4%) | 6 (33.3%) | 0.031¹ |
disorientation | 48 (54.5%) | 43 (60.6%) | 5 (29.4%) | 0.030¹ |
hallucination | 17 (19.3%) | 14 (19.4%) | 3 (18.8%) | 0.999¹ |
focal neurological sign | 27 (27.8%) | 23 (29.1%) | 4 (22.2%) | 0.772¹ |
COVID-19 acute encephalopathy features |
CAM | 2.97 (± 1.05) | 3.18 (± 0.89) | 2.06 (± 1.21) | 0.0013 |
RASS ≤ -3 | 8 (8.2%) | 7 (8.9%) | 1 (5.6%) | 0.1681 |
mutism | 14 (14.4%) | 12 (15.2%) | 2 (11.1%) | 0.999¹ |
severe encephalopathy* | 72 (74.2%) | 67 (84.8%) | 5 (27.8%) | < 0.001¹ |
duration of encephalopathy (days) | 25.8 (± 59.3) | 27.9 (± 65.5) | 16.9 (± 12.4) | 0.0183 |
Biological results in the blood |
C-reactive protein (mg/l) | 89.85 (39.2-163.4) | 89.6 (41.9–163.6) | 90.1 (9.7–159.3) | 0,5042 |
leucocytes (/mm3) | 9.88 (7.3–12.2) | 9.84 (7.2–12) | 11.35 (8.5–14.2) | 0,2152 |
lymphocytes (/mm3) | 0.82 (0.5–1.1) | 0.79 (0.5–1) | 0.97 (0.7–1.2) | 0,2302 |
segmented neutrophils (/mm3) | 7.47 (5.5–10.4) | 7.35 (5.1–10) | 9.09 (6.4–11.7) | 0,2722 |
monocytes (/mm3) | 0.53 (0.3–0.8) | 0.52 (0.3–0.8) | 0.58 (0.4–0.7) | 0,9982 |
thrombocytes (/mm3) | 298 (232-354.5) | 307 (241–356.8) | 264 (167–319) | 0,3452 |
Brain MRI |
missing values** | 35 | 27 | 8 | |
leucoencephalopathy | | | | 0.050¹ |
0 | 23 (31.5%) | 17 (28.3%) | 6 (46.2%) | |
1 | 32 (43.8%) | 30 (50.0%) | 2 (15.4%) | |
2 | 7 (9.6%) | 6 (10.0%) | 1 (7.7%) | |
3 | 11 (15.1%) | 7 (11.7%) | 4 (30.8%) | |
stroke (DWI lesion) | 34 (47.2%) | 26 (44.1%) | 8 (61.5%) | 0.359¹ |
hyperT2 lesion (number) | 7.12 (± 6.22%) | 7.3 (± 6.2) | 6.31 (± 6.54) | 0.6233 |
microbleed (number) | 3.03 (± 7.43%) | 2.9 (± 7.75) | 3.62 (± 5.92) | 0.7143 |
number of vessels with endotheliitis*** | | | | < 0.001¹ |
0 | 9 (14.5%) | 1 (1.9%) | 8 (80.0%) | |
1 | 18 (29.0%) | 16 (30.8%) | 2 (20.0%) | |
2 | 16 (25.8%) | 16 (30.8%) | 0 | |
3 | 19 (30.6%) | 19 (36.5%) | 0 | |
circumferential endotheliitis*** | 48 (77.4%) | 48 (92.3%) | 0 | < 0.001¹ |
endotheliitis*** | | | | < 0.001¹ |
unilateral | 26 (41.9%) | 24 (46.2%) | 2 (20.0%) | |
bilateral | 27 (43.5%) | 27 (51.9%) | 0 | |
stenosis | 3 (4.1%) | 3 (5.0%) | 0 | |
EEG slowing | 47 (53.4%) | 40 (55.6%) | 7 (43.8%) | 0.420¹ |
Epidemiological features |
intensive care unit | 62 (63.9%) | 50 (63.3%) | 12 (66.7%) | 0.9993 |
hospitalization time | 43.5 [25.0-61.5] | 43.5 [25.0–61.0] | 42.5 [28.0–66.0] | 0.7962 |
mRS at discharge | | | | 0.0081 |
0 | 9 (9.3%) | 8 (10.1%) | 1 (5.6%) | |
1 | 23 (23.7%) | 13 (16.5%) | 10 (55.6%) | |
2 | 16 (16.5%) | 12 (15.2%) | 4 (22.2%) | |
3 | 22 (22.7%) | 21 (26.6%) | 1 (5.6%) | |
4 | 14 (14.4%) | 13 (16.5%) | 1 (5.6%) | |
5 | 1 (1.0%) | 1 (1.3%) | 0 (0.0%) | |
6 (death) | 12 (12.4%) | 11 (13.9%) | 1 (5.6%) | |
mRS at discharge ≥ 3 | 49 (50.5%) | 46 (58.2%) | 3 (16.7%) | 0.002¹ |
This table presents patient characteristics at the time of COVID-19 onset (general examination data and COVID-19 pulmonary imaging status) and at COVID-19 acute encephalopathy paroxysm (neurological signs, encephalopathy features, epidemiological features, biological data from blood and CSF, brain MRI data and EEG results).
¹ Fisher's exact test. Table results were given in number of patients (percentage of total number of patients per group).
2 Mann-Whitney u test. Table results were given in median (± interquartile ratio).
3 t-test. Table results were given in median (± interquartile ratio).
* severe encephalopathy was defined on a RASS < −3 at worst presentation ─ meaning deep sedation, no response to voice but possible movement or eye, opening to physical stimulation; or on a CAM score ≥ 3 among patients with a RASS ≥ -3 ─ meaning displaying 3 out of 4 items among symptoms fluctuation, inattention, thought disturbance, and altered alertness.
** Missing value. Brain MRI were missing in many patients because of the inability to perform these tests due to patient compliance at the acute phase of COVID-19 encephalopathy.
***The term “endotheliitis” referred to homogeneous gadolinium contrast enhancement of the inner part of the vessel wall (injected brain MRI) without stenosis. Circumferential endotheliitis referred to contrast enhancement of the vessel wall greater than 50% of the circumference.
Abbreviations: CAM = Confusion Assessment Method, COVID-19 AE = COVID-19 acute encephalopathy, DWI = diffusion-weighted imaging, OSA = obstructive sleep apnea, mRS = modified Rankin Scale, RASS = Richmond Agitation Sedation Scale.
Neurological signs at COVID-19 AE paroxysm were also presented in Table 2. Patients with OSA versus no OSA presented more often with inattention (93.7% vs. 66.7%; p = 0.005), thought disturbance (82.4% vs. 41.2%; p = 0.001), alertness trouble (54.4% vs. 22.2%, p = 0.026), perseveration (64.4% vs. 33.3%; p = 0.031), and disorientation (60.6% vs. 29.4%; p = 0.030). Patients with OSA exhibited more often severe COVID-19 AE on neurological examination compared to those without OSA (84.8% vs. 27.8%; p < 0.001). The mean CAM score was significantly different in the two groups: 3.18 (standard deviation (SD) ± 0.89) in the OSA group versus 2.06 (± 1.21) in the no OSA group (p = 0.001). The mean duration of COVID-19 AE significantly differed between the two groups: 27.9 (± 65.5) days in the OSA group versus 16.9 (± 12.4) in the no OSA group (p = 0.018).
The two groups did not significantly differ in blood biological data (Table 2). Notably, the mean serum concentration of C-reactive protein was 89.85 (39.2-163.7) mg/L. Only 37 patients (38.1%) had a CSF analysis (supplemental data). Of these, CSF white blood cell count was normal in 35 patients (94.6%): all patients in the no OSA group had white blood cell count ≤ 1 /cm3 while 18 (54%) patients of the OSA group have white blood cell count ≥ 2 /cm3. The mean CSF white blood cell count was 4.3 (± 15.5) /cm3 in the OSA group compared to 0.5 (± 0.6) /cm3 in the no OSA group (p = 0.056). The mean CSF lymphocyte count was 2.1 (± 2.0) /cm3 in the OSA group compared to 0.9 (± 0.1) /cm3 in the no OSA group (p = 0.005). The mean CSF macrophages count was 1.1 (± 1.3) (%) in the OSA group compared to 0.4 (± 0.5) (%) in the no OSA group (p = 0.301), but 7 patients (21%) of the OSA group have more than 1% of activated macrophages compared to none in the no OSA group. The rest of CSF analyses did not evidence any difference between the 2 groups. RT-PCR for SARS-CoV-2 was negative for all patients in the CSF.
Injected MRI was available for 62 patients (63.9%) — 52 (65.8%) in the OSA group and 10 (55.6%) in the no OSA group (Table 2). Noteworthy, intracranial vessel gadolinium enhancement was observed in 53 patients (85.5%), with predominance in the OSA group (98.1%) compared to no OSA group (20%) (p < 0,001). The vast majority of the vessel enhancement was circumferential (48 patients, 77.4%) and found on vertebral arteries without sign of stenosis or downstream ischemia. Among patients with intracranial vessel gadolinium enhancement, the number of enhancing vessels was significantly higher in the OSA group compared to the no OSA group ─ respectively 35 patients (67.3%) in the OSA group versus none in the no OSA group with more than one enhancing vessel (p < 0,001). Bilateral intracranial enhancing vessel involvement was only described in the OSA group ─ 27 patients (51.9%, p < 0,001). Cerebral microbleeds were reported in 56 of 97 patients (57.7%). The mean number of microbleeds by patients was 3.03 (± 7.43) with no difference in the two groups. Finally, there were no differences in term of brain MRI parenchyma abnormalities (acute stroke, T2-FLAIR hyperintensities, microbleeds) between the two groups. The Fig. 2 illustrates MRI sequences of one patient from the OSA group with evidence of intracranial vessel gadolinium enhancement without vascular stenosis nor any parenchymal lesion.
No ictal discharge was reported at electroencephalogram (EEG), while EEG slowing was noticed in 47 of 97 patients (53.4%) with no difference between the groups.
The modified Rankin scale (mRS) at discharge was statistically significantly different between the two groups (p = 0.008). Forty-six patients (58.2%) from the OSA group presented mRS ≥ 3 meaning moderate to severe deficit or death, as compared to three patients (16.7%) in the no OSA group (p = 0.002). Twelve patients included in the current series died during their hospitalization, 11 in the OSA group (13.9%) and 1 in the no OSA group (5.6%, p = 0.45). The median length of hospital stay was similar between the two groups (44 (27-61.8) days) without any difference between the two groups: 44.5 (26.5–61.2) in the OSA group compared to 42.5 (30-62.8) in the no OSA group (p = 0.594).
Using a stepwise forward multiple logistic regression model, high risk for significant OSA based on the mNOSAS score was selected by the model with a 14 times risk of developing a severe COVID-19 AE (OR = 14.52; 95% CI [4.37–48.22]; p < 0.001), explaining 19.9% of the variability of the severity. Using the same method, the age at admission (OR = 1.07; 95% CI [1.01–1.13]; p = 0.018), the mNOSAS score (OR = 1.17; 95% CI [0.99–1.38]; p = 0.058), reduced dyspnea sensation at admission (OR = 0.28; 95% CI [0.098–0.79]; p = 0.017) and the severity of COVID-19 AE (OR = 7.37; 95% CI [1.87–29.0]; p = 0.004) were selected by the model to be associated with the risk of developing severe disability at discharge (mRS score ≥ 3), explaining 29.1% of the variability of the disability at discharge.