In children with sickle cell disease (SCD), obstructive sleep apnoea (OSA) is common and the degree of overnight hypoxia is closely correlated with disease severity, particularly frequency of painful crises, risk of stroke and presence of pulmonary hypertension. No studies have examined the prevalence or effects of OSA in adults with SCD.
All adults with SCD attending routine sickle cell out-patient clinics were offered screening for OSA using the Epworth Sleepiness Score (ESS) questionnaire. All of those in whom the score was >10 or in whom there was a high clinical suspicion of OSA were offered overnight domiciliary oximetry. Oximetry traces were both analysed by computer and manually. Computerised analysis was used to quantify night time hypoxic load (mean nocturnal SpO2 and sleep time with SpO2 below 90%) and frequency of desaturation (number of times per hour the SpO2 dropped by >4%, or Overnight Desaturation Index, ODI). Manual analysis of each trace was performed by four independent sleep physicians who assigned a diagnosis of normal, OSA, non-OSA nocturnal hypoxia or inadequate based on the pattern of the oxygen saturation and heart rate traces. These parameters were then correlated with measures of disease severity and presence of complications taken from the patient's medical record. Statistical analysis was performed using the t-test and linear regression.
93 patients completed the ESS, with 34 patients identified for subsequent oximetry (26 ESS>10, 8 clinical suspicion). 22 patients went on to have subsequent oximetry. 17/22 of recordings (77%) were abnormal of which 11 showed OSA (65%) and 6 showed nocturnal hypoxia without OSA (35%). Nocturnal hypoxic load, measured as mean overnight oxygen saturation, was correlated with glomerular filtration rate (r=0.5; p=0.0008) and pulmonary artery systolic pressure estimated by echocardiography (r=0.71; p=0.0001). Furthermore, urine protein level correlated with ODI (r=0.5 p=0.0007) and mean overnight oxygen saturation (r=0.35; p=0.02). In addition, mean overnight oxygen saturation was 87% in male patients that experienced priapism and 94% in those who did not (p=0.004). Mean ODI was 10/hr in patients with priapism and 4 in those without (p=0.008). There were no correlations demonstrated between either hypoxic load or ODI with the frequency of painful crises, frequency of hospital admissions or avascular necrosis. Lung function test parameters, history of chest syndrome, hydroxyurea therapy and Epworth score did not predict either hypoxic load or ODI. However, daytime SpO2 were highly predictive of nocturnal hypoxic load (r=0.69 p=0.0000007) but not ODI.
OSA and non-OSA nocturnal hypoxia are common in adults with sickle cell disease. The degree of nocturnal hypoxia is associated with impaired renal function, proteinuria, priapism and pulmonary hypertension. Daytime SpO2 appears to be a better predictor of night time hypoxic load than daytime somnolence. Further work is needed to investigate the effect of interventions such as nocturnal oxygen therapy or continuous positive airway pressure on the associations reported.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.