Lina Palaiodimou, MD
The role of sleep-disordered breathing (SDB) is well established in the development of a first stroke or death from any cause. Previous studies have shown that SDB is associated with an increased incidence of stroke or death from any cause, and this association is independent of other cardiovascular and cerebrovascular risk factors. This condition appears both in prestroke patients and in poststroke patients, and is more often obstructive than central. An association between SDB and recurrent strokes is also being presumed, but there are no sufficient prospective data to demonstrate and support this association.
The study of Brown et al. is an attempt to enrich the scarce data regarding the interaction between SDB and recurrent strokes. More specifically, the primary endpoint of this study is to investigate the association between SDB and recurrent ischemic stroke and secondary endpoints are: possible association between SDB and all-cause poststroke mortality and possible influence of ethnicity on the interaction between SDB and outcome measures (recurrent stroke or mortality). For that purpose, the investigators designed a prospective study of 842 patients who suffered from an index ischemic stroke and underwent a sleep apnea study shortly after the event. Additionally, patients had to be above 45 years old and be a resident of Nueces County, as the study was limited in 7 acute care hospitals of this certain county. Demographics, stroke risk factors, clinical variables and the REI (which is the sum of apneas plus hypopneas per hour of sleep apnea study duration) were recorded. Patients were followed until the first recurrent stroke, death or the last follow-up date, whichever came first. Proportional hazard models were conducted, both unadjusted and adjusted, to assess the association between REI and recurrent stroke or death. Finally, the interaction of ethnicity, REI and each outcome was statistically analyzed.
Before and after adjustment for demographics, hypertension, diabetes mellitus, body mass index and initial stroke severity, the investigators proved that REI is associated with recurrent ischemic stroke with a hazard ratio of 1.02 per one-unit increase in REI. Analyzing a 20-unit increase in REI, they demonstrated that such an increase is associated with recurrent stroke with a hazard ratio of 1.51. In contrast, REI was not proven to be associated with death from all causes, neither in the unadjusted analysis nor the adjusted one. The same applied for a 20-unit change in REI and mortality. Regarding ethnicity, the Mexican American one was associated with increased rates of recurrent ischemic stroke (not significantly) and mortality (significantly), but REI was not proven to confound the ethnicity and outcomes-of-interest association.
The limitations of the study of Brown et al. are the following: First, there were some violations of the proportional hazards assumption in the mortality model, as is highlighted by the authors. Second, the study is considered multicentered, but actually all the centers were located in only one county. What is more, the age restriction (above 45 years old) as an eligibility criterion is not justified by the authors. It also seems rather surprising that atrial fibrillation and coronary artery disease were not associated with SDB, an association otherwise well documented in the literature. Another methodological shortcoming is the possible underestimation of other potential confounders, such as atrial fibrillation, that should have been considered in the adjusted analysis.
Despite the limitations, the study of Brown et al. underscores the association between the severity of SDB (measured after the first ischemic stroke) and the risk for a recurrent stroke. This association gains importance by the fact that SDB is a factor potentially modified and treated, allowing to improve secondary stroke prevention. The next step would be the design of large randomized-controlled trials, in order to investigate whether treatment of SDB will reduce the risk of stroke recurrence.
