Because of the relatively high cardiovascular event rates amongst patients who have established cardiovascular disease, the numbers of patients that need to be enrolled in a secondary prevention trial are generally in the order of 20-30% of the number needed for a primary prevention trial. While a secondary prevention trial design may target somewhat different pathogenic mechanisms for cardiovascular disease than a primary prevention study (Fig 1), it nevertheless has the potential to provide highly relevant information with which to inform clinical practice and reduce cardiovascular morbidity and mortality. Furthermore, it increases the overall feasibility of successfully completing such a trial, given the smaller number of patients and shorter period of follow-up required, and thus, ultimately, lower overall trial cost.

A secondary prevention trial of OSA treatment requires access to a large number of patients with co-existing cardiovascular disease and OSA who are willing to be entered with a 50:50 chance of being allocated to active OSA treatment or usual care control. It was considered that this would not be easily achieved by relying solely on patients referred to sleep medicine services. As well as the numbers of patients referred to sleep services being lower in comparison to the numbers referred to cardiovascular clinics, it is likely that the proportion of sleep clinic patients with co-existing cardiovascular disease is likely to be relatively low (ie in the order of 10% (6)). Even more important, however, is that sleep clinic patients will, in the main, be symptomatic and expect to be offered OSA treatment. The alternative method of recruiting patients directly from cardiovascular clinics is more attractive, although it has its own challenges. It is known that the prevalence of OSA is about 30 to 60% in high cardiovascular risk populations (25-29). Thus, cardiovascular clinics are potentially a rich source of patients for SAVE. However, we were aware that this would either require cardiovascular clinician researchers to develop new skills in sleep apnea diagnosis and CPAP treatment or for respiratory/ sleep and cardiovascular physicians to form strong collaborative working relationships at the local site level. In the end, it was decided to invite both sleep/ respiratory and cardiovascular clinicians to participate in the study, to provide training and support in sleep diagnostics and CPAP therapy where needed, and to encourage collaboration between the medical disciplines at national and local hospital levels.

It has become apparent to experts in the sleep field (30) that in-laboratory polysomnography (PSG) is not a costeffective method to screen and diagnose OSA large numbers of subject that are required for a study such as SAVE. As well as it being a high-cost and labour intensive test to perform, there is considerable cost and effort required to standardise PSG recording and scoring techniques between laboratories (or centralise scoring), which in a study with multiple sites, was considered prohibitive. Furthermore, we were aware that many of our potential recruitment sites in China did not have access to PSG and many Australian sites with PSG facilities were already heavily booked with clinical work. There has been increasing evidence that simplified home screening devices can be used to identify cases of at least moderate-severe OSA with a high degree of certainty, at least in populations with a high pre-test probability of disease. Since the prevalence of OSA in the proposed study population (i.e. patients with established coronary- or cerebro-vascular disease) was likely to be high, it was decided to validate a simple, automated 2-channel (oximetry and nasal pressure) screening device (the ApneaLink, ResMed, Sydney, Australia) in a high cardiovascular risk population in China with the view to using this device in the main study. Chinese and Australian clinician scientists collaborated on this validation study which was completed in Shanghai in early 2008. It was found in 143 high cardiovascular risk patients that the automatically calculated oxygen desaturation (oximetry) and apnea-hypopnea (nasal pressure) indices had equally high diagnostic accuracy for moderate-severe OSA when compared with full PSG simultaneously performed in the patients’ homes (Gantner et al Respirology in press). ApneaLink Oximetry had a lower technical failure rate (e.g. loss of signal because of sensor displacement) than nasal pressure recordings making oximetry the preferred primary diagnostic method for identifying OSA patients for SAVE. The nasal pressure trace is used to exclude patients whose predominant pattern of sleep disordered breathing pattern is symmetrical waxing and waning of flow indicating Cheyne-Stokes respiration.

The main ethical issue relevant to the SAVE study relates to the withholding of CPAP treatment in approximately half the patients who screen positive for OSA, when they may stand to benefit from reduction in daytime sleepiness, and improved driving safety and quality of life with therapy. Since patients in SAVE are recruited from cardiovascular clinics, almost all participants will be unaware of having had OSA prior to the SAVE screening diagnostic evaluations. While previous studies have shown that the great majority of patients with co-existing cardiovascular disease and OSA have little to no daytime symptoms, to minimise any safety concerns it was decided to exclude patients who held a commercial drivers, who demonstrated marked daytime sleepiness (defined as Epworth Sleepiness Scale score > 15), or who reported a fallasleep accident or near miss accident in the 12 months prior to enrolment. An independent Data Safety Monitoring Board (DSMB) was established to monitor the rates of self-reported accidents in the CPAP-treated and non CPAP-treated groups at regular intervals during the course of the study. At the time of enrolment, all patients are given a full explanation of the possible symptomatic benefits of CPAP and given the option to seek treatment outside the trial if they wish, and both the patient and their responsible physician must be comfortable about them being randomly allocated to CPAP treatment or no CPAP treatment.

It was considered that adherence to CPAP therapy might be a significant problem in a long-term study of several years duration, particularly considering that the majority of subjects were likely to report little or no daytime sleepiness and therefore would be unlikely to experience significant symptomatic benefit. In addition it is known from clinical studies that as many as 30% of sleep apnea patients refuse CPAP treatment outright or in the first few weeks of therapy (31). To exclude those patients who were unwilling or unlikely to adhere to CPAP therapy, we decided to use a one-week sham CPAP run-in phase. Sham CPAP is designed to deliver only a very low, non-therapeutic pressure to the airway, and previous clinic studies have shown similar short-term adherence levels with sham and active CPAP. We argued that if patients were sufficiently motivated and able to wear a CPAP mask for a minimum average period of 3 hours per night, they would likely comply long-term with active treatment during the trial.

The trial is led by an Executive Committee who has overall responsibility for the design and proper conduct of the study. Day-to-day operational matters are decided by an Operations Committee. A Principal Investigator (PI) for each participating country or region advises the Executive and Operations Committees on relevant national regulatory issues, clinical practice standards, and patient recruitment strategies. In China, the national PI is assisted by a China PI committee consisting of a small group of experts in sleep and respiratory medicine and stroke and cardiovascular medicine. A DSMB has regular oversight of the study and is responsible for safeguarding the interests of trial participants, assessing the safety and efficacy of the interventions during the trial, and for monitoring the overall conduct of the trial. The DSMB will undertake 2 planned interim analyses and will provide recommendations about stopping or continuing the trial to the trial Executive Committee. Further information regarding the structure and composition of the various SAVE trial committees is provided in Appendix A.

RD McEvoy (PI), CA Anderson (co-PI) , RR Grunstein, B Neal, S Redline, L Palmer, SG Huang, NS Zhong, JG Wang, J Hedner, G Lorenzi-Filho, N Ramakrishnan.

E Heeley (Chair), RD McEvoy, CA Anderson, Antic NA

GG Jennings (Chair), L Wong , G Marks , S Heritier (statistician)

NS Zhong (Chair), BY Chen, QY He, SG Haung, JG Wang, H Yining