We recruited 16 patients with COPD according to clinical history; FEV,/FVC% <0.55 (repeated measurements); and daytime PaO* below 9 kPa (67.5 mm Hg, repeated measurements), ie, low enough to be associated with a reasonable likelihood of desaturation during sleep. One patient was excluded due to intercurrent disease between the second and third study nights and one patient was analyzed separately because he had an abnormal number of sleep apneas. Thus, 14 patients remained for complete evaluation.

The patients were not allowed to use hypnotics or sedatives during the study and ten days before the start of the study. Each patient was studied on three separate nights, one week apart. Because of the presumed risk of severe hypoxemia during hypnotic-induced sleep, the placebo night was always the first, a fact that the patient and the laboratory technician were not aware of. On the two following study nights, the patient received either nitrazepam, 5 mg, or flunitrazepam, 1 mg, in a randomized crossover doubleblind fashion. This design allowed us to identify any patient with SaOz <70 percent during more than one minute (arbitrarily chosen limit) during the placebo night and to withdraw him from the rest of the study. The laboratory technician was also instructed to rouse the patient if Sa02 fell below 60 percent for more than 5 min or below 50 percent for more than 1 min during any of the nights. No patient, however, was withdrawn or roused because of these reasons. read more
Alcohol- or caffeine-containing beverages were not allowed from 10 am on study days. The patients arrived at the sleep laboratory at about 10 pm. Application of the instruments was finished at about 11 pm when measurements were started to provide a baseline reading of awake Sa02. The patients then swallowed the tablet and rested awake in bed during 15 min before they were encouraged to sleep.