I read with interest the study by Pendlebury et al 1 describing an overall worsening in mild to moderate obstructive sleep apnoea (OSA) over a mean follow up period of 17 months. However, data from my clinical centre, a university hospital sleep laboratory in a medium sized metropolitan area (population approximately 500 000) which performs over 500 polysomnograms/year, fail to confirm their findings completely.

Files were reviewed from all patients undergoing polysomnography in the University of Arizona Sleep Disorders Center from 1 January 1983 to 31 December 1996 who had two studies performed more than two years apart, and who had OSA on their initial study (apnoea/hypopnoea index (AHI) ⩾5 events/hour of total sleep time). Those who underwent surgery, such as tracheostomy or uvulopalatopharyngoplasty, to treat their sleep apnoea were excluded. Seventeen patients (14 men) met these criteria and their files were reviewed for clinical sleep symptoms, indications for studies, age at the time of initial polysomnography, sex, and weight, in addition to results of the polysomnography.

The mean (SD) age of the patients was 53 (12) years and the mean interval between studies was 6.9 (2.9) years (range 2.5–12.5). During this interval 11 patients received either no treatment, were prescribed protriptyline, or were given instructions to lose weight; four were treated with nasal continuous positive airway pressure; one was instructed on how to avoid sleeping supine; and one received nocturnal oxygen. Although there was a trend for weight to have increased between the two studies (100.7 (34.5) kg vs 103.6 (35.4) kg, p = 0.16), the change in mean initial AHI was not statistically significant (33.5 (27.7) vs 38.5 (29.2), p = 0.55). Similarly, the mean initial apnoea/hypopnoea time (AHT, min/hour total sleep time) was 13.0 (10.7) and subsequently was 13.6 (10.2) on the second study (p = 0.87).

Changes in AHI or AHT did not correlate with age, weight changes, or interval between studies. Changes in AHI, however, exhibited considerable variability within the study cohort. Using the criteria of a 25% change in AHI employed by Pendlebury et al,1 seven patients worsened, four patients improved, and six were stable. In one patient who was instructed only to lose weight, the AHI decreased from 57.5 to 0 over a five year interval with a corresponding weight loss of only 4 kg. Conversely, in another patient who was also instructed to lose weight, the AHI increased from 7.4 to 72.9 over an 11 year period.

Whether OSA is indeed a progressive condition and is associated with greater cardiovascular morbidity and mortality has important implications for the care of patients and for public health policy. These data do not confirm the findings of Pendlebury et al 1 of an overall increase in OSA with time. However, they are consistent with the variability in clinical course exhibited in their cohort. Combined with the observations of others,1-3 these data suggest that the clinical course of OSA is quite variable and that OSA is not necessarily a progressive condition. As advocated by Calverley,4 additional information as might be provided by large prospective studies such as the Sleep Heart Health Study5 is required to identify factors associated with progression and remission of OSA.