- Maurice M Ohayon, directora,
- Christian Guilleminault, professorb,
- Robert G Priest, head, academic department of psychiatryc,
- Malijai Caulet, scientista
- a Centre de Recherche Philippe Pinel de Montreal Montreal Quebec H1C 1H1 Canada
- b Stanford University School of Medicine Sleep Disorders Clinic and Research Centre Stanford California USA
- c University of London Imperial College School of Medicine at St Mary's Paterson Centre London W2 1PD
- Correspondence and reprint requests to: Dr Ohayon
- Accepted 29 November 1996
Objectives: To determine the prevalence of snoring, breathing pauses during sleep, and obstructive sleep apnoea syndrome and determine the relation between these events and sociodemographic variables, other health problems, driving accidents, and consumption of healthcare resources.
Design: Telephone interview survey directed by a previously validated computerised system (Sleep-Eval).
Setting: United Kingdom.
Subjects: 2894 women and 2078 men aged 15-100 years who formed a representative sample of the non-institutionalised population.
Main outcome measures: Interview responses.
Results: Forty per cent of the population reported snoring regularly and 3.8% reported breathing pauses during sleep. Regular snoring was significantly associated with male sex, age 25 or more, obesity, daytime sleepiness or naps, night time awakenings, consuming large amounts of caffeine, and smoking. Breathing pauses during sleep were significantly associated with obstructive airways or thyroid disease, male sex, age 35-44 years, consumption of anxiety reducing drugs, complaints of non-restorative sleep, and consultation with a doctor in the past year. The two breathing symptoms were also significantly associated with drowsiness while driving. Based on minimal criteria of the International Classification of Sleep Disorders (1990), 1.9% of the sample had obstructive sleep apnoea syndrome. In the 35-64 year age group 1.5% of women (95% confidence interval 0.8% to 2.2%) and 3.5% of men (2.4% to 4.6%) had obstructive sleep apnoea syndrome.
Conclusions: Disordered breathing during sleep is widely underdiagnosed in the United Kingdom. The condition is linked to increased use of medical resources and a greater risk of daytime sleepiness, which augments the risk of accidents. Doctors should ask patients and bed partners regularly about snoring and breathing pauses during sleep.
Disordered breathing during sleep is related to several health problems and may have important daytime repercussions
The prevalence of disordered breathing during sleep has not been well known in the United Kingdom until now
Middle aged men are at higher risk of reporting snoring, breathing pauses during sleep, or obstructive apnoea syndrome
Daytime sleepiness, poor sleep, obesity, and the use of healthcare resources are highly correlated with disordered breathing during sleep
Obstructive sleep apnoea syndrome is widely unrecognised, and consumption of healthcare resources is higher in this specific population, raising the question of the cause of this higher consumption
There have been several epidemiological studies of snoring but none has been conducted on a large representative sample of a major European population. Such investigations are of interest, as regular, heavy snoring is the most noticeable feature associated with disordered breathing during sleep and there is a known association with hypertension, cerebrovascular accidents, and coronary artery disease.1 2 3 4 5 6 We report a telephone questionnaire survey in 1994 to determine the prevalence of snoring and breathing pauses during sleep in a representative sample of the United Kingdom population.
Subjects and methods
The target population was all non-institutionalised residents of the United Kingdom aged 15 or over (roughly 45 709 600 people). A representative sample was obtained by a stratified probabilistic approach using 1991 census data to determine distribution among the 11 areas of the United Kingdom and the Kish selection method7 used to elect the person to be interviewed within each targeted household. One of eight different selection tables was randomly assigned to a household before the number was called. Based on the number of people in the household and their sex and age, the table indicated which member should be interviewed. Subjects who did not speak English, who had impaired hearing or a speech impediment, or who were too ill to be interviewed were excluded. Interviews were completed with 4972 subjects (79.6% of those approached). The highest rate of completed interviews was in Northern Ireland (86.8%) and the lowest rate in East Midlands (78.2%; 2=4.019, P<0.05).
Interviews were conducted by BPS Teleperformance, Birmingham, which specialises in large telephone surveys. Interviews were directed by the Sleep-Eval knowledge based system,8 9 a computer program designed to provide homogeneous and standardised evaluations. The system is a previously validated, non-monotonic level 2 expert system with a causal reasoning mode.10 11 The program selects the questions and displays them on a monitor. The interviewer reads each question to the subject, then enters his or her response. Expected responses vary with the questions, from a simple “yes” or “no” or “present,” “absent,” or “unknown” to answers on a five point scale or requiring use of the keyboard by the interviewer–for example, to record name and duration of illness.
The system is based on a logical reasoning module that poses questions in a manner adapted to the specific individual. Sleep-Eval pre-emptively eliminates irrelevant questions based on prior responses. For example, a subject who is completely satisfied with his or her quality of sleep will not be asked about the impact of sleep related problems.
Data from the 1991 census pertaining to the non-institutionalised population aged 15 or over were used as the standard population. The weighting procedure was adjusted for sample design and took into account the geographic distribution of the sample. The unweighted sample comprised 2894 women and 2078 men ranging from 15 to 100 years of age. After weighting, the sample consisted of 52.3% women and 47.7% men.
These calculations were performed for all variables and the results presented as weighted percentages. 95% Confidence intervals were also calculated. Univariate analyses (by 2 tests) and multivariate analyses were performed with spss statistical software. Colinearity problems between variables (that is, information redundancy) were checked. The method of indicator contrasts12 was used to determine which categories of the independent variables were significantly associated with the presence of snoring and breathing pauses during sleep. Odds ratios were calculated according to the different categories with the cut off point for significance set at 5%.
The International Classification of Sleep Disorders (1990)13 provided the criteria for identifying sleep disorders.
A total of 40.3% of the population (2004 subjects) reported snoring regularly, men more often than women, the prevalence of snoring increasing with age up to 55. Breathing pauses during sleep were reported by 3.8% of the sample (190 subjects) and also increased with age (table 1). The association of both snoring and breathing pauses was reported by 2.5% of the sample (124 subjects). A total of 7.8% of the population (386 subjects) did not know whether they snored or had breathing pauses during sleep.
Regression analysis (table 2) indicated that snoring was significantly associated with being an obese (body mass index (kg/m2) 30) married man aged 25 or more. Snoring was also significantly associated with daytime sleepiness, napping, night time awakenings, high caffeine intake (6 cups of tea or coffee daily), and smoking.
Breathing pauses were significantly associated with being a 35-44 year old man taking anxiolytics, who had been diagnosed with obstructive airways or thyroid disease, and who had consulted a doctor at least once during the past year. Table 3) gives the odds ratios. When snoring and breathing pauses during sleep were reported together there was a significant association with being an obese (body mass index 30; odds ratio 2.9 (95% confidence interval 2.3 to 3.5)) man (odds ratio 4.4 (3.9 to 4.8)) with leg pain (3.1 (2.5 to 3.8)), difficulty maintaining sleep (2.9 (2.4 to 3.4)), and usually not sleeping fully supine (4.4 (3.1 to 5.6)). The model also identified as significant variables urinary problems (3.9 (2.9 to 4.8)), high blood pressure (2.5 (1.8 to 3.1)), daytime sleepiness (2.3 (1.9 to 2.7)), and daily intake of more than six cups of caffeinated beverages (1.8 (1.2 to 2.4)).
A nationwide survey with such a large sample is based on interview responses. To evaluate further the responses to the questions we used the criteria (A + B + C) outlined in the International Classification of Sleep Disorders to define diagnoses of obstructive sleep apnoea syndrome and investigate the independent variables associated with this diagnosis. As above, the logistic regression model indicated a significant association with being an obese (body mass index 30; odds ratio 2.0 (95% confidence interval 1.4 to 2.5)) man (3.8 (3.3 to 4.2)) with difficulty maintaining sleep (4.0 (3.5 to 4.5)), daytime sleepiness (3.8 (3.3 to 4.2)), high blood pressure (2.8 (2.2 to 3.5)), presence of leg pain (2.7 (2.1 to 3.3)), and non-restorative sleep (1.9 (1.4 to 2.3)). Despite the frequency of sleep related complaints among subjects with snoring and breathing pauses only 18.2% (n=31) of the subjects with breathing pauses and 9.2% of the snorers (n=185) believed they had a sleep problem.
Our survey allowed us to evaluate the association between reports of snoring and breathing pauses and three different major healthcare related problems.
Driving accidents–In our representative sample 5.3% of drivers had an accident during the preceding year. However, there was no significant difference between snorers (4.6%), subjects with breathing pauses (6.1%), and other subjects (5.9%). Reports of falling asleep at the wheel, however, were significantly more frequent in subjects who reported breathing pauses (6.2%) and regular snoring (4.3%) than in other subjects (2.4%; 2=8.593, P<0.05).
Healthcare resource consumption–The percentage of snorers who had consulted a doctor at least once in the past 12 months did not differ significantly from that of non-snorers (62.2% v 60.2%). However, there was a significant difference between subjects with breathing pauses and those without (81.0% v 60.8%; 2=12.385, P<0.001). Also health resource consumption was significantly higher in subjects reporting breathing pauses during sleep. Thirty one per cent of subjects reporting breathing pauses had sought medical help six times or more in the past 12 months compared with only 12.0% of regular snorers and 11.9% of non-snorers (2=27.013; P<0.001). Numbers of admissions to hospital reported by 11.0% of the sample disclosed a trend only in respect of the small number of subjects with breathing pauses during sleep. Admissions were reported by 11.4% of snorers and 18.8% of subjects with breathing pauses during sleep.
Treatment for physical illness not linked by doctor to sleep related problem–At the time of interview 15.5% of subjects were being treated for a physical illness. The rate was significantly higher in subjects reporting breathing pauses during sleep (39.8%) than in snorers (16.9%) and other subjects (14.0%; 2=30.384, P<0.001). Treated or untreated hypertension was also significantly more frequently reported by subjects with breathing pauses during sleep (13.8%) than by snorers (8.6%) and other subjects (5.6%; 2=21.504, P<0.005).
Prevalence of obstructive sleep apnoea syndrome
Based on International Classification of Sleep Disorders criteria, which included daytime sleepiness, the prevalence of obstructive sleep apnoea syndrome in the sample (criteria A + B + C) was 1.9%. The prevalence of obstructive sleep apnoea syndrome in subjects aged 35-64 years was 1.5% in women (95% confidence interval 0.8% to 2.2%) and 3.5% in men (2.4% to 4.6%).
Though there have been other studies of snoring and sleep apnoea syndrome in the United Kingdom,14 15 16 so far as we know this is the first study with such a large representative sample of the general population. Some findings were expected, based on our knowledge of the pathophysiology of obstructive sleep apnoea–for example, the association between obesity or thyroid disease and breathing pauses during sleep. That obesity is a significant risk factor is not surprising given the extensive published data, and there is enough evidence to show that disordered breathing during sleep is an independent risk factor for hypertension.17 This survey re-emphasises the common association between disordered breathing during sleep and reports of disrupted nocturnal sleep, non-restorative sleep, daytime sleepiness, greater intake of caffeinated beverages, and drowsiness while driving.
A surprising finding was that the sleep disorder was as often labelled “daytime sleepiness” as “insomnia.” A diagnosis of insomnia may preclude recognition of the sleep related breathing problem and explain our finding that breathing pauses during sleep were significantly associated with anxiolytic drugs, which would be contraindicated for patients with breathing problems during sleep. The association between sleep bruxism and disordered breathing during sleep included in the logistic regression is supported by clinical observation and seems to be related to the disproportionate maxillomandibular anatomy presented by obstructive sleep apnoea patients, particularly those with family histories.18 The prevalence of obstructive sleep apnoea syndrome in our sample was much higher than the prevalence (0.3%; 95% confidence interval 0.07% to 0.9%) reported in men aged 35-65 from a survey in 1990 in Wheatley, near Oxford.19 This can be explained by the criteria used at that time, which probably identified only a severely affected population. Our results are very similar to those of Young et al in the United States (who found a prevalence of 2% in women and 4% in men)20 and Gislason et al in Iceland.21 Our study indicates that obstructive sleep apnoea syndrome is still widely unrecognised in the British Isles. It also suggests that consumption of healthcare resources may be higher in this specific population, raising the question of the cause of this higher consumption.
Finding This work was supported by a grant from the Synthelabo Group.
Conflict of interest None.