We surveyed the UK distribution of five factors commonly associated with obstructive sleep apnoea (OSA) to produce an overall risk map that could be used to predict relative prevalence estimates. The weighting and mapping of selected risk factors associated with OSA highlighted significant regional variation in predicted prevalence estimates. These data provide the first attempt to systematically map the UK for OSA and identify areas where the condition is likely to be more prevalent. The data show a significant mismatch between areas identified as having a high predicted prevalence estimate and the distribution of existing sleep centres.
- Sleep apnoea
Statistics from Altmetric.com
What is already known on this subject
It has been established that obstructive sleep apnoea (OSA) is frequently associated with obesity, diabetes, age, hypertension and male gender. However, in contrast to those five associated factors, no systematic nationwide registry has collected data to identify regional prevalence estimates for OSA in the UK.
What this study adds
This systematic analysis of five factors commonly associated with OSA maps and identifies, for the first time, relative predicted prevalence estimates for the whole UK. This dataset identifies regional variations and inequalities accessing local sleep services, emphasising the importance of developing future strategies to deliver healthcare and sleep service provision nationwide.
Obstructive sleep apnoea (OSA) is a major healthcare challenge in the UK. It is affecting an estimated 4% of middle-aged men, 2% of middle-aged women, and its strong association with obesity means that the prevalence has risen significantly over the last two decades.1 ,2 Similarly, the increased incidence of OSA with older age means the ageing population is also likely to increase the burden of OSA.
OSA is frequently associated with diabetes, cardiovascular disease, hypertension and depression, and there is an increased risk of road traffic accidents associated with sleepiness. OSA can impact significantly on quality of life and public health, and cause significant healthcare costs, particularly when the diagnosis is delayed. The efficient diagnosis and treatment of OSA should therefore form an important part of any public health strategy. Despite this, however, few of the prevalence data required for effective OSA management planning exist, and around 80% of OSA cases remain undiagnosed and untreated.
Predicting areas in which OSA is likely to be highly prevalent could help local health authorities formulate strategies to minimise these impacts, by allowing the comparison of actual diagnosis rates with predicted prevalence estimates to ensure that awareness campaigns and service provision are sufficient to meet the predicted needs of the local population.
To do this, we aimed to systematically survey and analyse five factors commonly associated with OSA—obesity, gender, age, hypertension and diabetes—in order to plot a map of relative predicted prevalence estimates for OSA across the UK. By overlaying these data with details of current available sleep services, we hypothesised that we would be able to compare predictions of OSA prevalence estimates with the distribution of current sleep service provision, thereby providing vital information to health authorities for the development and implementation of future healthcare strategies.
Patients and methods
In 2012 the British Lung Foundation (BLF) commissioned a specialist mapping company (Lovell Johns, UK) to create a set of maps showing the location of known sleep services in the UK and highlighting areas of potential high OSA risk, or ‘hotspots’. Anticipating that the maps would be of greatest value to local healthcare commissioners and policymakers, the BLF requested that the maps be divided along local health authority boundaries (NHS Health Boards (HBs) in Scotland, NHS Local Health Boards (LHBs) in Wales, NHS Health and Social Care Trusts (HSCTs) in Northern Ireland and NHS Clinical Commissioning Groups (CCGs) in England, with predicted CCG boundaries used where final boundaries were yet to be agreed).
To identify the most appropriate data with which to develop the maps, advice was sought from key experts in sleep medicine through the BLF's OSA Advisory Group. Following peer group input and an item reduction phase, the five most common factors associated with OSA were identified as:
Obesity: prevalence of sleep apnoea is directly correlated with obesity levels.
Type 2 diabetes: recent reports have indicated that the majority of these patients have OSA.
Age: OSA is more prevalent in older people.
Hypertension: hypertension can be caused by OSA.
Sex: OSA is more common in men than in women.
Results were stated as mean (SD) for normally distributed data and as median (IQR) for non-normally distributed data, unless otherwise indicated. Comparisons of data were made using unpaired t tests and correlation was described using the Pearson coefficient. Results with p<0.05 were considered to be of significance.
We gathered data for 239 NHS administrative areas in the UK (213 CCGs in England, 14 HBs in Scotland, 7 LHBs in Wales and 5 HSCTs in Northern Ireland) with a mean population of 261 604 (155 614), ranging from 20 110 (Orkney) to 1 203 870 (Greater Glasgow & Clyde).
There was significant regional variation in predictive prevalence estimates for OSA (figure 1). Areas of relatively high predicted prevalence estimates were Wales, the North East and large parts of East Anglia and Lincolnshire. Large urban areas in England and Scotland and a number of counties around London were among the areas with the lowest predicted prevalence estimates. Although age did not correlate with diabetes (R2=0.078) or obesity (R2=0.097), there was a close relationship between age and hypertension (R2=0.753, p<0.001).
The total number of identified sleep units was 289, with only 50 centres offering polysomnography (confirmed) and the other centres offering limited respiratory studies, pulse oximetry screening or continuous positive airway pressure. In 76 centres it was not possible to determine which diagnostic modality was offered. There was a wide range of available sleep centres per CCG health area, ranging from no sleep centres in 66 CCG health areas to 9 in one large urban area; on average, there were 1.2 (1.3) centres per health authority area. People in areas without sleep centres but who lived close to an urban area had a shorter travel distance to access sleep services. Others, specifically more rural areas or islands, had far greater distances to travel to the nearest available sleep service.
This research systematically analysed five factors commonly associated with OSA for the whole UK to identify and predict, for the first time, estimated OSA prevalence throughout the country. The maps reveal regional variations that are markedly different from those for obesity and economic deprivation.
The results reveal a concerning mismatch between predicted healthcare requirements and service delivery. Furthermore, most UK sleep centres are established in large urban areas characterised by younger populations that suffer from limited comorbidities. Patients in non-urban areas, close to large centres, are linked with sleep centres by existing infrastructure and public transport. However, patients in more remote rural areas face the problem of long travel distances to their nearest centre.
‘Hotspots’ of potentially high prevalence have, in the past, been similarly identified for conditions other than sleep disorders. Such exercises have seen NHS healthcare delivery strategies adjusted accordingly to try and address the healthcare needs identified. However, sleep medicine is still evolving and, despite an increasingly obese and ageing population with an intensifying likelihood of developing comorbidities such as diabetes and hypertension, the UK's need for a nationwide delivery strategy for sleep service provision has not yet been adequately addressed. The current provision of sleep services in the UK does not meet the healthcare requirements of the population and varies remarkably.
In England, the number of diagnostic sleep studies per 1000 population varies remarkably from 0.1 to 7.8 (60-fold; this compares with 10–780 sleep studies per 100 000 population).3 Although the number of sleep studies in the UK has increased in most of the health areas since 2004,4 this still represents the highest variation in service provision in any branch of respiratory healthcare, and is a reflection of both a low rate of symptom recognition and referral from primary care, and a lack of services to which patients can be referred. It indicates a need to improve understanding of the prevalence of OSA, raise awareness of the healthcare burden and consequences of the condition, assess the capacity of sleep services and promote alternative strategies to deliver healthcare for sleep conditions in the UK.
Polysomnography is available in 50 UK sleep centres, which is equivalent to one centre for every 1.25 million citizens. The number of citizens per sleep centre is around 10 times higher than in the USA where 2461 accredited sleep centres offering polysomnography cater for a population of just over 310 million.5
Limitations of the study
The data for the five factors associated with OSA used in this analysis might not be complete. However, a GP register was available nationwide for several conditions including diabetes and hypertension, but could be limited by the access to the service. Age, gender and obesity distribution were reliably populated from statistical surveys. Our data are therefore consistent with current nationwide datasets and are the most complete compendium.
The collated dataset of UK sleep centres is the best collection currently available. However, this was not an exhaustive survey of sleep centres and will reflect omissions from input sources. The data on level of service may be inconsistent due to the differing definitions of services used by these sources and will require permanent updates.
A significant challenge when determining estimates for predicted prevalence was to develop a robust method for combining the identified risk factors into an appropriately weighted cumulative ‘factor score’. Although the weighting of different variables is a valid statistical method, there is patchy evidence for the presumed difference in the importance of each factor for the development of OSA; the weighting method was devised on expert advice rather than on conclusive evidence. Similarly, the fact that some diseases such as hypertension and diabetes are not independent of age or obesity complicates the matter.
These data identify significant variations and inequalities in provision of sleep services within the UK. Remote areas might benefit from establishing outreach clinics from network provision of healthcare with tertiary services, national awareness campaigns and, potentially, the use of modern media and systematic screening to avoid delays in diagnosis and treatment of a highly prevalent disorder.
The authors acknowledge the input of the British Lung Foundation OSA Advisory Group in advising on the identification and weighting of the five factors used to estimate predicted OSA prevalence in this exercise (please see online supplement for more details on the advisory group). The authors gratefully acknowledge the input of Dr Abdel Douiri, Senior Lecturer in Medical Statistics, King’s College London.
Contributors Conception and design, or analysis and interpretation of data: JS, IJ, AM, JH, DP, AW. Drafting the article or revising it critically for important intellectual content: JS, AM, JH, JH, AW. Final approval of the version to be published: JS, IJ, AM, JH, DP, AW. Guarantor: JS.
Funding British Lung Foundation (Charity).
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.