There is strong evidence from both national and international research that insomnia is highly prevalent in the general population, with a significant adverse impact on physical health, mental health and quality of life, Reference Waterman and Selsick1 and a high economic impact. Reference Wickwire, Shaya and Scharf2 For example, insomnia has repeatedly been shown to increase the risk of developing new mental illnesses (such as depression) in people with no psychiatric history, and to reduce the efficacy of treatment of comorbid mental disorders, exacerbate their severity and precipitate relapses. Reference O’Regan and Selsick3 Whereas insomnia used to be mainly considered a consequence of disorders such as depression or chronic pain, it is no longer separated into ‘primary’ and ‘secondary’ insomnia due to the often bidirectional relationship between insomnia and other disorders, and the fact that research has consistently demonstrated the effectiveness of treating insomnia directly regardless of comorbidity or order in which the insomnia and comorbidity occurred. Reference Waterman and Selsick1 A significant body of high-quality research consistently demonstrates that cognitive–behavioural therapy for insomnia (CBT-I) leads to durable improvements in insomnia symptoms (around 60–80% experience clinically significant improvements), Reference Perlis, Posner, Riemann, Bastien, Teel and Thase4 as well as in a wide range of comorbid conditions. Reference Waterman and Selsick1 This is particularly important for insomnia that has become chronic, since research suggests that chronic insomnia, which is classified as a mental disorder in many diagnostic manuals, rarely remits spontaneously. Reference Fatima, Bucks, Mamun, Skinner, Rosenzweig and Leschziner5 It is also highly cost-effective Reference Wickwire, Shaya and Scharf2 with a low burden on resources, Reference Waterman and Selsick1 leading to it becoming the first-line recommended treatment by bodies such as the American College of Physicians Reference Qaseem, Kansagara, Forciea, Cooke and Denberg6 and the UK’s National Institute for Health and Clinical Excellence (NICE), which recommends it for insomnia that has become chronic or is likely to become chronic. 7 However, in many countries, including the UK, Reference Waterman and Selsick1 USA Reference Koffel, Bramoweth and Ulmer8,Reference Bhaskar, Hemavathy and Prasad9 and Western European countries, Reference Ellis, Ferini-Strambi, García-Borreguero, Heidbreder, O’Regan and Parrino10 chronic insomnia is known to be under-treated due to a paucity of available services providing CBT-I and lack of awareness of the treatment. In the UK, psychoeducation on ‘sleep hygiene’ is more commonly provided; however, research shows that this is ineffective for most people. Reference Waterman and Selsick1,Reference Irish, Kline, Gunn, Buysse and Hall11
Previous research into insomnia prevalence has mostly revolved around cross-sectional and longitudinal survey studies, Reference Bhaskar, Hemavathy and Prasad9,Reference Ohayon12 and suggests that its prevalence in England is 6–40%, Reference Calem, Bisla, Begum, Dewey, Bebbington and Brugha13 with similar findings internationally. Reference Ohayon12 This wide variation in prevalence is based upon the definitions of insomnia used, with ‘insomnia symptoms’ showing the highest prevalence and clinically diagnosed chronic insomnia at around 6% prevalence. Reference Ohayon12,Reference Calem, Bisla, Begum, Dewey, Bebbington and Brugha13 Prevalence is higher in patients receiving healthcare, with an estimated level of 69% among primary care patients. Reference Koffel, Koffel and Gehrman14 Some prior research has sought to identify risk factors associated with poor sleep, from UK Biobank data Reference Fatima, Bucks, Mamun, Skinner, Rosenzweig and Leschziner5 and longitudinal or cross-sectional surveys, Reference Morphy, Dunn, Lewis, Boardman and Croft15 and the following were associated with poor sleep: socioeconomic and lifestyle factors (including socioeconomic disadvantage, lower education level, shift work and unhealthy lifestyle), demographic factors (including minority ethnic background, increasing age and female gender) and comorbid illness (including unspecified chronic illness, pain, diabetes, obesity, hypertension, respiratory symptoms, mental illness, substance misuse and menopause). Reference Fatima, Bucks, Mamun, Skinner, Rosenzweig and Leschziner5,Reference Bhaskar, Hemavathy and Prasad9,Reference Morphy, Dunn, Lewis, Boardman and Croft15–Reference Yardi and Adsule20 However, to our knowledge, no prior study has carried out a detailed analysis of medical records to establish the real-world prevalence of insomnia, insomnia-related health inequalities or local unmet need regarding insomnia. This study’s aim was to establish these in North Central London, using routinely collected primary care data.
Method
Data extraction and criteria
In the UK, general practice (otherwise known as ‘primary care’) medical records are summarised using codes (previously by the coded system Read 21 and now by SNOMED 22 ), which can be utilised to summarise patient health and clinical interactions, and are useful for extraction of summarised data.
Coded data were extracted from general practice records across three North Central London boroughs (Camden, Islington and Haringey) in December 2021. Data were sourced from a data warehouse, operated by the North East London Commissioning Support Unit to house pseudonymised general practice data. Data were extracted for all patients aged 15+ years and registered with 102 of the 105 local general practices, encompassing 765 035 patients; 3 practices and 9% of patients from these participating practices were excluded due to opting out of data sharing. Data were aggregated at the point of extraction.
From discussions with local general practitioners (GPs), the authors are aware that GPs do not always code for insomnia when a patient presents with insomnia symptoms, perhaps particularly if that patient has also presented with a comorbid illness such as depression, when the depression may be coded instead. Preliminary analysis also supports this issue, because 2553 patients were prescribed tablets from a list of medications used only for insomnia (zopiclone, zolpidem, temazepam and melatonin) without having a code for insomnia. Therefore, the prevalence values sought are estimates.
For the purpose of this study, a patient was considered to have insomnia symptoms if they met one of three criteria:
-
criterion 1 – the presence of a code indicating insomnia;
-
criterion 2 – the presence of a code indicating insomnia treatment;
-
criterion 3 – a sedative medication recently prescribed for sleep.
For criteria 1 and 2, a search was conducted for Read and SNOMED codes containing the following strings: ‘sleep*’, ‘insomn*’, ‘sedat*’ and ‘hypnot*’. L.Z.W. (psychiatrist and insomnia specialist) selected those codes that were relevant to insomnia, and this was cross-checked by another sleep medicine specialist. Codes were excluded that indicated only hypersomnia, sleep apnoea or abnormal behaviours during sleep. A total of 53 codes were identified that indicated an insomnia diagnosis (criterion 1), and 8 codes that indicated treatment for insomnia (criterion 2) (see Appendix 1); 31 codes were excluded (see Appendix 2).
For criterion 3, data were extracted pertaining to hypnotic medications that had been prescribed, and patients were included if those medications were assumed by the authors to have been prescribed for sleep. Based on L.Z.W.’s and the sleep medicine specialist’s clinical experience of which sedative medications are commonly prescribed for sleep in the UK, the following were included:
-
(a) temazepam, zopiclone, zolpidem and melatonin (which are usually used only as sleeping tablets);
-
(b) promethazine, mirtazapine, amitriptyline and trazadone (which are often given for insomnia, for other disorders or for insomnia comorbid with other disorders).
For patients with the latter prescriptions, they were excluded (i.e. the medication was presumed not to have been prescribed for insomnia) if the patient also had a code recorded on their GP record over the past 5 years for another illness for which that medication was commonly used, as follows (for more details, see Appendix 3):
-
(a) promethazine if they also had a code for anxiety or agitation, or were on the Quality and Outcomes Framework (QOF) 23 register for severe mental illness (SMI);
-
(b) mirtazapine or trazadone if they also had a code for anxiety or were on the QOF depression register;
-
(c) amitriptyline if they were on the QOF depression register.
Two different estimates were determined: a ‘best estimate’, which determined caseness based upon patients’ histories of being coded as having insomnia, being referred for insomnia treatment or being prescribed medication for insomnia; and a ‘lowest estimate’, which excluded patients whose history of medication for insomnia was less certain. In choosing the criteria, the authors acknowledge that both may be underestimates (see Discussion). Specifically, the following criteria were used:
-
(a) ‘Lowest estimate’ prevalence values included patients who had at least one of the following:
-
(i) a code for insomnia over the past 5 years;
-
(ii) a prescription of temazepam, zopiclone, zolpidem or melatonin in the past 3 months (these are medications used only as sleeping tablets);
-
(iii) a code for CBT-I treatment/referral in the past 6 months;
-
(iv) a code for other insomnia treatment, such as sleep hygiene, in the past 5 years.
-
-
(b) ‘Best estimate’ prevalence values additionally included patients who had:
-
(i) a prescription for another medication commonly used for insomnia, without also having a code for a non-insomnia illness for which that medication is a common treatment.
-
Rationale for time frames for coded records
Coded records going back 5 years from the date of extraction were included to estimate the current prevalence of insomnia, to strike a balance between inclusion of previous diagnostic activity covering a sufficient period while also having a lower likelihood that the included patients’ insomnia had already resolved. This was also the time frame used for other disorders for which codes were extracted (relating to criterion 3 and to investigation of insomnia in patients with comorbid illness). When considering codes for insomnia treatment as an indication of current prevalence, only codes for CBT-I within the past 6 months were included, since CBT-I is highly effective, and therefore for those with a CBT-I code prior to this, many would have already received treatment and become insomnia free. Reference Perlis, Posner, Riemann, Bastien, Teel and Thase4 However, when considering codes for alternative insomnia treatments (such as sleep hygiene), 5 years was again used as the time frame because research suggests that these treatments are largely ineffective, so patients receiving them can be assumed to still have insomnia. Reference Ellis, Ferini-Strambi, García-Borreguero, Heidbreder, O’Regan and Parrino10 Sedative medications were considered if these were prescribed within the previous 3 months, because this is the maximum duration of individual prescriptions in UK primary care and, again, for its balance between not being too under- or over-inclusive.
Extraction of socioeconomic, comorbidity and demographic data
Data were extracted pertaining to body mass index and six other mental and physical health comorbidities, which were selected based on prior knowledge or suspicion of their association with insomnia, and on the feasibility of data extraction. Data relating to area-based deprivation, age, gender and ethnic group were also extracted to allow analysis of variation by these factors. These were all presented as aggregated data for analysis.
Analysis
Data were analysed using descriptive statistics routinely used by local public health teams in needs analysis. Prevalence was calculated separately for variables pertaining to demographic information and deprivation, which were compared with the mean prevalence for the sample; 95% confidence intervals were used as a test of significance.
Results
Data were extracted and analysed for 765 035 patients, as described above. Summary tables for the descriptive statistics are shown in Appendix 4.
Prevalence of insomnia
The ‘best estimate’ for insomnia prevalence was 4.3% (95% CI [4.24, 4.33]) and ‘lowest estimate’ was 3.5% (95% CI [3.45, 3.53]).
We excluded 0.5% of all patients from the list of those prescribed sleeping tablets, despite those individuals having been prescribed medication commonly used for insomnia, due to their having a comorbid condition such as depression, anxiety or SMI.
Insomnia prevalence by demographic factors and geographical deprivation
Prevalence was 4.9% for women and 3.7% for men. Prevalence increased steadily with age, being highest in those aged 85–90 years (10.8%, 95% CI [10.1, 11.1]) and lowest in those aged 15–19 years (1.4%, 95% CI [1.3, 1.5]), as shown in Fig. 1. There was significant variation by ethnic group (Fig. 2) and deprivation quintile (Fig. 3), with the highest prevalence being found in the most deprived quintile (5.2%, 95% CI [5.1, 5.3]) and those of Bangladeshi ethnicity (7.3%, 95% CI [6.9, 7.8]).
Fig. 1 ‘Best estimate’ insomnia prevalence by age group.
Fig. 2 ‘Best estimate’ insomnia prevalence by ethnicity.
Fig. 3 ‘Best estimate’ insomnia prevalence by geographical deprivation quintile.
Insomnia prevalence by long-term condition
Insomnia prevalence was significantly higher in patients with any of the six comorbidities for which data were extracted (Fig. 4): chronic obstructive pulmonary disease (COPD; 17.5%, 95% CI [16.8, 18.4]), SMI (16.6%, 95% CI [16.0, 17.3]), depression (14.1%, 95% CI [13.8, 14.3]), anxiety disorders (12.7%, 95% CI [12.5, 12.9]), diabetes mellitus (11.8%, 95% CI [11.5, 12.2]) and hypertension (11.1%, 95% CI [10.8, 11.3]). Prevalence also increased with increasing weight category (Fig. 5), with overweight patients having a higher prevalence (5.5%, 95% CI [5.4, 5.6]) than the mean, and a yet higher prevalence for those in the obese category (8.8%, 95% CI [8.6, 9.0]).
Fig. 4 ‘Best estimate’ insomnia prevalence by long-term condition. QOF, Quality and Outcomes Framework; SMI, severe mental illness; COPD, chronic obstructive pulmonary disease.
Fig. 5 ‘Best estimate’ insomnia prevalence by body mass index (BMI).
Treatment rates
It was shown that 1.7% (95% CI [1.2, 2.3]) of people with a code or medication indicating insomnia over the previous 6 months had been referred for CBT-I treatment during that time period, and 5.6% (95% CI [4.7, 6.7]) had a code for another intervention such as sleep hygiene during that time period.
Variation among general practices
Significant variation in insomnia prevalence occurred between general practices. The highest ‘best estimate’ prevalence was found for the Camden Health Improvement Practice (9.9%, 95% CI [7.5, 12.9]), which is the practice supporting Camden’s homeless population (the other two boroughs do not have a dedicated practice for their homeless populations).
Discussion
The ‘best estimate’ of insomnia prevalence from North Central London’s primary care data was 4.3%, with the ‘lowest estimate’ not being much lower (3.5%). It is possible that the inclusion of patients based on sedative medications that could also have been used to treat other disorders (such as mirtazapine and promethazine), even when excluding those with certain comorbid conditions, risked the inclusion of some patients who did not have insomnia. Therefore, the ‘lowest estimate’ excluded this criterion within the case definition, to check the extent to which that criterion had impacted the results. The difference between the best and ‘lowest estimate’ was 0.8%, demonstrating that the prevalence was still high even following removal of the criterion of having a sedating medication prescription. As described below, there are multiple reasons why both may actually be underestimates. These findings suggest that insomnia is as common as other illnesses that receive a high share of focus and healthcare resources in many countries.
There was significant variation in prevalence based on the demographic factors investigated. The higher prevalence in women and steady increase with age are consistent with prior research. Reference Bhaskar, Hemavathy and Prasad9,Reference Ohayon12,Reference Morphy, Dunn, Lewis, Boardman and Croft15–Reference Roy, Bhattacharjee, Chakraborti and Singh19 Given that insomnia rarely improves spontaneously without treatment once it has become chronic, Reference Fatima, Bucks, Mamun, Skinner, Rosenzweig and Leschziner5 and that it is under-treated (both within this sample and nationally Reference Waterman and Selsick1 ), it is understandable that the proportion of people with insomnia would gradually accumulate with increasing age. Increasing rates of insomnia with age may also relate to increasing rates of comorbid illness with age, which warrants further research. Some ethnic groups (especially Bangladeshi, White Irish and Black Caribbean), and those living in the most deprived areas, had a significantly higher prevalence of insomnia compared with the mean. This may reflect health inequalities relating to differences in the prevalence of insomnia, differences in GP access/help-seeking behaviour or the influence of patients’ demographic or socioeconomic factors on how GPs diagnose and code. The Bangladeshi community in this area is known to have generally poor health compared with other groups. 24 This variation in prevalence by sociodemographic factors (including increasing with sociodemographic disadvantage and age, and being higher in women) is consistent with the broader international literature, suggesting patterns similar to those seen in other countries. Reference Morin, Bei, Bjorvatn, Poyares, Spiegelhalder and Wing25
Prevalence was significantly higher in patients with the seven comorbid mental and physical health conditions for which data were extracted (COPD, diabetes mellitus, hypertension, depressive disorders, anxiety disorders, SMI and obesity), consistent with prior research, Reference Fatima, Bucks, Mamun, Skinner, Rosenzweig and Leschziner5,Reference Bhaskar, Hemavathy and Prasad9,Reference Morphy, Dunn, Lewis, Boardman and Croft15,Reference Dodge, Cline and Quan16,Reference Yardi and Adsule20 with the highest prevalence being in those with COPD and SMI. It is likely that other chronic illnesses may also be associated with a high prevalence of insomnia, as well as risk factors such as alcohol misuse. This highlights the importance of actively screening for insomnia in patients with various comorbid disorders. Prior research has suggested that untreated comorbid insomnia renders mental disorders more resistant to treatment and at a higher risk of relapse. Reference O’Regan and Selsick3 Therefore, particular attention should be paid to screening for, and treating, insomnia in those with mental illness.
Referral rates for CBT-I were low (1.7% for those indicated as having insomnia). It is likely that patients with insomnia who have not been referred for treatment would be less likely to have received an insomnia code, so the proportion of insomnia treated may be even lower. This is concerning, given that CBT-I is the NICE-recommended first-line treatment for insomnia in the UK 7 and the negative health, quality of life and economic implications of untreated insomnia. Reference Waterman and Selsick1 From local knowledge, this is probably due to a combination of limited service availability, unclear referral pathways and competing demands in primary care settings. These findings should be interpreted in the context of the significant pressures facing primary care, where clinicians manage multiple complex needs within limited time. This is also consistent with previous studies conducted in other countries, which highlight the lack of availability and accessibility of CBT-I treatment. Reference Morin, Bei, Bjorvatn, Poyares, Spiegelhalder and Wing25
Of those who were referred for therapy, were prescribed sleeping tablets or had other treatments for insomnia (such as sleep hygiene), many did not have a diagnosis code for insomnia, which may reflect variation in clinical coding practices and system-level documentation processes. Furthermore, there was significant variation in the apparent prevalence between practices. This may reflect true variation in local prevalence and/or differences in coding and management approaches between practices.
Strengths and limitations
A limitation of using GP data to determine disease prevalence is that this relies both on patients presenting to their GP when they have symptoms and on GPs diagnosing and coding problems correctly, and presentation may vary across different patient groups. In this study, all prevalence estimates may be underestimates because not all those with insomnia would have presented to their GP Reference Ohayon12,Reference Liu, Zhang, Lam, Yu, Li and Zhou26 (and many may instead be using self-help or over-the-counter treatments); Reference Liu, Zhang, Lam, Yu, Li and Zhou26 in addition, it is likely that insomnia is under-coded and under-treated in routine primary care, reflecting both service limitations and the many competing demands faced by clinicians. Additionally, prevalence estimates excluded those without an insomnia code who were prescribed sleeping tablets but had these stopped within the past 3 months; those who have died or moved out of the area since presenting with insomnia; and those taking less commonly prescribed sleeping tablets (e.g. diazepam). Furthermore, they excluded people for whom sedative medication was prescribed that could have been intended for a different mental illness (0.8%), even though insomnia commonly exists alongside other mental illnesses (meaning that many of these excluded patients may have actually been prescribed those medications for insomnia). Another limitation of this study is that free text could not be accessed to confirm prevalence, and patients were counted as having insomnia only if this could be identified from coded data or prescription information. This restricted insight into the severity or duration of patients’ insomnia symptoms, or whether they met criteria for a chronic insomnia diagnosis, and therefore whether CBT-I would have been indicated. The availability of solely aggregated data limited the complexity of the analysis that could be carried out, including the possibility of any regression analysis.
However, this study also has a number of strengths. It analysed a very substantial data-set across three large boroughs – the largest study the authors have identified that investigated insomnia prevalence or insomnia-related health inequality. With a large sample size also comes higher generalisability to the rest of the UK and international populations. To partially overcome the challenges associated with determining insomnia prevalence from the data, two different prevalence estimates were provided (‘best estimate’ and ‘lowest estimate’), which demonstrated that even when most sedative medications were removed from case-finding, the prevalence was found to be high.
Recommendations
The high prevalence and low treatment rates found in this study demonstrate the importance of increasing treatment rates. Under-treatment of insomnia is likely to be a national problem, Reference Waterman and Selsick1 and has also been reported in other countries; Reference Koffel, Bramoweth and Ulmer8–Reference Ellis, Ferini-Strambi, García-Borreguero, Heidbreder, O’Regan and Parrino10,Reference Morin, Bei, Bjorvatn, Poyares, Spiegelhalder and Wing25 therefore, the recommendations given below are likely to be relevant outside of North Central London. Reference Koffel, Bramoweth and Ulmer8–Reference Ellis, Ferini-Strambi, García-Borreguero, Heidbreder, O’Regan and Parrino10 Given the health economic impact of untreated insomnia and the cost-effectiveness of CBT-I, allocating resources to increasing effective treatment is warranted to save money elsewhere. Reference Wickwire, Shaya and Scharf2
Increasing treatment rates requires a combination of a greater number of referrals for appropriate treatment and greater availability of accessible treatment services. While CBT-I is currently available to all patients in the UK via remote sessions from a small number of sleep clinics that can be accessed regardless of address, 27,28 service capacity will need to be expanded to accommodate referral numbers as they increase. Currently, some National Health Service (NHS) Talking Therapy (formally called ‘IAPT’) services provide CBT-I, but this is a postcode lottery since insomnia is not included in the NHS Talking Therapies Manual of national service requirements. 29
As such, the following measures are recommended.
Training
-
(a) Mental health and primary care clinicians may benefit from attending training courses or webinars on insomnia management (which may not always have been included in medical and specialty training curricula), and this topic should be added to curricula in the future. Brief strategies from CBT-I can then be relayed to patients during their consultations, while they are awaiting formal CBT-I.
-
(b) To expand the availability of trained CBT-I therapists, healthcare professionals working in primary care and mental health services should be encouraged to undergo more comprehensive training to deliver formal CBT-I, although this training is still not time-intensive: for example, some UK sleep clinics run 1- to 3-day courses. Reference Waterman and Selsick1 Non-clinicians, such as psychological well-being practitioners (PWPs) working in NHS Talking Therapy services, could also be trained to deliver CBT-I under the supervision of qualified therapists, as has been done successfully previously. Reference Cape, Leibowitz, Whittington, Espie and Pilling30
-
(c) Facilitating clinician familiarity with CBT-I services and local referral pathways may help strengthen treatment delivery in routine care. Reference Waterman and Selsick1
Detection of chronic insomnia
-
(a) Clinicians in primary and secondary care should actively enquire about insomnia, especially in regard to those with advancing age, socioeconomic deprivation, mental illness, obesity, diabetes mellitus and respiratory disease – groups for which the prevalence was shown to be up to 20% in this study. This active enquiry is important, since research suggests that up to 40–45% of individuals with insomnia do not make their clinicians aware of this. Reference Ohayon12,Reference Liu, Zhang, Lam, Yu, Li and Zhou26
-
(b) Responses to initial screening questions, such as ‘Do you have problems with falling or staying asleep?’ and those about frequency, chronicity, daytime impairment and sleeping medication use, could then prompt the use of standardised questionnaires, such as the Insomnia Severity Index, Reference Bastien, Vallières and Morin31 and screening for comorbid sleep disorders. 32
Treatment pathways
-
(a) Clear local insomnia treatment pathways should be established, differentiating between clinicians in primary and secondary care. Following this study, a pathway was created in North Central London by an Insomnia Working Group (comprising GPs, sleep specialists and pharmacists), for which the group has given permission to be adapted and used elsewhere. 32
Chronic insomnia treatment services
-
(a) CBT-I treatment services should be commissioned so that they are accessible to patients nationally. The probable inequalities in insomnia related to factors such as deprivation and ethnicity should be acknowledged, and attention should be paid to designing and running services that facilitate equitable access, including consideration of the location and cultural appropriateness of services.
-
(b) As with other common mental disorders, a stepped-care model is recommended, Reference Waterman and Selsick1,Reference Koffel, Bramoweth and Ulmer8 such as:
-
(i) Step 1: self-help or guided self-help – for example, the Sleepio app is NICE-recommended and cost-effective 33,Reference Freeman, Sheaves, Goodwin, Yu, Nickless and Harrison34 but has high drop-out rates and is not suitable for everyone. 35 At the time of writing, the Sleepful app, 36 developed by the Clinical Sleep Research Unit at Loughborough University (UK), is freely available and recommended by some sleep experts, although it has not undergone clinical trials.
-
(ii) Step 2: therapist-led individual or group CBT-I (e.g. in talking therapies services and secondary care mental health services).
-
(iii) Step 3: referral to a specialist sleep clinic.
In a stepped-care model, patients would start off at step 1 and move up the steps if previous steps are ineffective. Only those with a clear need for specialist input would enter the pathway at a higher step, making the model more cost-effective.
-
-
(c) To enable this, chronic insomnia should be incorporated into the NHS Talking Therapies Manual of national service requirements, 29 so that CBT-I becomes a core treatment offered by all Talking Therapies services. This would help eliminate the current national postcode lottery and ensure that specialist sleep clinics are reserved for patients with more complex needs. Reference Waterman and Selsick1
-
(d) Medication for insomnia may be considered at different steps in treatment – either combined with or as a second-line alternative to CBT-I – as described in the North Central London Insomnia in Adults Clinical Pathway. 32
Population health management and coding
-
(a) Screening and management of insomnia could be supported by embedding it within population health management registries – for example, those focused on ageing well, mental illness and multimorbidity. Additional prompts, such as markers or pop-ups within GP clinical systems, may also help prompt identification and intervention.
-
(b) Efforts to support consistent coding approaches among practices may improve patient care and enable better system-wide population health management and research capabilities. The list of current insomnia-related SNOMED codes shown in Appendix 1 may provide a useful starting point; however, the range of insomnia-related codes available to GPs is extensive, and none clearly distinguish between acute and chronic insomnia – a distinction that is essential, because the treatment for chronic insomnia (a recognised mental disorder) differs from that for brief, time-limited insomnia. Reference Fatima, Bucks, Mamun, Skinner, Rosenzweig and Leschziner5,32 Therefore, coding and system prompts should be refined to support this differentiation.
Public health strategies
-
(a) There is increasing evidence that, for those with chronic insomnia, CBT-I has a role in primary, secondary and tertiary prevention of comorbid disorders, Reference Waterman and Selsick1 such as depression Reference Irwin, Carrillo, Sadeghi, Bjurstrom, Breen and Olmstead37–Reference Blom, Jernelöv, Kraepelien, Bergdahl, Jungmarker and Ankartjärn39 and asthma. Reference Luyster, Ritterband, Sereika, Buysse, Wenzel and Strollo40 Therefore, public health bodies should consider insomnia as part of Joint Strategic Needs Assessments. Insomnia should be included within national and local prevention strategies, in keeping with the UK NHS Long Term Plan. 41
Future research
-
(a) Future research that includes access to patient-level data at the analysis stage may be beneficial, especially if it were to make use of artificial intelligence, such as natural language processing, to identify and confirm cases from the free text available. Such analysis would enable more detailed exploration into how and why prevalence varies by demographic factors, deprivation and comorbidity, the timeline of when insomnia is reported in relation to comorbidity and the association between insomnia and other health conditions. Further research looking at associations between demographic or deprivation factors and diagnosis and treatment rates could establish how, and the extent to which, health inequalities have contributed towards variation between general practices.
-
(b) The current study could be repeated following implementation of some of the above recommendations, and replicated in other regions of the UK and internationally.
Data availability
The non-identifiable data were accessed as part of an National Health Service (NHS) data-sharing agreement for public health analysis, and the authors no longer have access to the disaggregated raw data that were used. However, the aggregated data used were a result of data processing carried out by U.O. and these data will be kept for a minimum of 5 years.
Acknowledgements
The authors thank all the clinical and public health experts from North Central London Integrated Care System and North Central London’s primary care network who have supported this project from inception, contributed towards the interpretation of the data and been involved in discussing and implementing recommendations for improving the diagnosis and treatment of insomnia. In particular, we thank Islington general practitioner and ‘Live Well’ clinical lead Dr Karl Roberts for all his input. L.Z.W. also thanks sleep psychiatry experts Dr David O’Regan and Dr Hugh Selsick for their ongoing contributions and mentorship.
Author contributions
L.Z.W. conceived the idea for the project, formulated the research questions and designed the study’s overall framework. She identified the specific data required for collection, planned the methodology and structured the analysis approach. L.Z.W. also conceptualised the Discussion section, including formulating recommendations based on the study findings, and took the lead in writing the initial draft of this article. U.O. extracted the data and transferred them from coded medical records to aggregated summarised tables, including breaking them down into the categories discussed in the Method section. F.O.M.H. undertook the analysis and co-created graphs and charts alongside L.Z.W. S.D. provided project supervision. All authors read and contributed towards the final draft.
Funding
S.D., on behalf of North Central London Integrated Care Board, successfully bid for funding for a Health Education England-funded Population Health Fellowship, which was awarded to L.Z.W., who undertook this project.
Declaration of interest
L.Z.W. initiated and ran a cognitive–behavioural therapy for insomnia service for existing patients at South London and Maudsley NHS Foundation Trust, but has received no extra funding or payment for doing so. Following later revisions of this article, L.Z.W. received honoraria from Idorsia for speaking at educational sessions for clinicians; L.Z.W. had no relationship with Idorsia prior to creating the North Central London Insomnia in Adults Clinical Pathway. U.O., S.D. and F.O.M.H. have no conflicts of interest to declare.
Ethical standards
Ethical approval was not required, because the primary purpose of this study was public health analysis to inform strategic needs assessment and commissioning rather than academic research, and it used routinely collected, non-identifiable data. Permission to publish the findings was granted by the Chief Medical Officer of North Central London Integrated Care Board.
Appendix 1: List of SNOMED Clinical Terms (CTs) included (and their associated Read Code Version 2)
Insomnia codes:
Insomnia treatment codes:
Appendix 2: List of SNOMED Clinical Terms (CTs) excluded (and their associated Read Code Version 2)
Appendix 3: SNOMED codes to identify comorbid mental illness
-
(a) Anxiety: a list of codes for anxiety were taken from https://clinicalcodes.rss.mhs.man.ac.uk/medcodes/article/78/. Reference Hoile, Tabet, Smith, Bremner, Cassell and Ford42
-
(b) Agitation: SNOMED Clinical Term ‘Restlessness and agitation’ (Concept Code ID: 274647009).
-
(c) Codes indicating severe mental illness, depression, chronic obstructive pulmonary disease, hypertension and diabetes mellitus were taken from the list of codes from the relevant Quality Outcomes Framework (QOF) disease registry business rules. 23
Appendix 4: Summary tables for descriptive statistics
Table 4a ‘Best estimate’ insomnia prevalence by age group
Table 4b ‘Best estimate’ insomnia prevalence by ethnicity
Table 4c ‘Best estimate’ insomnia prevalence by geographical deprivation quintile
Table 4d ‘Best estimate’ insomnia prevalence by long-term condition (LTC)
QOF, Quality and Outcomes Framework; SMI, severe mental illness; COPD, chronic obstructive pulmonary disease.
Table 4e ‘Best estimate’ insomnia prevalence by body mass index (BMI)
Table 4f ‘Best estimate’ insomnia prevalence by gender
Table 4g ‘Best estimate’ insomnia prevalence in all patients
Open access
eLetters
No eLetters have been published for this article.