G HOW SHOULD WE ORGANISE SCREENING FOR DIABETIC RETINOPATHY?
The Health Technology Board for Scotland (HTBS) was established as a Special Health Board on 1 April 2000 under the terms of the National Health Service Scotland (Scotland) Act 1978. The Scottish Executive created HTBS as an “arm’s length body”, ensuring that HTBS would have the necessary freedom to establish its credibility, authority and effectiveness. HTBS provides an agreed way of assessing and introducing health technologies to improve patient’s care as outlined in “Designed to care”.3 As the single source of authoritative advice on the clinical and cost effectiveness of new and existing health technologies, it equips NHSScotland to make sound judgements about value for money.
The first three assessments selected by HTBS were:
- Treatments for Alcohol Dependency
The 1998 Scottish Health Survey indicates that 12% of men and 5% of women are ‘problem drinkers’.
- Diabetic Retinopathy
The assessment of diabetic retinopathy supports the plans of NHSScotland to establish a national screening strategy to raise the standard of diabetes care.
- Positron Emission Tomography (PET) scanning
PET scanning may be beneficial in the management of cancer, specifically lung cancer, colorectal cancer and lymphoma.
Assessing PET scanning is in line with NHSScotland’s action plan, Our National Health, which identifies the need for early detection and the subsequent need to improve cancer screening programmes throughout Scotland.
1 The Health Technology Board for Scotland Advice on the Organisation of Diabetic Retinopathy Screening Programmes in Scotland 2002
In 2002, the Health Technology Board for Scotland published its health technology assessment for organisation of services for diabetic retinopathy screening.
THE PURPOSE OF THE SCOTTISH NATIONAL SCREENING PROGRAMME FOR DIABETIC RETINOPATHY
The second report of the UK NSC (UK NSC, 2000) defines screening as:
“A public health service in which members of a defined population, who do not necessarily perceive they are at risk of, or are already affected by, a disease or its complications, are asked a question or offered a test, to identify those individuals who are more likely to be helped than harmed by further tests or treatment in order to reduce the risk of a disease or its complications.”
Clearly, diabetic retinopathy screening accords well within this definition. For the national diabetic retinopathy screening programme in Scotland the goal is to reduce the rate of avoidable visual loss from diabetic retinopathy using an effective and efficient, quality assured process that takes account of patients’ needs and preferences.
The objectives of the Scottish screening programme may be stated quite simply as:
Primary objective
The detection of referable (potentially sight-threatening) retinopathy so that it can be treated.
Secondary objective
The detection of lesser degrees of diabetic retinopathy. This can have implications for the medical management of people with diabetes in terms of blood pressure and glycaemic control, important risk factors for STDR.
It should be noted that sensitivity for the detection of the earliest features of retinopathy (i.e. any retinopathy) might be low in the screening programme and so regular checks are essential.
GENERAL EYE EXAMINATIONS
In the fiscal year 2000/2001, Scottish health service statistics showed that optometrists performed 35,347 general eye examinations on people with diabetes.4 This is only 20% of the estimated diabetic population.
Diabetic retinopathy screening will not obviate the need for a regular general eye examination to monitor changes in refraction and to detect other eye diseases. However, if patients are undergoing annual eye screening, it may be possible to increase the time between general eye examinations to more than one year. The College of Optometrists guidelines (2001b) on this should be followed.
EXECUTIVE SUMMARY, HEALTH TECHNOLOGY ASSESSMENT REPORT 1 2002
Background to Diabetic Retinopathy Screening and this Assessment
The Health Technology Board for Scotland (HTBS) estimates that there are approximately 150,000 people with diabetes in Scotland.
Approximately 5-10% of all people with diabetes have sight-threatening retinopathy.5 Diabetic retinopathy is the biggest single cause of blindness and visual impairment in Scotland among people of working age. The rising prevalence of diabetes means that it will remain a major health and economic problem in Scotland.
The personal and social costs of blindness in terms of higher liability to dependence, potential loss of earning capacity, and increased likelihood of greater social support needs, are significant for individuals, for the caring services and for society.
In its early stages, diabetic retinopathy is symptom free and progression of disease can be prevented by laser treatment; so early detection by regular screening is beneficial.
A comprehensive survey by HTBS has shown that there is wide variation in the provision of diabetic retinopathy screening across Scotland. No National Health Service (NHS) Board has all the components in place to undertake quality assured population screening for diabetic retinopathy. Initiatives are underway to establish the screening service, but these are generally at early stages of development.
Our National Health: a plan for action, a plan for change6 recognised that NHS Scotland should create a national screening strategy for diabetic retinopathy. The Scottish Diabetes Framework7 recognises eye care as one of its most urgent priorities and declares the target that all people with diabetes will have their eye status (retinopathy) recorded on the local diabetes register by September 2003.
This HTBS Health Technology Assessment aims to determine the most effective and efficient approach to achieving, implementing and sustaining a quality assured, national screening programme for diabetic retinopathy that takes account of patient requirements.
Health Technology Assessment Evidence
This Health Technology Assessment used systematic literature searching to identify evidence published in scientific literature. It also used evidence submitted by professional groups, patient groups, manufacturers, other interested parties and experts. It also undertook primary research with patients to elicit their views and preferences.
For clinical effectiveness, guidelines produced by the National Institute for Clinical Excellence (NICE), the Scottish Intercollegiate Guidelines Network (SIGN) and the UK National Screening Committee (UK NSC) were used as the starting point for assessment. Additional relevant studies were identified and added to the overall analysis.
The patient issues component used published scientific literature, educational materials from patient groups, patient surveys, discussions at the HTBS patient workshops and focus group work undertaken by HTBS in Scotland.
For the economic evaluation, information was obtained from existing UK diabetic retinopathy screening programmes and a comprehensive systematic literature review was carried out. A large number of scientific, technical and medical databases and the websites of key economic research groups in the UK, and abroad, were searched. From this, 1,388 citations were identified initially as being of possible use. Full text versions of 114 were obtained. 27 of these were used to inform modelling and 3 provided significant information for the model.
Clinical Effectiveness
Studies that screened people with diabetes and used a Gold Standard of seven-field stereoscopic photography or biomicroscopy by ophthalmologists were evaluated in detail for the clinical effectiveness analyses.
Failure rates were summarised according to individual study specifications.
Sensitivity and specificity estimates were combined for relevant combinations of screening technology/operator using receiver operating curves. Sensitivities were directly compared by standardising to a uniform value of specificity.
Meta-analysis was performed by modality/operator on all studies that fulfilled prestated selection criteria.
There are two main approaches to screening for diabetic retinopathy: ophthalmoscopy and biomicroscopy (slit lamps) or retinal photography with subsequent grading. All sensitivities and specificities were calculated for the detection of referable (sight-threatening) retinopathy.
Direct ophthalmoscopy does not achieve sufficient sensitivity to act as a screening test for sight-threatening disease and therefore should not be the basis of the Scottish national programme. However, it may be used for opportunistic screening in those who persistently default from the systematic national programme.
Indirect ophthalmoscopy (biomicroscopy) using a slit lamp has been shown to be sensitive and specific enough to be viable as a model for a national screening programme when used by appropriately trained individuals. It carries the disadvantage that there is no permanent record of the image for quality assurance or for monitoring progressive changes. However, biomicroscopy will be essential for screening failures from other modalities.
Retinal photography, with one or two fields (photographs), has been shown to achieve high sensitivity and specificity for sight-threatening disease. Advantages of digital photography are ease of image acquisition and storage, and quality assurance. The image may be transmitted electronically, facilitating external quality assurance. Consequently, digital retinal photography is the screening modality of choice.
Some eye pupils are small and need to be dilated with eye drops (mydriasis) before screening is performed. Furthermore, if more than one image per eye is required mydriasis is essential because of constriction of the pupil caused by the first photographic flash.
This HTBS Health Technology Assessment has found no clear evidence that mydriasis or the routine use of more than one image significantly alters the sensitivity or specificity of screening for the detection of referable (sight-threatening) retinopathy.
Studies using older (not digital) retinal cameras indicate that the proportion of unusable images is probably slightly lower when mydriasis is used and new studies suggest that this also applies to digital retinal cameras.
HTBS considered the evidence on imaging failure rates with and without mydriasis. In the most recent study of digital cameras, the failure rate was 20%. This is judged to be acceptable in the context of the three-stage failsafe procedure being recommended.
Studies canvassing patient opinion have suggested that mydriasis may reduce attendance for retinopathy screening because of its temporary effects on vision.
If mydriasis is used, tropicamide is the recommended agent. It must be administered by a professional complying with the Patient Group Directions (section 3.5.3) or on a named patient basis. The possible effects of the mydriatic agent should be clearly communicated to patients.
The ultra wide field scanning laser ophthalmoscope. It is a form of scanning laser ophthalmoscope with a field of view of 200 degrees internal angle which does not require mydriasis. Due to the paucity of evidence available on the use of this technology in diabetic retinopathy screening it is not recommended for use in the national screening programme until its technical failure rate and accuracy can be reliably determined.
Organisational Issues
The main organisational features of the proposed national screening programme in Scotland are:
- strong quality assurance mechanisms;
- systematic call/recall of all eligible patients (see point 6);
- trained professionals;
- recorded outcomes and robust quality assurance;
- integration with the overall process of care for those with diabetes; and
- evaluation and research as an integral part of the programme.
The national screening programme must be organised within current health service structures in Scotland, under the auspices of the National Services Division (NSD), who will ensure a consistent co-ordinated approach to the implementation of the national programme according to national quality standards.
Diabetic retinopathy screening is just one component of diabetes care and to be effective, the national screening programme must be integrated with routine diabetes care as outlined in the Scottish Diabetes Framework. Tight glycaemic control and careful blood pressure control both reduce the development and progression of diabetic retinopathy in type 1 and type 2 diabetes. Clinicians responsible for ongoing diabetic care must be fully informed of results, not only of sight-threatening retinopathy requiring referral to the ophthalmologist but also of any retinopathy.
The NHS Quality Improvement Scotland (NHSQIS) will work with the NSD to develop quality standards for this national screening programme, using generic screening standards being developed for other national screening programmes in Scotland, with additional items specific to diabetic retinopathy.
NHS Boards will have responsibility for implementation of the programme in their area to meet the needs of local people and for the monitoring of screening performance.
All patients diagnosed with either type 1 or type 2 diabetes mellitus and aged over 12 years, or post-puberty should have annual examinations of the retina.
No upper age limit is suggested, but those who are already undergoing regular reviews by an ophthalmologist, those who are medically unfit to receive laser treatment (as determined by their general practitioner) or who are completely blind, will not benefit from screening. However, for those under the care of an ophthalmologist, it is essential that the specialist retinal examinations are fully integrated into the medical record and call/recall systems for screening.
In the medium term, a fully integrated call/recall system will be developed as part of the national Information Management and Technology (IM&T) system Scottish Care Information – Diabetes Collaboration (SCI-DC). In the short-term, local systems should be designed to complement this emerging system.
Screening using higher resolution digital cameras (1,365 x 1,000 pixels) is recommended, with images graded at capture resolution (i.e. not compressed). Image transfer should use a direct digital route to avoid degradation of quality. The image should be graded on a computer with a 19 or 21 inch cathode ray tube (CRT) monitor.
Any suitably trained, accredited and competent professional (diabetologist, ophthalmologist, optometrist or retinal screener) can grade the digital images, supported by second opinions, if necessary, from ophthalmologists and/or diabetologists. The same staff may be used for both grading and screening given suitable training for both roles.
A training, accreditation and continuing education programme is being developed for health professionals, to accompany this national screening programme. It aims to achieve uniform high standards, whatever the background of the health professional. It was piloted in two Board areas in Scotland in spring 2002 and should be used as a framework for all training in Scotland.
A standard grading nomenclature for diabetic retinopathy is essential for consistent grading and so the newly defined Scottish Diabetic Retinopathy Grading System should be used.
Those patients who have sight-threatening retinopathy should be referred to a specialist eye clinic at the most convenient ophthalmology department and treated according to the Royal College of Ophthalmologists guideline’s.
The clinical IM&T functions of the retinopathy screening programme should be consistent with the national Information Technology (IT) system for diabetes care that is being established in Scotland. Furthermore, the screening result and image should be incorporated into the computerised medical record.
Optometrists are well suited to be part of the national screening programme; for the first and second level grading and screening with digital retinal cameras or for slit lamp screening of those not amenable to digital cameras. Optometrists should be linked with the national IM&T systems (e.g. NHSnet) and must adhere to the national quality assurance processes.
Patient Issues
The individuals involved in this screening programme are unlike those involved in most other screening programmes because they are already undergoing routine medical care for their condition. Also, unlike other screening programmes, patients are of both sexes, come from a wide age range and there is a higher prevalence in some ethnic minorities.
In common with allaspects of diabetes, patients must be empowered to help manage their disease; this requires support from, and collaboration between, clinical and patient organisations.
Diabetic retinopathy screening will be just one component of the annual screening programme for people with diabetes. Consequently, the goal should be to synchronise this screening visit with other healthcare visits. However, this will need to be balanced against local capacity and the recognition that some patients would like a choice of venue and appointment time.
Patient preferences for diabetic retinopathy screening include a desire for clear, timely information about all aspects of screening, choice of screening venue and appointment time and a desire to be treated as an individual, rather than just ‘another eye’.
Patient research has revealed that mydriasis is an undesirable feature of screening and some individuals would not return for screening if mydriasis was used.
Patients should be informed of the possible need for mydriasis and its effects before attending the screening visit. It should be clearly explained that there will be an increased sensitivity to light and that driving is not recommended for at least two hours after mydriasis, but that effects may last longer in some individuals.
A variety of methods should be used to inform patients about the screening process and to encourage screening attendance. It will be particularly important to encourage those who have never previously attended screening to do so and to prepare specific approaches to address all groups of patients (e.g. the young, those from ethnic minorities, etc.).
Research is required to determine the most efficient methods to increase screening uptake. For example, it has been shown that the issuing of more than two written reminders is not effective, but that contact with a health professional may help to overcome fears associated with screening.
Appointment cards and patient information should be available in accessible formats (large print, disk, audio).
General practitioners and patients should be informed of results in a timely fashion. The timeframe should be agreed at the outset of the national programme as one of the quality standards.
Economic Evaluation
The economic evaluation uses cost minimisation analysis to calculate the cost per screen for seven possible screening options. Five options of these assume a mydriatic national screening programme whilst the other two options are based on a non-mydriatic programme. The options also vary by location (hospital, GP’s surgery or within a modified van) and by staff numbers (single or double staffing for the mydriasis options).
An economic evaluation has been used to model differences in the outcomes of two possible screening programmes on the existing population with diabetes. The model compares the effectiveness, measured in mean increase in sight days, of moving to either a non-mydriatic or a mydriatic national screening programme.
The utility gains, in terms of mean sight years, are expressed as quality adjusted life years (QALY), using values reported in the NSC report. Combining the QALY data with the costing information yields a cost per QALY of moving from an opportunistic programme to a national screening programme based on either mydriasis or non-mydriasis. The analysis also shows the cost per QALY of moving from a systematic non-mydriatic based programme to a systematic mydriatic-based national screening programme.
Costings are based on information obtained from several large area diabetic retinopathy screening programmes within the UK. This allows comparison between the seven possible screening options. The costings assume a working year of 200 days during which units are operational, with an unfilled slot percentage of 5%. Staff costs take a midpoint from the relevant NHSScotland salary scale. Capital costs are from manufacturer quotes and have been annualised at a 6.0% real discount rate. There is no clear evidence on how patient travel and attendance costs vary between screening options. It has thus been assumed that average patient costs are the same for each option.
The final stage in the economic evaluation is to provide financial forecasts of the costs to NHSScotland of adopting the recommended three-stage protocol for a national screening programme for people with diabetes.
The following table presents the costs per screen for the seven screening options:
|
Cost per Graded Screen (including fixed costs) |
Mobile GP-based one staff |
Mobile GP-based two staff |
Mobile Van-based one staff |
Hospital-based two staff |
Hospital-based one staff |
| Mydriatic |
£32.28 |
£33.11 |
£30.06 |
£27.94 |
£26.56 |
| Non-Mydriatic |
– |
– |
£21.09 |
– |
£21.04 |
To cost optometrist screening, £1.49 grading and £10.45 fixed costs should
be added to the local optometrist’s fees.
The components that have most influence on these costs are the patient turnaround
times, and, for mobile units, the daily drive time required. Patient turnaround
times for the base cases are 20 minutes for mydriatic photography with one staff
member, 15 minutes for mydriatic photography with two staff members and 10 minutes
for non-mydriatic photography. The base case daily drive time is two hours.
These assumptions have been varied in the report to allow NHS Boards to calculate
the likely costs within their areas, and to compare these with community optometrist
charges.
The cost minimisation analysis indicates that single staffed, hospital facilities
and single staffed, van-based, mobile units are cheaper and that a non-mydriatic
screening programme is cheaper than a mydriatic screening programme, if the
faster turnaround times can be achieved.
For a programme based on non-mydriatic screening, the average costs of hospital-based
screening are similar to the costs of operating a van-based facility. Local
circumstances should determine which service is used. For rural areas, the costs
of a mobile service should be compared to the level of optometrist fees. Major
factors will be acceptable daily drive times and patient travel time.
The modelling shows that moving to systematic screening from an opportunistic
programme is cost-effective for all people with diabetes.
Adopting a three-stage process for the national screening programme is estimated
to cost NHSScotland approximately £3.7 million in the first year and £1.9 million
per annum thereafter. The screening programme will result in more people with
diabetes requiring some form of treatment to improve their sight. The additional
annual treatment costs could be around £65,000.
The HTBS Proposed Model for Diabetic Retinopathy Screening
A quality assured national diabetic retinopathy screening programme is proposed
that is sufficiently flexible to accommodate the needs of patients living in
all communities (urban, rural and island) in Scotland.
Following evaluation and analysis of data and evidence available up to January
2002 on clinical effectiveness, organisational issues, patient issues and economics,
HTBS proposes that the national systematic screening programme for diabetic
retinopathy in Scotland uses the following three-stage process:
- Macular single-field digital retinal photography, without mydriasis, for
each eye. - If there is a technical failure, macular single-field digital retinal photography,
with mydriasis, for each eye. - If there is a technical failure with mydriatic digital photography, biomicroscopy
with a slit lamp.
Visual acuity, with refractive correction if required, should be recorded for
each eye.
HTBS believes this sequential and pragmatic model optimises clinical effectiveness,
cost-effectiveness and patient preferences. Evidence suggests that in approximately
80% of people, images suitable for grading and detection of referable (sight-threatening)
retinopathy will be obtained through undilated pupils, so mydriasis will not
be needed in the majority of patients. However, no patient will be denied mydriasis
when it is necessary and patients known to require mydriasis should start at
the second stage. This sequential, potentially three-stage, process is felt
to be both efficient and failsafe.
The screening/grading will be performed by appropriately trained, accredited
and competent professionals.
A national survey (Appendix 2) indicates that a large number of local schemes
exist in Scotland but none meets the required specifications of a national scheme.
It is important that the introduction of the national screening programme for
diabetic retinopathy does not disadvantage these existing schemes but allows
for their enhancement to meet the approved quality assured specifications.
Screening must be accessible to all patients, whether they receive community-based
and/or hospital-based diabetic care. HTBS has made no restrictive recommendations
on the organisation of the programme in any area, or precluded any professional
groups from participating in the screening programme. The local implementation
must allow easy access for patients and may include services in diabetes centres,
primary healthcare facilities, mobile vans or with community optometrists.
Several important research questions have been identified in this Health Technology
Assessment. One of the key questions relates to the performance of the sequential
three-stage screening model. This will be addressed in the first year of the
roll out of the programme, taking account of data arising from this programme
and that available elsewhere, particularly in the rest of the UK. This will
allow modifications to be made to improve the efficiency of the Scottish screening
programme.
HTBS recommends that the national programme for diabetic retinopathy screening
is achieved by building upon established local systems: evolution rather than
revolution, with best practice and learning shared across Scotland. This will
be achieved with the help of the Scottish Diabetic Group who will be taking
forward implementation of this programme in 2002.