DIABETIC RETINOPATHY SCREENING SERVICES IN SCOTLAND: RECOMMENDATIONS FOR IMPLEMENTATION
Diabetic Retinopathy Screening: Software to Support Diabetic Retinopathy Screening
52. The diabetic retinopathy programme must be supported by appropriate information management and technology. To ensure consistency of standards across the country NHS Boards will need to adopt the same or compatible software. A single integrated system to cover all the required functions of screening has many attractions but practical considerations (including cost constraints) may dictate a solution based upon separate inter-linking modules. The exact specification of the software which should be adopted will not be clear until a full specification has been developed and its procurement undertaken. This section of the report sets out the principles which this exercise must follow and recommendations about how the process should be managed.
53. Although it is not possible at this stage to specify the detail of the software solution, it is possible to describe the component functions and explain how they must interact. Figure 2 below presents an overview of the system. The key elements are:
Interface with the Local Diabetes Register
Patient management and recall
Grading and reporting images
Audit and statistical analysis
Key statement 22: A fully comprehensive software solution is required for the effective and efficient provision of diabetic retinopathy screening. Such a system must integrate or interface the functions of image acquisition, call-recall, grading and quality assurance. On behalf of NHS Boards and Trusts, NSD should commission work to define the full specification of the system and bring it into operation.
Key statement 23: Until the software to support DRS is developed, piloted and implemented, NHS Boards should offer DRS to patients using standard clinic appointment systems.
54. Interface with the Local Diabetes Register. It is the responsibility of the NHS Board to create and maintain a Diabetes Register. In time, all NHS Boards will have a clinical management system (CMS) which will include data for all patients with diabetes in their area. In most cases, this system will be SCI-DC, ( see Annex G). Demographic information from the register will be used to prime the call/recall database. The results of screening, including images, will be fed back into the register. Pragmatic solutions will need to be employed until these local registers are fully operational. Software will need to be able to accommodate areas which choose not to use SCI-DC. The use of the CHI number is essential in all aspects of interface and interchange between all components of the systems involved in providing this service.
Key statement 24: Diabetic retinopathy screening is an integral part of diabetes care and must be fully linked to the local clinical management system. The demographic data to populate the DRS system will be supplied by the local clinical management system. The results of the screening process (including a compressed image) must be fed back into the local clinical management system. DRS software will be compatible with SCI-DC.
55. Patient management and recall. The system will generate all call/recall standard letters and will be responsible for the management of patients who do not attend (DNA) or cannot attend (CNA). A list of the required letters is set out below ( see Box 2). The system needs to keep track of all the letters generated and alert the screening co-ordinator when actions are overdue. The system will control the booking and arrangement of all screening sessions. It will also provide data for monitoring national standards for clinic uptake and revisit targets. It will be able to pass on data containing action recommendations as well as interface with ophthalmology department records so that feedback on treatment and referral outcomes can be monitored. The system needs to be "time aware" on these movements and provide reports to support the monitoring national standards. It must generate all letters to patients, GPs, Optometrists, Ophthalmologists and Diabetologists.
Figure 2: System overview showing how the components of retinopathy screening interact
Key statement 25: To avoid the risk of missing some patients, it is essential to adhere to the guiding principles of a standardised approach with integral failsafe arrangements and effective monitoring and evaluation. NSD will specify the standards and principles.
56. Image capture. The software will be expected to capture images in acceptable (accredited) file format, resolution and quality, edit or add patient information on the front end and label images on capture. It should allow images captured by community optometrists and distributed by CD to be imported. Further information about cameras and image transmission is included above ( see paragraph 45). It is possible that a supplier wishing to provide software may also wish to supply cameras and associated software. This may have advantages for the NHS in terms of avoiding disputes about responsibilities if problems arise. Any procurement exercise will clearly specify the standards to be delivered but will be open to proposals which secure optimum value.
57. Grading and reporting images. The software will need to support manipulation and enhancement of images using methods that do not alter the original image (such as zoom facility, contrast/brightness, red free - the ability to take red out of the image); allow flexible input of grading criteria complying with national standards; keep a record of who graded the image; produce a grading report to send to the GP and keep for screening service records; produce a patient report; and allow images from previous screening episodes to be viewed on the same screen as images currently being graded. There must be the capability to record changes at secondary or tertiary grading in either classification or in the action to be taken. Changes should remain distinct at each subsequent stage and not overwrite previous stages. The software must be capable of operating for individual patients or for whole clinics. A flag system must be incorporated so that sequential grading can identify and log urgent cases. This will allow higher level graders to extract appropriate cases for early attention. Further information about grading issues is included above ( see paragraph 43).
58. Archiving and backup. Archiving and backup software must be able to produce archives on either removable media or over networks. Archived material must include images and clinical data and must be in industry standard formats which can be read by generic software; not solely by the software which created them.
59. Audit and statistical analysis. It will be necessary to be able to interrogate the system so that all information remains accessible. Audit data will be determined according to the national standards and the system must be capable of producing queries on each parameter stored such that audit of individual screeners, graders, centres, cameras or other relevant components is possible. Automated printed reports are useful for regular enquiries but the system also needs the ability to export data for ad hoc reporting. The system must be capable of exporting summarised clinical data, grading results, outcomes and service levels with selection of batches of images for QA. Audit must respect patient confidentiality.
60. Quality assurance. Since a proportion of all images will be re-evaluated for quality assurance purposes, the dataset must allow recording of two screening results for each image with a field to record which interpretation is adjudicated correct in the case of discrepancies. Random selection of a set of images for QA should be achievable within the IT system. To enable quality assurance and the management of images captured in the community, an ability to export images to, and import from CD is necessary. Further information about quality assurance is included below ( see paragraph 65).
61. IT support. The exact standards for communication with other software systems to be specified in the national screening programme have still to be defined. Exportation from a system must be sufficiently simple for local IT staff to operate. It is essential that any approved systems comply with recognised standards for integrity and security of data. This implies effective backup, product maintenance contracts and product update procedures.
Developing and Procuring Software
62. The complexity and risk associated with developing and procuring this key piece of software and the importance of consistent application of standards and quality assurance demands a single Scottish solution to support the DRS programme.
Key statement 26: Fully comprehensive software to support diabetic retinopathy screening should be procured centrally rather than by each NHS Board independently. The commissioning process will determine how this will be provided in practice - what will be purchased centrally and what will be the responsibility of NHS Boards.
Key statement 27: In order to ensure consistent standards of service delivery and quality assurance, all NHS Boards must either use the nationally developed software, or satisfy NSD that an alternative system can deliver fully comparable results.
63. NSD will develop the system specification and seek information from suppliers. It is likely that a number of options will be offered both in terms of the software itself and how this might be supplied (e.g. via a Prime Contractor arrangement or a Managed Service approach). Issues such as system maintenance will also be considered.
Key statement 28: The full specification of the software and proposals to deliver it will be produced by NSD by July 2003.
64. Taking the process from the development of the software specification, through a procurement exercise to the implementation phase will take some time. During this period NHS Boards will need to prepare the local infrastructure (e.g. ensuring the quality and completeness of their patient register, purchase cameras and train staff) whilst at the same time continuing to screen patients. The DRS collaborative network ( see paragraph 85) is expected to have an important role in facilitating the sharing of ideas about how to establish an efficient local DRS service. Smaller Boards and the island Boards in particular are urged to explore opportunities to work jointly with other areas to achieve best value.
Diabetic Retinopathy Screening: Quality Assurance and Standard Setting
Key statement 29: A fully comprehensive software solution to support diabetic retinopathy screening will be made available by September 2004 and fully implemented by no later than September 2005.
65. Quality Assurance is defined as: 'improving performance and preventing problems through planned and systematic activities including documentation, training and review'. All NHS Boards are currently reviewing their screening systems following publication of the HTBS report. However, as the recent survey showed, very few local services incorporate any quality assurance measures to check screening quality. Uniform national standards are required to ensure that all patients receive adequate retinopathy screening. National standards for the DRS programme will be defined by NHS Quality Improvement Scotland. Additional operational details will be defined nationally and included in the DRS Manual.
66. NHS Quality Improvement Scotland has the remit to develop and run a quality assurance process for clinical services provided by NHSScotland. For screening programmes, NHS Quality Improvement Scotland has an additional responsibility to ensure adequate quality assurance is in place for each programme. NHS Quality Improvement Scotland will produce a set of clinical standards specifically for diabetic retinopathy screening (DRS).
67. Existing NHS Quality Improvement Scotland standards for cervical and breast screening programmes and general diabetes services, in conjunction with information from the HTBS Technology Assessment report, will be used to produce the draft DRS standards. The standards development process will also be guided by the recommendations contained in this report.
Key statement 30: The NHS Quality Improvement Scotland Working Group on Diabetic Retinopathy Screening should produce standards for DRS. These standards will cover the following topics: responsibilities, service specification, communication, patient information, call/recall and failsafe, the screening process, quality assurance, referral and treatment. Draft standards should be published by May 2003.
68. The NHS Quality Improvement Scotland draft standards will be reviewed following formal consultation and piloting, and in the light of evidence emerging from the DRS programme. There may be merit in allowing this process to be extended to allow the standards to be fully tested in practice before being finalised. Reviews will be undertaken once DRS services are in operation. A decision regarding the timing of review visits will need to take account of the extent to which DRS programmes have been established.
69. The standards published by NHS Quality Improvement Scotland will include elements common to all national screening programmes (e.g. register, call/recall issues) and those that relate solely to diabetic retinopathy (e.g. quality assurance of graders and equipment). These standards will underpin the quality assurance arrangements for the DRS programme. However, the detail of how these standards should be applied in practice will need to evolve as the programme develops and as lessons are learned. Guidance on quality assurance will be made available in the DRS Manual which will be revised in the light of experience.
70. Although the focus must be on implementing national standards, the introduction of a consistent quality assurance regime must be tempered by pragmatism during the development of the DRS programme and in light of the uncertainty surrounding a number of issues including the minimum proportion of digital photographs which should be quality assured by a level 3 grader.
71. No screening programme can be foolproof and it is inevitable that some images with signs of retinopathy, or more seriously, sight-threatening retinopathy will be missed. Quality assurance is required in order to minimise the number of errors. It is estimated that approximately a tenth of all images viewed within the DRS programme will be for the purposes of quality assurance (i.e. level 3 grader or equivalent, checking a proportion of images graded by level 1 graders and level 2 graders, not previously referred to the level 3 grader). The performance of level 3 graders will be monitored through regular audit undertaken by the DRS Network Co-ordinator which will allow any outliers to be swiftly identified. The standards for quality assurance will be determined by NHS Quality Improvement Scotland and may be revised in the light of experience.
Key statement 31: Quality assurance requires that for each grader randomly sampled images are checked by a level 3 grader. Initial estimates suggest that a minimum of 200 images per grader per annum will be required. This will be finalised in the NHS Quality Improvement Scotland Standard.
72. If it becomes a reality, automated grading would allow a larger number of images per grader to be regraded. Indeed, it may in time prove possible to automate all level 1 grading. This will be kept under review as the technology develops.
73. One of the most important standards for assuring the quality of the DRS programme will be to ensure that all graders and screeners are properly trained and accredited. All graders must work to the same standards. Grading standards will be maintained by regular proficiency testing.
Key statement 32: NSD should be responsible for the national organisation of proficiency testing.
74. NHS Boards will be accountable for monitoring and performance management of the screening service and for overseeing the day-to-day "quality control" QA. Other monitoring arrangements are outlined below ( see paragraph 90).Diabetic Retinopathy Screening: Impact on Ophthalmology Services
75. Any national screening programme will be expected to have some impact on ophthalmology services. The National Screening Committee's report suggested this impact might be very considerable as they estimated that, for a population of 600,000 (prevalence of diabetes 2.5%) 8% would be referred in the first year with referrals falling to about 3% only by year 4. However, based on the structure of the screening programme recommended for Scotland and the insights gained in Grampian and Tayside (two health boards with considerable experience of operating a screening service) it is believed that the impact on ophthalmology will be manageable. Ophthalmologists have a central role in the quality assurance of the DRS programme and as level 3 graders. NHS Boards must provide adequate protected clinical time for these functions. In Grampian one weekly session of protected ophthalmology time for quality assurance and level 3 grading has been provided. A number of ophthalmologists will also be involved in the provision of training.
Key statement 33: The net effect on ophthalmology services of the introduction of a national diabetic retinopathy screening programme is likely to be modest. The impact should, however, be monitored by NHS Boards.
Key statement 34: Protected staff time for quality assurance and level 3 grading must be provided.
76. The Grampian screening programme started in February 2002. For the first 3,000 patients screened, the referral rate for patients new to ophthalmology has been 4.2%. Similar figures have also been found in Tayside where using a similar grading system to the National Screening Committee (NSC). New referrals to ophthalmology have consistently been in the range of 3-4% over an eight-year period, screening 3,500-4,500 patients per annum. In Lanarkshire, new referrals to ophthalmology currently run at 2.5%.
77. In Grampian, the commonest referable grade has been diabetic maculopathy (see table 1 below).
Table 1: Referable Grades (including patients already attending ophthalmology)
Source: Grampian Retinal Screening Programme.
78. The Scottish Diabetic Retinopathy Grading Scheme 2003 enables referrals to ophthalmology to be seen according to clinical priority. However, this will be of little value if waiting times for referrals to ophthalmology are excessive. This should be monitored.
Key statement 35: Referrals to ophthalmology from screening should be seen according to clinical priority as determined by the screening photograph. Waiting times should be audited by NHS Boards.
79. Single field non-mydriatic photography has yielded a technical failure rate of 8-9% in an established screening programme over 12 years (Tayside). A recent research project showed that subsequent mydriasis reduces the failure rate to around 4%. Research in Grampian has shown that one-field photography has a lower technical failure rate than two field (3.5% vs. 4.5%). The main reason for technical failure (despite the use of mydriasis) appears to be cataract.
80. Patients for whom it is not possible to obtain an image from a digital camera even with mydriasis require slit-lamp biomicroscopy. It is expected that these patients will be examined within the screening system in a separate session by one of the level 2 graders, or by a community optometrist. Ophthalmology departments would not be able to contain the workload within existing resources if technical failures were to be referred to them.
Key statement 36: Technical failures (i.e. ungradeable images) must be contained within the screening programme in order to prevent ophthalmology services from being overwhelmed.
Returning Patients to the Screening Programme
81. The facility for ophthalmologists to discharge patients safely back to the screening programme after treatment will be crucial to allow ophthalmology departments to absorb the impact of the screening programme. Stable patients who at present have to attend hospital ophthalmology clinics for regular review will in future attend the screening programme.
Key statement 37: The screening programme should provide capacity for ophthalmologists to return to screening those patients who, after treatment or otherwise, no longer have referable retinopathy and who may therefore re-enter the screening cycle without continuing to attend an ophthalmologist.
82. The DRSIG recommends the following guidelines to help identify patients currently attending hospital ophthalmology clinics who are suitable for transfer to the screening programme:
Patients with moderate background retinopathy can be discharged to the screening programme provided there are facilities for 6-monthly follow-up within the programme.
Patients with observable maculopathy can be discharged to the screening programme provided there are facilities for 6-monthly follow-up within the programme.
Patients with maculopathy or macular oedema can be discharged following successful laser treatment if they no longer meet the criteria for referral.
Patients with complete regression of proliferative retinopathy can be discharged following successful laser treatment if they no longer meet the HTBS criteria for referral.
Patients with proliferative retinopathy who have had laser treatment but who have residual new vessels can be discharged to the screening programme provided there is photographic evidence that there has been no progression of the new vessels over a period of 6 months.
The presence or absence of laser burns should have no effect on the decision to refer, monitor or discharge patients.
All patients that are discharged must have baseline digital retinal photographs sent to the screening programme.
Finally, the DRS programme director must have the right to refuse discharged patients if it is felt that their retinal photographs suggest continued observation by the hospital ophthalmologist is appropriate.
Linking Ophthalmology to the Screening Programme
83. In order to support the screening programme effectively, ophthalmology departments should be provided with the necessary equipment and IT.
Diabetic Retinopathy Screening: National Support and Managing
Key statement 38: All ophthalmology services must have access to digital photography. In addition, ophthalmologists will require access to the Diabetes Clinical Management System (SCI-DC) and NHS Network.
84. Providing retinopathy screening to people with diabetes is the responsibility of NHS Boards. However, in order to assist boards to meet their obligations and to ensure that consistent standards are applied and monitored across the country, central co-ordination and support is required to:
provide guidance and to help disseminate and share knowledge
support quality assurance
ensure that appropriate training is available.
85. The DRSIG consider that the most appropriate mechanism to fulfil these tasks is a collaborative network guided by NSD.
Key statement 39: A DRS collaborative network should be established to support and facilitate the implementation of DRS across Scotland. This network should be directed by an executive group comprising individuals from NHS Boards, the various relevant professions involved in the retinal screening programme and patient representatives.
Key statement 40: A full time DRS network co-ordinator should be appointed by NSD to initiate and support the network and facilitate exchange of information between screening centres.
Key statement 41: By June 2003 a lead clinician should be appointed to act as chairman of the executive group of the DRS collaborative network. A part-time commitment of one half-day session per week is suggested for a period of three years.
86. The collaborative network might develop into a more formal Managed Clinical Network. This will depend upon the needs of the service and consideration of the relationship with the diabetes MCNs in each NHS Board area.
87. One of the main responsibilities of the DRS network co-ordinator would be to prepare and maintain a DRS Manual for use by NHS Boards which will contain detailed guidance about establishing and running a DRS service; standardised patient information, letters and other materials, including:
Most recent statement of policy for Scotland (from SEHD circular)
Detailed Specification of the service to be delivered
NHS Quality Improvement Scotland national standards
Detailed guidelines for effective call/recall arrangements (including exclusion criteria)
Specification of the national minimum data set required to be submitted for the annual Scottish Diabetes Survey
List of lead contacts in each NHS Board area
Summary of HTBS recommendations
Professional guidelines on (a) cameras; (b) mydriasis; (c) reasons for exclusion from screening; (d) criteria for referral to ophthalmology including recommendations for urgency of referral for different circumstances; (e) others taken from DRSIG report
Patient and professional information leaflets (templates and good practice examples)
Other guidelines and protocols that represent good practice and are approved by the Executive Group of the DRS collaborative network.
88. The manual is largely a collation of information and guidance from other sources, especially the HTBS report. The objective is to publish a first edition by September 2003. Thereafter, the manual would require to be kept up to date by regular revision.
Key statement 42: A DRS Manual for NHS Boards should be published and maintained by NSD, on behalf of the DRS collaborative network. The DRS Manual should contain detailed guidance on establishing and managing a DRS programme and include a collation of all relevant standards and policies, standard letters and contact details. A first edition of the DRS Manual should be published by September 2003.
89. A first objective for NHS Boards in implementing retinopathy screening was highlighted in the Scottish Diabetes Framework. The Framework recognised eye care as one of the first stage priority issues, and set a target that all people with diabetes should have their eye status (retinopathy) recorded on the local diabetes register by September 2003. The 2002 Scottish Diabetes Survey (12) showed that by September 2002 of those patient registered, almost three-quarters had been screened, but that over a quarter of patients either had never been screened, or had no recorded result ( see Annex C).
Monitoring the DRS Programme
90. The DRS programme will be monitored in five complementary ways which together will ensure that national standards are effectively applied:-
Quality control - reviewing and checking of all the aspects of performance to determine that they meet the agreed standards.
External quality assurance - regular audit by the National DRS Co-ordinator to ensure that routine quality control systems are working effectively.
National Review - periodic NHS Quality Improvement Scotland review of standards.
SMR Returns - mandatory collection of key performance indicators compiled nationally by ISD.
Other reporting mechanisms - Performance Assessment Framework, annual Scottish Diabetes Survey, Diabetes Annual Report produced by NHS Boards, local clinical governance arrangements).
Key statement 43: The main method for assessment of the performance of NHSScotland in the operation of DRS will be the reviews undertaken by NHS Quality Improvement Scotland which will evaluate apply the published Diabetic Retinopathy Screening Standards. Review visits will commence once DRS services are in operation. In the interim, information about progress towards improving DRS will be published as part of the NHS Quality Improvement Scotland review of the diabetes services, in the annual Scottish Diabetes Survey and in the Diabetes Annual Reports published by NHS Boards.
Key statement 44: ISD should be involved in defining the national dataset to be used by the DRS programme. A subset of the dataset should be identified as key performance indicators to be included as a routine SMR return.
Review of Diabetes Services
91. Diabetes services will be subject to a NHS Quality Improvement Scotland review during 2003. The diabetes standards to be reviewed were published in final form in October 2002. (13) The existing standards which relate to eye care are set out in Box 1 opposite. The standards specific to DRS referred to above (paragraph 65) are in addition to these more general service standards.
Box 1: Diabetes Clinical Standards defined by NHS Quality Improvement Scotland (formerly the Clinical Standards Board for Scotland)
Clinical Standards: Diabetes
Standard Statement. All people with diabetes are offered annual or more frequent examination, where clinically indicated, to monitor the management and progression of their condition. There is intervention as required, and support for the modification of lifestyle factors.
There is a protocol to ensure that all people with diabetes are offered review of the following indicators on an annual basis, or more frequently where clinically indicated, from diagnosis.
4. Eye examination for diabetic retinopathy according to HTBS recommendations. (Essential)
17. Referring practitioners (including optometrists, with patient consent) are given feedback regarding the outcome of their referrals. (Desirable)
Clinical Management: Eyes
Standard Statement. All people with diabetes who have identified signs of developing diabetes-related, sight-threatening retinopathy, according to HTBS grading recommendations are referred to an ophthalmologist for assessment and, if necessary, treatment.
1. There is a referral process to a consultant ophthalmologist-led service for people with diabetes with identified signs of developing diabetes-related, sight-threatening retinopathy according to HTBS grading recommendations.
2. All people whose eye examination has revealed retinopathy have their glycaemic control and blood pressure reviewed and treated as clinically indicated.
3. All people with active proliferative diabetic retinopathy are offered laser treatment.
Clinical Standards Board for Scotland (2002). Clinical Standards: Diabetes. Second Edition
Measure of Success: Number of People Being Screened
92. One of the key measures of the success of the DRS programme will be the proportion of people with diabetes screened each year. This aspect of the performance of NHS Boards will be monitored but it is realised that screening rates depend on the decisions of individual patients whether or not they wish to attend.
Key statement 45: People with diabetes are strongly encouraged to attend for screening, but attendance cannot be compulsory. Although NHS Boards are required to offer screening to all appropriate patients, they cannot be held responsible for those who choose not to attend. Research is required to explore why some people do not take up the offer of screening.