This study identified screening interventions that eye care service providers, policy makers and health service commissioners would find feasible and acceptable to use in a trial of glaucoma screening and for implementation into practice if screening was effective. The economic modelling component of this study ruled out candidate tests that would be highly resource intensive and provided a short list of candidate strategies that had the greatest potential to be cost-effective. These involved testing in a primary care setting (because they are embedded in the community and have the infrastructure in place) delivered by technically trained personnel using a simple testing strategy - a measure of intraocular pressure (a known risk factor for glaucoma) , and either a measure of the visual field or photography of the optic nerve.
The screening configurations identified in this study meet one of the key principles of screening - that a simple, safe and affordable test is required before a screening programme can be implemented [30, 16]. There is tension, however, when determining the most appropriate screening test, between a public health and a clinical perspective. The public health perspective requires that a screening test be sufficiently precise to distinguish those who probably do and do not have the disease at an acceptable cost, whereas the clinical view point often tends towards the desire for precision and reliability. There are many sophisticated imaging tests for diagnosing glaucoma, but these are not affordable for a public health intervention. Similarly, there are a variety of technologies to evaluate the peripheral visual fields. From the currently available tests, fundus photography or simple visual field testing combined with a measure of intraocular pressure meet the economic criteria. More precise technologies or examinations would then be required to examine those who screen positive. Currently, the recognition of glaucoma on fundus photography is a skilled task and would require manual grading by experienced observers. Automated grading systems for retinal images have been developed for diabetic retinopathy screening , and have been shown to be less costly and as effective as manual grading . Similarly, there is potential for the development of automated systems for glaucoma recognition from retinal photographs . The performance of photography, with and without automated grading, compared with simple visual field testing would need to be compared within the intervention arm in a future glaucoma screening trial or in a parallel companion study. Screening tests suitable for self-assessment are in the early stages of development, and should be considered as an option when the technologies have advanced sufficiently.
This study is one of the first to show the formal integration of qualitative and quantitative methodologies (as promoted in the MRC guidance) in the development of a complex intervention and specifically the use of a qualitative enquiry to restructure the care pathways of an economic model. The qualitative exploration enabled agreement to emerge for some aspects of the proposed intervention, whilst allowing reasons for disagreement to be explored in-depth. The detailed information that underpinned the lack of consensus (e.g. the disagreements between the professional groupings) informs the implementation planning for any of the proposed interventions, and aided interpretation of the results of the economic analysis in particular highlighting that a more universal coverage of screening all individuals above a certain minimum age would represent a more appropriate balance between equality of access and cost-effectiveness. The economic modelling facilitated the reduction of a wide selection of suggested interventions to a shortlist representing those most likely to be cost-effective. The study also brought together the perspectives of a wide range of stakeholders (including ophthalmologists, optometrists, GPs, nurses, technicians and policy makers) who would all have input into a screening programme for glaucoma, and this breadth of perspective adds to the veracity of our findings. Also, while previous attempts to provide high quality evidence for the case of screening for glaucoma have concentrated on the technical and technological aspects of design, evaluation and assessment, our research has also considered the implications of a screening programme on the complex organisational context in which the management of glaucoma exists.
Whilst our project was successful in facilitating the development of an intervention for future testing in a trial of glaucoma screening, it did have a number of limitations. In particular, recruitment proved more difficult in practice than had been hoped. Whilst ophthalmologists and optometrists were easier to recruit, nurses, technicians and general practitioners were more difficult. In the case of nurses and technicians, the difficulty appeared to be related to confidence in speaking as one who is located relatively low in the clinical 'hierarchy', whilst difficulties in recruiting general practitioners' in research is well recognised . Additionally, despite the use of telephone interviews, we were unable to recruit as many respondents with practice experience of isolated and rural locations.
The MRC revised guidance outlines well the appropriate steps for the development of a complex intervention . Integral to its tenets is the recommendation for the synthesis of qualitative and quantitative methods, as well as the need to consider implementation issues through participants' accounts of their perspectives and experiences. The integration process, however, required significant iteration between the disciplinary approaches to ensure that the qualitative data could be synthesised into an economic model. The specification of particular screening interventions was a challenge for the qualitative researchers because this was not a standard way of interpreting the data produced. Nevertheless, two complementary mechanisms to synthesising the qualitative data were used - firstly to systematically sort and describe the respondents' opinions about preferred components of the intervention (to allow direct usage in the economic model) and secondly to use the richness of the data provided by the qualitative data to undertake an interpretative analysis of the results of the economic model and around the implementation considerations in which the preferences were nested. The importance of interplay between disciplines in mixed methods research has been emphasised in the literature, particularly by O'Cathain and colleagues .
We have identified screening strategies that are acceptable in a service provider and policy context, but they may not be acceptable to health care users. Additional research is required to elicit the beliefs of health service users regarding the acceptability of these screening interventions and their current use of eye care services to identify the most likely behaviour change interventions for maximising attendance for glaucoma testing. These findings will provide additional information, in terms of potential uptake of any screening interventions and current uptake of current eye care services, to be integrated into the economic model to inform the feasibility of any future glaucoma screening trial. This research is underway.