We present here the analysis of a very substantial national data set which is not based on sampling methodology and which report incidence of new cases of certification for visual impairment. We have for comparison, analysis of a previous data set collected nine years previously using essentially the same methods. Since this is not a sample survey, it is not appropriate to use sampling theory in the analysis and observed differences are actual not estimates. Biases clearly exist as to who and who does not become certified and it must be remembered that these data are hospital not population based since an individual has to access the hospital eye service to be seen by a consultant ophthalmologist. Essentially the same biases will have occurred at both time points and the same forms and eligibility criteria were in place. There may have been some drift in the threshold for certification by Consultants over time and this is most likely to have been a lowering of threshold. Over the period there have been a number of campaigns by agencies representing the interests of the visually impaired to highlight the importance of the registration process in facilitating the delivery of Social Service support to those who need it and there have been no formal audits published on eligibility for registration to our knowledge. We cannot know the extent to which this has happened because there is no objective information on visual function on Part 5 of the old BD8 form. This may account for some of the observed increase in registration. However, it is not obvious how this might be disease specific and one might expect such an increase to occur across all causes.
The population has aged considerably even in 9 years however we have provided age specific estimates to allow assessment of this. The population change may have had a differential effect by cause since AMD becomes exceedingly common in the very elderly [9].
None of this is likely to explain the near doubling of incidence of certifiable sight loss due to diabetic retinopathy in people over 65. This finding is of importance since it is contrary to what one might hope and expect with increased efforts being made for the detection and treatment of the condition. One explanation is that diabetics themselves are living longer but remain at risk of the disabling consequences of the disease.
Is an increase in diabetic blindness plausible? There have been increases observed for other diabetic complications such as diabetic nephropathy. UK data has suggested a doubling in incidence of childhood- onset type 1 diabetes between 1966 and 2000 [10]. The prevalence of type II diabetes is increasing; overweight/obesity being the single most important predictive factor for the development of diabetes [11].
Is an increase in AMD blindness likely or possible? There was an increase in 1990–91 as compared to previous years which is present despite the possible effects of previous miscoding [12]. If age effects have been adequately controlled for, there may yet be an absolute increase in the incidence of the disease. The only modifiable risk factor for the disease convincingly demonstrated from epidemiological studies is tobacco smoking [13]. Changes in the patterns of smoking over previous years, particularly the enormous increase in the prevalence of cigarette consumption after the Second World War may be emerging in increasing incidence of the disease.
Much criticism has been directed towards the validity and coverage of the data collected during BD8 certification [14–17]. It is estimated that approximately 53 % of eligible patients have not been registered blind or partially sighted despite consultation with an ophthalmologist. However, it must be noted that cross sectional studies are by their nature bound to detect under-registration because there is often a necessary delay between the onset of certifiable visual loss and the offer of registration. Individuals need time to come to terms with their loss of vision and consultants need time to determine whether an individual is certifiable. One should not underestimate the distress faced by an individual when told that their condition is certifiable for registration as blind or partially sighted. Measures of vision show variability – visual acuity can fluctuate in an individual with diabetes, and the point at which someone becomes certifiable due to visual field loss in glaucoma is not always easy to determine. These figures can surely be useful as indicators or minimum estimates of the incidence of severe sight loss in the population for planning preventive health care strategies and prioritising research particularly for irreversible causes.
The WHO stresses the importance of collecting within-country data on causes of visual blindness for use in priority-setting and resource allocation [18]. The future with regards analysis of data on the causes of visual impairment in England and Wales is unclear.
At the end of October 2000, a bulletin was circulated to all Directors of Social Services in England and Chief Executives of Health Authorities and Trusts by the Department of Health (DH) indicating that the DH no longer required the part 5 (the epidemiological return) of the form to be sent to ONS and that part 5 would be omitted for future reprint. In November 2003, following consultation with service users and key stakeholders, form BD8 was replaced by the Certificate of Vision Impairment 2003 in England – the form is currently undergoing trial and there is no commitment from the Department of Health for future analysis of data on causes of severe visual loss. A similar situation exists in Wales although their CVI has yet to be launched.