Background/aims Advances in therapy have improved outcomes for patients with neovascular age-related macular degeneration (nAMD). Prompt access to treatment is a priority and may be used as a key performance indicator. In this study, we investigate how social deprivation may impact on access to services, treatment and visual impairment registration.
Methods Patients were identified retrospectively through the Certificate of Visual Impairment system for the University Hospitals Birmingham Medical Retina service. The Index of Multiple Deprivation (IMD) 2007 score was calculated for each patient. The impact of deprivation, age, gender and ethnicity on key stages in the care pathway was assessed.
Results 120 patients were identified. Patients with greater social deprivation were under-represented, had worse visual acuity at first presentation (correlation of the better-seeing eye with IMD 0.225 (p=0.013)) and had sight-impairment registration earlier (correlation −0.246; p=0.007). Deprivation did not affect time to first appointment, and was not associated with a higher rate of non-attendance.
Conclusions The late presentation and under-representation of patients with greater social deprivation is a serious concern. Our study strongly suggests that this vulnerable group is encountering barriers in accessing treatment in nAMD, and that these occur prior to entry into the Hospital Eye Service.
- Public health
Statistics from Altmetric.com
Age-related macular degeneration (AMD) is the leading cause of blind registration in England and Wales,1 and as the elderly population continues to expand, the number of people with visual loss secondary to AMD is predicted to rise.2 Recent advances in the treatment of neovascular AMD (nAMD) have resulted in vastly improved visual outcomes,3 and this, combined with the identification of various risk factors for the development of AMD, means that it has been listed in the Public Health Outcome Framework as a ‘preventable sight loss indicator’.4 It is therefore important that services are designed to ensure patients with nAMD receive prompt referral to the Hospital Eye Service (HES) for risk factor amelioration and disease treatment. Any barriers to this process existing because of inequalities in level of social deprivation must be addressed.
There is a well-established link between the degree of social deprivation and many systemic conditions including hypertension, heart disease and diabetes,5 as well as overall morbidity and life expectancy.6 Social deprivation has also been linked to a number of ocular conditions, such as primary angle closure glaucoma,7 diabetic retinopathy,8 retinal detachment9 and late presentation of primary open-angle glaucoma (POAG).10 It has also been found to be associated with worse visual acuity (VA) at presentation for cataract surgery,11 and paediatric visual impairment registration.12
The Index of Multiple Deprivation (IMD) 2007 is a well-validated quantifier which provides a relative ranking of areas across England according to their level of social deprivation. In order to identify pockets of deprivation, the IMD 2007 looks at small areas known as Lower Super Output Areas (LSOA); each contains an average population of 1500 people and can be identified from a person's postcode. Each LSOA is given an overall national rank according to its level of deprivation. This ranking is calculated by incorporating information from the following domains: income, employment, health deprivation and disability, education skills and training, barriers to housing and services, crime and living environment.
The purpose of our study was to assess social deprivation in a cohort of patients with sight loss due to severe nAMD. Additionally, we sought to establish whether social deprivation affects the quality of care received by these patients following referral to the HES.
We retrospectively identified all patients who had been registered for a CVI for sight impairment and severe sight impairment due to nAMD, between November 2007 and June 2012 in a teaching hospital setting in the UK; this was conducted as part of a service evaluation of the Medical Retina Service of the University Hospitals Birmingham NHS Foundation Trust, UK. Patients were identified via the hospital's own log of CVI registrations, and data was collected from the records of each patient to ascertain demographic information including age, self-determined ethnicity, gender and postcode. We identified the time taken from community referral to initial eye clinic appointment, and whether there had been any patient non-attendances (‘Did Not Attends’ (DNA)) or cancellations prior to this appointment. We noted the LogMAR (Logarithm of the Minimum Angle of Resolution) VA of each eye at presentation to the eye clinic and then at the time of CVI registration. We also ascertained each patient's treatment pathway in terms of time taken from clinic appointment to first treatment, any DNAs for treatment and how many treatments were received prior to CVI registration. The term ‘study eye’ was used to describe the eye with nAMD which generated the referral to the HES. Where the patient was referred with bilateral active nAMD then the ‘study eye’ was deemed to be the one which was prioritised to receive treatment first.
The postcode of the patient's home address was linked to the LSOA and given an IMD 2007 score. The distribution of these scores for our patient cohort was then compared with the distribution for our catchment area within Birmingham. From this we ascertained quintiles ranging from 1 (most socially deprived) to 5 (least socially deprived). The observed and expected frequencies for these quintiles were compared using a χ2 test. The association between two continuous variables, such as IMD score and VA, was measured using Spearman's rank correlation coefficient. The difference between two categorical variables, such as whether a patient DNA or not, was tested using Mann–Whitney U Test. Multiple linear regression was used to identify predictors of continuous variables such as time to first treatment.
There were 120 patients (35 men and 85 women) meeting the inclusion criteria for the study (nAMD, registered for CVI between November 2007 and June 2012, resident in Birmingham). The mean age at time of referral to the HES was 83.3 years (range 67–95 years) and the majority of patients were classed as white British (n=112, 93%) with six white non-British patients and two of mixed ethnicity.
Patients with greater social deprivation are under-represented in this cohort of patients with severe nAMD
We identified a significant difference between the observed and expected frequencies for the different quintiles of social deprivation (χ2 27.33; 4 degrees of freedom; p<0.001) with an under-representation of patients from areas of greater deprivation (quintile 1) and a corresponding over-representation from areas of lower deprivation (quintiles 4 and 5) (table 1).
Potential delays in assessment of nAMD are not influenced by social deprivation
For all patients (n=120), the median time from referral to first appointment given by the HES was 27 days (IQR 50 days). There was no significant correlation found between IMD score and time to first appointment given by the hospital (0.032; p=0.735). DNAs and advised cancellations were equally represented with no significant association between DNA and/or cancellation and IMD score (Mann–Whitney for DNA: U=528 (p=0.515); for cancellation: U=584 (p=0.196); for DNA and cancellation: U=939 (p=0.344)).
Social deprivation has an adverse effect on the level of VA at presentation to the hospital eye service
The median VA at presentation to the HES was 0.8 (IQR=0.48) in the study eye and 0.6 (IQR=0.5) in the better eye. The VA at presentation showed a weak positive correlation with social deprivation, that is, the greater the deprivation, the lower the vision. For the study eye, this correlation was 0.185 (p=0.043), and for the better-seeing eye, this increased to 0.225 (p=0.013). The VA at presentation also showed a positive correlation with age, being 0.262 (p=0.005) for the study eye but not statistically significant for the better-seeing eye (correlation of 0.166 (p=0.077)).
Social deprivation does not affect the treatment pathway in nAMD
The median time from presentation to the HES to first treatment was 24 days (IQR=79 days). Deprivation was not associated with any additional delay in this process, with no significant correlation between IMD and time to first treatment (−0.162; p=0.126). Additionally, deprivation did not affect the rate of DNAs and/or cancellations for treatment (Mann–Whitney for DNA, U=1603 (p=0.77); for cancellation, U=1572 (p=0.37); for DNA and cancellation, U=1547 (p=0.65)).
In 17% of patients (n=25), no treatment was given, either through the patient declining or not fulfilling National Institute of Health and Care Excellence (NICE) guidelines. There was no association between deprivation and likelihood of not receiving treatment (Mann–Whitney U=1157.500 (p=0.846)). Overall, the median number of treatments given prior to CVI registration was 4 (IQR=5). Although not significant, there was a trend towards fewer treatments administered prior to registration in the higher deprivation group (−0.167; p=0.069).
Social deprivation is associated with earlier registration of sight loss
In our cohort, 73.3% (n=88) of patients were registered as sight impaired, and 26.7% (n=32) as severely sight impaired. The median time from presentation to CVI registration was 290 days (IQR=669 days). Increasing deprivation was associated with earlier CVI registration, with an inverse correlation of IMD score and time to registration (−0.246; p=0.007).
Median VA of the better eye at time of registration was 0.65 (IQR=0.30) for those being registered as sight impaired, and 1.01 (IQR=0.40) for severely sight impaired. Deprivation did not affect the level of VA at which registration occurred, with no significant correlation between VA and IMD score for those being registered as severely sight impaired (−0.152; p=0.407) or sight impaired (0.139; p=0.196).
Role of deprivation within multivariate models of the patient pathway in severe nAMD
Multivariate regression models were generated comprising the input variables of gender, age and deprivation. We removed ethnicity from this model, as our cohort was almost exclusively Caucasian (98.3% n=118). Neither gender, age, nor IMD score were predictors for time from referral to first appointment (p=0.906) or time to first treatment (p=0.108). However, all variables were significant when assessing time to CVI registration (p=0.004). Age was the strongest predictor of earlier registration (p=0.004); as age increased by 1 year, the time to registration decreased by 17.6 days. Age, gender and IMD score accounted for 9% of the variability in time to CVI registration (adjusted R2=0.091).
Social deprivation has been linked to a variety of eye conditions,7–12 and the association between socioeconomic status and visual impairment is well known.13 In this study, we examined the link between social deprivation and nAMD in a population attending a university hospital in Birmingham, UK.
We found that patients with greater social deprivation are under-represented in a cohort of patients certified with visual impairment due to severe nAMD. A number of studies have looked at the association between elements of social deprivation and early AMD (in contrast with nAMD in our study) with mixed findings. The Beijing Eye Study14 and The Singapore Malay Eye Study15 demonstrated a significant association with lower education level, although not with housing type and income. Similarly, the Age-Related Eye Disease Study (AREDS)16 found that education level was inversely related to dry AMD and nAMD. Other studies, namely The Visual Impairment Project,17 The Framingham Eye Study18 and The National Health and Nutrition Examination Survey19 have shown no relationship between education level and AMD.
The explanation for the disparity between these studies is multifactorial. Overall, the studies are hugely diverse in terms of aims, populations and methods, and cannot be specifically compared with our study or each other. They do, however, give an indication of how elements of social deprivation may interact with AMD.
We recognise that, although our study cohort of CVI-registered patients is a vitally important subpopulation of AMD sufferers, the potential exists for there to be a number of biases present because of the decision to use this group. Factors which affect the likelihood of CVI registration may also be affected by the level of social deprivation, for example, the natural progression of the disease and the response to treatment. Another consideration is patient engagement with the community and HESs, that is, patients with greater deprivation may not seek help and may not request registration. It is also possible that registration has been offered but declined more frequently in patients with greater deprivation. In 1968, Graham et al20 reported a postal survey which estimated that up to 29% of patients who were eligible for visual impairment registration were not registered, and that ‘many who were registrable but non-registered were opposed to registration considering it a form of charity’. More recently, Robinson et al21 reported that 12% of eligible patients declined registration; and there are some findings in this and the subsequent study by Barry and Murray22 to suggest that this might be more common in certain ethnic groups, although the influence of social deprivation was not considered.
As our study only surveys patients from community referral onwards, we can only make inferences about barriers to accessing services before this stage. Similarly, our study only provides data on patients who are registered, and not on those who were eligible but have declined registration. Our experience is that, in contrast with historical studies, patient refusal of registration is now very uncommon, and is therefore unlikely to have significantly contributed to under-representation of particular groups in our study.
Of particular concern is the possibility that our findings reflect earlier mortality in those with greater social deprivation. A review of the IMD 2010 data in the West Midlands by Eayres et al23 demonstrated that the mean life expectancy between Middle Super Output Areas (MSOA) in Birmingham varied from 69.2 to 85.1 years for males and 76.5–87.9 years for females with a strong negative correlation between life expectancy and greater deprivation. It is therefore possible that earlier mortality in the more deprived cohort is contributing to their under-representation in this study.
We also recognise that there is potential discordance between actual social deprivation and inferred deprivation when using the IMD 2007 which looks at postcodes rather than the individual's scores. However, the system has been well validated, and is the preferred tool of the UK government for assessing social deprivation within populations. We avoided potential referral bias, existing because of the presence of other hospitals available to the same population, by ensuring that the catchment area for our patient population accurately reflected the Birmingham reference population.
We found no association between IMD score and the number of DNAs or cancellations. Previous studies have shown that patients of a lower socioeconomic status tend towards a higher rate of non-attendance.24 Our study design only captures data on patients who associate with the HES as they need to attend at least once in order to be CVI registered. It is likely, therefore, that there is a cohort of more socially deprived patients who consistently fail to engage with either community or HESs. This is supported by the significant under-representation of patients with higher social deprivation in this study.
There was no significant correlation between IMD score and the time to first hospital appointment or IMD score and time to first treatment, or the decision not to treat. This is reassuring, as it suggests that the hospital appointment system and the interaction between doctor and patient when deciding whether or not to treat is not biased towards a particular postcode area.
Our study found an association between worse VA at time of presentation and higher level of social deprivation. This is in contrast with Acharya et al25 who found no evidence of an effect. This difference may reflect the greater severity of disease in our sample compared to the Acharya cohort of consecutive presentations of nAMD.
Our study showed that social deprivation and increasing age were associated with worse VA at presentation and shorter time to CVI registration. Given that the level of VA at which registration occurred was not associated with deprivation, it is likely that the ‘earlier’ registration is because of a later presentation rather than a bias by the ophthalmologist towards early registration for those with higher social deprivation or greater age.
Our findings are consistent with the studies which describe late presentation of POAG10 and worse VA at presentation for cataract surgery11 in more socially deprived populations. It is interesting to note that all these common causes of sight loss are painless. In our previous qualitative work on reasons for presentation to the HES, we observed that sight loss alone was not always regarded as a sufficient driver to seek help, whereas, painful sight loss was.26 Therefore, health awareness programmes must emphasise the fact that major causes of sight loss (such as AMD) are painless, but still must be regarded as serious.
Although not the primary objective of this study, the association of increasing age with late presentation and with shorter time to CVI is important, particularly in the context of barriers to accessing eye care. It has been demonstrated that patients greater than 65 years old are less likely to attend free sight testing,27 and that the elderly are more likely to lack awareness of blind registration and the support that comes with this.28 It would appear, therefore, that increasing age (like higher social deprivation) is another significant risk factor for delay in presentation to the HES.
Further studies are required to build upon these initial results and look at different cohorts of patients with AMD, for example, new referrals with nAMD of all visual levels. It would also be interesting to look more closely into the IMD domains and try to identify which elements of social deprivation have the largest effect on outcome. This could also be assessed along with an investigation of factors known to be linked with social deprivation and AMD, for example, smoking.
This is, to our knowledge, the first study to investigate the role of social deprivation in nAMD from initial referral through to CVI registration. Given that nAMD is a cause of preventable sight loss, as per the Public Health Outcome Framework,4 future healthcare policy should recognise that social deprivation acts as a potential barrier to accessing healthcare for patients with this blinding condition. Future studies will look at different cohorts of patients and explore potential solutions for inequality of healthcare provision.
MDT and AKD share the role of senior author.
Contributors HES: substantial contribution to conception and design, acquisition of data, analysis and interpretation of data, drafting and revising the article and final approval of the version to be published. PAM and ML: substantial contributions to study design and acquisition of data. PS: substantial contributions to conception and design, interpretation of data and final approval of the version to be published. MSH: substantial contribution to statistical analysis of data. NG, HP, MDT: substantial contribution to revising the article and final approval of the version to be published. AKD: substantial contribution to conception and design, analysis and interpretation of data, drafting and revising the article and final approval of the version to be published.
Competing interests None.
Ethics approval This study was conducted as an approved service evaluation of the Medical Retina Service of the University Hospitals Birmingham NHS Foundation Trust, UK.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.