Do Black and Asian individuals wait longer for treatment? A survival analysis investigating the effect of ethnicity on time-to-clinic and time-to-treatment for diabetic eye disease

Diabetic retinopathy remains one of the principal causes of vision loss in adults of working age in developed countries, with a considerable health and socioeconomic impact [1]. A systematic review of 35 population-based studies of people with diabetes reported the prevalence of diabetic retinopathy, proliferative diabetic retinopathy (PDR), vision-threatening diabetic retinopathy and diabetic macular oedema (DMO) as 35%, 7%, 10% and 7%, respectively, although the actual rates varied widely among different ethnic groups [2]. In the UK, the incidence of diabetes is six times higher in people of South Asian origin and up to three times higher in people of African and African Caribbean origin [3]. Previous studies have also identified a higher prevalence of diabetic retinopathy and DMO in people of ethnic Black and Hispanic descent compared with people of White descent [4‐7]. It is estimated that up to 95% of vision loss in diabetes is preventable or treatable, if detected early. Hence there is a pressing need to identify at-risk individuals [8].

There are also clinical and wider concerns that individuals from ethnic minority groups, particularly those of Black ethnicity, are treated differently in healthcare settings. Black individuals experience lower quality pain management, delays in receiving care within the Emergency Department, disparities in accessing cardiac tests, diagnoses and procedures, and are less likely to be placed on renal transplant waiting lists compared with their White counterparts [9‐12]. Reasons for this variation are complex and multifactorial but remain poorly understood.

There is also a paucity of data on whether treatment is delayed in ethnic minority groups, and how this impacts health outcomes. The aim of this study was to explore ethnic variation within a metropolitan population by testing time-to-clinic, time-to-treatment and rates of vision loss in individuals referred to hospital with a similar severity of diabetic eye disease.

A total of 2062 referrals for 1798 people with diabetes were made from the screening service to the hospital diabetic eye clinic between 1 October 2013 and 31 December 2017. After applying the eligibility criteria, a final cohort of 1676 people remained. A flow chart outlining the reasons for exclusion is provided in Fig. 1.

Despite a similar burden of disease at referral that might require treatment (Black vs White people, respectively: 8% and 10% at retinopathy grade R3A; 93% and 86% at maculopathy grade M1), time-to-treatment was significantly longer in Black people than in White people. Black people did ultimately have similar treatment rates, but these treatments took longer to occur. However, no association was found between social deprivation and time-to-treatment.

Similarly, despite a similar distribution of routine and urgent referrals, the time to first hospital eye clinic review, from referral, was longer in Asian people compared with their Black or White counterparts. A longer time-to-clinic could be a contributory factor to the reported sixfold higher incidence of diabetic retinopathy in Asian vs non-Asian populations [3]. Unlike their Black counterparts, there was no significant difference in time-to-treatment between Asian and White people. A longer time-to-clinic may result in a greater severity of disease at first presentation, which could lead to expedited treatment. This requires further exploration but might explain why time-to-treatment may not have been significantly delayed in Asian people.

Previous studies have shown that minority ethnic groups are twice as likely to have sight-threatening diabetic retinopathy, with South Asians being three times more likely than White people to be registered blind due to diabetic retinopathy [23‐25]. In this study, ≥10 ETDRS letters was chosen as the minimum threshold for a clinically important vision loss to the individual, while also minimising any confounding generated by test–retest variability. Despite disparities in time-to-clinic and time-to-treatment, there was no statistically significant difference in the rates of vision loss among Black, White and Asian people. However, the proportion of individuals who reached a ≥10 ETDRS letter drop was lower in White individuals (112/495, 23%) compared with their Black (213/866, 25%) and Asian (44/169, 26%) counterparts (ESM Table 4). While these differences are small and difficult to interpret in low numbers, they do suggest that ethnic minorities may be experiencing higher rates of vision loss, in keeping with previously published reports.

The number of individuals who DNA any hospital eye clinic appointment during the study period was only 58 (3%) of the total number of referrals (1798). This is likely to be an underestimate of the total DNA rate, as some individuals may only have attended the hospital after re-invitation to clinic or even re-referral. However, in order to maximise the sample size, individuals who had failed to attend at least one clinic appointment were not excluded from the analysis. Of interest, White people were underrepresented in this DNA cohort compared with other ethnic groups, especially Asians.

Several factors are known to increase hospital non-attendance and loss to follow-up including ethnicity, increasing age, reduced mobility, poor baseline vision and fear of the clinical outcome [26‐29]. Language and cultural barriers may also play an important role, particularly among South Asians, in whom work and family commitments are reported to be prioritised over health-seeking behaviours and such factors might contribute to a delayed initial presentation to clinic [30]. While a long distance between home and the hospital is a recognised factor, this is unlikely to have had a significant impact in this study, given that the DESP clinic is co-located within the main hospital eye clinic.

The burden of comorbid diseases and disability may also influence attendance at clinic and treatment appointments. In the USA, Black people are almost four times as likely to develop renal failure than their White counterparts [31]. Despite only constituting around 13% of the general population, 35% of all patients receiving dialysis for renal failure are of Black origin [31]. Black people not only experience a higher prevalence of type 2 diabetes but also suffer from a disproportionate burden of serious complications, such as hypertension and stroke, with associated disabilities, compared with White people [32]. Competing medical appointments, hospitalisations and transport difficulties could all potentially hinder timely treatment of eye disease.

Variation in the provision of healthcare is widely reported among minority ethnic groups but is poorly understood [9‐12]. Reported factors include a lack of engagement with healthcare providers and physician bias. Higher levels of physician distrust have also been reported among male patients and those from ethnic minorities or low socioeconomic backgrounds [33, 34]. A lack of perceived benefit, particularly in the absence of any visual symptoms in early diabetic eye disease, may also contribute to reduced engagement with hospital eye services and a reluctance to accept treatment, such as laser or intravitreal injections.

Racial stereotypes and unconscious bias are known to influence physician behaviour towards minority ethnic groups, particularly with respect to treatment choices offered [35‐37]. A recent study of over 80,000 patients with type 2 diabetes in England found that Black patients were 50% less likely and Asian patients 15% less likely to be prescribed newer medications such as sodium–glucose cotransporter 2 inhibitors and glucagon-like peptide-1 agonists, compared with their White counterparts [38]. Similarly, Black people with age-related macular degeneration (AMD) were found to be 23% less likely to receive anti-VEGF treatment and 18% less likely to have regular eye examinations for AMD compared with White people [39].

Referrer bias in our study is unlikely as the DESP fundus image graders did not have ethnicity data readily available, and high numbers of Black people were referred to the hospital service. In face-to-face clinic reviews, physician bias is possible and this could be one possible reason for the delayed time-to-treatment seen in Black people; however, this is hard to confirm or quantify.

Limitations of this study include an over-representation of individuals from areas of social deprivation, constraining analysis of the impact of social deprivation on time-to-treatment. As only small numbers of Asian and other non-Black minority ethnic groups were present in this study, we could not explore specific at-risk subgroups, such as people of South Asian and Hispanic ethnicity. A lack of refracted BCVA in all visits may reduce the accuracy of visual acuity measurements but it might be expected to affect all groups similarly and PHVA usually overcomes any uncorrected refractive error. The cause of the fall in visual acuity also could not be determined and while it is reasonable to assume this was most often due to diabetic eye disease, other conditions could co-exist, such as glaucoma and AMD, which are known to vary among ethnic groups [40]. Only individuals who attend annual eye screening were included in this study but this accounts for 81.5% of the total diabetic population living within the London Boroughs of Lambeth and Southwark [22]. Data collection was retrospective but that may be helpful when studying patient and clinician behaviour, as prospective data collection might influence behaviour. No data were collected on those already under the care of hospital eye services and those excluded from annual eye screening due to very low vision. Hence, the true prevalence of low vision among individuals with diabetes is likely to be greater, as previously reported [25].

In summary, despite similar disease severity at the point of referral, this study suggests that Black people were either less likely to be offered treatment in a timely manner or were less likely or unable to accept an offer of treatment. This warrants further investigation and future studies could consider gathering the following qualitative data from individuals with diabetes: (1) their knowledge of the underlying health condition; (2) their beliefs surrounding treatments for diabetic eye disease; and (3) their experience of interactions with ophthalmologists in hospital. It would also be helpful to know if the ethnicity of a patient somehow influences the clinical approach of the attending ophthalmologist. Improved understanding may reduce the apparent inequality of healthcare provision based on a patient’s ethnicity.