Analysis of the prevalence of diabetic retinopathy in a telemedicine screening programme from 2018 to 2023

Abstract

Purpose

Diabetic retinopathy (DR) is the most common ocular complication of diabetes. Screening programmes using fundus examination with retinal imaging are a useful tool for early diagnosis of DR and an improved disease prognosis. The purpose of this study is to analyse the number of patients attended in the DR screening programme of the Castilla y León health system and to estimate DR prevalence and severity.

Methods

An observational study of patients previously registered as diabetic who participated in the screening programme between 2018 and 2028 was conducted. Patients were classified according to their diagnosis in non-evaluable test, non-evaluable patient, no DR, DR (mild, moderate, severe, proliferative, and suspected diabetic macular oedema) and other pathologies.

Results

The total number of consultations in the program up to the present date has been 31,901, increasing from 3,934 in 2018 to 9,508 in 2023. The estimated prevalence of DR is 15.41%. 1.58% of patients were referred to their reference hospitals due to DR or other pathologies requiring urgent treatment. The remaining patients were managed by the program.

Conclusions

The prevalence of DR appears to be slightly lower than that reported in other published series, but no significant differences were observed between different program years, overall assessment, or different forms of DR or suspected diabetic macular oedema.

Resumen

Antecedentes y objetivos

La retinopatía diabética (RD) es la complicación ocular más frecuente de la diabetes. La realización de programas de cribado mediante exploración del fondo de ojo con retinografías permite el diagnóstico precoz de la RD y un mejor pronóstico de la enfermedad. El objetivo de este trabajo es analizar el número de pacientes atendidos en el programa de cribado del Sistema Sanitario de Castilla y León y estimar la prevalencia y grados de severidad de la RD.

Material y métodos

Se realizó un estudio observacional de los pacientes previamente censados como diabéticos que participaron en el programa de cribado entre 2018 y 2023 estableciendo los siguientes resultados: prueba no valorable, paciente no valorable, no RD, RD (leve, moderada, severa, proliferante y sospecha de edema macular diabético) y otras patologías no diabéticas encontradas.

Resultados

El número de consultas realizadas en el programa, hasta la fecha ha sido de 31.901 aumentando desde las 3934 del año 2018 a las 9508 del año 2023. La prevalencia estimada de la RD es del 15,41%. Se han derivado un 1,58% de los pacientes por RD o por otras patologías que requerían tratamiento urgente a sus hospitales de referencia. El resto ha sido controlado por el programa.

Conclusiones

La prevalencia de la RD resulta ser un poco más baja que la de otras series publicadas y no se han observado diferencias significativas entre los diferentes años del programa ni en el cómputo global, ni en las diferentes formas ni en las sospechas de Edema Macular Diabético.

Introduction

Diabetic retinopathy (DR) is the most frequent ocular complication of diabetes¹ and one of the main causes of preventable blindness in the world.^2,3 It is estimated to affect approximately 22.7% of diabetics, with more than 100 million patients with DR worldwide.⁴ Furthermore, it is calculated that the current global prevalence of diabetes is 10.5% of the population between 20 and 79 years old, and this prevalence is expected to go up to 12.2% by 2045,^5,6 so the number of patients with DR will also increase significantly in the coming years.⁴

The most apparent lesions in a fundus examination are microvascular, caused by damage to the endothelial cells of the retinal capillaries due to cellular malfunction resulting from insulin deficiency.⁷ This endothelial malfunction causes saccular dilatations of the capillaries, called microaneurysms, which are usually the first clinical-ophthalmoscopic sign of DR.^7,8

As the condition progresses, intraretinal hemorrhages (caused by ruptures in the capillaries) and exudates (extracellular accumulations of lipids, proteins, and lipoproteins filtered by the malfunctioning capillaries) are usually added as the main clinical signs of DR.⁸ Finally, if there are areas of ischemia, neovascularization phenomena occur, which are the most severe forms. And in almost all stages, fluid can accumulate in the macula, leading to diabetic macular edema (DME), the most common cause of central vision loss.⁹

In general, in many cases, even severe ones, there may not be a significant reduction in vision. Due to this, fundus examination under pharmacological mydriasis is the tool considered the gold standard for the diagnosis of DR.^10,11 Furthermore, this evaluation also allows the classification of DR into degrees of severity.¹²

Early diagnosis and adequate treatment of DR allow for a better prognosis of the disease, thus reducing the probability of visual disability.¹³ Therefore, screening in diabetic patients is an effective tool in the prevention of blindness due to DR.¹⁴

Within these screenings, the use of telemedicine with retinal photographs has proven to be a valid tool with a better cost-effectiveness ratio than traditional screenings performed directly by the ophthalmologist.^15,16 In addition, the use of the Joslin Vision Network (JVN) capture protocol, with 3 retinal images of 45 ° (centered on the macula, superior temporal, and nasal),¹⁷ allows for adequate sensitivity and specificity compared to the in-person fundus examination by a retina specialist ophthalmologist.¹⁸

At the end of 2017, the Castilla y León Health System (SACYL) (Spain) launched a DR screening program. The program is based on the capture of 3 retinal photographs following the JVN protocol using non-mydriatic fundus cameras after pupillary dilation.¹⁸ The fundus cameras are operated by nursing staff previously certified by the Centro de Lectura del Instituto de Oftalmología Aplicada (IOBA-RC), and the retinal photographs are sent to the reading center and evaluated by a team of certified optometrists and ophthalmologists. The implementation of this screening program with a first reading by certified optometrists has proven to be effective and cost-effective.¹⁹ The program began in the provinces of Valladolid and Palencia and has now also been extended to Burgos and Salamanca.

The objective of this work is to conduct a descriptive analysis of the number of patients evaluated in the program during the period 2018–2023, to estimate the prevalence of DR in this population and its degrees of severity, using the international classification of DR.¹² The percentage of patients who could not be adequately photographed, mainly due to the presence of media opacities, has also been assessed.

Material and methods

The SACYL DR screening program began in 2018 with the aim of screening for DR in the diabetic population of Castilla y León through the capture and analysis of fundus images. This screening program includes the type 1 and type 2 diabetic population who are not already on therapy for their DR or in ophthalmological check-ups for any other reason.

Each health area has 1 or 2 capture centers where the fundus camera is located. Currently, Valladolid has 6 capture centers; Palencia, 3; Burgos, 2; and Salamanca has 1. In all areas, Topcon TRC-NW8 models (Topcon, Japan) were installed, except in the 2 rural centers of eastern Valladolid, where Topcon-NW400 is used, and in a rural western Valladolid center where Zeiss Visucam 200 (Carl Zeiss Meditec, Germany) was installed.

The image acquisition protocol, as already indicated, is that of the JVN, taking 3 images of each eye at 45 ° (centered on the macula, superior temporal, and nasal),¹⁸ to which an image of the fundus reflex is added, focused on the pupil, to assess the existence of opacities in the ocular media that may affect the quality of the retinal photographs. The images are captured in the health centers by nursing staff (previously certified in retinal photography capture by the IOBA-RC) after dilating the patients' pupils with one drop of 10 mg/mL tropicamide. The instillation of tropicamide improves the quality of the images obtained.²⁰

Fig. 1 illustrates the flow chart followed by diabetic patients within the DR screening program. Patients are referred by their primary care physician (PCP) to the telemedicine consultation for image capture by certified personnel. These images, along with the clinical history and visual acuity with correction, are sent to the reading center using the RISOftal software (IRE, Spain); they are sent to the IOBA-RC, where they undergo a two-level evaluation. The first level is the reading by IOBA-RC certified optometrists. To obtain this certification, optometrists (the training is also aimed at ophthalmology residents, ophthalmology fellows, nurses, and PCP) receive theoretical and practical training in retinal photograph reading lasting 80 h. Finally, they undergo an evaluation in which they must correctly label 100 retinal photographs, with a maximum of 5% errors allowed to pass the certification process favorably. Certified optometrists classify the images into:

• 1

  Non-evaluable test (because the photographer has taken an incorrect image with poor quality, it is not possible to evaluate the presence or absence of retinal disease). The patient must be recalled for the capture of good quality retinal photographs.

• 2

  Non-evaluable patient (due to the presence of opacities in the anterior pole or insufficient pupil size, it is not possible to evaluate the presence or absence of retinal disease). The patient is scheduled for an in-person ophthalmological evaluation.

• 3

  No diabetic retinopathy (no signs of DR or other diseases are observed). The patient will be scheduled for a review through the program according to the established protocol.

• 4

  Presumably disease (signs of DR and/or other conditions are observed).

Fig. 1.

Flow chart of patients from the screening program.

Only in the presence of disease do the images proceed to a second reading, while in the other 3 cases, the patient receives a report through their PCP.

The second reading is performed by a retina specialist ophthalmologist, who evaluates the images classified as abnormal in the first reading and issues their diagnosis. Cases of severe and proliferative DR and suspected DME, as well as other diseases that the ophthalmologist may consider, are urgently referred to their reference hospital. For their part, patients with mild and moderate DR or other conditions that do not require specialized ophthalmological evaluation are scheduled for review within the screening program following the general protocol.

For the observational study, data from all patients who participated in the screening program from January 2018 through December 2023 were used. Data were stored in Microsoft Access (Microsoft Office, 2019, USA), and statistical analysis was performed with SPSS 27.0 (SPSS, USA) using the chi-square test to analyze differences in diagnoses between the different years of the study, with p < 0.05 considered statistically significant.

Results

Fig. 2 shows the number of consultations performed in the program each year. From 2018 through 2023, the number of consultations performed went up by 141.7%.

Fig. 2.

Representation of consultations performed in the screening program per year from 2018 through 2023.

Table 1 breaks down the number of consultations performed in each health area. Similarly, Fig. 3 illustrates the percentage of each diagnosis in all health areas. The program began in the provinces of Valladolid (divided into 2 rural and 2 urban areas) and Palencia (divided into 1 rural and 1 urban area). In all of them, except for the rural area of Palencia, a significant increase in the number of consultations performed is observed, and with some exceptions, this increase is constant year after year from the start of the program until 2023. The table also shows the extension of the program to the provinces of Burgos (with 2 areas, Burgos capital and Medina de Pomar) and Salamanca (one single area) and the inclusion of the first patients in these provinces in 2023.

Table 1.

Consultations conducted in the diabetic retinopathy screening program by health area.

Health Area	2018	2019	2020	2021	2022	2023
Valladolid East Urban	2,294	2,355	1,254	2,414	2,584	3,147
Valladolid East Rural	390	843	218	187	357	1,259
Valladolid West Urban	582	345	207	516	1,037	1,801
Valladolid West Rural	272	426	66	118	484	423
Palencia Urban	288	939	572	1,228	1,747	1,944
Palencia Rural	108	500	4	0	57	131
Burgos Urban	0	0	0	0	0	468
Medina de Pomar	0	0	0	0	0	275
Salamanca	0	0	0	0	0	60
Total	3,934	5,408	2,321	4,463	6,266	9,508

Fig. 3.

Representation of the percentage of diagnoses in each health area.

Table 2 illustrates the number of diagnoses, and the percentage of the total consultations performed since 2018; 31,901 evaluations of diabetic patients have been conducted, among whom 4,165 (13.06%) were diagnosed with DR of varying severity, and 502 (1.57%) were referred for DR and other diseases to ophthalmology consultations. The prevalence of DR found, without considering non-evaluable tests or other conditions, was 15.41% (Fig. 4), with no significant differences being observed between the different years or between the different degrees of severity.

Table 2.

Summary of diagnoses and their percentage over the total consultations conducted in the diabetic retinopathy screening program from 2018 through 2023.

	2018	2019	2020	2021	2022	2023	p-value
Patients managed	3,934	5,408	2,321	4,463	6,266	9,508
Non-evaluable test	121 (3.08%)	5 (0.09%)	3 (0.13%)	6 (0.13%)	18 (0.29%)	48 (0.50%)	< 0.01
Non-evaluable patient	284 (7.22%)	637 (11.78%)	232 (10.00%)	340 (7.62%)	492 (7.85%)	546 (5.74%)	< 0.01
No diabetic retinopathy	2,605 (66.22%)	3,664 (67.75%)	1,593 (68.63%)	3,229 (72.35%)	4,528 (72.26%)	7,239 (76.14%)	< 0.01
Mild DR	352 (8.95%)	503 (9.30%)	220 (9.48%)	441 (9.88%)	670 (10.69%)	952 (10.01%)	0.05
Moderate DR	64 (1.63%)	127 (2.35%)	53 (2.28%)	73 (1.64%)	113 (1.80%)	139 (1.46%)	< 0.01
Severe DR	19 (0.48%)	22 (0.41%)	16 (0.69%)	26 (0.58%)	35 (0.56%)	41 (0.43%)	0.47
Proliferative DR	6 (0.15%)	10 (0.18%)	2 (0.09%)	7 (0.16%)	10 (0.16%)	9 (0.09%)	0.70
Suspected DME	34 (0.86%)	39 (0.72%)	28 (1.21%)	35 (0.78%)	47 (0.75%)	72 (0.76%)	0.32
Other diseases	449 (11.41%)	401 (7.41%)	174 (7.50%)	306 (6.86%)	353 (5.63%)	462 (4.86%)	< 0.01
Referrals for other diseases	3 (0.08%)	14 (0.26%)	1 (0.04%)	7 (0.16%)	6 (0.10%)	13 (0.14%)	0.10

Diagnoses in bold in the original are those requiring patient referral.

DME, Diabetic Macular Edema; DR, Diabetic Retinopathy; A p-value < 0.05 is considered statistically significant.

Fig. 4.

Representation of patients diagnosed with diabetic retinopathy vs those without diabetic retinopathy per year from 2018 through 2023.

The percentage of non-evaluable tests remained <0.50%, except for 2018, which showed a value of 3.08% (p < 0.01). Regarding non-evaluable patients (generally due to the presence of media opacities), the percentage ranged from 5.74% up to 11.78%. The use of different fundus cameras did not influence the proportion of non-evaluable tests and patients (p = 0.12).

Regarding referrals to reference hospitals, the percentage remained constant over the years (p > 0.32). Among these, the most common were those caused by suspected DME (between 0.75% and 1.21%), followed by severe DR (between 0.41% and 0.69%) and proliferative DR (between 0.09% and 0.18%). The same thing occurs with referrals for other diseases, whose percentage did not show statistically significant differences (p = 0.10), ranging from a minimum of 0.04% in 2020 to a maximum of 0.26% in 2019. Of note, the percentage of patients referred to hospitals is <2%.

In contrast, diagnoses of other ocular diseases showed a decrease from 11.41% in 2018 up to 4.86% in 2023 (p < 0.01). The other main diseases detected in the screening program are AMD, epiretinal membranes, and choroidal nevi. However, knowing the exact percentage of these is complex, as they are not defined in the computer program, and their diagnosis depends on ophthalmological criteria. In turn, the objective of the program is not to screen for ocular disease but to focus on DR.

Discussion

Diabetes is a disease whose general and ocular complications generate a high economic burden for society.²¹ The global incidence rate of diabetes in people between 20 and 79 years old is estimated at around 10.5%,^5,6 being somewhat higher in Spain (14.8%), which is the second European country with the highest prevalence.⁶ Of note, a significant percentage of these diabetic patients do not know they are diabetic.^22,23

Castilla y León is a Spanish autonomous community with a population of 2,383,703 inhabitants, according to the latest census of the National Institute of Statistics (2023).²⁴ This represents an approximate diabetic population of 352,788 people, which is likely to be higher as the prevalence of type 2 diabetes mellitus increases with age.⁶ This number of eye examinations is unmanageable from the traditional point of view of in-person ophthalmological consultations. Furthermore, traditional screening systems have the problem of lack of compliance (adherence) by a significant number of patients.²⁵ In contrast, the use of telemedicine generates good acceptance by patients^26,27 and allows for a better cost-effectiveness ratio.^15,16

This program is based on systematic screening, as there is a census of diabetic patients, at least those diagnosed. Of note, the number of patients managed increases year after year since the start of the program, except for 2020 and 2021 due to the decrease in health care activity caused by the COVID-19 pandemic.²⁸

Although the population prevalence of DR is estimated at around 20% globally,⁴ in this study, the prevalence found was 15.41% (Fig. 4), somewhat similar to the meta-analysis by Romero-Aroca et al.,²⁹ who found a prevalence of 15.28% in Spain. Valpuesta et al.²⁶ reported a slightly higher prevalence (16.4%) in a rural area of this same screening program, although this may be due to the higher mean age of that population area, as they themselves indicate.

The heterogeneity and the different ways of presenting the results of the various screening programs make their comparison difficult. Other similar screening programs conducted in public health systems of other Spanish autonomous communities found that the first reading is done by PCPs with a single retinal photograph centered on the macula, as occurs, for example, in the Canary Islands (Spain). However, in this case, only 30% of the images referred to the ophthalmologist showed DR.³⁰ In Andalusia (Spain), screening is conducted following the JVN protocol, and the first reading is performed by PCPs or endocrinologists, showing a cumulative incidence rate of 11% in 15 years of the program.³¹ In the rural area of western Asturias (Spain), the images are captured by trained nursing staff following the JVN protocol, and the first reading is done by PCPs, finding a DR incidence rate of 12.05%.³²

A recent study on the North American population²³ has shown that the prevalence of DR is 26%, and that those forms with a high risk of blindness approach 5.06%. But once again, the different methodologies make comparative studies very complicated.

Perhaps the best comparison can be drawn with a study by this group conducted in 1993³³ in a rural environment and with different methodologies, yet the incidence rate of DR was 20.9% (95%CI, 19.6–22.3%), which is a higher rate than the current one. Furthermore, a 5.7% rate of DME was detected vs the current values that do not exceed 1%, and severe forms of NPDR and PDR accounted for 3.4% vs <0.5% currently. These data will need to be further investigated, but it may be due to the inclusion of diabetes and DR in the community's health plans.

In the present study, other interesting data can also be observed, such as the decrease over the years in the number of non-evaluable tests, which may be due to the greater experience of the photographers in taking images, or that the diagnoses of other diseases have decreased, which does not have an easy explanation beyond the fact that the population is more aware of their eye check-ups.

The main limitation of this study is that, despite the existence of a census of diagnosed diabetic patients, this is not comprehensive, and it is presumed that many cases are undiagnosed. This gives the screening an opportunistic characteristic and, therefore, makes it difficult to extrapolate our results to the overall diabetic population of Castilla y León, although they may offer an approximation.

The future lines of the program involve extending it to all health areas of the autonomous community, since due to the geographical and population characteristics of the community, Castilla y León could greatly benefit from the implementation of an active screening program across its territory.

In conclusion, we can assume that the prevalence of DR has decreased compared with 30 years ago,³³ and that it remained at stable values across the years of the program, despite the fact that the diabetic population has presumably grown due to aging, among other factors. It can be assured that DR screening through telemedicine is an effective tool for conducting the periodic fundus evaluation of diabetic patients in the context of an extense autonomous community and aging population, allowing the detection of cases that require referral and emergency treatment to reduce waiting lists in ophthalmology services.

Patient consent

To be able to conduct the study, informed consent was obtained from the patients prior to their participation. All data were treated in full compliance with current regulations regarding data protection (Organic Law 3/2018, of December 5, on the Protection of Personal Data and Guarantee of Digital Rights).

Ethical considerations

The study was conducted in full compliance with the World Medical Association's Code of Ethics (Declaration of Helsinki).

Funding

Study funded by the Castilla y León Health System (SACYL), through the direct granting of a subsidy to Fundación Universidad of Valladolid for the development of the diabetic retinopathy screening program.