TYPE Original Research PUBLISHED 16 August 2022 DOI 10.3389/fmed.2022.945245 Diabetic retinopathy as a OPEN ACCESS predictor of cardiovascular EDITED BY Anna Maria Roszkowska, University of Messina, Italy morbidity and mortality in REVIEWED BY Hsin-Hua Chen, subjects with type 2 diabetes Taichung Veterans General Hospital, Taiwan Adisu Birhanu Weldesenbet, Joan Barrot1,2,3,4 , Jordi Real2,3 , Bogdan Vlacho2,5*, Haramaya University, Ethiopia Pedro Romero-Aroca6 , Rafael Simó7,8,9 , Didac Mauricio2,9,10,11 , *CORRESPONDENCE Bogdan Vlacho Manel Mata-Cases2,9,12 , Esmeralda Castelblanco2,13 , bogdan.vlacho@gmail.com Xavier Mundet-Tuduri2,4* and Josep Franch-Nadal2,9,14 Xavier Mundet-Tuduri mundetx@gmail.com 1 Primary Health Care Center Dr. Jordi Nadal i Fàbregas (Salt), Gerència d’Atenció Primària, Institut Català de la Salut, Girona, Spain, 2 Diabetis des de l’Atenció Primária (DAP)-Cat Group, Unitat SPECIALTY SECTION de Suport a la Recerca Barcelona, Fundació Institut Universitari per a la Recerca a l’Atenció Primària This article was submitted to de Salut Jordi Gol i Gurina (IDIAPJGOL), Barcelona, Spain, 3 Fundació Institut Universitari per a la Ophthalmology, Recerca a l’Atenció Primària de Salut Jordi Gol i Gorina (IDIAPJGOL), Barcelona, Spain, a section of the journal 4 Departament of Medicine, Universitat Autònoma de Barcelona, Barcelona, Spain, 5 Hospital de la Frontiers in Medicine Santa Creu i Sant Pau, Barcelona, Spain, 6 Ophthalmology Service, University Hospital Sant Joan, RECEIVED 16 May 2022 Institut de Investigacio Sanitaria Pere Virgili (IISPV), University of Rovira and Virgili, Reus, Spain, 7 ACCEPTED 26 July 2022 Diabetes and Metabolism Research Unit, Department of Endocrinology, Vall d’Hebron University PUBLISHED 16 August 2022 Hospital, Vall d’Hebron Research Institute, Barcelona, Spain, 8 Department of Medicine, Faculty of Medicine, Autonomous University of Barcelona, Barcelona, Spain, 9 Centro de Investigación CITATION Biomédica en Red de Diabetes y Enfermedades Metabólicas Asociadas (CIBERDEM), Barcelona, Barrot J, Real J, Vlacho B, Spain, 10 Department of Endocrinology and Nutrition, Hospital de la Santa Creu i Sant Pau, Romero-Aroca P, Simó R, Mauricio D, Barcelona, Spain, 11 Departament of Medicine, University of Vic—Central University of Catalonia, Vic, Mata-Cases M, Castelblanco E, Spain, 12 Centre d’Atenció Primària La Mina, Gerència d’Àmbit d’Atenció Primària de Barcelona, Mundet-Tuduri X and Franch-Nadal J Institut Català de la Salut, Barcelona, Spain, 13 Division of Endocrinology, Metabolism and Lipid (2022) Diabetic retinopathy as Research, John T. Milliken Department of Medicine, School of Medicine, Washington University a predictor of cardiovascular morbidity in St. Louis, St. Louis, MO, United States, 14 Primary Health Care Center Raval Sud, Gerència d’Àmbit and mortality in subjects with type 2 d’Atenció Primaria, Institut Català de la Salut, Barcelona, Spain diabetes. Front. Med. 9:945245. doi: 10.3389/fmed.2022.945245 COPYRIGHT © 2022 Barrot, Real, Vlacho, This study aimed to evaluate the predictive value of diabetic retinopathy Romero-Aroca, Simó, Mauricio, (DR) and its stages with the incidence of major cardiovascular events and Mata-Cases, Castelblanco, Mundet-Tuduri and Franch-Nadal. This all-cause mortality in type 2 diabetes mellitus (T2DM) persons in our large is an open-access article distributed primary healthcare database from Catalonia (Spain). A retrospective cohort under the terms of the Creative Commons Attribution License (CC BY). study with pseudo-anonymized routinely collected health data from SIDIAP The use, distribution or reproduction in was conducted from 2008 to 2016. We calculated incidence rates of other forums is permitted, provided major cardiovascular events [coronary heart disease (CHD), stroke, or both— the original author(s) and the copyright owner(s) are credited and that the macrovascular events] and all-cause mortality for subjects with and without original publication in this journal is DR and for different stages of DR. The proportional hazards regression analysis cited, in accordance with accepted academic practice. No use, distribution was done to assess the probability of occurrence between DR and the study or reproduction is permitted which events. About 22,402 T2DM subjects with DR were identified in the database does not comply with these terms. and 196,983 subjects without DR. During the follow-up period among the subjects with DR, we observed the highest incidence of all-cause mortally. In the second place were the macrovascular events among the subjects with DR. In the multivariable analysis, fully adjusted for DR, sex, age, body mass index (BMI), tobacco, duration of T2DM, an antiplatelet or antihypertensive drug, and HbA1c, we observed that subjects with any stage of DR had higher risks for all of the study events, except for stroke. We observed the highest probability Frontiers in Medicine 01 frontiersin.org Barrot et al. 10.3389/fmed.2022.945245 of all-cause death events (adjusted hazard ratios, AHRs: 1.34, 95% CI: 1.28; 1.41). In conclusion, our results show that DR is related to CHD, macrovascular events, and all-cause mortality among persons with T2DM. KEYWORDS diabetic retinopathy, macrovascular complication, primary healthcare, real word data analyses, mortality Introduction risk of mortality from all causes compared to subjects with diabetes who do not have retinopathy (RR: 2.33, 95% CI: 1.92; Diabetes mellitus has become one of the major public 2.81) (30–32). health challenges globally, both in developed and developing So far, to the best of our knowledge, there is a lack of studies countries (1). The increasing prevalence of diabetes worldwide is evaluating the relation between DR and major cardiovascular caused by a complex interplay of socioeconomic, demographic, events from primary healthcare settings. In this study, we aimed environmental, and genetic factors. The International Diabetes to evaluate the predictive value of DR, its severity with the Federation (IDF) estimates that 10.5% of adults between 20 and incidence of major cardiovascular events [coronary heart disease 79 years have diabetes, equating to 537 million people (2). The (CHD) and stroke], and all-cause mortality in subjects with IDF also estimates that 643 million adults will live with diabetes T2DM in a Mediterranean region. in 2030 (11.3% of the population), and the number will reach 783 million (12.2%) by 2045. The main burden of diabetes results from its complications. Materials and methods Diabetic retinopathy (DR) is a microvascular and neurodegenerative complication whose prevalence increases The study included a retrospective cohort of subjects with with disease duration and causes a high risk of severe visual T2DM from the SIDIAP database (Sistema de Información impairment and blindness (3, 4). According to recently para el desarrollo de la Investigación en Atención Primaria). published studies, there is considerable variability in the The SIDIAP database routinely collects pseudo-anonymized prevalence of DR. The meta-analysis of Yau et al. reports that health data from users who attend the primary healthcare one in every three subjects with diabetes will present some centers of the Catalonian Health Institute (Institut Català de degree of DR (5) compared to 27% in the last review and la Salut, ICS). The ICS is the primary healthcare provider the analysis of the IDF from the studies published in recent in Catalonia (Spain), covering about 80% (5,564,292 persons) years (6). of the Catalonian population. The SIDIAP database contains Diabetes can cause numerous complications that weaken comprehensive patient data, such as visits with healthcare health, lower quality of life, and cause early death. People professionals, diagnoses, demographic information, clinical with diabetes are two to three times more likely to develop variables, laboratory test results, prescriptions, referrals to cardiovascular disease, and the risk of death doubles compared specialists and hospitals, and medication obtained from to people without diabetes (7). pharmacies. For this analysis, data were extracted covering The consequences of late detection of DR go beyond a 10-year period. The inclusion period was defined from 1 the resulting suboptimal visual acuity (8, 9). We have January 2008 to 31 December 2012. The follow-up period was extensive evidence that associates DR with other micro- and until the data extraction end date (31 December 2017) or a macrovascular complications of diabetes. Recently, it was discontinuation event (death or any other database dropout). reported that DR is associated with subclinical atherosclerosis (10–14), macrovascular comorbidities such as coronary disease (15–20) and cerebrovascular accident (21–26). Of interest, Participants some studies have reported an association between cognitive impairment and the incidence of dementia [risk ratio (RR), All subjects aged between 30 and 89 years with a diagnosis 1.3; 95% confidence interval (CI), 1.27–1.58] (27, 28), and the of T2DM, defined as the presence of International Classification relation between DR and neurodegeneration diseases such as of Diseases, 10th Revision (ICD-10) diagnostic codes: E11, E14 Parkinson’s disease has been proposed but remains unclear (29). (and their subcodes), and screening for fundus photography In addition, a recent meta-analysis with observational studies (to determine the presence of DR), were included in the found that subjects with diabetes who have DR have an increased analysis. We excluded those subjects with other types of Frontiers in Medicine 02 frontiersin.org Barrot et al. 10.3389/fmed.2022.945245 diabetes (type 1 diabetes, gestational diabetes, secondary, or Statistical analysis other), without diagnostic codes for T2DM, subjects with cardiovascular events, and/or absence of fundus photography. A descriptive statistical analysis was carried out, The eligible participants were followed up for at least 5 years or summarizing the quantitative parameters with the mean until the discontinuation event. and its standard deviation (SD), median or interquartile range, and the qualitative variables with frequency and percentage. We used an opportunistic sampling technique to capture all Variables persons meeting the study inclusion criteria. To assess the association of the main study events considering the time of For all the included subjects, the presence or absence follow-up, a time-to-event complete case analysis (CCA) was of DR was assessed using reports pertaining to digital 45◦ performed, adjusting Cox proportional hazards models (34). color fundus images. Two photographs were taken for Unadjusted hazard ratios (UnAHRs) and adjusted hazard ratios each eye, the macula-centered and between the macula and (AHRs), 95% CI, and p-value were estimated and summarized. the optic nerve. We only included fundus photography We used Kaplan–Meier survival curves to graphically visualize from patients screened at primary healthcare centers. the cumulative incidence for study events during the follow-up DR was classified into different stages according to the period in each group. Different study events were fully adjusted Early Treatment Diabetic Retinopathy Study (ETDRS) considering potentially confounding clinical variables such as classification (33) as no apparent retinopathy (NDR), mild DR, sex, age, BMI, tobacco, duration of T2DM, antiplatelet non-proliferative retinopathy (NPDR), moderate NPDR, or antihypertensive drug treatment, and HbA1c. Additionally, severe NPDR, proliferative diabetic retinopathy (PDR), and we performed a model adding lipid-lowering and antidiabetic diabetic macular edema (DME). The DR diagnosis was drugs to the previous model. To treat missing data, a multiple taken from the worst-affected eye, and from the most recent imputation analysis (MICE) was performed with the statistical photograph in case there was more than one screening during package (35) using ten replicates and five iterations. The the inclusion period. Also, at inclusion, we collected socio- CCA and MICE analyses were compared in a sensitivity demographic variables (age, gender), toxic habits (current analysis, adjusting the models for different variables (Model tobacco use), and clinical variables related to diabetes [age 1: unadjusted; Model 2: adjusted for age and sex; Model 3: at diagnosis of diabetes, diabetes duration, and glycated adjusted for sex, smoking, antiplatelet or antihypertensive hemoglobin levels (HbA1c)]. Data on cardiovascular risk drug treatment, and BMI; Model 4: fully adjusted). P-values factors [body mass index (BMI), blood lipids, total cholesterol, less than 0.05 were considered statistically significant without low-density lipoprotein (LDL) cholesterol, high-density using the correction for multiple comparisons for the multiple lipoprotein (HDL) cholesterol, non-HDL cholesterol, blood events analyzed. We used the cox.zph function in the survival pressure, pulse pressure] were collected. Additional data package (R statistical software) to check the proportional were gathered on medication. Obesity was defined as a hazards assumption of Cox models (34). Statistical analysis was BMI ≥ 30 kg/m2 . The estimated glomerular filtration rate performed with R statistical software, version 3.6.1.1 (eGFR) was calculated according to the CKD Epidemiology Collaboration (CKD-EPI) equation. Chronic kidney disease (CKD) was defined as an estimated glomerular filtration rate Institutional review board statement (eGFR) < 60 ml/min/1.73 m2 calculated using the CKD-EPI equation and/or a ratio of albumin/creatinine (CAC) in This study was approved by the Institutional Review Board urine ≥ 30 mg/g. (or Ethics Committee) of IDIAP Jordi Gol i Gurina Foundation During the follow-up period, we collected data on (protocol code P13/028 and date of approval 03/04/2013). mortality by any cause and by severe cardiovascular events such as stroke (defined by ICD-10 codes and subcodes for “cerebral infarctions” and/or “transient cerebral ischemic Results attacks and related syndromes”) or CHD defined by ICD-10 codes and subcodes for “angina pectoris” and/or At the end of the inclusion period, 219,385 (86.5%) subjects “acute myocardial infarction and/or subsequent myocardial met the study inclusion criteria and were included in the infarction” and/or “certain current complications following analysis. We identified 22,402 (10.2%) persons with any stage acute myocardial infarction” and/or “chronic ischemic heart DR. Figure 1 shows the study flow diagram. The mean age disease.” Additionally, we created a composite event as a (SD) of the subjects was 64.6 (± 11.6) years; the cohort combination of stroke and/or CHD, named as “macrovascular events.” The definition of the study variables and codes are summarized in Supplementary Table 1. 1 https://www.r-project.org Frontiers in Medicine 03 frontiersin.org Barrot et al. 10.3389/fmed.2022.945245 FIGURE 1 Study flow chart. contained a slightly higher proportion of men than women higher proportion (28.3%) of subjects with LDL < 100 mg/dl (55.7%) (Table 1). and higher mean HDL (HDL cholesterol). Subjects with DR Compared to those with normal fundus photography, had a poorer renal profile than subjects without DR (eGFR, subjects who had DR (any stage) were older and had 1.6 70.9 (22.4) ml/min/1.73 m2 vs. 78.8 (19.7) ml/min/1.73 m2 , times longer diabetes duration. Moreover, the DR group had respectively). The proportions of subjects (32.4%) with a smaller proportion of smokers and a higher proportion eGFR < 60 ml/min/1.73 m2 were higher than those without (53.4%) of subjects with BMI < 30 kg/m2 . Furthermore, the DR. We observed 1.6 times higher CKD prevalence among lipid profile was better in the DR group, with lower total the subjects with DR than the subjects without DR. Regarding cholesterol, triglycerides, and LDL cholesterol levels, and a good glycemic control (HbA1c < 7%), statistically significant Frontiers in Medicine 04 frontiersin.org Barrot et al. 10.3389/fmed.2022.945245 differences were observed among the groups, in favor of subjects shortest time until the event was observed among the subjects without DR (61.3%) compared with subjects with DR (42.0%). with CHD and DR. In the bivariate analysis, we observed statistically significant, unadjusted hazard ratios for all of the events, with the risk for the Association between diabetic study events being higher among the subjects with DR. When retinopathy and study events stratified for the stage of DR, the same tendency was observed. The highest risk for the study events was observed among the Table 2 shows the epidemiology for different study events subjects with proliferative DR. We did not observe statistically for both subjects with or without DR. During the follow- significant HRs among the persons with DME and stroke events. up period among the subjects with DR, mortality (all cause) In the multivariable CCA, fully adjusted for DR, sex, was the most common event, followed by macrovascular age, BMI, tobacco use, duration of T2DM, an antiplatelet or events, CHD, and then stroke. The same pattern was observed antihypertensive drug, and HbA1c, we observed that subjects among persons without DR, but with a lower incidence. The with any stage of DR had higher risks for all of the study events. TABLE 1 Clinical characteristics of the subjects at inclusion. All N = 219,385 Without diabetic With diabetic retinopathy N = 196,983 retinopathy N = 22,402 Male sex (%) 122,280 (55.7) 109,984 (55.8)* 12,296 (54.9)* Age (years), mean (SD) 64.6 (11.6) 64.4 (11.5)* 67.0 (11.6)* Current tobacco use (%) 58,602 (26.0) 52,839 (28.1)* 5,763 (26.9)* Clinical variables, mean (SD) Diabetes duration (years) 5.18 (5.27) 4.84 (4.95)* 8.10 (6.81)* BMI (kg/m2 ) 30.6 (5.22) 30.7 (5.20)* 30.2 (5.33)* SBP (mmHg) 134 (14.8) 134 (14.5)* 137 (16.3)* DBP (mmHg) 77.1 (9.59) 77.3 (9.51)* 76.0 (10.2)* Heart rate (beat/min) 76.5 (12.3) 76.4 (12.2)* 77.4 (12.4)* Laboratory variables, mean (SD) HbA1c (%) 7.18 (1.53) 7.12 (1.49)* 7.76 (1.75)* Total cholesterol (mg/dL) 198 (40.3) 198 (40.1)* 192 (41.5)* HDL cholesterol (mg/dL) 49.0 (12.8) 49.0 (12.8)* 49.5 (13.4)* LDL cholesterol (mg/dL) 117 (33.8) 118 (33.7)* 112 (34.4)* Triglycerides (mg/dL) 168 (122) 169 (123)* 160 (112)* GFR (CKD-EPI; ml/min/1.73 m2 ) 78.0 (20.1) 78.8 (19.7)* 70.9 (22.4)* Albumin-to-creatinine ratio 32.9 (131) 28.9 (114)* 67.5 (226)* Comorbidities (%) Dyslipidemia 104,442 (47.6) 93,347 (47.4)* 11,095 (49.5)* Hypertension 132,756 (60.5) 117,830 (59.8)* 14,926 (66.6)* CKD 31,584 (14.4) 26,623 (13.5)* 4,961 (22.1)* Concomitant treatment (%) Antihypertensive drugs 131,967 (60.2) 117,143 (59.5)* 14,824 (66.2)* Antiplatelet drugs 52,396 (23.9) 44,581 (22.6)* 7,815 (34.9)* Antidiabetics drugs 160,166 (73.0) 141,309 (71.9)* 18,857 (84.2)* Lipid-lowering drugs 103,059 (47.0) 92,143 (46.8)* 10,916 (48.7)* BMI, body mass index; CKD, chronic kidney disease; GFR, glomerular filtration rate; HbA1c, glycosylate hemoglobin; SD, standard deviation. *p < 0.001. TABLE 2 The overall incidence for the study events between both groups. Events Groups of Patients/year Event-free survival time Events Event rate (1,000 subjects (median in years) (number) patients/year) Mortality No DR 1017964.6 4.950 17,527 17.2 DR 112524.5 4.750 3,364 29.9 CHD No DR 998687.0 4.846 6,360 6.3 DR 109527.5 4.586 1,058 9.7 Stroke No DR 1006859.7 4.893 3,637 3.6 DR 110843.5 4.652 563 5.1 Macrovascular No DR 988167.7 4.786 9,758 9.9 DR 107933.7 4.498 1,584 14.7 CHD, coronary heart disease; stroke, macrovascular: CHD and/or stroke; event rate, incidence per 1,000 person-year. Frontiers in Medicine 05 frontiersin.org Barrot et al. 10.3389/fmed.2022.945245 The probability of an event occurring was highest for all-cause cardiovascular events at inclusion, during the follow-up of death (AHR: 1.34, 95% CI: 1.28; 1.41) and lowest for stroke 7.8 years, the authors observed that the presence of DR was (AHR: 1.09, 95% CI: 0.97; 1.24). Similar results were observed associated with the occurrence of CHD events (HR: 2.07, when adding the lipid-lowering and antidiabetic drugs to the 95% CI: 1.38; 3.11) and fatal CHD (HR: 3.35, 95% CI: additional model. The Kaplan–Meier survival curves for study 1.40; 8.01) (16). Compared with the results observed in our events are shown in Figure 2, and the results for the unadjusted study, during the follow-up over an average of 4.8 years, we and adjusted HR are summarized in Table 3. observed lower but statistically significant adjusted risks for CHD (HR:1.27, 95% CI: 1.16; 1.39). In a recently reported systematic review and meta-analysis of cohort studies, including Sensitivity analysis 17,611 patients without a previous history of CV disease at baseline, DR was associated with an increased risk of Figure 3 and Supplementary Table 2 show the HR observed CV disease (relative risk (RR): 2.42, 95% CI: 1.77; 3.31) in in the sensitivity analysis. In the MICE, similar HRs for different diabetes (17). This risk was especially elevated among the study events were observed to those from the CCA. When T1DM persons (RR: 3.59, 95% CI: 1.79; 7.20) compared with comparing the two analyses for different adjusted models, we T2DM persons (RR: 1.81, 95% CI: 1.47; 2.23) (17). Regarding observed that HRs for the different events in the MICE analysis the severity of DR, only a few studies have evaluated the were going in the same direction as the CC analysis. association between the stage of DR and CV events. The studies done by Gimeno-Orna et al. and Targher et al. reported an increased risk between NPDR and RDP and the incidence Discussion of non-fatal or fatal cardiovascular events, independently of other known cardiovascular risk factors (HR:1.71, 95% CI: The results of our analysis with data collected from primary 1.1; 2.66 and HR: 2, 95% CI: 1.1; 3.56, respectively) (18, healthcare centers in Catalonia show a positive association 19). Additionally, the Japan Diabetes Complications Study between the DR and major cardiovascular events (CHD and (JDCS) that included 2,033 subjects with T2DM with an 8- stroke) and all-cause mortality. year follow-up period found an increased risk for CHD even Coronary heart disease, stroke, and DR have common during the initial stages of DR (mild-to-moderate NPDR) risk factors, such as hyperglycemia, hyperlipidemia, and after adjusting for traditional CV risk factors (HR:1.69, hypertension. Several studies have observed an association 95% CI: 1.17; 2.97) (20). Our study observed a similar between DR and macrovascular disease and subclinical tendency but with a lower HR for NPDR and CHD in a atherosclerosis (10–14). Some studies have reported that DR fully adjusted model. is a strong determinant of carotid intima-media thickness Evidence suggests that subjects with diabetes and DR and arterial stiffness in patients with T2DM (12, 13). In the seem to be at a high risk of ischemic stroke, with a Atherosclerosis Risk in Communities (ARIC) cohort study meta-analysis that included 19 observational cohort studies based on a population of 1,524 people with T2DM without of 81,452 diabetic patients reporting that the presence of FIGURE 2 Survival curves and different study events and stages of diabetic retinopathy. (A1) All-cause mortality; (B1) coronary heart disease; (C1) stroke; (D1) macrovascular complications; (A2) all-cause mortality and different states of DR; (B2) coronary heart disease and different states of DR; (C2) stroke and different states of DR; (D2) and different states of DR and different states of DR. Frontiers in Medicine 06 frontiersin.org Barrot et al. 10.3389/fmed.2022.945245 TABLE 3 Hazard ratios for events among the study groups complete cases analysis. General Stage of DR Group Group with Non-proliferative Proliferative Diabetic macular Unknown stage* without DR DR diabetic retinopathy diabetic edema (DME) N = 2,216 N = 196,983 N = 22,402 (NPDR) N = 19,048 retinopathy (PRD) N = 475 N = 663 Mortality, n (%) 17,527 (8.90) 3,364 (15.0) 2,794 (14.7) 144 (21.7) 76 (16.0) 350 (15.8) Unadjusted HR 95% CI Ref. 1.75 [1.69;1.82] 1.69 [1.62;1.76] 2.82 [2.39;3.32] 2.44 [1.95;3.06] 1.91 [1.72;2.12] [LL; UL] **Adjusted HR 95% CI Ref. 1.34 [1.28;1.41] 1.34 [1.27;1.42] 1.68 [1.31;2.15] 1.71 [1.27;2.31] 1.19 [1.03;1.38] [LL; UL] ***Adjusted HR 95% CI Ref. 1.34 [1.27;1.41] 1.34 [1.26;1.41] 1.66 [1.30;2.13] 1.69 [1.25;2.28] 1.19 [1.03;1.38] [LL; UL] CHD, n (%) 6,360 (3.23) 1,058 (4.72) 904 (4.75) 38 (5.73) 22 (4.63) 94 (4.24) Unadjusted HR 95% CI Ref. 1.84 [1.70;1.99] 1.51 [1.41;1.62] 2.00 [1.45;2.75] 1.82 [1.20;2.76] 1.39 [1.13; 1.70] [LL; UL] **Adjusted HR 95% CI Ref. 1.27 [1.16;1.39] 1.30 [1.19;1.43] 1.17 [0.71;1.92] 0.80 [0.38; 1.68] 1.11[0.86;1.44] [LL; UL] ***Adjusted HR 95% CI Ref. 1.28 [1.17;1.40] 1.31 [1.19;1.44] 1.18 [0.72;1.92] 0.81 [0.38; 1.69] 1.11[0.86;1.44] [LL; UL] Stroke, n (%) 3,637 (1.85) 563 (2.51) 468 (2.46) 26 (3.92) 9 (1.89) 60 (2.71) Unadjusted HR 95% CI Ref. 1.75 [1.64;1.86] 1.36 [1.24;1.50] 2.41 [1.64;3.55] 1.30 [0,68;2.50] 1.56 [1.20;2.01] [LL; UL] **Adjusted HR 95% CI Ref. 1.09 [0.97;1.24] 1.11 [0.97;1.26] 1.65[0.95;2.86] 0.74 [0.27;1.99] 0.93[0.65;1.33] [LL; UL] ***Adjusted HR 95% CI Ref. 1.09 [0,97;1,24] 1.11 [0.97;1.26] 1.65[0.95;2.85] 0.74 [0.28;1.98] 0.93[0.65;1.33] [LL; UL] Macrovascular, n (%) 9,758 (4.95) 1,584 (7.07) 1,341 (7.04) 61 (9.02) 30 (6.32) 152 (6.86) Unadjusted HR 95% CI Ref. 1.75 [1.64; 1.86] 1.47 [1.39; 1.55] 2.12 [1.65; 2.73] 1.62 [1.13; 2.31] 1.48 [1.26; 1.73] [LL; UL] **Adjusted HR 95% CI Ref. 1.22 [1.13; 1.31] 1.24 [1.15;1.34] 1.29 [0.88; 1.89] 0.79 [0.44; 1.44] 1.08 [0.87;1.33] [LL; UL] ***Adjusted HR 95% CI Ref. 1.22 [1.14; 1.31] 1.25 [1.15;1.35] 1.30 [0.89; 1.90] 0.80 [0.44; 1.44] 1.08 [0.88;1.33] [LL; UL] DR, diabetic retinopathy; CHD, coronary heart disease; NDR, no apparent diabetic retinopathy; NPDR, non-proliferative diabetic retinopathy; PDR, proliferative diabetic retinopathy; DME, diabetic macular edema; ref, reference group; HR, hazard ratio; 95% CI, 95% confidence interval; LL, lower limit; UL, upper limit. *Subjects having diabetic retinopathy by diagnostic code but without fundus photography/stage of DR. **Adjusted for DR, sex, age, BMI, tobacco, duration of T2DM, an antiplatelet or antihypertensive drug, and HbA1c. ***Adjusted for DR, sex, age, BMI, tobacco, duration of T2DM, an antiplatelet, antihypertensive, lipid-lowering, or antidiabetic drugs, and HbA1c. DR was associated with an increased risk of stroke (HR: methodological differences or diagnostic codes used to identify 1.25, 95% CI: 1.12; 1.39). Subgroup analysis for the type the cases with stroke. of diabetes yielded a pooled HR of 1.29 and 95% CI: Regarding all-cause mortality, our study shows that the 1.10; 1.50 in T2DM (21). In a secondary analysis of presence of any degree of DR is correlated with an increased patients enrolled in the Action to Control Cardiovascular risk of this event. In a study from the United States with Risk in Diabetes (ACCORD) Eye Study in an adjusted Cox 4,777 adults, the authors found that those persons with mild regression model, DR was independently associated with and moderate/severe retinopathy had an increased risk of all- incident stroke (HR:1.52, 95% CI:1.05; 2.20) (22). Another cause mortality (HR: 1.81, 95% CI: 1.29; 2.55 and HR: 4.14, prospective cohort study with 1,617 middle-aged people with 95% CI: 1.77; 9.69, respectively) (30). In another meta-analysis diabetes observed that during a follow-up of 7.8 years, DR from ten observational studies with 11,239 diabetic patients, was associated with an increased risk of ischemic stroke the authors reported a doubling in the risk of mortality due (HR: 2.34, 95% CI: 1.13; 4.86) (24). In our study, we did to CVD in subjects with diabetes and severe DR (31). Similar not observe associations between DR or stage of DR and results were also obtained in the analysis of other studies stroke in fully adjusted models, which could be due to the with similar follow-up periods in a meta-analysis of 19 studies Frontiers in Medicine 07 frontiersin.org Barrot et al. 10.3389/fmed.2022.945245 FIGURE 3 Hazard ratios for cardiovascular events and mortality according retinopathy Sensitivity analysis. M1, unadjusted; M2, sex, age; M3, sex, smoking, diabetes, antiaggregants, antihypertensive, BMI; M4, sex, age, smoking, diabetes, duration, antiaggregants, antihypertensive, HBA1C, BMI; M5, age, smoking, diabetes, duration, antiaggregants, antihypertensive, antidiabetes, libid-lowering drugs, HBA1C, BMI. encompassing 142,625 participants; the risk ratio (RR) for all- to define each study event. Another limitation is the numerical cause mortality with DR was 2.33 (95% CI: 1.92; 2.81) compared imbalance between the different groups due to the study design to subjects without DR. According to the different degrees and lack of randomization. Moreover, we should acknowledge of DR in subjects with T2DM, the RR of all-cause mortality there is an inherent possibility of a certain risk of detection bias varied. The RDNP risk was 1.38 (1.11–1.70), while the risk of related to the frequency of medical visits. To minimize this bias, PDR was 2.32 (1.75; 3.06) (32). Our analysis only observed we designed the study to include only DM subjects who had statistically significant risks for all-cause mortality stratified at least one visit. We included the subjects in the study on the for different stages of DR; this risk was highest among the day of diabetic eye screening. Therefore, all the subjects were persons who had PDR. active users of the system. The main strength of our study was Our study has some potential limitations inherent in the large study population which increases the statistical power observational studies with routinely collected healthcare data. and external validity. Similarly, the results of our study confirm Firstly, missing data for the study variables is an important the results of previous studies. limitation. We used pseudo-anonymized routinely collected In conclusion, our results show that DR is associated with health data, where patients were visited as part of the regular an increased risk of cardiovascular disease and death among healthcare surveillance. To test the effect of this limitation, we persons with T2DM. These findings indicate the importance of did a CCA and multiple imputations of missing data, and the early screening, identification, and proper treatment of subjects results were consistent. Due to the database characteristics, there with DR to reduce the risk of macrovascular disease and death. is a possibility of registering errors related to the diagnostic code Further functional studies are needed to evaluate the biological for DR. Therefore, we used the fundus photography results and background of these complications of diabetes and the potential combined them with diagnostic codes. This limitation could also use of stages of DR as an early marker for major adverse be applied to the study events. We used a wide spectrum of codes cardiovascular events. Frontiers in Medicine 08 frontiersin.org Barrot et al. 10.3389/fmed.2022.945245 Data availability statement received advisory and/or speaking fees from Astra-Zeneca, Ascensia, Boehringer Ingelheim, GSK, Lilly, MSD, Novartis, The data analyzed in this study is subject to the following Novo Nordisk and Sanofi; he has received research grants licenses/restrictions: Restrictions apply to the availability of to the institution from Astra-Zeneca, GSK, Lilly, MSD, some or all data generated or analyzed during this study Novartis, Novo Nordisk, Sanofi and Boehringer. JF-N has because they were used under a license. The corresponding received advisory and/or speaking fees from Astra-Zeneca. author will, on request, detail the restrictions and any Ascensia, Boehringer Ingelheim, GSK, Lilly, MSD, Novartis, conditions under which access to some data may be provided. Novo Nordisk and Sanofi; he has received research grants to Requests to access these datasets should be directed to XM-T, the institution from Astra-Zeneca, GSK, Lilly, MSD, Novartis, mundetx@gmail.com. Novo Nordisk, Sanofi and Boehringer. XM-T has received advisory and/or speaking fees from Boehringer Ingelheim, Lilly, MSD, Novartis. MM-C has received advisory honorarium Ethics statement from Astra-Zeneca, Bayer, Boehringer Ingelheim, GSK, Lilly, MSD, Novartis, Novo Nordisk, and Sanofi; he has received This study was approved by the Institutional Review Board speaker honorarium from Astra-Zeneca, Bayer, Boehringer (or Ethics Committee) of IDIAP Jordi Gol i Gurina Foundation Ingelheim, GSK, Lilly, Menarini, MSD, Novartis, Novo Nordisk, (protocol code P13/028 and date of approval 03/04/2013). and Sanofi; he has received research grants to the institution Written informed consent for participation was not required for from Astra-Zeneca, GSK, Lilly, MSD, Novartis, Novo Nordisk this study in accordance with the national legislation and the and Sanofi. institutional requirements. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential Author contributions conflict of interest. JB: conceptualization. XM-T: methodology and resources. JR: software, formal analysis, and data curation. JR, BV, and JB: validation. JB and BV: writing—original draft preparation and Publisher’s note review and editing. DM, PR-A, RS, JR, EC, JB, MM-C, and JF-N: supervision. All authors have read and agreed to the published All claims expressed in this article are solely those of the version of the manuscript. authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or Acknowledgments claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. We thank Amanda Prowse for her editorial support of this manuscript. Supplementary material Conflict of interest The Supplementary Material for this article can be JB has received speaking fees from Boehringer Ingelheim, found online at: https://www.frontiersin.org/articles/10.3389/ Astra-Zeneca, Lilly, MSD, Novo Nordisk, Sanofi. DM has fmed.2022.945245/full#supplementary-material References 1. Zimmet P, Alberti KG, Magliano DJ, Bennett PH. Diabetes 3. Solomon SD, Chew E, Duh EJ, Sobrin L, Sun JK, VanderBeek BL, et al. Diabetic mellitus statistics on prevalence and mortality: Facts and fallacies. retinopathy: A position statement by the American diabetes association. Diabetes Nat Rev Endocrinol. (2016) 12:616–22. doi: 10.1038/nrendo. Care. (2017) 40:412–8. doi: 10.2337/dc16-2641 2016.105 4. Simó R, Hernández C. Neurodegeneration in the diabetic eye: New insights 2. International Diabetes Federation. IDF diabetes atlas. 10th ed. Brussels: and therapeutic perspectives. Trends Endocrinol Metab. (2014) 25:23–33. doi: 10. International Diabetes Federation (2021). 1016/j.tem.2013.09.005 Frontiers in Medicine 09 frontiersin.org Barrot et al. 10.3389/fmed.2022.945245 5. Yau JW, Rogers SL, Kawasaki R, Lamoureux EL, Kowalski JW, Bek T, et al. type 2 diabetic patients. Diabet Med. (2008) 25:45–50. doi: 10.1111/j.1464-5491. Global prevalence and major risk factors of diabetic retinopathy. Diabetes Care. 2007.02327.x (2012) 35:556–64. doi: 10.2337/dc11-1909 20. Kawasaki R, Tanaka S, Tanaka S, Abe S, Sone H, Yokote K, et al. Risk of 6. Thomas RL, Halim S, Gurudas S, Sivaprasad S, Owens DR. IDF diabetes cardiovascular diseases is increased even with mild diabetic retinopathy: The Japan atlas: A review of studies utilising retinal photography on the global prevalence of diabetes complications study. Ophthalmology. (2013) 120:574–82. doi: 10.1016/j. diabetes-related retinopathy between 2015 and 2018. Diabetes Res Clin Pract. (2019) ophtha.2012.08.029 157:107840. doi: 10.1016/j.diabres.2019.107840 21. Hu K, Jiang M, Zhou Q, Zeng W, Lan X, Gao Q, et al. Association of diabetic 7. Almdal T, Scharling H, Jensen JS, Vestergaard H. The independent effect retinopathy with stroke: A systematic review and meta-analysis. Front Neurol. of type 2 diabetes mellitus on ischemic heart disease, stroke, and death: (2021) 12:626996. doi: 10.3389/fneur.2021.626996 A population-based study of 13,000 men and women with 20 years of 22. Wong KH, Hu K, Peterson C, Sheibani N, Tsivgoulis G, Majersik JJ, et al. follow-up. Arch Intern Med. (2004) 164:1422–6. doi: 10.1001/archinte.164.1 Diabetic retinopathy and risk of stroke: A secondary analysis of the ACCORD eye 3.1422 study. Stroke. (2020) 51:3733–6. doi: 10.1161/STROKEAHA.120.030350 8. Scanlon PH, Aldington SJ, Stratton IM. Delay in diabetic retinopathy screening 23. Petitti DB, Bhatt H. Retinopathy as a risk factor for nonembolic stroke in increases the rate of detection of referable diabetic retinopathy. Diabet Med. (2014) diabetic subjects. Stroke. (1995) 26:593–6. doi: 10.1161/01.str.26.4.593 31:439–42. doi: 10.1111/dme.12313 24. Cheung N, Rogers S, Couper DJ, Klein R, Sharrett AR, Wong TY. Is diabetic 9. Klein R, Klein BE, Moss SE. Epidemiology of proliferative diabetic retinopathy. retinopathy an independent risk factor for ischemic stroke? Stroke. (2007) 38:398– Diabetes Care. (1992) 15:1875–91. doi: 10.2337/diacare.15.12.1875 401. doi: 10.1161/01.STR.0000254547.91276.50 10. Cheung N, Wong TY. Diabetic retinopathy and systemic vascular 25. Hägg S, Thorn LM, Putaala J, Liebkind R, Harjutsalo V, Forsblom CM, complications. Prog Retin Eye Res. (2008) 27:161–76. doi: 10.1016/j.preteyeres.2007. et al. Incidence of stroke according to presence of diabetic nephropathy and 12.001 severe diabetic retinopathy in patients with type 1 diabetes. Diabetes Care. (2013) 36:4140–6. doi: 10.2337/dc13-0669 11. Liu Y, Teng X, Zhang W, Zhang R, Liu W. Association between diabetic retinopathy and subclinical atherosclerosis in China: Results from a community- 26. Hankey GJ, Anderson NE, Ting RD, Veillard AS, Romo M, Wasik M, et al. based study. Diab Vasc Dis Res. (2015) 12:366–72. doi: 10.1177/1479164115591744 Rates and predictors of risk of stroke and its subtypes in diabetes: A prospective observational study. J Neurol Neurosurg Psychiatry. (2013) 84:281–7. doi: 10.1136/ 12. Saif A, Karawya S, Abdelhamid A. Retinopathy is a strong determinant of jnnp-2012-303365 atherosclerosis in type 2 diabetes: Correlation with carotid intima-media thickness. Endocr Pract. (2015) 21:226–30. doi: 10.4158/EP14390 27. Hugenschmidt CE, Lovato JF, Ambrosius WT, Bryan RN, Gerstein HC, Horowitz KR, et al. The cross-sectional and longitudinal associations of diabetic 13. Rema M, Mohan V, Deepa R, Ravikumar R. Association of carotid intima- retinopathy with cognitive function and brain MRI findings: The action to control media thickness and arterial stiffness with diabetic retinopathy: The Chennai urban cardiovascular risk in diabetes (ACCORD) trial. Diabetes Care. (2014) 37:3244–52. rural epidemiology study (CURES-2). Diabetes Care. (2004) 27:1962–7. doi: 10. doi: 10.2337/dc14-0502 2337/diacare.27.8.1962 28. Exalto LG, Biessels GJ, Karter AJ, Huang ES, Quesenberry CP Jr., Whitmer 14. Alonso N, Traveset A, Rubin E, Ortega E, Alcubierre N, Sanahuja J, et al. RA. Severe diabetic retinal disease and dementia risk in type 2 diabetes. J Alzheimers Type 2 diabetes-associated carotid plaque burden is increased in patients with Dis. (2014) 42 Suppl 3:S109–17. doi: 10.3233/JAD-132570 retinopathy compared to those without retinopathy. Cardiovasc Diabetol. (2015) 14:33. doi: 10.1186/s12933-015-0196-1 29. Mauricio D, Vlacho B, Barrot de la Puente J, Mundet-Tudurí X, Real J, Kulisevsky J, et al. Associations between diabetic retinopathy and Parkinson’s 15. Pearce I, Simó R, Lövestam-Adrian M, Wong DT, Evans M. Association disease: Results from the catalonian primary care cohort study. Front Med between diabetic eye disease and other complications of diabetes: Implications for (Lausanne). (2022) 8:800973. doi: 10.3389/fmed.2021.800973 care. A systematic review. Diabetes Obes Metab. (2019) 21:467–78. doi: 10.1111/ dom.13550 30. Frith E, Loprinzi PD. Retinopathy and mortality. Diabetes Spectr. (2018) 31:184–8. doi: 10.2337/ds17-0010 16. Cheung N, Wang JJ, Klein R, Couper DJ, Sharrett AR, Wong TY. Diabetic retinopathy and the risk of coronary heart disease: The atherosclerosis risk 31. Xu XH, Sun B, Zhong S, Wei DD, Hong Z, Dong AQ. Diabetic retinopathy in communities study. Diabetes Care. (2007) 30:1742–6. doi: 10.2337/dc07- predicts cardiovascular mortality in diabetes: A meta-analysis. BMC Cardiovasc 0264 Disord. (2020) 20:478. doi: 10.1186/s12872-020-01763-z 17. Guo VY, Cao B, Wu X, Lee JJW, Zee BC. Prospective association between 32. Zhu XR, Zhang YP, Bai L, Zhang XL, Zhou JB, Yang JK. Prediction of risk diabetic retinopathy and cardiovascular disease-a systematic review and meta- of diabetic retinopathy for all-cause mortality, stroke and heart failure: Evidence analysis of cohort studies. J Stroke Cerebrovasc Dis. (2016) 25:1688–95. doi: 10.1016/ from epidemiological observational studies. Medicine (Baltimore). (2017) 96:e5894. j.jstrokecerebrovasdis.2016.03.009 doi: 10.1097/MD.0000000000005894 18. Gimeno-Orna JA, Faure-Nogueras E, Castro-Alonso FJ, Boned-Juliani B. 33. Wu L. Classification of diabetic retinopathy and diabetic macular edema. Ability of retinopathy to predict cardiovascular disease in patients with type 2 World J Diabetes. (2013) 4:290. doi: 10.4239/wjd.v4.i6.290 diabetes mellitus. Am J Cardiol. (2009) 103:1364–7. doi: 10.1016/j.amjcard.2009. 34. Therneau TM, Grambsch PM. Modeling survival data: Extending the cox 01.345 model. New York, NY: Springer (2000). doi: 10.1007/978-1-4757-3294-8 19. Targher G, Bertolini L, Zenari L, Lippi G, Pichiri I, Zoppini G, et al. Diabetic 35. Buuren S, Groothuis-Oudshoorn C. MICE: multivariate imputation by retinopathy is associated with an increased incidence of cardiovascular events in chained equations in R. J Stat Softw. (2011) 45:1–67. doi: 10.18637/jss.v045.i03 Frontiers in Medicine 10 frontiersin.org