Association between cataract surgery and diabetic retinopathy incidence and progression: an 11-year population-based retrospective cohort study Authors: Ya-Chu Chang, Jen-Hung Wang, Wei-Tse Chen, Cheng-Jen Chiu Published: 2026 ABSTRACT Abstract AbstractObjectivesTo evaluate the association between cataract surgery and the incidence and progression of diabetic retinopathy (DR) in a population-based cohort of patients with diabetes.DesignA nationwide population-based retrospective cohort study.SettingA nationwide administrative claims database in Taiwan covering more than 99% of the population, with longitudinal follow-up from 2010 to 2021.ParticipantsAdult patients (≥20 years) with diabetes mellitus and cataract were identified. Patients with prior cataract surgery before the first recorded cataract diagnosis were excluded. A total of 15 360 patients who underwent first-time cataract surgery and 15 846 patients without surgery were included. After two-stage matching (1:2 age-matching and sex-matching followed by 1:1 propensity score matching), 7837 matched pairs were analysed.InterventionsNot applicable.Primary and secondary outcome measuresThe primary outcomes were (1) incident DR and (2) DR progression within 6 months and 1 year after the index date. Incident DR was defined as new-onset non-proliferative DR (NPDR) or proliferative DR (PDR). DR progression was defined as transition from no DR to NPDR/PDR or from NPDR to PDR. Outcomes were identified using diagnostic codes recorded at least twice in outpatient visits or once during hospitalisation.ResultsCataract surgery was associated with higher risks of both incident DR and DR progression within 1 year. In the 1:2 matched cohort, the adjusted ORs were 3.40 (95% CI 2.65 to 4.36) for incident DR and 3.11 (95% CI 2.47 to 3.92) for DR progression. Similar findings were observed in the 1:1 propensity score-matched analysis. Absolute event rates at 1 year showed DR progression in approximately 3.1% of patients undergoing surgery compared with 1% in the non-surgical group. Increased risk was also associated with more intensive antidiabetic treatment.ConclusionCataract surgery in patients with diabetes was associated with an increased risk of incident DR and disease progression, particularly within the first postoperative year. These findings support the consideration of cataract surgery as a clinically relevant risk window for enhanced retinal monitoring. Incorporating surgical history into DR risk stratification and screening strategies may improve early detection and inform postoperative care pathways.Keywords: Diabetic retinopathy, Cataract and refractive surgery, DIABETES & ENDOCRINOLOGY, Retrospective Studies STRENGTHS AND LIMITATIONS OF THIS STUDY This study used a large, nationwide administrative database covering over 99% of the population, enhancing representativeness and reducing selection bias. This study used a large, nationwide administrative database covering over 99% of the population, enhancing representativeness and reducing selection bias. A two-stage matching strategy, including age-matching and sex-matching followed by propensity score matching, was applied to improve comparability between groups and reduce confounding. A two-stage matching strategy, including age-matching and sex-matching followed by propensity score matching, was applied to improve comparability between groups and reduce confounding. Both incident diabetic retinopathy (DR) and disease progression were evaluated, providing a comprehensive assessment of retinal outcomes. Both incident diabetic retinopathy (DR) and disease progression were evaluated, providing a comprehensive assessment of retinal outcomes. The use of claims-based diagnostic codes may have led to misclassification or under-detection of mild DR and lacked detailed clinical information such as haemoglobin A1c levels and imaging findings. The use of claims-based diagnostic codes may have led to misclassification or under-detection of mild DR and lacked detailed clinical information such as haemoglobin A1c levels and imaging findings. The follow-up period was limited to 1 year, which may not capture long-term progression of DR after cataract surgery. The follow-up period was limited to 1 year, which may not capture long-term progression of DR after cataract surgery. Introduction Diabetes mellitus (DM) is a rapidly growing global health burden and a significant cause of vision impairment worldwide. Among its microvascular complications, diabetic retinopathy (DR) remains a leading cause of blindness among working-age adults, affecting approximately 20%–35% of individuals with diabetes globally.1–3 The prevalence of DR increases with longer diabetes duration and poorer glycaemic control, highlighting the close relationship between systemic disease severity and retinal outcomes.2 Beyond its clinical consequences, vision loss secondary to DR carries substantial public health implications. Visual impairment due to DR contributes to productivity loss, reduced quality of life and increased healthcare expenditure globally.4 5 The economic burden associated with diabetic eye disease is projected to rise in parallel with the increasing prevalence of diabetes, particularly in ageing societies. Therefore, identifying modifiable or procedure-related risk factors for DR development and progression is essential not only for individual patient care but also for optimising healthcare resource allocation. In parallel, cataract is one of the most common ocular comorbidities in patients with diabetes, occurring earlier and progressing more rapidly than in non-diabetic individuals.6–8 Population-based studies have shown that patients with diabetes have a significantly higher likelihood of undergoing cataract surgery during their lifetime compared with the general population.7 Consequently, cataract and DR frequently coexist, and cataract surgery is often required to restore vision in patients with diabetes. Optimal management of these patients is challenging, since surgery can improve media clarity for DR treatment but might also impact the course of retinopathy. Cataract surgery is among the most frequently performed surgical procedures worldwide, with millions of operations conducted annually.9 In countries with universal health coverage, such as Taiwan, access to cataract surgery is high, potentially amplifying the population-level impact of any postoperative complications or disease acceleration. Even a modest increase in DR risk following cataract surgery could translate into a substantial increase in retinal care demand at the health system level. Although modern cataract surgery using phacoemulsification is generally considered safe and effective, its impact on the diabetic retina remains controversial. While advances in surgical techniques have reduced operative trauma, concerns persist that cataract surgery may still precipitate the development of new DR or exacerbate pre-existing retinopathy.10–13 Several cohort studies and population-based analyses have reported a higher incidence of DR or faster DR progression following cataract surgery.11–15 A recent systematic review of 15 studies (7287 patients) found that, overall, DR progression risk was about 46% higher after cataract surgery relative to controls, and the risk of sight-threatening DR (proliferative DR (PDR) or diabetic macular oedema) was about 84% higher in operated eyes.12 Another large database study from Asian populations demonstrated an increased risk of non-proliferative DR (NPDR) development after cataract extraction among patients with diabetes without pre-existing retinopathy.14 Proposed mechanisms include surgical disruption of the blood–aqueous barrier, leading to a pro-inflammatory state and a surge in vascular endothelial growth factor (VEGF) and other cytokines, which may accelerate microvascular damage in the diabetic retina.16 17 These findings suggest that cataract surgery may act as an independent risk factor for onset of DR or worsening in susceptible patients. In contrast, evidence from the Early Treatment Diabetic Retinopathy Study (ETDRS) suggested that lens surgery was associated with, at most, a borderline increase in the risk of DR progression after multivariable adjustment.18 Another paired-eye analysis reported no significant difference in DR progression between operated and non-operated eyes after modern phacoemulsification.19 Evidence from other research suggests that uneventful phacoemulsification does not significantly exacerbate the progression of DR in the postoperative period.20 These discrepancies may be attributable to differences in study design, sample size, follow-up duration, baseline DR severity and the extent of adjustment for systemic and ocular confounders. Importantly, the modifying effects of diabetes severity and prior ocular interventions remain insufficiently clarified. Large-scale, long-term studies that simultaneously examine cataract surgery, diabetes severity, systemic comorbidities and prior ocular treatments within a single analytical framework are scarce, particularly in Asian populations, where the diabetes burden is high, and patient profiles may differ from Western cohorts. Therefore, this 11-year population-based retrospective cohort study with a relatively large sample size aims to clarify the potential association between cataract surgery and DR, including both incidence and progression, with particular attention to interactions among other well-established risk determinants. By addressing these inter-relationships (eg, diabetes medication, comorbidities and history of retinal photocoagulation or injections), the study seeks to generate meaningful evidence to guide more holistic management strategies for diabetic individuals undergoing cataract surgery, optimising visual outcomes while minimising the risk of sight-threatening retinopathy. Methods Study design and data source This population-based retrospective cohort study used data from the Longitudinal Generation Tracking Database (LGTD), a subset of Taiwan’s National Health Insurance Research Database (NHIRD), accessed with authorisation from the Ministry of Health and Welfare. The NHIRD contains comprehensive claims data on inpatient and outpatient medical services, including diagnoses, procedures, prescriptions, and surgical treatments. Covering more than 99% of Taiwan’s population, the NHIRD is a well-established resource for large-scale population-based research. For this study, we used the LGTD 2010, which comprises medical records of 2 million individuals randomly sampled from approximately 23.8 million National Health Insurance beneficiaries, with longitudinal follow-up data spanning from 2010 to 2021. This study was approved by the Research Ethics Committee of Hualien Tzu Chi Hospital, Buddhist Tzu Chi Medical Foundation (Institutional Review Board (IRB) No. 113–169 B) and was conducted in accordance with the prespecified study protocol (online supplemental file 1). The requirement for written informed consent was waived by the IRB because the study used de-identified secondary data, thereby ensuring the protection of patient privacy. This study was conducted and reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines (online supplemental file 2).21 Study cohorts This study included patients aged ≥20 years who were diagnosed with both DM and cataract during the study period (online supplemental table S1). The initial cohort was identified from 2 000 123 individuals in the general population of the NHIRD. To ensure data validity, we excluded patients who had undergone cataract surgery before their first recorded cataract diagnosis, as well as those who died during the study period. Among the eligible patients, 15 360 underwent cataract surgery for the first time during the study period, whereas 15 846 did not receive cataract surgery. For patients who underwent cataract surgery, the admission date for the surgical procedure was defined as the index date. To ensure comparability between groups, a two-stage matching strategy was applied, consisting of 1:2 age-matching and sex-matching followed by 1:1 propensity score matching (PSM). After matching, 7847 patients comprised the cataract surgery group, while 15 694 patients formed the first comparison group and 7837 patients formed the second comparison group (figure 1). The index date of each matched control was assigned to be the same as that of the corresponding patient in the cataract surgery group. To avoid correlated measurements, only one eye per patient was included in the analysis. Flow diagram of patient selection and cohort construction. DM, diabetes mellitus; DR, diabetic retinopathy; LGTD, Longitudinal Generation Tracking Database; M, months; Y, year. Covariates and outcomes Baseline characteristics of eligible patients were retrieved from the LGTD. Pre-existing comorbidities, prior ocular treatments and baseline antidiabetic medication use—identified as potential confounders based on previous literature—were incorporated into the analysis.3 14 15 18 Comorbidities were identified using International Classification of Diseases (ICD)-9-CM and ICD-10 diagnostic codes and included hypertension, hyperlipidaemia, diabetic nephropathy, coronary artery disease, cerebrovascular disease, chronic kidney disease and cancer (online supplemental table S1). A comorbidity was considered present if it was recorded as a discharge diagnosis or documented in at least two outpatient visits within the year preceding the index date. Prior ocular treatments were defined as receipt of intravitreal injection (IVI) or panretinal photocoagulation (PRP) within 1 year before the index date (online supplemental table S2). Baseline medication use was defined as prescriptions with a cumulative duration of ≥30 days during the year prior to the index date (online supplemental table S3). Based on baseline antidiabetic medication use, patients were categorised into four groups: no antidiabetic medication, oral hypoglycaemic agents (OHA) only, insulin only and combined OHA plus insulin therapy (OHA+insulin). The first primary outcome was the incidence of DR, defined as a new diagnosis of NPDR or PDR occurring within 6 months or 1 year after the index date. The second primary outcome was DR progression, defined as a change in disease status from no DR to NPDR or PDR or from NPDR to PDR within 6 months or 1 year after the index date. An outcome event was considered present if the corresponding diagnosis was recorded at least twice during outpatient visits or once during an inpatient admission. Patients were followed from the index date until the occurrence of an outcome event, death or 1 year after the index date, whichever came first. Statistical analysis To ensure comparability between the cataract surgery group and the comparison group, patients were initially matched on age and sex. In the second stage, PSM was performed to further balance baseline characteristics and reduce potential confounding. Propensity scores were estimated using multivariable logistic regression that incorporated age, sex and baseline comorbidities, including hypertension, hyperlipidaemia, diabetic nephropathy, coronary artery disease, cerebrovascular accident, chronic kidney disease and cancer to estimate each patient’s probability of undergoing cataract surgery. A 1:1 nearest-neighbour matching algorithm without replacement was applied, with a calliper width of 0.2 SD of the logit of the propensity score. Baseline characteristics between the two groups were compared using Student’s t-test for continuous variables and the χ2 test or Fisher’s exact test for categorical variables as appropriate. Multivariable conditional logistic regression was subsequently used to identify factors associated with the incidence or progression of DR. Statistical significance was defined as a two-sided p value <0.05. E-value was adopted to quantify the minimum strength of an unmeasured confounder that would be required to fully explain the observed association.22 Larger E-values indicate that stronger unmeasured confounding would be necessary to negate the observed effect estimate. All statistical analyses were performed using SAS software V.9.4 (SAS Institute, Cary, North Carolina, USA) and Stata V.16 (StataCorp, College Station, Texas, USA). Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting or dissemination plans of this research. Results Study population After a two-stage matching strategy, the study population consisted of cohorts generated through 1:2 age-matching and sex-matching followed by 1:1 PSM. The demographic and clinical characteristics of the two groups are presented in table 1. Table 1Demographics1:2 Age/sex (n=23 541)1:1 PSM (n=15 674)Cataract surgeryTotalP valueCataract surgeryTotalP valueNoYesNoYesN15 694784723 5417837783715 674Age65.96±10.3565.96±10.3665.96±10.35166.00±10.2765.97±10.3565.98±10.310.829Age group (years)–––1–––0.448 20–657244 (46.2%)3622 (46.2%)10 866 (46.2%)3556 (45.4%)3613 (46.1%)7169 (45.7%) 65–806712 (42.8%)3356 (42.8%)10 068 (42.8%)3434 (43.8%)3356 (42.8%)6790 (43.3%) ≥801738 (11.1%)869 (11.1%)2607 (11.1%)847 (10.8%)868 (11.1%)1715 (10.9%)Sex–––1–––0.441 Male7134 (45.5%)3567 (45.5%)10 701 (45.5%)3513 (44.8%)3561 (45.4%)7074 (45.1%) Female8560 (54.5%)4280 (54.5%)12 840 (54.5%)4324 (55.2%)4276 (54.6%)8600 (54.9%)DR status–––<0.001*–––<0.001* No DR15 398 (98.1%)7210 (91.9%)22 608 (96.0%)7674 (97.9%)7205 (91.9%)14 879 (94.9%) NPDR188 (1.2%)295 (3.8%)483 (2.1%)100 (1.3%)294 (3.8%)394 (2.5%) PDR108 (0.7%)342 (4.4%)450 (1.9%)63 (0.8%)338 (4.3%)401 (2.6%)Antidiabetic medications (%)–––<0.001*–––0.005* None10 525 (67.1%)4909 (62.6%)15 434 (65.6%)5053 (64.5%)4907 (62.6%)9960 (63.5%) OHA only4418 (28.2%)2428 (30.9%)6846 (29.1%)2370 (30.2%)2423 (30.9%)4793 (30.6%) Insulin only140 (0.9%)75 (1.0%)215 (0.9%)70 (0.9%)75 (1.0%)145 (0.9%) OHA+Insulin611 (3.9%)435 (5.5%)1046 (4.4%)344 (4.4%)432 (5.5%)776 (5.0%)Prior IVI (%)59 (0.4%)146 (1.9%)205 (0.9%)<0.001*33 (0.4%)146 (1.9%)179 (1.1%)<0.001*Prior PRP (%)49 (0.3%)179 (2.3%)228 (1.0%)<0.001*32 (0.4%)178 (2.3%)210 (1.3%)<0.001*Hypertension (%)7850 (50.0%)4425 (56.4%)12 275 (52.1%)<0.001*4456 (56.9%)4416 (56.4%)8872 (56.6%)0.519Hyperlipidaemia (%)5769 (36.8%)3067 (39.1%)8836 (37.5%)0.001*3043 (38.8%)3062 (39.1%)6105 (39.0%)0.756Diabetic nephropathy (%)1143 (7.3%)788 (10.0%)1931 (8.2%)<0.001*747 (9.5%)779 (9.9%)1526 (9.7%)0.389CAD (%)2046 (13.0%)1173 (15.0%)3219 (13.7%)<0.001*1216 (15.5%)1170 (14.9%)2386 (15.2%)0.306CVA (%)1442 (9.2%)652 (8.3%)2094 (8.9%)0.025*634 (8.1%)651 (8.3%)1285 (8.2%)0.621CKD (%)743 (4.7%)551 (7.0%)1294 (5.5%)<0.001*522 (6.7%)542 (6.9%)1064 (6.8%)0.525Cancer (%)1022 (6.5%)512 (6.5%)1534 (6.5%)0.970465 (5.9%)511 (6.5%)976 (6.2%)0.128Number of ophthalmology OPD visits (within 6 months)0.95±1.857.52±4.203.14±4.21<0.001*0.97±1.887.52±4.204.24±4.61<0.001*Number of ophthalmology OPD visits (within 1 year)1.91±3.419.85±6.304.56±5.92<0.001*1.95±3.469.85±6.305.90±6.43<0.001*Data are presented as n or mean±SD deviation.*p value <0.05 was considered statistically significant after test.CAD, coronary artery disease; CKD, chronic kidney disease; CVA, cerebrovascular accident; DR, diabetic retinopathy; IVI, intravitreal injection; NPDR, non-proliferative diabetic retinopathy; OHA, oral hypoglycaemic agents; OPD, outpatient department; PDR, proliferative diabetic retinopathy; PRP, panretinal photocoagulation. Demographics Data are presented as n or mean±SD deviation. p value <0.05 was considered statistically significant after test. CAD, coronary artery disease; CKD, chronic kidney disease; CVA, cerebrovascular accident; DR, diabetic retinopathy; IVI, intravitreal injection; NPDR, non-proliferative diabetic retinopathy; OHA, oral hypoglycaemic agents; OPD, outpatient department; PDR, proliferative diabetic retinopathy; PRP, panretinal photocoagulation. Under the 1:2 age-matched and sex-matched design, the mean age and sex distribution were identical between groups (all p values=1). However, patients in the cataract surgery group exhibited significantly higher rates of antidiabetic medication use and prior ocular treatments. Following 1:1 PSM, no significant differences were observed in the distribution of comorbidities between the two groups (all p values >0.05). These findings indicate that the PSM procedure effectively reduced baseline imbalances and potential confounding. In addition, patients in the cataract surgery group also underwent more intensive postoperative ophthalmic follow-up than those in the comparison group. Accordingly, the number of ophthalmology outpatient visits within 6 months and 1 year were significantly higher in the cataract surgery group (both p values <0.001). Primary outcome analysis In the analysis of DR progression and incidence within 6 months and 1 year (table 2), the cataract surgery group demonstrated significantly higher progression and incidence rates under both study designs (all p values <0.05). Specifically, the progression rate in the cataract surgery group was approximately 2% within 6 months and 3% within 1 year, whereas the comparison group exhibited progression rates of approximately 0.5% and 1% over the same time intervals. After restricting the analysis to patients without DR at baseline, the cataract surgery group continued to show a significantly higher incidence rate compared with the comparison group. These findings remained consistent in analyses stratified by NPDR and PDR (online supplemental table S4). The incidences of NPDR and PDR in the cataract surgery group were approximately 1.7% and 0.7% within 6 months, increasing to 2.4% and 1.1% within 1 year, respectively. In contrast, the corresponding incidences in the comparison group were approximately 0.4% and 0.1% within 6 months, increasing to 0.8% and 0.3% within 1 year, respectively. Table 2Comparison of DR progression and incidence within 6 months/1 year1:2 Age/sex (n=23 541)1:1 PSM (n=15 674)Cataract surgeryTotalP valueCataract surgeryTotalP valueNoYesNoYesProgression (n=23 541)Progression (n=15 674)Progression at 6 months (%)77 (0.5%)175 (2.2%)252 (1.1%)<0.001*41 (0.5%)173 (2.2%)214 (1.4%)<0.001*Progression at 1 year (%)161 (1.0%)243 (3.1%)404 (1.7%)<0.001*87 (1.1%)241 (3.1%)328 (2.1%)<0.001*Incidence (n=22 608) Incidence (n=14 879)Incidence at 6 months (%)77 (0.5%)165 (2.3%)242 (1.1%)<0.001*41 (0.5%)164 (2.3%)205 (1.4%)<0.001*Incidence at 1 year (%)160 (1.0%)230 (3.2%)390 (1.7%)<0.001*86 (1.1%)229 (3.2%)315 (2.1%)<0.001*Data are presented as n(%).*p value <0.05 was considered statistically significant after test.DR, diabetic retinopathy; PSM, propensity score matching. Comparison of DR progression and incidence within 6 months/1 year Data are presented as n(%). p value <0.05 was considered statistically significant after test. DR, diabetic retinopathy; PSM, propensity score matching. Factors associated with progression and incident DR within 1 year Multivariable conditional logistic regression analyses were performed to identify factors associated with DR progression and incidence. Under the 1:2 age-matched and sex-matched design, cataract surgery and antidiabetic medication use were significantly associated with DR progression (table 3). Patients who underwent cataract surgery had a higher risk of DR progression within 1 year compared with the comparison group (adjusted OR (aOR) 3.11, 95% CI 2.47 to 3.92; p <0.001). In addition, patients receiving antidiabetic medications exhibited an increased risk of DR progression compared with those not receiving antidiabetic treatment, with aORs of 3.50 for OHA only, 5.49 for insulin only and 6.61 for OHA+insulin therapy (95% CI 2.49 to 4.91, 1.51 to 19.88, and 3.66 to 11.95, respectively; all p <0.001). Similar associations were observed under the 1:1 PSM design. Table 3Factors associated with diabetic retinopathy progression within 1 yearFactors1:2 Age/sex (n=23 541)1:1 PSM (n=15 674)CrudeAdjustedCrudeAdjustedOR (95% CI)P valueOR (95% CI)P valueOR (95% CI)P valueOR (95% CI)P valueCataract surgery (yes vs no)3.11 (2.54 to 3.81)<0.001*3.11 (2.47 to 3.92)<0.001*2.79 (2.18 to 3.57)<0.001*2.76 (2.10 to 3.62)<0.001*Antidiabetic medications(%) NoneReference Reference Reference Reference OHA only3.36 (2.53 to 4.48)<0.001*3.50 (2.49 to 4.91)<0.001*3.83 (2.49 to 5.89)<0.001*3.97 (2.46 to 6.42)<0.001* Insulin only4.07 (1.31 to 12.60)0.015*5.49 (1.51 to 19.88)0.010*6.98 (1.58 to 30.84)0.010*5.95 (1.05 to 33.76)0.044* OHA+insulin7.55 (4.54 to 12.56)<0.001*6.61 (3.66 to 11.94)<0.001*6.69 (3.54 to 12.62)<0.001*5.79 (2.88 to 11.65)<0.001*Prior IVI (yes vs no)2.46 (0.84 to 7.17)0.0991.55 (0.45 to 5.35)0.4912.00 (0.60 to 6.64)0.2581.24 (0.29 to 5.21)0.769Prior PRP (yes vs no)2.00 (0.79 to 5.04)0.1410.49 (0.17 to 1.43)0.1899.00 (1.14 to 71.04)0.037*1.72 (0.16 to 18.99)0.660Hypertension (yes vs no)1.16 (0.91 to 1.48)0.2400.81 (0.59 to 1.11)0.190 Hyperlipidaemia (yes vs no)1.33 (1.03 to 1.70)0.027*1.03 (0.76 to 1.41)0.836 Diabetic nephropathy (yes vs no)2.24 (1.56 to 3.21)<0.001*1.19 (0.75 to 1.89)0.460 CAD (yes vs no)0.80 (0.54 to 1.19)0.2710.80 (0.50 to 1.28)0.355 CVA (yes vs no)1.39 (0.89 to 2.16)0.1441.12 (0.67 to 1.86)0.678 CKD (yes vs no)1.67 (0.99 to 2.80)0.0540.80 (0.42 to 1.54)0.503 Cancer (yes vs no)0.72 (0.42 to 1.23)0.2280.54 (0.28 to 1.02)0.059 Data are presented as OR (95% CI). *p value <0.05 was considered statistically significant after test.CAD, coronary artery disease; CKD, chronic kidney disease; CVA, cerebrovascular accident; IVI, intravitreal injection; OHA, oral hypoglycaemic agents; PRP, panretinal photocoagulation; PSM, propensity score matching. Factors associated with diabetic retinopathy progression within 1 year Data are presented as OR (95% CI). p value <0.05 was considered statistically significant after test. CAD, coronary artery disease; CKD, chronic kidney disease; CVA, cerebrovascular accident; IVI, intravitreal injection; OHA, oral hypoglycaemic agents; PRP, panretinal photocoagulation; PSM, propensity score matching. The results of the DR incidence analysis are summarised in table 4. Under both study designs, cataract surgery and antidiabetic medication use were significantly associated with an increased risk of incident DR, while cancer was identified as an additional significant factor under the 1:2 age-matched and sex-matched design. Patients who underwent cataract surgery had a higher risk of DR incidence within 1 year compared with the comparison group, with aORs of 3.40 and 3.03 under the 1:2 matched and 1:1 PSM designs, respectively (95% CI 2.65 to 4.36 and 2.26–4.06; all p <0.001). Similarly, patients receiving antidiabetic medications had a higher risk of incident DR than those without antidiabetic treatment, with aORs of 3.65 (OHA only), 5.92 (insulin only) and 7.26 (OHA+insulin) under the 1:2 age-matched and sex-matched design (95% CI 2.56 to 5.20, 1.52 to 23.05 and 3.76 to 14.04, respectively; all p <0.001). Comparable findings were observed under the 1:1 PSM design. In contrast, patients with cancer had a lower risk of incident DR within 1 year compared with those without cancer (aOR 0.49, 95% CI 0.25 to 0.94; p=0.032). Table 4Factors associated with incident diabetic retinopathy within 1 year among patients without diabetic retinopathyFactors1:2 Age/sex (n=22 608)1:1 PSM (n=14 879)CrudeAdjustedCrudeAdjustedOR (95% CI)P valueOR (95% CI)P valueOR (95% CI)P valueOR (95% CI)P valueCataract surgery (yes vs no)3.32 (2.68 to 4.11)<0.001*3.40 (2.65 to 4.36)<0.001*2.93 (2.26 to 3.81)<0.001*3.03 (2.26 to 4.06)<0.001*Antidiabetic medications (%) NoneReference Reference Reference Reference OHA only3.49 (2.60 to 4.69)<0.001*3.65 (2.56 to 5.20)<0.001*3.74 (2.40 to 5.84)<0.001*4.20 (2.54 to 6.95)<0.001* Insulin only4.96 (1.50 to 16.39)0.009*5.92 (1.52 to 23.05)0.010*7.17 (1.60 to 32.11)0.010*6.49 (1.06 to 39.74)0.043* OHA+insulin8.65 (4.92 to 15.20<0.001*7.26 (3.76 to 14.04)<0.001*7.76 (3.80 to 15.83)<0.001*7.11 (3.25 to 15.54)<0.001*Prior IVI (yes vs no)2.19 (0.58 to 8.36)0.2490.66 (0.13 to 3.42)0.617NANANANAPrior PRP (yes vs no)7.12 (0.79 to 64.38)0.0807.71 (0.50 to 118.16)0.142NANANANAHypertension (yes vs no)1.24 (0.96 to 1.60)0.0960.87 (0.62 to 1.21)0.395 Hyperlipidaemia (yes vs no)1.36 (1.05 to 1.76)0.020*1.11 (0.80 to 1.55)0.520 Diabetic nephropathy (yes vs no)2.54 (1.72 to 3.74)<0.001*1.25 (0.76 to 2.06)0.381 CAD (yes vs no)0.81 (0.54 to 1.21)0.3000.80 (0.49 to 1.31)0.373 CVA (yes vs no)1.47 (0.94 to 2.31)0.0921.31 (0.76 to 2.25)0.325 CKD (yes vs no)1.78 (1.02 to 3.09)0.041*0.89 (0.44 to 1.82)0.756 Cancer (yes vs no)0.71 (0.41 to 1.22)0.2140.49 (0.25 to 0.94)0.032* Data are presented as OR (95% CI).*p value <0.05 was considered statistically significant after test.CAD, coronary artery disease; CKD, chronic kidney disease; CVA, cerebrovascular accident; IVI, intravitreal injection; OHA, oral hypoglycaemic agents; PRP, panretinal photocoagulation; PSM, propensity score matching. Factors associated with incident diabetic retinopathy within 1 year among patients without diabetic retinopathy Data are presented as OR (95% CI). p value <0.05 was considered statistically significant after test. CAD, coronary artery disease; CKD, chronic kidney disease; CVA, cerebrovascular accident; IVI, intravitreal injection; OHA, oral hypoglycaemic agents; PRP, panretinal photocoagulation; PSM, propensity score matching. Subgroup analysis The results of the subgroup analyses of DR progression within 1 year, stratified by age group, sex and baseline DR status, are summarised in online supplemental table S5. Under both study designs, patients who underwent cataract surgery had a significantly higher risk of DR progression within 1 year than those in the comparison group, with aORs ranging from 1.97 to 5.38 (all p <0.001). Similar findings were observed after restricting the analysis to patients without DR at baseline, as shown in online supplemental table S6. Discussion Our population-based retrospective cohort study highlights that cataract surgery in patients with diabetes is associated with a higher risk of both new-onset DR and the progression of existing DR. Within 6 months and 1 year postoperatively, patients who underwent cataract surgery showed significantly greater rates of DR development and DR worsening compared with non-surgical controls. These findings suggest that cataract surgery may be associated with an increased risk for initiating DR in diabetics and accelerating the advancement of retinopathy in eyes already affected. Notably, the detrimental effect was observed as early as 6 months post-surgery and remained evident at 1 year, underscoring a significant impact of cataract surgery on the diabetic retina. Historically, early studies in the era of cataract extraction reported high rates of postoperative DR worsening, including accelerated retinopathy progression and complications like vitreous haemorrhage.23 24 With the advent of phacoemulsification, surgical trauma and inflammation have been reduced; some modern series even suggested cataract removal may have minimal effect on DR course.15 Supporting this view, further evidence suggested that the risk of post-cataract DR exacerbation is negligible when using modern phacoemulsification techniques in the absence of surgical complications.20 Nevertheless, a growing body of recent evidence indicates that cataract surgery can indeed precipitate DR changes.11–14 Our results align with numerous cohort and population-based studies that reported higher incidence or faster progression of DR following cataract surgery. In a study conducted in Sydney, diabetic eyes undergoing phacoemulsification had roughly double the 1-year DR progression rate compared with non-operated eyes.11 Another population-based study from Singapore noted that any history of cataract surgery was associated with a higher 6-year risk of developing DR (adjusted risk ratio=1.7).13 Similarly, an Asian database analysis found a significant increase in new-onset DR after cataract extraction (adjusted HR=1.48) in patients without prior retinopathy.14 Consistently, our matched analysis demonstrated approximately a threefold increased risk of DR development and progression at 12 months in the operated eyes compared with controls. This magnitude of effect appears somewhat greater than that reported in the above studies, and the difference may reflect potential ethnic or healthcare differences in our Taiwanese cohort. It is also possible that closer postoperative surveillance in surgical patients led to earlier detection of mild DR that might have gone unrecognised in non-surgical patients, representing a potential detection bias noted in prior studies. However, the magnitude and consistency of the association after matching and multivariable adjustment suggest that surveillance bias alone is unlikely to fully account for the findings. Overall, the cumulative evidence suggests that cataract surgery may contribute to DR development and progression in susceptible individuals. Cataract surgery, even when uncomplicated, induces an intraocular inflammatory response that can disrupt the blood-ocular barriers. Patel et al reported that immediately after phacoemulsification, intraocular levels of pro-inflammatory and pro-angiogenic factors rise significantly.25 For instance, 1 day post surgery, aqueous levels of VEGF, interleukin-1 and other cytokines can increase several-fold and remain elevated for weeks, and such cytokine surges following uneventful cataract surgery may take up to 1 month to subside. These inflammatory mediators are known to contribute to retinal vascular permeability and neovascularisation, potentially exacerbating incipient DR or triggering new lesions. It is likely that in susceptible individuals, surgical insult accelerates the microvascular damage of diabetes through such inflammatory pathways, leading to progression from no DR to NPDR or from NPDR to PDR in the early postoperative period. Our finding that DR progression was already evident by 6 months in some patients supports the notion of an acute inflammatory-driven insult superimposed on the ongoing chronic disease process. Crucially, our study highlights that the severity of the diabetic state itself remains a dominant driver of retinal outcomes after surgery. We found that patients on more intensive diabetes treatment regimens (insulin or combined therapy) had significantly higher odds of DR progression and incidence, independent of surgical status. Those requiring insulin had about a sixfold greater risk of retinopathy progression compared with diet-controlled diabetics, and combination therapy was associated with an even higher risk. This echoes the well-established relationship between poor glycaemic control and DR development. Henricsson et al, for example, noted that among operated eyes, DR progression was significantly associated with higher haemoglobin A1c (HbA1c) levels and insulin use, whereas well-controlled diabetic patients often underwent cataract surgery without any acceleration of retinopathy.15 From a clinical management standpoint, our study reinforces the importance of optimising diabetic care before and after cataract surgery. In our cohort, prior cancer diagnosis was associated with a lower observed incidence of DR after surgery. This finding should be interpreted cautiously, as it may reflect residual confounding, survivor bias or differential healthcare use rather than a true protective effect. Further studies are needed to clarify this association. Furthermore, common systemic comorbidities (hypertension, hyperlipidaemia, diabetic nephropathy, coronary artery disease, stroke and cancer) did not emerge as independent risk factors for DR progression in our adjusted analyses. Pre-existing retinopathy and other ocular factors also critically influence outcomes. Eyes with moderate to severe DR at the time of cataract surgery are widely recognised to be at higher risk of postoperative progression.26 Because PRP or IVI is typically indicated for patients with more advanced DR, we further examined the association between prior PRP or IVI and subsequent DR incidence or progression. After adjustment, the distributions of prior PRP and IVI were comparable between groups, and no significant differences in DR incidence or progression were observed within 1 year following surgery. However, the number of patients with prior PRP or IVI in our cohort was relatively small, and the corresponding CIs were wide, indicating that these subgroup analyses were likely underpowered to detect clinically meaningful associations in either direction. This finding differs from some clinical guidelines that advocate preoperative intervention to reduce ocular ischaemic drive and inflammatory load.27 28 Given the limited power of our analysis, this null result should be interpreted with caution and should not be taken as evidence either supporting or refuting the effectiveness of preoperative treatment in stabilising high-risk eyes. Larger, adequately powered studies specifically designed to evaluate the role of preoperative PRP or IVI are warranted. Taken together, our findings suggest that the primary determinants of postoperative retinopathy outcomes are the intrinsic severity of diabetes and the surgical insult itself, whereas other covariates play a comparatively smaller role after adjustment. These observations not only inform individual risk stratification but also raise important considerations at the population and health system levels. From a health systems perspective, cataract surgery may serve as a clinically identifiable trigger point for intensified DR surveillance. Current screening strategies are primarily based on baseline retinopathy severity and glycaemic control, with limited consideration of procedural events. Our findings suggest that incorporating surgical history into risk stratification models could enable more targeted and timely retinal monitoring, particularly during the first postoperative year. Given the high global volume of cataract surgery and the increasing prevalence of diabetes, even a modest increase in postoperative DR risk could translate into substantial demand for retinal services, including increased use of retinal imaging, intravitreal therapy and specialist care. In resource-constrained settings, such increases may challenge healthcare system capacity, highlighting the need for proactive surveillance strategies. Preventive approaches, including optimisation of glycaemic control before surgery and structured postoperative retinal follow-up, may help mitigate this burden. In addition, disparities in access to retinal care—particularly in rural or underserved areas—may further influence long-term outcomes, underscoring the importance of integrating cataract surgical pathways with DR screening programmes to ensure continuity of care. Although this study was based on a Taiwanese population, similar associations have been reported in both Asian and Western populations, supporting the external validity of our findings, although further validation in more diverse healthcare systems remains warranted. The strengths of our study include the provision of real-world evidence derived from a national administrative database, thereby enhancing external validity compared with single-centre clinical series. We examined both DR incidence and progression, providing a comprehensive assessment of retinal outcomes in patients with diabetes with and without cataract surgery. However, several limitations merit discussion. First, our study relied on diagnosis codes from an administrative database to define DR outcomes. This may under-report milder retinopathy, while subtle retinal changes might have been missed or unrecorded, and some misclassification is possible. Second, although we used antidiabetic medication intensity as a proxy for diabetes severity and observed clear associations with the outcomes, we lacked data on important clinical variables such as HbA1c levels, duration of diabetes and ocular imaging findings. The possibility of residual or unmeasured confounding cannot be excluded. To assess the potential impact of unmeasured confounding, we calculated the E-value for our primary finding (table 3). The E-values were 5.67 and 4.96, respectively, indicating that the observed OR could be explained away only by an unmeasured confounder that was associated with both the exposure and the outcome by an OR of 5.67-fold or 4.96-fold, respectively, above and beyond the measured covariates. Confounding of this magnitude is unlikely in this clinical context, suggesting that our findings are robust to unmeasured confounding. Third, the follow-up in our study was limited to 1 year. While this is a relevant timeframe for postoperative changes, it may be insufficient to capture late-onset DR progression or PDR development that could occur several years after surgery. Some studies suggest that differences in DR outcomes become more evident over longer periods.29 Thus, our findings primarily inform the short-to-intermediate term risk. Additionally, socioeconomic factors and lifestyle behaviours, which may influence both diabetes control and healthcare use, were not available in the database. Fourth, the possibility of a detection bias cannot be ruled out. Patients undergoing cataract surgery received more intensive postoperative ophthalmic follow-up and retinal examinations leading to a higher detection rate of DR than non-surgical patients. Nevertheless, the greater extent of the strength of the association observed in our study might not be eliminated after detection bias is considered. Finally, this study was conducted using a Taiwanese population-based database, and differences in healthcare systems, ethnicity and clinical practice patterns may limit direct generalisability to other settings. However, given the high global prevalence of diabetes and the widespread use of cataract surgery, as well as consistent findings reported in other populations, our results are likely applicable to similar healthcare contexts. Despite these limitations, our study offers valuable population-level evidence, reflecting ‘real-world’ patient management outside of clinical trials. These findings support the integration of postoperative retinal monitoring into comprehensive diabetes care pathways and highlight the importance of multidisciplinary coordination between cataract surgeons and retinal specialists. Conclusion In this large population-based retrospective cohort study, cataract surgery in patients with diabetes was associated with an increased risk of both incident DR and progression of existing retinopathy within 1 year. Our findings suggest that postoperative retinal outcomes are influenced by both the surgical procedure and the underlying severity of diabetes, whereas other systemic comorbidities and prior ocular treatments appeared to have a comparatively smaller impact after adjustment. These results highlight the importance of optimising systemic glycaemic control and ensuring structured postoperative retinal surveillance, particularly during the first year after surgery. Integrating cataract surgical care into comprehensive diabetes management pathways may help improve visual outcomes and reduce the burden of sight-threatening retinopathy at the population level. Supplementary material Acknowledgements The authors used ChatGPT (OpenAI, San Francisco, California, USA) to assist with language editing. All scientific content was reviewed and verified by the authors, who take full responsibility for the accuracy and integrity of the work. Footnotes Funding: The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors. Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2026-120170 ). Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Data availability free text: The datasets analysed during the current study are available from the Taiwan National Health Insurance Research Database (NHIRD), managed by the Ministry of Health and Welfare, Taiwan. However, restrictions apply to the availability of these data, which were used under licence for the current study and are therefore not publicly available. Data are available from the Ministry of Health and Welfare upon reasonable request and with permission from the relevant authorities. Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. Data availability statement Data are available upon reasonable request. Data may be obtained from a third party and are not publicly available. References Supplementary Materials Data Availability Statement Data are available upon reasonable request. Data may be obtained from a third party and are not publicly available.