The retrospective population-based cohort study was conducted in accordance with the Declaration of Helsinki and was approved by both the National Health Insurance Administration of Taiwan and the Institutional Review Board of Chung Shan Medical University (identification code: CS-17075). Supported by the Taiwan National Health Research Institutes, the NHIRD contains medical information on insurance claims from nearly the entire population of Taiwan. These claims data were obtained from the Longitudinal Health Insurance Database 2005 version (LHID), which includes information on 2 million individuals randomly selected from the NHIRD documents for 2005. The LHIDs were connected from 1 January 2000 until 31 December 2016, and the International Classification of Diseases, Ninth Revision (ICD-9) and International Classification of Diseases, Tenth Revision (ICD-10) were applied for disease diagnosis and identification. Information on the patients’ medication history, demographic characteristics, socioeconomic conditions, and locations of residence are also accessible from the NHIRD.
All the subjects included in the current study were selected in 2005, and the relevant data for the participants were traced from 2000 to 2016. Patients were regarded as a case with DM if (1) their medical documents demonstrated a history of type 2 DM; (2) a blood glucose test, a glycosylated haemoglobin test, or an oral glucose tolerance test was scheduled before the diagnosis of DM; or (3) DM was diagnosed by an internal medicine or family medicine department. To more accurately elucidate the association between DM and related retinopathy and cataracts, the following criteria were applied to exclude certain impaired ocular conditions: (1) diagnosed with legal blindness before the diagnosis of DM; (2) scheduled any type of eyeball removal surgery before the diagnosis of DM; (3) receipt of a diagnosis of ophthalmic tumours before the diagnosis of DM; (4) receipt of a diagnosis of prominent ocular trauma at any time; (5) scheduled cataract surgery, diagnosed with pseudophakia, and scheduled Nd:YAG capsulotomy before the index date; (6) diagnosed with myopia, which has previously been associated with cataract formation ; (7) diagnosed with DR before the diagnosis of DM; (8) event is before the index date; and (9) receipt of a diagnosis of DM before 2005 (to exclude patients with extremely long DM disease periods). After exclusion, the study group was set as DR individuals who (1) had a diagnosis of DR, (2) scheduled fundus photography before the DR diagnosis and (3) scheduled optical coherence tomography before the DR diagnosis. In addition, every subject in the study group was age- and sex-matched with four DM patients without DR and non-DM individuals, which served as the DM control group and non-DM control group, respectively. Patients with DR who could not be matched with either four DM patients without DR or four non-DM patients were excluded, and the same exclusion criteria applied to the study group were also used before matching in both the DM control and non-DM control groups. The index date was regarded as the date of DR diagnosis in the study group, and the same day was also set as the index date in the matched DM control group and non-DM control group. Furthermore, the patients in the DR population were regarded as PDR if (1) NPDR-related diagnostic codes were present and (2) PDR-related codes were present with an additional diagnosis of retinal neovascularisation, vitreous haemorrhage, tractional retinal detachment or neovascular glaucoma. The parameters between the NPDR and PDR subgroups were evaluated in the following analyses.
Main outcome measurement
The development of sight-threatening cataracts was regarded as the primary outcome in the current study, and it was based on the emergence of senile/complicated/diabetic cataract-related diagnostic codes plus the receipt of cataract surgery after the index date. Cataract-related diagnostic codes that clearly indicate the underlying aetiology were not included in the current study to prevent confusion or overestimation. In addition, only those subjects who received those diagnostic codes by an ophthalmologist were recognised as having achieved the outcome and were included in the study.
Demographic data and comorbidities
To increase the homogeneity of the health status of each subject, we also assessed the influence of age, sex and the following systemic diseases in the multivariable analysis: hypertension, ischaemic heart diseases, hyperlipidaemia, congestive heart failure, cerebrovascular disease, dementia, chronic pulmonary disease, rheumatoid arthritis and osteoarthritis, kidney disease, liver cirrhosis, alcoholic liver disease, gout, atopic dermatitis, and allergic otolaryngologic diseases. Similarly, dry eye diseases (DEDs), uveitis, glaucoma, age-related macular degeneration (AMD) and the scheduling of trans pars plana vitrectomy (TPPV) were considered in the multivariate model to standardise the ocular condition. Since certain medications may lead to the development of cataracts, the following medications were also included in our analysis: systemic steroids, including prednisolone, methylprednisolone, hydrocortisone, triamcinolone, and dexamethasone; topical steroids, including prednisolone, fluorometholone, betamethasone, triamcinolone, dexamethasone, phenothiazines, and amiodarone; statins, including rosuvastatin, atorvastatin, simvastatin, pravastatin, lovastatin, fluvastatin and pitavastatin; and metformin. We traced the data in the NHIRD longitudinally from the index date of each participant to (1) the date of sight-threatening cataract development, (2) withdrawal from the National Health Insurance program, or (3) the time of 31 December 2016.
SAS version 9.4 (SAS Institute Inc., NC, USA) was applied for all statistical analyses. After performing age- and sex-matching (1:4 ratio) for the study group, DM-control group and non-DM control groups, the chi-square test and independent t test were applied to compare the differences in age, sex, comorbidities and medications, including hypoglycaemic agents, between the study and control groups, and the independent t test was also used to analyse differences in ophthalmic characteristics, including the percentage of diabetic macular oedema (DME), ratio of laser photocoagulation and ratio of intravitreal injection, between the patients with NPDR and the individuals with PDR. Then, a Poisson regression was conducted to calculate the incidence rate, crude relative risk and related 95% confidence intervals (CIs). We conducted a Cox proportional hazard regression to calculate the adjusted hazard ratios (aHRs) of sight-threatening cataracts among the three groups by incorporating the aforementioned parameters in the multivariate analysis, which included age; sex; the duration of DM; the presence of hypertension, ischaemic heart diseases, hyperlipidaemia, congestive heart failure, cerebrovascular disease, dementia, chronic pulmonary disease, rheumatoid arthritis and osteoarthritis, kidney disease, liver cirrhosis, alcoholic liver disease, gout, atopic dermatitis, allergic otolaryngologic diseases, DED, uveitis, glaucoma, and AMD; scheduling of TPPV; and the use of systemic steroids (including prednisolone, methylprednisolone, hydrocortisone, triamcinolone, and dexamethasone), topical steroids (including prednisolone, fluorometholone, betamethasone, triamcinolone, dexamethasone, phenothiazines, and amiodarone), statins (including rosuvastatin, atorvastatin, simvastatin, pravastatin, lovastatin, fluvastatin and pitavastatin), and metformin. To investigate the effect of DR severity on the development of sight-threatening cataracts, the study group was further divided into NPDR and PDR subgroups in the multivariate model, and the aHR of each subgroup was analysed. Moreover, a sensitivity analysis with the aHR of sight-threatening cataracts stratified by age, sex, severity of DR and DM interval (less than 2 years, 2 to 5 years and more than 5 years) was performed for the subgroups. Additionally, we drew Kaplan–Meier curves to present the cumulative probability of sight-threatening cataracts among the study, DM control and non-DM control groups. Subsequently, the log rank test was used to evaluate the significance among all three survival curves. Since almost all the individuals enrolled in the NHIRD belong to the Han/Taiwanese population, ethnicity was thus not considered a possible confounding factor. Statistical significance was set at P < 0.05.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.