Diabetic Nephropathy, Glycemic Control and Pseudophakic Macular Edema After Cataract Surgery

Cataract and IOL Surgery, Cataract (Post-operative Cataract Surgery Follow-up), Diabetes Mellitus, Type 2, Diabetic Nephropathies, Pseudophakic Macular Edema

pseudophakic macular edema, diabetic nephropathy, phacoemulsification, HbA1c, central macular thickness, optical coherence tomography, UACR, eGFR, KDIGO, cataract surgery

This retrospective cohort study examines whether the kidney complications of diabetes (diabetic nephropathy) and blood-sugar control before surgery help predict swelling of the central retina, called pseudophakic macular edema (PME), after cataract surgery in people with type 2 diabetes. PME is a common cause of blurred vision after otherwise successful cataract surgery, and people with diabetes are at higher risk. Doctors currently estimate this risk mainly from the HbA1c blood-sugar level and the stage of diabetic eye disease (retinopathy). It is not yet known whether diabetic kidney disease, measured by the urine albumin-to-creatinine ratio (UACR) and estimated kidney function (eGFR), adds independent information about who will develop PME. Using medical records already collected during routine care at Thammasat University Hospital, the investigators will measure central retinal thickness on optical coherence tomography (OCT) scans before uncomplicated phacoemulsification cataract surgery with a lens implant, and again 1 and 3 months afterward. PME is defined as at least a 30% increase in central macular thickness. Patients with and without baseline diabetic nephropathy will be compared, and statistical modeling will identify the best preoperative HbA1c threshold for predicting PME. The results are intended to improve preoperative risk counseling and to help identify diabetic patients who may need closer monitoring after cataract surgery. Because only existing records are used, the study involves no additional visits, tests, or risks for patients.

Background and rationale: Pseudophakic macular edema (PME) is a leading cause of suboptimal visual recovery after otherwise uncomplicated cataract surgery, and diabetes mellitus is an established risk factor. Preoperative risk assessment currently relies mainly on glycemic control (HbA1c) and diabetic retinopathy stage. Diabetic nephropathy and diabetic retinopathy share a common microvascular pathophysiology, yet it has not been established whether baseline diabetic nephropathy carries information about PME risk independently of retinopathy stage and HbA1c. The HbA1c threshold of 7% widely cited in clinical practice also derives from receiver operating characteristic analysis that was not stratified by renal status. Objectives: The study tests whether preoperative HbA1c and baseline diabetic nephropathy severity are independent predictors of PME after uncomplicated phacoemulsification in patients with type 2 diabetes, and derives an optimized preoperative HbA1c cut-point for PME prediction, stratified by nephropathy status. Setting and data sources: The study is conducted at Thammasat University Hospital, a tertiary academic centre in Pathum Thani, Thailand. Data are abstracted from the hospital electronic medical record, the laboratory information system, the operative-procedure database, and archived spectral-domain optical coherence tomography (OCT) examinations. Cohort entry (T = 0) is the date of cataract surgery. The surgical window is January 2020 to December 2025, with a pre-specified extension back to January 2018 if the target sample is not reached. Because the analysis uses pre-existing clinical records only, there is no patient contact, no additional visit or investigation, and no assigned intervention. Measurement of the exposure: Renal status is classified using the KDIGO 2022 framework from the preoperative spot urine albumin-creatinine ratio and the estimated glomerular filtration rate, with eGFR calculated by the CKD-EPI 2021 creatinine equation. Diabetic retinopathy severity is recorded using the ETDRS classification and lens hardness using the LOCS III nuclear opalescence grade. Measurement of the outcome: Central macular thickness is measured on Spectralis (Heidelberg Engineering) or Cirrus HD-OCT (Carl Zeiss Meditec) using the manufacturer's automated macular mapping protocol (macular cube 512 x 128 or an equivalent high-density raster scan), and the device is recorded for each patient. Scans are graded independently by two ophthalmologists masked to clinical and laboratory data. Inter-rater reliability is quantified by the intraclass correlation coefficient, and disagreements are resolved by a senior retinal specialist. Scans of inadequate quality (signal strength below the manufacturer threshold, segmentation failure, or motion artifact) are withdrawn from analysis and logged. Data quality control: Data are captured on a standardized electronic case report form with an accompanying data dictionary, implemented in REDCap with range checks, field-type validation, and logic checks. A second investigator verifies abstracted values against the source record: 100% verification of the first 20 to 30 records to confirm process stability, then random audit of at least 15% of records for the remainder of the study. Discrepancies are resolved by joint review, with the principal investigator adjudicating any that remain unresolved. Sample size justification: The target enrollment was derived from a two-proportion z-test for unequal allocation (2:1 unexposed to exposed), assuming PME incidence of 5% in unexposed and 15% in exposed eyes, a two-sided alpha of 0.05 and 80% power, with a 15% allowance for incomplete or unusable records. The resulting target provides at least 10 events per candidate predictor for the multivariable model. Statistical analysis: Continuous variables are summarized as mean with standard deviation or median with interquartile range according to normality (Shapiro-Wilk test and Q-Q plots), and categorical variables as frequency and percentage. The primary analysis is multivariable logistic regression with PME as the dependent variable. HbA1c, urine albumin-creatinine ratio, eGFR, diabetic retinopathy stage, diabetes duration, and insulin use are forced into the model a priori; other variables are screened univariably at p \< 0.20; and the model is refined by backward elimination with likelihood-ratio testing. Results are reported as adjusted odds ratios with 95% confidence intervals, model fit is assessed by the Hosmer-Lemeshow goodness-of-fit test and Nagelkerke R-squared, and multicollinearity by the variance inflation factor. Receiver operating characteristic analysis with the Youden index identifies the optimal preoperative HbA1c cut-point in the whole cohort and separately in the nephropathy and non-nephropathy subgroups; areas under the curve are estimated with 95% confidence intervals by the DeLong method and compared between subgroups using the DeLong test for correlated curves. Pre-specified subgroup analyses are by KDIGO albuminuria category (A1, A2, A3) and eGFR stage (G1 to G2 versus G3 to G5). Sensitivity analyses exclude eyes with severe non-proliferative or proliferative retinopathy, apply the alternative Level 2 threshold for central macular thickness increase, and test an HbA1c by nephropathy status interaction term. Missing-data patterns are assessed with Little's MCAR test; multiple imputation by chained equations (20 imputations) is applied to variables with less than 30% missing data, variables with more than 30% missing are excluded from the primary analysis, and complete-case analysis is reported as a sensitivity analysis. Analyses use SPSS v28.0, R v4.3 or later (pROC and mice packages), and MedCalc v23 or later, with two-sided p \< 0.05 considered statistically significant. Reporting: The study is reported in accordance with the STROBE statement for observational studies and the TRIPOD statement for multivariable prediction-model studies.

No related papers found yet.

Thailand