Study participants were recruited from the baseline visit of the Inter99 study, a randomised non-pharmacological intervention study of ischaemic heart disease prevention [7]. An age- and sex-stratified random sample of 12,934 eligible individuals aged 30–60 years was invited to participate, of whom 6,906 (53.4%) volunteered. In brief, the Inter99 Eye study examined 970 participants from the Inter99 study between 1999 and 2001, including a random sample of 563 participants and a skewed sample of 407 participants with a high-risk profile defined by a compound cardiovascular risk score [8]. Diabetes-related data were available for 937 participants. The present analysis excluded six participants, three with type 1 diabetes (serum C-peptide < 100 pmol/l) and three with ungradeable or missing fundus photographs, thus leaving 205 participants with type 2 diabetes for analysis. In 17 participants lens fluorometry was unavailable (lens opacities in three, instrument failure in 14).
All participants gave their written informed consent. The protocols were in accordance with the Helsinki Declaration and approved by the local medical ethics committees. The Inter99 study was registered in 2005 at ClinicalTrials.gov (registration no. NCT00289237).
After pupil dilation, participants underwent an ophthalmic examination including cataract assessment and 7-field non-stereoscopic 60-degree digital fundus photography (TRC-50X camera; Topcon, Tokyo, Japan) with a 1,024 × 1,024 pixel CV-1000 back-piece (AngioVision 1000, MediVision, Tel Aviv, Israel) in red-free illumination (Wratten 54 filter; Eastman Kodak, Rochester, NY, USA) and fovea-centred stereoscopic colour diapositive photography (Ektachrome Elite 100; Eastman Kodak). Fundus characteristics were assessed by an ophthalmologist (I. C. Munch) masked to all systemic variables. The photographs were graded for retinal lesions and classified according to the Early Treatment Diabetic Retinopathy Study (ETDRS) retinopathy severity scale. The maximum grade in any of the seven standard fields in either of a participant’s eyes defined the participant’s retinopathy level.
Lens fluorescence was measured using a non-invasive ocular fluorometer (Fluorotron, OcuMetrics, Mountain View, CA, USA). Excitation was from 430 to 490 nm and detection 530–630 nm. Lens fluorescence was measured along the optical axis of the eye and measurements were corrected for loss of light by absorption in the lens. The mean of six scans, three per eye, was used in analysis. All fluorescence values were calibrated to an external solution of fluorescein and expressed as equivalent concentration of fluorescein in ng/ml (ng f-eq/ml). The coefficient of variation for repeated measurements of lens fluorescence is below 3% [3].
Participants were defined as having diabetes mellitus when fasting plasma glucose (FPG) was 7.0 mmol/l or higher or when 2 h plasma glucose during the oral glucose tolerance test was 11.1 mmol/l or higher or if participants responded to the study questionnaire that they had been diagnosed with diabetes by a physician.
The lower of two blood pressure recordings made after at least 5 min of rest was used for data analysis. Smoking was assessed by questionnaire, daily smoking being defined as an average consumption of ≥7 g of tobacco per day for more than 1 year.
Multiple logistic regression and general linear modelling (proc LOGISTIC and GLM, SAS software package version 9.1, SAS institute, Cary, NC, USA) was used for data analysis. Two-tailed p values and 95% CIs were calculated by the use of Wald’s test. Trends were estimated as the slope when categorical variables of interest were treated as quantitative variables. Log-linear assumptions were tested by likelihood ratio tests. Tests for interaction were performed by adding the cross product to the model.