Individualized optimization of the screening interval for diabetic retinopathy: a new model

Individualised screening for diabetic retinopathy with proliferative retinopathy and macular oedema as separate end points

PURPOSE

A number of algorithms have been developed to calculate screening intervals for diabetic retinopathy on the basis of individual risk factors. However, these approaches have not considered proliferative diabetic retinopathy (PDR) and diabetic macular oedema (DME) as separate end points and death as competing risk.

METHODS

A multi-state survival model with death as competing risk was used to predict the screening interval for diabetic retinopathy based on information about all 2446 patients from a well-defined population who had started treatment for either PDR or DME during 25 years. The performance of the model was tested on the existing database and at seven screening sites on patients who had not developed a treatment requiring condition.

RESULTS

Testing on the existing database showed that at a risk level of 2% the algorithm could predict a screening interval with a success rate higher than 90% and a 1.75 times average prolongation of the screening interval without failing to detect the development of verified PDR og DME. The model was limited to a diabetes duration shorter than 40 years and depended on knowledge of relevant risk factors. At the other participating screening sites the algorithm predicted shorter intervals than the screener.

CONCLUSIONS

Algorithms for individualised screening for diabetic retinopathy can prolong screening intervals without losing patients who develop a vision threatening condition. The calculation of screening intervals requires access to relevant risk factors and should be developed on large data sets that reflect the population in which the algorithm should be used.

Role of Systemic Factors in Improving the Prognosis of Diabetic Retinal Disease and Predicting Response to Diabetic Retinopathy Treatment

Topic

To review clinical evidence on systemic factors that might be relevant to update diabetic retinal disease (DRD) staging systems, including prediction of DRD onset, progression, and response to treatment.

Clinical relevance

Systemic factors may improve new staging systems for DRD to better assess risk of disease worsening and predict response to therapy.

Methods

The Systemic Health Working Group of the Mary Tyler Moore Vision Initiative reviewed systemic factors individually and in multivariate models for prediction of DRD onset or progression (i.e., prognosis) or response to treatments (prediction).

Results

There was consistent evidence for associations of longer diabetes duration, higher glycosylated hemoglobin (HbA1c), and male sex with DRD onset and progression. There is strong trial evidence for the effect of reducing HbA1c and reducing DRD progression. There is strong evidence that higher blood pressure (BP) is a risk factor for DRD incidence and for progression. Pregnancy has been consistently reported to be associated with worsening of DRD but recent studies reflecting modern care standards are lacking. In studies examining multivariate prognostic models of DRD onset, HbA1c and diabetes duration were consistently retained as significant predictors of DRD onset. There was evidence of associations of BP and sex with DRD onset. In multivariate prognostic models examining DRD progression, retinal measures were consistently found to be a significant predictor of DRD with little evidence of any useful marginal increment in prognostic information with the inclusion of systemic risk factor data apart from retinal image data in multivariate models. For predicting the impact of treatment, although there are small studies that quantify prognostic information based on imaging data alone or systemic factors alone, there are currently no large studies that quantify marginal prognostic information within a multivariate model, including both imaging and systemic factors.

Conclusion

With standard imaging techniques and ways of processing images rapidly evolving, an international network of centers is needed to routinely capture systemic health factors simultaneously to retinal images so that gains in prediction increment may be precisely quantified to determine the usefulness of various health factors in the prognosis of DRD and prediction of response to treatment.

Financial Disclosures

Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Development and validation of predictive risk models for sight threatening diabetic retinopathy in patients with type 2 diabetes to be applied as triage tools in resource limited settings

Background

Delayed diagnosis and treatment of sight threatening diabetic retinopathy (STDR) is a common cause of visual impairment in people with Type 2 diabetes. Therefore, systematic regular retinal screening is recommended, but global coverage of such services is challenging. We aimed to develop and validate predictive models for STDR to identify 'at-risk' population for retinal screening.

Methods

Models were developed using datasets obtained from general practices in inner London, United Kingdom (UK) on adults with type 2 Diabetes during the period 2007-2017. Three models were developed using Cox regression and model performance was assessed using C statistic, calibration slope and observed to expected ratio measures. Models were externally validated in cohorts from Wales, UK and India.

Findings

A total of 40,334 people were included in the model development phase of which 1427 (3·54%) people developed STDR. Age, gender, diabetes duration, antidiabetic medication history, glycated haemoglobin (HbA1c), and history of retinopathy were included as predictors in the Model 1, Model 2 excluded retinopathy status, and Model 3 further excluded HbA1c. All three models attained strong discrimination performance in the model development dataset with C statistics ranging from 0·778 to 0·832, and in the external validation datasets (C statistic 0·685 - 0·823) with calibration slopes closer to 1 following re-calibration of the baseline survival.

Interpretation

We have developed new risk prediction equations to identify those at risk of STDR in people with type 2 diabetes in any resource-setting so that they can be screened and treated early. Future testing, and piloting is required before implementation.

Funding

This study was funded by the GCRF UKRI (MR/P207881/1) and supported by the NIHR Biomedical Research Centre at Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology.

Diabetic retinopathy screening in the emerging era of artificial intelligence

Diabetic retinopathy is a frequent complication in diabetes and a leading cause of visual impairment. Regular eye screening is imperative to detect sight-threatening stages of diabetic retinopathy such as proliferative diabetic retinopathy and diabetic macular oedema in order to treat these before irreversible visual loss occurs. Screening is cost-effective and has been implemented in various countries in Europe and elsewhere. Along with optimised diabetes care, this has substantially reduced the risk of visual loss. Nevertheless, the growing number of patients with diabetes poses an increasing burden on healthcare systems and automated solutions are needed to alleviate the task of screening and improve diagnostic accuracy. Deep learning by convolutional neural networks is an optimised branch of artificial intelligence that is particularly well suited to automated image analysis. Pivotal studies have demonstrated high sensitivity and specificity for classifying advanced stages of diabetic retinopathy and identifying diabetic macular oedema in optical coherence tomography scans. Based on this, different algorithms have obtained regulatory approval for clinical use and have recently been implemented to some extent in a few countries. Handheld mobile devices are another promising option for self-monitoring, but so far they have not demonstrated comparable image quality to that of fundus photography using non-portable retinal cameras, which is the gold standard for diabetic retinopathy screening. Such technology has the potential to be integrated in telemedicine-based screening programmes, enabling self-captured retinal images to be transferred virtually to reading centres for analysis and planning of further steps. While emerging technologies have shown a lot of promise, clinical implementation has been sparse. Legal obstacles and difficulties in software integration may partly explain this, but it may also indicate that existing algorithms may not necessarily integrate well with national screening initiatives, which often differ substantially between countries.

Using Machine Learning Techniques to Develop Risk Prediction Models for the Risk of Incident Diabetic Retinopathy Among Patients With Type 2 Diabetes Mellitus: A Cohort Study

OBJECTIVE

To construct and validate prediction models for the risk of diabetic retinopathy (DR) in patients with type 2 diabetes mellitus.

METHODS

Patients with type 2 diabetes mellitus hospitalized over the period between January 2010 and September 2018 were retrospectively collected. Eighteen baseline demographic and clinical characteristics were used as predictors to train five machine-learning models. The model that showed favorable predictive efficacy was evaluated at annual follow-ups. Multi-point data of the patients in the test set were utilized to further evaluate the model's performance. We also assessed the relative prognostic importance of the selected risk factors for DR outcomes.

RESULTS

Of 7943 collected patients, 1692 (21.30%) developed DR during follow-up. Among the five models, the XGBoost model achieved the highest predictive performance with an AUC, accuracy, sensitivity, and specificity of 0.803, 88.9%, 74.0%, and 81.1%, respectively. The XGBoost model's AUCs in the different follow-up periods were 0.834 to 0.966. In addition to the classical risk factors of DR, serum uric acid (SUA), low-density lipoprotein cholesterol (LDL-C), total cholesterol (TC), estimated glomerular filtration rate (eGFR), and triglyceride (TG) were also identified to be important and strong predictors for the disease. Compared with the clinical diagnosis method of DR, the XGBoost model achieved an average of 2.895 years prior to the first diagnosis.

CONCLUSION

The proposed model achieved high performance in predicting the risk of DR among patients with type 2 diabetes mellitus at each time point. This study established the potential of the XGBoost model to facilitate clinicians in identifying high-risk patients and making type 2 diabetes management-related decisions.

The SMART Framework: Selection of Machine learning Algorithms with ReplicaTions - a Case Study on the Microvascular Complications of Diabetes

Over 34 million people in the US have diabetes, a major cause of blindness, renal failure, and amputations. Machine learning (ML) models can predict high-risk patients to help prevent adverse outcomes. Selecting the 'best' prediction model for a given disease, population, and clinical application is challenging due to the hundreds of health-related ML models in the literature and the increasing availability of ML methodologies. To support this decision process, we developed the Selection of Machine-learning Algorithms with ReplicaTions (SMART) Framework that integrates building and selecting ML models with decision theory. We build ML models and estimate performance for multiple plausible future populations with a replicated nested cross-validation technique. We rank ML models by simulating decision-maker priorities, using a range of accuracy measures (e.g., AUC) and robustness metrics from decision theory (e.g., minimax Regret). We present the SMART Framework through a case study on the microvascular complications of diabetes using data from the ACCORD clinical trial. We compare selections made by risk-averse, -neutral, and -seeking decision-makers, finding agreement in 80% of the risk-averse and risk-neutral selections, with the risk-averse selections showing consistency for a given complication. We also found that the models that best predicted outcomes in the validation set were those with low performance variance on the testing set, indicating a risk-averse approach in model selection is ideal when there is a potential for high population feature variability. The SMART Framework is a powerful, interactive tool that incorporates various ML algorithms and stakeholder preferences, generalizable to new data and technological advancements.

The Role of Telemedicine, In-Home Testing and Artificial Intelligence to Alleviate an Increasingly Burdened Healthcare System: Diabetic Retinopathy

In the presence of the ever-increasing incidence of diabetes mellitus (DM), the prevalence of diabetic eye disease (DED) is also growing. Despite many improvements in diabetic care, DM remains a leading cause of visual impairment in working-age patients. So far, prevention has been the best way to protect vision. The sooner we diagnose DED, the more effective the treatment is. Thus, diabetic retinopathy (DR) screening, especially with imaging techniques, is a method of choice for vision protection. To alleviate the burden of diabetic patients who need ophthalmic care, telemedicine and in-home testing are used, supported by artificial intelligence (AI) algorithms. This is why we decided to evaluate current image teleophthalmology methods used for DR screening. We searched the PubMed platform for papers published over the last 5 years (2015–2020) using the following key words: telemedicine in diabetic retinopathy screening, diabetic retinopathy screening, automated diabetic retinopathy screening, artificial intelligence in diabetic retinopathy screening, smartphone diabetic retinopathy testing. We have included 118 original articles meeting the above criteria, discussing imaging diabetic retinopathy screening methods. We have found that fundus cameras, stable or mobile, are most commonly used for retinal photography, with portable fundus cameras also relatively common. Other possibilities involve the use of ultra-wide-field (UWF) imaging and even optical coherence tomography (OCT) devices for DR screening. Also, the role of smartphones is increasingly recognized in the field. Retinal fundus images are assessed by humans instantly or remotely, while AI algorithms seem to be useful tools facilitating retinal image assessment. The common use of smartphones and availability of relatively cheap, easy-to-use adapters for retinal photographs augmented by AI algorithms make it possible for eye fundus photographs to be taken by non-specialists and in non-medical setting. This opens the way for in-home testing conducted on a much larger scale in the future. In conclusion, based on current DR screening techniques, we can suggest that the future practice of eye care specialists will be widely supported by AI algorithms, and this way will be more effective.

Retinal oxygen saturation changes progressively over time in diabetic retinopathy

PURPOSE

According to cross-sectional studies, oxygen saturation is elevated in retinal vessels in diabetic patients. We evaluated how retinal oxygenation (metabolic marker), vessel diameters and retinopathy grade (structural markers) change over time in diabetic patients.

DESIGN

Prospective cohort study following screening in a hospital setting.

METHODS

Retinal oximetry images were acquired in 214 patients with the Oxymap T1 oximeter. Imaging was repeated after a median of 3.0 years (range 0.76-6.8 years). Oxygen saturation and vessel diameters were measured in the right eye. Semiquantitative grading of retinopathy according to international guidelines and red lesion count were performed on fundus photographs.

RESULTS

Retinopathy grade according to the international semiquantitative grading system was unchanged. Arteriolar saturation increased by 0.75±0.15 percentage points per year of follow-up (p<0.0001). Venular saturation increased by 1.74±0.26 percentage points per year (p<0.0001) and arteriovenous difference decreased by 0.99±0.20 percentage points per year (p<0.0001). Arteriolar diameters decreased by 2.7±8.5μm (p<0.0001) between visits and venular diameters decreased by 2.4±9.1μm (p = 0.0002). Median increase in red lesion count between visits was 2 lesions (range -128 to 212 lesions, p<0.0001). The change in red lesion count and change in diameters did not correlate with the length of follow-up (p>0.44).

CONCLUSIONS

Oxygen saturation in larger retinal vessels can increase and arteriovenous difference can decrease over time in diabetic patients without any observable changes in retinopathy grade. The results suggest that changes in retinal oxygen saturation may precede progression of diabetic retinopathy or that oxygen saturation is more sensitive to disease progression than retinopathy grade.

Quality assessment of a screening algorithm for diabetic retinopathy: prolongation of screening intervals with minimal effect of HbA1c and arterial hypertension on the length of intervals

PURPOSE

At Steno Diabetes Center Copenhagen (SDCC), diabetic retinopathy (DR) screening intervals are based on quantification of retinal lesions. Screening intervals are, for the milder forms of DR, prolonged to 2-3 years. The purpose of the present study was to evaluate the effect of the prolongation on developing unexpected events and to evaluate the effect of HbA_1c and arterial hypertension.

METHODS

We assessed 18 972 screening intervals from 6000 patients from 1/1-2003 to 1/5-2017 for occurrence of unexpected events, defined as: (1) DR progression requiring treatment, at the following screening date, and (2) DR-related hospital contact within the planned interval. We modelled the effect of several risk factors for developing unexpected events in a Cox regression. Furthermore, we assessed the risk of unexpected events in a logistic regression analysis using cubic splines to model the effect of HbA_1c , stratified by arterial hypertension status.

RESULTS

16 283 (86%) intervals followed the planned interval and among those, only 86 (0.5%) experienced unexpected events. Intervals of dysregulated patients (86% of all intervals) did not experience more unexpected events, compared with well-regulated patient intervals (Hazard Ratio: 1.12, 95% CI: 0.55-2.27). We found a nonlinear effect of HbA_1c on the risk of unexpected events which peaked around HbA_1c levels of 80 mmol/mol. Having arterial hypertension slightly increased the risk of unexpected events.

CONCLUSIONS

The present study supports the validity of the current algorithm. We found no increased risk of unexpected events among dysregulated intervals but a nonlinear effect of HbA_1c . Age, diabetes duration and diabetes type were significantly associated with unexpected events.

Prediction models for development of retinopathy in people with type 2 diabetes: systematic review and external validation in a Dutch primary care setting

AIMS/HYPOTHESIS

The aims of this study were to identify all published prognostic models predicting retinopathy risk applicable to people with type 2 diabetes, to assess their quality and accuracy, and to validate their predictive accuracy in a head-to-head comparison using an independent type 2 diabetes cohort.

METHODS

A systematic search was performed in PubMed and Embase in December 2019. Studies that met the following criteria were included: (1) the model was applicable in type 2 diabetes; (2) the outcome was retinopathy; and (3) follow-up was more than 1 year. Screening, data extraction (using the checklist for critical appraisal and data extraction for systemic reviews of prediction modelling studies [CHARMS]) and risk of bias assessment (by prediction model risk of bias assessment tool [PROBAST]) were performed independently by two reviewers. Selected models were externally validated in the large Hoorn Diabetes Care System (DCS) cohort in the Netherlands. Retinopathy risk was calculated using baseline data and compared with retinopathy incidence over 5 years. Calibration after intercept adjustment and discrimination (Harrell's C statistic) were assessed.

RESULTS

Twelve studies were included in the systematic review, reporting on 16 models. Outcomes ranged from referable retinopathy to blindness. Discrimination was reported in seven studies with C statistics ranging from 0.55 (95% CI 0.54, 0.56) to 0.84 (95% CI 0.78, 0.88). Five studies reported on calibration. Eight models could be compared head-to-head in the DCS cohort (N = 10,715). Most of the models underestimated retinopathy risk. Validating the models against different severities of retinopathy, C statistics ranged from 0.51 (95% CI 0.49, 0.53) to 0.89 (95% CI 0.88, 0.91).

CONCLUSIONS/INTERPRETATION

Several prognostic models can accurately predict retinopathy risk in a population-based type 2 diabetes cohort. Most of the models include easy-to-measure predictors enhancing their applicability. Tailoring retinopathy screening frequency based on accurate risk predictions may increase the efficiency and cost-effectiveness of diabetic retinopathy care.

REGISTRATION

PROSPERO registration ID CRD42018089122.

Reduced baseline diameter and contraction of peripheral retinal arterioles immediately after remote ischemia in diabetic patients

PURPOSE

Remote ischemic conditioning (RIC) implies that transient ischemia in one organ can affect blood flow and protect from ischemia in another remote organ such as the retina. The purpose of the present study was to investigate the effect of RIC on the diameter of retinal arterioles in patients with diabetic retinopathy and whether this effect differs among peripheral and macular vessels.

METHODS

In twenty type 1 diabetes patients aged 20-31 years, the Dynamic Vessel Analyzer (DVA) was used to measure diameters of peripheral and macular arterioles during rest, isometric exercise, and flicker stimulation. Measurements were obtained before, immediately after, and 1 h after RIC, and were compared to responses obtained from normal persons.

RESULTS

The reduced baseline diameter (p < 0.009) and contraction of peripheral retinal arterioles during isometric exercise (p = 0.025) observed immediately after RIC in normal persons were absent in the studied diabetic patients.

CONCLUSIONS

RIC affects the diameter of peripheral but not macular arterioles in normal persons, but the response is abolished in diabetic patients.

TRIAL REGISTRATION

NCT03906383.

Personalized risk-based screening for diabetic retinopathy: A multivariate approach versus the use of stratification rules

AIMS

To evaluate our proposed multivariate approach to identify patients who will develop sight-threatening diabetic retinopathy (STDR) within a 1-year screen interval, and explore the impact of simple stratification rules on prediction.

MATERIALS AND METHODS

A 7-year dataset (2009-2016) from people with diabetes (PWD) was analysed using a novel multivariate longitudinal discriminant approach. Level of diabetic retinopathy, assessed from routine digital screening photographs of both eyes, was jointly modelled using clinical data collected over time. Simple stratification rules based on retinopathy level were also applied and compared with the multivariate discriminant approach.

RESULTS

Data from 13 103 PWD (49 520 screening episodes) were analysed. The multivariate approach accurately predicted whether patients developed STDR or not within 1 year from the time of prediction in 84.0% of patients (95% confidence interval [CI] 80.4-89.7), compared with 56.7% (95% CI 55.5-58.0) and 79.7% (95% CI 78.8-80.6) achieved by the two stratification rules. While the stratification rules detected up to 95.2% (95% CI 92.2-97.6) of the STDR cases (sensitivity) only 55.6% (95% CI 54.5-56.7) of patients who did not develop STDR were correctly identified (specificity), compared with 85.4% (95% CI 80.4-89.7%) and 84.0% (95% CI 80.7-87.6%), respectively, achieved by the multivariate risk model.

CONCLUSIONS

Accurate prediction of progression to STDR in PWD can be achieved using a multivariate risk model whilst also maintaining desirable specificity. While simple stratification rules can achieve good levels of sensitivity, the present study indicates that their lower specificity (high false-positive rate) would therefore necessitate a greater frequency of eye examinations.

Prognostic prediction models for diabetic retinopathy progression: a systematic review

With the increasing incidence of diabetic retinopathy and its improved detection, there is increased demand for diabetic retinopathy treatment services. Prognostic prediction models have been used to optimise services but these were intended for early detection of sight-threatening retinopathy and are mostly used in diabetic retinopathy screening services. We wanted to look into the predictive ability and applicability of the existing models for the higher-risk patients referred into hospitals. We searched MEDLINE, EMBASE, COCHRANE CENTRAL, conference abstracts and reference lists of included publications for studies of any design using search terms related to diabetes, diabetic retinopathy and prognostic models. Search results were screened for relevance to the review question. Included studies had data extracted on model characteristics, predictive ability and validation. They were assessed for quality using criteria specified by PROBAST and CHARMS checklists, independently by two reviewers. Twenty-two articles reporting on 14 prognostic models (including four updates) met the selection criteria. Eleven models had internal validation, eight had external validation and one had neither. Discriminative ability with c-statistics ranged from 0.57 to 0.91. Studies ranged from low to high risk of bias, mostly due to the need for external validation or missing data. Participants, outcomes, predictors handling and modelling methods varied. Most models focussed on lower-risk patients, the majority had high risk of bias and doubtful applicability, but three models had some applicability for higher-risk patients. However, these models will also need updating and external validation in multiple hospital settings before being implemented into clinical practice.

Evidence-based Danish guidelines for screening of diabetic retinopathy

PURPOSE

Diabetic retinopathy (DR) is among the leading causes of visual loss in the working-age population. It is generally accepted that screening of DR is cost-effective and can detect DR before it becomes sight-threatening to allow timely treatment.

METHODS

A group of retinal specialists was formed by the Danish Ophthalmological Society with the aim to formulate contemporary evidence-based guidelines for screening of DR in order to implement these in the Danish screening system.

RESULTS

We hereby present evidence for DR-screening regarding (1) classification of DR, (2) examination techniques, (3) screening intervals and (4) automated screening. It is our recommendation that the International Clinical Retinopathy Disease Severity Scale should be used to classify DR. As a minimum, mydriatic two-field disc- and macular-centred images are required. In the case of suspected clinically significant diabetic macular oedema, supplementary optical coherence tomography can increase the diagnostic accuracy. There is solid evidence to support a flexible, individualized screening regimen. In particular, it is possible to prolong screening intervals to 24-48 months for patients with no or mild nonproliferative diabetic retinopathy (NPDR), but it is also possible to use extended intervals of 12-24 months for patients with moderate NPDR given that these are well-regulated regarding glycaemic control (HbA1c ≤ 53 mmol/mol) and blood pressure (≤130/80 mmHg). Automated screening of DR is encouraging but is not ready for implementation at present.

CONCLUSION

Danish evidenced-based guidelines for screening of DR support high-quality imaging and allow flexible, individualized screening intervals with a potential for extension to patients with low risk of DR progression.

Retinal oxygen saturation is an independent risk factor for the severity of diabetic retinopathy

BACKGROUND

The oxygen saturation in larger retinal vessels has been shown to increase with increasing diabetic retinopathy (DR) grade and to help predict the effect of antivascular endothelial growth factor treatment in patients with diabetic maculopathy. However, it is unknown to what extent the increased oxygen saturation co-varies with other risk factors and whether it is an independent risk factor for the severity of DR.

METHODS

Seven hundred and twenty-two successive patients referred for specialist evaluation of diabetic retinopathy including retinal oximetry were studied. Multiple regression analysis was used to investigate whether oxygen saturation in the larger retinal arterioles and venules contributed to the severity of diabetic retinopathy, independently of gender, age, diabetes duration, diabetes type, body mass index, blood pressure, haemoglobin A1c, visual acuity and central retinal thickness.

RESULTS

The included parameters could explain less than 15% of the variation in retinopathy grade. Approximately, one-third of the explained variation was related to the retinal oxygen saturation.

CONCLUSIONS

Prospective studies are needed to evaluate whether retinal oxygen saturation is predictive for the development of diabetic retinopathy and how it interacts with other biomarkers and risk factors over time.

Spatial distribution of early red lesions is a risk factor for development of vision-threatening diabetic retinopathy

AIMS/HYPOTHESIS

Diabetic retinopathy is characterised by morphological lesions related to disturbances in retinal blood flow. It has previously been shown that the early development of retinal lesions temporal to the fovea may predict the development of treatment-requiring diabetic maculopathy. The aim of this study was to map accurately the area where lesions could predict progression to vision-threatening retinopathy.

METHODS

The predictive value of the location of the earliest red lesions representing haemorrhages and/or microaneurysms was studied by comparing their occurrence in a group of individuals later developing vision-threatening diabetic retinopathy with that in a group matched with respect to diabetes type, age, sex and age of onset of diabetes mellitus who did not develop vision-threatening diabetic retinopathy during a similar observation period.

RESULTS

The probability of progression to vision-threatening diabetic retinopathy was higher in a circular area temporal to the fovea, and the occurrence of the first lesions in this area was predictive of the development of vision-threatening diabetic retinopathy. The calculated peak value showed that the risk of progression was 39.5% higher than the average. There was no significant difference in the early distribution of lesions in participants later developing diabetic maculopathy or proliferative diabetic retinopathy.

CONCLUSIONS/INTERPRETATION

The location of early red lesions in diabetic retinopathy is predictive of whether or not individuals will later develop vision-threatening diabetic retinopathy. This evidence should be incorporated into risk models used to recommend control intervals in screening programmes for diabetic retinopathy.

Screening Intervals for Diabetic Retinopathy and Implications for Care

PURPOSE OF REVIEW

The purpose of this study is to review the evidence that lower risk groups who could safely be screened less frequently for sight-threatening diabetic retinopathy (DR) than annually.

RECENT FINDINGS

Data have demonstrated that people with no DR in either eye are at a low risk of progression to sight-threatening DR over a 2-year period (event rate 4.8 per 1000 person years), irrespective of whether the screening method is one-field non-mydriatic or two-field mydriatic digital photography. Low risk has been defined as no retinopathy on two consecutive screening episodes or no retinopathy on one screening episode combined with risk factor data. The risk of an extension to 2 years is less than 5 per 1000 person years in a population with a national screening programme, and the general standard of diabetes care is relatively good, whether low risk is defined as no retinopathy on two consecutive screening episodes or no retinopathy on one screening episode combined with other risk factor data. The definition used in different populations is likely to depend on the availability of data.

The role of retinopathy distribution and other lesion types for the definition of examination intervals during screening for diabetic retinopathy

PURPOSE

It has previously been shown that the intervals between screening examinations for diabetic retinopathy can be optimized by including individual risk factors for the development of the disease in the risk assessment. However, in some cases, the risk model calculating the screening interval may recommend a different interval than an experienced clinician. The purpose of this study was to evaluate the influence of factors unrelated to diabetic retinopathy and the distribution of lesions for discrepancies between decisions made by the clinician and the risk model.

METHODS

Therefore, fundus photographs from 90 screening examinations where the recommendations of the clinician and a risk model had been discrepant were evaluated. Forty features were defined to describe the type and location of the lesions, and classification and ranking techniques were used to assess whether the features could predict the discrepancy between the grader and the risk model.

RESULTS

Suspicion of tumours, retinal degeneration and vascular diseases other than diabetic retinopathy could explain why the clinician recommended shorter examination intervals than the model. Additionally, the regional distribution of microaneurysms/dot haemorrhages was important for defining a photograph as belonging to the group where both the clinician and the risk model had recommended a short screening interval as opposed to the other decision alternatives.

CONCLUSIONS

Features unrelated to diabetic retinopathy and the regional distribution of retinal lesions may affect the recommendation of the examination interval during screening for diabetic retinopathy. The development of automated computerized algorithms for extracting information about the type and location of retinal lesions could be expected to further optimize examination intervals during screening for diabetic retinopathy.

Development of a cost-effectiveness model for optimisation of the screening interval in diabetic retinopathy screening

BACKGROUND

The English NHS Diabetic Eye Screening Programme was established in 2003. Eligible people are invited annually for digital retinal photography screening. Those found to have potentially sight-threatening diabetic retinopathy (STDR) are referred to surveillance clinics or to Hospital Eye Services.

OBJECTIVES

To determine whether personalised screening intervals are cost-effective.

DESIGN

Risk factors were identified in Gloucestershire, UK using survival modelling. A probabilistic decision hidden (unobserved) Markov model with a misgrading matrix was developed. This informed estimation of lifetime costs and quality-adjusted life-years (QALYs) in patients without STDR. Two personalised risk stratification models were employed: two screening episodes (SEs) (low, medium or high risk) or one SE with clinical information (low, medium-low, medium-high or high risk). The risk factor models were validated in other populations.

SETTING

Gloucestershire, Nottinghamshire, South London and East Anglia (all UK).

PARTICIPANTS

People with diabetes in Gloucestershire with risk stratification model validation using data from Nottinghamshire, South London and East Anglia.

MAIN OUTCOME MEASURES

Personalised risk-based algorithm for screening interval; cost-effectiveness of different screening intervals.

RESULTS

Data were obtained in Gloucestershire from 12,790 people with diabetes with known risk factors to derive the risk estimation models, from 15,877 people to inform the uptake of screening and from 17,043 people to inform the health-care resource-usage costs. Two stratification models were developed: one using only results from previous screening events and one using previous screening and some commonly available GP data. Both models were capable of differentiating groups at low and high risk of development of STDR. The rate of progression to STDR was 5 per 1000 person-years (PYs) in the lowest decile of risk and 75 per 1000 PYs in the highest decile. In the absence of personalised risk stratification, the most cost-effective screening interval was to screen all patients every 3 years, with a 46% probability of this being cost-effective at a £30,000 per QALY threshold. Using either risk stratification models, screening patients at low risk every 5 years was the most cost-effective option, with a probability of 99-100% at a £30,000 per QALY threshold. For the medium-risk groups screening every 3 years had a probability of 43-48% while screening high-risk groups every 2 years was cost-effective with a probability of 55-59%.

CONCLUSIONS

The study found that annual screening of all patients for STDR was not cost-effective. Screening this entire cohort every 3 years was most likely to be cost-effective. When personalised intervals are applied, screening those in our low-risk groups every 5 years was found to be cost-effective. Screening high-risk groups every 2 years further improved the cost-effectiveness of the programme. There was considerable uncertainty in the estimated incremental costs and in the incremental QALYs, particularly with regard to implications of an increasing proportion of maculopathy cases receiving intravitreal injection rather than laser treatment. Future work should focus on improving the understanding of risk, validating in further populations and investigating quality issues in imaging and assessment including the potential for automated image grading.

STUDY REGISTRATION

Integrated Research Application System project number 118959.

FUNDING DETAILS

The National Institute for Health Research Health Technology Assessment programme.

Early detection of type 2 diabetes mellitus and screening for retinopathy are associated with reduced prevalence and severity of retinopathy

PURPOSE

To explore whether the prevalence and severity of retinopathy differ in diabetes cohorts diagnosed through screening as compared with conventional health care.

METHODS

A total of 257 diabetes patients, 151 detected through screening and 106 through conventional clinical care, were included. Retinopathy was evaluated by fundus photography. The modified Airlie House adaptation of the Early Treatment Retinopathy Study protocol was used to grade the photographs. Averages of clinically collected fasting blood glucose (FBG), blood pressure and body mass index values were compiled from diabetes diagnosis until the eye examination. Blood chemistry, smoking habits and peripheral neuropathy were assessed at the time of the eye examination.

RESULTS

Among the screening-detected patients, 22% had retinopathy as compared to 51% among those clinically detected (p < 0.0001). In a multivariate analysis, patients with retinopathy were more likely to have increased average FBG (OR 1.42, 95% CI 1.19-1.70 per mmol/l) and peripheral neuropathy (OR 2.75, 95% CI 1.40-5.43), but less likely to have screening-detected diabetes (OR 0.31, 95% CI 0.17-0.57). Similar results were found using increasing severity grade of retinopathy as outcome. The cumulative retinopathy prevalence for the screening-detected diabetes cohort as compared with the clinically diagnosed cohort was significantly lower from 10 years' follow-up and onwards (p = 0.0002).

CONCLUSIONS

Among patients with screening-detected diabetes, the prevalence of retinopathy and increasing severity of retinopathy were significantly lower than among those who had their diabetes diagnosed through conventional care, even when other risk factors for retinopathy such as duration, hyperglycaemia and blood pressure were considered. Early detection of diabetes reduces prediagnostic time spent with hyperglycaemia. In combination with early and regular screening for retinopathy, more effective prevention against retinopathy can be provided.

Toward Big Data Analytics: Review of Predictive Models in Management of Diabetes and Its Complications

Diabetes is one of the top priorities in medical science and health care management, and an abundance of data and information is available on these patients. Whether data stem from statistical models or complex pattern recognition models, they may be fused into predictive models that combine patient information and prognostic outcome results. Such knowledge could be used in clinical decision support, disease surveillance, and public health management to improve patient care. Our aim was to review the literature and give an introduction to predictive models in screening for and the management of prevalent short- and long-term complications in diabetes. Predictive models have been developed for management of diabetes and its complications, and the number of publications on such models has been growing over the past decade. Often multiple logistic or a similar linear regression is used for prediction model development, possibly owing to its transparent functionality. Ultimately, for prediction models to prove useful, they must demonstrate impact, namely, their use must generate better patient outcomes. Although extensive effort has been put in to building these predictive models, there is a remarkable scarcity of impact studies.

A classification model for predicting eye disease in newly diagnosed people with type 2 diabetes

Diabetic retinopathy may be present at the time type 2 diabetes is diagnosed, and initial screening encompassing an eye examination performed by an ophthalmologist or optometrist is therefore recommended. However, proper screening for retinopathy may be challenging in many parts of the world. We hypothesized that simple, commonly available patient characteristics can be used to identify patients at high risk for having retinopathy. We investigated data from multiple years extracted from the National Health and Nutrition Examination Survey which holds information about blood glucose and eye examinations. Individuals with hitherto undiagnosed diabetes were classified according to the presence or absence of retinopathy. Linear classification was used to predict which patients had retinopathy at the time of diagnosis. A total of 266 individuals with undiagnosed diabetes were identified from the cohorts. Of these, 222 individuals had no sign of retinopathy, whereas 44 had mild or moderate non-proliferative retinopathy. Using information regarding HbA1c, BMI, waist circumference, age, systolic blood pressure, urinary albumin, and urinary creatinine, we were able to construct a model that predicts the presence of retinopathy with a positive predictive value of 22% and a negative predictive value of 99%. Only one true positive (1/44) with mild non-proliferative retinopathy was falsely classified. A classification model using readily available patient information and routine biochemical measures can be used to identify patients at high risk of having retinopathy at the time their diabetes is diagnosed. The model may be used to identify high-risk patients for retinopathy screening.

External validation of a risk assessment model to adjust the frequency of eye-screening visits in patients with diabetes mellitus

AIMS

To validate a sight-threatening diabetic retinopathy (STDR) risk assessment model to adjust the frequency of eye-screening visits in patients with diabetes mellitus.

METHODS

Retrospective follow-up study of patients with diabetes mellitus attending a diabetes center. Anonimyzed data on gender, type and duration of diabetes, HbA1c, blood pressure and the presence and grade of diabetic retinopathy were gathered to estimate risk for STDR for each individual's worse eye over time by means of a prediction model. Receiver operating characteristics (ROC) analysis was performed to determine the diagnostic ability of the model, and a calibration graph was done to see the model fit.

RESULTS

508 screening intervals were analyzed, median diabetes duration was 10years, 87% were type 2 diabetes mellitus, and 3.1% developed STDR before the next screening visit. The area under the ROC curve was 0.74, and the calibration graph showed that model had a good fit. The reduction in screening frequency was 40% compared with fixed annual screening.

CONCLUSIONS

Current prediction model used to estimate the risk of developing STDR in patients with diabetes performed well. A personalized screening frequency for diabetic retinopathy could be implemented in practice.

Extending the diabetic retinopathy screening interval beyond 1 year: systematic review

To determine whether the recommended screening interval for diabetic retinopathy (DR) in the UK can safely be extended beyond 1 year. Systematic review of clinical and cost-effectiveness studies. Nine databases were searched with no date restrictions. Randomised controlled trials (RCTs), cohort studies, prognostic or economic modelling studies which described the incidence and progression of DR in populations with type 1 diabetes mellitus or type 2 diabetes mellitus of either sex and of any age reporting incidence and progression of DR in relation to screening interval (vs annual screening interval) and/or prognostic factors were included. Narrative synthesis was undertaken. 14 013 papers were identified, of which 11 observational studies, 5 risk stratification modelling studies and 9 economic studies were included. Data were available for 262 541 patients of whom at least 228 649 (87%) had type 2 diabetes. There were no RCTs. Studies concluded that there is little difference between clinical outcomes from screening 1 yearly or 2 yearly in low-risk patients. However there was high loss to follow-up (13–31%), heterogeneity in definitions of low risk and variation in screening and grading protocols for prior retinopathy results. Observational and economic modelling studies in low-risk patients show little difference in clinical outcomes between 1-year and 2-year screening intervals. The lack of experimental research designs and heterogeneity in definition of low risk considerably limits the reliability and validity of this conclusion. Cost-effectiveness findings were mixed. There is insufficient evidence to recommend a move to extend the screening interval beyond 1 year.

The influence of background diabetic retinopathy in the second eye on rates of progression of diabetic retinopathy between 2005 and 2010

Purpose: The Gloucestershire Diabetic Eye Screening Programme offers annual digital photographic screening for diabetic retinopathy to a countywide population of people with diabetes. This study was designed to investigate progression of diabetic retinopathy in this programme of the English NHS Diabetic Eye Screening Programme.

Methods: Mydriatic digital retinal photographs of people with diabetes screened on at least 2 occasions between 2005 and 2010 were graded and included in this study if the classification at first screening was no DR (R0), background DR in one (R1a) or both eyes (R1b). Times to detection of referable diabetic retinopathy (RDR) comprising maculopathy (M1), preproliferative (R2) or proliferative retinopathy (R3) were analysed using survival models.

Results: Data were available on 19 044 patients, 56% men, age at screening 66 (57–74) years (median, 25th, 75th centile). A total of 8.3% of those with R1a and 28.2% of those with R1b progressed to any RDR, hazard ratios 2.9 [2.5–3.3] and 11.3 [10.0–12.8]. Similarly 7.1% and 0.11% of those with R1a progressed to M1 and R3, hazard ratios 2.7 [2.3–3.2] and 1.6 [0.5–5.0], compared to 21.8% and 1.07% of those with R1b, hazard ratio 9.1 [7.8–10.4] and 15.0 [7.1–31.5].

Conclusions: The risk of progression is significantly higher for those with background DR in both eyes than those with background retinopathy in only one or in neither eye.

Predicted impact of extending the screening interval for diabetic retinopathy: the Scottish Diabetic Retinopathy Screening programme

AIMS/HYPOTHESIS

The aim of our study was to identify subgroups of patients attending the Scottish Diabetic Retinopathy Screening (DRS) programme who might safely move from annual to two yearly retinopathy screening.

METHODS

This was a retrospective cohort study of screening data from the DRS programme collected between 2005 and 2011 for people aged ≥12 years with type 1 or type 2 diabetes in Scotland. We used hidden Markov models to calculate the probabilities of transitions to referable diabetic retinopathy (referable background or proliferative retinopathy) or referable maculopathy.

RESULTS

The study included 155,114 individuals with no referable diabetic retinopathy or maculopathy at their first DRS examination and with one or more further DRS examinations. There were 11,275 incident cases of referable diabetic eye disease (9,204 referable maculopathy, 2,071 referable background or proliferative retinopathy). The observed transitions to referable background or proliferative retinopathy were lower for people with no visible retinopathy vs mild background retinopathy at their prior examination (respectively, 1.2% vs 8.1% for type 1 diabetes and 0.6% vs 5.1% for type 2 diabetes). The lowest probability for transitioning to referable background or proliferative retinopathy was among people with two consecutive screens showing no visible retinopathy, where the probability was <0.3% for type 1 and <0.2% for type 2 diabetes at 2 years.

CONCLUSIONS/INTERPRETATION

Transition rates to referable diabetic eye disease were lowest among people with type 2 diabetes and two consecutive screens showing no visible retinopathy. If such people had been offered two yearly screening the DRS service would have needed to screen 40% fewer people in 2009.

A simple risk stratification for time to development of sight-threatening diabetic retinopathy

OBJECTIVE

The American Diabetes Association and the English NHS Diabetic Eye Screening Program recommend annual screening for diabetic retinopathy (DR) with referral to ophthalmology clinics of patients with sight-threatening DR (STDR). Using only longitudinal data from retinal photographs in the population-based NHS Diabetic Eye Screening Program in Gloucestershire, we developed a simple means to estimate risk of STDR.

RESEARCH DESIGN AND METHODS

From 2005, 14,554 patients with no DR or mild nonproliferative DR only at two consecutive annual digital photographic screenings were categorized by the presence of DR in neither, one, or both eyes at each screening and were followed for a further median 2.8 years.

RESULTS

Of 7,246 with no DR at either screening, 120 progressed to STDR, equivalent to an annual rate of 0.7%. Of 1,778 with no DR in either eye at first screening and in one eye at second screening, 80 progressed to STDR, equivalent to an annual rate of 1.9% and to a hazard ratio (HR) of 2.9 (95% CI 2.2-3.8) compared with those with no DR. Of 1,159 with background DR in both eyes at both screenings, 299 progressed to STDR equivalent to an annual rate of 11% and an HR of 18.2 (14.7-22.5) compared with individuals with no DR.

CONCLUSIONS

Combining the results from 2 consecutive years of photographic screening enables estimation of the risk of future development of STDR. In countries with systematic screening programs, these results could inform decisions about screening frequency.

[Screening for diabetic retinopathy in type 2 diabetes: a critical review of an annual routine]

BACKGROUND

The German diabetes guideline published in 2010 recommends an annual retinopathy screening for all type 2 diabetics. Patients' and physicians' questions about the need for this routine procedure prompted our critical review.

METHODS

Based on guidelines from six industrial nations, recent scientific studies, and from the type 2 disease management programme of North Rhine-Westphalia we reviewed alternative retinopathy screening strategies and their implications for over- and underuse.

RESULTS

A comparison of the English, Australian, US, Swedish, Canadian and German guidelines shows that a fixed screening interval is recommended in four countries, while an individualised, risk profile-adjusted screening interval of up to two to three years is favoured in two countries (Sweden, Canada). Current studies indicate that diabetes patients without retinopathy may safely be screened every two to three years without adverse health outcomes if performed consequently.

CONCLUSIONS

An individualised retinopathy screening based on the patient's risk factor profile may be a reasonable alternative to achieve the best possible health outcome and to avoid overuse.

Individual risk assessment and information technology to optimise screening frequency for diabetic retinopathy

AIMS/HYPOTHESIS

The aim of this study was to reduce the frequency of diabetic eye-screening visits, while maintaining safety, by using information technology and individualised risk assessment to determine screening intervals.

METHODS

A mathematical algorithm was created based on epidemiological data on risk factors for diabetic retinopathy. Through a website, www.risk.is , the algorithm receives clinical data, including type and duration of diabetes, HbA(1c) or mean blood glucose, blood pressure and the presence and grade of retinopathy. These data are used to calculate risk for sight-threatening retinopathy for each individual's worse eye over time. A risk margin is defined and the algorithm recommends the screening interval for each patient with standardised risk of developing sight-threatening retinopathy (STR) within the screening interval. We set the risk margin so that the same number of patients develop STR within the screening interval with either fixed annual screening or our individualised screening system. The database for diabetic retinopathy at the Department of Ophthalmology, Aarhus University Hospital, Denmark, was used to empirically test the efficacy of the algorithm. Clinical data exist for 5,199 patients for 20 years and this allows testing of the algorithm in a prospective manner.

RESULTS

In the Danish diabetes database, the algorithm recommends screening intervals ranging from 6 to 60 months with a mean of 29 months. This is 59% fewer visits than with fixed annual screening. This amounts to 41 annual visits per 100 patients.

CONCLUSION

Information technology based on epidemiological data may facilitate individualised determination of screening intervals for diabetic eye disease. Empirical testing suggests that this approach may be less expensive than conventional annual screening, while not compromising safety. The algorithm determines individual risk and the screening interval is individually determined based on each person's risk profile. The algorithm has potential to save on healthcare resources and patients' working hours by reducing the number of screening visits for an ever increasing number of diabetic patients in the world.