BMJ Open Diabetes Research & Care

Epidemiology/Health services research

Prevalence and risk factors for diabetic retinopathy at diagnosis of type 2 diabetes: an observational study of 77 681 patients from the Swedish National Diabetes Registry

Sheyda Sofizadeh, Katarina Eeg-Olofsson, Marcus Lind



To assess the prevalence of diabetic retinopathy (DR) in persons with newly diagnosed type 2 diabetes (T2D) to understand the potential need for intensified screening for early detection of T2D.

Research design and methods

Individuals from the Swedish National Diabetes Registry with a retinal photo <2 years after diagnosis of T2D were included. The proportion of patients with retinopathy (simplex or worse) was assessed. Patient characteristics and risk factors at diagnosis were analyzed in relation to DR with logistic regression.


In total, 77 681 individuals with newly diagnosed T2D, mean age 62.6 years, 41.1% females were included. Of these, 13 329 (17.2%) had DR.

DR was more common in older persons (adjusted OR 1.03 per 10-year increase, 95% CI 1.01 to 1.05) and men compared with women, OR 1.10 (1.05 to 1.14). Other variables associated with DR were OR (95% CI): lower education 1.08 (1.02 to 1.14); previous stroke 1.18 (1.07 to 1.30); chronic kidney disease 1.29 (1.07 to 1.56); treatment with acetylsalicylic acid 1.14 (1.07 to 1.21); ACE inhibitors 1.12 (1.05 to 1.19); and alpha blockers 1.41 (1.15 to 1.73). DR was more common in individuals born in Asia (OR 1.16, 95% CI 1.08 to 1.25) and European countries other than those born in Sweden (OR 1.11, 95% CI 1.05 to 1.18).


Intensified focus on screening of T2D may be needed in Sweden in clinical practice since nearly one-fifth of persons have retinopathy at diagnosis of T2D. The prevalence of DR was higher in men, birthplace outside of Sweden, and those with a history of stroke, kidney disease, and hypertension.


  • Diabetic retinopathy at diagnosis of type 2 diabetes (T2D) is used as a surrogate marker to indicate late detected T2D, but contemporary and population-based studies are sparse.


  • The study reveals that a significant proportion (17.2%) of individuals newly diagnosed with T2D in Sweden already have DR at diagnosis, indicating that a significant proportion of patients have had long-term hyperglycemia before diagnosis.


  • Attention is needed in clinical practice in Sweden regarding screening for T2D in persons with a risk profile and further research is urgently needed regarding potential benefits of structured screening in the population.
From the article


Diabetic retinopathy (DR) is the most common microvascular complication of diabetes.1High blood glucose levels are a critical risk factor for DR, and the risk and severity of DR are directly related to glycated hemoglobin A1c (HbA1c) level over time in both type 1 diabetes and type 2 diabetes (T2D).2–5Since DR typically develops over several years, individuals with DR at diagnosis of T2D generally have elevated blood glucose levels long before diagnosis.6Hypertension in conjunction with hyperglycemia is also a well-established risk factor for DR progression.7Other risk factors that have been associated with retinopathy in persons with T2D are Body Mass Index (BMI), dyslipidemia, insulin treatment, and nephropathy.8–11

The Swedish National Diabetes Registry (NDR) includes the majority of persons with T2D within the country.12Diabetes care in Sweden has significantly improved over time and more patients are reaching glucose control targets. Given intensive treatment in patients with newly diagnosed T2D, undetected hyperglycemia before diagnosis of T2D may be at least as harmful or more so than after diagnosis of T2D. When T2D is undetected individuals may unknowingly have glycemic levels clearly above targets, while after diagnosis modern diabetes care enables patients in many instances to achieve HbA1c targets associated with low risk of diabetes complications.12Early hyperglycemia can also be detrimental over time by virtue of legacy effects, and before diagnosis patients do not receive the same level of attention in terms of screening and treatment for complications.13 14

Prevalence of retinopathy at diagnosis of T2D has been used as a surrogate marker for late detected T2D in several other studies.15The aim of the current study was to evaluate to what extent DR exists in persons with newly diagnosed T2D in Sweden and to investigate factors related to increased risk of DR among patients included in the NDR, which includes the absolute majority of persons with T2D in Sweden.


In this nationwide study from Sweden, using DR at diagnosis of T2D as a marker for late detected T2D, almost one-fifth of patients had DR at diagnosis of T2D. DR was more common in men, individuals born in Asia, and those with a history of stroke and kidney disease. High SBP and elevated HbA1c levels were also associated with DR. A higher proportion of patients with normal weight had DR at diagnosis of T2D compared with those who were overweight or obese. DR was less common in individuals with previous CHD.

Prevalence of retinopathy as an indicator for late detected T2D has been used in several earlier studies.15However, contemporary population-based studies of the prevalence of DR are overall spars. In a UK-based study examining newly diagnosed persons with T2D until year 2017, the prevalence of DR ranged from 14% to 25% depending on whether pre-diabetes had been recorded as diagnosis or not before diagnosis of T2D.28A systematic review and meta-analysis including studies generally performed more than 10 years ago found that the pooled prevalence of DR at diagnosis of T2D was 14.6% (95% CI 11.9% to 17.3%).15Some studies have reported that DR is present in up to 15%–20% of patients at the time of diagnosis of T2D, while others have reported that DR is present in around 5%–10%.6 15 29–33

Hyperglycemia and hypertension are risk factors for DR in persons with established T2D as confirmed in randomized settings.2 7Studies have also reported hyperglycemia and hypertension to be more common in patients with DR at diagnosis of T2D.1 15 34DR at diagnosis of T2D has also been reported to be more common in persons with renal complications whereas smoking has shown divergent associations.5 32In different populations of individuals with DR has been more common in men compared with women.35

Experience from clinical practice and studies in type 1 diabetes, where the initial hyperglycemia is generally more abrupt, suggest that hyperglycemia generally needs to exist over a long period of time before DR appears.3 4Data indicate that diabetes is generally present for at least 5 years before signs of retinopathy appear, and it may be more than 10 years after diagnosis of diabetes before clinical diagnosis of DR.6That almost one-fifth of patients in the current study had DR at diagnosis of T2D indicates that long-standing hyperglycemia before diagnosis of T2D is relatively common in Sweden, and hyperglycemia increases risk of complications at diagnosis of T2D. Furthermore, legacy effects of earlier hyperglycemia may worsen prognosis after diagnosis compared with persons with early detection.13 14Moreover, many individuals do not receive treatments for preventing diabetes complications before diagnosis of T2D such as lipid-lowering and antihypertensive drugs, lifestyle advice, and screening programs for complications.23It is possible that diabetes complications and mortality can be reduced during this high-risk phase if diabetes is detected early, and intensive prevention programs are started. ACE inhibitors and angiotensin-2 receptor blockers are likely beneficial in preventing or slowing the progression of early DR.36Further, studies indicate that the use of antiplatelet/anticoagulant medications may reduce the risk of developing non-proliferative DR among patients with T2D while fibrates may benefit diabetic macular edema.36 37

Diabetes care in Sweden has significantly improved over time with a large proportion of persons with T2D obtaining a target HbA1c level <52 mmol/mol (6.9%).38However, that a relatively large proportion of patients have DR at diagnosis of T2D indicates that strategies for detecting T2D at earlier stages need to improve. Although diabetes care for persons with established T2D has substantially improved over time, detecting diabetes at an early stage has not achieved corresponding success.12When clearly elevated glucose levels exist before diagnosis, the harm due to legacy effects will likely not be evident until later years.13 14

Guidelines suggest that overweight and obese individuals should be screened for T2D.23 39Other individuals in focus are first-degree relatives of individuals with T2D, that is, having a hereditary component. Specific risk scores exist that can be used for screening for T2D.40However, clearly structured programs for screening risk groups are lacking in most countries, while screening is generally random and, in many instances, may be missed. In the ADDITION study, structured screening for T2D was evaluated, but clear benefits on a population level could not be confirmed.41More research is needed into implementing structured screening programs for at-risk persons with T2D to detect disease at an early stage. Currently, by greater focus in clinical practice by extended screening of T2D, it may also be possible to detect pre-diabetes and prevent T2D more efficiently through lifestyle interventions.42

In the current study, most risk factors for DR at diagnosis were expected. However, we did not expect that those with high BMI were less likely to have DR compared with those with normal weight. It is possible that individuals with normal BMI who end up developing T2D may be screened later for T2D after a more long-term hyperglycemia. It was also of interest that individuals born in Asia and then migrating to Sweden had higher risk of DR at diagnosis of T2D compared with those born in Sweden. One possible explanation is that this patient group may be less informed regarding T2D risk factors and need for screening. Another is that disease progression differs since persons born in Asia who are not overweight or obese generally develop T2D more often compared with those born in Western countries.43 44Retinopathy progression has shown to be more common in certain ethnic groups in earlier studies including Indian, Pakistani, and South Asian African ethnic groups.10 45 46

One strength of the current study is the population-based design where the NDR covers the majority of persons with T2D in Sweden. A limitation is that 44% of the newly diagnosed had no data available in the NDR on retinopathy less than 2 years after diagnosis of T2D and were therefore not included in the current analysis. However, patient characteristics were similar overall among included and excluded patients indicating major selection bias is not likely. Although some patient characteristics differed between the included and excluded patients, they were overall numerically small, except for HbA1c where a somewhat greater difference existed at 58.1 mmol/mol (7.5%) vs 54.3 mmol/mol (7.1%). Mean HbA1c was somewhat lower among excluded patients possibly indicating slightly lower prevalence of DR in this population. Nevertheless, even if a lower proportion of excluded patients had DR, the overall proportion of patients having DR would still be relatively high. It is unclear to what extent those patients without data on DR in the NDR lacked a retinal screening or if results of screening had not been recorded. The NDR is dependent on health professionals registering information on retinopathy in the NDR based on clinical eye examinations. The study was limited that a minority of patients had information on albuminuria, creatinine levels, and grading of retinopathy at the time of diagnosis of T2D and these variables were therefore not included in the analyses.

Since a large proportion of persons with T2D in Sweden reach HbA1c targets, indicating high overall quality of diabetes care compared with many other countries, similar challenges in terms of detecting persons with T2D at an early stage of hyperglycemia seem likely in other European countries and parts of the world. It also seems likely that slightly lowering glycemic targets (eg, from 52 mmol/mol to 48 mmol/mol) in patients with established T2D, often intensively debated, may have relatively little influence on prognosis,16whereas many individuals with much higher levels remain undetected in turn leading to complications already at diagnosis. Therefore, we view early detection of T2D as a key challenge to resolve in the field of T2D.

In conclusion, intensified screening for T2D in clinical practice is needed in Sweden since almost one-fifth of these persons have retinopathy at diagnosis indicating long-standing hyperglycemia. The prevalence of DR was higher in certain patient groups including men, birthplace outside of Sweden, and those with a history of stroke, kidney, disease, and high SBP. Further research is needed to develop efficient strategies and programs to not only screen for T2D at random in clinical practice but also more structured screening to detect T2D earlier. This is of particular concern since many persons may have hyperglycemia before diagnosis and are not targets of efficient prevention strategies for complications before diagnosis.





www red DiabetologNytt