Article Text

Download PDFPDF
Epidemiology, risk factors and management of paediatric diabetic retinopathy
  1. Marla B Sultan1,2,
  2. Carla Starita3,
  3. Kui Huang1
  1. 1Pfizer Inc, New York, New York, USA
  2. 2Department of Ophthalmology, The New York Eye and Ear Infirmary, New York, New York, USA
  3. 3Pfizer Ltd, Tadworth, Surrey, UK
  1. Correspondence to Marla B Sultan, Pfizer Inc, 235 East 42nd Street, 8th Floor, New York, NY 10017, USA; marla.b.sultan{at}


Diabetic retinopathy (DR), a common complication of both type 1 and type 2 diabetes, is rarely expressed at a level greater than background retinopathy during childhood and adolescence. Epidemiological studies in paediatric diabetic patients together with data from the Diabetes Control and Complications Trial have demonstrated the importance of glycaemic control in delaying or preventing the development of DR; thus, the incidence of DR has declined somewhat over the past two decades. Both prepubertal and postpubertal years with diabetes contribute to the overall probability of DR development. In addition to duration of disease and degree of glycaemic control, other risk factors for DR development include elevated blood pressure, lipid profiles, serum levels of advanced glycation endproducts, evidence for early stage atherosclerosis, increased calibre of retinal blood vessels and several genetic factors, such as enzymes involved in glucose and lipid metabolism. Annual screening is recommended, with mydriatic stereoscopic fundus photography being the most sensitive detection method. Both pathophysiology and treatment in paediatric populations are essentially the same as described for adult populations, with treatment usually not required until adulthood is reached.

  • Adolescent
  • child
  • diabetic macular oedema
  • diabetic retinopathy
  • retina

Statistics from


Diabetic retinopathy (DR) is a major cause of vision loss in Western countries and is distinguished among other causes of visual morbidity in that it is reported across all age groups ≥40 years.1 Owing to the time required for its expression after the onset of diabetes, clinical retinopathy is rarely seen in paediatric populations. However, there is considerable literature examining the development of DR during childhood and adolescent years and the measures that are available to ameliorate its progression.2–4 This review focuses on the epidemiology, pathophysiology and treatment of DR in paediatric populations.


Since several years of diabetes usually are required for the development of clinically relevant complications, the clinical manifestations of DR have been reported rarely in children and adolescents.2 3 5–7 Thus, a common parameter in paediatric studies is the presence of background retinopathy and a minimum presence of microaneurysm.4

The majority of published epidemiological studies have involved patients with type 1 diabetes3 5 8–14 with findings dependent on disease duration and age group. In a prevalence study of early diabetes complications in which the dominant effect of disease duration was excluded from the calculations, results indicated that increasing age and later pubertal staging influenced the prevalence of retinopathy.12 In 201 Australian children12 aged <15 years with type 1 diabetes (median duration, 6 years; median haemoglobin A1c (HbA1c), 8.7%), background retinopathy was detected in 8% of children <11 years of age and in 28% >11 years of age.3 Other reports of the prevalence of background retinopathy in type 1 diabetic patients have varied from as low as 4.5% (prepubertal) in 504 French children and adolescents aged 10–18 years14 to 22% for the presence of any retinopathy in 99 Tanzanian children aged 5–18 years.13

The institution of intensive glycaemic control, as shown in the Diabetes Control and Complications Trial (DCCT),15 16 has assisted in significantly reducing the occurrence of DR, while the rising incidence of type 2 diabetes17–19 in paediatric populations has increased it. In a cross-sectional analysis of complications of diabetes mellitus in 878 Australian adolescents matched for age and disease duration (median age, 14.6 years; median duration, 7.5 years) with type 1 disease, the occurrence of background retinopathy declined significantly from 49% to 31% to 24% for the time periods 1990–1994, 1995–1998, and 1999–2002, respectively (p<0.0001).20 The presence of clinical retinopathy also declined from 12% in the earliest period to 3% over the last two periods combined. This may reflect the fact that from 1999 to 2002, more diabetic patients were treated with three or more insulin injections per day (p<0.0001), while insulin dose also increased (p<0.0001); however, median HbA1c was unchanged.20 The authors suggested that frequent injections may have minimised wide glucose level fluctuations and harmful postprandial spikes. A more recent review concluded that data supporting harmful postprandial hyperglycaemia remain at best equivocal21 so that the underlying reason for the reduction in the occurrence of clinical retinopathy remains unclear.

Relatively few epidemiological studies have examined the incidence of childhood retinopathy in type 2 diabetes.22 In Australian youths aged <18 years with type 1 diabetes (n=1433) and type 2 disease (n=68), the rate of retinopathy (20% vs 4%; p=0.04) and the mean duration of diabetes in patients (6.8 vs 1.3 years; p<0.001) with type 1 disease were at least five times greater than that in patients with type 2 disease.7 This significant difference in the rate of retinopathy may be solely related to the duration of diabetes. In a study of New Zealanders of predominantly non-European descendants, the frequencies of background and sight-threatening retinopathy were each 4% in type 2 patients (mean age at diagnosis, 20 years) after a mean of 3 years' duration.23 In another study focusing on 1052 New Zealand Maori with diabetes, 51 (5%) were diagnosed before age 30 years.24 Among this younger cohort, the average age of diagnosis was 12.4 years for type 1 and 19.4 years for type 2 diabetes. Frequencies of retinopathy among type 1 and type 2 patients were 17% (3/18) and 35% (9/26), respectively, after mean durations of 8.4 and 10.1 years; these differences were not statistically significant.24 In comparison, in a study of 40 Caribbean Hispanic and African American type 2 diabetic children, diagnosed at a mean age of 15.5 years and examined at an average disease duration of 21.7 months, only 1 (2.5%) had background retinopathy.25 Similarly, an investigation of microvascular complications in Pima Indians, who have a strong predisposition to type 2 diabetes, reported that among 928 youths aged 15–19 years who did not have retinopathy at a baseline examination, 30 (3%) subsequently developed it. The mean follow-up in this population was 14 years.26

Risk factors

Time since diagnosis of diabetes mellitus

Two key parameters affecting prognosis are duration of diabetes and degree of glycaemic control. The earliest signs of background DR rarely occur before the fifth year of disease with the prevalence reaching 50% by year 10.2 Although there have been occasional case reports of proliferative DR (PDR) in adolescents,15 clinical retinopathy is extremely rare before the onset of puberty.2 27 28 The likelihood of developing vision-limiting pathology increases with age.27 In a 24-year study of patients diagnosed with diabetes mellitus before the age of 15 years, 84% of the 194 patients without baseline retinopathy developed it over the study course, with 5% progressing to PDR. By contrast, 20% of the 97 patients having DR at baseline progressed to PDR, and 3 of them became blind.29 Similarly, in the Wisconsin Epidemiological Study of Diabetic Retinopathy, among the population diagnosed with diabetes at <30 years of age (mean 14.6 years), the prevalence of DR was 17% at <5 years duration, increasing to 97.5% after 15 years; PDR varied from 1.2% at <10 years, increasing to 67% after ≥35 years.30

Degree of glycaemic control

The degree of glycaemic control has been the most clearly established parameter affecting the development of retinopathy. Maintenance of glycaemic control is influenced by a series of unique challenges in children and adolescents, including hormonal fluctuations, psychological effects of weight gain and puberty, nutritional problems, complexities of family relationships,31–34 and, in some cases, impaired cognitive function.33 35 These issues impose special responsibilities on the primary care provider and requires support from family, peer group and school.

Several studies have assessed the influence of long-term glycaemic control on DR in children and adolescents with type 1 diabetes. To date, the DCCT in which intensive and conventional therapy were compared with a mean 6.5 years follow-up, has been the most comprehensive.36 A subset of this study included 195 adolescents (ages 13–17 years) for whom intensive treatment was found to reduce the risk of developing clinically significant retinopathy by 53% in those with no retinopathy at baseline and to delay the progression in those with mild retinopathy at baseline by 70%.

In an observational study of adolescent patients who had completed the DCCT, the Epidemiology of Diabetes Interventions and Complications (EDIC) Research Group reported benefits after 4 years in patients who received intensive treatment compared with those who received conventional treatment, despite similar HbA1c levels between groups (8.38% vs 8.45%, respectively).37 After 10 years, these benefits had disappeared for those diagnosed during adolescence16; however, they remained for those diagnosed in adulthood.16 38 This difference in long-term outcomes may reflect a lesser sensitivity to glycaemic control among the younger cohort in combination with higher HbA1c levels.16 The reasons for the metabolic memory effect observed in the DCCT/EDIC, allowing for sustained benefit even years after cessation of intensive control, are unknown; one potential mechanism of advanced glycation endproduct (AGE) accumulation is reduced.39 40 Prevention of hyperglycaemia-induced epigenetic modification of DNA has also been proposed, and will be evaluated in the EDIC epigenetics study.16 Finally, the Diabetes Incidence Study in Sweden, also found glycaemic control to be a significant predictor of the incidence of retinopathy 10 years after diagnosis in young adults (HbA1c 8.1±1.5% and 6.8±1.2% for those with and without retinopathy, respectively; p<0.001).41 Taken together, these studies provide strong evidence that maintaining strict glycaemic control, while avoiding hypoglycaemia, limits the development and progression of DR. The fact that the metabolic memory effect observed in the DCCT/EDIC is less pronounced in children and adolescents than in adults indicates that maintaining strict control is even more important in the younger cohort.


Of considerable interest is the role of puberty in the development of DR. One early report had suggested that prepubertal years of diabetes were not significant in contributing to DR progression,42 but numerous studies, predominantly in patients with type 1 diabetes, have since concluded that the duration of prepubertal years of diabetes does indeed contribute to the cumulative disease impact.8–11 43 44 However, the available data do suggest that not all years are equal in this respect. Comparative studies indicate that progression is more rapid in the postpubertal years concomitant with sexual maturation.11 43 45 In a detailed examination, a sparing effect was found in patients <5 years of age after which the risk of clinical retinopathy increased for each year of prepubertal and postpubertal duration.44

Blood pressure and lipid profiles

Several major clinical trials in adults have provided evidence that blood pressure-lowering drugs,46–53 as well as therapies to improve lipid profiles54 55 can lessen the risk of DR progression.56 57 With respect to lipid profiles, a comparative study of paediatric patients with and without retinopathy found that high-density lipoprotein cholesterol was related to the development of retinopathy.58 A study of 1869 adolescents with type 1 diabetes reported the association of DR development and progression with higher systolic and diastolic blood pressures.59 Moreover, studies have shown that some antihypertensive medications, notably the ACE inhibitors and angiotensin receptor blockers, may exert beneficial effects in DR that are independent of blood pressure lowering.53 57 60 61 Possible physiological mechanisms for these benefits include neuronal protection,62 increased insulin sensitivity,63 anti-inflammatory actions,64 inhibition of blood-retinal barrier breakdown65 and restoration of the activity of glyoxalase-I, an enzyme essential for AGE detoxification.66

Advanced glycation endproducts

AGEs are believed to contribute to the deleterious effects of hyperglycaemia on the microvasculature67 and elevated AGE serum levels have been measured in children and adolescents with type 1 diabetes.68 69 These levels are even higher in patients with background or preproliferative retinopathy or microalbuminuria.69 Plantar fascia thickness, a measure of collagen glycation that can be ascertained non-invasively by ultrasound, also is correlated with subsequent development of retinopathy in adolescents with type 1 diabetes.70

Evidence for early stages of atherosclerosis

Endothelial dysfunction has been confirmed in children and adolescents3 with diabetes and has been correlated to defects in folate metabolism71 and lipid status.72 In a study with adolescent type 1 diabetic patients, flow-mediated dilatation and intima-media thickness, two parameters of endothelial function, were abnormal among diabetic patients with retinopathy in comparison with non-diabetic controls as well as with diabetic patients without retinopathy.73 Moreover, elevated serum concentrations of elastin-derived peptides, arterial wall breakdown products, have also been found in diabetic children with microvascular complications compared with those without these complications.74

Associations with retinal vascular parameters

Numerous studies have assessed the relationship between effects of diabetes on retinal arterial and venous calibre with increases in both parameters, in particular venous calibre, being related to the development or progression of DR.56 Evidence has been adduced that these correlations are relevant in the paediatric population for both type 1 and type 2 diabetes. In a prospective cohort study of 645 type 1 diabetic patients, 43% developed retinopathy over a median of 2.5 years, with patients in the quartile of the largest retinal arteriolar calibre showing a threefold higher risk of DR than those in the lowest quartile.75 Similarly, in a study comparing adolescents with type 2 diabetes to age-matched controls without diabetes, venous calibre was significantly increased in the diabetic group.76 The mechanisms by which changes in vessel calibre are related to development of DR remain to be established. Arteriolar dilation may contribute to the aetiology of DR by leading to elevated capillary pressure and consequent oedema56 but it is also possible that the relationship is strictly correlative.

In addition to examining the relationship between retinal vessel calibre and the development of DR, the issue of fractal dimension of the retinal vasculature has also been examined in paediatric patients. An initial report indicated that prevalence of DR was significantly correlated with increased fractal dimension.77 However, a subsequent prospective study by the same group found that for young diabetic patients without DR, development of DR over 2.9 years was not related to fractal dimension, suggesting that increased fractal dimension was a consequence, rather an antecedent, of DR.78

Genetic correlations

The rapid advances in sequencing technology have prompted studies for specific markers and genome-wide associations that may predict DR or be targets for therapeutic intervention. A number of candidate gene mutations or their regulatory regions have already been revealed.79 80 It is likely that many of these associations will be similar in both paediatric and adult populations; however, as discussed in the section on glycaemic control, there appears to be age-dependent differences in the disease expression. Notably, several studies have focused on the paediatric cohort.3 81–85

To date, an increased risk for complications in adolescents with type 1 diabetes mellitus has been associated with genes for enzymes involved in the regulation of oxidation products. These include aldose reductase,82 glutathione-S-transferase,86 manganese superoxide dismutase86 and paraoxonase.81 83 A link to elevated DR risk has also been reported for concomitant polymorphisms in genes for enzymes involved in folate metabolism.84 Conversely, a polymorphism in the gene for interleukin 6 in adolescents with type 1 diabetes has been found to have a protective effect.85 Further studies will likely permit a better delineation of the metabolic pathways related, either positively or negatively, to DR development in paediatric populations with diabetes.

Aetiology, pathophysiology and clinical manifestations

The aetiology of the vascular damage characteristic of DR/diabetic macular oedema (DMO) is believed to involve protein kinase C activation, polyol and AGE accumulation, and angiogenic growth factor upregulation.87–90 The first morphological alterations, characteristic of background retinopathy are microaneurysms in areas of hypoperfusion followed by blood-retinal barrier damage and consequent hard lipid exudate deposition, nerve fibre infarctions and haemorrhage. The subsequent development of the neovascularisation characteristic of PDR, in response to local ischaemia, is especially vision threatening since new blood vessels are prone to leakage and haemorrhage.91

Factors contributing to the impaired vascular integrity include pericyte loss, basement membrane thickening and the proliferation and death by apoptosis of endothelial cells.92 This is accompanied by inflammatory cell influx that contributes to both capillary blockage and dropout, induces endothelial cell death and promotes pathological neovascularisation.92 93 One key mediator of several underlying processes is vascular endothelial growth factor (VEGF), the most potent known promoter of vascular permeability,94 which furthers blood-retinal barrier breakdown, the disruption of tight junctions95 and inflammatory cell influx.93 VEGF is also a potent endothelial cell mitogen,96 accelerating the endothelial cell hyperplasia of non-proliferative (NP) DR and of pathological angiogenesis of the proliferative phase.90 97

Diabetic retinopathy is categorised into three stages of severity, ranging from early, middle (typical in paediatric patients with DR) to advanced (involving aberrant blood vessel growth on the retina) (table 1).98 Diabetic macular oedema (DMO), the principal cause of vision loss in NPDR,99 can develop at any stage and is defined as clinically significant based on the presence of retinal thickening or the development of hard exudates within 500 μm of the macular centre, or of retinal thickening 1 disc area or larger within 1 disc diameter of the centre.100 In recent studies, an operational definition of centre-involved macular oedema has also been employed, including the presence of definite retinal thickening involving the centre of the macula and assessed to be the main cause of visual loss, together with retinal thickness measured on optical coherence tomography ≥250 μm in the central subfield.101–103

Table 1

Stages of diabetic retinopathy (with permission, Aiello et al 199898)

While the principal retinal changes in DR/DMO are microvascular, neural retinal function can also be affected.104 Evidence of focal retinal neuropathy has been found to precede the appearance of retinopathy in paediatric patients. This includes impairments in contrast sensitivity,105 colour vision,106 visual field defects107 and delayed implicit times on multifocal electroretinogram.108


Direct and indirect ophthalmoscopy, non-mydriatic fluorescein angiography and mydriatic stereoscopic fundus photography are screening tools for DR, the latter two being the most sensitive.28 Mydriatic stereoscopic fundus photography was reported to be four times more sensitive than direct ophthalmoscopy alone for detecting DR and is the preferred screening method.28 Haemorrhages and microaneurysms seen through the ophthalmoscope are often the earliest abnormality detected and occur without visual loss unless macular oedema also is present.2 Recently, optical coherence tomography has been employed to assess retinal thickness in adolescent patients with type 2 diabetes.109

Annual screening for DR recently has been recommended by the International Society for Paediatric and Adolescent Diabetes in patients aged 11 years after diabetes of 2 years' duration and from 9 years of age with diabetes of 5 years' duration, yet the optimal screening interval is unclear.4 An earlier review of the various screening recommendations27 concluded that annual screening should begin 3–5 years postdiagnosis in patients aged 9–10 years. However, in another study involving 668 children and adolescents with type 1 diabetes, the risk of developing retinopathy did not increase significantly 1 year after initial screening in any of the evaluated groups (aged <11 years, ≥11 years or higher-risk (10 years of diabetes duration or HbA1c >10% at any assessment)).28 DR did increase significantly after 2 years in the older patients but not until 6 years in the youngest group; no patients had proliferative retinopathy or required treatment. The authors suggested that a 2-year interval would be sufficient to detect vision-threatening DR.28 It is unclear as to whether or not these recommendations also apply to children and adolescents with type 2 diabetes, which is becoming increasingly more prevalent.17–19 In any case, a visit to an ophthalmologist should go beyond just a retinal examination and include counselling, education regarding the benefits of optimal metabolic control,110 and be used to establish a rapport with a patient who will require life-long ophthalmic care.


There is very little information available regarding the treatment of paediatric patients for DR/DMO. The Early Treatment of Diabetic Retinopathy Study evaluating photocoagulation for the treatment of DR and DMO included only adults aged 18–70 years.110 Since the earliest signs of DR rarely occur before the fifth year of disease, with the prevalence reaching 50% by the tenth year, clinical retinopathy is extremely rare before the onset of puberty.2 In one study of DR progression in children and adolescents, none required laser therapy or surgery.28 Studies suggest that few paediatric patients have DR of sufficient severity to require treatment. At this time, focal or panretinal photocoagulation is the treatment of choice for PDR or DMO, with surgical options for those with late-stage PDR.2 56 There has been recent approval of anti-VEGF therapy for the treatment of DMO; however, there is no information with regard to paediatric use at this time, and it is not approved for the paediatric population.111


Although clinical DR remains rare among paediatric patients with diabetes, regular monitoring for the appearance of complications is essential given the progressive nature of the disease and the chronic nature of the underlying hyperglycaemia. One group, the International Society for Paediatric and Adolescent Diabetes, suggests that annual screening for DR be conducted in patients aged 11 years after diabetes of 2 years' duration and from 9 years of age with diabetes of 5 years' duration. The single most important intervention remains prevention through adherence to strict glycaemic control achieved without inducing hypoglycaemic complications. However, at least one study has shown that adherence to the clinical guidelines, which include HbA1c monitoring and dilated eye examinations in paediatric patients, is suboptimal.112 113 The authors call for the conduct of additional studies to further evaluate adherence to evidence-based guidelines in children.


The authors thank Lauren Swenarchuk of Zola Associates for providing editorial support, including updating the literature search, incorporating new data into the manuscript as instructed by the authors and styling the paper for journal submission.



  • Funding This study was supported by Pfizer Inc.

  • Competing interests MBS, CS and KH are all employees of Pfizer Inc.

  • Provenance and peer review Commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.