When should screening for diabetic retinopathy begin for children with type 1 diabetes?

Treatment of diabetic retinopathy

Visual impairment and blindness as a result of diabetic retinopathy (DR) are amongst the most feared complications of diabetes. However, the prevalence of sight-threatening diabetic retinopathy (STDR) has been slowly decreasing[1–4]. Recently, it has been reported that diabetes is no longer the leading cause of blindness in the working age population in the United Kingdom[5]. These observations may reflect the cumulative impact of better management of diabetes, the introduction of screening programs aiming to identify STDR, and more active and effective ophthalmologic management. Good glycemic and blood pressure control are pivotal in the primary prevention of DR, with some evidence of direct benefit from fibrates when used in those with dyslipidemia[6,7]. The introduction of intensive insulin therapy to optimize glycemic control in children aged 13–17 years with type 1 diabetes (T1DM) was seen to reduce the risk of developing DR by 53%[8]. The benefit of such intensive management in the adolescent years remained evident many years later (legacy effect), even when HbA_1c values were similar for the intensively treated and control groups[9]. Currently, the treatment for STDR, which encompasses proliferative DR (PDR) and selected cases of severe non-proliferative DR (pre-proliferative DR (PPDR), and maculopathy, is primarily by laser photocoagulation and, more recently, in conjunction with intravitreal injections of inhibitors of vascular endothelial growth factors (anti-VEGF). The relatively recent addition of anti-VEGF treatment has improved visual outcomes in those with PDR and clinically significant macular edema (CSMO)[10]. If these measures are deemed inadequate then vitrectomy may be required. It is generally acknowledged that DR remains asymptomatic until it reaches an advanced stage (STDR) and that the benefit from treatment with laser photocoagulation is best achieved the earlier the diagnosis and treatment are delivered. This is the basis for the introduction of screening for DR, which has also been shown to be cost-effective, especially in comparison to the societal cost of blindness[11]. The detection of any DR will help to prompt the need for improving and maintaining good glycemic control to prevent progression to STDR. In the context of this review, glycemic control is usually worse in adolescents compared with younger-age children [12]

Guidelines for DR screening in children and young persons with diabetes

DR screening programs have been instituted in several countries over the last 20 years, aiming for the early detection and close monitoring of retinal changes related to diabetes in an attempt to prevent loss of vision or blindness as the primary goal and secondarily, to instigate interventions to prevent the progression of non-STDR. National structured and community-based DR screening programs for DR have been operational in the United Kingdom since 2003. Currently, screening is offered to all persons with diabetes from the age of 12, and carried out on an annual basis, as recommended by the Royal College of Ophthalmologists[13]. However, the question remains unresolved as to when is the most appropriate age to begin screening for DR in persons with T1DM. Recommendations vary between programs/countries, based either on the duration of diabetes or age of the children (Table 1). The American Academy of ophthalmology currently recommends annual screening for all with a duration of diabetes of more than 5 years, whilst the American Academy of Paediatrics recommends an initial screening 3–5 years after diagnosis of diabetes if over age 9 and annually thereafter, and the American Diabetes Association also recommends that screening begin 3–5 years after diagnosis of diabetes and once the child is 10 years old. However, in Europe, the International Society for Pediatric and Adolescent Diabetes (2011) recommends screening for DR from the age of 11 after 2 years of diabetes and from the age of 9 for those with diabetes of 5 years or more. The Retinopathy Working Party recommends screening from the onset of puberty[14]. In Scandinavia, Finland begin DR screening when patients enter puberty [15], and Sweden commence DR screening from the age of 10 years[16].

Table 1. Previous publications reporting prevalence of DR in children with T1DM.

Publication	Study setting/population characteristics	Screening details	Clinical parameters measured	Prevalence of DR	Significant findings	Conclusions
Kubin 2011 [17]	Finland Population-based 297 persons with T1DM Mean age: 11.9 years Mean duration of diabetes: 4.9 years	Fundus photographs following dilation Canon CF 60 1 × 60° image per eye captured	Duration of diabetes, gender, stage of puberty (using Tanner classification), relative height for age, relative weight for height, blood pressure, HbA1c, urinary albumin content, and lipids	Prevalence of DR: 11.8%, all had mild BDR Prevalence of DR in males: 7.6%, females: 16.5% Youngest age at which DR was found was 8 years	DR was associated with older age, a longer duration of diabetes, a higher HbA1c, and gender.	Signs of DR in childhood may act as a risk marker for florid DR in early adulthood. Pinpointing those people at risk for intensified treatment and close follow-up are crucial in preventing visual impairment and blindness
Kernell 1997 [18]	Sweden Population-based 780 persons aged <15 years at diagnosis with a duration of diabetes <12 years from 8 different regions 557 persons were photographed Median age: 14.6 years Duration of diabetes: 8 years	Fundus photographs 3 × 45° fields per eye. Graded using modified Airlie house protocol (KROC study) by 3 ophthalmologists	Weight, height, insulin dose (IU/kg), body weight, number of insulin doses per day	Prevalence of DR: 14.5% 11.8% had BDR, 2.2% PPDR, and 0.2% PDR. Youngest age at which PPDR was found was 11.8 years and PDR 21.5 years. Prevalence of DR by age 8–10 years: 5% 23–25 years: 71% Pubertal stage 1 6% Pubertal stage 5 18%	More advanced stages of DR (40–65) were associated with a longer duration of DM	BDR and PPDR were found in children before puberty. Our recommendation would be to begin screening for DR from the age of 10 years. Although these early retinal changes do not need photocoagulation, it is important to diagnose DR as soon as possible in order to intensify treatment and normalize metabolic control. If screening is performed annually from age 10 instead of biennially from age 15, it could prevent blindness in at least one patient with a cost-benefit of more than £200,000 for the society.
Donaghue 1997 [19]	Diabetes clinics Australia Study A—38 prepubertal persons and 140 pubertal persons with T1DM aged 10–14 years with a duration of diabetes of 3–12 years. Study B—193 persons with T1DM aged 15–22 years with prepubertal-onset diabetes	Fundus photography 7 fields per eye captured using a Topcon camera following dilation. Grading was performed using an adaption of the ETDRS protocol	HbA1c, albumin excretion rate	Prevalence of DR, Study A Prepubertal: 27% Pubertal: 29% Study B Prepubertal 15–22 years 52%	No significant difference for DR between the 2 groups
Kullberg 2002 [20]	Sweden Hospital clinics 440 persons with T1DM aged <36 years at diagnosis 224 persons aged <15 years at onset and 216 persons aged 15–36 years. Fundus photos on 390 patients	Color fundus photography 3 ×  45° 35 mm films per eye following dilation. Images graded by 2 ophthalmologists using the modified Airlie house protocol	Data on HbA1c, blood pressure, antihypertensive treatments, insulin use, urinary albumin concentration, body weight, and height were collected from medical records.	Prevalence of DR: 29% Youngest age of DR: 9 years, Youngest age of DR with more than Ma’s or Hm’s: 18 years. Prevalence with DR by age at onset of DM 0–19 years: 27% 20–35 years: 32.6%		Risk of DR varies with different age at onset, independently of differences in the duration of glycemic control.
Maguire 2005 [21]	Australia Children’s hospital 668 children and adolescents with T1DM. Analysis divided by age at screening 50 persons aged <11 years Median HbA1c: 8.5% 618 persons aged ≥11 years median HbA1c: 8.7%	Fundus photography 7 field stereoscopic following dilation with a Topcon camera. Graded using the ETDRS protocol		Prevalence of DR <11 years: 16% ≥11 years: 22% No PDR and no laser
LeCaire 2006 [22]	Wisconsin, USA Population-based 474 persons with T1DM	Fundus photographs 30° 7 field imaging following dilation Graded using WESDR protocol	Height, weight, waist:hip circumference, blood pressure, and HbA1c	Prevalence of DR Min-severe NPDR: 43% Prevalence of DR by duration of DM 4 years: 6%, 7 years: 23%, 9 years: 47% 14 years: 73%	Risk of DR increased with increasing duration of diabetes and HbA1c levels
Flack 1996 [23]	Finland Population-based 182 children			Prevalence of DR increased from 10.8% to 28% during a follow-up of 2.5 years. DR diagnosed at a mean age of 15.3 years and duration of DM 8.9 years.	Increased HbA1c levels during puberty was the only factor that differentiated these patients from age, sex, puberty, and duration of diabetes matched controls.
Geloneck 2015 [24]	Philadelphia, USA Hospital-based 370 children aged <18 years with T1DM or T2DM Mean age: 11.2 years, duration: 5.2 years	Dilated fundoscopy performed by pediatric ophthalmologists		No DR seen in 338 persons with T1DM and 32 persons with T2DM Mean age at first eye exam: 11.2 years		DR is rare in children regardless of duration and control of DM. Screening examinations for T1DM could begin at the age of 15 years or at 5 years after the diagnosis of DM, whichever occurs later, unless the child is judged by the endocrinologist as being at unusually high risk. We think that the value of screening for such mild disease is questionable, considering the large number of normal examinations being performed. A more reasonable screening target may be identification of sight-threatening DR, which might be close to requiring treatment.
Cahill 2000 [25]	Republic of Ireland Hospital clinic 34 persons with juvenile-onset diabetes of ≥10 years	Fundus photography 7 field following dilation. Graded in a retinal grading center by ETDRS protocols	Age at diagnosis of DM, duration of DM, recent HbA1c, hypertension treatment, and microalbuminuria obtained from medical records.	Prevalence of DR: 44%, all minimal-mod BDR	Those with DR had significantly higher HbA1c levels. No other differences found	Due to coincident lifestyle changes and the probability of long duration of disease, accurate detection of early DR in juvenile-onset diabetes with disease duration of 10 years or more is important
Demirel 2013 [26]	Turkey Pediatric hospital clinic 155 persons with T1DM for 2 years aged ≥11 years Mean age: 14.4 (2.1) years, duration: 6.3 (2.9) years, mean HbA1c: 8.4%	Fundus examination performed by an ophthalmologist following mydriasis. Fundus imaging carried out in suspected cases.	Presence of DR, nephropathy, and peripheral neuropathy, duration of diabetes, HbA1c, BMI, dyslipidemia, total cholesterol, triglycerides, LDL-cholesterol, HDL-cholesterol, microalbuminuria, and hypertension	No DR found		Persons with poor metabolic control, dyslipidemia, and hypertension should be screened 2 years after diagnosis of DM or after 11 years of age. Those with good metabolic control without hypertension or dyslipidemia screening could be postponed to 5 years after a diagnosis of DM.
Dujic 2009 [27]	Serbia Ophthalmology clinic 33 persons with juvenile-onset DM 1992–2007 Mean age: 23.1 (6.4) years, duration: 17.4 (7.4) years.	After dilation direct and indirect ophthalmoscopy performed by ophthalmologists	Age, age at diagnosis of DM, time when eye complications occurred, and treatment modalities.	Prevalence of DR NPDR: 41.4%, PDR: 13.1%, 10.1% had macula involvement 11 of 66 eyes had a vitreous Hm, 2 persons had neovascular glaucoma, and 1 person had rubeosis iridis		Screening for DR should begin at age 10.
Florkowski 2001 [28]	New Zealand Population-based 286 persons with T1DM and age at onset <20 years	Fundoscopy through dilated pupils by an experienced observer		Prevalence of DR was 37.4%	Those with DR were older, and had diabetes for longer, with higher HbA1c levels compared to those without DR.	The association between age at diagnosis and the effects of extended duration of exposure to diabetes suggests that targeting of resources to younger patients is appropriate.
Holl 1998 [29]	Germany Ophthalmology clinic 441 children with T1DM Median age: 15.5 years Duration of diabetes: 6.3 years.	Fundus examination in the ophthalmology department following dilation 2 × 60° fields per eye captured using a Canon 60 degree camera (Canon, Ota, Tokyo, Japan) and Kodak 64 ASA film (Kodak, Rochester, New York, New Jersey, USA)		Prevalence of mild NPDR: 16.3% Moderate NPDR: 0.9% Shortest duration of DM when DR seen was 2.2 years Youngest age when DR seen was 5.5 years		Screening could be conducted biennially in children, provided there are no other risk factors.
Olsen 2004 [30]	Denmark Prospective nation-wide study 339 children and adolescents with T1DM 304 had prepubertal onset diabetes <11.7 years in girls and <12.9 years in boys	Fundus photography with central grading	Demographic data, HbA1c, albumin excretion rate	Prevalence of DR: 57.6% Prevalence of DR in those aged 12–15 years: 17.7%, 16–20 years: 45.4% >20 years: 67.6%	DR was significantly associated with poor long-term metabolic control and duration of DM	Prepubertal diabetes duration contributes to the overall risk of developing microvascular injury but is modest compared to post-pubertal duration.

ETDRS: early treatment of diabetic retinopathy study; DR: diabetic retinopathy; BDR: backrground diabetic retinopathy; PDR: proliferative diabetic retinopathy; PPDR: pre-proliferative diabetic retinopathy; DM: diabetes mellitus; Ma's: microaneurysms; Hm's: haemorrhages; T1DM: type 1 diabetes mellitus; T2DM: type 2 diabetes mellitus; WESDR: Wisconsin epidemiological study of diabetic retinopathy; NPDR: non-proliferative diabetic retinopathy; LDL: low density lipoprotein; HDL: high density liproprotein.

Prevalence of DR in children

The evidence indicates that the prevalence of DR in children is low and is extremely rare prior to puberty[18,31]. The prevalence of DR has been found in children and young persons with diabetes ranges between 10.5% and 57.6% depending on the inclusion age and duration of diabetes at the time of screening and adopted methods of screening for DR (Table 1) [17–30].

The youngest age at which DR has been observed was 5 years, [29] and the youngest age at which STDR was reported was 15 years, with the shortest duration of diabetes being 5 years, [32] with only five cases of STDR in children <18 years[29,32]. However, in a recent study of 370 children with diabetes aged <18 years, no cases of DR were found, [24] prompting the suggestion that screening for DR be delayed to the age of 15 years or once duration of diabetes is 5 years, whichever comes later and also on the basis that few require treatment for STDR at an earlier age. Some clinicians have also suggested that screening for DR at such a young age may theoretically have a psychosocial impact, creating stress and anxiety, which is an area that requires further examination.

In our experience involving 1770 children and young people, aged 12–18 years, with T1DM, undergoing screening for the first time in the Diabetic Retinopathy Screening Service for Wales (DRSSW), the prevalence of DR was 17.4%, consisting of 17.1% non-STDR and 0.3% STDR[33]. Of those with STDR, one person was found to have PPDR, four had maculopathy, and none had PDR. The youngest person with STDR was aged 14 years. About 11.9% presented with non-STDR aged 12–13 years. After adjusting for gender and duration of diabetes, those aged 14–16 years and ≥17 years were 2.8-fold and 2.9-fold, respectively, more likely to develop DR compared to those aged ≤13 years. Diagnosing non-STDR at this early stage allows appropriate enhancement of diabetes management in an attempt to near-normalize metabolic control to avoid any further deterioration of the DR[8].

Recommendations

It is imperative to assess the positive and negative impact of screening for DR on children and young persons with T1DM. In general, screening programs have the potential to generate anxiety. Further investigation should be undertaken to fully evaluate the early introduction of screening in this population. Whilst it is recognized that few type 1 subjects require laser photocoagulation for STDR before mid-teens, the earlier detection of DR affords an opportunity to improve glycemic control and the other putative risk factors, including blood pressure, to prevent progression to STDR. However, what clinical interventions and clinical benefits resulted from the early detection of DR remains unclear. Further debate is now required to review in detail the available evidence, together with input from patients, parents/guardians, pediatric diabetologists, and ophthalmologists to reach a consensus as to what age DR screening should commence in children and young persons with T1DM. Delaying the introduction of screening beyond the age of 12 years cannot at present be justified without the necessary evidence, which will be the subject of future research.

Financial and competing interests disclosure

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.