Arch Dis Child 96:963-968 doi:10.1136/adc.2011.210807
  • Original articles

Prevalence of retinopathy in Finnish children and adolescents with type 1 diabetes: a cross-sectional population-based retrospective study

  1. Aura Falck1
  1. 1Department of Ophthalmology, Oulu University Hospital, Oulu, Finland
  2. 2Department of Paediatrics, Oulu University Hospital, Oulu, Finland
  1. Correspondence to Aura Falck, Department of Ophthalmology, Oulu University Hospital, PO Box 21, Oulu 90029 OYS, Finland; aura.falck{at}
  • Accepted 30 May 2011
  • Published Online First 29 June 2011


Aim A population-based study was carried out to evaluate the prevalence and risk factors of diabetic retinopathy (DR) in children with type 1 diabetes (T1D) in The Northern Osthrobothnia Hospital District, Finland. The aim was to compare the current prevalence and the risk factors with those obtained in a study performed in a similar setting 17 years earlier.

Methods and patients The prevalence of DR was evaluated from fundus photographs in a cross-sectional manner in children and adolescents with T1D (n=297) living in the Northern Osthrobothnia Hospital District on 1 January 2007.

Results The prevalence of DR was 7.6% (12/158) in males and 16.5% (23/139) in females in the present study and 7.3% in males and 12.9% in females in the former study. The mean age of the patients was 11.9 and 11.8 years, and the mean duration of diabetes was 4.9 and 5.0 years in the present and the former study, respectively. DR was associated with older age (p<0.001), longer duration of diabetes (p<0.001), higher glycated haemoglobin A1c (GHbA1c) (9.3% in those with DR vs 8.3% in those without DR, p=0.001, or 78 vs 67 mmol/mol, respectively) and female sex (p=0.016); in a logistic regression analysis, these factors explained 35% of DR. These risk factors are essentially the same as identified in the cohort 17 years earlier. GHbA1c levels had not significantly improved during that time.

Conclusions The overall prevalence of DR among children with T1D was 11.8% (35/297) showing no decrease over the past 17 years; in girls, DR was diagnosed more often in the present than in the former study, but there was no change in the prevalence among the boys. Glycaemic control had remained unchanged.


The incidence of childhood type 1 diabetes (T1D) in Finland has increased fourfold since the 1950s, being 64/100 000/year in children under 15 years of age in 2005,1 and thus it is higher in Finland than in any other country in the world.1 The incidence has increased most rapidly among the children under 10 years of age, and this trend is continuing, but the reason for this is unknown.1

Diabetic retinopathy (DR) is a common microvascular complication of diabetes.2 3 It may lead to impaired vision or even permanent blindness; DR is the cause of 10% of the cases of impaired vision among working-age people in Finland.4

Hyperglycaemia, long duration of diabetes, onset of puberty and pregnancy are known risk factors for the onset and progression of DR.2 3 5 DR is rare in children before puberty,6 7 and as far as we are aware, no cases of vision-threatening DR have been reported in prepubertal children. However, the onset of DR during childhood and adolescence is not rare6 7 and may precede vision-threatening DR during early adulthood.

What is already known on this topic

  • The onset of diabetic retinopathy (DR) during childhood and adolescence is not rare, but vision-threatening retinopathy seldom develops before early adulthood.

  • Hyperglycaemia, long duration of diabetes and onset of puberty are risk factors for the onset and progression of DR in children and adolescents.

What this study adds

  • The prevalence of retinopathy remained unchanged in type 1 diabetes children from the same region in two cohorts 17 years apart, with no improvement in metabolic control.

  • The population-based prevalence of DR has increased in girls in 17 years, and is currently twice as high as the prevalence in boys.

The prevalence of DR among children living in The Northern Osthrobothnia Hospital District was evaluated in 1989–1990 in a population-based study. Among children aged 4.6–16.9 years, 21/216 (9.7%) had DR in fundus photographs, and 22/216 did not attend the examination.6

The possibilities of good glycaemic control of T1D have improved since the year 1990. Human insulin, the insulin analogues, insulin pumps as well as continuous monitoring of blood glucose have been introduced. The health-awareness and education levels of the population including families with a child with T1D have improved, and knowledge about dietary matters has increased.

The present study was carried out to evaluate the current prevalence of childhood DR in The Northern Osthrobothnia Hospital District. The aim of the study was to compare today's prevalence and the risk factors for DR with those obtained in the earlier study.

Patients and methods

This population-based study comprised 297 children and adolescents with T1D living in the Northern Osthrobothnia Hospital District on 1 January 2007. The patients were born during the years 1990–2001. The clinical care of all these patients was provided by the Paediatric diabetes clinic of the Oulu University Hospital, Oulu, Finland. The prevalence of DR was evaluated in 2006–2007 with fundus photographs in a cross-sectional manner.

Since 1992, the Paediatric diabetes clinic at the Oulu University Hospital has referred all children aged 10 years or older annually for fundus photography to the Department of Ophthalmology to detect DR. Children younger than 10 years of age are evaluated with fundus photography after 5 years' duration of diabetes. During the present study, young children with T1D were encouraged to attend fundus photography earlier than in the routine practice in order to obtain a more complete cross-sectional analysis of the prevalence of DR.

A total of 251/297 (85%) of the children from the clinic underwent fundus photography during 2006–2007. For the children who had been photographed twice during the study period, the photographs taken closer to 1 January 2007 were included in this study. Thirty-four of the young children and/or those with a very short duration of diabetes at the beginning of 2007 had their first fundus photography in 2008. Seven children had undergone fundus photography in 2005, and subsequent photographs were taken in 2008. Five children had not attended photography at all by the end of 2008.

The pupils were dilated by cyclopentolate 10 mg/ml eye-drops prior to photography. The photographs were taken with a Canon CF 60 UVI camera with and without a green filter producing black-and-white and colour pictures. Sixty-degree images centred in the macula were taken. The digital images were viewed on a computer screen using the Neagen (Neagen Oy, Oulu, Finland) program.

The children had clear media, and the quality of the images was excellent. The images were viewed independently by two of the authors (MK, AF), and the findings were classified using a modification of the Early Treatment Diabetic Retinopathy Study classification.8 The lesion had to be seen in both black-and-white and colour images in order to be classified as DR. The viewers agreed on the diagnosis in all but one instance. In that case, the images were viewed again together to come to the final consensus on classification for that patient.

Each patient's age and duration of diabetes at the date of fundus photography were recorded. Sex, stage of puberty using Tanner classification9 10 (prepubertal/pubertal/postpubertal), relative height for age (height (standard deviation score) SDS) and relative weight for height based on the Finnish children's growth charts,11 blood pressure (BP, mm Hg) and glycated haemoglobin A1c (GHbA1c) (ADVIA 2400 Chemistry System; Siemens Healthcare Diagnostics, Deerfield, IL, USA) as well as urinary albumin content (normal range <25 mg/l) and blood lipids (high-density lipoprotein and low-density lipoprotein) were records from the closest visit to the paediatrician. The results are given separately for children <11 years of age and ≥11 years of age. Since pubertal staging was performed systematically for 240 out of 297 patients during their paediatric outpatient visits, some of the results are shown based on both age and pubertal stage (prepubertal vs pubertal). The age of 11 years was chosen as an approximation for the onset of puberty based on some previous studies in the field of T1D.12 13

The amount of hyperglycaemia in the two cohorts could not be compared directly, as the method of monitoring the glycated haemoglobin was changed from GHbA1 to GHbA1c in 1992. Linear regression revealed the formula GHbA1c=1.581+0.588×GHbA1, which is used for conversion of the GHbA1 values to the equivalent GHbA1c values. GHbA1c are shown both in the Diabetes Control and Complications Trial (DCCT) (%) and the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) (mmol/mol) units.14

The study was approved by the Ethics Committee of the Oulu University Hospital, Oulu, Finland.

Statistical methods

The mean and SD are given for continuous variables. A χ2 test and independent-samples t test were used for statistical comparison of groups when appropriate. Logistic regression was used for determining risk factors of DR, and the OR and CI are provided. A p value of <0.05 was considered statistically significant.


A total of 251 of 297 (85%) children or adolescents with T1D, 134 boys and 117 girls, underwent fundus photography in 2006–2007. Thirty-three children had DR in one or both fundi, while 218 did not exhibit any signs of DR. DR was classified as mild in all cases: one or a few microaneurysm(s) (eight cases) or/and small haemorrhage(s) (25 cases). The changes were unilateral in 23/33 (70%) cases.

The first fundus photography did not take place until the year 2008 in 34 cases (18 boys, 16 girls) including many of the youngest children or those with a very short duration of T1D. These patients were 8.8±2.2 years old when first photographed after 3.9±1.8 years of diabetes duration. None of these children showed any signs of DR in their first fundus photography in 2008.

Seven subjects had undergone fundus photography in 2005 and had the next photographs taken in 2008. Five of them showed no DR in either examination, while one girl displayed DR both times, and another girl exhibited DR in the second but not in the first examination. It seems likely that those individuals showing no DR in either examination did not have DR at the beginning of 2007. The patient having DR in both photographs was assumed to have had DR at the beginning of 2007 as well. The patient who developed the first DR changes between 2005 and 2008 was assumed to have had DR at the beginning of 2007.

In the cross-sectional study, the lesions of DR were found more often and at an earlier age in girls than in boys. The girls had also been younger than the boys when DR was first diagnosed, as shown in a Kaplan–Meier curve (figure 1).

Figure 1

Kaplan–Meier curve illustrating the age at onset of diabetic retinopathy (DR) in boys (1) and girls (2). Cum, cumulative.

The youngest patient with DR was an 8-year-old girl who had had diabetes since 15 months of age. She had a unilateral haemorrhage in her fundus photographs. Her GHbA1c values had varied between 7.9% and 11.9% (63–107 mmol/mol), and the mean GHbA1c (of 12 measurements) was 9.5% (80 mmol/mol). The next youngest patients with DR were three girls and one boy aged 10.5–10.9 years. Each of them displayed unilateral retinal changes. In this group of patients, the duration of T1D was 4.2–8.9 years, and the latest GHbA1c values were 8.5–11.0% (69–97 mmol/mol). The shortest duration of T1D before the detection of DR was 1.7 years in a postpubertal 16-year-old boy with a unilateral haemorrhage, the most recent GHbA1c 7.5% (58 mmol/mol) and mean value 8.0% (64 mmol/mol) during the course of T1D.

DR was associated with older age (p<0.001), longer duration of diabetes (p<0.001), elevated mean GHbA1c value during the years of T1D (p<0.001), latest GHbA1c value (p=0.001), onset of puberty (p<0.001) and female sex (p=0.016), but not with albuminuria (p=0.584), height SDS (p=0.980) or relative weight for height (p=0.166).

The mean BP was higher in the pubertal patients with DR than those without DR, but the systolic or diastolic BP did not statistically significantly associate with DR. The pubertal boys had a significantly higher systolic BP than the girls, 126±13 versus 120±12 mm Hg, respectively, but the mean BP or diastolic BP did not differ (p=0.559 and p=0.868, respectively). The association of various factors to DR in children under 11 years of age, children and adolescents 11 years of age or older, and the pubertal children is summarised in table 1.

Table 1

Association of various factors to diabetic retinopathy in children aged <11 years and ≥11 years, and in children with onset of puberty

In the binary logistic forward stepwise regression analysis, the factors explaining DR were age (OR 1.2, 95% CI 1.0 to 1.4), duration of T1D (OR 1.3, 95% CI 1.1 to 1.5), mean GHbA1c of the years with T1D (OR 2.1, 95% CI 1.5 to 3.1) and gender (OR 2.3, 95% CI 1.0 to 5.4). The model explained 35% of DR.

Comparison with the 1989–1990 study

The number of diabetic children and adolescents living in The Northern Osthrobothnia Hospital District had increased by 38% between the years 1989/90 and the beginning of 2007. The age groups of interest consisted of 297 patients in the latter and 216 subjects in the former cohort. The proportion of boys in the new cohort was 53% (158/297), and in the old cohort it was 57% (123/216). The demographics of the two cohorts are presented in table 2.

Table 2

Demographics and glycaemic control of the patients with type 1 diabetes (T1D) with or without retinopathy in the 1989–1990 cohort and in the 2007 cohort

DR lesion(s) were observed in 33 cases among the patients having photography in 2006–2007, and in 35 including the information of those having photography in 2008 and extrapolating back to the likely situation at the start of 2007. The prevalence per 297 was thus 11.1% at the minimum, and most likely 11.8%. Five young children did not undergo fundus photography. In the earlier study, the prevalence of DR was 9.7% (21/216) with information on 22 subjects missing.

The mean age of the subjects was similar in the two cross-sectional studies, and the duration of diabetes was about the same. The presence of DR associated with older age, longer duration of diabetes, onset of puberty and the GHbA1c values. DR was clearly more likely to be present in females than in males in the new study, and the trend was similar in the former study. Glycaemic control was not more favourable in boys.

In the present study, the latest GHbA1c values were 8.8±1.4% (73±12 mmol/mol) for those without DR, and interestingly 8.8±1.0% (73±8 mmol/mol) also for those (N=9) with microaneurysm(s) only, but 10.3±0.4% (89±3 mmol/mol) for those with one or more haemorrhage(s) with or without microaneurysm(s) (p<0.001). The GHbA1c equivalent in the former study indicates that glycaemic control of patients was similar to that of the patients in the recent study.

Smoking habits were not obtained from all the subjects. Thus, 14/23 girls with DR did not smoke, and 1/23 did smoke; 1/12 boys did smoke, while 5/12 did not. Smoking habits were not recorded from the older cohort at all.

All cases of DR were mild. In the older cohort, nine of 21 (43%) had changes in one eye only, while 25/35 (71%) had unilateral changes in the new cohort.


The prevalence of T1D has increased all over the world and especially in Scandinavian countries during the past decades.1 15 This increase is reflected in the present study by a 38% increase in number of children and adolescents with T1D of defined age and living area. The population living in The Northern Osthrobothnia Hospital district has increased by 11% (from 343 301 in 1990 to 383 708 in 2006)16 at the same time. Thus, the increase in the number of children with T1D cannot be completely explained by the increased population in the region.

Some recent studies have detected a decreasing trend in prevalence of DR in children or young adults with T1D17 18 compared with previous reports.19 In Denmark, a significant improvement in the GHbA1c (0.08% per year) was reported among children with T1D between 1996 and 2006; this was associated with increased self-monitoring of blood glucose, while association with prevalence of DR was not studied.20 A Swedish survey showed a decline in the prevalence of severe DR over an interval of 25–30 years, owing mostly to modern diabetes care.21 In an Australian study, a decline of the prevalence of early DR and simultaneous decrease in the GHbA1c level in adolescents over time from 1990 to 2006 was observed.22

In the present study, it is shown that in Northern Finland, the prevalence of DR remained unchanged, and the metabolic control of the children with T1D living in the same defined area did not improve markedly between 1990 and 2007: DR was present in 9.7% at the earlier time point and in 11.8% at the later time point. These results emphasise again the crucial role of good glycaemic control in preventing and delaying complications.23

The risk factors for DR remained the same in the analysis of the two cohorts, and one of the most important was female sex. Interestingly, in both cohorts the girls appeared to have better glycaemic control than the boys but a higher risk for DR. The trend is towards an increased prevalence of DR in girls between 1990 and 2007. Given that the metabolic control was not better in boys, other factors than glycaemic control alone may be important. Earlier pubertal growth with enhanced growth hormone during puberty may significantly reduce insulin sensitivity and change insulin metabolism as shown in previous studies.24 These well-characterised metabolic and hormonal changes may predispose girls to earlier manifestations of the microvascular complications.

A higher incidence of retinopathy in girls has been reported by others also,25 but this finding is not universal. There are studies in which DR is equally common in boys and girls with T1D,26 and those showing male sex to be a risk factor for developing retinopathy27 or nephropathy.28

In this study, a comparison of the prevalence of DR in two carefully defined and characterised young cohorts with T1D from the Northern Osthrobothnia Hospital district in Finland has been presented.

There are, however, limitations in the present study. First, some of the data such as pubertal stage and BP were collected retrospectively, and data needed for the analysis of these risk factors were not available in the patient notes in all cases. This is also true for some other predisposing factors such as smoking. Smoking is unlikely to be a risk factor given to the cohort age, but it has a clear adverse effect on blood vessels. We also noticed that there was a lack of data on children's smoking habits in the previous cohort, but evaluation of this risk factor had been introduced during the clinical visits of the more recent cohort, which we find is clinically a very important change. It is also worth mentioning that some detailed information on individual diabetes care such as current insulin dosing and current insulin regime was not available. Second, pubertal staging was available in 80% of the study population. We chose the age of 11 for the landmark of pubertal onset; out of these 240 children, 10 were pubertal under 11 years of age, and 11 over 11 years of age were prepubertal. Although the classification is not ideal, we find that it gives a relatively good approximation of the timing of puberty in this cohort. We also analysed the data based on the pubertal staging, and this did not change the results in the more recent cohort, as shown in table 1. There is, however, some evidence that the prevalence of vascular complications may be underestimated or overestimated, owing to definitions of puberty based on age versus pubertal staging.29 Third, the study sample size was not calculated for the detection of the possible differences between the cohorts. This was due to the fact that both the previous and the present data are population-based, and all patients of defined age with T1D from the same clinical centre are included.

In the study performed in Oulu in 1989–1990, the 60° black-and-white and colour fundus photographs were reproduced as paper prints and colour slides. The resolution in these was better than that in the digital images taken at the time of the present study, even though the resolution and quality of these were still very good. The slightly higher prevalence of DR in the present study cannot be explained as a result of improved quality of images leading to better diagnostics.

Even though all cases of DR diagnosed in the paediatric population of this study were mild and did not require treatment, the onset of DR during childhood and adolescence is concerning. The population-based prevalence had not diminished during the past 17 years, even though the possibilities to treat diabetes as well as the awareness of the importance of good glycaemic control have improved. In girls, the prevalence of DR has increased for unclear reasons. Although there have been changes towards more intensified insulin regime as multiple insulin dosing with insulin analogues and modern insulin pumps have been introduced, we could not see any difference in the glycaemic control between these two cohorts. It is most likely that there are many other factors that influence the individual level and are beyond this survey.

The characteristics of patients developing rapidly progressive florid DR were recently discussed in an editorial by Agardh. The typical features include young age, poor metabolic control and early signs of rapid progression of DR. One fourth of people with T1D receiving treatment for proliferative DR are estimated to have such high-risk DR.30 Pinpointing these individuals for intensified treatment and close follow-up is crucial in preventing visual impairment and blindness. Whether signs of DR in childhood act as a risk marker of the development of florid DR in early adulthood is not quite clear and can only be determined in a well-designed follow-up study which still needs to be conducted. It may be that those paediatric patients having high levels of hyperglycaemia and showing signs of appearing DR are the patients who are at increased risk. A reliable risk profile of the people with T1D most likely to become visually impaired would be a useful tool in order to achieve correct prioritising among patients with DR.

In conclusion, no significant differences or critical improvement were found between the previous and the recent cohort in the prevalence of DR or the glycaemic control. For a clinician, this finding of ‘no change’ emphasises the importance of the continuous quality control in diabetes care.


  • Funding Oulu University Hospital.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval Ethics approval was provided by the Oulu University Hospital.

  • Provenance and peer review Not commissioned; externally peer reviewed.


Free sample

This recent issue is free to all users to allow everyone the opportunity to see the full scope and typical content of ADC.
View free sample issue >>

Don't forget to sign up for content alerts so you keep up to date with all the articles as they are published.