Objective To assess the risk of severe retinopathy of prematurity (ROP) requiring treatment in different ethnic groups.
Design Retrospective observational study on ROP screening and treatment. It involved a cross-sectional review of all eligible infants over a seven-year period. Statistical tests used were the Kruskal-Wallis test and Mann-Whitney U test. Logistic regression was used to control for any differences in birth weight and gestational age.
Setting City Hospital and Birmingham Women's Hospital, Birmingham, UK.
Results 1690 preterm infants underwent ROP screening. Birth weight was lower in black (1142.5 g) and Asian infants (1180 g) when compared to white infants (1196.5 g). Gestational age was lower in black infants (28.5 weeks) compared to Asian and white infants (both 29 weeks). Compared to white infants, the odds of severe ROP requiring treatment was higher in Asian infants (odds ratio (OR): 2.52; 95% CI 1.41 to 4.50) and black infants (OR: 2.51; 95% CI 1.30 to 4.86). The additional risk from ethnicity was present even after adjusting for birth weight and gestational age (adjusted OR for Asian vs white infants: 2.45; 95% CI 1.334 to 4.497); (adjusted OR for black vs white infants: 2.0; 95% CI 1.004 to 4.014).
Conclusions Ethnicity is a risk factor for severe ROP. Asian and black infants have a higher risk of developing threshold ROP compared to white infants.
Statistics from Altmetric.com
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.
Retinopathy of prematurity (ROP) is a major cause of childhood visual impairment.1 The World Health Organization's “Vision 2020 programme” has identified ROP as an important cause of blindness in both high-income and middle-income countries.2 Low birth weight and gestational age are known risk factors for ROP.1 3 Other factors like illness severity and supplemental oxygen therapy have been associated with severe ROP.1 4 It has also been suggested that preterm infants with better survival rates are more likely to develop severe ROP.5
The West Midlands region in the UK has a large cosmopolitan population comprising different ethnic groups. We observed that Asian and black infants were more likely to develop threshold ROP. The aim of this study was to test this hypothesis on a large cohort of preterm infants. We sought to determine if there was a difference in the incidence of ROP requiring treatment among different ethnic groups.
This was a retrospective observational study. We undertook a cross-sectional review of seven years of ROP screening and treatment from 1 January 1999 to 31 December 2005. It involved all eligible infants from two neonatal units in Birmingham, the Women's Hospital and City Hospital. The Women's Hospital has a level 3 neonatal unit while City Hospital has a level 2 unit. They work together with transfer of the more sick cases to the Women's Hospital. The two units together account for the majority of ROP screening and treatment in Birmingham, and their catchment area is representative of the local ethnic population. All infants were under the care of one consultant (LB) with an interest in ROP. The data were collected from the computerised hospital database as well as the admission records of these neonatal units. Data on ROP treatment were collected as a part of an ongoing audit.
What is already known on this topic
Low birth weight and gestational age are risk factors for ROP.
What this study adds
Ethnicity is also a risk factor for severe ROP.
Asian and black infants have a higher risk of threshold ROP.
The screening and treatment of ROP was based on the UK guidelines by the Royal College of Ophthalmologists (RCO) and the British Association of Perinatal Medicine.6 Accordingly, infants with birth weights less than 1501 g, and/or gestational age less than 32 weeks were screened. After dilating the pupils, retinal examination was performed using an indirect ophthalmoscope and a 28-dioptre lens. The stage of ROP was defined according to the international classification.7 The threshold for treatment was five or more contiguous clock hours or eight or more cumulative clock hours of stage three ROP in the presence of “plus” disease. In 2003, the threshold for treatment was modified following the recommendations of the “Early Treatment for ROP Cooperative Group”.8 The modified threshold was any ROP with plus disease in zone 1, stage 3 ROP without plus disease in zone 1 and stage 3 ROP with plus disease in zone 2. This modification to treatment was uniformly applied to all infants. Hence it is unlikely that it would have led to any bias or differences in treatment to any particular group. All treatments were carried out using diode laser through a portable indirect ophthalmoscope.
The Kruskal-Wallis test and Mann-Whitney U test, as appropriate, were used to see whether there was any difference between the three ethnic groups in terms of birth weight and gestational age. Logistic regression analysis was used to control for any differences in birth weight and gestational age. The analyses were performed using the SPSS software (v 14). A probability level of <0.05 was considered significant.
During the study period, there were 5450 admissions to the two neonatal units. Of these, 1690 infants were screened for ROP. The majority of the screened cases were white infants (table 1). The Asian group composed of infants of South Asian origin (India, Pakistan and Bangladesh). Among the black group, the majority were Caribbean. Infants classified in the ‘other’ group were mostly of mixed ethnicity. In 2% infants, the ethnicity was not declared or not known. This group was too small to make meaningful comparison with the rest of the study group. Since the white, Asian and black groups together comprised 94% of the screened cases, the detailed statistical analyses were confined to these groups. Birth weight and gestational age did not follow normal distribution. Hence, non-parametric tests were used to compare the ethnic groups.
The birth weight (table 2) showed a significant difference between the three ethnic groups (Kruskal-Wallis χ2=11.017, df=2, p=0.004). Compared to white infants, the median birth weight was significantly lower in black infants (95% CI 28.00 to 130.00; Mann-Whitney U test, p=0.002) and Asian infants (95% CI 2.00 to 87.00; Mann-Whitney U test, p=0.037).
The frequency distribution of birth weight based on ethnicity is shown in table 3. Although black infants were slightly over-represented in the birthweight category less than 750 g, the combined distribution in the categories less than 1000 g was similar for the three ethnic groups.
A comparison of the gestational age (table 2) showed a significant difference between the three ethnic groups (Kruskal-Wallis χ2=7.511, df=2, p=0.023). The gestational age was significantly lower in black infants when compared to Asian infants (Mann-Whitney U test, p=0.011) and white infants (Mann-Whitney U test, p=0.022).
The number of infants who developed severe ROP needing treatment is summarised in table 1. There were no infants who did not receive treatment because of refusal or poor health. The odds of treatment in Asian infants were two and a half times higher than the white infants (odds ratio (OR): 2.52; 95% CI 1.41 to 4.50). Similarly the odds of treatment in black infants were also two and a half times higher than the white infants (OR: 2.51; 95% CI 1.30 to 4.86).
After adjusting for the variation in birth weight and gestational age, through logistic regression, the Asian and black infants still had a higher risk when compared to the white infants (Asian vs white, OR: 2.45; 95% CI 1.334 to 4.497; p<0.005), (black vs white, OR: 2.0; 95% CI 1.004 to 4.014; p<0.05). As the lower limit of the 95% CI of OR for black vs white infants is close to 1, it is possible that the difference had a relatively minor effect in clinical terms.
A further subgroup analysis for comparing the ethnic groups by different birthweight categories was performed, but this did not yield meaningful results because of small numbers in some subgroup categories.
We looked at the proportion of infants who were small for gestational age (SGA), and found a higher rate among Asian (36.19%) than white (29.45%) or black infants (27.81%).
For all infants who were eligible for ROP screening, we determined the proportion that failed to survive while in hospital. There was no significant difference between the groups, with survival to discharge being 88.56% (95% CI 86.31% to 90.49%) for white infants, 90.05% (95% CI 86.82% to 92.55%) for Asian infants and 89.51% (95% CI 89.51% to 92.64%) for black infants.
Following laser photocoagulation for threshold ROP, five infants needed top-up treatment. All treated cases had successful outcome with complete regression of ROP. No infant developed adverse retinal outcome (progression to stage 4/5 ROP). Hence in terms of final structural outcome, there was no difference between the different ethnic groups.
The results of this study suggest that there are ethnic variations in progression to threshold ROP. Asian and black infants are more likely to need treatment than white infants.
There have been very few studies on ethnicity and ROP.5 9,–,11 We believe that this is the largest study on Asian infants. It is also the largest study on ethnicity and ROP from the UK. In our literature search, we found two previous studies from the UK. Sabri et al did not find any statistical evidence for a difference in the incidence of severe ROP among infants of South Asian ethnic origin compared with those of white origin.9 Their study was based on 355 neonates, of which 10 developed threshold ROP. Ng et al reported on the epidemiology and ethnic variation of ROP in the East Midlands, and found that severe ROP was more likely in Asian infants (14.1%) than in white infants (2.7%).5 This difference was thought to be because of better survival of Asian infants. Although this is similar to our findings, the results cannot be directly compared since the screening criteria and infant survival rates at that time were different from our study period. Both these studies had smaller numbers compared to our study.
American studies have reported a higher rate of threshold ROP among white than African-American infants. One study showed that the incidence among white infants (7.4%) was twice that of black infants (3.2%).10 Their criteria involved screening infants less than 1251 g, which is different from the RCO guidelines in the UK. In addition, the difference in the demographics of their population, the screening protocol and infant survival 20 years back makes direct comparison with our study difficult.
In a more recent study, Lang et al11 reported an increased risk of threshold ROP in a darkly pigmented race, the Alaskan natives. They found that compared to the lightly pigmented non-natives, threshold ROP was more common in the Alaskan natives and Asian Pacific races.
There are a few possible explanations for the ethnic variations in ROP progression found in our study. Asian and black infants were noted to have better survival rates in some studies.5 12 Hence, preterm infants in these ethnic groups are more likely to survive and hence develop severe ROP. Although we found a similar pattern, the difference in the survival rates was not significant. Some studies have shown that black infants are more likely to be SGA.10 13 14 We found a higher rate of SGA among Asian infants, which might be responsible for the increased rate of severe ROP in this group. It must be noted that our estimates of SGA are based on growth charts that do not take ethnicity into account.
The differential risk may be related to pigmentation. Animal models of ROP have shown that pigmented strains of rats are more susceptible to ischaemia-induced retinal neovascularisation than the albino counterparts.15 Recently there has been an increased interest in the role of genetic factors in ROP.16 It is possible that genetic variations among different ethnic groups could influence ROP progression.
There are limitations to our study. Since it was a retrospective study, we were unable to collect other data such as details of oxygen therapy and severity of systemic illness. It is therefore not possible to assess the impact of these factors on progression of ROP. Subgroup analysis for different birthweight categories did not yield meaningful results in our study. A multicentre trial could help achieve larger numbers required for such detailed analysis.
Our study shows a differential risk of progression to threshold ROP among ethnic groups. We hope that further research and better understanding of risk factors and genetic markers in ethnic groups would help improve our understanding of racial variation in ROP progression.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.