May 2005, VOLUME115 /ISSUE 5

Characteristics of Infants With Severe Retinopathy of Prematurity in Countries With Low, Moderate, and High Levels of Development: Implications for Screening Programs

  1. Clare Gilbert, MD*,
  2. Alistair Fielder, FRCOphth,
  3. Luz Gordillo, MD§,
  4. Graham Quinn, MD, MSCE,
  5. Renato Semiglia, MD,
  6. Patricia Visintin, MD#,
  7. Andrea Zin, MD**,
  8. on behalf of the International NO-ROP Group
  1. *London School of Hygiene and Tropical Medicine, London, United Kingdom
  2. Imperial College London, London, United Kingdom
  3. §Hospital Nacional E. Rebagliati, Lima, Peru
  4. Children's Hospital of Philadelphia, Philadelphia, Pennsylvania
  5. Hospital Enrique Sotomayor, Guayaquil, Ecuador
  6. #Hospital Oftalmológico Santa Lucía, Buenos Aires, Argentina
  7. **Instituto Fernandes Figueira Fiocruz, Rio de Janeiro, Brazil


Objective. Retinopathy of prematurity (ROP) is a potentially avoidable cause of blindness in children. The proportion of blindness as a result of ROP varies greatly among countries depending on their level of development, being influenced by the availability of neonatal care, neonatal outcomes, and whether effective screening and treatment programs are in place. The objective of this study was to compare characteristics of premature infants who developed severe ROP between 1996 and 2002 in highly developed countries with less developed countries.

Methods. This was an observational study. A questionnaire was completed by ophthalmologists in countries with low, moderate, and high development rankings (3 highly developed countries and from 10 less well-developed countries) who screen for ROP in which they supplied birth weights and gestational ages (GAs) of infants who were treated for threshold ROP or identified with more advanced stages of the disease. Birth weights and GAs of infants with severe ROP were measured.

Results. The mean birth weights of infants from highly developed countries ranged from 737 to 763 g compared with values ranging from 903 to 1527 g in less developed countries. Mean GAs of infants from highly developed countries ranged from 25.3 to 25.6 weeks compared with 26.3 to 33.5 weeks in less developed countries. A total of 13.0% of 1091 infants from poorly developed countries exceeded United Kingdom screening criteria; 3.6% exceeded a criteria of <34 weeks’ GA and/or <1750 g birth weight.

Conclusions. These findings suggest that larger, more mature infants are developing severe ROP in countries with low/moderate levels of development compared with highly developed countries. ROP screening programs need to use criteria that are appropriate for their local population.

  • retinopathy of prematurity
  • screening
  • development index

Retinopathy of prematurity (ROP) is a potentially avoidable cause of blindness in children. The proportion of blindness as a result of ROP varies greatly among countries (Table 1), 1,2 being influenced both by levels of neonatal care (in terms of availability, access, and neonatal outcomes) and by the availability of effective screening and treatment programs. This raises important questions concerning strategies to reduce the incidence of blindness as a result of ROP, which should include ensuring that all infants who are at risk are examined in screening programs.


Proportion of Childhood Blindness as a Result of ROP in Countries Ranked by Their UNDP HDI

In highly developed, industrialized countries (ie, those ranked highly by the United Nations Development Programme [UNDP] on the basis of their Human Development Index [HDI]), the population of premature infants who are currently at risk for the advanced stages of ROP that requires treatment is extremely premature, with birth weights almost always <1000 g.3,4 However, this has not always been the case: during the first epidemic of ROP, in the 1940s and 1950s, larger, more mature infants became blind from retrolental fibroplasia (the term used earlier to describe the condition). At that time, the mean birth weight of affected infants in the United Kingdom was 1370 g (range: 936–1843 g) and in the United States was 1354 g (range: 770–3421 g).5 More premature infants generally did not survive. The population of infants who are at risk for blinding ROP, therefore, has changed over time in highly developed countries. These changes can be attributed to a better understanding of the risk factors and pathogenesis of ROP, leading to improvements in neonatal care. More conservative use of supplemental oxygen, meticulous monitoring of blood oxygen levels, and aggressive management of instability of the infant are probably the most important factors that are responsible for the lower risk in more mature infants. In many countries of Latin America and the Former Socialist Economies of Eastern Europe (ie, those that are moderately developed, with HDI rankings in the range 31–100), ROP is emerging as a major cause of blindness (Table 1). This has been referred to as the “third epidemic.”1 Indeed, it is thought that two thirds of the 50000 children who are blind from ROP worldwide live in Latin America.6,7 There are several possible reasons for this third epidemic: first, birth rates and rates of premature birth are higher8,9; second, neonatal care may be compromised as a result of lack of resources, leading to higher rates of severe ROP not only in extremely premature infants but also in larger, more mature infants; and third, because of lack of awareness, skilled personnel, and because of financial constraints, screening and treatment programs are not in place in all neonatal units in many cities. Indeed, in Lima, Peru, a city with 8 million residents, 1 of the authors (L.G.) single-handedly provides the ROP service for the whole city, and in Rio de Janeiro, Brazil, another author (A.Z.) examines infants in 7 of the largest units in the city that cater to two thirds of all premature births. In poorly developed countries (ie, those with UNDP rankings <100, which includes most of sub-Saharan Africa and much of Asia), blindness from ROP is virtually unknown. South Africa is the exception, where ROP accounts for 11% of blindness in children.10 In the majority of other sub-Saharan African countries, services for the care of premature infants are not well developed, and preterm infants do not survive long enough to develop severe ROP.11 Accurate delineation of the population of infants who are at risk for potentially blinding ROP is essential, providing the evidence on which to base guidelines regarding which infants need to be examined. The United Kingdom, the United States, and Canada,4,1216 along with several other countries, have developed evidence-based screening criteria, which continue to be reviewed as the population of infants who are at risk changes over time. Information on the population of infants who develop treatable ROP is required from less developed countries, to develop screening programs that include all premature infants who are at risk. The purpose of this study was to describe and compare the birth weight and gestational age (GA) of infants who develop severe ROP in countries with a range of development indices to determine whether screening criteria used in highly developed countries would be applicable worldwide. We did not attempt to describe rates of severe ROP or to explore other risk factors for ROP, particularly those in relation to neonatal care. In this study, we use the term “severe ROP” to encompass infants who are treated for threshold stage 3 disease as well as those with stages 4 and 5 ROP, or cicatricial disease.


For the purposes of this study, countries were grouped into 3 categories according to their rank for the year 2002, based on their UNDP HDI.2 “Highly developed” countries are those ranked among the top 30 most developed countries in the world, where ROP accounts for 3% to 11% of blindness in children; countries that are ranked 31 to 100 are termed “moderately developed,” and in these countries, ROP accounts for up to 60% of blindness; “poorly developed countries” are those that are ranked <100 and where blindness in children as a result of ROP is not currently a major problem.

Information on the population of infants with severe ROP was obtained via several different approaches. The following individuals were contacted: ophthalmologists who screen for ROP in Latin America and who had presented their data at workshops held annually in the region since 1997 (from Argentina, Brazil, Colombia, Ecuador, Mexico, and Peru); authors of papers describing large case series of infants who were screened for ROP in Canada14 or large series of infants with stage 5 disease (from India17); individuals who have undertaken research on ROP in the United Kingdom and the United States; an ophthalmologist who runs a low vision clinic for children in Argentina; other ophthalmologists who are known to the authors and screen for ROP (in Lithuania, Vietnam, and India); the ophthalmologist responsible for the national ROP screening program in Chile; and neonatologists who manage infants who are referred to a national ROP treatment center in Argentina. All were sent a questionnaire, in which they were asked to indicate the following: (1) the city, (2) the level of neonatal care provided, (3) the health sector providing the service (eg, government, private, social security), (4) and the screening criteria used (if relevant). They were also sent a spreadsheet in which they were asked to supply the birth weight and GA of individual infants who were treated for threshold ROP, who had presented too late for treatment, who had been treated for stage 5 ROP, or who were being treated in the low vision clinic. They were asked to supply data for infants who were seen during the preceding 1 to 2 years or, in the case of published series, for the years relevant to those series. Infants who were referred to the unit just for treatment were excluded, apart from the unit in Argentina, which is the national referral unit for treatment (“Argentina G”). Information on the number of infants who were examined during the relevant time period was not requested as we were not intending to report rates of ROP, which would be difficult to interpret in light of variation in case mix, level of care, and screening criteria. Ethical approval for release of these routinely collected data was obtained, when necessary.


Data on the birth weight and GA of infants who were treated for threshold ROP were provided from 3 highly developed countries (262 infants) and from a range of sources in 8 moderately developed and 2 poorly developed countries (1091 infants; Table 2). All but 1 of the individuals contacted supplied the data requested. Because of the different sources, the data presented in Table 3 and in Figs 1 and 2 are for infants with threshold disease or more advanced stages. All have been included, as all infants with stage 4, 5, or cicatricial ROP would have passed through the threshold, treatable stage of disease. The data show that the mean birth weights of infants with severe ROP in highly developed countries are lower than in moderately and poorly developed countries. In the 3 highly developed countries, the mean birth weight values all were <800 g, whereas the mean values for the other countries all were >1000 g, apart from Chile and Brazil (903 and 952 g, respectively). The mean GA values of infants with severe ROP in highly developed countries all were <26 weeks, which was lower than the values for the other countries, which ranged from 26.3 weeks in Lithuania to 33.5 weeks in Ecuador. Overall, 142 (13%) of 1091 infants in this study with severe ROP from moderately and poorly developed countries had birth weights and GAs exceeding those recommended for screening by the Royal College of Ophthalmologists of the United Kingdom (ie, birth weight <1500 g and/or GA <32 weeks).13 Only 5 (0.5%) infants had a GA of ≥37 weeks, whereas 39 (3.6%) exceeded a criterion of <34 weeks and/or <1750 g. Only 1 (0.4%) of the 262 infants who were treated for threshold disease in Canada, the United Kingdom, and the United States exceeded United Kingdom screening criteria.

Fig 1.

A, Birth weight and GA of infants reported with threshold disease from countries with high UNDP HDIs between 1996 and 2002. B, Birth weight and GA of infants reported with severe ROP from countries with low/middle HDIs between 1996 and 2002. The horizontal line indicates the boundaries of the United Kingdom screening data.

Fig 2.

A, Box plots of birth weights of infants reported with severe ROP from 13 countries with varying levels of development. B, Box plots of GAs of infants reported with severe ROP from 13 countries with varying levels of development.


Sources of Information and Screening Criteria


Birth Weight and GA of Infants Reported With Severe ROP in Countries Ranked by Their HDI Between 1996 and 2002


This study suggests that the population of infants who develop severe ROP in highly developed countries differs from those who are affected in less well-developed countries. Given the complex interaction between case mix, neonatal care, and survival rates, as well as variation in screening practices and follow-up rates of discharged infants, this represents the best snapshot of the situation currently possible. Several limitations of this study need to be acknowledged. Different methods of assessing GA may have been used, the data come from a variety of sources (individual neonatal units, a treatment referral center; national data, a low vision clinic, and a national surgical referral center), and the data provided may not be representative of all units in the countries concerned. There are, therefore, potential sources of bias. For example, data from the treatment referral center in Argentina may have underascertained very low birth weight/GA infants who were too sick to be transferred to the regional center for treatment. Another potential source of bias relates to differences in the interpretation of the different signs of ROP by the screening ophthalmologists and inaccuracies in recording data. These limitations are almost inevitable in retrospective studies of this nature. Variation in screening criteria may also explain the findings: some ophthalmologists in countries with moderate and poor levels of development follow the US guidelines, some follow the United Kingdom guidelines, and some have generated their own. One explanation for the “highly developed country” picture of some units in moderately developed countries (eg, Brazil) is that larger, more mature infants were not included in the screening program. However, it seems more likely that this reflects the excellent care that some infants are receiving. Variation in case mix may explain, in part, the findings but would not explain why larger, more mature infants are developing severe ROP. The implications of these findings are that screening guidelines developed in highly developed countries will not suit all situations. The finding that bigger, more mature infants are developing threshold ROP in less developed countries or present too late for treatment has led some ophthalmologists in Latin America to widen their inclusion criteria for diagnostic screening. Some now examine infants with birth weights <2000 g and/or GA <37 weeks, and in Argentina, risk factor criteria have been expanded to include blood transfusion with adult blood, sepsis, or >36 hours in oxygen. In Ecuador, the ROP screening program started in 1994, and initially only infants with birth weights <1500 g were examined. The criteria were changed the following year to <1901 and/or <37 weeks because several unexamined infants with birth weights >1500 g presented with inoperable stage 5 ROP. Since adopting the wider criteria, no infant has become blind from ROP because of screening failure. A recent publication from Bangkok, Thailand, supports the findings of our study: the authors suggested region-specific criteria of <1500 g or <33 weeks’ GA, emphasizing that these need to be evaluated and revised if necessary.18

The current picture of the infants who are affected by severe ROP in moderately and poorly developed countries seems to suggest a mixture of “first epidemic” risk factors (inadequately monitored oxygen) and the “second epidemic” risk factors (extreme prematurity), reflecting variation in levels of neonatal care. In this study, we did not attempt to address issues of variation in neonatal care in detail, but policies, practices, and levels of training and expertise do vary between countries and service providers. For example, in the unit in Brazil, all infants who receive oxygen are monitored, whereas in the unit in Peru, only ventilated infants who receive oxygen are being monitored. There is, therefore, an urgent need to monitor oxygen saturation levels in all infants who receive supplemental oxygen where this is currently not being done. Although this will not prevent all severe ROP, it will reduce its incidence, particularly in larger, more mature infants, as has happened in North America and Western Europe. From the ophthalmic perspective, there is a need to train and equip ophthalmologists to work with neonatologists and nurses so that comprehensive screening programs can be established in all units that admit high-risk infants, using local protocols developed from national guidelines. Establishing these services will not be straightforward, particularly in Latin America, where neonatal care is provided by a range of different service providers (government, private, charities, and through insurance schemes).

To provide the evidence on which to base decisions regarding appropriate screening criteria, standardize, prospective studies are needed to (1) identify the population of infants at risk and (2) determine whether the screening criteria need to vary according to case mix, levels of care, or neonatal outcomes. These issues are currently being explored in a prospective study in Rio de Janeiro, Brazil (principal investigators C.G. and A.Z.). If a single criterion of <37 weeks' GA were applied to the data presented in this article, then only a few infants with severe ROP would have been missed. However, increasing the GA criterion would significantly increase the number of infants who need examination, which may not be justifiable for the marginal gain. However, studies from the United Kingdom show that more mature infants need fewer examinations, and the majority of infants with GA >32 weeks are likely to need only 1 examination.19 Broadening the criterion, therefore, may not increase the workload as much as might be anticipated. Another issue is that more mature infants are likely to be discharged before the first examination, making implementation of a screening program challenging. Ideally, decisions about appropriate screening criteria, as well as the timing of examinations, should be based on locally available data. Until those criteria are established, it is advisable for those who undertake screening to err on the side of caution and use wide screening criteria and not rely on published criteria from highly developed countries. The implications of this study are that communities with relatively sparse health resources have the greatest ROP screening workload. As this situation is likely to continue for the foreseeable future, innovative methods of screening, perhaps using nonmedical personnel, need to be explored.


Salary support for Dr Gilbert was provided by Sight Savers International United Kingdom. The United Kingdom cohort study of infants with stage 3 ROP was funded by the Department of Health of England and Wales.

Other members of the International NO-ROP Group and other contributors: R.V. Azad, R. Bagdoniene, A.M. Benitez, L. Bouzas, M.M. Contreras, L.A. Galina, A. Gelzinis, L. Gopal, L. Haines, S. Jalali, A.D. Kairiene, X. Katz, S.K. Lee, J. Manzitti, M. Mier, M.T. Sepúlveda, J.C. Serrano, R. Sirtautiene, and N.X. Tinh.

All authors played a role in providing information, interpreting data, and writing the article. All have seen and commented on the final draft. Other members of the International NO-ROP group provided information and commented on drafts.


    • Accepted November 18, 2004.
  • Address correspondence to Clare Gilbert, MD, International Centre for Eye Health, Department of Infectious Tropical Diseases, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, United Kingdom. E-mail: clare.gilbert{at}
  • No conflict of interest declared.

ROP, retinopathy of prematurityUNDP, United Nations Development ProgrammeHDI, Human Development IndexGA, gestational age