Article Text


Retinopathy of prematurity—the “second lull”?
  1. Department of Epidemiology and International Eye Health, Institute of Ophthalmology, 11–43 Bath Street, London EC1V 9EL, UK clare.gilbert{at}

    Statistics from

    Ophthalmologists who trained in the 1960s and 1970s believed that retrolental fibroplasia was a tragic, iatrogenic problem of the 1940s and early 1950s. In the 1960s and 1970s screening programmes were generally not in place, as this was the period of the “first lull”—the first epidemic (due to unmonitored supplemental oxygen, often given routinely to all premature babies)1 was over, and the “second epidemic” (due to increased survival of very low birth weight babies) had not really started. The paper by Rowlandset al suggests that we may be entering a “second lull.”2 The study reports the changing incidence of retinopathy of prematurity (ROP) in babies in one unit in the United Kingdom over a 10 year period (1989–98), and shows a significantly lower incidence of stage 3–5 ROP in babies examined in the second 5 year period compared with the first. This is despite babies in the second cohort being more preterm with lower birth weights as a result of improved survival of extremely low birthweight babies. If the risk factors for ROP had remained constant over the study period, one would expect a higher incidence in the second cohort, given that birth weight and gestational age are the most important risk factors. The authors conclude that the lower incidence of ROP in babies born more recently is attributable to improvements in neonatal care, in particular to the use of prenatal steroids and surfactant. However, this statement has to be interpreted cautiously, as historical cohort studies are subject to inherent weaknesses.

    This paper raises several questions concerning the incidence of severe ROP. Firstly, has a fall in the incidence of ROP been reported in other studies, and if so, is this a universal trend? Secondly, if the trend is real, what are possible explanations? Thirdly, what are the implications for clinical practice and research, and lastly, are these applicable universally?

    Rowlands quotes studies from the USA, Denmark, and the UK which also report a reduction in the incidence of ROP over time, or where the incidence of ROP has remained stable despite more extremely premature babies being included. These studies seem to have provoked considerable controversy not only regarding whether the trend is real, but also over possible explanations. Comparison between neonatal units has to be interpreted with caution, as differences in the population of babies referred to the units may, in part, explain differences in incidence. However, there is an increasing body of evidence that suggests that the incidence and severity of ROP in babies <1250 g at birth is decreasing in industrialised countries. One has to be aware that this trend may not be universal—data from Latin American countries, and urban centres in developing countries such as India, show that babies with a much wider range of birth weights are developing threshold disease than in industrialised countries.34 The evidence suggests that in settings where neonatal care is inadequate, risk factors relevant during the “first epidemic” (that is, unmonitored supplemental oxygen) lead to severe ROP in more mature babies and the more premature babies are dying. Conversely, more sophisticated units reflect the “second epidemic,” with more premature babies developing ROP.5

    In the paper by Rowlands and colleagues the authors suggest that improved neonatal care explains the falling incidence in birthweight specific cohorts of babies, but how good is the evidence for this? Clinical trials of interventions in preterm babies are complicated by the fact that the organs of very premature babies are still differentiating as well as growing—the time of onset of the interventions as well as the dose, duration, and cessation of the intervention (for example, abrupt versus gradual), may all be relevant variables. Some trials have been designed to determine whether the intervention will prevent the outcome of interest, or to assess effectiveness once the condition has developed.

    A review of the Cochrane Collaboration Neonatal Group's list of systematic reviews shows that 106 reviews of clinical trials have been undertaken, of which 74 are for interventions targeted at preterm babies (website: The studies have reviewed clinical trials undertaken to investigate a wide range of interventions to improve neonatal outcomes, including control of insensible fluid loss; different methods of ventilation; different regimes for supplemental feeding; the role of synthetic or natural surfactants for the prevention or treatment of respiratory distress syndrome (RDS); dopamine and albumin treatment for hypotension; the timing, dose, and method of administration of steroids to prevent or treat chronic lung disease; the timing and speed of discontinuation of oxygen, and interventions to control body temperature. The wide range and quality of the trials, as well as the large number of systematic reviews, shows a highly commendable emphasis on evidence based medicine.

    Many of the studies undertaken did not include ROP as an main, or subsidiary, outcome but several did report the incidence of ROP. These studies can be divided into two groups: those which might have an impact on the development of ROP through stabilising the respiratory system and/or the newborn's physiology, and those which could arguably have a direct bearing on mechanisms implicated in the pathogenesis of ROP (that is, oxygen supplementation regimes, light exposure, antioxidant treatment). In the first group the following interventions were associated with a significantly lower risk of ROP:d-penicillamine treatment to prevent hyperbilirubinaemia, and treatment of infants with RDS with inositol (which promotes maturation of surfactant). Studies that did not show a statistically significant reduction in the incidence of ROP (usually severe ROP) include postnatal steroids given within 96 hours of birth, natural or synthetic surfactant given prophylactically to prevent RDS, or given as a treatment for established RDS. An insignificant increase in ROP was observed in studies where steroids were given more than 3 weeks after birth.

    In the second group of trials the incidence of ROP was less in babies in whom discontinuation of oxygen was gradual rather than abrupt, and where the use of oxygen was restricted. Early versus late discontinuation of oxygen did not have any impact on the development of ROP. A systematic review of studies restricting ambient light has also failed to demonstrate a reduction in ROP. A vitamin A supplementation trial showed a non-significant trend for a lower incidence of ROP in the supplemented group. Rather surprisingly, the Cochrane Neonatal Group has not undertaken a systematic review of the role of vitamin E in preventing ROP, where at least six clinical trials have been undertaken. A meta-analysis of these studies6 has not helped to resolve the current controversy, but it has been argued that an adequately powered clinical trial is justified, requiring a sample size of 2600 babies.7

    There are other, important clinical trials which are not included in the Cochrane Library. One is the STOP–ROP trial, which was designed to determine whether higher oxygen saturation once prethreshold disease had developed would prevent progression to threshold disease requiring treatment.8 Babies with prethreshold disease were randomly allocated to two groups: one group received supplemental oxygen so that oxygen saturation was maintained at 96–99% (supplemented group), while in the second group it was maintained at 89–94% (conventional group). The study showed that the rate of progression in the supplemental group was lower than in the conventional group, but the findings were not statistically significant. A subgroup of babies without “plus” disease at enrolment seemed to be more responsive to supplemental oxygen. Other, retrospective studies of oxygen supplementation have recently been reported, which showed better results. However, these studies were not randomised clinical trials and the findings need to be interpreted cautiously. The evidence that antenatal dexamethasone reduces the incidence and severity of ROP is also not derived from clinical trials.

    The rate of adverse events is another important aspect to consider in trials. In the STOP-ROP trial the rate of chronic lung disease was higher in the “supplemental” group than in the conventional group, and in the vitamin E trials there was a higher rate of a variety of adverse events (but not mortality) in supplemented babies. The potential for harm has to be balanced against potential for benefit, with implications not only for clinical practice, but also as further clinical trials may not be considered ethical.

    To summarise, many trials to assess interventions to improve neonatal outcomes have been undertaken, and some clearly show an impact on the development of ROP.

    What are the clinical implications of a falling incidence of severe ROP? Several authors have critically reviewed the criteria used for screening, which in the UK are a birth weight of ⩽1500 g, or a gestational age of <32 weeks (in the USA the criteria are a birth weight ⩽1500 g; gestational age <29 weeks, or babies weighing >1500 g with an “unstable clinical course”). As the vast majority of babies who develop threshold disease fall within these criteria, reducing the birthweight criterion to ⩽1250 g has been suggested.9 However, lowering the birthweight criterion would expose a small number of larger babies who are at risk, and research is needed to clearly establish relevant “sickness criteria” in babies weighing >1250 g.10 In an ideal world an objective test would be the best way of identifying babies at risk of severe ROP, and a recent study suggests that serum vascular endothelial growth factor (VEGF) may prove to be a useful predictive tool.11 In middle income countries and in urban settings in developing countries, screening criteria may need to vary widely depending on the level of care being provided, and some centres may need to include babies weighing ⩽1750 g at birth.3-5 Research into this important area is urgently needed so that the tragedy of the “first epidemic” can be avoided in countries which are expanding neonatal intensive care services.

    There are still many unknowns concerning the factors that initiate and promote the progression of ROP.12 Oxygen almost certainly plays an important part in both processes, and further research is warranted into optimum oxygen saturation levels, which may vary depending on postconceptual age. Other important areas of research include early nutritional interventions, exploration of objective tests which have a high positive predictive value for the development of severe ROP, and examination of the role of non-ophthalmologists for preliminary screening for ROP13 in settings where ophthalmologists skilled at indirect ophthalmoscopy are in short supply.

    View Abstract

    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.