Using chronological age alone may fail to diagnose threshold retinopathy of prematurity (ROP) in larger-birth-weight infants (1000-1250 g). Shaded area shows percent of infants with threshold disease potentially missed by screening at 8 weeks of chronological age. Adapted from Palmer et al.12
Using postconceptional age alone may fail to diagnose threshold retinopathy of prematurity in smaller-birth-weight infants (<750 g). Shaded area shows percentage of infants with threshold disease potentially missed by screening at 34 weeks of postconceptional age. Adapted from Palmer et al.12
Hutchinson AK, Saunders RA, O'Neil JW, Lovering A, Wilson ME. Timing of Initial Screening Examinations for Retinopathy of Prematurity. Arch Ophthalmol. 1998;116(5):608-612. doi:10.1001/archopht.116.5.608
To test a screening protocol for retinopathy of prematurity (ROP) that uses the dual criteria of postconceptional age and chronological age, rather than a single parameter, to determine precisely when to begin eye examinations.
We retrospectively reviewed medical records of 179 infants (326 eyes) who had undergone laser treatment for threshold ROP. We entered their chronological and postconceptional ages at treatment into a database and evaluated various screening parameters to determine the combination of criteria that would allow us to safely postpone the initial eye examinations.
Screening infants at 7 weeks of chronological age or 34 weeks of postconceptional age (whichever comes first), but not before 5 weeks of chronological age, seems to reliably detect the onset of threshold ROP while reducing the number of unnecessary early examinations.
Simultaneously applied dual criteria of chronological age and postconceptional age may be a superior method of determining when to initiate ROP examinations and is preferable to using either chronological age or postconceptional age alone.
THE IMPORTANCE of screening eye examinations for retinopathy of prematurity (ROP) has dramatically increased as clinical studies have shown improved visual outcomes in infants with severe, acute ROP after treatment with either transscleral cryotherapy1- 4 or transpupillary laser therapy.5- 9 In 1995, the Royal College of Ophthalmologists British Association of Perinatal Medicine recommended that the first eye examination take place between 6 and 7 weeks of chronological age in infants deemed at risk for ROP.10 More recently, the American Academy of Pediatrics, the American Association for Pediatric Ophthalmology and Strabismus, and the American Academy of Ophthalmology released a joint statement recommending that initial screening examinations be performed between 4 and 6 weeks of chronological age or 31 and 33 weeks of postconceptional age.11 Although the concept of considering both chronological age and postconceptional age represents an advance over previous approaches, these guidelines lack specificity. For example, a conservative examiner might screen all infants at 4 weeks of chronological age, regardless of postconceptional age, while a more liberal examiner could choose to screen infants at 33 weeks of postconceptional age regardless of chronological age. While both regimens adhere to current recommendations, we suspected that the more stringent protocol would result in many unnecessary examinations while the more liberal protocol might fail to diagnose the onset of threshold ROP in a substantial number of infants.
We propose more specific screening criteria that consider both chronological age and postconceptional age simultaneously to determine when to begin screening examinations for ROP. Based on published data from the Multicenter Clinical Trial of Cryotherapy for Retinopathy of Prematurity (CRYO-ROP study)12 we hypothesized that dual criteria would allow us to reliably diagnose the onset of threshold ROP while eliminating many early and potentially unnecessary eye examinations. This hypothesis was tested using a data set of 326 eyes diagnosed with and treated for threshold disease.
We retrospectively reviewed medical records of 179 infants (326 eyes) who underwent argon laser treatment to the avascular retina for threshold ROP. For this cohort, the definition of threshold in zone II was the same as the CRYO-ROP study: 5 contiguous or 8 cumulative clock hours of stage 3 ROP in the presence of plus disease (dilation and tortuosity of the posterior pole retinal vessels).1 However, more recent reports suggest that eyes with posterior disease are more likely to have adverse outcomes, possibly justifying earlier treatment.13,14 Thus, in eyes with retinal vascular maturity partially or completely limited to zone I, we considered threshold for treatment to be any stage 3 ROP in the presence of plus disease.
The cohort of infants studied was derived from a much larger group screened for ROP by one of us (J.W.O.) between 1993 and 1996. This physician is experienced in the diagnosis and treatment of ROP and received his residency and fellowship training at one of the participating centers of the CRYO-ROP study. During the study period, he performed approximately 50 screenings per month on infants with birth weights of 1500 g or less in the Phoenix, Ariz, metropolitan area. These examinations were conducted at a total of 11 different hospitals, although the majority of patients were derived from 3 major level III nurseries.
Infants were screened at 4 weeks of chronological age unless this occurred before the infant reached 31 weeks of postconceptional age. If so, the examination was postponed until 31 weeks of postconceptional age. In each case demographic characteristics, including birth weight, chronological age, and postconceptional age at the time of treatment, were entered into our database. Follow-up examinations were routinely performed weekly in infants with more severe ROP and all were treated within 72 hours of diagnosis of threshold disease; hence, we used the date of laser treatment to approximate the onset of threshold ROP.
Using these data, we then considered different screening protocols to determine which would be the most effective in diagnosing the onset of threshold ROP, yet require the fewest number of early examinations to reduce stress and potential trauma to these fragile, low-birth-weight infants.
Of the 179 infants treated for threshold ROP, one patient was excluded because of late referral. At the time of first examination (47 weeks of postconceptional age, 24 weeks of chronological age), evidence of cicatricial retinal detachment (stage IV) was already present. Table 1 presents the birth weights and estimated gestational ages at birth for our patient population and compares this with the population of patients with threshold ROP in the CRYO-ROP study.12 The infant subpopulations were statistically similar except for a larger proportion (27% vs 15%, P=.001) of infants weighing 1000 g or more at birth in our study. This disparity reflects our screening of infants up to 1500 g of birth weight, vs 1250 g in the CRYO-ROP study.
Table 2 gives the chronological and postconceptional ages at which our infants reached threshold ROP. Table 3 summarizes the effectiveness of various combinations of chronological age and postconceptional age in detecting the onset of threshold ROP. Based on this information, we established our optimum screening protocol. We recommend that to assure both efficiency and safety, initial screening should be performed at 7 weeks of chronological or 34 weeks of postconceptional age, whichever comes first. Furthermore, because no infant in our cohort or in the CRYO-ROP study reached threshold ROP before 5 weeks of chronological age, we added this as a third criterion.
Table 4 shows the number of weeks the initial screening examinations could have been delayed by using the aforementioned screening criteria rather than chronological age or postconceptional age alone. Our protocol is compared with screening at 5 weeks of chronological age and 31 weeks of postconceptional age, because this is when screening would be required using a single parameter to detect all threshold events at or before their onset.
In broad terms, the goals of an efficient screening program are to detect treatable disease at as high a rate as possible while simultaneously minimizing the number of examination results that either are normal or do not alter the course of clinical care. Larger babies (birth weight >1500 g) generally do not receive ophthalmologic examinations because the incidence of severe ROP is low and these infants are likely to have a favorable outcome even without treatment.14,15 Although all premature infants, regardless of birth weight or other risk factors, could be screened for ROP, we believe this would be an inefficient use of medical resources and cause unnecessary examination trauma to many infants. Insofar as possible, a good screening protocol should be carefully tailored to the population at risk and be modified in accordance with new data as they become available.11,12
The CRYO-ROP study systematically collected data on 4099 premature infants in the United States weighing less than 1251 g at birth.1 The CRYO-ROP protocol called for initial ophthalmologic examination at 4 or 6 weeks of chronological age with follow-up at least every 2 weeks until any ROP was found to be regressing or retinal vascular maturation had reached zone III (nasal retinal vessels at the ora serata). The purpose of these examinations was to document the progression of ROP and to identify infants who were randomization candidates for cryotherapy. Subsequent publications analyzing these data have shown in graphic form the time of onset of threshold disease in different subsets of infants based on birth weight, chronological age, and postconceptional age.12Figure 1 shows that larger infants (1000-1250 g) began reaching threshold disease at an earlier chronological age than smaller ones. Conversely, Figure 2 shows that the very smallest infants (<750 g) began reaching threshold disease at an earlier postconceptional age than their larger counterparts. Thus, using a single screening criterion for all premature infants would require examination of some who are not realistically at risk or, alternatively, forgoing examination on others who are at substantial risk. The following examples illustrate the nature of this dilemma when attempting to eliminate early examinations.
Screen all infants at 8 weeks of chronological age (Figure 1). Virtually all infants with birth weights less than 1000 g will be detected before the onset of threshold ROP. However, approximately 30% of larger infants (1000-1250 g) would first be examined after the onset of threshold disease.
Screen all infants at 34 weeks of postconceptional age (Figure 2). Almost all infants at birth weights greater than 750 g will be detected prior to the onset of threshold ROP. However, approximately 15% of the very smallest infants would be first examined after the onset of threshold disease.
Clearly, based on published data from the CRYO-ROP study, the number of early screening examinations cannot be safely reduced using a single screening criterion. Using chronological age alone will increase the risk of missing the onset of threshold ROP in larger-birth-weight infants. Using postconceptional age alone will increase the risk of missing the onset of threshold ROP in smaller-birth-weight infants.
In their recently released guidelines for ROP screening examinations, the American Academy of Pediatrics, American Association for Pediatric Ophthalmology and Strabismus, and American Academy of Ophthalmology invite modifications.10 Our data suggest that the first examination can often be safely delayed by using the combined criteria of chronological age and postconceptional age. Using chronological age alone, our infants would have required screening no later than age 5 weeks to diagnose ROP at or before the onset of threshold disease in every case (Table 2). Using postconceptional age alone, our infants would have required screening by age 31 weeks to diagnose ROP at or before the onset of threshold disease in every case (Table 2). In contrast, Table 3 presents the result of using dual screening criteria in the same cohort of infants with threshold ROP: performing the first screening examination at 7 weeks of chronological age or 34 weeks of postconceptional age, whichever comes first, yields only 2 eyes with threshold ROP at the time of their initial examination. However, lengthening either the chronological age or postconceptional age parameters by 1 week (to 8 weeks and 35 weeks, respectively) results in at least a few infants in whom threshold ROP would not have been diagnosed within 1 week of onset.
Occasionally, small for gestational age infants born at 31 or 32 weeks will fall within the 1500 g or smaller birth weight category and require screening for ROP. In these infants, our protocol would theoretically call for eye examinations at 2 or 3 weeks of chronological age. Because no infant in our study or the CRYO-ROP study developed threshold ROP before 5 weeks of chronological age, we added this as an exclusion to prevent additional unnecessary examinations. Thus, we recommend the initial screening examination for ROP be performed at 7 weeks of chronological age or 34 weeks of postconceptional age, whichever comes first, but not before 5 weeks of chronological age. These recommendations are in agreement with those recently published by the American Academy of Pediatrics, the American Association for Pediatric Ophthalmology and Strabismus, and the American Academy of Ophthalmology,11 but provide more specific and needed guidance to the examiner.
While timely detection of threshold ROP is clearly essential for any successful screening protocol, eliminating unnecessary and stressful examinations is also a worthwhile goal because these may be hazardous to the infant. Other than external ocular trauma, recognized risks include apnea and/or bradycardia (including cardiorespiratory arrest), adverse systemic effects of the dilating eyedrops, and nosocomial infection. In addition, ROP examinations potentially interfere with infant feeding schedules, disrupt the flow of the nursery, and can be upsetting to patients' families. With an estimated 52000 infants per year weighing 1500 g or less at birth and potentially requiring screening examinations for ROP,16 a substantial number of early examinations could be avoided. While the variability in individual nursery practices makes this number difficult to determine with accuracy, even eliminating 1 early examination in 50% of infants could reduce by 26000 the number of examinations performed nationwide, presumably at no measurable risk of failing to diagnose threshold ROP.
Applying dual criteria of chronological age and postconceptional age to determine the time of the first ROP examination is a potentially effective method of postponing the initial screening examination in some infants without failing to diagnose threshold ROP. Our data suggest that infants weighing 1500 g or less at birth could be safely screened for ROP at 7 weeks of chronological age or 34 weeks of postconceptional age, whichever comes first, but not before 5 weeks of chronological age. However, these conclusions are based on a specific data set from an isolated population of premature infants with threshold ROP who have demographic characteristics different from the larger population of patients studied in the CRYO-ROP study.1,12 Evidence for these differences include a higher rate of threshold ROP in our cohort of patients in spite of screening infants up to 1500 g of birth weight. In addition, improved survival rate of very-low-birth-weight infants has increased the frequency of zone I disease, which is now usually treated earlier than in previous years. Our patient population was composed almost exclusively of white and Hispanic infants who appear to develop severe ROP at a substiantially greater rate than black infants.17 Finally, it is uncommon for infants in the Phoenix metropolitan area to undergo blood transfusion with a hematocrit of greater than 0.20. We suspect that the presence of chronic anemia may play a role in the retinal hypoxia and contribute to the progression of ROP in this population.
While we are confident in the accuracy of our ophthalmologic findings, this study is nevertheless retrospective and several assumptions were necessary to allow data interpretation. The most important of these was using the date of treatment to approximate the onset of threshold ROP. If a substantial interval of time, such as 1 week, elapsed between the onset of threshold disease, ophthalmoscopic diagnosis, or the actual treatment, our conclusions might be different. A similar analysis of a more strictly controlled database, such as that found in the CRYO-ROP study, would be helpful in determining the validity of our recommendations.
Accepted for publication January 30, 1998.
This research was supported in part by an unrestricted grant to the Storm Eye Institute from Research to Prevent Blindness Inc, New York, NY.
Reprints: Richard A. Saunders, MD, Storm Eye Institute, Medical University of South Carolina, 171 Ashley Ave, Charleston, SC 29425-2236.