 |
|
Risk Analysis of Prethreshold Retinopathy of Prematurity
Robert J. Hardy, PhD;
Earl A. Palmer, MD;
Velma Dobson, PhD;
C. Gail Summers, MD;
Dale L. Phelps, MD;
Graham E. Quinn, MD, MSCE;
William V. Good, MD;
Betty Tung, MS; for the Cryotherapy for Retinopathy of Prematurity Cooperative Group
Arch Ophthalmol. 2003;121:1697-1701.
ABSTRACT
| |
Objective To present a new multifactorial algorithm to integrate important risk factors for unfavorable retinal outcome in eyes that reached prethreshold retinopathy of prematurity (ROP) in the Cryotherapy for Retinopathy of Prematurity (CRYO-ROP) study. A computerized risk model (RM-ROP2) was developed from this algorithm to identify high-risk prethreshold eyes for enrollment in the Early Treatment for Retinopathy of Prematurity randomized trial.
Methods Data were analyzed from 613 eyes (1 eye per infant) in the natural history cohort of the Multicenter Trial for Cryotherapy for Retinopathy of Prematurity. These eyes were selected from infants in whom 1 or both eyes progressed to prethreshold ROP. Eyes that progressed to threshold ROP and were randomized to cryotherapy were excluded from this study, but control eyes that reached threshold ROP were included. The course of ROP for 1 prethreshold eye for each infant was tracked until the evaluation of its structural outcome at 3 months' postterm.Tables present structural outcome by se lected risk characteristics. A multiple logistic risk model is used to summarize the combined effect of all of these known prognostic risk factors as they relate to structural outcome.
Results Eyes were classified by predicted outcome into 10 risk categories, lowest to highest. Both the observed and predicted outcomes in each category showed an increasingly unfavorable outcome when viewed from lowest to highest risk. Prethreshold ROP eyes were then divided into 2 groups, high-risk, eyes (risk, 0.15-1.0) and low-risk eyes (risk, <0.15). High-risk eyes had an unfavorable outcome of 36% at 3 months' postterm; whereas, only 5% of the low-risk eyes had an unfavorable outcome.
Conclusion The model effectively identifies prethreshold ROP eyes that have a relatively high risk and eyes that have a lower risk of an unfavorable structural outcome at 3 months.
INTRODUCTION
IN 1991, the Cryotherapy for Retinopathy of Prematurity Cooperative Group (CRYO-ROP) published detailed incidence figures for ROP by severity of disease and also tabulated them by an array of patient subgroupings such as birth weight, race, and others.1 In 1993, we reported the results of multiple logistic regression analysis of numerous independent variables, both for baseline characteristics of infants and by ROP classification features.2 That article quantified, for the first time, the risk effect of these variables on the likelihood of an eye reaching the threshold ROP (Table 1) for treatment, and the likelihood that an eye that reached threshold ROP and was untreated would progress to an unfavorable outcome 3 months later.2 We also have published outcome statistics for untreated ROP tabulated by baseline infant characteristics and numerous severity categories.3-4
|
|
|
|
Table 1. Definitions of Prethreshold ROP, Threshold ROP, and Unfavorable Physician's Summary Diagnosis
|
|
|
While the published risk analysis data lent objectivity to estimating prognosis, it remained difficult for physicians to quantify the risk for an individual infant. Major progress in that direction was achieved through the development of a mathematical algorithm, converted into a risk analysis computer program known as RM-ROP.5
As originally described, RM-ROP consists of 5 mathematical equations that provide a relationship among risk factors observed about the infant and the infant's retina as they correlate with structural outcome. The program is based on data from 4099 infants who weighed less than 1251 g at birth who composed the natural history cohort of the Multicenter Trial for Cryotherapy for Retinopathy of Prematurity.1-2 The RM-ROP analyzes data acquired during acute-phase ROP to calculate 2 risk estimates: (1) that an individual infant's eye will develop threshold ROP, and (2) if threshold is reached, that the eye with or without treatment will develop an unfavorable structural outcome (and very likely, blindness) 3 months later.
We have developed a different version of the program that calculates the risk of an unfavorable 3-month outcome for eyes at prethreshold ROP severity (Table 1), rather than waiting until the eye reaches threshold ROP. This new risk model, RM-ROP2, is the subject of this article. The RM-ROP2 has been used to select high-risk prethreshold ROP eyes for enrollment in a clinical trial, the Early Treatment for Retinopathy of Prematurity study sponsored by the National Eye Institute, National Institutes of Health, Bethesda, Md.6
METHODS
STUDY POPULATION
A group of infants was selected as follows from the natural history cohort of the CRYO-ROP study.3 The full natural history cohort consists of 4099 subjects who were enrolled from January 1, 1986, through November 30, 1987, as newborns with birth weights less than 1251 g. Beginning at 4 through 6 weeks of age, they were examined by specially trained and certified ophthalmologists every 2 weeks for the development of ROP, and later for outcome 3 months' postterm or after randomization. If ROP progressed to the prethreshold level, the examination interval was shortened to within 7 days. Data from the 731 infants who developed prethreshold or worse ROP in 1 or both eyes were eligible for inclusion in this study. (The remaining 3368 infants either had no ROP or developed ROP that remained less severe than the prethreshold category.) One eye of each infant was selected for the present analysis.
In cases in which only 1 eye reached prethreshold or worse ROP during the course of acute-phase ROP, that eye was chosen for study. For infants with both eyes reaching prethreshold ROP, the following selection criteria were used: (1) if neither eye reached the treatment threshold, 1 eye was selected at random; (2) if both eyes eventually reached threshold ROP, the eye randomized in the CRYO-ROP study to not receive cryotherapy (control eye) was included; and (3) if only 1 eye eventually reached threshold ROP severity and was randomized to serve as the control eye, that eye was included; conversely, if the eye was randomized to receive cryotherapy, the untreated prethreshold fellow eye was included. Following the eyes in groups 1 through 3 produces a natural history of untreated eyes through onset of prethreshold ROP, including those eyes that progressed to treatment threshold ROP but were untreated. This allows the risk model to properly estimate the risk of an unfavorable 3-month outcome.
These data are presented in tabular form by infant and eye characteristics. The multiple logistic risk model was used to examine the prognostic factors related to an unfavorable structural outcome. The multiple logistic equation is described and the coefficients of each of the factors included in the model are given. Calculation of the multiple logistic risk model was done using Stata Statistical Software: Release 7.0.7
RESULTS
Of the 731 infants who constituted the prethreshold CRYO-ROP study population, a subgroup of 613 infants had their outcome determined at 3 months. Outcomes were unavailable from the remaining 118 infants for the following reasons: (1) 43 died before the 3-month examination, (2) 66 were not examined because of parental refusal (n = 27) or loss of contact with the family (n = 39), and (3) 9 were examined but the examiner was unable to classify the outcome. The baseline characteristics of both the entire cohort and the subgroup of 613 eyes with 3-month outcome data are summarized in Table 2. The subgroup of infants with 3-month outcome data is overall remarkably similar to the base natural history cohort.
|
|
|
|
Table 2. Baseline Risk Characteristics*
|
|
|
The rates of unfavorable fundus outcome are given in Table 3 by various characteristics of the infants and their selected eyes. As expected from previously published risk factors, infants born in 1 of the participating hospitals had a lower rate of an unfavorable outcome than those born in nonparticipating hospitals (15.5% vs 21.1%, P = .13). Likewise, black infants had lower rates of unfavorable outcome than nonblack infants (8.6% vs 19.7%, P<.001). Singleton infants had an unfavorable outcome rate of 17.5%, whereas multiple-birth infants had a slightly lower rate of 14.5% (P = .44). The unfavorable outcome rate for female infants was 18.4%, while male infants had a slightly lower rate of 15.1% (P = .28). By birth weight categories, the unfavorable outcome rates vary from 23.7% for infants weighing less than 750 g at birth down to 9.0% for infants weighing 1000 to 1250 g at birth. Similarly, infants born more prematurely had higher rates of unfavorable outcome than infants born later in gestation.
|
|
|
|
Table 3. Unfavorable Outcome Rates at 3 Months for 613 Eyes That Had Untreated Prethreshold ROP
|
|
|
Table 3 also gives the outcome of the eyes categorized by the clinical findings at the time of designation as prethreshold ROP. Within each zone of disease, eyes with plus disease (defined as dilation and tortuosity of posterior pole retinal vessels) had a higher rate of unfavorable outcome than eyes without plus disease. Note that for ROP in zone II, eyes with stage 2 with plus disease had more unfavorable outcomes than eyes with stage 3 without plus disease. The percentage of unfavorable eyes varied only slightly with postconceptional time of onset of ROP; however, it decreased with increasing duration from the onset of ROP to prethreshold level, ranging from 27.4% when that interval was 1 week, to 6.4% when the interval was longer than 3 weeks.
To further evaluate these risk factors individually as well as their simultaneous influence on the outcome of ROP, a logistic risk analysis was applied to these data. The multiple logistic risk model RM-ROP2 for prethreshold eyes has the following mathematical form:
p = {1 + exp[-( +  1x1 + 2x2 + ... + kxk)]}-1.
Each xi is an infant or eye factor that increased (or decreased) the risk p, of having an unfavorable outcome. The iand are coefficients in the risk model that are estimated from these data. The i is the coefficient associated with xi and is a constant term. The function exp raises the expression in brackets to the base e = 2.71828. . . .
The results, given in Table 4, are summarized in terms of the regression coefficients and odds ratio for each factor. For example, the odds of an unfavorable outcome, p/(1 - p), for infants who have plus disease at the first prethreshold examination are 8.6 times higher than for those who do not have plus disease. Both birth weight and gestational age were important; however, neither reached statistical significance in the multiple variable analyses. Because these 2 factors are biologically related, they should be considered jointly rather than separately. When considered together as an indicator of prematurity, they become statistically significant (P = .002) as predictive of unfavorable outcome. Black race was an important predictor of reduced risk. When clinical characteristics of ROP in each eye were studied, zone and stage of disease were important factors along with plus disease. Also, the interval of ROP progression from onset to prethreshold level played a major role in the prognosis. All ROP characteristics that classified an eye at the prethreshold level were important indicators of outcome prognosis. However, the presence of zone I disease and the presence of plus disease were particularly strong indicators of a poor prognosis.
|
|
|
|
Table 4. Multiple Logistic Coefficients for 613 Natural History Eyes (Untreated) in the CRYO-ROP Study
|
|
|
The multiple logistic model summarizes the experience of this group of prethreshold eyes. Figure 1 shows the fit of the model in 10 risk categories each containing approximately one tenth of the sample. Hence, there are approximately 60 eyes in each risk category (lowest to highest). Both the observed and expected percentages of eyes that had an unfavorable outcome increased from the lowest to the highest decile of risk. They are also similar within each decile category ( 28 = 5.43, P = .71).
|
|
|
|
Observed and expected percentage of eyes with unfavorable 3-month outcome by decile of risk (10 risk categories). The number of eyes with retinopathy of prematurity studied was 613 with approximately 60 eyes per decile. Both the observed and expected percentage of eyes that had an unfavorable 3-month outcome increased from the lowest to the highest decile of risk.
|
|
|
A further evaluation of the results was obtained by dividing the prethreshold ROP eyes into 2 groups using a cutoff point for the risk at 0.15. On the upper side of risk were those with a risk of an unfavorable outcome of 0.15 to 1.00 vs those with a risk less than 0.15. During the planning of the Early Treatment for Retinopathy of Prematurity study, the span of cutoff point values between 0.10 and 0.20 was examined. The value of 0.15 was chosen as the enrollment level of risk for the study because it included within the high-risk group of 235 eyes, 68% that progressed to threshold ROP and/or an unfavorable outcome. It also appeared to minimize assigning the eyes for early treatment that did not progress to threshold ROP or an unfavorable outcome by reducing this to 32% (other cutoff point information available from the corresponding author).
Data showing the results for these 2 risk groups are given in Table 5, further classifying eyes according to International Classification of ROP categories. For eyes with a risk 0.15 to 1.00, 36% had an unfavorable 3-month structural outcome. When the risk was less than 0.15, 5% had an unfavorable 3-month structural outcome. This striking separation in the unfavorable structural outcome rate was fairly consistent across the International Classification of ROP categories, with high-risk eyes showing more adverse outcomes than low-risk eyes. If we look at the progression to conventional threshold ROP, for eyes designated as high risk, 63% progressed to the conventional threshold ROP for treatment and for eyes designated as low risk, 14% progressed to threshold ROP.
|
|
|
|
Table 5. Outcome for High-Risk and Low-Risk Eyes With ROP by ICROP Classification at First Determination of Prethreshold ROP*
|
|
|
COMMENT
The CRYO-ROP study has reported various prognostic factors that play a role in acute-phase ROP.2 A further article from the CRYO-ROP study described using the profile of these factors to develop a series of risk models.6 These RM-ROP models were designed as part of a novel system to aid physicians in assessing the risk of ROP in individual infants. The risk of reaching threshold ROP could be calculated with RM-ROP at birth, at onset of ROP, and at onset of prethreshold ROP. Then, at threshold ROP the risk of an unfavorable outcome at 3 months with or without treatment could be calculated.
In this article, again using data from the CRYO-ROP multicenter trial, we focus on a newer computer program (RM-ROP2) based on the natural history of the latter part of acute-phase ROP, from the time an eye is identified as having prethreshold ROP until the ROP either resolves or progresses to an unfavorable structural outcome at 3-months' follow-up. Using RM-ROP2 in a clinical setting requires a biweekly schedule of eye examinations by an experienced and knowledgeable examiner, beginning at 4 through 6 weeks of life and continuing until prethreshold ROP develops. Data on an infant's demographic characteristics and the progression of ROP are used to calculate the risk that the infant's eye would have an unfavorable fundus outcome at 3 months without surgical intervention for threshold ROP. The RM-ROP2 integrates all the information about the infant and the ROP observed in the eye examination into a single risk estimate. The coefficients in Table 4 provide the relative weighting of the demographic and clinical factors that was mathematically derived from the database.
The clinical course and visual outcome are generally favorable for eyes that do not progress to prethreshold ROP.4 However, variability in outcome occurs once an eye reaches prethreshold ROP, making RM-ROP2 useful in predicting risk. To illustrate the use of RM-ROP2, presume that an infant has reached prethreshold ROP. At this point an assessment of risk becomes desirable because of the wide range of possible risk levels and the potential for reaching the standard treatment threshold. Further, suppose that the infant is white, born in a participating CRYO-ROP center, a singleton birth, with a birth weight of 640 g and a gestational age of 26 weeks. The ROP began at 34 weeks, and prethreshold ROP was first observed at 36 weeks in the form of stage 3 with plus disease in zone II, but without the requisite number of clock hour sectors to comprise threshold ROP. All this demographic and clinical information are entered into the RM-ROP2 system. The RM-ROP2 uses its mathematical risk model to simultaneously consider the pertinent characteristics of the infant at birth and later ocular characteristics at prethreshold ROP (Table 2 and Table 4). Based on the assumed information in the example, RM-ROP2 provides an estimate that the risk of the eye's having an unfavorable outcome is 0.45.
The Early Treatment for Retinopathy of Prematurity study used the RM-ROP2 system to identify a select group of prethreshold ROP infants with a relatively high risk of an unfavorable outcome. These selected high-risk prethreshold eyes were randomized to either receive immediate peripheral laser ablation, or to have prethreshold ROP managed in the conventional way, with treatment only if the retinopathy progressed to threshold ROP severity. Based on analyses of the CRYO-ROP data, a cutoff point of 0.15 was chosen as the minimum risk level to test the efficacy of earlier intervention at prethreshold level vs later treatment (if required, at threshold). The selection of the cutoff point and the RM-ROP2 model were solely based on prethreshold ROP eyes in the CRYO-ROP study. A companion article in this issue of the ARCHIVES reports the application of the RM-ROP2 model and the results of the Early Treatment for Retinopathy of Prematurity multicenter clinical trial.8
AUTHOR INFORMATION
| |
Corresponding author and reprints: Robert J. Hardy, PhD, University of Texas–Houston Health Science Center, School of Public Health, Coordinating Center for Clinical Trials, 1200 Herman Pressler St, Room E827, Houston, TX 77030.
Submitted for publication August 27, 2003; final revision received October 7, 2003; accepted October 9, 2003.
This study was supported by grant U10 EY05874, from the National Eye Institute, National Institutes of Health, Bethesda, Md.
From the School of Public Health, University of Texas–Houston (Dr Hardy and Ms Tung); Casey Eye Institute, Oregon Health & Science University, Portland (Dr Palmer); Department of Ophthalmology, University of Arizona, Tucson (Dr Dobson); Departments of Ophthalmology and Pediatrics, University of Minnesota, Minneapolis (Dr Summers); Departments of Pediatrics and Ophthalmology, University of Rochester, Rochester, NY (Dr Phelps); Division of Pediatric Ophthalmology, The Children's Hospital of Philadelphia, University of Pennsylvania School of Medicine, Philadelphia (Dr Quinn); and the Smith-Kettlewell Eye Research Institute, San Francisco, Calif (Dr Good). The authors have no relevant financial interest in this article. A list of the members of the Cryotherapy for Retinopathy of Prematurity Cooperative Group can be found at http://www.nei.nih.gov./neitrials/static/study32.htm#Clinical Centers.
REFERENCES
1. Palmer EA, Flynn JT, Hardy RJ, et al, for the Cryotherapy for Retinopathy of Prematurity Cooperative Group. Incidence and early course of retinopathy of prematurity. Ophthalmology. 1991;98:1628-1640.
ISI
| PUBMED
2. Schaffer DB, Palmer EA, Plotsky DF, et al, for the Cryotherapy for Retinopathy of Prematurity Cooperative Group. Prognostic factors in the natural course of retinopathy of prematurity. Ophthalmology. 1993;100:230-237.
ISI
| PUBMED
3. Cryotherapy for Retinopathy of Prematurity Cooperative Group. The natural ocular outcome of premature birth and retinopathy: status at one year. Arch Ophthalmol. 1994;112:903-912.
FREE FULL TEXT
4. Editorial Committee, for the Cryotherapy for Retinopathy of Prematurity Cooperative Group. Natural history ROP: Multicenter Trial of Cryotherapy for Retinopathy of Prematurity: ocular outcome at 5 years in premature infants with birth weights less than 1251g. Arch Ophthalmol. 2002;120:595-599.
FREE FULL TEXT
5. Hardy RJ, Palmer EA, Schaffer DB, Phelps DL, Davis BR, Cooper CJ, for the Multicenter Trial of Cryotherapy for Retinopathy of Prematurity Cooperative Group. Outcome-based management of retinopathy of prematurity J AAPOS. 1997;1:46-54. [published correction appears in J AAPOS. 1977;1:137].
6. Good WV, Hardy RJ, for the ETROP Multicenter Study Group. The Multicenter Study of Early Treatment for Retinopathy of Prematurity (ETROP) [guest editorial]. Ophthalmology. 2001;108:1013-1014.
FULL TEXT
|
ISI
| PUBMED
7. Stata Corp. Stata Statistical Software: Release 7.0. College Station, Tex: Stata Corp; 2001.
8. Early Treatment for Retinopathy of Prematurity Cooperative Group. Revised indications for the treatment of retinopathy of prematurity: results of the Early Treatment for Retinopathy of Prematurity randomized trial. Arch Ophthalmol. 2003;121:1684-1696.
FREE FULL TEXT
 CiteULike  Connotea  Del.icio.us  Digg  Reddit  Technorati  Twitter
What's this?
RELATED ARTICLES
Revised Indications for the Treatment of Retinopathy of Prematurity: Results of the Early Treatment for Retinopathy of Prematurity Randomized Trial
Early Treatment for Retinopathy of Prematurity Cooperative Group
Arch Ophthalmol. 2003;121(12):1684-1694.
ABSTRACT
| FULL TEXT
Preliminary Results of Treatment of Eyes With High-Risk Prethreshold Retinopathy of Prematurity in the Early Treatment for Retinopathy of Prematurity Randomized Trial
Alistair R. Fielder
Arch Ophthalmol. 2003;121(12):1769-1771.
EXTRACT
| FULL TEXT
THIS ARTICLE HAS BEEN CITED BY OTHER ARTICLES
|
Early Weight Gain Predicts Retinopathy in Preterm Infants: New, Simple, Efficient Approach to Screening
Hellstrom et al.
Pediatrics 2009;123:e638-e645.
ABSTRACT
| FULL TEXT
Progression to Severe Retinopathy Predicted by Retinal Vessel Diameter Between 31 and 34 Weeks of Postconception Age
Rabinowitz et al.
Arch Ophthalmol 2007;125:1495-1500.
ABSTRACT
| FULL TEXT
Changes in Pericytes and Smooth Muscle Cells in the Kitten Model of Retinopathy of Prematurity: Implications for Plus Disease
Hughes et al.
IOVS 2007;48:1368-1379.
ABSTRACT
| FULL TEXT
The further refinement of ROP treatment. Commentary on the Early Treatment of Retinopathy of Prematurity study: structural findings at 2 years of age
Fielder and Wilkinson
Br J Ophthalmol 2007;91:136-137.
FULL TEXT
Screening for Retinopathy of Prematurity: The Promise of New Approaches
Good
Arch Ophthalmol 2006;124:1775-1776.
FULL TEXT
The Early Treatment for Retinopathy of Prematurity Clinical Trial: presentation by subgroups versus analysis within subgroups.
Hardy et al.
Br J Ophthalmol 2006;90:1341-1342.
FULL TEXT
The Early Treatment for Retinopathy Of Prematurity Study: structural findings at age 2 years
The Early Treatment for Retinopathy of Prematurity
Br J Ophthalmol 2006;90:1378-1382.
ABSTRACT
| FULL TEXT
Prevalence and course of strabismus in the first year of life for infants with prethreshold retinopathy of prematurity: findings from the Early Treatment for Retinopathy of Prematurity study.
VanderVeen et al.
Arch Ophthalmol 2006;124:766-773.
ABSTRACT
| FULL TEXT
Outcome of Eyes Developing Retinal Detachment During the Early Treatment for Retinopathy of Prematurity Study (ETROP)
Repka et al.
Arch Ophthalmol 2006;124:24-30.
ABSTRACT
| FULL TEXT
Premature for Life
Fielder
Arch Ophthalmol 2005;123:392-394.
FULL TEXT
Clinical Application of the Revised Indications for the Treatment of Retinopathy of Prematurity
Jalali et al.
Arch Ophthalmol 2005;123:407-408.
FULL TEXT
The Dilemma of Exercising Clinical Judgment in the Treatment of Retinopathy of Prematurity--Reply
Good and Hardy
Arch Ophthalmol 2005;123:409-410.
FULL TEXT
The Early Treatment for Retinopathy of Prematurity Study: Better Outcomes, Changing Strategy
Phelps and on behalf of the ETROP Cooperative Group
Pediatrics 2004;114:490-491.
FULL TEXT
Revised Indications for the Treatment of Retinopathy of Prematurity: Results of the Early Treatment for Retinopathy of Prematurity Randomized Trial
Early Treatment for Retinopathy of Prematurity Coo
Arch Ophthalmol 2003;121:1684-1694.
ABSTRACT
| FULL TEXT
Preliminary Results of Treatment of Eyes With High-Risk Prethreshold Retinopathy of Prematurity in the Early Treatment for Retinopathy of Prematurity Randomized Trial
Fielder
Arch Ophthalmol 2003;121:1769-1771.
FULL TEXT
|
