Percentage of individuals with anylevel of functional limitations within each attribute of the Health UtilitiesIndex Mark 3. Comparison of results from the 87 children in the randomized(RZ) group who had a visual acuity of 20/200 or worse at age 10 years, the157 children in the RZ group who had 1 or more sighted eyes at age 10 years,and the 102 children in the reference group who did not develop retinopathyof prematurity (ROP).
Distribution of health-related quality-of-life(HRQL) scores for the health states of the 86 children in the randomized (RZ-blind/lowvision) group who had visual acuity of 20/200 or worse at age 10 years andcomplete Health Utilities Index Mark 3 results, the 157 children in the randomizedgroup (RZ-sighted) who had 1 or more sighted eyes at age 10 years, and the102 children with birth weights less than 1251g who did not develop ROP (NoROP/reference group) and who were participants at one study center in thenatural history portion of the multicenter trial of Cryotherapy for Retinopathyof Prematurity study.
CRYO-ROP Cooperative Group. Health-Related Quality of Life at Age 10 Years in Very Low-Birth-WeightChildren With and Without Threshold Retinopathy of Prematurity. Arch Ophthalmol. 2004;122(11):1659-1666. doi:10.1001/archopht.122.11.1659
To describe parental perspectives on health status and health-relatedquality of life (HRQL) at age 10 years in children with birth weights lessthan 1251 g who participated in the multicenter Cryotherapy for Retinopathyof Prematurity (CRYO-ROP) study.
In 244 participants in the randomized CRYO-ROP trial and 102 CRYO-ROPparticipants who did not develop ROP, the Health Utilities Index (HUI) systemwas used to characterize health status for the following 8 attributes: vision,hearing, speech, ambulation, dexterity, emotion, cognition, and pain. Usinga utility formula, HRQL was determined for each child on a scale from 0.0(dead) to 1.00 (perfect health).
The proportion of the ROP-randomized group with limitations in 4 attributesor more was 20.6% compared with 2.0% for the no-ROP group. Within the ROP-randomizedgroup, the proportion of “sighted” children with limitations in4 attributes or more was 6.4% vs 46.5% in the “blind/low vision”group. The median HRQL score for the ROP-randomized children was lower thanfor the no-ROP children (0.72 vs 0.97, P<.001);the median HRQL score for the sighted-randomized children was 0.87 vs 0.27for the blind/low vision children (P<.001).
Threshold ROP was associated with functional limitations in health attributesand reduction in HRQL scores at age 10 years. Furthermore, among childrenwho developed threshold ROP, a greater reduction in HRQL scores was foundamong children with a poor visual outcome compared with those with bettersight.
Preterm birth may have a lifelong influence on general health, academicabilities, and behavior.1- 3 Inearlier studies, these conclusions were based on standard psychometric measuresand histories elicited from parents and teachers. More recently, Saigal andcolleagues4,5 have used a multiattributeapproach, the Health Utilities Index (HUI) system,6- 8 toreport the health status and health-related quality of life (HRQL) in a cohortof extremely low-birth-weight (ELBW, <1000 g) children at age 8 years,compared with a cohort of full-term children of the same age. In that study,the health status was determined retrospectively from objective psychometricmeasures, and a utility equation was applied based on preferences from a randomsample of 194 parents of school-aged children. Results indicated that themean HRQL score was significantly lower in the ELBW group than in the referencegroup, with higher variability of scores in the ELBW group. Particularly notablein the ELBW group were limitations in cognition (58%), sensation (vision,hearing, and speech) (48%), mobility (21%), and self-care (17%) compared with28%, 11%, 1%, and 0%, respectively, for children in the full-term referencegroup. A significantly higher proportion of the ELBW children had limitationsin multiple attributes.5
Retinopathy of prematurity (ROP) is a significant and relatively commonmorbidity of extremely preterm birth and is sometimes associated with otheradverse perinatal events such as periventricular leukomalacia,9 intraventricularhemorrhage,9- 11 bronchopulmonarydysplasia,12- 14 andnecrotizing enterocolitis,12- 14 aswell as visual impairment.14- 18 However,comparisons of HRQL in preterm infants with and without ROP have not beenreported.
Recently, 247 children with birth weights less than 1251 g who had thresholdROP during the neonatal period and who participated in the randomized trialof cryotherapy for severe (threshold) ROP (CRYO-ROP) returned for a 10-yearfollow-up.19- 21 Aspart of the 10-year examination, an interview of parents/guardians was conductedto assess the health status and HRQL of these children. During the same period,identical 10-year follow-up examinations were conducted on a reference cohortof 102 children who were also participants in the CRYO-ROP study at 1 of the23 study centers, who had birth weights less than 1251 g, but who did notdevelop ROP in the neonatal period.20,21
The purpose of this article is to provide HRQL results for childrenin the cohort that developed threshold ROP, as well as for a concurrent referencegroup of children who did not develop ROP. We hypothesized that children inthe threshold cohort would have lower HRQL scores than the generally healthierchildren with larger birth weights in the reference group, and that childrenwhose threshold ROP led to blindness or very low vision would have lower HRQLscores than those children with threshold ROP in whom good visual functionwas preserved.
Subjects were 2 groups of 10-year-old children who had birth weightsless than 1251 g and who were participants in the multicenter CRYO-ROP study,in which subjects were enrolled from January 1, 1986, through November 30,1987. One group was the 255 surviving children (291 were initially enrolledat 23 centers) who had developed threshold ROP (defined as stage 3 ROP inzone 1 or in zone 2 with 5 continuous or 8 cumulative sectors of fibrovascularproliferation in the presence of plus disease)22 duringthe neonatal period and who were participants in the randomized trial of cryotherapy.Although detailed neurological information is unavailable for these subjects,data obtained at the 5 ½- and 8-year examinations indicated that functionallimitations were present in many of the children who had developed thresholdROP.23,24 Two hundred forty-seven(96.8%) of the 255 surviving children in this group returned for the 10-yearfollow-up examination.19 The Health UtilitiesIndex Mark 3 (HUI3)23 questionnaire was administeredto the parents of 244 and complete answers to the HUI3 questionnaire wereobtained from parents of all but 1 of the children. The second group (thereference group) consisted of a concurrent sample of 104 CRYO-ROP study participantsat the Philadelphia study center who did not develop ROP during the neonatalperiod.20,21 One hundred two (98.1%)of these children returned for the 10-year follow-up examination and responsesto the HUI3 questionnaires were obtained from the parents of all 102 children.This study was approved by the institutional review boards at all participatinginstitutions and written informed consent was obtained from parents/guardiansat study enrollment, prior to randomization, and prior to each follow-up phase.
For some analyses, children in the randomized group were subdividedinto a “blind/low vision” group and a “sighted” group.The blind/low vision group consisted of 86 children whose parents providedcomplete HUI3 results and 1 child whose parents failed to provide data forthe emotion and cognition attributes of the HUI3. All had visual acuity (VA)scores in the better eye that were classified as an unfavorable outcome inthe CRYO-ROP study.19,21 Thisincluded VA of 20/200 or worse as measured by the Early Treatment DiabeticRetinopathy Study (ETDRS) distance VA charts (n = 17),19,25 gratingacuity worse than 6.4 cycles per degree as measured by the Teller acuity cardprocedure (n = 13),19,25 orby vision too poor to be measured with the ETDRS charts or the Teller acuitycards (minimal pattern vision, defined as detection of the 2.2-cm-wide stripeson the Low Vision Teller acuity card, light perception only, or no light perception)(n = 56). The sighted group consisted of 157 children (64.3%) whohad VA classified as a favorable outcome in the CRYO-ROP study.19 Onehundred fifty-four had VA at age 10 years that was better than 20/200 in atleast 1 eye when tested with ETDRS logarithm of the minimum angle of resolution(logMAR) distance VA charts.19,26 Threeadditional children in the sighted group were not testable with the ETDRScharts but had VA of 6.4 cycles per degree or better when tested with theTeller acuity card procedure.19,25 Amongthe 102 children in the no-ROP group, all but 1 child had VA better than 20/200in both eyes and were testable with the ETDRS charts. The 1 child who couldnot be tested with the ETDRS charts provided Teller acuity card results of9.6 cycles per degree in the better eye, indicating that it was appropriateto consider this child as “sighted.”
A complete eye examination was performed by a study-certified ophthalmologist.Monocular VA, with spectacle correction if needed, was measured by 1 of 2study-certified testers who were masked to the treatment status and currentocular condition of the child’s eyes.19 Thetester performed a standardized protocol using ETDRS logMAR distance VA charts.19,26 If the child could not be testedwith the ETDRS chart, monocular grating acuity was assessed using the Telleracuity card procedure.19,25
Parents or caregivers of the children were interviewed by study personnelwho had been trained (by S.S.) in the administration of the questionnaire.The interviewers followed a written script (based on the HUI3 ClassificationSystem)27 that was administered in exactlythe same order to all respondents.28,29 Thehealth status was measured from the perspective of the parent or guardianproxy respondents who reported both the health and the functional level withineach attribute according to their best judgment. The interview for measuringthe health status took less than 10 minutes for each study participant.
The HUI Classification System was used to report the health status ofthe subjects.6,30 The HUI is ageneric, comprehensive, valid, and reliable system. It has demonstrated thecapability of discriminating among levels of functional deficits among childrenwith cancer31,32 and a populationof ELBW school-aged children and controls.4,5 Inthe current study, the version of the system used was the HUI3 ClassificationSystem, which superseded the HUI2 Classification System used in the earlierstudies.4,5,28,31,32 TheHUI3 Classification System consists of 8 attributes: vision, hearing, speech,ambulation, dexterity, emotion, cognition, and pain. These attributes wereempirically selected from a list of 15 attributes considered most importantby parents and children.33 Each attribute haseither 5 or 6 defined levels of severity, ranging from normal function tosevere dysfunction (Table 1). The healthstatus of a particular patient may be described as a profile by an 8-elementvector (x1,x2, . . . x8, in which xi describes thelevel of functioning [1-5 or 1-6] for attribute i).
Results from each of the 8 attributes were classified as “normal,”that is, no functional limitations on any of the several levels of functionallimitations, or “abnormal” (Table1).8,27 For the 8 attributes,normal was defined as follows:
Vision: Able to see wellenough to read ordinary newsprint and recognize a friend across the street,with (level 2) or without (level 1) glasses. (In view of the high prevalenceof high refractive errors in the participants of this study, we elected toconsider levels 1 and 2 of the HUI3 Classification System as normal when describingthe population for this attribute. However, for calculation of utility scores,levels 1 and 2 were entered separately into the equation.)
Hearing: Able to hearwhat is said in a group conversation with at least 3 other people, withouta hearing aid (level 1).
Speech: Able to be understoodcompletely when speaking with strangers or friends (level 1).
Ambulation: Able to walkaround the neighborhood without difficulty and without walking equipment (level1).
Dexterity: Full use of2 hands and 10 fingers (level 1).
Emotion: Happy and interestedin life (level 1).
Cognition: Able to remembermost things, think clearly, and solve day-to-day problems (level 1).
Pain: Free of pain anddiscomfort (level 1).
The HUI3 Classification System was designed to have full structuralindependence.8 Thus, a child who is blind mayhave normal ambulatory abilities, even though navigation may be affected.Similarly, dexterity may be normal although the child may be unable to exerciseit owing to blindness.
The above HUI3 Classification System can be linked directly with preference-basedscoring models called “multiattribute utility functions.” Thissystem converts descriptive measures of ability or disability within eachattribute obtained from respondents into a measure of overall HRQL based ona mathematical utility formula developed by Torrance and colleagues.7,27 The formula for HUI3 was developedin 1994 using preference measurements collected from a random sample of 504adults living in Hamilton, Ontario, using a combination of the Visual AnalogScale (feeling thermometer) and Standard Gamble Technique (chance board).The HRQL score obtained, thus, is an estimate of the score that would be obtainedhad utility for the health state been measured directly from a random sampleof the general population using the Standard Gamble instrument.8 Inthis study, therefore, the HRQL scores represent the health status as reportedin an interview by parents of less than 1251-g birth-weight infants in thecontext of the values of general public adults.
The HUI3 multiattribute utility scores provide a spectrum from 0.00(“dead”) to 1.00 (“perfect health”). The HUI3 scaleis defined for the interval –0.36 to 1.00.8 Negativescores represent states considered “worse than dead.” Furtherdetails on the calculation of HRQL scores are described elsewhere.27
As utility scores are continuous measures with interval scale properties,these data are reported as mean and standard deviation of the group mean.The statistical significance of the distribution differences in the HRQL scoresbetween the groups was assessed using the 2-sample Wilcoxon rank sum (Mann-Whitney)test.
Table 2 provides demographic informationfor the randomized and no-ROP groups of children, as well as for the subgroupsof blind/low vision and sighted children who comprised the randomized group.As reported previously,20,21 althoughall children in both groups had to be less than 1251 g birth weight, childrenwho developed threshold ROP and participated in the randomized trial had meanlower birth weights and mean gestational ages than did study participantswho did not develop ROP. Children in the randomized group who had favorableVA in at least 1 eye at age 10 years had birth weights that were higher thanthose of children who were in the blind/low vision group. However, the remainingdemographic characteristics were similar for the 2 subgroups of children inthe randomized trial, though they differ in comparison to children in theno-ROP group in racial demographics and in whether they were born in a studyhospital (inborn vs outborn). Table 2 alsogives the median VA scores of children in each group and indicates that themedian VA of the children in the sighted-randomized group, all of whom wereclassified as having favorable VA outcomes for the primary study analysis,was poorer than the median VA of the children in the no-ROP group.
Based on interviews with parents or guardians, Figure 1 shows the proportion of children with functional limitationsat any level in each of the 8 attributes of the HUI3 for the blind/low visionand sighted subgroups of children in the randomized cohort and for the no-ROPreference group. In 5 of the attributes (vision, hearing, speech, ambulation,and dexterity), the no-ROP group had minimal functional limitations (<4.0%for each attribute). In contrast, parents or guardians of the children inthe randomized group who were blind or had low vision reported high ratesof functional limitation in vision (94.3%), speech (42.5%), ambulation (51.7%),and dexterity (32.2%) in their children. The corresponding figures for thechildren in the randomized group who were sighted indicated lower rates offunctional limitations ranging from 4.5% for dexterity to 16.6% for ambulation.Rates of functional limitation in the hearing attribute were similar: 3.4%for the blind/lowvision–randomized group and 3.8% for the sighted-randomizedgroup.
Rates of functional limitations reported by parents or guardians forthe cognition attribute were high for all 3 groups of children. Parents orguardians of nearly one third (31.4%) of children in the no-ROP group, almosttwo thirds (64.0%) of children in the randomized group who were blind/lowvision, and half (51.0%) of children in the randomized group who were sightedindicated that their children had functional limitations in the cognitionattribute.
Rates in the pain attribute were similar for all 3 groups while ratesin the emotion attribute were similar for the no-ROP reference group (13.7%)and the sighted-randomized group (15.3%), but somewhat higher (25.6%) forchildren in the blind/low vision randomized group.
Table 3 provides information onthe frequencies of the number of attributes that are affected in the individualchild for the randomized and no-ROP cohorts.Two (2.3%) of 86 children in theblind/low vision–randomized group, approximately one third of childrenin the sighted-randomized group, and half of children in the no-ROP groupwere judged by parents or guardians to be normal for all 8 attributes. Theproportion of children in the blind/low vision–randomized group whoseparents or guardians indicated limitations for 4 or more attributes was 46.5%,in contrast to 6.4% for children in the sighted-randomized group and 2.0%for children in the no-ROP group.
Table 4 gives a description ofoverall HRQL scores of children for the randomized and no-ROP cohorts. MedianHRQL score for the blind/low vision–group (0.27) is significantly lowerthan the median HRQL score for the sighted-randomized group (0.87, P<.001). Although the median score for the sighted-randomized groupis significantly lower than the median HRQL score for the no-ROP referencegroup (0.97, P<.001), the difference is much smaller.All 3 groups of children showed a broad range of scores, from below 0 to 1.00,despite differences in median scores.
Figure 2 shows the variabilityin the distribution of HRQL scores within the blind/low vision–randomized,sighted-randomized, and no-ROP reference groups. In the blind/low vision–randomizedgroup, very few children had HRQL scores above 0.65, and most children hadscores in the lower ranges, including a substantialproportion with scoresbelow 0. In contrast, a substantial proportion of children in the sighted-randomizedgroup and most children in the no-ROP group had scores above 0.65. Twenty-fivechildren (29.1%) in the blind/low vision–group, 2 children (1.3%) inthe sighted-randomized group, and the 1 child (1.0%) in the no-ROP referencegroup had scores that were below 0.
In a study of HRQL in 8-year-old children, Saigal and colleagues4,5 found significantly lower HRQL scoresfor preterm children with birth weights less than 1000 g compared with a groupof full-term children of the same age (median HRQL score 0.88 vs 1.0, P<.0001). However, few children in this cohort had significantvisual morbidity. The present study evaluated HRQL in 2 groups of 10-year-oldpreterm children with birth weights less than 1251 g and found that the HRQLscores were related both to the presence of threshold ROP and to the visualoutcome following threshold ROP. Overall, the highest HRQL scores were foundfor the group of children who did not develop ROP, who, on average, had birthweights that were 262 g higher and gestational ages that were 3.3 weeks greaterthan those of children with ROP in the randomized group, despite similar enrollmentcriteria. This is reflective of the nature of ROP that is more prevalent andmore severe in the most immature infants. Furthermore, the higher HRQL scoresin the no-ROP reference group may reflect the better visual acuities thatwere found in this group of children (Table 1), as well as their somewhat increased maturity at birth comparedwith the children in the randomized group.
The lowest HRQL scores were observed in those children who developedthreshold ROP in the neonatal period and were bilaterally severely visuallyimpaired or blind. The group of children who developed severe ROP in the neonatalperiod and retained VA better than 20/200 showed a pattern of HRQL resultsthat resembled those of children in the no-ROP reference group, albeit witha lower median score and a broader distribution of scores, even though theyhad an mean birth weight of only 817g, an average gestational age of only26.5 weeks, and median VA 20/40 in comparison to the median VA of 20/20 inthe no-ROP group
Within the group who developed threshold ROP, there was a marked differencein both the median scores and the distribution of scores between those childrenwho had VA of 20/200 or worse in both eyes (blind/low vision) and those whohad VA that was better than 20/200 in at least 1 eye (sighted) (Figure 2). In fact, the distribution of HRQL scores for the sighted-randomizedgroup was more similar to that in the no-ROP group than it was to the distributionof scores in the blind/low vision–randomized group. Scores greater than0.95, which are associated with, for example, having to wear glasses but otherwisebeing in perfect health, were found in 33.8% of children in the sighted-randomizedgroup and in 52.0% of children in the no-ROP reference group, but in only2.3% of children in the blind/low vision randomized group. At the lower endof the distribution, 1.3% of children in the sighted-randomized group and1% of children in the no-ROP group had scores that were below 0, in comparisonto 29.1% of children in the blind/low vision randomized group. Thus, nearlyone third of children in the blind/low vision-randomized group had scoresindicating the health status of the child was viewed as being worse than dead.
The utility formula applied to the aforementioned data was based onthe perspective of adults in the general population. The HRQL scores of thechildren in the current study might have been higher had the preferences beenelicited directly from the children. Furthermore, many of the low scores obtainedin the present study might well be because of the perception of the adultsfrom the general population, from whose preferences the formula was derived,that blind children would likely have poor functioning in their daily lives.In another study, Saigal and colleagues28,29 interviewed12- to 16-year-old ELBW adolescents (and their parents) to determine theirhealth status and measured preferences with the chance board. The resultsshowed HRQL results that were fairly positive, even though a significant proportionof the children had been diagnosed as having neurosensory impairments.28,29 Saigal et al suggested that the childrenand their parents may have learned to accept their disabilities and may haverecalibrated their personal expectations. Although utility scores obtainedfrom patients and/or parents are clearly relevant for decision making at anindividual level, utility scores based on community or general public values,as used in the current study, may be more appropriate for assessing the efficacyof interventions and for allocation of health care resources for clinicalprograms.34
Another way of comparing HRQL outcome among the 3 groups is to examinethe number of attributes in which the parents or guardians judged that functionallimitations were present (Table 3).In the blind/low vision–randomized group, only a very small percentageof children were judged to have no attributes affected, while one third ofthe sighted-randomized and more than half of the no-ROP group were judgedto have no attributes affected. Thus, the HRQL scores among the sighted-randomizedgroup were closer to those of the no-ROP reference group than to those inthe blind/low vision–group. This similarity between the sighted-randomizedand no-ROP groups, despite differences in the birth weights, gestational ages,and VAs of the children, is likely owing to the lower burden of visual andother associated morbidities in the sighted group compared with the blind/lowvision–group. Furthermore, it is possible that the differences in VAbetween the sighted-randomized group (median VA 20/40) and the no-ROP group(median VA 20/20) are of little consequence in the quality of everyday lifeof a 10-year-old child.
An encouraging finding for the blind/low vision–group is thatdespite the very low HQRL scores in many children in this group, more than25% of these children were judged by their parents or guardians to have functionallimitations in only 1 attribute (Table 3),and in 22 of 157 of these children the single attribute that showed a functionallimitation was vision. It is not surprising that some of the substantial differencesin HRQL scores shown by the sighted vs the blind/low vision–subgroupsof children who had a history of threshold ROP are likely to be compoundedby the sequelae of perinatal morbidities in this immature population.11,12,35,36 Thatis, low vision or blindness may not be the only cause of poor health status,but simply one of many outcomes that may occur in the sickest, most fragilepreterm babies. Indeed, poor or low vision may be a “marker” forbrain injury.23 In a recent study, severe retinopathy,ultrasonographic signs of brain injury, and bronchopulmonary dysplasia stronglypredicted the risk of subsequent death or neurosensory impairment.37 Unfortunately, data on respiratory distress syndrome,periventricular leukomalacia, intraventricular hemorrhage, and other severemorbidities of preterm birth were not recorded in the CRYO-ROP study.
It is important to recognize that the children in the CRYO-ROP studywere born in 1986 and 1987. It is certainly reasonable to speculate that theoutcomes of survivors of neonatal intensive care nurseries from 2003 may besomewhat better than those of children who are 16 and 17 years old. Visionloss among current survivors would be expected to be less, due to the useof cryotherapy or laser therapy for threshold ROP,38 aswould other aspects of perinatal morbidity, owing to improvements in neonatalcare. Both eyes were never treated in this randomized control study, thoughbilateral treatment is generally done when both eyes develop threshold ROP.When both eyes do receive treatment, the child’s chance of having 1eye with a good outcome is increased.
The present study provides data on several dimensions of health statusin a large cohort of 10-year-old children who had birth weights less than1251 g and who were participants in the CRYO-ROP study.19- 21 Thestudy followed a rigorous methodology, and the attrition rate was remarkablylow. To our knowledge, this is the first time that HRQL has been measuredin a large number of formerly premature children with and without significantvisual impairments. Clearly, significant visual impairment, often associatedwith multiple morbidities related to extreme immaturity, results in a substantiallygreater long-term burden for the individual, the family, and society. Knowledgeof the quality of life in these children is desirable for both the medicalcommunity and society as a whole, as these children may require prolongedand expensive care in the neonatal period and beyond. Such assessments areimportant for quality assurance, medical decision making, counseling of parents,and resource allocation.
Correspondence: Graham E. Quinn, MD, MSCE,Division of Pediatric Ophthalmology, The Children's Hospital of Philadelphia,One Children's Center, Philadelphia, PA 19104 (email@example.com).
Reprints: CRYO-ROP Study Headquarters,Oregon Health & Science University, Casey Eye Institute, 3375 SW TerwilligerBlvd, Portland, OR 97239-4197.
Submitted for Publication: July 14, 2003; finalrevision received January 16, 2004; accepted May 14, 2004.
Funding/Support: This study has been supportedin part by grant U10 EY05874 from the National Eye Institute, National Institutesof Health, Department of Health and Human Services; Bethesda, Md.
Financial Disclosure: None.
Previous Presentation: Some of the materialin this article was presented at the annual meeting of the American Associationof Pediatric Ophthalmology and Strabismus; March, 25, 2003; Waikoloa, Hawaii.
Acknowledgment: We are grateful to David Feeny,PhD, University of Alberta, Calgary, and William Furlong, PhD, McMaster University,Montreal, Quebec, for their advice regarding measurement of HRQL and for theirpermission to use the HUI3 questionnaires.