Infantile Cataract in the Collaborative Perinatal Project: Title and subTitle BreakPrevalence and Risk Factors FREE

INFANTILE CATARACT is a rare disorder that accounts for a substantial proportion of childhood blindness.^{1 - 2} The condition is defined as lens opacity present at birth or detected within the first year of life. Infantile cataract produces optical blur of the retinal image, thus altering the quality of sensory information available to the child during sensitive periods of visual system development. Left uncorrected, it may result in deprivation amblyopia—a suppression of central vision.^{3 - 4} In economically developed nations, the prevalence of congenital and infantile cataract is approximately 6.0 per 10 000 infants surviving to early childhood (95% confidence interval[CI], 1.6-15.3); prevalence of childhood blindness due to cataract is 0.1 to 0.4 per 10 000 infants.⁵ In otherwise healthy children, infantile cataract may exist either as an isolated ocular defect or as part of a set of isolated visual system anomalies. Infantile cataract may occur in children with multisystem disorders or dysmorphic syndromes, either as the sole ocular anomaly or as part of a set of associated visual system defects.⁶

Pathogenic factors implicated in infantile cataractogenesis are multiple and diverse; many are rare and remain uncharacterized.⁵ There are genetic,^{7 - 11} environmental,² and metabolic⁶ causes. Approximately half of all bilateral and all of unilateral infantile cataract cases are idiopathic.¹ Hereditary causes are attributed in approximately 40% of cases with a known cause.¹ Environmental or metabolic factors are attributed in approximately 30% of all cases.^{2 ,5}

The purposes of the current study were (1) to estimate the prevalence of 4 types of infantile cataract in subjects surviving the neonatal period(first 28 days of life) from a large and geographically diverse US cohort; and (2) to investigate a set of demographic, lifestyle, obstetric, and infant-based risk factors for isolated infantile cataract. We used robust longitudinal data collected using standardized, masked-examination protocols from the Collaborative Perinatal Project (CPP) of the National Institute of Neurological Disorders and Stroke (Bethesda, Md)¹² to examine these issues.

A cohort of 55 908 pregnancies were enrolled in the CPP at 12 US university-affiliated medical centers from January 2, 1959, to December 31, 1965. A 95.9% follow-up from enrollment to childbirth was attained across all centers. Details of the sampling design were reported in a previous publication.¹³

Prospective longitudinal information on demographic, lifestyle, prenatal, perinatal, and neonatal factors was obtained from standardized personal interviews and medical examinations. Each mother was interviewed to obtain medical history and socioeconomic information about herself, her family, the infant's father, and his family at the prenatal checkup. Study obstetricians recorded the results of prenatal examinations, histories, and laboratory tests at all visits; information about labor and delivery was documented. Study pediatricians performed examinations of participants at the time of delivery, and at 4 months, 1 year, and 7 years of age. Collaborative Perinatal Project staff obtained detailed reports of consultations with nonstudy physicians and summaries of hospitalizations from patients' records across the first 7 years of life.

A pediatric neurologist or a pediatrician with training in neurology performed a standardized examination to identify physical, neurological, cognitive, and sensory defects at birth and at age 1 year; as part of this process, all neonates received a dilated funduscopic examination with direct ophthalmoscopy. Examining physicians were masked to information on prenatal and perinatal events. A funduscopic examination was required of all children with suspected neurologic or sensory abnormalities at 1 year of age. All children received funduscopic examinations at 7 years of age. Pupils were dilated if the fundi could not be adequately visualized. To improve case ascertainment, a member of our group (E.Y.C.) performed a detailed retrospective review of medical records for all children classified with infantile cataract. Infantile cataract was confirmed (1) if an ophthalmologist's medical report provided the diagnosis; or (2) if there were consistent pediatric or neurologic reports at the neonatal examination and during the follow-up examinations.

We estimated prevalence and calculated exact CIs¹⁴ on the 53 724 infants who survived the neonatal period. We report prevalence using 4 categories of cataract: (1) isolated; (2) part of a set of isolated ocular or visual system anomalies; (3) the sole ocular anomaly within a set of multisystem anomalies; and (4) part of a set of ocular, visual system, and other systemic anomalies.

We conducted multivariate risk factor analyses for confirmed cases of isolated infantile cataract (of nontraumatic or environmental origin) in a 3-step process. Participants with no history of cataract served as the comparison group. First, we identified risk factors from a set of variables with potential etiologic associations described in the biomedical and epidemiologic literature.^{6 ,15 - 16} The next step was to conduct bivariate analyses with exact methods. The Fisher exact test (SAS 8.0; SAS Institute, Cary, NC) was used for dichotomously coded variables. The χ² permutation test (StatXact version 4.0; CYTEL Inc, Cambridge, Mass) was used for polychotomously coded variables. We considered variables significant at P≤.30 on bivariate tests as putative correlates or predictors.¹⁷ These variables were then analyzed in a time-oriented regression modeling technique with exact methods (LogXact version 4.0, CYTEL Inc). The technique involved entering temporally ordered sets of exposure variables into a series of logistic models characterized by periods of risk (prenatal, perinatal, and postnatal periods). If a variable from one period was an independent predictor of outcome, it was retained throughout all subsequent models. Since many of the prenatal, perinatal, and postnatal factors were highly associated, the earliest occurring factors may have been eliminated or "displaced" from the final model by later-occurring ones if traditional regression methods were used. Data were incomplete (at least 30% missing) for toxemia and prior pregnancy loss variables; this precluded use in the multivariate models.

The original CPP database contained positive infantile cataract diagnoses for 110 of the 53 724 subjects who survived the neonatal period. Our final sample contained 73 subjects with confirmed cataract of nontraumatic origin and 53 639 subjects with no history of idiopathic, hereditary, or environmentally based cataract. Figure 1 illustrates the composition of the study sample.

Table 1 contains prevalence estimates for 4 cataract categories of subjects surviving the neonatal period. Infantile cataract was present in one of the following categories: (1) in an isolated form in 32 (43.8%) of 73 subjects with confirmed disease; (2) within a set of generalized ocular or visual system anomalies for 10 (13.7%) of 73 subjects; (3) as the sole ocular anomaly within a set of multisystem anomalies for 7 (9.6%) of 73 subjects; and (4) within a set of ocular, visual, and other systemic anomalies or syndromes for 24 (32.9%) of 73 subjects.

Thirty-eight subjects (52.1%) had unilateral disease. Thirty-six subjects(49.3%) were female. Proportions of subjects within each of the 4 cataract categories did not vary with laterality or sex. Cataract laterality varied by sex in that 24 (63.2%) of 38 subjects with unilateral cataract and 12 (34.3%) of 35 subjects with bilateral cataract were female (Fisher exact test P≤.02). Cataract was diagnosed during the neonatal period in 35 subjects (47.9%). These subjects included 11 (34.3%) from the isolated cataract group, 4 (40.0%) from the generalized ocular or visual system anomalies group, 3 (42.8%) from the isolated cataract plus multisystem (nonocular/nonvisual) anomalies group, and 17 (70.8%) from the generalized ocular or visual system anomalies plus multisystem anomalies and syndromes group. When we excluded subjects with confirmed congenital rubella syndrome, the resulting prevalence of any cataract was 11.2 per 10 000 (95% CI, 8.5-14.4). Prevalence of unilateral and bilateral disease was 6.1 (95% CI, 4.2-8.6) and 5.0 (95% CI, 3.3-7.3) per 10 000, respectively.

Forty-one subjects had a major malformation and cataract that was not trauma based (Figure 1), and 22(52%) of these had unilateral disease. Of the 18 subjects (44%) in this nonisolated cataract group who had a medical condition previously associated with cataract, 12 (67%) had bilateral cataract. Thirteen of these had mothers with prenatal rubella infection. There were 2 cases with metabolic disorders (galactosemia), 2 cases with inherited renal disease (Lowe syndrome), and 1 case with a chromosomal abnormality (trisomy 21). Seventeen (74%) of the 23 subjects with idiopathic cataract had unilateral disease. Microphthalmia was the most frequently associated ocular anomaly. Ten of the 11 subjects with microphthalmia had congenital rubella syndrome, and the majority of subjects with an associated ocular anomaly were identified to have some infectious, metabolic, or hereditarily based comorbidity.

Table 2 reports the distributions of correlates and potential predictors of isolated infantile cataract.

For all models, we retained maternal age (<35 years vs ≥35 years) as a demographic covariate. Results were essentially identical in situations where either parity, gravidity, or duration of the first stage of labor was modeled with other covariates. These collinear terms were never simultaneously modeled.

Exact multiple logistic regression analysis consistently demonstrated a 3-fold increased likelihood of any isolated infantile cataract among low-birth-weight(LBW; birth weight ≤2500 g) infants. The lower values of exact CIs for these odds ratios [ORs] ranged from 1.3 to 1.5, and the upper values ranged from 7.3 to 8.6. Infantile cataract was 3.8 times as frequent (95% CI, 1.5-8.6; P<.001) among LBW infants than among those with a birth weight greater than 2500 g after controlling for the effects of demographics, lifestyle, obstetric, and infant-based factors.

No factor identified in bivariate analyses was independently associated with the odds of developing unilateral isolated infantile cataract after entry into the multivariate model.

Low birth weight was the only independent predictor of outcome (OR = 4.4; 95% CI, 1.2-13.9; P = .005) for isolated bilateral cataract.

Infantile cataract is a rare disorder. Children born with LBW are more likely to have solated infantile cataract of noninfectious origin than are their heavier peers. To our knowledge, this is the first study to report results from multivariate analyses on risk factors for isolated infantile cataract in a geographically diverse US population of children surviving the neonatal period. The CPP is the largest longitudinal prospective study to date that collected detailed information on infantile cataract and associated risk factors. The strength of evidence provided by the CPP data is bolstered by the study's large-scale longitudinal prospective design and standardized data collection protocol. All children in this study, whether affected or unaffected by cataract, were evaluated and tested using the same standardized medical history and examination protocols at the same ages. Follow-up continued until 7 years of age. Such consistent and comprehensive data collection allows for a more accurate comparison of outcome groups.

Physicians performing follow-up child examinations were masked to information on prenatal maternal risk exposure. Information about the pregnancy was collected before the diagnosis of cataract, and this reduced the chance of the recall bias that is often found in retrospective investigations. Although examiners were masked to information on prenatal and perinatal factors, observation of infant size at the neonatal examination allowed them to make inferences about risk for abnormal development. This may have produced an ascertainment bias to the extent that it may have led to more extensive examinations of the smallest infants. This possibility was minimized by requirement of a standardized dilated funduscopic examination. We minimized misclassification of affected and unaffected children by reviewing the medical records of all suspected and confirmed cases of infantile cataract. The effects of outcome misclassification(undiagnosed cataract) in the comparison group on ORs would be negligible, considering the large sample size.

A limitation of the study was that existing data did not allow accurate determination of the proportion of subjects with hereditary cataract. We made assumptions about familial patterns of disease from CPP interview data on parental visual-sensory defects. Anomalies existed in the father of one infant with isolated bilateral cataract; in no other instance were they noted. Family history was reported in ophthalmic examinations. Two subjects with hereditary cataract were identified from this source; both had bilateral disease and generalized ocular or visual system defects, and multisystem defects.

Owing to the small number of cases, some associations between risk factors and infantile cataract may have been missed, and the power of the study may have been limited. Although our methods were unable to completely address this issue of specification error, we applied exact regression methods as the best existing analytic approach. Another consideration is whether results that are based on 40-year-old data are applicable today. Although it is possible that the distribution of some risk factors may have changed over time, we have no reason to believe that the effects of LBW (or those factors causing it) on infantile cataract have changed.

In this large prospective study of children born from 1959 to 1965, the prevalence of all forms of infantile cataract unassociated with trauma was 13.6 per 10 000 (95% CI, 10.7-17.1). Unilateral cataract occurred at a prevalence of 7.1 per 10 000 (95% CI, 5.0-9.7). Bilateral disease occurred at a prevalence of 6.5 per 10 000 (95% CI, 4.5-9.1). These estimates are within the limits of those reported from the 2 large prospective cohort studies completed in economically developed nations (Table 3, Figure 2). The National Birth Cohort Study¹⁸ was an evaluation of 97% of all 10-year-old children born in the United Kingdom between April 5, 1970, and April 11, 1970. Prevalence of cataract was 5.6 per 10 000(95% CI, 2.2-11.2). As in the CPP, frequency of cataract was approximately equal for unilateral and bilateral disease. Unilateral cataract was observed in 2.3 per 10 000 (95% CI, 0.5-6.8), and bilateral cataract was observed in 3.1 per 10 000 (95% CI, 0.8-8.0). A regionally based cohort study was performed on all children born in 1984 in the Oxfordshire Health District(England).¹⁹ The study group examined medical records when participants were 2 to 5 years old. Prevalence of cataract was 6.0 per 10 000 (95% CI, 1.6-15.3).

Although both studies demonstrate the rare occurrence of infantile cataract, the prevalence estimates are somewhat lower than those reported in our study. This difference may be related to the decline of intrauterine rubella embryopathy resulting from public health immunization programs, which were implemented after completion of the CPP. In CPP infants with no known prenatal rubella exposure, overall prevalence was 11.2 per 10 000 (95% CI, 8.5-14.4). Within this same population, it is possible that women with subclinical infections may not have been identified, though their children were diagnosed with cataract. Prevalence differences may also be explained by variations in outcome ascertainment methodology. Since the CPP used a standardized, protocol-driven dilated examination, case finding was nearly maximized.

The CPP and the UK studies could be affected by underreporting. The milder forms of infantile cataract are more likely to be missed on the examination using the integrity of the red reflex than on examinations using biomicroscopy. Nevertheless, results of these studies indicate that the prevalence of clinically significant infantile cataract is low in economically developed countries.

Many regional population-based studies,^{20 - 21} and national^{22 - 25} and regional^{22 ,26} population- and hospital-based surveillance systems report data on congenital and infantile cataract in economically developed countries (Table 3, Figure 2). Comparison with the present study is inappropriate due to differences in study design and implementation. Comparison across surveillance systems is also inappropriate, as data sources and methods of outcome ascertainment differ.

Half of the CPP study subjects with infantile cataract were diagnosed during the neonatal period. Researchers from the Centers for Disease Control and Prevention (Altanta, Ga) have proposed that approximately 50% of infants with clinically significant cataract will remain undiagnosed until after the neonatal period.²² Rahi, Dezateux, and the British Congenital Cataract Interest Group (BCCIG)²⁷ reported that 51% of infants with congenital or infantile cataract were diagnosed during the neonatal period. The BCCIG reports on an active ophthalmic and pediatric surveillance system implemented nationally in the United Kingdom that suggests that approximately 50% of all children with incident cataract diagnoses from 1995 to 1996 were age 10 weeks or younger.²³ Apart from problems associated with varying case definitions and outcome ascertainment methods, these findings are important to consider in comparing cohort study results.

Low birth weight was independently associated with a higher likelihood of developing infantile cataract. In the CPP data, the LBW variable represents a population of infants born at either a low gestational age (≤ 36 weeks) or at ages greater than 36 weeks, with intrauterine growth restriction (IUGR). Estimates of gestational age at birth were based on maternal report of last menstrual period and were somewhat less accurate than those derived by current methods. The ponderal index ([birth weight × 100]/[birth length]³) was derived from anthropometric data to identify subjects with IUGR. On entry of the ponderal index into the final models for any isolated cataract, point estimates for LBW were not substantially altered. Furthermore, ponderal index did not independently predict outcome. We observed the same results in infants with isolated bilateral cataract. The relationship between size at birth and preterm birth was further investigated by stratifying outcome within the preterm and full term subsets of all LBW subjects. Unadjusted prevalence of isolated infantile cataract within LBW subjects was 16.5 per 10 000(95% CI, 7.6-31.3). Unadjusted prevalence of isolated infantile cataract within LBW preterm subjects was 7.1 per 10 000 (95% CI, 0.9-25.5). The prevalence of isolated infantile cataract in LBW infants born after 36 weeks was 26.8(95%CI, 10.8-55.1) per 10 000. Our results suggest that the LBW effect seems to be mediated more by size at birth (IUGR or the processes that influence it) than by preterm birth. After entry into the logistic model, none of the factors associated with unilateral cataract predicted outcome. This finding is similar to those reported by the BCCIG. In one report, BCCIG investigators assigned idiopathic cause in 76 (97%) of 78 isolated unilateral cases, and in 53 (55%) of 97 isolated bilateral cases.¹⁵

In conclusion, CPP prevalence estimates for the 4 categories of infantile cataract are within the limits of those reported by large-cohort studies in economically developed nations. Isolated infantile cataract and isolated bilateral infantile cataract occurred 3 to 4 times as often among in children born with LBW than among children born at greater than 2500 g in the CPP population.

Submitted for publication January 4, 2002; final revision received June 12, 2002; accepted July 8, 2002.

Corresponding author and reprints: John Paul SanGiovanni, ScD, Division of Epidemiology and Clinical Research, National Eye Institute, National Institutes of Health, Bldg 31, Room 6A52, 31 Center Dr, MSC 2510, Bethesda, MD 20892-2510(e-mail: jpsangio@nei.nih.gov).