Objectives. To analyze, in extremely low birth weight infants, associations between peak bilirubin concentration and evidence of brain damage, and between peak bilirubin concentration and blindness attributable to retinopathy of prematurity.
Methods. Retrospective study of 128 infants of ≤800 g birth weight and ≤27 weeks gestation born between 1980 and 1989 and discharged from a tertiary neonatal intensive care unit. After screening analyses, multivariable analyses were conducted to identify associations between blindness and peak bilirubin concentration (dichotomized at different levels to create 3 binary variables), and between severe adverse neurodevelopmental outcome at 18 months postterm age and peak bilirubin levels.
Results. Of 128 18-month survivors, 15 had severe visual loss attributable to retinopathy of prematurity, 21 had neurodevelopmental deficit, and 5 were deaf. Visual loss was significantly associated with low-peak serum bilirubin concentration (<9.4 mg/dL (<160 μmol/L) versus ≥9.4 mg/dL (odds ratio [OR] confidence interval [CI] 4.48 [1.15–17.43])), low gestational age (OR [CI] per week 1.95 [1.05–3.63]), and longer duration of phototherapy (OR [CI] per 10 hours 1.17 [1.02–1.33]). The association of neurodevelopmental impairment with grades 3 and 4 intraventricular hemorrhage was statistically significant (OR 5.39 [1.83–15.84]), but with high-peak serum bilirubin concentration ≥11.7 mg/dL (≥200 μmol/L), was not significant (OR 2.89 [0.87–9.53]).
Conclusions. In these infants, prolonged phototherapy and low-peak serum bilirubin concentrations were associated with severe visual loss attributable to retinopathy of prematurity. The findings should be interpreted with caution until the evidence is reinforced in other patient populations.
- extremely low birth weight infant
- retinopathy of prematurity
- partial pressure of oxygen
- intraventricular hemorrhage
- partial pressure of carbon dioxide
Low kernicterus rates at autopsy,1 and weak or no associations between peak serum bilirubin concentrations and neurodevelopmental impairment in survivors2–5 contribute to the perception that treated jaundice poses little risk to infants weighing <1500 g at birth.6 Also, in these infants, phototherapy appears to be effective and safe.2,,7,8
The risks of jaundice and of phototherapy have not been fully evaluated in extremely low birth weight (ELBW) infants (defined here as ≤800 g birth weight and ≤27 weeks gestation), because their survival is a recent phenomenon. Only 21 infants of ≤27 weeks gestation were included in the above trial of phototherapy.2,,8 Because of the availability and effectiveness2 of phototherapy, few opportunities may exist to observe bilirubin encephalopathy. On the other hand, phototherapy, by reducing body bilirubin and therefore endogenous antioxidant,9–11 may promote free radical mediated injury and increase the risk of oxygen radical disease.12
Motivated by the known neurotoxicity of bilirubin, and the hypothetical protective effect of bilirubin against oxidant injury, we analyzed associations between jaundice and outcome in a cohort of ELBW infants. The primary outcome variables were neurodevelopmental impairment and severe visual loss attributable to retinopathy of prematurity (ROP).
This retrospective study was conducted in a tertiary care center for outborn infants and included previously described ELBW infants admitted between 1980 and 1989.13 The additional data collected from the patient records for the present study were peak serum bilirubin concentration, total hours of phototherapy, and the number of pH measurements less than 7.25 during the period that the serum bilirubin concentration exceeded 5.9 mg/dL (100 μmol/L). According to our practice guidelines,14 serum bilirubin concentrations were to be measured every 6 to 8 hours. Peak serum bilirubin concentration was defined as the highest observed value, measured by the Jendrassik-Grof method to 1983, and since 1983, by a dry slide technique on a Kodak Echtachem Analyzer (Eastman Kodak, Rochester, NY) (the methods correlate well15).
Follow-up methods and definitions of impairment were described previously.13 Our neonatal follow-up clinic has used standardized procedures with prospective intent to perform outcome studies since 1975.16 Major neurodevelopmental impairment was defined as one or more of the following: Bayley corrected mental development index ≤68, spastic cerebral palsy of any limb(s) sufficient to delay walking beyond 18 months of corrected age, any degree of athetoid cerebral palsy, or severe unilateral or bilateral hearing loss partially corrected with hearing aids. Blindness attributable to ROP was defined as the presence of a condition known invariably to cause blindness such as total bilateral retinal detachment, or a diagnosis of blindness in the records of the multidisciplinary Neonatal Follow Up Clinic by 18 months adjusted age. The hospital charts of all children attending the Clinic who were diagnosed to be blind were retrieved and reviewed in detail by an ophthalmologist (P.M.). The charts contain records of the ophthalmology clinic (but not of ophthalmologists' private offices) and of hospitalizations for eye surgery. The staff of the follow-up and ophthalmology clinics were not masked to the neonatal diagnoses of the infants, but were not aware of their neonatal serum bilirubin values or of phototherapy exposures.
Before October 1985 we had no guidelines for phototherapy. Since then, the following guidelines (ranges) for initiating and maintaining phototherapy have been used for infants of birth weight ≤1000 g:14 <36 hours of age, 6.4 to 9.9; (110–170); 36 to 72 hours, 8.5 to 10.5; (145–180); 72 to 120 hours, 10.5 to 11.7; (180–200); and >120 hours, 11.1 to 12.3 mg/dL (190–210 μmol/L). White lights were used, and the infants were masked throughout. Phototherapy was administered continuously, and discontinued as indicated by reduced serum bilirubin concentration. No changes were made in illumination of the neonatal intensive care unit during the period of study, nor were isolettes covered. The guidelines for exchange transfusion were, at <72 hours of age, 11.7, (200) and at >72 hours, 12.6 mg/dL (215 μmol/L). Downward adjustment of the criteria was recommended for hemolytic disease or coexisting acidosis.
Most infants had umbilical artery catheters for blood gas monitoring for the first week after birth, and for those without catheters, transcutaneous partial pressure of oxygen (Po2) measurement was the preferred noninvasive method for monitoring oxygen therapy. Owing to reported associations between various parenteral nutrients and peroxide levels, especially with photoexposure,17–19 we reviewed our parenteral feeding practices during the study period. With one exception, enteral and parenteral feeding practices did not change systematically during the 1980s: in 1984 the parenteral multivitamin preparation was changed from MVI 1000 to MVI Pediatric, representing a reduction in the dose of riboflavin. Throughout the study period, the multivitamin preparation was added to the amino acid solution. Parenteral solutions in bottles and tubing were not protected from light. Parenteral feeding was usually initiated on day 2, but intravenous lipid was started between 2 to 5 days of age, as ordered by the staff neonatologist. Vitamin E was added routinely only before 1982. No preparations containing benzyl alcohol (possibly associated with kernicterus1) were used in our unit.
Surfactant therapy, introduced in October 1989, was given to 1 patient included in the study. Head ultrasound imaging was done frequently during the first few weeks after birth from 1980 onwards, but the device used until 1985 did not detect periventricular leukomalacia. Corticosteroids were rarely used for established lung disease in the 1980s. Cryotherapy, introduced in 1989, was performed on 1 patient in this cohort. Binary variables for peak bilirubin were defined arbitrarily at 3 bilirubin levels: <9.4 versus ≥9.4, <10.5 versus ≥10.5, and <11.7 versus ≥11.7 mg/dL (160, 180 and 200 μmol/L).
Infants with neurodevelopmental impairment were compared with those without impairment, and blind infants with nonblind infants. The demographic, obstetric, and neonatal resuscitation variables and measures of severity of illness at 48 hours of age listed in our previous article13 were analyzed, as well as the bilirubin and phototherapy data and the additional pH variable mentioned above. Risk factors for brain and retinal complications of prematurity were identified in the literature.3,20–23 The Kruskal-Wallis test compared continuous variables, and the χ2 or Fisher's exact test compared categorical variables. The association between duration of phototherapy and peak serum bilirubin concentration was examined by linear regression analysis, as was the association between the duration of oxygen therapy and of ventilation with blindness, duration of phototherapy and peak serum bilirubin.
Multiple Logistic Regression Analyses
Variables with P values <.10 in univariate analyses were entered into multiple logistic regression analyses together with the binary bilirubin variable that had the strongest association with the outcome in question in univariate analyses. A backward elimination procedure was used to further reduce the number of variables in the multiple logistic regression model to those with P values <.10. Analyses were then repeated entering interaction terms as well. A significant association was defined by a P value <.05 in multivariable analysis.
Of 287 infants of birth weight ≤800 g admitted during 1980 through 1989, 142 died by 2 years of age.13 Of the 145 infants discharged, 9 were lost to follow-up, 100 had no major impairment, 7 had gestational ages ≥28 weeks (excluded), and 36 had major impairment at 18 to 24 months postterm age. Of the 36 infants with major impairment, 21 had neurodevelopmental deficit (11 with severe developmental delay, 10 with severe cerebral diplegia; 1 of each also with blindness), 15 were blind (2 with neurodevelopmental deficit, and 2 with deafness), and 5 were deaf (3 with other major adverse outcomes). The 1 patient who received cryotherapy for ROP was later diagnosed to be blind. One infant whose file was lost was not included in the present study.
The blind infants had the following findings: all were 26 weeks gestation or less (4 infants were 22–23 weeks, 7 were 24 weeks, 3 were 25 weeks, and 1 was 26 weeks gestation). Many of the patients had some of their ambulatory ophthalmologic care delivered outside the hospital, hence only intermittent information on the ocular status of the child was available to the reviewer from the hospital chart. Because of the retrospective nature of this study and the variable follow-up it was not possible to state accurately the time of onset of the severe visual loss. Ophthalmologic sequelae in these children included the following: 8 eyes progressed to phthisis bulbi and 2 more manifested corneal opacification with no view of the intraocular contents. Eleven eyes had retinal detachments of which 7 eyes had associated cataract formation and 4 had clear lenses. Retinal distortion with macular heterotopia was apparent in 5 eyes. One eye with severe angle closure glaucoma resulting from ROP required enucleation for pain relief. One eye retained ambulatory vision after surgical reattachment. Finally 2 eyes (in different subjects) had reasonable but reduced vision in the setting of retinal distortion; one was 20/40 with −12.00 + 1.00 × 180° and the other was 20/70 with −4.25 + 2.25 × 90°. Six of the above eyes underwent retinal reattachment procedures with lensectomy and vitrectomy; 4 of these necessitated an open sky approach with a penetrating keratoplasty. One eye received cryotherapy that was followed by vitreous hemorrhage and subsequent progression to phthisis bulbi.
The mean (SD) gestational age was 24.5 (1.1) weeks and birth weight 696 (76) g. Fifty-seven percent of the infants were female, 19 (15%) were twins, and 20 (16%) were born by cesarean section. Bronchopulmonary dysplasia occurred in 118 infants (92%), the mean (SD) period of ventilatory support was 56 (31) days, and of supplemental oxygen therapy 130 (137) days. Grades 3 or 4 intraventricular hemorrhages (IVH) were found in 22 infants (17%). The mean peak serum bilirubin concentration was 9.4 (2.5) mg/dL (161  μmol/L), age of onset of phototherapy 30 (17) hours, and duration of phototherapy 86 (48) hours. The average number of serum bilirubin measurements in the 24 hours before plus the 24 hours after the peak measurement was 6.4 (mean 7.5 hours between measurements). All but 4 infants received phototherapy (none of the 4 had major adverse outcomes). Ten infants had exchange transfusions for jaundice, mostly on days 3 to 4. One of the infants who received exchange transfusion had neurosensory deafness, another had blindness attributable to ROP.
Associations With Adverse Outcomes
Variables with P values <0.10 for at least one of the outcomes are listed in Tables 1 and 2. Peak serum bilirubin concentration <9.4 versus ≥9.4 mg/dL (160 μmol/L) was associated with blindness, and peak serum bilirubin concentration ≥11.7 versus <11.7 mg/dL (200 μmol/L) with neurodevelopmental impairment, both with Pvalues <.1 (Tables 1 and 2). IVH grades 3 and 4 were also associated with both adverse outcomes with P values <.1 (Table 1). The duration of phototherapy increased with increase in peak serum bilirubin (P = .0005). No association was found between blindness or adverse neurodevelopmental outcome with high or low Po2 or partial pressure of carbon dioxide (Pco2) values at 8 and 48 hours of age. Nor was an association found between blindness, duration of phototherapy, and peak bilirubin on the one hand and duration of ventilatory support or of oxygen therapy (measures of bronchopulmonary dysplasia) on the other. The 2 infants with deafness as a sole major adverse outcome had peak serum bilirubin concentrations of 12.4 and 11.3 mg/dL (213 and 198 μmol/L). No child developed athetoid or mixed cerebral palsy.
Multiple Logistic Regression Analyses
Blindness Attributable to ROP
Blindness was significantly associated with peak bilirubin <9.4 mg/dL (160 μmol/L) (OR [CI] 4.48 [1.15–17.43]), duration of phototherapy (OR per 10 hours [CI] 1.17 [1.02–1.33]), and gestational age (OR per week [CI] 1.95 [1.05–3.63]) (Table 2). A minimum blood glucose concentration ≥2.2 mmol/L in the first 48 hours after birth, and IVH (grades 3 and 4) were not statistically significantly associated with blindness (P = .134 and .350, respectively). The interaction term peak bilirubin-duration of phototherapy was not statistically significantly associated with blindness.
This outcome was significantly associated with IVH grades 3 and 4 (OR [CI] 5.39 [1.83–15.84])(Table 3). For the peak serum bilirubin concentration ≥11.7 mg/dL (200 μmol/L) variable, the OR [CI] was 2.89 [0.87–9.53], and the OR [CI] for the number of low pH measurements (per low pH measure) was 1.11 [0.99–1.25]. The interaction term peak bilirubin-IVH was not statistically significant.
The pivotal findings in this retrospective study are that blind infants tended to have low maximum serum bilirubin concentrations (defined as <160 μmol/L, <9.4 dl/dL)(OR 4.48 [1.15–17.43]), and long exposures to phototherapy (OR 1.17 [1.02–1.33]). Multivariate statistical analysis reinforced these findings by controlling for the most important and well-recognized covariate (gestational age),23 and provided evidence that low peak bilirubin and prolonged phototherapy may be associated with blindness, independently of one another. These findings suggest that infants with severe visual loss tended to receive more hours of phototherapy than was needed to achieve target serum bilirubin concentrations. In general however, infants with high peak serum bilirubin concentration (defined as ≥200 μmol/L, ≥11.7 dl/dL) had long exposures to phototherapy however, suggesting that phototherapy was used appropriately in most infants. We did not study other reported associations of blindness attributable to ROP such as hyperoxia, hypoxia, hypercarbia, and hypocarbia adequately, as only one time point was analyzed. Severe IVH also was not associated with blindness, particularly in multivariate analysis that controlled for peak bilirubin and duration of phototherapy (Table 2). In addition, severity of bronchopulmonary dysplasia (as measured by days of mechanical ventilation and days of oxygen therapy), and number of episodes of sepsis were not associated with blindness.
With regard to neurodevelopmental impairment, most studies have reported a paucity of associations (including gestational age in the range analyzed here).13,,20 As in other studies, severe IVH stands out13,,20 (Table 3), but it represents an intermediate outcome rather than a cause. Absence of association between neurodevelopmental impairment and both acidosis and hyperbilirubinemia (Table 3) may be explained by insufficient power, particularly as cases with cerebral palsy and severe developmental delay were not analyzed separately. Further investigation of acidosis and hyperbilirubinemia is warranted.
We analyzed blindness rather than ROP as the outcome variable because some of our patients were transferred to secondary level neonatal units before their maximal grade of ROP was reached, resulting in missing ROP data. By contrast, our Neonatal Follow-up Clinic, which followed all admitted patients regardless of their length of neonatal stay, had a low attrition rate.15
ROP is described as an oxygen radical disease.12 Premature infants have fewer intracellular defense mechanisms against oxygen radicals than term infants.24 The degree of immaturity has been reported to be the most important factor explaining free radical-mediated lipid peroxidation.12 Oxygen radical disease may be mitigated by natural antioxidants, a large proportion of which in neonates resides in bilirubin.10,,11 Thus, by promoting bilirubin excretion, phototherapy may reduce antioxidant activity. Two studies of the relationship between ROP and serum bilirubin concentration produced conflicting results, but both had limited sample sizes. An alternative explanation of the observed association between low serum bilirubin levels and blindness is that oxygen radical mediated disease may represent an effect rather than a cause of oxidant injury.25 We did not analyze the rate of rise of bilirubin postnatally in our cohort owing to the initiation of phototherapy in some cases after a single serum bilirubin measurement.
Associations between phototherapy and blindness may be explained also by direct effects of ambient light on the unshielded immature eye,26 or by photo-oxidation of the lipid and vitamin content of parenteral solutions in transparent administration sets,17–19 producing peroxides toxic to the retina. Light may have similar effects on plasma circulating in cutaneous blood vessels.
The above lines of evidence in conjunction with the results of this study raise questions about the safety of phototherapy in ELBW infants. Empirical studies of the relationship between ROP with both serum bilirubin concentration27,,28 and light exposure29–32 produced conflicting results. However, based on the recent report of a prospective randomized multicenter study of light reduction to the eyes only, it seems unlikely that blindness results from the direct effect of ambient nonphototherapy light falling on the retina.32 Conclusions cannot be drawn about the safety of phototherapy on the basis of studies of eye shielding alone. Analyses of phototherapy exposures, serum bilirubin concentrations, or serum oxidant activity were not reported, and the number of extremely immature infants at high risk for blindness attributable to ROP was relatively small.
In this study, neurodevelopmental impairment was not significantly associated with high peak serum bilirubin concentration. The neurotoxicity of bilirubin is well-established, but measures of risk are uncertain, especially in ELBW infants. Premature infants are probably more susceptible to kernicterus than term infants.33 Soon after the introduction of intensive care in the 1960s, low-bilirubin kernicterus (autopsy-diagnosed kernicterus in low birth weight infants with low-peak serum bilirubin concentrations) was reported, but attributed to severe illness.34 Reports of associations between periventricular leukomalacia (a major cause of cerebral diplegia, the typical cerebral palsy in surviving preterm infants) and high serum bilirubin concentrations have been mixed, but a trend appears to exist.3,,21,22 We did not collect and analyze data on this intermediate outcome, owing to small numbers (it has been diagnosable by ultrasound only since 1985). In infants <32 weeks gestation and/or <1500 g birth weight, Van de Bor et al5 reported associations between peak serum bilirubin concentration and risk of adverse outcome at 2 years of age, which was not significant at 5 years of age. Other studies have not found associations between serum bilirubin concentration and neurotoxicity in low birth weight survivors3,,4,35 and at autopsy,1,,36 possibly explained by design limitations. Recently, during an era of high utilization of phototherapy and few exchange transfusions, the incidence rate of kernicterus at autopsy was 4%.1 The results of studies that found no association between peak bilirubin and neurodevelopmental impairment3or kernicterus in preterm infants may reflect the effective control of jaundice, especially since the introduction of phototherapy, rather than its innocuousness. The above observations may not be applicable to the extremely immature infants reported here, few of which survived in the past.
Observational studies can generate hypotheses and aid in designing clinical trials. Our findings raise three therapeutic questions in ELBWs: should parental nutrition fluids be shielded from light, what is the importance of shielding the eyes from light, and should exposure of the infant to phototherapy be limited in some way? If our observations are reinforced by others, clinical trials to answer these questions might be considered.
Based on observed oxidation of nutrients in parenteral solutions exposed to light,17–19 reports of associations between measures of lipid peroxidation in infant serum and death or bronchopulmonary dysplasia,18,,37 and the likelihood that shielding of parenteral solutions is safe, it may be difficult to justify a clinical trial of this intervention. Particular caution may be advisable with exposure to light of mixtures of multivitamin preparation with intravenous lipid solution, which together may increase peroxide formation. Although our data did not address the question of eye-shielding, the results do suggest that this should be done meticulously, which is not practiced universally.38 In view of the possibility that too much phototherapy is harmful to infants of birth weight ≤800 g and ≤27 weeks gestation, a therapeutic trial comparing liberal with conservative phototherapy criteria in this population might be considered. Our criteria (see under “Methods”) were relatively conservative.38 A clinical trial would need to assign weights to the disparate outcomes (neurodevelopmental impairment and blindness), balance the risks of these adverse outcomes, and plan to avoid both. To avoid both potential adverse outcomes, a relatively very large sample size would be needed. It may not be feasible to perform such a trial for logistic reasons. A prospective multicenter observational study may provide the additional data needed to construct practice guidelines for phototherapy for tiny infants.
- Received November 4, 1997.
- Accepted June 12, 1998.
Reprint requests to (M.P.) Division of Neonatology, Department of Paediatrics, Hospital for Sick Children, 555 University Avenue, Toronto, Ontario, M5G 1X8, Canada.
- ELBW =
- extremely low birth weight •
- ROP =
- retinopathy of prematurity
- Watchko JF,
- Claasen D
- Brown AK,
- Kim MH,
- Wu PYK,
- Bryla DA
- Graziani LJ,
- Mitchell DG,
- Kornhauser M,
- et al.
- O'Shea TM,
- Dillard RG,
- Klinepeter KL,
- Goldstein DJ
- Van de Bor M,
- Ens-Dokkum M,
- Schreuder AM,
- Veen S,
- Brand R,
- Verloove-Vanhorick SP
- Watchko JF,
- Oski FA
- Lipsitz PJ,
- Gartner LM,
- Bryla DA
- Bryla DA
- Stocker R,
- Yamamoto Y,
- McDonagh AF,
- Glazer AN,
- Ames BN
- ↵Perlman M, Kirpalani H, Schmidt B, James A. Guidelines for Neonatal Treatment. Toronto, Ontario, Canada: The Hospital for Sick Children; 1985
- Wu TW,
- Dappen GM,
- Spayd RW,
- Sundberg MW,
- Powers DM
- ↵Fitzhardinge PM. Follow-up studies of the high-risk newborn. In: Avery GB, ed. Neonatology. Pathophysiology and Management of the Newborn. 3rd ed. Philadelphia, PA: Lippincott; 1987:400–417
- ↵Lavoie J-C, Bélanger S, Spalinger M, Chessex P. Admixture of a multivitamin preparation to parenteral nutrition: the major contributor to in vitro generation of peroxides. Pediatrics. 1997;99(3). URL: http://www.pediatrics.org/cgi/content/full/99/3/e6
- Yu VYH,
- Downe L,
- Astbury J,
- Bajuk B
- Trounce JQ,
- Shaw DE,
- Levene MI,
- Rutter N
- ↵Watts JL. Retinopathy of Prematurity. In: Sinclair JC, Bracken MB, eds. Effective Care of the Newborn Infant. Oxford: Oxford University Press; 1992:617–638
- ↵Sadda Yu YS Sr, de Juan E Jr, Rencs EV, Green WR, Gottsch JD. Photosensitization-induced retinopathy in the newborn beagle. Invest Ophthalmol Vis Sci. 1994;35:1202–1211
- Ackerman B,
- Sherwonit E,
- Williams J
- Lopes JM,
- Braz RRT,
- Moreira MEEL,
- et al.
- Reynolds JD,
- Hardy RJ,
- Kennedy KA,
- Spencer R,
- van Heuven WAJ,
- Fielder AR,
- for the Light Reduction in Retinopathy of Prematurity (LIGHT-ROP) Cooperative Group
- Ahlfors CE
- Gartner LM,
- Snyder RN,
- Chabon RS,
- et al.
- Turkel SB,
- Guttenberg ME,
- Moynes DR,
- Hodgman JE
- Turkel SB,
- Miller CA,
- Guttenberg ME,
- Moynes DR,
- Hodgman JE
- Hansen TWR
- Copyright © 1998 American Academy of Pediatrics