Prophylactic phototherapy for preventing jaundice in preterm or low birth weight infants (Review)

Transcription

1 EVIDENCE-BASED CHILD HEALTH: A COCHRANE REVIEW JOURNAL Evid.-Based Child Health 8:1: (2013) Published online in Wiley Online Library (onlinelibrary.wiley.com). DOI: /ebch.1898 Prophylactic phototherapy for preventing jaundice in preterm or low birth weight infants (Review) Okwundu CI, Okoromah CAN, Shah PS This is a reprint of a Cochrane review, prepared and maintained by The Cochrane Collaboration and published in The Cochrane Library 2012, Issue 1

2 TABLE OF CONTENTS HEADER ABSTRACT PLAIN LANGUAGESUMMARY SUMMARY OFFINDINGS FORTHE MAIN COMPARISON BACKGROUND OBJECTIVES METHODS RESULTS Figure Figure Figure Figure Figure Figure Figure Figure DISCUSSION AUTHORS CONCLUSIONS ACKNOWLEDGEMENTS REFERENCES CHARACTERISTICS OFSTUDIES DATA AND ANALYSES Analysis 1.1. Comparison 1 Prophylactic phototherapy versus control, Outcome 1 Exchange transfusion Analysis 1.2. Comparison 1 Prophylactic phototherapy versus control, Outcome 2 Neurodevelopmental impairment. 237 Analysis 1.3. Comparison 1 Prophylactic phototherapy versus control, Outcome 3 Cerebral palsy Analysis 1.4. Comparison 1 Prophylactic phototherapy versus control, Outcome 4 Peak serum bilirubin Analysis 1.5. Comparison 1 Prophylactic phototherapy versus control, Outcome 5 Sensorineural hearing loss Analysis 1.6. Comparison 1 Prophylactic phototherapy versus control, Outcome 6 Peak serum bilirubin (by time of intervention) Analysis 1.7. Comparison 1 Prophylactic phototherapy versus control, Outcome 7 Serum bilirubin greater than 10 mg/dl Analysis 1.8. Comparison 1 Prophylactic phototherapy versus control, Outcome 8 Serum bilirubin greater than 15 mg/dl Analysis 1.9. Comparison 1 Prophylactic phototherapy versus control, Outcome 9 Duration of phototherapy (hours). 244 Analysis Comparison 1 Prophylactic phototherapy versus control, Outcome 10 Patent ductus arteriosus Analysis Comparison 1 Prophylactic phototherapy versus control, Outcome 11 Retinopathy of prematurity Analysis Comparison 1 Prophylactic phototherapy versus control, Outcome 12 Intraventricular haemorrhage Analysis Comparison 1 Prophylactic phototherapy versus control, Outcome 13 Duration of hospital stay (days). 247 Analysis Comparison 1 Prophylactic phototherapy versus control, Outcome 14 All-cause mortality HISTORY CONTRIBUTIONS OFAUTHORS DECLARATIONS OFINTEREST SOURCESOFSUPPORT DIFFERENCES BETWEEN PROTOCOL AND REVIEW INDEX TERMS

3 [Intervention Review] Prophylactic phototherapy for preventing jaundice in preterm or low birth weight infants Charles I Okwundu 1, Christy AN Okoromah 2, Prakeshkumar S Shah 3 1 Faculty of Health Sciences, University of Stellenbosch, Cape Town, South Africa. 2 Department of Paediatrics and Child Health, College of Medicine, University of Lagos, Lagos, Nigeria. 3 Department of Paediatrics and Department of Health Policy, Management and Evaluation, Rm 775A, University of Toronto, Toronto, Canada Contact address: Charles I Okwundu, Faculty of Health Sciences, University of Stellenbosch, Cape Town, Western Cape, 7500, South Africa. ciokwundu@sun.ac.za. Editorial group: Cochrane Neonatal Group. Publication status and date: New, published in Issue 1, Review content assessed as up-to-date: 31 March Citation: Okwundu CI, Okoromah CAN, Shah PS. Prophylactic phototherapyfor preventing jaundice in pretermor lowbirth weight infants. Cochrane Database of Systematic Reviews 2012, Issue 1. Art. No.: CD DOI: / CD pub2. Background ABSTRACT Low birth weight and premature infants are at major risk for exaggerated hyperbilirubinaemia and jaundice that can lead to bilirubin encephalopathy. Phototherapy is the most common treatment for neonatal hyperbilirubinaemia and could be most effective in preventing the sequelae of hyperbilirubinaemia if initiated prophylactically. Objectives To evaluate the efficacy and safety of prophylactic phototherapy for preterm (< 37 weeks gestational age) or low birth weight infants (birth weight < 2500 g). Search methods We searched the Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library, Issue 3) on 31 March 2011, MEDLINE (1950 to 31 March 2011), EMBASE (1980 to 31 March 2011) and CINAHL (1982 to 31 March 2011). Selection criteria Randomised controlled trials or quasi-randomised controlled studies evaluating the effects of prophylactic phototherapy for preterm or low birth weight infants. Data collection and analysis Two authors independently obtained data from published articles. We performed fixed-effect meta-analysis for the outcomes: rate of exchange transfusion, cerebral palsy or other neurodevelopmental impairment, peak serum bilirubin level and all-cause mortality. Main results Nine studies of 3449 participants were included. The rate of exchange transfusion was reduced in one study with liberal transfusion criteria (risk ratio (RR) 0.20; 95% confidence interval (CI) 0.13 to 0.31) but not in the other two more recent studies with stringent criteria (typical RR 0.66; 95% CI 0.19 to 2.28). There was no statistically significant difference in the rate of cerebral palsy (typical RR 0.96; 95% CI 0.50 to 1.85; two studies, 756 participants). However, one large study that reported on neurodevelopmental impairment 206

4 (a composite outcome including cerebral palsy) found a slightly lower rate of neurodevelopmental impairment with prophylactic phototherapy (RR 0.85; 95% CI 0.74 to 0.99; 1804 participants). The prophylactic phototherapy group had lower peak bilirubin levels (mean difference (MD) -2.73; 95% CI to -2.57; six studies, 2319 participants) and had fewer neonates with peak unconjugated serum bilirubin levels > 10 mg/dl (typical RR 0.27; 95% CI 0.22 to 0.33; three studies, 1090 participants) or peak unconjugated serum bilirubin levels > 15 mg/dl (typical RR 0.13; 95% CI 0.07 to 0.23; four studies, 1116 participants). There was no statistically significant difference in the rate of all-cause mortality between the two groups (typical RR 1.08; 95% CI 0.93 to 1.26; four studies, 3044 participants). Authors conclusions Prophylactic phototherapy helps to maintain a lower serum bilirubin concentration and may have an effect on the rate of exchange transfusion and the risk of neurodevelopmental impairment. However, furtherwell-designed studies are neededto determine the efficacy and safety of prophylactic phototherapy on long-term outcomes including neurodevelopmental outcomes. PLAIN LANGUAGE SUMMARY Prophylactic phototherapy for preventing jaundice in preterm very low birth weight infants Preterm (< 37 weeks gestational age) or low birth weight (LBW; birth weight < 2500 g) infants have a greater risk of developing jaundice compared to term or normal birth weight infant. This can be concerning because jaundice (caused by high levels of serum unconjugated bilirubin) could lead to permanent brain damage and/or death. In this review we evaluated the efficacy and safety of prophylactic phototherapy in preventing jaundice in preterm or LBW infants. A total of nine clinical trials representing 3449 infants were included. The findings suggest that phototherapy initiated soon after birth (within 36 hours) for preterm or low birth weight infants may prevent the serum bilirubin from reaching a level that would require exchange transfusion and may reduce the risk of impairment of brain and central nervous system development. However, further welldesigned studies are needed to evaluate theeffects of prophylactic phototherapy on brain and central nervous system development and other long-term outcomes. 207

5 SUMMARY OF FINDINGS FOR THE MAIN COMPARISON [Explanation] Prophylactic phototherapy for jaundice in preterm or low birth weight infants Patient or population: preterm or low birth weight infants Settings: neonatal units Intervention: prophylactic phototherapy Control: standard care Outcomes Illustrative comparative risks* (95% CI) Relative effect (95% CI) No of participants (studies) Assumed risk Corresponding risk Control Prophylactic phototherapy Exchange transfusion Medium-risk population RR 0.22 (0.15 to 0.34) 120 per per 1000 (18 to 41) 2946 (3 studies) Cerebral palsy Medium-risk population RR 0.96 (0.5 to 1.85) 84 per per 1000 (42 to 155) 756 (2 studies) Peak serum bilirubin The mean peak serum bilirubin (all infants) in the intervention groups was 2.73 lower (2.89 to 2.57 lower) 2319 (6 studies) Serum bilirubin greater than 10 mg/dl Medium-risk population RR 0.27 (0.22 to 0.33) 600 per per 1000 (132 to 198) 1090 (3 studies) Quality of the evidence (GRADE) moderate 1 moderate 1 low 2,3 moderate 4 Comments 208

6 high 1687 (2 studies) Medium-risk population RR 0.93 (0.77 to 1.13) Retinopathy of prematurity 225 per per 1000 (173 to 254) moderate (4 studies) All-cause mortality Medium-risk population RR 1.08 (0.93 to 1.26) 157 per per 1000 (146 to 198) *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; RR: Risk ratio GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate. 1 There was no mention of blinding of the outcome assessors in two of the studies. 2 In one of the studies participants were allocated alternately. The method of randomisation was unclear in two of the studies. 3 There is marked heterogeneity in the estimates of effect across the studies. 4 There was unclear allocation concealment in two of the studies. 5 There was unclear sequence generation and allocation concealment in two of the studies. 209

7 BACKGROUND Description of the condition Neonatal jaundice is the yellow discolouration of the skin and sclera of neonates caused by hyperbilirubinaemia. Bilirubin is a yellow pigment that is produced in the body during the normal recycling of aging red blood cells. About 50% of term and 80% of preterm infants develop jaundice in the first week of life (Kumar 1999). Jaundice is a common cause of readmission to hospital after early discharge of newborn infants (Gale 2001). The most common form of hyperbilirubinaemia observed in neonates is due to unconjugated or unbound bilirubin. High levels of unconjugated bilirubin are managed by treatment of the underlying cause (e.g. dehydration and sepsis), provision of phototherapy, administration of immunoglobulin (in certain cases of haemolytic anaemia) and, in very severe cases, exchange transfusion. If left untreated, extreme elevation of unconjugated bilirubin can lead to permanent brain damage and/or death (Hansen 2005). Several risk factors have been identified for the occurrence of exaggerated or severe unconjugated hyperbilirubinaemia. Low birth weight and preterm birth are major risk factors for exaggerated hyperbilirubinaemia. Clinically significant levels that warrant treatment occur in approximately 50% to 80% of preterm neonates (Almeida 2004). Preterm infants are at risk of bilirubin encephalopathy at lower total serum bilirubin levels than mature infants. High levels of unconjugated serum bilirubin can cross the blood brain barrier and can be deposited in the basal ganglia of the brain resulting in a condition termed kernicterus. Clinically, the term kernicterus is used to describe the chronic form of bilirubin encephalopathy and includes symptoms such as athetoid cerebral palsy (uncontrollable movement of the face, body, arms and legs), hearing loss, failure of upward gaze and dental enamel dysplasia. The exact level of unconjugated serum bilirubin that is neurotoxic is unclear. There is marked variability in the level of bilirubin at which kernicterus will occur. Kernicterus has been reported at autopsy in infants in the absence of markedly elevated levels of unconjugated serum bilirubin (Turkel 1980). Description of the intervention Phototherapy is the most common treatment for unconjugated hyperbilirubinaemia in neonates. Phototherapy leads to photoisomerisation of bilirubin into a water-soluble form that can be excreted by the kidney. Phototherapy effectively decreases the unconjugated bilirubin and decreases the need for exchange transfusion (Maisels 1998). Generally, phototherapy is considered a safe intervention. However, itisassociated with complications such as bronze baby syndrome if used in patients with conjugated hyperbilirubinaemia and it may result in dehydration due to insensible water loss. It may also cause damage to developing retina (Rubaltelli 1983). The major counter argument to the use of prophylactic phototherapy is that bilirubin is a powerful antioxidant (McDonagh 1990) and may have a physiological role in neonates (Hegyi 1994). It has also been suggested that the low levels of bilirubin achieved by the use of prophylactic phototherapy reduce the antioxidant level and may lead to the development of retinopathy of prematurity. Several other neonatal outcomes such as chronic lung disease and neurological injury (intraventricular haemorrhage, periventricular leukomalacia) could be affected by lack of antioxidant activity. In addition, alteration of fluid homeostasis associated with the use of phototherapy may affect closure of a patent ductus arteriosus. How the intervention might work Unconjugated hyperbilirubinaemia is very common in preterm and low birth weight neonates and because there is no way to predict a safe level of bilirubin in very low birth weight (VLBW) infants, prophylactic use of phototherapy immediately after birth has been suggested. It has been hypothesised that phototherapy could be used to prevent unconjugated hyperbilirubinaemia and clinical jaundice, thereby leading to improved neurodevelopmental outcomes. Why it is important to do this review Treatment of neonatal hyperbilirubinaemia is usually based on the measurement ofthetotalserumbilirubin levels (Bratlid 2001)and the serum bilirubin level used as an indication for treatment varies across healthcare institutions, from country to country and across continents. The use of prophylactic phototherapy in preterm or low birth weight infants is not an uncommon practice (Hansen 1996). However, there is currently no evidence to support the use of phototherapy for the prevention of jaundice caused by unconjugated hyperbilirubinaemia. OBJECTIVES To evaluate the efficacy and safety of prophylactic phototherapy in preterm or low birth weight infants. METHODS Criteria for considering studies for this review Types of studies 210

8 Randomised or quasi-randomised controlled studies that have evaluated the effects of prophylactic phototherapy in preterm and/ or low birth weight infants. Types of participants Preterm infants (< 37 weeks gestation). Low birth weight infants (< 2500 grams), within the first 36 hours of birth. Originally (in the protocol) the focus of the review was narrower (to include very low birth weight infants; < 1500 g birth weight), however, in order not to lose valuable information, we made a post hoc decision to include any study that involved low birth weight (< 2500 g birth weight) or preterm infants. We excluded studies of infants with a known cause that can lead to significant hyperbilirubinaemia, such as ABO incompatibility, Rh incompatibility, minor blood group incompatibility or G-6PD deficiency. Patent ductus arteriosus (PDA) (clinical diagnosis) or PDA requiring treatment with cyclo-oxygenase inhibitor or surgery. Retinopathy of prematurity (ROP) (any stage and severity (stages 3 or 4) defined according to the International Classification of Retinopathy of Prematurity). Intraventricular haemorrhage (IVH) (any grade and severity (grade 3 or 4) as defined by Papile) (Papile 1978). Periventricular leukomalacia (PVL) (diagnosed by ultrasound or magnetic resonance imaging (MRI)). Duration of hospital stay (days). Mortality during neonatal intensive care unit (NICU) stay: kernicterus-related mortality; all-cause mortality. Types of interventions We included studies in which prophylactic phototherapy was compared with control. We defined prophylactic phototherapy as initiation of phototherapy before the bilirubin has reached a prespecified level (according to the study criteria) at which therapeutic use of phototherapy is indicated. The control refers to the standard treatment whereby phototherapy is started when the serum bilirubin has reached to a pre-specified level (according to the study criteria) or no use of prophylactic phototherapy. Types of outcome measures Primary outcomes Clinical kernicterus. Pathological kernicterus. Need for exchange blood transfusion. Long-term outcomes assessed at any age beyond one year of age by a validated cognitive, motor, language or behavioural/ school/social interaction/adaptation test. Search methods for identification of studies See: Cochrane Neonatal Group methods used in reviews. We formulated a comprehensive and exhaustive search strategy in order to identify all relevant studies, regardless of language or publication status (published, unpublished, in press and in progress). Electronic searches We searched the following electronic databases until 31 March The Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library, Issue 3). 2. MEDLINE (1966 to 31 March 2011). 3. EMBASE (from 1980 to 31 March 2011). 4. CINAHL (from 1982 to 31 March 2011). The search strategy included the search terms Prophylactic AND phototherapy AND prevention AND jaundice OR hyperbilirubinemia OR hyperbilirubinaemia preterm OR premature AND very low birth weight OR VLBW and infants OR neonates. There were no language restrictions in the searches. Secondary outcomes Highest serum bilirubin level during first seven days of life (mg/dl). Duration of phototherapy in hours. Fluid intake/requirement (cc/kg/day) in the first week after birth. Bronze baby syndrome. Searching other resources We performed handsearches of the reference lists of all pertinent reviews and studies. We contacted the National Institute of Child Health and Human Development (NICHD) experts in the field for unpublished and any ongoing studies. We also searched for any ongoing or prospective studies in registries using the websites of and 211

9 Data collection and analysis Selection of studies See Figure 1 for details of the study selection process. Two review authors (CO and CAN) independently read the titles, abstracts and descriptor terms of the search output fromthe different databases to identify potentially eligible studies. We obtained fulltext articles for all citations identified as potentially eligible. Both review authors (CO and CAN) independently inspected these to establish the relevance of the articles according to the prespecified criteria. We reviewed studies for relevance based on study design, types of participants, interventions and outcome measures. We gave reasons for excluding potentially relevant studies in an excluded studies table. We resolved discrepancies by consensus and involving third author (PS). Data extraction and management We designed a data extraction form. CO and CAN independently extracted data into the data extraction form individually. We extracted the following characteristics from each included study. Administrative details: author(s); published or unpublished; year of publication; year in which study was conducted; details of other relevant papers cited. Details of the study: study design; type, duration and completeness of follow-up; country and location of study informed consent and ethics approval. Details of participants: age, birth weight and number of participants. Details of intervention: age when phototherapy was commenced in both the intervention and control arm, duration of phototherapy, serum bilirubin level at which phototherapy was commenced or stopped, wavelength and light intensity. Details of outcomes: peak serum bilirubin, mean serum bilirubin, exchange blood transfusions, duration of hospital stay, duration of phototherapy, long-term neurodevelopmental outcomes, deaths and any adverse events. Assessment of risk of bias in included studies Two review authors independently used the Cochrane Collaboration tool for assessing the risk of bias for each individual study. For trials, the Cochrane tool assesses risk of bias in individual studies across six domains: sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting and other potential biases. We judged the risk of bias for the different domains in each included study as high risk of bias, low riskofbias or unclearriskofbias. Sequence generation Low risk: investigators described a random component in the sequence generation process such as the use of a random number table, coin tossing, cards or envelope shuffling, etc. High risk: investigators described a non-random component in the sequence generation process such as the use of odd or even date of birth, algorithm based on the day/date of birth, hospital or clinic record number. Unclear risk: insufficient information to permit judgement of the sequence generation process. Allocation concealment Low risk: participants and the investigators enrolling participants cannot foresee assignment (e.g. central allocation; or sequentially numbered, opaque, sealed envelopes). High risk: participants and investigators enrolling participants can foresee upcoming assignment (e.g. an open random allocation schedule (e.g. a list of random numbers); or envelopes were unsealed or nonopaque or not sequentially numbered). Unclear risk: insufficient information to permit judgement of the allocation concealment or the method not described. Blinding Low risk: blinding of the participants, key study personnel and outcome assessor, and unlikely that the blinding could have been broken, or lack of blinding unlikely to introduce bias. No blinding in the situation where non-blinding is not likely to introduce bias. High risk: no blinding, incomplete blinding and the outcome is likely to be influenced by lack of blinding. Unclear risk: insufficient information to permit judgement of the adequacy or otherwise of the blinding. Incomplete outcome data Low risk: no missing outcome data, reasons for missing outcome data unlikely to be related to true outcome, or missing outcome data balanced in number across groups. High risk: reason for missing outcome data likely to be related to true outcome, with either imbalance in number across groups or reasons for missing data. Unclear risk: insufficient reporting of attrition or exclusions. 212

10 Selective reporting Low risk: a protocol is available which clearly states the primary outcome as the same as in the final trial report. High risk: the primary outcome differs between the protocol and final trial report. Unclear risk: no trial protocol is available or there is insufficient reporting to determine if selective reporting is present. Other forms of bias Low risk: there is no evidence of bias from other sources. High risk: there is potential bias present from other sources (e.g. early stopping of trial, fraudulent activity, extreme baseline imbalance or bias related to specific study design). Unclear risk: insufficient information to permit judgement of the adequacy or otherwise of other forms of bias. Measures of treatment effect We conducted measures of treatment effect data analysis using Review Manager (RevMan) version (RevMan 2008). We calculated outcome measures for dichotomous data (e.g. exchange transfusion, death, peak serum bilirubin > 10 mg/dl) as a risk ratio with 95% confidence intervals. We calculated continuous data (e.g. peak serum bilirubin concentration, duration of phototherapy, duration of hospital stay) using the mean difference and standard deviations. Assessment of heterogeneity Where studies were found to be methodologically or clinically comparable, we pooled trial results in a meta-analysis. As we anticipated the presence of statistical heterogeneity, we combined data using the random-effects model. We calculated whether statistical heterogeneity is present using the Chi² test for homogeneity (P < 0.1 was considered significant). We quantified the impact of statistical heterogeneity using the I² statistic, which describes the percentage of total variation across studies due to heterogeneity rather than sampling error (Higgins 2003). We explored possible causes of significant heterogeneity by looking critically at the characteristics of the various studies. Data synthesis We summarised all eligible studies in RevMan 5. The two review authors extracted the data and entered all data into the current version of RevMan 5. Both review authors rechecked all of the entries. We conducted meta-analysis where appropriate and if the studies were found to be clinically homogenous. For comparable studies, we summarised the findings using a random-effects model. We calculated summary estimates, when applicable: risk ratio (RR) and risk difference (RD) for dichotomous outcome measures and mean difference (MD) for continuous outcomes, with their respective 95% confidence intervals (CI). If an intervention was beneficial or harmful, we calculated number needed to treat to benefit (NNTB) or harm (NNTH). Subgroup analysis and investigation of heterogeneity We carried out subgroup analysis where appropriate. For one of the outcomes, we stratified participants according to birth weight (very low birth weight and extremely low birth weight infants). We also performed a sensitivity analysis to explore the effect of trial quality on the results by excluding those studies with high risk of bias and assessed the effect of this on the overall results. RESULTS Description of studies See: Characteristics of included studies; Characteristics of excluded studies. See: Characteristics of included studies; Characteristics of excluded studies. Results of the search See: Prisma flow diagram Figure 1 for details of the search output from each of the databases. We identified nine studies that met our inclusion criteria. These studies included 3449 participants with a median trial sample size of 96 (ranging from 22 (Curtis 1985) to 1974 participants (Morris 2008). 213

11 Figure 1. Study flow diagram. Included studies A total of nine studies were included in this review (Lucey 1968; Giunta 1969; Hodgman 1970; Brown 1985; Curtis 1985; Leite 2004; Tripathi 2006; Jangaard 2007; Morris 2008). We have included details for each trial in the table Characteristics of included studies. Six of the studies were conducted in the USA (Lucey 1968; Giunta 1969; Hodgman 1970; Brown 1985; Curtis 1985; Morris 2008), while the others were conducted in Brazil (Leite 2004), Canada (Jangaard 2007) andindia(tripathi 2006). The most recent trial was conducted between 2002 and 2007 (Morris 2008), while the oldest trial was conducted before 1967 (Lucey 1968). Four studies (Lucey 1968; Giunta 1969; Hodgman 1970; Brown 1985) were conducted in the late 1960s and early 1970s. There were two multicentre studies (Brown 1985; Morris 2008) and these had the highest number of study participants (922 and 1974 respectively). Types of participants All the included studies evaluated the efficacy of phototherapy in preventing jaundice in preterm (< 37 weeks gestational age) low birth weight infants (< 2500 g). The upper limit of birth weight of the study participants differed in the different studies (< 2500 gforgiunta 1969 and Lucey 1968, < 2000 g for Brown 1985; Leite 2004; andhodgman 1970, < 1500 g for Jangaard 2007, < 1250 g for Tripathi 2006 and Curtis 1985, and < 1000 g (Morris 2008). The mean gestational ages of the study participants were as follows: 27.3 weeks for Curtis 1985, 251 and 254 days (phototherapy and control group respectively) for Giunta 1969, 34.4 and 34.3 weeks (phototherapy and control group respectively) for Hodgman 1970, 28.1 and 28.7 weeks (phototherapy and control group respectively) for Jangaard 2007, < 37 weeks (all participants) for Leite 2004 and Brown 1985, 36 and 35 weeks (phototherapy and control group respectively) for Lucey 1968, and 26 weeks (all participants) for Morris Types of intervention For most of the studies (Lucey 1968; Giunta 1969; Curtis 1985; Leite 2004; Tripathi 2006; Jangaard 2007) phototherapy was started within 12 hours of age in the intervention group. Other studies initiated phototherapy by 24 +/- 12 hours of age (Brown 1985), 24 hours of age (Hodgman 1970) and at 12 to 36 hours (Morris 2008). Types of outcome measures All studies reported outcomes comparing prophylactic (early initiation) use of phototherapy with late initiation or therapeutic use of phototherapy. One of the studies (Brown 1985) also evaluated the efficacy of phototherapy both in preventing and in the treatment of already established hyperbilirubinaemia in infants weighing less than 2000 g and more than 2000 g respectively. However, for the purpose of this review, we extracted only outcome data for participants where phototherapy was evaluated for prevention rather than treatment of hyperbilirubinaemia. Follow-up Two of the studies (Jangaard 2007; Morris 2008) reportedout- comes at 18 months of age and another study (Brown 1985) reported outcomes up to six years later. The rest of the studies only reported short-term outcomes. 214

12 Excluded studies Reasons for excluding potentially eligible studies (Elliot 1974; Romagnoli 1996; Sanches 2004; Yaseen 2005) are summarised in the table of Characteristics of excluded studies. Risk of bias in included studies A summary of the assessment of risk of bias based on the criteria outlined in Higgins 2008 is given in Figure 2 and Figure 3. Additionally, a brief descriptive analysis of the studies is provided below. In general, the overall methodological quality of the included studies was acceptable. Figure 2. Methodological quality graph: review authors judgements about each methodological quality item presented as percentages across all included studies. 215

13 Figure 3. Methodological quality summary: review authors judgements about each methodological quality item for each included study. 216

14 Allocation Allocation of participants to either intervention or control group was reported as random in seven studies (Giunta 1969; Brown 1985; Curtis 1985; Leite 2004; Tripathi 2006; Jangaard 2007; Morris 2008); however, onlyfour ofthesementioned themethod used for generating the randomisation sequence (Brown 1985; Leite 2004; Jangaard 2007; Morris 2008). The sequence generation was not clear in three studies. Three studies did not mention any random process for allocation (Lucey 1968; Giunta 1969; Elliot 1974). In these studies participants were allocated alternately into the intervention or control group. Only three studies reported their method of allocation sufficiently clearly to determine that allocation was concealed (Brown 1985; Jangaard 2007; Morris 2008). Blinding Only one study reported blinding of the outcomes assessor for the treatment arm (Morris 2008). Incomplete outcome data For all the studies included in this review, randomised participants were either reported or analysed in the group to which they were allocated irrespective of the outcome. Selective reporting The study protocol was available for only two of the included studies (Brown 1985; Morris 2008) and all the outcomes specified in the protocol were reported in these two studies. Selective outcome reporting could not be assessed for the rest of the included studies. Other potential sources of bias Baseline characteristics were similar in all the studies and all the studies were completed as planned. The sources of funding for the various studies were not likely to influence the outcomes. Of all the nine included studies, we had access to only two of the study protocols (Brown 1985; Morris 2008). Thus it is hard to assess whether all prespecified outcomes were reported or not for seven of the studies. Effects of interventions See: Summary of findings for the main comparison Prophylactic phototherapy for jaundice in preterm or low birth weight infants We presented results based on outcome measures as outlined in the review protocol. We also reported other outcomes reported in the primary studies that we deemed to be relevant to the review question. We conducted meta-analysis where we considered it to be appropriate. Prophylactic phototherapy versus control (COMPARISON 1) Primary outcomes Incidence of clinical kernicterus None of the studies provided data on this outcome. Incidence of pathological kernicterus None of the studies provided data on this outcome. Exchange blood transfusion Three studies (Brown 1985; Tripathi 2006; Morris 2008) reported data on exchange blood transfusion and contributed 2946 participants. The studies used different thresholds for exchange transfusion (a lower threshold was used in the Brown 1985 study, while Morris 2008 and Tripathi 2006 used a higher threshold). Therefore, we created subgroups based on higher or lower thresholds for exchange transfusion. Brown 1985 reported a statistically significant reduction in exchange transfusion (risk ratio (RR) 0.20; 95% confidence interval (CI) 0.13 to 0.31) while Morris 2008 and Tripathi 2006 did not find any significant difference in the rate of exchange transfusion (RR 0.66; 95% CI 0.19 to 2.28). Overall, prophylactic phototherapy was associated with a 78% reduction in the rate of exchange blood transfusion (typical RR 0.22; 95% CI 0.15 to 0.34), though there was moderate statistical heterogeneity (Chi² = 3.27, df = 2 (P = 0.19); I² = 39%) (Figure 4). 217

15 Figure 4. Forest plot of comparison: 1 Prophylactic phototherapy versus control, outcome: 1.1 Exchange transfusion. Long-term outcomes assessed at any age beyond one year of age by a validated cognitive, motor, language or behavioural/school/social interaction/adaptation test Neurodevelopmental impairment This was a composite outcome reported by one study (Morris 2008). Neurodevelopmental impairment was assessed at 18 to 22 months of age and was defined to include blindness, severe hearing loss and moderate or severe cerebral palsy. Prophylactic phototherapy reduced the rate of neurodevelopmental impairment (RR 0.85; 95% CI 0.74 to 0.99, risk difference (RD) -0.04; 95% CI to -0.00). Cerebral palsy Two studies (Brown 1985; Jangaard 2007) reported on this outcome. There was no significant difference in the incidence of cerebral palsy assessed at one year of age (Brown 1985) andat18 months of age (Jangaard 2007) (typical RR 0.96; 95% CI 0.50 to 1.85) (Figure 5). Jangaard 2007 also reported separately on the rate of cerebral palsy among infants weighing < 1000 g. There was also no significant difference between the two arms in this subgroup (RR 0.29; 95% CI 0.04 to 2.27). Secondary reports emanating from Brown 1985 at six-year follow-up also showed that there was no significant difference in the rate of cerebral palsy between the phototherapy and control group. 218

16 Figure 5. Forest plot of comparison: 1 Prophylactic phototherapy versus control, outcome: 1.3 Cerebral palsy. Sensorineural hearing loss One study (Brown 1985) at six-year follow-up reported on sensorineural hearing loss and found no statistically significant difference between the phototherapy group and the control group for infants weighing less than 2000 g (RR 0.31; 95% CI 0.07 to 1.50). Secondary outcomes Peak serum bilirubin level during the first seven days of life Six studies (Lucey 1968; Curtis 1985; Leite 2004; Tripathi 2006; Jangaard 2007; Morris 2008) reported on this outcome and contributed 2319 participants. The mean peak serum bilirubin was significantly lower in the phototherapy group compared to the control group for all infants weighing < 2500 g (mean difference (MD) -2.73; 95% CI to -2.57). However, statistical heterogeneity was substantial and highly significant (I² = 91%) (Figure 6). This is possibly due to the differences in population, timing of assessment and methods of measurement of bilirubin. We conducted a subgroup analysis for this outcome by birth weight and age of participants at time of initiation of phototherapy. 219

17 Figure 6. Forest plot of comparison: 1 Prophylactic phototherapy versus control, outcome: 1.4 Peak serum bilirubin. Subgroup analyses 1. Birth weight Two studies (Jangaard 2007; Morris 2008) reported peak serum bilirubin for infants weighting < 1000 g. The peak serum bilirubin was significantly lower in the phototherapy group compared to the control group (MD -2.79; 95% CI to -2.62). Four studies (Curtis 1985; Tripathi 2006; Jangaard 2007; Morris 2008) reported on the peak serum bilirubin for infants weighing less than 1500 g and also found a statistically significant difference in the phototherapy and control group (MD -2.69; 95% CI to ) (Figure 6). There was moderate statistical heterogeneity (I² = 34%) in the less than 1000 g birth weight group. 2. Age of participants at time of initiation of phototherapy In five of the studies (Lucey 1968; Curtis 1985; Leite 2004; Tripathi 2006; Jangaard 2007) phototherapy was initiated within 12 hours of birth in the intervention group. Pooling results from these studies shows that the peak serum bilirubin was significantly lower in the phototherapy group compared to the control group (MD -2.06; 95% CI to -1.53). Phototherapy was initiated within 12 to 36 hours of birth in one study (Morris 2008). The peak serum bilirubin was significantly lower in the phototherapy group compared to the control group in this study (MD -2.80; 95% CI to -2.63). The difference between commencing phototherapy within 12 hours and between 12 and 36 hours of birth was not significantly different. Serum bilirubin > 10 mg/dl This outcome was considered post hoc. Three studies (Hodgman 1970; Brown 1985; Leite 2004) reported incidence of serum bilirubin > 10 mg/dl. There was a statistically significantly lower number of neonates with serum bilirubin > 10 mg/dl in the phototherapy group compared to the control group (typical RR 0.27; 95% CI 0.22 to 0.33) with moderate heterogeneity (I² = 51%) (Figure 7). 220

18 Figure 7. Forest plot of comparison: 1 Prophylactic phototherapy versus control, outcome: 1.7 Serum bilirubin > 10 mg/dl. Serum bilirubin > 15 mg/dl This outcome was also considered post hoc. Three studies (Giunta 1969; Hodgman 1970; Brown 1985) reported on this outcome. There was a statistically significant reduction in the number of neonates with serum bilirubin > 15 mg/dl in the phototherapy group compared to the control group (typical RR 0.13; 95% CI 0.07 to 0.23). Duration of phototherapy in hours Three studies (Curtis 1985; Jangaard 2007; Morris 2008)reported on this outcome. The mean duration of phototherapy was longer in the prophylactic phototherapy group compared to the control group (MD hours; 95% CI to hours). Fluid intake/requirement in the first week after birth None of the studies provided data on this outcome. Incidence of retinopathy of prematurity Three studies reported on retinopathy of prematurity (Brown 1985; Jangaard 2007; Morris 2008). There was no statistically significant difference in the incidence of retinopathy of prematurity in the phototherapy group compared to the control group in two of the studies (Jangaard 2007; Morris 2008) (typical RR 0.93; 95% CI 0.77 to 1.13). Secondary reports emanating from the multicentre trial of Brown 1985 at one-year follow-up shows that 11 infants had severe visual and retinal changes, two in the phototherapy group and nine in the control group. One of the infants in the control group had retinopathy of prematurity. However, the causes of the visual impairments were unrelated and could not be attributed to a single cause. Intraventricular haemorrhage Two studies (Jangaard 2007; Morris 2008) presented data on the incidence of intraventricular haemorrhage and meta-analysis revealed no statistically significant difference between groups (typical RR 0.94; 95% CI 0.80 to 1.10). Incidence of bronze baby syndrome None of the studies provided data on this outcome. Incidence of patent ductus arteriosus (PDA) (clinical diagnosis) or PDA requiring treatment with cyclo-oxygenase inhibitor or surgery One study (Morris 2008) reportedtheincidenceofpda.there was no statistically significant difference in the incidence of PDA in both groups (RR 0.94; 95% CI 0.85 to 1.03). Periventricular leukomalacia [diagnosed by ultrasound or magnetic resonance imaging (MRI)] None of the studies provided data on this outcome. Duration of hospital stay (days) This outcome was reported in two studies (Jangaard 2007; Morris 2008). There was no difference in the duration of hospitalisation between groups (MD -3 days; 95% CI -7 to +1 days). 221

19 All-cause mortality (before discharge) Four studies reported this outcome (Hodgman 1970; Brown 1985; Tripathi 2006; Morris 2008). There was no statistically significant difference in the risk of mortality between groups (typical RR 1.08; 95% CI 0.93 to 1.26). In the group of infants weighing < 1000 g, there was also no statistically significant difference in the rate of mortality between the phototherapy and control group (RR 1.07; 95% CI 0.91 to 1.25) (Figure 8). Figure 8. Forest plot of comparison: 1 Prophylactic phototherapy versus control, outcome: 1.14 All-cause mortality. GRADE assessment We assessed the evidence according to GRADE criteria, although this was not stipulated in the protocol. See Summary of findings for the main comparison for the main comparison. Using the GRADE tool, we evaluated the evidence provided by the studies and rated this for each outcome identified as critical or important. DISCUSSION Summary of main results Nine studies were eligible for this review. Most of the studies compared early initiation of phototherapy (prophylactic phototherapy) with late commencement of phototherapy in preterm low birth weightneonateswithoutany evidence of haemolytic jaundice due to ABO incompatibility. There was reduction in the number of infants who required exchange transfusion; however, there was marked heterogeneity with one study using a lower threshold for exchange transfusion and reporting a significantly higher number of exchange transfusions in the control group. There was also a significantreductioninpeakserumbilirubinandalowernumber of infants with bilirubin level > 10 mg/dl or 15 mg/dl. Duration of phototherapy was longer in prophylactic phototherapy group by more than two days. There was no statistically significant difference in the rate of cerebral palsy and sensorineural hearing loss. However, one of the studies which reported on neurodevelopmental impairment (a composite outcome including cerebral palsy) found a slightly lower rate of neurodevelopmental impairment in the phototherapy group. Findings from this review also suggest that prophylactic phototherapy does not significantly reduce the rate of all-cause mortality. The low incidence of long-term outcomes such as cerebral palsy could be because clinical consider- 222

20 ations required exchange blood transfusion to prevent bilirubin from reaching very high levels. This could explain the lack of significant difference in long-term outcomes between the prophylactic phototherapy and control group. Also, the power of the studies to detect the difference in the categorical outcomes should be considered. We used the GRADE approach for rating quality of evidence to assess the quality of evidence for seven of the outcomes that were considered to be very important. The quality of evidence was moderate for most of the outcomes, suggesting that further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Overall completeness and applicability of evidence All the studies directly answered the question of whether prophylactic phototherapy can be used to prevent hyperbilirubinaemia and jaundice in preterm and low birth weight infants. Seven of the studies were conducted in high-income settings (USA and Canada) and two of the studies were conducted in low and middleincome settings (India and Brazil). However, there is no reason to believe that the findings are not applicable to other parts of the world. Two of the studies provided data on follow-up of participants beyond the neonatal period up to 18 months, with low incidence of long-term outcomes. At the time of writing up this review, we did not identify and are not aware of any ongoing or planned studies aiming to answer this question. The available evidence is fairly large for short-term outcomes of peak serum bilirubin and number of infants crossing a threshold bilirubin level. However, not enough studies have reported data on long-term outcomes. Reduction of serum bilirubin by approximately 3 mg/dl for an increase in phototherapy duration by two days will needto be judged in the context of the individual setting. Data from the Morris 2008 report revealed that there was no difference in their primary composite outcome of death or neurodevelopmental impairment between two groups (52% versus 55%; risk ratio (RR) 0.94; 95% confidence interval (CI) 0.87 to 1.02; risk difference (RD) -0.03; 95% CI to 0.01) and there was a trend towards an increase in mortality in the preplanned subgroup of 751 to 1000 g infants (39% in phototherapy group versus 34% in control group, RR 1.13; 95% CI 0.96 to 1.34) which may offset the benefit gained of reduction in neurodevelopmental disability in extremely low birth weight infants. However, overall, there was no difference in the rate of all-cause mortality between the phototherapy and control group. Quality of the evidence The methodological quality varied across the included studies. Most of the studies did not state the method used for sequence generation and most likely these studies are unblinded. In addition, the sample sizes varied across all the studies and most of the studies did not provide any description of the sample size calculation. Two multicentre, large-scale randomised studies had no risk of bias. Potential biases in the review process We conducted comprehensive searches of both journal and conference databases to ensure that all relevant published and unpublished studies were identified. We also scanned through the reference lists of all the identified studies to look for any relevant study titles. There were no language or date limits in our search for studies. We repeated the search for studies just before publication to identify any new studies. It is therefore unlikely that our search missed any relevant study. We also reduced potential bias in the conduct of this review by having two of the authors independently scan through the search output, extract data and assess the methodological quality of each study. In the meta-analysis, we pooled data from different studies even though the studies were slightly different in terms of gestational age of participants, birth weight and wavelength of phototherapy used. However, we did not consider these differences to be a source of bias, but the results of from the meta-analysis can be considered to be less direct. Our protocol was aimed at very low birth weight infants; however, after reviewing studies available on the topic we modified the population to include preterm and low birth weight infants. AUTHORS CONCLUSIONS Implications for practice Based on the available data there is evidence that prophylactic phototherapy prevents a significant rise in unconjugated hyperbilirubinaemia, reduces the need for exchange transfusion and may reduce long-term neurodevelopmental impairment. There was no effect on all-cause mortality, duration of hospital stay or increase in duration of phototherapy. Longer duration of phototherapy could incur additional cost and the benefit needs to be weighed against the cost and complications. Implications for research The quality of evidence assessed using the GRADE approach suggests that further research is likely to have an important impact on our confidence in the estimate of effect for most of the outcomes and may change the estimate. Therefore, further well-designed studies are warranted to determine both the short and long-term effectsof prophylacticphototherapy, including a cost-effectiveness 223