Patterns of neurobehavioral functioning in school-aged survivors of neonatal jaundice and hypoxic-ischemic encephalopathy in Kilifi, Kenya: A cross-sectional study

Background: Studies in high-income countries have reported that school-aged children who survive neonatal jaundice (NNJ) and hypoxic-ischemic encephalopathy (HIE) develop long-term neurocognitive problems. However, less is known about the patterns of functioning in school-aged survivors of NNJ and HIE in sub-Saharan Africa. This study examined patterns of functioning in school-aged children who survived NNJ and HIE in Kilifi, Kenya. Methods: This is a cross-sectional study that included 107 survivors of NNJ/HIE (64 with NNJ, 43 with HIE), aged 6-12 years, admitted to Kilifi County Hospital on the Kenyan Coast. The Gross Motor Function Classification System (GMFCS), Adapted Communication Profile, Raven’s Coloured Progressive Matrices (RCPM) and an epilepsy screening tool were used to assess gross motor function, communication function, intellectual functioning, and epilepsy, respectively. Results: Most of the survivors of NNJ (95.2%) and HIE (95.3%) had no impairments in gross motor functioning. A small percentage of the children in the NNJ and HIE groups had profound problems in their communication (4.7% and 4.7%); expressive communication function (4.7% and 4.7%); social functions (3.1% and 2.3%); receptive communication (4.7% and 2.3%); and communicative effectiveness (4.7% and 2.3%). Cognitive impairment was reported in 10.9% and Open Peer Review


Introduction
Neonatal jaundice (NNJ) and hypoxic-ischemic encephalopathy (HIE) are common insults during the neonatal period. These insults have both short-term and long-term impacts on children's functioning [1][2][3][4][5][6][7][8] . The global incidence of severe NNJ is estimated at 9.9 per 10,000 live births among children 9 . Africa has the highest burden of severe NNJ with incidence rates of 667.8 per 10,000 live births (95% CI 603-738) 9 . HIE is caused by different factors, such as uterine rupture, placenta abruption, cord prolapse, maternal hypotension, and obstructed labour, which either impair the supply of blood and oxygen to the brain before, during or immediately after the birth of the baby 10,11 . The incidence of HIE globally is estimated at 1.5 per 1000 live births (95% CI 1.3-1.7) 12 and it is associated with poor neurocognitive outcomes 4,5,13-17 .
The overall burden of NNJ and HIE in neonates admitted to Kilifi County Hospital in Kenya increased between 1990 and 2008 significantly by 6% and 11%, respectively 18 . They were the second and third most common neonatal conditions after sepsis (13%) 18 . In 2015, 32% of neonatal mortality was caused by HIE and birth trauma in Kenya 19 .
Studies in high-income countries have reported that schoolaged children who survive NNJ and HIE develop adverse longterm neurocognitive outcomes 4,6,16,[20][21][22][23][24][25][26][27][28][29][30] , although long-term outcomes of NNJ tend to be less severe. However, to the best of our knowledge, there are no studies on long-term neurocognitive outcomes in school-aged children who survived NNJ and HIE in sub-Saharan Africa (SSA) despite the high burden of NNJ and HIE in this context. This study investigated the patterns of functioning in school-aged children who survived NNJ or HIE in Kilifi, Kenya.

Study design
This is a cross-sectional study that examined the neurobehavioral patterns of functioning of school-aged children (6-12 years) who survived NNJ and HIE.

Study site
This study was conducted at the Centre for Geographic Medicine Research-Coast (CGMR-C) located in Kilifi County, at the North Coast of Kenya. The study used the Kilifi Health and Demographic Surveillance System (KHDSS) to identify and recruit a well-defined cohort of school-aged children who were admitted to the Kilifi County Hospital in the first 28 days of life with NNJ or HIE and for whom neonatal data were available. Participants were recruited and assessed from September to December 2017. Assessments were carried out by trained research assistants under the supervision of a psychologist (DM) at the CGMR-C neuro-assessment unit, during which participants were accompanied by their mother or a primary caregiver in the absence of the mother.

Participants and procedures
We utilized the KHDSS to identify and trace survivors. Participants were recruited in the study if they had a diagnosis of NNJ or HIE during the first 28 days of life; were aged between 6 to 12 years at the time of follow-up; parental consent was obtained; and they were living within the area covered by the KHDSS. Participants were excluded if they did not consent to the study.

Diagnosis
The diagnosis of NNJ was based on clinical laboratory measurement of total serum bilirubin (TSB) as well as medical history and examination during the first 28 days of life. NNJ was defined as a TSB level of >85 µ/mols/l recorded to the clinical notes. Severe hyperbilirubinemia was defined as TSB of >250 µ/mols/l. HIE diagnosis was based on the clinical diagnosis recorded by the clinicians at discharge. Severe disability was defined as the impairment in body structure which results in significant loss and difficulty for a participant to perform a task 31 .

Screening tools
A set of screening tools were used to describe the level of functioning and patterns of disability among participating children. Anthropometric data (weight, height, head circumference, and Mid Upper Arm Circumference) were obtained based on the World Health Organization (WHO) standards 32 . Screening assessments were done for gross motor functioning, communication functioning, intellectual disability, and epilepsy. The participants were screened for hearing and visual acuity using an auditory brainstem response machine 33 and the Snellen and E-Chart, respectively. Almost all the participants had normal hearing and vision functioning except one who had mild vision problems.
The Gross Motor Function Classification System (GMFCS) was used to measure gross motor functioning. The GMFCS tool was devised by Peter Rosenbaum and colleagues to determine the level that best describes a participant's current abilities and limitations in gross motor function 34 . The GMFCS classifies children into 5 levels: Level I, the child can walk to various places and climb stairs without using rails and can jump and run with ease, although some children might have limitations in motor coordination while performing such gross motor functions; Level II, the child has limitations in outdoor activities; Level III, the child needs support to move; Level IV, the child needs technological assistance to move; Level V, the child's movement is completely restricted, and they need complete assistance to move. The caregiver is asked to choose the best description of their child, which shows the child's level of gross motor functioning. The GMFCS has good interrater agreement [Kappa 0.76 to 0.88; intraclass correlation coefficient (ICC) ranging from 0.89 to 0.95] 35 .
Communication functioning was assessed using the Adapted Communication Profile 36 . This tool captures the child's language abilities through a caregiver report. The caregiver is asked questions about the child's communication abilities and asked to indicate the level of problems his/her child has for the subscales social communication functions, receptive communication functions, and communication effectiveness, and this is rated using scores of 0= not a problem, 1= a bit of a problem, and 3= a big problem. The scores are then summed for each participant. The Adapted Communication Profile is contextually relevant and has previously been used with children in Kilifi 37 ; however, its psychometric properties in Kenya are yet to be established.
The Raven's coloured progressive matrices (RCPM) 38 was administered to assess intellectual functioning. The RCPM is made up of a series of patterns with a missing part, which the participant completes by choosing from several options. The multi-choice items require abstract reasoning. This test has been validated and previously used in children in Kilifi, Kenya 39 and has good internal consistency (Cronbach alpha = 0.81) and test-retest reliability (ICC = 0.77) 39 . The test has good construct validity in the Kenyan population 40 .
The epilepsy screening tool 41 was used to screen for epilepsy in this study sample. This tool was validated using a three-stage screening methodology for detecting active epilepsy in Kilifi, Kenya 41 . Active epilepsy was defined as two or more unprovoked seizures occurring within the last 12 months 42 , or on anti-epileptic treatment.

Study size
As per the KHDSS records by December 2017, of the 280 cases with NNJ admitted between 2005 and 2012, 17 (6.1%) children died before discharge, 15 (5.4%) died after discharge, while 67 (23.9%) had migrated from the KHDSS and their survival could not be determined. Of the 378 neonates who were admitted with HIE between 2005 and 2012, 117 (31.0%) died before discharge, and 16 (4.2%) died after discharge. However, 79 (20.9%) had migrated from the KHDSS and their survival could not be determined.
The recruitment, and assessment processes are indicated in Figure 1. Out of the 658 survivors of NNJ and HIE, 347 survivors were identified, 121 were followed up and visited at home for recruitment, and 107 participants aged 6-12 years were included in this study.

Statistical analysis
Data were collected, entered, and managed using REDCap, an electronic data capture tool hosted at the CGMR-C, and analysed using STATA (version 15) 43 . The anthropometric variables Weight-for-Age (WAZ) and Height-for-Age (HAZ) were standardized using WHO Anthro plus 44 . An abnormal nutritional status (stunted growth or underweight) was considered if the z-scores obtained from WHO Anthro plus were below -2 standard deviation (SD). Descriptive statistics such as means, medians, and percentages were used to describe sample characteristics and to summarize gross motor, language, and intellectual functioning and history of epilepsy. The cognitive and epilepsy outcomes of these children were compared to the normative data obtained from a study conducted in 2016 with 11,223 children aged 6 to 9 years randomly selected from the Kenyan community to estimate the burden of neurological impairments 45 . Cognitive impairment was defined as total Ravens Z-scores below -2 SD. The 95% confidence interval (CI) were calculated using the Clopper-Pearson exact method. A sub-analysis was conducted with participants with severe hyperbilirubinemia on all outcomes and comparable results were obtained. Therefore, we report data with all participants with TSB of >85 µ/mols/l.

Demographic characteristics of participants
In this study, of the 107 included participants, 64 survived NNJ (31 females and 33 males) and 43 survived HIE (29 females and 14 males). The median age of the participants was 10 [interquartile range (IQR) = 5-12] years. These participants had normal anthropometric measures; none of the participants were underweight or had stunted growth. All participants had normal hearing and visual functioning except one who had visual acuity problems. The mean WAZ was -1.3 (SD = 0.9) and -1.0 (SD = 1.6), while the mean HAZ was -1.1 (SD = 1.1) and -0.8 (SD = 1.5) for NNJ and HIE, respectively. Table 1 presents a summary of these results (see underlying data 46 ).   functions, social communication functions, receptive communication and communicative effectiveness, respectively (Table 2).

History of epilepsy
As shown in  (Table 3).

Discussion
This study investigated the patterns of neurobehavioral functioning in children who survived NNJ and HIE in Kilifi, Kenya. However, the children in that study were also reported to have a substantial risk of comorbidities such as cerebral palsy (20%) and epilepsy (20%), unlike in our study where 1.6% of the NNJ survivors had active epilepsy. The difference in the prevalence of epilepsy in these two studies could be because of difference in the severity of NNJ. Unlike in the current study where severe hyperbilirubinemia was defined as TSB of >250 µ/mols/l, the participants in Martínez-Cruz and colleagues' study had severe NNJ defined as an increase in bilirubin by >0.5 mg/dL and >0.3 mg/dL per hour in term and preterm infants, respectively, and required exchange transfusion. Therefore, the SNHL could be a result of the loss and alterations of neurons caused by the motor disorders and deposition of bilirubin in the nuclei involved in the auditory pathway. Similar to the findings of Kara et al. 48 who evaluated children aged 3 to 5 years who survived NNJ, the current study did not find any visual abnormality in survivors of NNJ.  24 , who did not find any cognitive impairment in their sample, our results show that survivors of HIE have poorer cognitive outcomes compared to the normative sample. The difference in results could be due to the difference in age groups in the two samples. Since the current study had an older age group, it is likely that the cognitive deficits reflect a cumulative effect. It should be noted that the cognitive outcomes for this sample (median age 10 years) were compared to normative data from a younger sample (age 6-9 years), which implies that the number of children with cognitive impairment found in our study may be an underestimation.
Lastly, it should be noted that the normal functioning reported in this study was found in a sample consisting of children who survived beyond age 6. There is a possibility that the cases with worse outcomes died before age 6 years thus, their data are unavailable for this study. The mortality rates of children with neonatal insults in SSA are high due to limited quality care. For instance, in severe NNJ needing exchange transfusion or HIE requiring hypothermia, provision of adequate personnel, monitoring facilities, and finances are limited, unlike in developed countries where there are available resources and personnel to provide quality care. Therefore, it is likely that most children with mild impairment survived the neonatal insults.

Limitations of the study
These study findings should be interpreted taking several limitations into account. First, most of the children with severe disabilities may have died before they reached the age of 6-12 years, thus this represents a survivor cohort. Only two out of the 107 participants were severely disabled, the prevalence of severe disability in the sample was 1.9% (0.46 -7.32). Data collection on severe disability was discontinued. Second, the motor and communication assessment tools used in this study were screening instruments, which may not have captured important aspects of these outcomes. Third, potential risk factors such as socioeconomic factors, maternal education and maternal mental health, and parenting factors that are likely to affect neurodevelopmental outcomes were not considered in this study. Lastly, due to inconsistencies in clinical documentation of bilirubin levels at admission and Apgar scores, it was not possible to add estimates of the severity of illness in the children we followed up and those whom we were not able to follow-up.

Conclusion
Based on the screening tools used, survivors of HIE and NNJ in Kilifi, Kenya, do not experience challenges in motor and communication functioning. Additionally, their nutritional status was normal. However, a substantial proportion of them are likely to have impaired cognition compared to the normative sample. It is likely that the children who were followed up had mild impairment while those with severe outcomes did not survive.

Ethical statement
Ethical approval for this study was granted by the Kenya Medical Research Institute Scientific and Ethics Review Unit (SERU); protocol number 092/3470. The primary caregivers of the children were informed about the study and their written informed consent for them and their children to take part in the study was obtained. Assent was also obtained from the children who took part in the study. Additional permission was obtained from the Kilifi County Office, and the Kilifi County Director of Education as most of the participants were school going children. Confidentiality and anonymity were maintained in all stages of data management and analysis. In this cross-sectional retrospective study, the authors set out to investigate the neurodevelopmental outcome of infants born with neonatal jaundice and hypoxic ischemic encephalopathy in low-resource setting and for that the authors are to be commended. This information is urgently needed if we are to get the appropriate resources, infra-structure and governmental support to eliminate or at least significantly decrease these diseases and the longterm consequences for children who survive these insults as neonates. This article is especially noteworthy in that the study team includes experts in neurodevelopment from the Kenya Medical Research Institute (KEMRI).

Data availability
However, there are several significant issues with the study that significantly impact the interpretation of the results. First, it is unclear how the authors define neonatal jaundice and hypoxic ischemic encephalopathy. Notably, a serum bilirubin of 85 micromoles/liter would not be expected to cause any problems except perhaps for the extremely low-birth weight infants. Even 250 micromoles/liter would rarely be expected to cause motor or developmental problems. Furthermore, the complete absence of any children with deafness along with the lack of other markers of severe neonatal jaundice such as the need for exchange transfusion or even phototherapy make it unlikely that the population included in this study were at risk for neurodevelopmental delay from neonatal jaundice. The only possible marker of acute bilirubin encephalopathy they recorded is death. While death certainly is a marker of severe neonatal jaundice, it can be the result of multiple other pathological processes. Unfortunately, the authors do not clarify whether the death was related to jaundice or not, leaving this uncertainty open to interpretation. It follows that the lack of markers for acute bilirubin encephalopathy confounds any possible relationship between the observed (or lack thereof) neurodevelopmental problems and hyperbilirubinemia. In fact, low levels of bilirubin (as reported in this study) have been postulated to be neuroprotective not harmful. The study would have been much more meaningful had the authors looked at children who had had suffered from truly severe neonatal hyperbilirubinemia and/or acute bilirubin encephalopathy.
Diagnosis of hypoxic ischemic encephalopathy by "clinical diagnosis" at discharge is also too nonspecific to have any idea of the group of neonates being discussed. The authors give no indication of the degree of hypoxic ischemic encephalopathy and thus, as with neonatal jaundice, the interpretation of a possible causal relationship with neurodevelopmental delay in school age children remains uninterpretable. In addition, the authors do not clarify the training of the clinicians making the diagnosis of HIE or how this diagnosis was validated, further complicating any possible meaningful interpretation of the observed results.
There is no comment regarding neonates who had both neonatal jaundice and hypoxic ischemic encephalopathy. Asphyxiated neonates would be expected to higher risk for acute bilirubin encephalopathy and long-term problems, but this group is not addressed by the authors.
Because of the flaws mentioned above, the study as currently presented, does not add sustainably to our understanding of the true magnitude of neurodevelopmental problems from either neonatal jaundice or hypoxic ischemic encephalopathy. The study may be meaningful If the authors add substantial clarification of their patient populations and demonstrate why the selected populations would be expected to potentially have an increased risk of neurodevelopmental problems from either neonatal jaundice or hypoxic encephalopathy.

If applicable, is the statistical analysis and its interpretation appropriate? Partly
Are all the source data underlying the results available to ensure full reproducibility? Yes

Are the conclusions drawn adequately supported by the results? No
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: My primary area of research is severe neonatal jaundice, its preventions, diagnosis, treatment and follow-up. I also have interest in breast feeding in LMICs and bubble CPAP/respiratory support beyond the neonate.
We confirm that we have read this submission and believe that we have an appropriate level of expertise to state that we do not consider it to be of an acceptable scientific standard, for reasons outlined above. Notably, a serum bilirubin of 85 micromoles/litre would not be expected to cause any problems except perhaps for the extremely low-birth weight infants. Even 250 micromoles/litre would rarely be expected to cause motor or developmental problems. Furthermore, the complete absence of any children with deafness along with the lack of other markers of severe neonatal jaundice such as the need for exchange transfusion or even phototherapy make it unlikely that the population included in this study were at risk for neurodevelopmental delay from neonatal jaundice. The purpose of this study was to assess more subtle impairments, such as cognitive impairment and communication since this has not been studied in children surviving these insults in sub-Saharan Africa. Our inclusion of children with TSB>85 µ/mols/l is based on two facts: First, this is the level at which jaundice is reliably detected in the neonate. It is the definition used by the American Academy of Pediatrics for hyperbilirubinaemia, and other authors (Avery, 2005;. Second, there are considerable difficulties in establishing gestational age  and time of birth and the severity of hypoxic-ischemic encephalopathy of neonates admitted to hospitals serving rural areas in sub-Saharan Africa, where most births occur at home. Moreover, there is considerable debate about the criteria for a safe level of bilirubin in sick neonates Varughese, 2019). This is one of the few studies to provide data that suggests that few problems develop in neonates who have bilirubin levels between 85 and 250 µmol/l/. 6. There is no comment regarding neonates who had both neonatal jaundice and hypoxic ischemic encephalopathy. Asphyxiated neonates would be expected to higher risk for acute bilirubin encephalopathy and long-term problems, but this group is not addressed by the authors. Because of the flaws mentioned above, the study as currently presented does not add sustainably to our understanding of the true magnitude of neurodevelopmental problems from either neonatal jaundice or hypoxic-ischemic encephalopathy. We did not have these cases in this cohort. None of the participants in this study had a diagnosis of both NNJ and HIE as those with a combined diagnosis were excluded from this cohort study (page 4).

Reviewer 2: Deirdre Murray
1.The definition of hyperbilrubinaemia is very vague. They seem to have included all infants with a bilirubin level > 85. The authors do not give a time for this, beyond stating that the measurement took place in the first 28 days. The majority of infants have some level of jaundice, and will reach this level The bilirubin was measured on admission, since many of the neonates were born at home, it was difficult to determine the exact age in hours. Moreover, the inclusion criteria included neonates up to 30 days of age as per the definition of neonatal period as used in other studies . However, despite the inclusion criteria of 0-30 days, the median age of the participants at admission was 3 [interquartile range (IQR) = 0-8] days (page 6-7). ○ 2. They state that they have done a sub analysis in the severe group, but do not tell us the numbers or the results in this group, who are actually the more interesting.
We have provided the actual numbers of children with hyperbilirubinemia. A subanalysis was conducted with 25 participants with severe hyperbilirubinemia on all outcomes, and similar results were obtained (page 6).
○ 3. Surprisingly with this low level of hyperbilirubinaemia the outcomes are poor which makes me wonder whether these infants had other underlying diagnoses, such as sepsis, prematurity, IUGR?
We have addressed this issue in the methods section as follows: Some of the participants had sepsis and preterm birth as a secondary diagnosis. For NNJ 23 had neonatal sepsis, and 6 were preterm. In the HIE group, 5 had sepsis, and 2 were preterm. However, based on another study that we conducted, sepsis did not appear to aggravate the developmental outcomes of children with neonatal jaundice and sepsis ) (page 5).
○ 4. The follow up rate was low. they state that 347 survivors were identified and 121 were followed up. Please explain why the other 227 were not followed? This explanation is given under limitations in the discussion section as follows: Given that this study was designed to screen out children with severe disability (they could not be able to carry out tasks during assessment) and only two out of the 107 participants were severely disabled, the prevalence of severe disability in the sample was 1.9% (0.46 -7.32). Data collection on children with severe disability was discontinued (page 10) as we had reached the needed sample size to determine the severity of disability in these children. . 7. The first paragraph of the discussion is repeated twice in the second paragraph.
We have revised this part and deleted the repetition as needed.

Reviewer 3: Bolajoko O Olusanya
1.The clinical profile of the participants as neonates is quite deficient and does not provide an objective basis for evaluating the risks of neurodevelopmental disorders. For example, the operational definitions of NNJ used in the study are rarely associated with neurodevelopmental disorders. NNJ is generally benign except in children with or at risk of acute bilirubin encephalopathy (ABE). Since the authors acknowledged inconsistencies in clinical documentation of bilirubin levels at admission (and presumably on discharge also), it would have been useful to identify those who received phototherapy and/or exchange transfusion as proxies for identifying participants with severe NNJ. This is even more crucial in a developing country like Kenya where delays in receiving appropriate care are not uncommon (see Olusanya et al. (2014)1. The purpose of this study was to assess more subtle impairments, such as cognitive impairment and communication since this has not been studied in children surviving these insults in sub-Saharan Africa. Our inclusion of children with TSB>85 µ/mols/l is based on two facts: First, this is the level at which jaundice is reliably detected in the neonate. It is the definition used by the American Academy of Pediatrics for hyperbilirubinaemia, and other authors (Avery, 2005;. Second, there are considerable difficulties in establishing gestational age  and time of birth and the severity of hypoxic-ischemic encephalopathy of neonates admitted to hospitals serving rural areas in sub-Saharan Africa, where most births occur at home. Moreover, there is considerable debate about the criteria for a safe level of bilirubin in sick neonates Varughese, 2019). This is one of the few studies to provide data that suggests that few problems develop in neonates who have bilirubin levels between 85 and 250 µmol/l/. 3. It is unclear why the authors opted for auditory brainstem response (ABR) in these school-aged children rather than pure-tone audiometry which is a more accurate and common measure of auditory threshold, especially in resource-limited settings. The authors need to provide details of the type of ABR and the methodology employed for hearing screening in their population. We have added details on the type of audiometry machine and the methodology employed: The participants were screened for hearing and visual acuity using a pure-tone audiometry machine-Kamplex model R17A AUD Type 3   The definition of hyperbilrubinaemia is very vague. They seem to have included all infants with a bilirubin level > 85. The authors do not give a time for this, beyond stating that the measurement took place in the first 28 days. The majority of infants have some level of jaundice, and will reach this level. They state that they have done a sub analysis in the severe group, but do not tell us the numbers or the results in this group, who are actually the more interesting.
Surprisingly with this low level of hyperbilirubinaemia the outcomes are poor which makes me wonder whether these infants had other underlying diagnoses, such as sepsis, prematurity, IUGR?
Similarly no grades are given for HIE. However this is understandably difficult in a retrospectively study. More detailed neonatal information regarding these infants; Apgar score, severity of encephalopathy would be more informative if available.
The follow up rate was low. they state that 347 survivors were identified and 121 were followed up. Please explain why the other 227 were not followed?
They have stated in their discussion the other major limitation: the fact that it is likely that a high proportion of children; those with moderate or severe HIE are likely to have died. If the authors could focus on this as a study of outcome following HIE, reporting mortality and survival, with outcome to 10 years then this would be a very valuable article and would add significantly to the literature.
Minor points: How did the WAZ and HAZ scores compare to the general population? Do they have this data?
The first paragraph of the discussion is repeated twice in the second paragraph.

Are sufficient details of methods and analysis provided to allow replication by others? No
If applicable, is the statistical analysis and its interpretation appropriate? Yes for neurodevelopmental delay from neonatal jaundice. The purpose of this study was to assess more subtle impairments, such as cognitive impairment and communication since this has not been studied in children surviving these insults in sub-Saharan Africa. Our inclusion of children with TSB>85 µ/mols/l is based on two facts: First, this is the level at which jaundice is reliably detected in the neonate. It is the definition used by the American Academy of Pediatrics for hyperbilirubinaemia, and other authors (Avery, 2005;. Second, there are considerable difficulties in establishing gestational age  and time of birth and the severity of hypoxic-ischemic encephalopathy of neonates admitted to hospitals serving rural areas in sub-Saharan Africa, where most births occur at home. Moreover, there is considerable debate about the criteria for a safe level of bilirubin in sick neonates Varughese, 2019). This is one of the few studies to provide data that suggests that few problems develop in neonates who have bilirubin levels between 85 and 250 µmol/l/. 6. There is no comment regarding neonates who had both neonatal jaundice and hypoxic ischemic encephalopathy. Asphyxiated neonates would be expected to higher risk for acute bilirubin encephalopathy and long-term problems, but this group is not addressed by the authors. Because of the flaws mentioned above, the study as currently presented does not add sustainably to our understanding of the true magnitude of neurodevelopmental problems from either neonatal jaundice or hypoxic-ischemic encephalopathy. We did not have these cases in this cohort. None of the participants in this study had a diagnosis of both NNJ and HIE as those with a combined diagnosis were excluded from this cohort study (page 4).

Reviewer 2: Deirdre Murray
1.The definition of hyperbilrubinaemia is very vague. They seem to have included all infants with a bilirubin level > 85. The authors do not give a time for this, beyond stating that the measurement took place in the first 28 days. The majority of infants have some level of jaundice, and will reach this level The bilirubin was measured on admission, since many of the neonates were born at home, it was difficult to determine the exact age in hours. Moreover, the inclusion criteria included neonates up to 30 days of age as per the definition of neonatal period as used in other studies . However, despite the inclusion criteria of 0-30 days, the median age of the participants at admission was 3 [interquartile range (IQR) = 0-8] days (page 6-7). ○ 2. They state that they have done a sub analysis in the severe group, but do not tell us the numbers or the results in this group, who are actually the more interesting.
We have provided the actual numbers of children with hyperbilirubinemia. A subanalysis was conducted with 25 participants with severe hyperbilirubinemia on all outcomes, and similar results were obtained (page 6).
3. Surprisingly with this low level of hyperbilirubinaemia the outcomes are poor which makes me wonder whether these infants had other underlying diagnoses, such as sepsis, prematurity, IUGR?
We have addressed this issue in the methods section as follows: Some of the participants had sepsis and preterm birth as a secondary diagnosis. For NNJ 23 had neonatal sepsis, and 6 were preterm. In the HIE group, 5 had sepsis, and 2 were preterm. However, based on another study that we conducted, sepsis did not appear to aggravate the developmental outcomes of children with neonatal jaundice and sepsis ) (page 5).
○ 4. The follow up rate was low. they state that 347 survivors were identified and 121 were followed up. Please explain why the other 227 were not followed? This explanation is given under limitations in the discussion section as follows: Given that this study was designed to screen out children with severe disability (they could not be able to carry out tasks during assessment) and only two out of the 107 participants were severely disabled, the prevalence of severe disability in the sample was 1.9% (0.46 -7.32). Data collection on children with severe disability was discontinued (page 10) as we had reached the needed sample size to determine the severity of disability in these children.
○ 5. They have stated in their discussion the other major limitation: the fact that it is likely that a high proportion of children; those with moderate or severe HIE are likely to have died. If the authors could focus on this as a study of outcome following HIE, reporting mortality and survival, with outcome to 10 years then this would be a very valuable article and would add significantly to the literature. Plans to study the mortality and survival of NNJ and HIE are underway in future studies. However, in this study, our main research question in the current study was to understand the neurobehavioral patterns of survivors of NNJ and HIE. ○ 7. The first paragraph of the discussion is repeated twice in the second paragraph. We have revised this part and deleted the repetition as needed.

Reviewer 3: Bolajoko O Olusanya
1.The clinical profile of the participants as neonates is quite deficient and does not provide an objective basis for evaluating the risks of neurodevelopmental disorders. For example, the operational definitions of NNJ used in the study are rarely associated with neurodevelopmental disorders. NNJ is generally benign except in children with or at risk of acute bilirubin encephalopathy (ABE). Since the authors acknowledged inconsistencies in clinical documentation of bilirubin levels at admission (and presumably on discharge also), it would have been useful to identify those who received phototherapy and/or exchange transfusion as proxies for identifying participants with severe NNJ. This is even more crucial in a developing country like Kenya where delays in receiving appropriate care are not uncommon (see Olusanya et al. (2014)1. The purpose of this study was to assess more subtle impairments, such as cognitive impairment and communication since this has not been studied in children surviving these insults in sub-Saharan Africa. Our inclusion of children with TSB>85 µ/mols/l is based on two facts: First, this is the level at which jaundice is reliably detected in the neonate. It is the definition used by the American Academy of Pediatrics for hyperbilirubinaemia, and other authors (Avery, 2005;. Second, there are considerable difficulties in establishing gestational age  and time of birth and the severity of hypoxic-ischemic encephalopathy of neonates admitted to hospitals serving rural areas in sub-Saharan Africa, where most births occur at home. Moreover, there is considerable debate about the criteria for a safe level of bilirubin in sick neonates Varughese, 2019). This is one of the few studies to provide data that suggests that few problems develop in neonates who have bilirubin levels between 85 and 250 µmol/l/. 3. It is unclear why the authors opted for auditory brainstem response (ABR) in these school-aged children rather than pure-tone audiometry which is a more accurate and common measure of auditory threshold, especially in resource-limited settings. The authors need to provide details of the type of ABR and the methodology employed for hearing screening in their population. We have added details on the type of audiometry machine and the methodology employed: The participants were screened for hearing and visual acuity using a pure-tone audiometry machine-Kamplex model R17A AUD Type 3

Bolajoko O Olusanya
Centre for Healthy Start Initiative (HSI-Centre), Lagos, Nigeria This cross-sectional study set out to determine long-term neurodevelopmental disorders associated with survivors of neonatal jaundice (NNJ) and hypoxic-ischemic encephalopathy (HIE) at school age. The Gross Motor Function Classification System (GMFCS), Adapted Communication Profile, Raven's Coloured Progressive Matrices (RCPM) and an epilepsy screening tool were used to assess gross motor function, communication function, intellectual functioning, and epilepsy, respectively. The participants were also screened for hearing and visual acuity using an undisclosed auditory brainstem response instrument and the Snellen and E-Chart, respectively. The principal findings reported by the authors were that children who survived NNJ and HIE have normal vision, hearing, motor functioning, and communication functioning, but have poorer intellectual functioning compared to the normative sample.
Conceptually, this study was intended to fill a critical gap in available research evidence on the long-term sequelae of NNJ and HIE in sub-Saharan Africa. However, the validity of the study and the reported findings are compromised by the following major methodological drawbacks: The clinical profile of the participants as neonates is quite deficient and does not provide an objective basis for evaluating the risks of neurodevelopmental disorders. For example, the operational definitions of NNJ used in the study are rarely associated with neurodevelopmental disorders. NNJ is generally benign except in children with or at risk of acute bilirubin encephalopathy (ABE). Since the authors acknowledged inconsistencies in clinical documentation of bilirubin levels at admission (and presumably on discharge also), it would have been useful to identify those who received phototherapy and/or exchange transfusion as proxies for identifying participants with severe NNJ. This is even more crucial in a developing country like Kenya where delays in receiving appropriate care are not uncommon (see Olusanya et al. (2014) 1 ).

1.
The study suggests that the clinical diagnosis of HIE was based on Apgar scores. Please clarify and report the criteria for HIE.

2.
It is unclear why the authors opted for auditory brainstem response (ABR) in these schoolaged children rather than pure-tone audiometry which is a more accurate and common measure of auditory threshold especially in resource-limited settings. The authors need to provide details of the type of ABR and the methodology employed for hearing screening in their population.

3.
neonate. It is the definition used by the American Academy of Pediatrics for hyperbilirubinaemia, and other authors (Avery, 2005;. Second, there are considerable difficulties in establishing gestational age  and time of birth and the severity of hypoxic-ischemic encephalopathy of neonates admitted to hospitals serving rural areas in sub-Saharan Africa, where most births occur at home. Moreover, there is considerable debate about the criteria for a safe level of bilirubin in sick neonates Varughese, 2019). This is one of the few studies to provide data that suggests that few problems develop in neonates who have bilirubin levels between 85 and 250 µmol/l/.
We have amended the document to include the above information. 6. There is no comment regarding neonates who had both neonatal jaundice and hypoxic ischemic encephalopathy. Asphyxiated neonates would be expected to higher risk for acute bilirubin encephalopathy and long-term problems, but this group is not addressed by the authors.
Because of the flaws mentioned above, the study as currently presented does not add sustainably to our understanding of the true magnitude of neurodevelopmental problems from either neonatal jaundice or hypoxic-ischemic encephalopathy. We did not have these cases in this cohort. None of the participants in this study had a diagnosis of both NNJ and HIE as those with a combined diagnosis were excluded from this cohort study (page 4).

Reviewer 2: Deirdre Murray
1.The definition of hyperbilrubinaemia is very vague. They seem to have included all infants with a bilirubin level > 85. The authors do not give a time for this, beyond stating that the measurement took place in the first 28 days. The majority of infants have some level of jaundice, and will reach this level The bilirubin was measured on admission, since many of the neonates were born at home, it was difficult to determine the exact age in hours. Moreover, the inclusion criteria included neonates up to 30 days of age as per the definition of neonatal period as used in other studies . However, despite the inclusion criteria of 0-30 days, the median age of the participants at admission was 3 [interquartile range (IQR) = 0-8] days (page 6-7). ○ 2. They state that they have done a sub analysis in the severe group, but do not tell us the numbers or the results in this group, who are actually the more interesting.
We have provided the actual numbers of children with hyperbilirubinemia. A subanalysis was conducted with 25 participants with severe hyperbilirubinemia on all outcomes, and similar results were obtained (page 6).
○ 3. Surprisingly with this low level of hyperbilirubinaemia the outcomes are poor which makes me wonder whether these infants had other underlying diagnoses, such as sepsis, prematurity, IUGR?
We have addressed this issue in the methods section as follows: Some of the participants had sepsis and preterm birth as a secondary diagnosis. For NNJ 23 had ○ neonatal sepsis, and 6 were preterm. In the HIE group, 5 had sepsis, and 2 were preterm. However, based on another study that we conducted, sepsis did not appear to aggravate the developmental outcomes of children with neonatal jaundice and sepsis ) (page 5). 4. The follow up rate was low. they state that 347 survivors were identified and 121 were followed up. Please explain why the other 227 were not followed?
This explanation is given under limitations in the discussion section as follows: Given that this study was designed to screen out children with severe disability (they could not be able to carry out tasks during assessment) and only two out of the 107 participants were severely disabled, the prevalence of severe disability in the sample was 1.9% (0.46 -7.32). Data collection on children with severe disability was discontinued (page 10) as we had reached the needed sample size to determine the severity of disability in these children.
○ 5. They have stated in their discussion the other major limitation: the fact that it is likely that a high proportion of children; those with moderate or severe HIE are likely to have died. If the authors could focus on this as a study of outcome following HIE, reporting mortality and survival, with outcome to 10 years then this would be a very valuable article and would add significantly to the literature. Plans to study the mortality and survival of NNJ and HIE are underway in future studies. However, in this study, our main research question in the current study was to understand the neurobehavioral patterns of survivors of NNJ and HIE. ○ 7. The first paragraph of the discussion is repeated twice in the second paragraph. We have revised this part and deleted the repetition as needed.

Reviewer 3: Bolajoko O Olusanya
1.The clinical profile of the participants as neonates is quite deficient and does not provide an objective basis for evaluating the risks of neurodevelopmental disorders. For example, the operational definitions of NNJ used in the study are rarely associated with neurodevelopmental disorders. NNJ is generally benign except in children with or at risk of acute bilirubin encephalopathy (ABE). Since the authors acknowledged inconsistencies in clinical documentation of bilirubin levels at admission (and presumably on discharge also), it would have been useful to identify those who received phototherapy and/or exchange transfusion as proxies for identifying participants with severe NNJ. This is even more crucial in a developing country like Kenya where delays in receiving appropriate care are not uncommon (see Olusanya et al. (2014)1. The purpose of this study was to assess more subtle impairments, such as cognitive impairment and communication since this has not been studied in children surviving these insults in sub-Saharan Africa. Our inclusion of children with TSB>85 µ/mols/l is based on two facts: First, this is the level at which jaundice is reliably detected in the neonate. It is the definition used by the American Academy of Pediatrics for hyperbilirubinaemia, and other authors (Avery, 2005;. Second, there are considerable difficulties in establishing gestational age  and time of birth and the severity of hypoxic-ischemic encephalopathy of neonates admitted to hospitals serving rural areas in sub-Saharan Africa, where most births occur at home. Moreover, there is considerable debate about the criteria for a safe level of bilirubin in sick neonates Varughese, 2019). This is one of the few studies to provide data that suggests that few problems develop in neonates who have bilirubin levels between 85 and 250 µmol/l/.  3. It is unclear why the authors opted for auditory brainstem response (ABR) in these school-aged children rather than pure-tone audiometry which is a more accurate and common measure of auditory threshold, especially in resource-limited settings. The authors need to provide details of the type of ABR and the methodology employed for hearing screening in their population. We have added details on the type of audiometry machine and the methodology employed: The participants were screened for hearing and visual acuity using a pure-tone audiometry machine-Kamplex model R17A AUD Type 3  and the Snellen and E-Chart, respectively. For audiometric testing, first, we talked to the participants while walking towards the sound-proof assessment room to assess how well they are hearing. We then inspected their ear canals using an otoscope. We then instructed the participants to push the button when they hear a sound through the headphones and tested to see if the instructions were clear. We started at 1000 Hz and decreased the level by 10dB until no response was obtained. We then increased the level by 5 Db steps until a reply was captured again. We did these steps until the lowest level at which the participant responded was received. We continued with this procedure at 2000 Hz, 4000 Hz, 500 Hz, 250 Hz, and 125 HZ for both ears. Almost all the participants had normal hearing and vision functioning except one who had mild vision problems (page 4-5). ○