Outcomes of neonatal hyperbilirubinemia and associated factors at a tertiary hospital in Ghana

Abstract

Introduction

Neonatal Hyperbilirubinemia (NH) is one of the most common clinical presentations in the first week of life.

Aims

This study sought to determine the incidence, clinical presentation and risk factors associated with mortality, among neonates presenting with unconjugated NH at the Komfo Anokye Teaching Hospital in Ghana.

Methodology

This was a hospital-based prospective analytical cohort study conducted from October to December 2020. A total of 223 newborns admitted to the nursery with NH were recruited, mothers were interviewed, and patients were followed-up until discharge. Data was collected and analyzed with R statistical software. Associations were determined as crude and adjusted odds ratios using univariate and multivariate binomial logistic regression.

Results

The incidence of NH was 244 per 1000 admissions (95% CI: 21.7–27.3) with a case fatality rate of 8.1% (95% CI: 4.8–12.4). Independent predictors of increased mortality were older age at admission [AOR = 1.28(1.05–1.57), p = 0.015], gestational age between 35 and 37 weeks [AOR = 4.89(1.04–25.8), p = 0.047], and the presence of abnormal posture at admission [AOR = 6.40(1.37–35.9), p = 0.023]. The presentations associated with survival till discharge were primiparity in the mother [AOR = 0.06(0.01, 0.32), p = 0.003], and exclusive breastfeeding at the time of admission [AOR = 0.09 (0.02–0.46), p = 0.004].

Conclusion

The incidence of hyperbilirubinemia among neonates admitted to the Mother Baby Unit at KATH is high and gestational ages between 35 and 37 weeks, abnormal posture and older age at presentation were identified to be significantly associated with mortality.

Introduction

Neonatal hyperbilirubinaemia (NH) is one of the most common clinical presentations in the first week of life [1]. Approximately 60% of term infants and 80% of preterm infants present with unconjugated hyperbilirubinemia during the neonatal period [2]. This condition results from the accumulation of unconjugated bilirubin in the blood due to an imbalance between bilirubin conjugation and excretion [1]. Rapid accumulation of bilirubin poses a unique problem since an elevated level of unconjugated bilirubin is potentially toxic to the developing central nervous system resulting in severe and irreversible long-term consequences or even death [1].

NH is estimated to be highest in the African region, with rates of 667.8 per 10,000 live births reported in two systematic reviews, compared to incidences of 3.2 per 10,000 and 4.4 per 10,000 live births in Europe and the Americas, respectively [3, 4]. Higher prevalence rates have been reported in some countries with one teaching hospital in Ghana reporting a rate of 32.9% [5].

NH is one of the leading causes of admissions in the first week of life worldwide [1], and delays in presentation or in the initiation of its treatment could result in bilirubin encephalopathy - related complications and mortality. In spite of the presence of risk factors at birth in most cases, the condition also tends to be one of the major causes of readmissions during the neonatal period [6], which emphasizes the need for health professionals to promptly identify and monitor such neonates before discharging them home too early. The high readmission rate for NH after an initial early discharge [6] also highlights the need for a more aggressive assessment of newborns before early discharge [7].

In order to achieve the Sustainable Development Goals in Low- and Middle-Income Countries (LMICs) like Ghana, it is important to reduce the neonatal mortality rate which has stagnated for some time. Reducing deaths and disabilities from NH is one way of achieving this goal [8]. Unfortunately, there are no national figures for NH in most LMICs, including Ghana. Since NH contributes significantly to both neonatal morbidity and mortality, it is vital to have representative data for developing countries to inform resource allocation. Most studies done are retrospective making it difficult to estimate the incidence of NH. This study thus sought to determine the incidence of NH at one of the biggest teaching hospitals in Ghana. Additionally, we aimed to determine the risk factors associated with adverse outcomes specifically death, among neonates presenting with unconjugated hyperbilirubinaemia to the Komfo Anokye Teaching Hospital (KATH) in Ghana.

Methodology

Study design

This was a hospital-based prospective analytical cohort study conducted from October 2020 to December 2020.

Study site

The study was conducted at KATH, the main tertiary hospital responsible for the referral needs of the middle to northern belts of Ghana. KATH is a 1200-bed-capacity second-largest hospital in the country and is located in the capital city of the Ashanti Region. The Mother Baby Unit (MBU), a 130-cot capacity unit, the setting for the study, serves as both a 24-hour emergency/intensive as well as continuous care setting for newborns up to 2 months old. The unit has five paediatricians/neonatologists who are complimented by other health cadres. Admissions are made up of about 55% inborn deliveries with the remaining 45% being referrals. The number of admissions per month averages between 400 and 500 neonates [9], and the services provided include phototherapy, exchange blood transfusion (EBTs), and mechanical ventilation. The hospital, however, does not have the capacity to provide in-utero therapy to fetuses that develop hyperbilirubinemia.

Study population

Infants aged less than or equal to 28 completed days who had unconjugated hyperbilirubinemia requiring phototherapy or EBT based on the United Kingdom (UK) National Institute for Health and Clinical Excellence (NICE) Guidelines [10, 11] were included in the study. Neonates requiring inpatient admission for monitoring of total serum bilirubin (TSB) levels were also included. Infants with conjugated hyperbilirubinemia defined as serum conjugated fraction > 15% of the TSB or known previous episodes of unconjugated hyperbilirubinemia were excluded from the study.

Sample size

Our primary objective was to determine the factors associated with mortality in newborns with hyperbilirubinemia and our secondary objective was to determine the incidence of NH. In the absence of a previous study indicating the mortality rate in this cohort, we assumed a rate of 15%. To determine a minimum difference in prevalence of any factor of 25%, a minimum of 202 babies with neonatal jaundice was required for the study using a type I error rate of 0.05 and a power of 80%.

Sample size calculation

The prevalence rate of neonates with unconjugated hyperbilirubinaemia at the Neonatal Unit is 25%, and the anticipated study population for this study was 400–500 per month. Consequently, in consideration of the population (less than 10,000), a maximum rate of 500 multiplied by the intended number of months for the study was used in calculating the sample size.

In using the formula:

$$\:\varvec{n}=\frac{{\varvec{Z}}^{2}\varvec{p}\left(1-\varvec{p}\right)}{{\varvec{d}}^{2}}$$

where $\:Z\:=\:1.96$, $\:p\:=\:0.25$ and $\:d\:=\:0.06$, we obtain $\:n\:=\:200.083\approx\:201$

n is the required sample size;

Z is the Z statistic for the level of confidence at 5%, type 1 error = 1.96;

p is the estimated prevalence; the expected proportion in population based on previous study;

$$\:q\hspace{0.17em}=\hspace{0.17em}1\:\--\:p;$$

d is the precision of the estimate.

The expected sample population was 500 for each of the intended 4 months of study = 2000,

therefore, for this population which is less than 10,000 neonates, the expected sample size for this population was:

$$\:\frac{n}{1+\frac{n}{N}}=\:\frac{201}{1+\frac{201}{2000}}=182.64\approx\:183$$

With an anticipated attrition rate of 10%, the final sample size will be 202.

Study Variables

Study variables included sociodemographic factors of mother and baby, clinical characteristics and laboratory parameters as shown in Tables 1, 2 and 3.

Study procedure

There were two categories of neonates recruited into the study; those who were already on admission when jaundice developed with the jaundice identified by a healthcare professional, and those who presented to the unit with signs of hyperbilirubinemia during the study period. All babies presenting to the unit were received at the triaging area. An initial assessment was performed on the patient, which involved ensuring that the airway was patent and maintainable, breathing was spontaneous and regular, circulation adequate and patient stabilized. A complete history was then obtained and a thorough physical examination was carried out. Neonates who were suspected of having hyperbilirubinemia were taken through a physical examination where the skin, eyes and mucous membranes were inspected for a yellow discoloration, by blanching the skin on the forehead, gums, palms of the hands and soles of the feet. Clinical assessment included checking and documenting the vitals; the heart and respiratory rates, temperature (> 37.5⁰C considered as fever) and oxygen saturation. All temperatures were measured via the axillary route and only pre-ductal oxygen saturation was recorded with the Mindray VS600 patient monitor.

A Ballard score assessment was done for those less than 72 h old to ascertain their gestational ages, and baby’s weight was checked and documented. Baseline laboratory tests included the TSB, complete blood count, glucose 6 phosphate dehydrogenase (G6PD) screening and blood grouping. A baby was deemed to meet the eligibility criteria if upon plotting the TSB value on the UK NICE guidelines bilirubin threshold graphs [12], the baby needed phototherapy or phototherapy and EBT. Once that was confirmed, the parents/caregivers were approached, the study explained to them and informed consent was obtained for participation in the study. This study used the UK NICE guidelines which are widely accepted for use in all gestational ages as a reference guide for determining when to initiate and discontinue phototherapy or to perform an EBT, since the study setting was in a tertiary hospital and therefore admits neonates of all gestational ages.

All potential participants for the study were selected consecutively. Afterwards, the caregiver was interviewed using a structured case report form to gather data on sociodemographics, the clinical history of the current illness of the baby, whether or not baby had received only breastmilk, accompanying clinical features and other relevant history. Neonates enrolled into the study were followed up each admission day, all the laboratory results traced, and the relevant data was collected up until discharge. Upon discharge, the mother/caregiver counselled again about the condition and a date given for review before being sent home. Neonates who underwent EBT were referred for follow-up for audiology and neurodevelopment.

Data management and statistical analyses

The total number of admissions of neonates into the MBU over the study period was used as the denominator for calculating the incidence of NH. Basic descriptive analysis stratified by mortality was done for quantitative continuous variables such as the baby’s age at presentation, birth/current weight and maternal age. The various neonatal and maternal clinical and demographic features were stratified by outcome and compared using the Pearson’s chi-square test, Fisher’s exact or the Wilcoxon rank sum test as appropriate. Factors that had a p-value of < 0.2 were selected and subjected to a multivariate stepwise logistic regression to select the best model to predict mortality. In so doing the variance inflation factor was first used to determine the collinearity between the various predictors. For the final model, a p-value of 0.05 was considered as a statistically significant relationship.

Results

There was a total of 913 admissions from October 2020 to December 2020, with 225 neonates admitted and managed for NH. Out of these 225 neonates, 223 including 9 sets of twins were recruited into the study, with two of the mothers unable to give consent to be part of the study, because the fathers were unavailable to help make the final decision for consent. This gives an incidence of 244 per 1000 admissions (95% CI: 21.7–27.3) for NH at the KATH MBU.

Out of the 223 neonates, 18 (8.1%, 95% CI: 4.9–12.4) died with the remainder being discharged home.

Males constituted 63.2% while the birth weight of the majority of the neonates ranged between 2.5 kg to less than 4 kg (52.5%) or < 2.5 kg (44.8%). The median (IQR) age at admission into the study was 22 h (1–80) with 95% exclusively breastfed. The most common neurological observation was poor suck which was observed in 23.3% of neonates. Thirty-three per cent (33.3%) had ABO incompatibility, 8.4% rhesus incompatibility and 38.1% G6PD deficiency. A total of 11.7% had EBT done on them as shown in Table 1. The median (IQR) age of the mothers was 29 years (25, 34) with 22% being single. 5.8% had no formal education with 21.5% being unemployed (Table 2).

The univariate neonatal factors associated with mortality were weight on admission, age at admission, gestational age, fever present at admission, poor suck/inability to suckle, seizures, abnormal tone, abnormal posture, having a shrill cry, lethargy, and elevated total serum bilirubin. The only maternal factor associated with a positive neonatal outcome was primiparity.

In the multivariate model, a primiparous mother (AOR = 0.06, 95% CI: 0.01–0.32, p = 0.003), and exclusive breastfeeding (AOR = 0.09, 95% CI: 0.02–0.46, p = 0.004) were protective against mortality (Table 3). Conversely, age at admission (AOR = 1.28, 95% CI: 1.05–1.57, p = 0.015) and gestational age between 35 and 37 weeks (AOR = 4.89, 95% CI: 1.04–25.8, p = 0.047) were significantly associated with mortality compared to those with gestational age 38 completed weeks and above. Additionally, the presence of abnormal posture (AOR = 6.40, 95% CI: 1.37–35.9, p = 0.023) was significantly associated with neonatal mortality (Table 3).

Babies with weights between 2.5 and 4.0 kg on admission, admission age above 4 days, gestational age of 38 weeks and above, fever, poor suck, seizures, abnormal muscle tone, and posture, and a TSB of 446µmol/L at presentation were all determined to be significantly associated with death. (Table 3).

Table 1 Relationships between newborn outcomes and clinicopathological features

The median (IQR) age for the mothers in the study was 29 years (25, 34), with less than a quarter of the mothers in the study being single 50 (22.4%). With the exception of infants of primiparous mothers being more likely to be discharged home, none of the maternal sociodemographic, clinical or obstetric characteristics were determined to be significantly linked to discharge or death.

Table 2 Relationships between newborn outcomes and maternal characteristics

After multivariate regression, primiparity and exclusive breastfeeding were significantly linked to neonatal survival, whereas older age at admission, gestational age of 35–37 weeks, and the presence of abnormal posture were significantly associated with neonatal mortality.

Table 3 Predictors of mortality in patients with neonatal hyperbilirubinemia

Discussion

The incidence of NH, which is 24.4% in this study, falls within the incidence range of 23–60% determined by a multicenter study conducted in Nigeria across 10 neonatal units [13]. This is also comparable to the 28.1% determined in a systematic review examining the prevalence of the condition in SSA [4].

On the other hand, the incidence determined in the present study was lower than the 37.3% and 45.6%, reported by Lake et al. from Mekelle City, Northern Ethiopia [14], and Abd Elmoktader et al. from Faiyum, Egypt [15] respectively. In these studies, however, different, relatively lower and fixed TSB cut-off values were used for the determination of NH that qualified for admission and management. Also, late preterm babies were excluded by Lake et al. Perhaps, the differences in TSB cut-off values and the exclusion of late preterm babies by Lake et al. could explain the significant differences in the incidence rates. A South African study [16] done at a national district hospital determined the prevalence of NH to be 55.2% and even though they employed the use of the NICE guidelines for their management, the transcutaneous bilirubinometer (TCB) was used to measure the bilirubin levels of their neonates. However, the use of the TCB in measuring the bilirubin levels of neonates has been known to overestimate serum bilirubin levels, especially in black infants [17, 18], hence this could have led to overestimation of the number of neonates with this condition which could explain the high incidence they reported. This study used TSB, which is the gold standard for measuring the serum bilirubin levels for all neonates [19].

There was a male preponderance of 63.2% and this figure is consistent with the 62.6% reported by Asefa et al. [20] in northern Ethiopia and the 62.5% reported by Osuorah et al. [21] from southeastern Nigeria. The higher frequency of NH in male infants is consistent with findings in other similar studies. This male preponderance effect perhaps gives credence to the theory that male newborns tend to have higher levels of bilirubin [22].

The median age at admission was 22 h, which is less than what Bizuneh et al.¹⁷ determined, with a median (IQR) admission age of 3 ± 2 days in a case-control study done in Ethiopia, and the 2 ± 2 days determined by Asefa et al. [20] in another Ethiopian study. The study by Bizuneh et al. [23] employed specific TSB cut-off ranges for term and all preterm infants (> 205 µmol/L in term neonates and > 257 µmol/L in preterm neonates) regardless of their gestational and postnatal ages. Perhaps this could explain the slightly higher admission ages, especially since the different gestational ages among preterm infants are likely to build up their TSBs to the cut-off ranges at different rates. This current study used different TSB cut-off values based on the gestational and postnatal ages.

Gestational age was significantly associated with mortality, with neonates with gestational ages of 38 weeks and more tending to have a higher chance of dying in this study. Even though prematurity is a recognized risk factor for NH [1, 24], a major contributing factor to the reduced association with mortality determined in this study may be because most of these babies are more likely to be on admission in the first couple of weeks of life thereby increasing the probability of early detection and prompt initiation of treatment for the NH. Since term infants are more likely to be discharged home earlier, there is the likelihood of the baby developing the NH at home, thereby increasing the chances of the condition being missed, resulting in late presentation in a severe state. This could also be a potential explanation for the birthweight category of 2.5-<4.0 kg being significantly related to an increased risk of mortality since most term infants tend to fall within this weight category.

Exclusive breastfeeding was found to be significantly associated with neonatal survival in this study with 93.7% of mothers practising it at the time of reporting to the hospital, which was on averagely within the first week of the baby’s life. Two Ghanaian studies reported prevalence rates of 63.4% and 61.1% for mothers who initiate exclusive breastfeeding within an hour of life [25, 26]. This is much lower than the 93.7% identified in this study and could be attributed to the target group of mothers that were questioned in both studies. These studies focused on mothers of infants aged between 0 and 59 months and between 0 and 24 months, hence there could have been challenges with recall bias, especially among mothers with older infants in the group, thereby resulting in underestimation of the rates. This study questioned mothers of infants who were less than a month old, hence making it easier for them to recall information correctly, possibly explaining the high rate found in this current study. However, despite the many documented benefits of breastfeeding, some studies have found a strong association between breastfeeding and an increased risk of NH [27, 28]. Numerous studies have determined that supporting optimal breastfeeding practices and encouraging mothers to continue breastfeeding, even in the face of hyperbilirubinemia far outweighs the harm, hence the need not to interrupt breastfeeding [28,29,30]. A longitudinal birth cohort study done in China revealed that strict exclusive breastfeeding done for the first 6 months of life in babies with severe NH may reduce the potential neurological damage caused by severe NH [31].

Fever, abnormal muscle tone, inability to suckle, abnormal posture, seizures, lethargy, and shrill cry were all significantly associated with neonatal mortality in this study. Most of these clinical features are consistent with signs of bilirubin encephalopathy [32] which is of concern as this indicates that a significant proportion of neonates in our study presented late to the health facility by which time complications had already set in. Osuorah et al. [21] in a study conducted at a Nigerian teaching hospital involving 48 neonates with severe hyperbilirubinemia, identified refusal to suckle, fever, and depressed or absent primitive reflexes as the most common clinical features on presentation. These findings are consistent with bilirubin encephalopathy and similar to what was found in the present study. The median (IQR) age of mothers in our study was 29 years ranging from 25 to 34 years, and this is very similar to the mean maternal age of 29.3 years reported by Asefa et al. in Ethiopia [20], a median maternal age of 30 years by Ekwochie et al. in Nigeria [33], and a mean maternal age of 29.4 years by Ameghan Aho et al. in Ghana [34]. A plausible explanation could be that this age range is the peak reproductive age bracket hence the majority of neonates will be born to mothers in this category.

Almost half of the mothers (48.4%) in the study were earning less than the prevailing minimum wage during the study period, at which time the monthly minimum wage in Ghana was approximately GhS320 (US $50) [35]. Even though education, income and employment status are powerful factors associated with maternal health-seeking behaviours [36, 37], these were not found to be significantly linked to the baby’s outcome in this study.

Infants of primiparous mothers with hyperbilirubinemia in this study (45.3%) were determined to be more likely to be discharged home. This is much higher than the 26.9% reported by Ameghan-Aho et al. [34] in another Ghanaian study but significantly lower than the 78.2% reported by Demis et al. [38] from Ethiopia. However, in the study by Ameghan-Aho et al., the researchers recruited expectant mothers from a referral centre as well as a primary health clinic whereas, this study recruited mothers from just one big referral hospital. The Ethiopian study recruited from six major referral hospitals which could have likely resulted in mothers with a history of maternal and/or newborn problems before, hence their higher figure.

After multivariate regression, primiparity and exclusive breastfeeding were determined to be associated with good neonatal outcomes whereas admission age, gestational ages between 35 and 37 weeks and abnormal posture were associated with an increased risk of neonatal mortality in this study. Babies who were older on admission were determined to be more likely to die compared to those who were brought in early. This could perhaps be attributed to a lack of awareness of the signs and symptoms of hyperbilirubinaemia and succedent late presentation.

Early preterms have more compromised liver, however, these babies are more likely to be on admission during the period when they are most at risk of newborn jaundice. This allows for early detection and prompt intervention by healthcare workers hence improving survival among these babies. The approach to the management of late preterm infants tends to be similar to the term infant even though their livers are still not as mature as that of term infants, and their ability to conjugate and excrete bilirubin is compromised [22]. Despite this, similarities in the size, weight and other physical characteristics between late preterm and term infants imply that they are likely to be discharged early thereby putting them at a higher risk of late detection and delayed initiation of treatment for certain neonatal danger signs such as NH [39]. These could explain why this category of preterm infants was identified to have an increased risk of death among neonates with hyperbilirubinemia.

Since abnormal posture is a recognized symptom of bilirubin encephalopathy [32], it is no surprise that its presence in this study was significantly associated with an increased risk of neonatal mortality. This could be another indication of the lack of knowledge of the early signs and symptoms of neonatal hyperbilirubinaemia and poor health-seeking behaviour of mothers/ caregivers for their newborns.

Limitations

This study is a single-center / facility-level study; hence the generalization of the results might not be appropriate. Recognized risk factors like UGT1A1 polymorphisms could not be measured in the current study.

Conclusion

In conclusion, the incidence of hyperbilirubinemia among neonates admitted to the Mother Baby Unit at KATH is high, with about one in every 4 babies presenting with the condition. The median admission age was significantly higher in those who died compared to patients who survived. Gestational age, the presence of abnormal posture and older age at admission were significantly associated with mortality, with neonates with a gestational age between 35 and 37 weeks having a higher chance of dying in this study. Exclusive breastfeeding and primiparity were however determined to be significantly associated with discharge/survival.

Recommendations

• There is the need to avoid early newborn discharge, especially for late preterm infants. This will reduce the incidence of missed cases of NH and also reduce the number and severity of cases presenting late.

• There is a need for health education during antenatal visits with emphasis on the benefits of exclusive breastfeeding, how to help mothers identify signs of NH and the importance of good health–seeking behaviour.