Article Text
PDF
Abstract
Background Since most infants are usually discharged before age 48–72 hours, peak bilirubin levels will almost always occur after discharge. Parents may be the first to observe the onset of jaundice after discharge, but visual assessment is unreliable. The jaundice colour card (JCard) is a low-cost icterometer designed for the assessment of neonatal jaundice. The objective of this study was to evaluate parental use of JCard to detect jaundice in neonates.
Methods We conducted a multicentre, prospective, observational cohort study in nine sites across China. A total of 1161 newborns ≥35 weeks of gestation were enrolled in the study. Measurements of total serum bilirubin (TSB) levels were based on clinical indications. The JCard measurements by parents and paediatricians were compared with the TSB.
Results JCard values of parents and paediatricians were correlated with TSB (r=0.754 and 0.788, respectively). The parents’ and paediatricians’ JCard values 9 had sensitivities of 95.2% vs 97.6% and specificities of 84.5% vs 71.7% for identifying neonates with TSB ≥153.9 µmol/L. The parents’ and paediatricians’ JCard values 15 had sensitivities of 79.9% vs 89.0% and specificities of 66.7% vs 64.9% for identifying neonates with TSB ≥256.5 µmol/L. Areas under the receiver operating characteristic curves of parents for identifying TSB ≥119.7, ≥153.9, ≥205.2, and ≥256.5 µmol/L were 0.967, 0.960, 0.915, and 0.813, respectively, and those of paediatricians were 0.966, 0.961, 0.926 and 0.840, respectively. The intraclass correlation coefficient was 0.933 between parents and paediatricians.
Conclusion The JCard can be used to classify different levels of bilirubin, but it is less accurate with high bilirubin levels. The JCard diagnostic performance of parents was slightly lower than that of paediatricians.
- Jaundice
- Neonatology
Data availability statement
Data are available upon reasonable request. All free text entered will be published.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Bilirubin concentrations in term infants with physiological jaundice peak around the third and fifth day of life, when most affected babies would have been discharged from the hospital. Parents may be the first to observe the onset of severe hyperbilirubinaemia.
Visual assessment alone is not a reliable method for determining the degree of bilirubin.
WHAT THIS STUDY ADDS
The jaundice colour card (JCard) can be used to classify different levels of bilirubin, but the diagnostic efficacy of JCard decreased as the concentration of bilirubin increased.
The JCard diagnostic performance of parents was slightly lower than that of paediatricians.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
The JCard is less accurate with high bilirubin levels. To avoid missed diagnosis of severe hyperbilirubinaemia, we suggested that parents use JCard value 9 to identify severe hyperbilirubinaemia. This should be verified in future studies.
Introduction
Jaundice affects at least 60% of full-term and 80% of preterm neonates during the first days of life.1 2 Severe hyperbilirubinaemia may progress to acute bilirubin encephalopathy or kernicterus, with a significant risk of mortality in newborns.3–6 Worldwide, it is estimated that extreme hyperbilirubinaemia, which is defined as a total serum bilirubin (TSB) level of ≥427.5 µmol/L, affects at least 481 000 late-preterm and term newborn infants annually, resulting in 114 000 deaths and over 63 000 survivors with moderate or severe long-term disability.7 8 Timely and appropriate treatment with phototherapy and/or exchange transfusion is effective to prevent bilirubin encephalopathy in the affected infants.9–11 Bilirubin concentrations in term infants with physiological jaundice peak around the third and fifth day of life, when most affected babies would have been discharged from the hospital. In addition, approximately one in three normal, breastfed infants remains clinically jaundiced at 2 weeks old.12 Hence, mothers may be the first to observe the onset of severe jaundice, especially in low/middle-income countries.10 13 14
Visual estimates alone of the degree of jaundice can be misleading.15 16 Hence, a low-cost screening tool is needed for parents to identify severe neonatal jaundice.17 Transcutaneous bilirubin (TcB) metres are on the market and successfully used by many healthcare professionals, but they are quite expensive. Different icterometers and optical bilirubin estimates obtained with a smartphone camera and processed with a smartphone application have been developed.18–22 Ingram icterometer appears to be the most widely reported tool to assist visual recognition of neonatal jaundice and it has good correlation with serum bilirubin concentrations.23–25 Recently, a two-colour transcutaneous icterometer (Bilistrip) and a novel icterometer (Bili-ruler) had been reported as potential tools for neonatal jaundice screening.19 20 Previously, we designed an eight-colour jaundice colour card (JCard) as an icterometer for parents to visually detect infant jaundice.26 For easier visual discrimination, we simplified it to six colours. In this study, we aimed to evaluate the practicability of the JCard as a potential screening tool to help parents identify neonatal jaundice.
Materials and methods
Study design and patients
We conducted a multicentre prospective study guided by the Standards for Reporting of Diagnostic Accuracy Studies statement.27 Newborns were recruited from nine hospitals in China between 1 October 2019 and 30 September 2021. Inclusion criteria were as follows: (a) newborns hospitalised requiring their serum bilirubin checked or jaundice diagnosis; (b) gestational age ≥35 weeks or age ≤28 days. Exclusion criteria included: (a) skin disease, haemoglobinopathy and severe congenital malformation, and (b) neonates who had received phototherapy.
JCard measurement
The JCard was adjusted from our original eight-colour card (figure 1).26 It was a sheet of thick paper with six blocks of printed colour with precise and graded hues. The colours of the blocks numbered 5, 7, 9, 12, 15 and 18, ranging from lightest to darkest, represented the colour of Han Chinese neonate skin up to severe yellow skin. The units of these numbers were defined as mg/dL. The colour blocks 5, 7, 9 and 15 were consistent with the original colour card. Before the measurement, a study assistant showed the parents how to use the JCard. The newborn was placed in a room with bright natural light, preferably near the window. During the 2-hour period before the TSB determination, the JCard was used by the neonate’s mother or father to measure jaundice at the infant’s cheek. The skin was pressed lightly and checked for signs of jaundice in ‘blanched’ skin, and the colour was compared with the JCard; the user chose the JCard value that most closely matches the underlying skin colour. If the exposed skin colour was darker than the previous colour block and brighter than the next colour block, the middle number was taken as the measurement value. When the skin yellowing depth reached or exceeded the stain 18, 18 was taken as the JCard measurement value. Within 10 min after parental measurement, a paediatrician blinded to the parent’s JCard value used the JCard to measure the degree of jaundice. The JCard was not used for clinical decision-making and no adverse events occurred during this intervention. The card also does not damage the baby’s skin.
Use of the jaundice colour card (JCard) on the cheek of an infant by a father. The skin was pressed lightly and checked for signs of jaundice in 'blanched' skin, and the colour was compared with the JCard.
TSB measurement
The measurement of TSB levels was based on clinical indications, such as jaundice. Blood was obtained by venepuncture, and collecting tubes were shielded from exposure to light. TSB levels were measured in the clinical chemistry laboratories of participating hospitals (online supplemental table 1). Significant hyperbilirubinaemia is defined as a TSB level of ≥205.2 µmol/L and severe hyperbilirubinaemia defined as TSB ≥340.0 µmol/L.
Supplemental material
Patient and public involvement
Members of the public were involved in several stages of the study including design and conduct. We received input from parents and implemented them in our study design. We intend to disseminate the main results to study participants and will seek public involvement in the development of an appropriate method of dissemination.
Statistical analysis
All data were analysed using MedCalc software V.13 (MedCalc Software, Mariakerke, Belgium). Participants missing either a JCard measurement or reference standard TSB test were excluded from the analysis. Simple descriptive statistics were used to calculate the percentage of characteristics of the study population and the mean and SD of the JCard value and TSB. Linear regression analysis was used to determine the association between TSB and JCard values measured by the parents and paediatricians. The Bland-Altman diagram with 95% CI was used to determine whether the difference between TSB and JCard values measured by parents and paediatricians was stable over all the measurement ranges of bilirubin. Since the highest colour block value of the JCard was 18 and the lowest colour block value was 5, children with 85.5 µmol/L≤TSB≤307.8 µmol/L were also analysed. We computed the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and positive and negative likelihood ratios, with 95% CI, for the TSB levels of interest (TSB ≥119.7, ≥153.9, ≥205.2, ≥256.5, ≥342.0 and ≥427.5 µmol/L). Subsequently, we generated a receiver operating characteristic (ROC) curve and compared the area under the curve (AUC) for each using the approach described by DeLong et al.28 We also calculated the intraclass correlation coefficient between the parents’ and paediatricians’ JCard measured values. All tests were two tailed, and statistical significance was set at p<0.05. Using an expected sensitivity of 90% and specificity of 80% for JCard and a 13.6% prevalence of hyperbilirubinaemia in this population,29 we estimated that a minimum sample size of 1017 newborns would be required, within 5% margin of error, at 95% CI.
Results
Participant characteristics
A total of 1161 newborns from nine hospitals were enrolled in this study (online supplemental figure 1). The mean TSB value in the study participants was 241.1±83.8 µmol/L (range 23.9–716.5 µmol/L). A total of 810 newborns had significant hyperbilirubinaemia. The clinical characteristics of the enrolled infants are presented in table 1 and online supplemental figure 2.
Clinical characteristics of the 1161 neonates
Correlation of JCard measurements and bilirubin concentrations
We analysed all children and observed that JCard measurements were highly correlated with TSB levels; the correlation coefficients for parents and paediatricians were 0.754 (0.728–0.777) and 0.788 (0.765–0.809), respectively (figure 2).
Dot plots of JCard values compared with reference standard TSB levels (μmol/L): (A) 5 (88.9±35.9), 7 (133.4±42.8), 9 (201.8±54.7), 12 (234.3±58.1), 15 (259.9±47.9), 18 (306.1±73.5); (B) 5 (78.7±32.5), 7 (114.6±37.6), 9 (152.2±51.3), 12 (212.0±53.0), 15 (256.5±46.2), 18 (309.5±70.1). JCard, jaundice colour card; TSB, total serum bilirubin.
Validity tests to detect different bilirubin levels
The sensitivity, specificity, PPV and NPV of JCard measurements by parents to detect different bilirubin levels are shown in table 2. The JCard value 9 had 95.2% sensitivity and 84.5% specificity for identifying a TSB ≥153.9 µmol/L. The JCard value 12 had 94.1% sensitivity and 72.1% specificity for identifying a TSB ≥205.2 µmol/L. The JCard value 15 had 79.9% and 93.9% sensitivity and 66.7% and 47.3% specificity for identifying TSB ≥256.5 µmol/L and ≥342.0 µmol/L, respectively.
Diagnostic accuracy of JCard measurements by parents for different TSB levels
The sensitivity, specificity, PPV and NPV of JCard measurements by paediatricians for the detection of different bilirubin levels are presented in table 3. The JCard value 9 had 97.6% sensitivity and 71.7% specificity for identifying a TSB ≥153.9 µmol/L. The JCard value 12 had 97.3% sensitivity and 69.0% specificity for identifying TSB ≥205.2 µmol/L. Lastly, the JCard value 15 had 89.0% and 98.8% sensitivity and 64.9% and 42.0% specificity for identifying TSB ≥256.5 µmol/L and ≥342.0 µmol/L, respectively.
Diagnostic accuracy of JCard measurements by paediatricians for different TSB levels
ROC curves
The areas under the ROC curve for parents who identified TSB ≥119.7, ≥153.9, ≥205.2, ≥256.5, ≥342.0, and ≥427.5 µmol/L were 0.967, 0.960, 0.915, 0.813, 0.791, and 0.886, respectively, while 0.966, 0.961, 0.926, 0.840, 0.817, and 0.894, respectively, for paediatricians (figure 3). The AUCs of parents were lower than those of paediatricians in identifying TSB ≥205.2 µmol/L and TSB ≥256.5 µmol/L (figure 3). There were no significant differences between the AUCs of parents and those of paediatricians in identifying TSB ≥119.7 µmol/L, TSB ≥153.9 µmol/L, TSB ≥342.0 µmol/L and TSB ≥427.5 µmol/L (figure 3).
ROC curves for JCard measurement to identify clinically relevant TSB thresholds: (A) ROC curve for JCard values to identify TBS ≥119.7 µmol/L; (B) ROC curve for JCard values to identify TBS ≥153.9 µmol/L; (C) ROC curve for JCard values to identify TBS ≥205.2 µmol/L; (D) ROC curve for JCard values to identify TBS ≥256.5 µmol/L; (E) ROC curve for JCard values to identify TBS ≥342.0 µmol/L; (F) ROC curve for JCard values to identify TBS ≥427.5 µmol/L. AUC, area under the curve; JCard, jaundice colour card; ROC, receiver operating characteristic; TSB, total serum bilirubin.
Agreements of JCard measurements and TSB levels
The Bland-Altman plot means and 95% limits of agreement (LoA) are shown in figure 4. For all newborns, the mean differences between TSB and JCard measurements by parents and paediatricians were −10.4 µmol/L (95% LoA: −120.3 to 99.4 µmol/L) and −2.5 µmol/L (95% LoA: −104.7 to 99.6 µmol/L), respectively. For newborns with 85.5 µmol/L≤TSB≤307.8 µmol/L, the mean JCard measurements by parents were nearly equal to that of TSB (95% LoA: −90.2 to 92.1 µmol/L); however, that of paediatricians was 8.5 µmol/L higher than that of TSB (95% LoA: −72.3 to 89.2 µmol/L).
A Bland-Altman plot of TSB and JCard values of parents and paediatricians: (A) Agreement between TSB and JCard values of parents for all newborns; (B) Agreement between TSB and JCard values of paediatricians for all newborns; (C) Agreement between TSB and JCard values of parents for newborns with 85.5 μmol/L ≤ TSB ≤ 307.8 μmol/L; (D) Agreement between TSB and JCard values of paediatricians for newborns with 85.5 μmol/L ≤ TSB ≤ 307.8 μmol/L. The lines represent mean difference and ±1.96 SD of the difference. JCard, jaundice colour card; TSB, total serum bilirubin.
Inter-rater reliability
The intraclass correlation coefficient was 0.933 (95% CI: 0.925 to 0.940) between parents and paediatricians. Moreover, 59.5% of scores by parents and paediatricians were in exact agreement, and 96.3% of scores by independent readers fell within 51.3 µmol/L (online supplemental table 2).
Discussion
Bilirubin encephalopathy still occurs in some economically underdeveloped areas of China because of the relative lack of medical resources and insufficient screening and monitoring of hyperbilirubinaemia.30 Among the newborns included in this study, 82 had severe hyperbilirubinaemia, of which 64 were aged ≥4 days. One 19-day-old newborn with extreme hyperbilirubinaemia (TSB=716.5 µmol/L) suffered from bilirubin encephalopathy. Although TcB measurement is a convenient, accurate and non-invasive method, it is unlikely that mothers will be offered the use of more reliable but substantially more expensive electronic TcB devices for home use even in high-income countries.31 Few studies have investigated parental use of visual aids to detect jaundice in neonates.19 26 32 33
In several previous studies, researchers have examined the accuracy of visual aids and their usefulness in the diagnosis of neonatal jaundice. Schumacher et al observed a linear correlation between serum bilirubin values and the readings on the icterometer in predominantly white neonates (r=0.63), and a cut-off value of ≥2.5 on the icterometer had a sensitivity and specificity of 73% and 65%, respectively, for identifying neonates with TSB ≥205.2 µmol/L, and sensitivity and specificity of 100% and 58%, respectively, for TSB ≥290.7 µmol/L.23 In a study in Turkey, the correlation coefficient between the TSB and Ingram jaundice metre was 0.78, and an icterometer score ≥3 had sensitivity (100%) and specificity (48%) for identifying TSB ≥222.3 µmol/L.24 Lee et al reported a novel icterometer and showed that the sensitivity and specificity of the nasal Bili-ruler score ≥3.5 for identifying TSB ≥188.1 µmol/L were 84.7% and 83.2%. The areas under the ROC curve for identifying TSB ≥188.1, ≥222.3, and ≥256.5 µmol/L were 0.90, 0.87, and 0.86, respectively.20 In the above studies, jaundice measurements were conducted by health workers. Olusanya et al reported a two-colour jaundice instrument for mothers and showed that it had a sensitivity of 91.4% and specificity of 23.6% for identifying neonates with TSB ≥205.2 µmol/L, sensitivity of 100% and specificity of 20.5% for TSB ≥290.7 µmol/L.19
In this study, we observed that the parental performance of the JCard in detecting TSB was similar to the ranges reported in previous studies: the sensitivity and specificity of the JCard value ≥9 for identifying TSB ≥153.9 µmol/L were 95.2% and 84.5% and the sensitivity and specificity of the JCard value ≥12 for identifying TSB ≥205.2 µmol/L were 94.1% and 72.1%. The areas under the ROC curve of the parents for identifying TSB ≥153.9, ≥205.2, and ≥256.5 µmol/L were 0.960, 0.915, and 0.813, respectively. Our study showed that the ability of parents using JCard to screen for hyperbilirubinaemia is similar to that of health workers or mothers using an Ingram icterometer or similar tools.
Our results showed that AUCs gradually decreased with the increase of JCard score, indicating that its diagnostic performance decreased with the deepening of colour cards. When JCard score was ≥15, the sensitivity and PPV of parents were low, so some infants with hyperbilirubinaemia were missed. Meanwhile, Bland-Altman plots showed that although the difference between the mean values of the JCard and TSB was small, the 95% LoA were large, suggesting that the JCard is not accurate enough to judge the degree of neonatal jaundice. However, we found that the JCard values 9 and 12 had 98.8% and 97.6% sensitivity for identifying a TSB ≥342.0 µmol/L, respectively. The JCard value 15 had 93.9% sensitivity for identifying TSB ≥342.0 µmol/L. The corresponding mean TSB for parental JCard value 9 was approximately 201.8 μmol/L and was equivalent to Ingram icterometer score 3 and Bilistrip B colour card. JCard value 9 is also equivalent to Bili-ruler score 3.5. Several studies used the cut-off values ≥3 on Ingram icterometers and 3.5 on Bili-ruler for identifying neonates with significant jaundice.20 24 34 Therefore, in order to avoid misdiagnosis of severe hyperbilirubinaemia, we suggested that parents bring the baby to the hospital promptly when JCard score is 9 and the paediatricians decides the TcB and TSB examination and treatment of the newborns. Of course, this is for healthy infants without any other associated problems that may increase susceptibility for bilirubin neurotoxicity. Whether this is correct and safe should be analysed in future studies. Additionally, the recommendation to go to the hospital when the score is 9 will result in many hospital visits.
In this study, we compared the predictive ability of JCard used by parents and paediatricians in recognising obvious jaundice and found that 96.3% of scores by parents and paediatricians fell within 51.3 µmol/L and the intraclass correlation coefficient was 0.933 (95% CI: 0.925 to 0.940) between them. There were no differences between parents and paediatricians in AUCs for detecting TSB ≥119.7, ≥153.9, ≥342.0 and ≥427.5 µmol/L. When the cut-off values at TSB were ≥205.2 and ≥256.5 µmol/L, the AUCs of the parents were lower than those of the paediatricians. This showed that although parents and paediatricians have good inter-rater reliability when using the JCard, experienced paediatricians may have more advantages in identifying 205.2 µmol/L<TSB<342.0 µmol/L. However, there was no significant difference between parents and paediatricians in identifying mild jaundice with TSB ≤153.9 µmol/L and severe hyperbilirubinaemia with TSB ≥342.0 µmol/L.
The JCard has some differences from other icterometers. First, its measurement site was different from those of other jaundice metres. During the neonatal period, skin disorders are quite common. Sebaceous gland hyperplasia is the most common cutaneous condition found in neonates (53%–89.4%) and the most common site of location is the nose.35 36 In the previous study, the nose was the recommended measurement site for icterometers, such as Ingram icterometer, Bilistrip and Bili-ruler. Therefore, some neonates with cortisol hyperplasia over the nose may not be suitable to use them to detect jaundice. Previously, we found that the cheek was the best measurement site for the JCard.26 Therefore, the JCard may be applicable to more newborns. Second, compared with the two-colour icterometer, the JCard has six different shades of yellow stripes and can more conveniently help parents judge the approximate degree of jaundice. Third, the production cost of the JCard is about $0.10 per device, which is lower than that of Bili-ruler ($1.0 per device).20
In the analysis of children with TSB ≥342.0 µmol/L, we observed that a newborn with TSB=410.2 µmol/L had a parent JCard value of 8 and had a paediatrician score of 15. It is not ruled out that the mother or father was a patient with colour vision abnormalities. Therefore, we recommend that two parents perform the test at home, and the highest value should be taken.
Our study had several limitations. First, it was hospital based, and the bilirubin level of the participants was generally high, which may not reflect the performance of the general population. Second, the parents who participated in this study were asked verbally if they had ophthalmic diseases such as colour blindness and weakness, but no ophthalmic tests were performed on them. Third, our colour comparison card was designed according to the Han Chinese neonate skin colour and was unsuitable for people with other skin colours. Fourth, our study was conducted at nine hospitals. TSB was not detected using the same biochemical instrument and there were no data of a quality control programme of TSB measurements. Fifth, we primarily recruited term infants and had a low number of preterm infants. Sixth, the JCard’s scores under natural and artificial light have not been compared. Finally, our study was conducted in the hospital, not in the family environment, which may not fully simulate the family setting.
In conclusion, our study shows that the JCard can be used to classify different levels of bilirubin, but it is less accurate with high bilirubin levels. To avoid missed diagnosis of severe jaundice, we suggested that parents use a low JCard score to identify severe hyperbilirubinaemia. Of course, this should be verified in future studies. While awaiting these data, the JCard should be used cautiously before general use in large populations in low/middle-income countries.
Data availability statement
Data are available upon reasonable request. All free text entered will be published.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved by the ethics committee of Wuxi Ninth People’s Hospital affiliated to Soochow University (registered as KT2019004). Participants gave informed consent to participate in the study before taking part.
Acknowledgments
We thank all the newborns and their parents who participated in the study, our dedicated study staff, and all the physicians, nurses, other healthcare providers, laboratory staff, and the staff of the Clinical Trial Management Public Platform of China for providing us with free data management.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors GX designed the study, enrolled participants, did the statistical analysis, interpreted the data, wrote the manuscript and is responsible for the overall content as guarantor. HZ, XD, FX, YLiu, HP, CW, YZhao and HY enrolled participants, collected the data, interpreted the data and revised the manuscript. MR, CM, HL, YaL, RM, LX, NC, XC, JWang, XX, RW, QJ, YoZ, GL, YuL, JK, HZ, YiZ, YY, YawL, YS, JL, SD, QL and JWei enrolled participants, collected data, reviewed the results and revised the manuscript. All authors approved the final manuscript for submission.
Funding The study was funded by Wuxi Ninth People’s Hospital affiliated to Soochow University (grant number 23074702).
Disclaimer The views expressed are those of the authors alone and do not necessarily reflect the views or policies of their respective institutions or organisations with whom they are affiliated.
Competing interests None declared.
Patient and public involvement Members of the public were involved in several stages of the study including design and conduct. We received input from parents and implemented them in our study design. We intend to disseminate the main results to study participants and will seek public involvement in the development of an appropriate method of dissemination.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.