Evaluation of Bilirubin Estimates in Newborns From Smartphone Digital Images in a Population in Botswana

Neonatal jaundice is a condition in newborns with elevated levels of the waste product bilirubin, also known as hyperbilirubinemia. It is primarily caused by a combination of the rapid turnover of red blood cells and the immature liver functions in the newborn. Neonatal jaundice affects approximately 60 to 80 % of newborns but is usually self-limiting and harmless. The condition is still potentially dangerous because bilirubin can accumulate in the basal ganglia of the brain, where it can cause permanent brain damage. Such brain damage, better known as kernicterus, can manifest as cerebral palsy, deafness, language difficulty, or in the worst cases, be fatal.

This study will be conducted at a hospital in Botswana, where the study aim to collect data from newborns with a broad range of bilirubin levels and high melanin content.

The investigators hypothesize that the Picterus JP will offer bilirubin level estimates that are highly correlated to bilirubin levels measured in blood samples in a population in Botswana. Following our hypothesis, the short-term goal for this project is to demonstrate that this screening method can be used to identify children with severe neonatal jaundice independent of skin color. Thus, providing a cheaper, patient-friendly, and more readily available means of neonatal jaundice detection.

The study will be a diagnostic accuracy cross-sectional study with quantitative methods of data collection.

The study population will comprise 150 newborns from the Neonatal Unit and the postnatal ward at Princess Marina Hospital, a number determined by a sample size calculation based on standard errors of the 95% limit agreement, as described by Bland and Alman.

The study objective of this study is two-folded. First, to determine the correlation between TSB measurement and Picterus image estimate of serum-bilirubin. Second, to determine the accuracy of the Picterus App used as a screening tool.

Following informed consent, background data such as birth weight, age on examination, gestational age, and type of feeding will be obtained. The skin type of the infant will be classified according to the newborn Neomar's scale score.

Transcutaneous bilirubin estimates will be performed over the sternum of the infant. A Dräger Jaundice Meter JM-105 will be used in this study.

Skin reflectance will be done using a CE-marked portable Konica Minolta spectrophotometer CM-700d. The spectrophotometer readings will be used to adjust the existing algorithm from Picterus to African skin and thereafter bilirubin estimates from the taken images will be determined.

The Picterus calibration card will be fixed in place with a disposable back sticker on the newborn's chest with the hole of the card placed over the infant's sternum. A validated smartphone with Picterus JP will then be used to collect digital images, as shown below. Once the phone is aligned with the card, two sets of 6 images of each newborn will be captured, 3 with flash and 3 without flash. After all the images are obtained, a unique ID will be displayed on the smartphone. This ID will be recorded on the case report forms and later used to pair clinical data and digital images.

A blood sample to determine TSB will be obtained within 60 minutes of obtaining the images and processed at the Department of clinical biochemistry in the hospital laboratory. In addition, TSB will be measured with neonatal bilirubin analyzer Neo-Bil Plus from DAS to assess the quality of the equipment. For the hospital laboratory analysis, 1 ml of blood is required and will be obtained either by venous puncture or heel prick. An additional 10 μl volume of blood will be collected in a heparinized capillary tube for analysis in the Neo-Bil Plus equipment.

The bilirubin estimates from the images taken will be determined. The images will be compared to the TSB, NeoBil and TcB measurements using the Pearson correlation coefficient. Sub-analysis for the different skin colors, gestational ages, days of life, and bilirubin levels will be performed for the TSB and image estimate. Sensitivity and specificity analysis will be calculated for different cut-off values using the ROC analysis. Systematic over- or under-estimation of bilirubin levels will be evaluated using Bland-Altman plots. To avoid bias, the developer of the product - Picterus AS - will not be involved in statistical work-up or even see the raw data.

The results from this study will have relevance for the potential validation of the innovation. Studies demonstrating the tool's potential in a high-burden setting will be necessary for users of the device. This is the first validation study following the study in Uganda. The results of this study will result in a published paper in a peer-reviewed journal. Further, the study's results will be communicated through popular media and seminars in both countries.