Effectiveness of a locally made eye protector for neonates treated with phototherapy at Queen Elizabeth Central Hospital (QECH), Malawi: a bench study Authors: Gillian Mwale, Thembikile Nyasulu, Tanaka Mudzengerere, Kondwani Kawaza Published: 2026 ABSTRACT AbstractBackgroundThe highest incidence of severe neonatal jaundice occurs in Africa at 667.8 per 10,000 live births. At the Queen Elizabeth Central Hospital (QECH), approximately 75% of neonates have clinical jaundice, with up to 15% requiring treatment. While phototherapy is the standard treatment, the intense blue light poses the risk of retinal damage, which has led to the routine practice of eye protection for neonates receiving this therapy. Nonetheless, QECH, like other low-resource settings, lacks standard eye shields, forcing staff to improvise with different materials to shield the neonate’s eyes. Whether these improvised methods are effective in that function is unknown, however. This study aims to develop a reusable eye shield and evaluate the effectiveness of the various eye protectors used for neonates receiving phototherapy at QECH.MethodsIn this bench study, improvised eye shields used between May and December 2023 were collected and compared to the newly developed eye shield using a gold-standard material. Twenty samples from each type of eye shield were measured using a light metre placed immediately beneath the material to measure the intensity of light irradiance at 10 cm, 15 cm, 25 cm, and 35 cm from the light source. Data were analysed using non-parametric statistical methods.ResultsOwing to the lack of standard eye care, nurses at QECH improvised eye protection for neonates using makeshift materials such as folded gauze, plaster, and surgical face masks. Among these, surgical masks were the least protective, with an average light irradiance of 12.78 µW/cm²/nm (95% CI 12.00–13.56) at the farthest distance. The shielding capacity of gauze and plaster materials was dependent on the thickness of the gauze used, which was determined by the available resources in the ward. The most protective gauze-based shields had an irradiance of 2.0 µW/cm²/nm, with an overall average of 5.96 µW/cm²/nm (95% CI 4.66–7.26). In comparison, the reusable cotton shield and the reference eye shield had no detectable light across all distances, registering a zero µW/cm²/nm light irradiance.ConclusionFor settings like QECH, where resources are limited and standard eye protection is unavailable for neonates undergoing phototherapy, the reusable cotton eye shield presents a viable alternative to improvised materials.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12887-026-07195-6.Keywords: Neonatal jaundice, Eye protection, Phototherapy Background Up to 80% of newborns will develop jaundice in the neonatal period [1]. While it is usually a benign condition, it can become severe and damage the neonate’s developing brain, leading to death or long-term neurological effects such as cerebral palsy [2, 3]. The highest incidence of this severe form of neonatal jaundice occurs in the African region (667.8 per 10,000 live births) [2, 4]. It is, therefore, paramount in this setting to promptly identify and institute therapy in jaundiced neonates requiring treatment. Phototherapy is usually the first line of treatment for hyperbilirubinemia [1]. Bilirubin level for age and gestational age at birth, amongst other factors, guide the commencement of phototherapy treatment [5]. Therapies such as intravenous immunoglobulins and exchange transfusion exist but are often unavailable in low-resource settings [1, 6]. Therefore, most clinically jaundiced neonates in this context receive phototherapy. When used appropriately, the benefits of phototherapy generally outweigh the known risks, which can include dehydration and temperature instability [7, 8]. However, the phototherapy light has potential to cause retinal damage, especially with direct and prolonged exposure [9–13]. The blue light induces free radical formation within the retina, triggering photochemical reactions that may result in cumulative retinal damage, with long-term effects on visual ability [10–12]. While clinically significant retinal damage has not been conclusively demonstrated in infants receiving phototherapy [14], the biological plausibility of photochemical retinal injury underpins the routine practice of shielding the neonate’s eyes during treatment. There is limited research on the appropriate neonatal eye protection for phototherapy. However, several manufactured eye shields are available commercially. These are usually single-use eye shields. Whilst these commercial eye shields are relatively affordable, some low-resource settings cannot provide their neonate population with such eye protection amidst competing resource priorities [6]. One Brazilian facility reported using X-ray film cut-outs and gauze strips taped to the neonate’s eyes to improvise eye shields for neonates undergoing phototherapy [15]. Considering the disadvantages of these methods, the same study developed an eye shield from cotton fabric. A separate study by Abdulkadir et al. demonstrated how when cotton fabric is used as an eye shield, black cotton fabric was more effective in shielding against blue light than other types of cotton [16]. Queen Elizabeth Central Hospital (QECH), located in southern Malawi in sub-Saharan Africa, is one such setting where neonatal jaundice is a common condition, and phototherapy is a frequent treatment modality. Cross sectional studies showed that 75% of neonates admitted to the QECH neonatal ward developed clinical jaundice and up to 15% were instituted on phototherapy treatment [17, 18]. Although shielding the neonate’s eyes during phototherapy is universal practice, the lack of resources, specifically eye shields, often hinders jaundiced neonates from receiving appropriate eye protection during treatment. Commercially available eye shields are unavailable, forcing nurses to improvise with materials like folded gauze, plaster and surgical face masks placed over the neonate’s eyes. Whether or not these improvised materials effectively shield the neonate’s eyes from blue light is unknown. Moreover, these methods utilise materials not designed to act as eye shields, leading to neonatal discomfort and increased hospital waste. This study, therefore, aimed to replace the current practice of neonatal eye protection at QECH by developing a reusable cotton eye shield. Studies whose outcomes included neonatal anthropometry [19–22], specifically head circumference, guided the dimensions of the cotton eye shield. The primary material used for the shield was black cotton as recommended by Abdulkadir et al. [16]. The cotton eye shield was then evaluated for its effectiveness in shielding the neonate’s eyes from phototherapy whilst comparing it to what is currently on the market and the improvised methods used at the QECH. Methods This bench study was conducted in the paediatrics department of the Queen Elizabeth Central Hospital in Blantyre, Malawi, from April to December 2023. The study tested only eye shields and did not include neonates. The study had two phases: the development and testing phases. Development phase Firstly, we reviewed published studies reporting neonatal head circumference at birth among Malawian populations [19–22]. Across these studies, head circumference values typically ranged from 31.8 cm to 36.3 cm. These measurements guided the dimensions of the eye shield and informed the development of multiple size options to accommodate variability in neonatal head size. Cotton fabric was procured locally and tailored according to these dimensions. The portion of the shield covering the eyes consisted of four layers of cotton fabric to maximise light attenuation. An elastic band, covered with an outer layer of cloth, was incorporated to secure the shield around the head while allowing adjustability and comfort. For the purpose of this study, the cotton shields were made to be 34 cm at full stretch, with the eye coverage portion being 12 cm by 4 cm. Considering that Abdulkadir et al. [16] reported a superiority of black cotton in blocking blue light, all eye shields were initially made of black cotton. After engaging the paediatrics department, however, it was found that the black colour may lead to negative cultural perceptions, so we modified the shield to have two inner layers of black cloth and two outer layers of colourful cotton fabric to allow for both the shielding effect of the black, and the paediatric appeal of colourful cotton fabric (Fig. 1). Locally developed reusable cotton eye shield. The eye covering section consists of four layers of fabric (two inner black layers for light attenuation and two outer coloured layers for acceptability), secured with an elastic band (34 cm at full stretch) based on neonatal head circumference data from local anthropometric studies Testing phase Twenty samples of each eye shield type were tested to capture potential variability in materials and construction while remaining feasible for systematic measurement. The investigators collected improvised eye shields used in the paediatric wards for neonates undergoing phototherapy. Nurses typically used clean absorbent gauze rolls (90 cm × 100 m; Hebei Kangji Medical Instrument Co. Ltd.) that were cut, rolled up, and secured over the eyes with medical plaster tape (ABENA, 5 m × 7.5 cm rolls). In other cases, surgical face masks (SAFRAN, 50 piece disposable surgical masks) were folded in half and secured over the eyes. Samples of both types of improvised shields were collected sequentially during the study period until the target sample size of 20 per type was reached. Twenty samples of the newly developed cotton eye shield were selected for testing and labelled as ‘Group A’ for simplicity. The improvised gauze-plaster and surgical face mask shields were labelled Group B and Group C, respectively. Commercial eye shields, ‘Group G’, were then purchased and used as the gold standard for comparison. The commercial shields were White 365 mm × 40 mm Benquan neonatal phototherapy eye shields, produced by Shandong Huashi Benquan Biotechnology Co. Ltd., with product certificate CE/ISO13485. A phototherapy machine (Brilliance Pro, Phoenix Medical Systems) was used to test the eye shields. Its built-in light meter measured irradiance through each material, with values recorded directly from the device display (µW/cm²/nm) as per manufacturer calibration. Measurements of “0” reflect readings displayed by the device and are interpreted as below the measurable range rather than absolute absence of light. The meter is designed to measure irradiance within the blue-light spectrum relevant to neonatal phototherapy, although detailed spectral sensitivity characteristics are defined by the manufacturer. Baseline irradiance (no shield) was measured at 35 cm, 25 cm, 15 cm, and 10 cm from the light source. These distances were selected to cover the range typically used in clinical phototherapy [23–26], spanning the minimum recommended proximity to the light source as well as greater distances commonly observed in practice, to assess shield performance across different exposure conditions. Each shield was tested at these distances with the meter positioned beneath the material to simulate the eye position. Two measurements were taken at each point, and the average value was used for statistical analysis. The cotton eye shields were then subjected to up to 15 consecutive laundry cycles to assess whether repeated washing affected their light-shielding ability, given their reusable nature. They were washed at the maximum machine temperature of 60 °C for 120 min using powdered surf detergent, and then tumble-dried on high heat. After each set of cycles, the shields were ironed. Then, the cotton eye shields were then tested for light irradiance again using the same procedure as before. The amount of light through each eye shield in the four groups was recorded into Microsoft Excel sheets and summarised into tables and graphs. Statistical analysis Statistical analyses were performed using GraphPad Prism Software. For each eye shield, light irradiance measurements were obtained in duplicate at each tested distance (35, 25, 15, and 10 cm), and the mean of the two readings was used for analysis. Data distributions were not assumed to be normal. Therefore, non-parametric methods were used. Comparisons between two independent eye shield groups at each distance were performed using the Mann–Whitney U test. This test assumes independence of observations, a continuous outcome variable, and similarly shaped distributions between groups. For each comparison, the Mann–Whitney U statistic and corresponding two-tailed p-value were calculated, with statistical significance defined as p < 0.05. As comparisons were pre-specified and performed separately at each distance, no adjustment for multiple comparisons was applied. Eye shield groups with no detectable light at all distances were excluded from inferential testing and are reported descriptively. Confidence intervals (95%) for mean irradiance values were calculated using standard parametric methods within GraphPad Prism. Each eye shield was measured at multiple distances; however, statistical comparisons were conducted separately at each distance using shield-level mean values (n = 20 per group per distance). The unit of analysis for each comparison was therefore the individual eye shield at a given distance. While this approach does not explicitly account for within-shield repeated measurements across distances, it was considered appropriate for this bench study where comparisons at clinically relevant distances were pre-specified. No formal adjustment for multiple comparisons was applied, and p-values should therefore be interpreted descriptively. Results A total 80 shields were tested, with 20 eye shields from each group. Groups A and G recorded a light irradiance of 0 µW/cm2/nm (as displayed by the device, indicating no detectable light within the instrument’s measurable range) at all distances. The baseline light irradiance in the absence of a shield is reported in Table 1. Table 1Average unshielded light irradiance at specified distances from the phototherapy light source. Measurements were obtained with the built-in light meter positioned directly under the light source at each distance without any intervening material. The identical irradiance values observed at 10 cm and 15 cm likely reflect limitations of the device readout, such as rounding or saturation at higher irradiance levels, and should therefore be interpreted as approximateDistance from lightLight irradiance mean (µW/cm2/nm)35 cm60.925 cm96.115 cm200.910 cm200.9 Average unshielded light irradiance at specified distances from the phototherapy light source. Measurements were obtained with the built-in light meter positioned directly under the light source at each distance without any intervening material. The identical irradiance values observed at 10 cm and 15 cm likely reflect limitations of the device readout, such as rounding or saturation at higher irradiance levels, and should therefore be interpreted as approximate The gauze-and-plaster eye shields demonstrated considerable variability in transmitted light irradiance, attributed to the differences in thickness, as illustrated in Fig. 2. A very strong negative correlation was observed between shield thickness and transmitted irradiance at all distances (Spearman’s ρ ranging from − 0.94 to − 0.98; all p < 0.0001), demonstrating a consistent reduction in light transmission with increasing material thickness. Furthermore, the measured light irradiance increased as the distance between the phototherapy light source and the shield decreased, with the highest irradiance observed at the shortest distances. Notably, even the thickest gauze-and-plaster combination of 13 mm was still permeable to light with a light irradiance of 2.0 µW/cm2/nm at the furthest distance. On average, the gauze and plaster combinations let through a light irradiance of 5.96 µW/cm2/nm (95% CI 4.66–7.26) at the furthest distance. Supplementary Table 1 shows the specific dimensions for each eye shield in Group B. Light irradiance transmitted through gauze-and-plaster eye shields (Group B) plotted against material thickness (mm) at four distances from the phototherapy light source (10 cm, 15 cm, 25 cm, and 35 cm). Each point represents an individual eye shield measured at a given distance. Colours indicate measurement distance and overlap in points reflects shields with identical thickness. A consistent reduction in transmitted irradiance with increasing thickness is observed across all distances The surgical face masks had the least shielding ability amongst the four groups with an average light irradiance of 12.78 µW/cm2/nm (95% CI 12.00–13.56) at the longest distance. In comparison, the gauze-and-plaster eye shields demonstrated lower transmitted irradiance overall, and this difference between Groups B and C was statistically significant at all distances (Mann–Whitney U test, p < 0.0001). The cotton eye shields had no light detected under each shield, with the light irradiance displayed as 0 µW/cm2/nm, at all distances. After being laundered 15 times each, the cotton eye shields retained the light irradiance of 0 µW/cm2/nm. Figure 3 demonstrates the mean light irradiance at each distance from light of the improvised and cotton eye shields when compared to the commercial eye shield. Mean light irradiance transmitted through each eye shield group at different distances from the phototherapy light source (10 cm, 15 cm, 25 cm, and 35 cm). Error bars represent 95% confidence intervals. The cotton eye shields (Group A) and commercial eye shields (Group G) showed no detectable light transmission at all distances, whereas improvised eye shields (gauze-and-plaster, Group B; surgical masks, Group C) demonstrated higher transmitted irradiance, particularly at shorter distances Discussion The commercial eye shield, which served as a gold standard for neonatal eye protection in this study, was found to have a light irradiance of 0 µW/cm²/nm at all distances. Because the device’s detection threshold is not specified, these readings should be interpreted as below the measurable range of the instrument rather than confirmation of complete elimination of light transmission. Nonetheless, this suggests that under the conditions tested, minimal or no phototherapy light would be expected to reach the neonate’s eyes during treatment, regardless of the distance from the light source. However, in resource-limited settings such as QECH, where such eye shields are unavailable, this is not the case. It was observed that the improvised eye shields made from gauze and plaster had the most varying shielding capacity, owing to the different thicknesses and techniques used by the nurses to make them. For example, during weeks of hospital-wide stock outs, the wards did not have sufficient materials and eye shields would be made with thinner layers of folded gauze and thus would be more permeable to light. Similarly, eye shields made from surgical face masks provided the least amount of protection against the blue light at all distances. Overall, these improvised eye shields fell short of the standard that was the commercial eye shields. This shortfall potentiates the neonate’s eye exposure to blue light for the duration of phototherapy treatment. On the other hand, the cotton eye shields developed in the study were seen to be comparable with what is used in other settings. They had no detectable phototherapy light at all tested distances, even after undergoing several laundry cycles. Although routine hospital sterilisation would typically involve autoclaving at higher temperatures than those used in this study, the absence of light transmission after repeated high-temperature washing and ironing suggests that the shielding material may be compatible with repeated cleaning processes; however, its performance under formal sterilisation procedures such as autoclaving requires further evaluation. Considering the high incidence of severe neonatal jaundice in this region [2], a substantial proportion of affected neonates require phototherapy, for which eye protection is a standard procedure. Although the literature provides limited guidance on specific eye shield design, commercially available products are designed to completely block phototherapy light. Additionally, while it is generally advised to follow manufacturer-recommended distances when placing neonates under the light, some guidelines suggest positioning the infant as close as 10 cm from the source to maximise treatment efficacy [23–25]. This underscores the importance of effective eye protection, particularly at shorter distances where exposure to blue light is greatest. With the cotton eye shields offering the same level of protection as the gold standard in this study, they could be a potential replacement for our current practice at the QECH, and other similar settings where commercial eye shields are not always available. Not only are these shields effective in blocking phototherapy light, but they also allow for materials like gauze and plaster to be used for their intended purposes. Although this study was conducted at a single tertiary hospital in Malawi, the findings may be relevant to other low- and middle-income country (LMIC) settings, where access to commercial neonatal eye shields is limited. Effective phototherapy in these contexts is often constrained by inconsistent equipment availability, variable light intensities, and resource limitations, highlighting the need for pragmatic solutions to protect neonates’ eyes during treatment [27]. The reusable cotton eye shield evaluated in this study was constructed from inexpensive, widely available materials, supporting its potential adaptability across diverse clinical environments. Further evaluation under different phototherapy systems and routine clinical conditions is warranted to confirm generalisability. Ethical considerations The eye shield that was proven to be more effective was lobbied to the department for use to prevent potential retinal damage in these neonates. Limitations Several limitations of this study should be considered. First, the study did not directly assess retinal damage associated with each eye shield, as the infrastructure for real-time retinal evaluation was unavailable. However, as a bench study, the primary aim was to evaluate light transmission rather than clinical outcomes. Secondly, the statistical analysis did not account for repeated measurements across distances for the same eye shields, which may limit independence of observations. Given that distances were treated as distinct experimental conditions in this bench-based evaluation, this approach was considered appropriate. In addition, shielding performance was evaluated using a single phototherapy unit (Brilliance Pro, Phoenix Medical Systems). As phototherapy devices may differ in spectral output and irradiance characteristics, these findings may not be directly generalisable to other systems. Finally, only two replicate measurements per condition were obtained, which may limit measurement precision and contribute to variability in the estimated irradiance and confidence intervals. Conclusion Overall, the cotton eye shields were shown to provide superior performance compared to improvised methods used at QECH in terms of measured transmitted irradiance. For similar settings, where commercial eye shields are not available to protect the jaundiced neonate receiving phototherapy, this reusable cotton eye shield is an effective and reusable alternative. To better protect the neonate from the potential retinal damage and its long-term effects, the cotton eye shield represents a practical and sustainable option for eye protection in jaundiced neonates undergoing phototherapy. Supplementary Information Supplementary Material 1. Acknowledgements Professor Kondwani Jambo, Cecilia Shammim Saidi, Malawi Liverpool Wellcome Programme Training and Development department. Abbreviations Queen Elizabeth Central Hospital Authors’ contributions GM and KK conceptualized the study and designed the protocol. TM, TN, and GM collected and analyzed the data. GM drafted the final manuscript and TM, TN, and KK revised it. Funding The study was funded through philanthropic support from the Hamish Ogston Foundation under the Liverpool School of Tropical Medicine in Collaboration with the Malawi Liverpool Wellcome Programme. Data availability The datasets generated and analysed during the study are available in the Dryad repository, available from: https://docs.google.com/spreadsheets/d/11pUcT8PVvp59JsR3Hfoa-9g47OpIUkohkkwqoKWTarI/edit?usp=sharing. Declarations Ethics approval and consent to participate The study obtained ethical approval from the Kamuzu University of Health Sciences Research and Ethics Committee, previously known as the College of Medicine Research and Ethics Committee. Reference number: U.08/22/3726. The ethical board ensured the ethical use of existing data to guide the make of the developed eye shield and safeguarding principles for all involved parties in this study. The study was carried out in accordance with the Declaration of Helsinki. The study only tested eye shields and did not enrol any neonates. Therefore, informed consent to participate was not required. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The datasets generated and analysed during the study are available in the Dryad repository, available from: https://docs.google.com/spreadsheets/d/11pUcT8PVvp59JsR3Hfoa-9g47OpIUkohkkwqoKWTarI/edit?usp=sharing.