Diagnostic accuracy of newborn foot length measurement in identifying low birth weight and preterm infants in Sidama Region, Ethiopia Authors: Fitsum Weldegebriel Belay, Fanuel Belayneh Bekele, Akalewold Alemayehu, Rekiku Fikre, Andrew Clarke, Sarah Williams, Hannah Richards, Yohannes Chanyalew Kassa Published: 2025 ABSTRACT AbstractBackgroundLow birth weight (LBW) and preterm birth significantly contribute to increased risks of neonatal mortality, particularly in sub-Saharan Africa. Early identification of and timely intervention for these vulnerable infants remains challenging. Simple anthropometric measurements, such as foot length (FL), have emerged as potential alternatives to identify small and vulnerable newborns. This study aims to assess the accuracy of FL as a screening tool to identify LBW and preterm babies and to establish appropriate cut-off points for the local context.MethodsA facility-based cross-sectional study was conducted at Hawassa University Comprehensive Specialised Hospital and Bushulo Mother, Newborn and Child Health Specialty Centre and Primary healthcare from 15 May 2024 to 6 July 2024; 396 mother-newborn dyads were included using a consecutive sampling technique. Data were collected through interview, medical record review and FL measurement using digital calliper. Birth weight was measured using calibrated digital weight scales. Descriptive statistics, correlation analysis and receiver operating characteristic (ROC) curves were used to assess the distribution of measured characteristics and determine optimal FL cut-off points for identifying LBW and preterm babies.ResultsThe study included 61 (15.4%) LBW and 48 (12.3%) preterm babies. FL showed a strong correlation with birth weight (r=0.79) and moderate correlation with gestational age (r=0.62). The optimal cut-off point for identifying LBW babies was 70.65 mm, yielding 93.4% sensitivity, 91.6% specificity and an area under receiver operator curve (AUC) of 0.97 (95% CI: 0.95 to 0.98). For preterm babies, the optimal cut-off was 69.95 mm, with 76.7% sensitivity, 91.1% specificity and an AUC of 0.91 (95% CI: 0.86 to 0.96).ConclusionFL has demonstrated excellent diagnostic accuracy to identify LBW and preterm newborns. Optimal accuracy of identifying LBW and preterm babies in similar populations can be achieved at 70.65 and 69.95 mm cut-off points, respectively. In designing FL-based screening tools, appropriate cut-off points for the required level of accuracy should be considered.Keywords: Neonatology, Health Policy, Developing Countries WHAT IS ALREADY KNOWN ON THIS TOPIC Globally, 15%–20% of all newborns are low birth weight (LBW) or preterm. Globally, 15%–20% of all newborns are low birth weight (LBW) or preterm. Neonatal death accounts for nearly half of all under-5 mortalities, with the highest rates in sub-Saharan Africa. Neonatal death accounts for nearly half of all under-5 mortalities, with the highest rates in sub-Saharan Africa. In Ethiopia, many newborns are not weighed at birth mainly because of high home delivery rates (46%). In Ethiopia, many newborns are not weighed at birth mainly because of high home delivery rates (46%). Evidence suggests that anthropometric measurements like foot length can be used to identify LBW and preterm babies. Evidence suggests that anthropometric measurements like foot length can be used to identify LBW and preterm babies. Previous studies suggest that the optimal cut-off point for using foot length to identify preterm and LBW babies varies across populations. Previous studies suggest that the optimal cut-off point for using foot length to identify preterm and LBW babies varies across populations. WHAT THIS STUDY ADDS Further evidence about the correlation between foot length, LBW and prematurity. Further evidence about the correlation between foot length, LBW and prematurity. Specific cut-off points that can be used for designing foot length-based screening tools in the study area. Specific cut-off points that can be used for designing foot length-based screening tools in the study area. At the optimal cut-off point, foot length showed excellent accuracy of 93.4% sensitivity, 91.6% specificity and area under receiver operator curve (AUC) 0.973 to identify LBW. At the optimal cut-off point, foot length showed excellent accuracy of 93.4% sensitivity, 91.6% specificity and area under receiver operator curve (AUC) 0.973 to identify LBW. The diagnostic accuracy to identify preterm was 76.7% sensitivity, 91.1% specificity and AUC 0.91. The diagnostic accuracy to identify preterm was 76.7% sensitivity, 91.1% specificity and AUC 0.91. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Further research is justified into utilisation of foot length measurement to screen for LBW and premature newborns born at home in Ethiopia. Further research is justified into utilisation of foot length measurement to screen for LBW and premature newborns born at home in Ethiopia. It informs selection of foot length cut-off points to be used considering sensitivity and specificity to screen for small babies in the study area. It informs selection of foot length cut-off points to be used considering sensitivity and specificity to screen for small babies in the study area. Generalisable cut-off points applicable for screening across wider populations need to be better understood and explored. Generalisable cut-off points applicable for screening across wider populations need to be better understood and explored. Introduction Low birth weight (LBW) is defined as birth weight below 2500 g and preterm as an infant delivered at a gestational age below 37 weeks. Globally, LBW and preterm newborns account for 15%–20% of all births.1 Child mortality has declined worldwide during the past three decades. However, the reduction in neonatal mortality has been slow, and neonatal deaths now contribute almost half (47%) of all under-5 mortalities. Neonatal mortality also varies markedly across regions and countries. The highest rate is in sub-Saharan Africa, with 27 neonatal deaths per 1000 live births.2 Preterm birth complications are the leading cause of death in children under the age of five.2 Preterm and LBW infants have a twofold to 10-fold increased risk of morbidity and mortality.3 These conditions are associated with stillbirth, neonatal complications (eg, preterm respiratory distress, hypothermia and hypoglycaemia), childhood stunting and an increased risk of chronic diseases (eg, cardiovascular disease).4,6 Saving the lives of small and vulnerable newborns begins with timely identification of preterm and LBW babies who need extra care. Birth weight measurement with calibrated weighing scales can enable timely provision of life-saving interventions7 8 and was one of the focal areas in the Every Newborn measurement improvement roadmap.9 10 Childbirth in a well-equipped health facility and skilled birth attendance is a key strategy to reduce maternal and newborn mortality.11 Despite this, about half (46%) of deliveries in Ethiopia take place outside of health institutions.12 As a result, a high number of infants do not get weighed at birth.13 In addition, LBW and preterm newborns delivered in health facilities where weighing scales are not functional or not available are unlikely to be identified and referred for additional lifesaving care.14 Also, gestational age often cannot be calculated because the mothers are not adequately assessed during pregnancy.15 Simple, low-cost instruments that are suitable for resource-limited settings are urgently required to ensure that these vulnerable infants are identified and referred for additional care. Simple and reliable anthropometric tools to identify preterm and LBW infants have been studied and tested. In a systematic review16 conducted on feasibility and diagnostic accuracy of neonatal anthropometric measurements to identify LBW and preterm infants, foot length (FL), mid-upper arm circumference, and head and chest circumferences had comparable diagnostic accuracy in identifying LBW and preterm babies. The results also showed FL measurement performed best for identifying preterm infants, with 82% sensitivity, 89% specificity and area under receiver operator curve (AUC of 0.91) at 7.2 cm optimal cut-off point. In identifying LBW babies, an FL of 7.3 cm was calculated as the optimal cut-off point (AUC of 0.89). There was also variation noted on the optimal cut-off points used across studies.17,21 FL may be the most feasible and safe method of measurement in low resource settings, as these anthropometries are simple and minimise risk of infant exposure to cold. The use of anthropometric parameters as screening tools for LBW and preterm babies is contextual, but to an extent that is not well understood.22 In designing an appropriate FL screening tool to identify small babies at birth, it is important to establish the appropriate cut-off point for the setting. The findings from this study will be used to design FL-based screening tools for community level identification of small babies born at home in southern Ethiopia. This hospital-based study aimed to assess the diagnostic accuracy of newborn FL to identify preterm and LBW infants and establish the optimal cut-off length for screening in this population. Methods and materials Study area and design A facility-based cross-sectional study was conducted at Hawassa University Comprehensive Specialised Hospital (HUCSH) and Bushulo Mother, Newborn and Child Health Specialty Centre and Primary healthcare (Bushulo PHC) from 15 May 2024 to 6 July 2024. These facilities are in Hawassa town, Sidama region, 273 km from Addis Ababa, Ethiopia. The facilities provide specialised maternal and newborn care and have neonatal intensive care units (NICUs). Eligibility criteria and study population The study included all mother–newborn dyads delivered at HUCSH and Bushulo PHC, as well as referred cases to these facilities within the study period. Mother–newborn pairs were excluded from the study if the mother did not consent to participate, and if the measurements were performed more than 24 hours after birth. Newborns with congenital malformations that could potentially affect FL measurement were also excluded from the study, as were seriously ill mothers who were unable to participate in the study within the 24 hour period after birth. Sampling Buderer’s formula23 was used to calculate the required sample size by using the following assumptions and estimates of FL diagnostic accuracy. We considered LBW prevalence of 14.0%; 95% confidence limit; 10% acceptable width; 84.2% sensitivity and 73.9% specificity to identify LBW infants.24 Based on these, we determined a final sample size of 407. The study employed a consecutive sampling technique to include mother–newborn dyads who fulfilled the inclusion criteria until the sample size was obtained. Data collection We employed three different data collection methods for this study. First, an interviewer-administered questionnaire was used to collect information from the mothers. Then, medical records were reviewed to extract relevant clinical information about the mothers and newborns. Finally, trained data collectors measured the FL of the newborns using a digital calliper. Data collectors recorded data using an electronic data collection form designed in Kobo Toolbox, a free data collection and management platform. The electronic form was deployed and accessed by a mobile phone application called ‘Kobo Collect’. Supervisors also used Kobo Toolbox to monitor the progress of data collection and forms submitted by the data collectors. Data collection and FL measurements were performed by three nurses and one midwife. The data collectors received intensive training for 3 days before the actual data collection. The principal investigators supervised the data collection on a daily basis. Measurements Birth weight was measured by calibrated digital infant weight scales. Healthcare providers attending the delivery performed the weight measurement and recorded the birth weight on the mother’s medical record to the nearest 10 g. The data collectors recorded the birth weight and clinical information from medical charts onto the Kobo electronic data collection form uploaded on their mobile phone. LBW was defined as a baby born weighing less than 2500 g. Preterm birth (premature birth) was defined as a baby born before 37 weeks gestation, based on the gestational age estimated by either early ultrasound or last menstrual period (LMP) methods. FL was measured by the maximum length of the right foot from the edge of the posterior point of the mid heel to the tip of the longest toe. The FL measurements were performed by trained data collectors using a digital FL measuring calliper. The FL measurement was recorded to the nearest 0.1 mm. Data analysis The data set was first exported as a SPSS file and transferred to Stata V.16.0 statistical software for analysis. Frequencies, percentages, means and SD were used for descriptive statistical analysis. Pearson’s correlation coefficients and simple linear regression analyses were used to assess the correlation of FL measurement with birth weights and gestational age. To determine the optimal FL cut-off points in identifying preterm and LBW infants, receiver operating characteristic (ROC) curve analyses were performed. The ROC analysis for preterm identification included infants whose gestational age (GA) was estimated by ultrasound and LMP methods. The area under the receiver operating characteristic curve (AUC) was used to measure the overall accuracy of the FL measurements in identifying LBW and preterm babies. Statistical significance was declared at p-value 0.05. Ethical considerations Prior to participation, mothers were provided with comprehensive information about the study’s purpose, and informed written consent was obtained from each participant. Information collected from mothers and their medical records was documented anonymously to protect confidentiality. Patient and public involvement There was no involvement of patients or the public in the design, conduct, reporting or dissemination plans of our research. Results Sociodemographic characteristics Three hundred and ninety-six mother–newborn dyads were included in the study with a response rate of 97.3%. Maternal age was normally distributed with a mean of 26.9 years and +5.1 years SD. There were 162 (40.9%) mothers in the age group 25–29 years and 93 (23.5%) were 20–24 years old. Mothers from urban areas were 331 (83.6%) of the total participants and almost all, 394 (99.49%) were married. About one-third, 122 (30.8 %) of the women had completed elementary level education, 107 (27.0%) had secondary educational level and 92 (23.2%) mothers had a first degree. One hundred and seventy-five (44.19%) of the mothers were housewives, with more than half, 232 (58.6%) from households with four to six family members (table 1). Government employees account for 75 (18.9%) of the women. Table 1Sociodemographic characteristics of the mothersCharacteristicsFrequencyPer centAge group <20 years235.8 20–24 years9323.5 25–29 years16240.9 30–34 years8321.0 35 years and above358.8Place of residence Urban33183.6 Rural6516.4Marital status Single10.3 Married39499.4 Widowed10.3Educational status of the mother No formal education61.5 Elementary education (grades 1–8)12230.8 Secondary education (grades 9–12)10727.0 Certificate and diploma5814.7 Degree9223.2 Masters112.8Occupation of the mother Student143.5 Farmer30.8 Merchant9624.2 Government employee7518.9 NGO employee102.5 Daily labourer61.5 Housewife17544.2 Other174.3Family size <48922.5 4–623258.6 >67518.9FL, foot length; NGO, non-governmental organisation. Sociodemographic characteristics of the mothers FL, foot length; NGO, non-governmental organisation. Characteristics of the mothers and newborn One hundred and twenty (30.3%) of the participating women were primigravida. Including the current birth, mothers with one live birth comprised 136 (34.3%), those with two 115 (29.0%) and those with three 70 (17.7%). Of the 260 mothers who had more than one live birth, 15 (5.8%) had LBW babies previously. The majority of mothers, 390 (98.5%) attended one or more antenatal care (ANC) during pregnancy. Of these, 234 (60.0%) had four to seven, and 106 (27.2%) had eight or more ANC visits. During their pregnancy, 374 (94.4%) mothers received iron-folic acid supplements (table 2). Table 2Characteristics of mothers and newbornVariablesFrequencyPercentageTotal number of pregnancies One12030.3 Two10125.5 Three7318.4 Four5714.4 Five and above4511.4Total number of live births One13634.3 Two11529.0 Three7017.7 Four4511.4 Five and above307.6LBW babies during previous deliveries (n=260) Yes155.8 No/do not know24594.2ANC attendance during current pregnancy Yes39098.5 No61.5Frequency of ANC visit during current pregnancy (n=390) 1–2 times143.6 3 times369.2 4–7 times23460.0 8+ times10627.2IFA intake during current pregnancy Yes37494.4 No225.6Place of delivery Health facility39298.9 Home41.1Birth companion* Husband36191.2 Household family members24862.6 Friends15438.8 Relatives24662.1 No one10.3Newborns from the current delivery Single37494.4 Twins194.8 Multiple30.8Sex of the screened newborn Male18647.0 Female21053.0*Multiple response.ANC, antenatal care; IFA, iron folic acid; LBW, low birth weight. Characteristics of mothers and newborn Multiple response. ANC, antenatal care; IFA, iron folic acid; LBW, low birth weight. Nearly all mothers, 392 (98.9%) gave birth in a health facility and 361 (91.2%) were accompanied by their husbands during birth. Three hundred and seventy-four (94.4%) deliveries were of single babies (table 2). Gestational age ranged from 27 to 44 weeks, with a mean of 38.83 weeks (SD+2.28), while birth weight was distributed by mean and SD of 3108.94 (+666.9) g. Birth weight ranges from a minimum of 1090 g to a maximum of 4700 g. Median FL measurement was 75.2 mm with IQR of 31 mm. The minimum FL measurement was 47.6 mm, while the maximum was 88.7 mm. Sixty-eight (17.5%) of the newborns were either preterm or LBW. Of the 389 mothers, those who delivered at 37 or more weeks of gestational age were 341 (87.7%), while 48 (12.3%) delivered preterm. Three hundred and thirty-five (84.6%) babies had birth weights within normal range, and 61 (15.4%) were classified as LBW (figure 1). Distribution of LBW and preterm infants. LBW, low birth weight. Almost all, 389 (98.2%) mothers attended delivery in the selected facilities. More than half, 224 (56.6%) of the mothers gave birth by caesarean section (CS), whereas 172 (43.4%) by spontaneous vaginal deliveries. From 389 mothers who delivered in the selected facilities, 340 (87.4%) had no medical complications. Among those who encountered multiple medical complications, pre-eclampsia was the most frequent case (18 (4.6%)). There were 54 (13.6%) newborns admitted to NICU. The study utilised the most accurate method to estimate gestational age when more than one method was recorded in the mothers’ medical chart. Of these, ultrasound gestational age estimation was recorded for 243 (62.5%) participants and LMP for 136 (34.9%). New Ballard Score and Fundal height, 5 (1.3%) were less commonly employed methods to estimate gestational ages (table 3). Table 3Maternal and neonatal medical record reviewVariableFrequencyPercentageDelivered at the selected facilities Yes38998.2 No71.8NICU admission Yes5413.6 No34286.4Type of delivery SVD17243.4 CS22456.6Medical complication (n=389) No complication34087.4 Pre-eclampsia184.6 Oligohydramnios71.8 APH61.5 premature rupture of membranes61.5 Placenta previa41.0 Other*143.6Method used to estimate GA Ultrasound24362.5 LMP13634.9 Fundal height51.3 New Ballard Score51.3*Breech presentation, retained placenta, seizure, diabetes mellitus, anaemia, hepatitis infection, HIV infection, malaria, GA.APH, acute postpartum haemorrhage; CS, caesarian section; GA, gestational age; LMP, last menstrual period; NICU, neonatal intensive care unit; SVD, spontaneous vaginal delivery. Maternal and neonatal medical record review Breech presentation, retained placenta, seizure, diabetes mellitus, anaemia, hepatitis infection, HIV infection, malaria, GA. APH, acute postpartum haemorrhage; CS, caesarian section; GA, gestational age; LMP, last menstrual period; NICU, neonatal intensive care unit; SVD, spontaneous vaginal delivery. Diagnostic accuracy of FL measurement to identify LBW and preterm babies Linear regression analysis showed a strong correlation between FL and birth weight (r=0.79, R2=0.63) and a moderate correlation with gestational age (r=0.62, R2=0.38). The FL digital measurement also significantly predicted both birth weight (beta coefficient=83.83, p<0.001) and gestational age (beta coefficient=0.23, p<0.001) (table 4). Table 4Correlation and simple linear regression analysis of FL with birth weight and GA parameters of neonatesPredictionCorrelation coefficient (95% CI)R2Beta-coffSEP valueBirth weight by FL0.79 (0.73 to 0.85)0.6383.833.23<0.001Gestational age by FL0.62 (0.53 to 0.68)0.380.230.015<0.001FL, foot length; GA, gestational age. Correlation and simple linear regression analysis of FL with birth weight and GA parameters of neonates FL, foot length; GA, gestational age. At the optimal 70.65 mm cut-off point, FL measurement was found to identify LBW babies at sensitivity of 93.4% and specificity of 91.6%. It also showed excellent diagnostic accuracy with AUC value 0.973 (0.96–0.99). The optimal cut-off point for FL measurement to identify preterm babies was 69.95 mm. At this cut-off, FL can identify preterm babies with a sensitivity of 76.7%, specificity of 91.1% and AUC value 0.91 (0.86–0.96) (figure 2).In addition table 5 summarizes the diagnostic accuracy of FL to identify preterm and LBW babies at different cut-off points(table 5). ROC curve showing diagnostic accuracy of FL digital calliper measurement to identify LBW (left) and preterm (right) babies. AUC, area under receiver operator curve; FL, foot length; LBW, low birth weight; ROC, receiver operating characteristic. Table 5Diagnostic accuracy of FL at serial cut-off points to identify preterm and LBW babiesOutcomeCut-off (mm)Sn%(95% CI)Sp%(95% CI)PPV%(95% CI)NPV%(95% CI)AUC%(95% CI)LBW≤6554.1(49.2 to 59.0)99.1(98.2 to 100)91.7(88.9 to 94.4)92.2(89.6 to 94.9)76.6(70.3 to 82.9)≤6657.4(52.5 to 62.2)98.8(97.7 to 99.8)89.7(86.8 to 92.7)92.7(90.2 to 95.3)78.1(71.8 to 84.4)≤6763.9(59.2 to 68.7)98.8(97.7 to 99.8)90.7(87.8 to 93.6)93.8(91.4 to 96.1)81.4(75.3 to 87.5)≤6877.0(72.9 to 81.2)97.9(96.5 to 99.3)87.0(83.7 to 90.3)95.9(93.9 to 97.8)87.5(82.1 to 92.8)≤6980.3(76.4 to 84.2)96.1(94.2 to 98.0)79.0(75.0 to 83.0)96.4(94.6 to 98.2)88.2(83.1 to 93.3)≤7088.5(85.4 to 91.7)93.7(91.3 to 96.1)72.0(67.6 to 76.4)97.8(96.4 to 99.3)91.1(86.9 to 95.4)≤7193.4(91.0 to 95.9)90.1(87.2 to 93.1)63.3(58.6 to 68.1)98.7(97.6 to 99.8)91.8(88.3 to 95.3)≤7296.7(95.0 to 98.5)83.3(79.6 to 86.9)51.3(46.4 to 56.2)99.3(98.5 to 100)90.0(87.0 to 93.0)≤7398.4(97.1 to 99.6)74.3(70.0 to 78.6)41.1(36.2 to 45.9)99.6(98.9 to 100)86.3(83.5 to 89.2)Preterm≤6553.5(48.5 to 58.5)98.2(96.9 to 99.5)79.3(75.2 to 83.4)94.3(91.9 to 96.6)75.6(68.3 to 83.4)≤6655.8(50.8 to 60.8)97.6(96.1 to 99.1)75.0(70.6 to 79.4)94.5(92.2 to 96.8)76.7(69.2 to 84.3)≤6758.1(53.2 to 63.1)97.0(95.3 to 98.7)71.4(66.9 to 76.0)94.8(92.5 to 97.0)77.6(70.1 to 85.1)≤6865.1(60.3 to 69.9)95.2(93.1 to 97.4)63.4(58.8 to 68.5)95.5(93.4 to 97.6)80.2(72.9 to 87.5)≤6967.4(62.7 to 72.2)93.1(90.6 to 95.7)55.8(50.8 to 60.8)95.7(93.7 to 97.8)80.3(73.1 to 87.5)≤7076.7(72.5 to 81.0)90.8(87.9 to 93.7)51.6(46.5 to 56.6)96.8(95.1 to 98.6)83.7(77.2 to 90.3)≤7179.1(74.9 to 83.2)86.6(83.2 to 90.0)43.0(38.1 to 48.0)97.0(95.3 to 98.7)82.8(76.4 to 89.2)≤7286.0(82.6 to 89.5)80.1(76.0 to 84.1)35.6(30.8 to 40.4)97.8(96.4 to 99.3)83.1(77.4 to 88.7)≤7388.4(85.1 to 91.6)71.1(66.6 to 75.7)28.1(23.6 to 32.7)97.9(96.5 to 99.4)79.7(74.3 to 85.2)AUC, area under receiver operator characteristics curve; FL, foot length; LBW, low birth weight; NPV, negative predictive values; PPV, positive predictive value; Sn, sensitivity; Sp, specificity. Diagnostic accuracy of FL at serial cut-off points to identify preterm and LBW babies AUC, area under receiver operator characteristics curve; FL, foot length; LBW, low birth weight; NPV, negative predictive values; PPV, positive predictive value; Sn, sensitivity; Sp, specificity. The diagnostic accuracy of the FL measurement to identify LBW babies did not show a significant difference between male and female newborns. The AUC value of FL measurement to identify LBW babies was 0.99 (0.98–1.0) for male and 0.95 (0.91–0.98) for females (figure 3). Similarly, the optimal cut-off point and accuracy of FL to identify preterm infants did not show a statistically significant difference across the gestational age estimation methods used. ROC curve showing diagnostic accuracy of FL digital calliper measurement to identify LBW for male (left) and female (right) newborns. AUC, area under receiver operator curve; FL, foot length; LBW, low birth weight; ROC, receiver operating characteristic. Discussion Our study revealed that FL is a robust predictor of both birth weight and gestational age. FL demonstrated a strong positive correlation with birth weight and a moderate correlation with gestational age. Regression analysis further confirmed FL as a significant predictor for these outcomes, and it exhibited excellent diagnostic accuracy in identifying LBW infants with sensitivity of 93.4%, specificity of 91.6% and AUC value 0.973 (0.957–0.989) at an optimal cut-off point of 70.65 mm. A high diagnostic performance in identifying preterm births was also found, with a sensitivity of 76.7%, specificity of 91.1% and AUC value 0.91 (0.86–0.96). The correlation between FL and birth weight in our study was in line with the reports from studies conducted in Ethiopia18 and Uganda.21 However, a stronger correlation between FL and birth weight was observed in our study when compared with other studies conducted in Africa.2025,31 This might be explained by the difference in measurement techniques used. The studies in line with our findings used a FL calliper and hard ruler, while the other studies used measuring tape. As our findings from the previously published meta-analysis showed, FL measurements conducted by hard ruler and calliper have better accuracy than measuring tape. The strength of the correlation between FL and gestational age in our study was similar to findings from a previous study conducted in Dire Dawa city, Ethiopia.29 However, our finding was lower than other studies conducted in Northwest Ethiopia and South Africa.18 32 Unlike the methods used to estimate gestational age in our study, the study conducted in Northwest Ethiopia used New Ballard Score estimations performed by senior paediatric residents.18 The study conducted in South Africa included only newborns whose weight was appropriate for their gestational age and those who had an early antenatal ultrasound performed before 23 weeks of gestation.32 This methodological difference, along with the variation of methods used for estimation of gestational age, may explain the discrepancies. Our finding of the correlation between FL and gestational age was also higher than findings from studies conducted in Northeast Ethiopia and Tanzania.20 33 Similarly, the difference in FL measurement techniques used by these studies (flexible tape and ruler) and the digital calliper methods used in our study could explain the discrepancy. The overall accuracy of FL in identifying LBW babies as measured by AUC value in our study concurred with similar reports by previous studies.21 29 31 However, the accuracy observed in our study was better.2024,26 28 In addition, none of the previous studies utilised the digital calliper measurement method we employed in our study, which could explain the higher accuracy we observed. These studies also reported different optimal cut-off points. This indicates a need to have context-specific evaluation of the most appropriate anthropometric measurement before using it to screen for LBW babies. However, the feasibility of undertaking this process multiple times at very localised levels is likely to be challenging in real-world applications and may present a barrier to mainstreaming and routine adoption. The extent to which cut-off points generated from data derived from regional populations may be generalised and still used effectively to screen newborns with sufficient sensitivity and specificity is not well understood. However, further work in this field will help in assessing the viability of such tools for integration into standard practice. The diagnostic performance of FL measurement in identifying preterm infants observed in this study was comparable with previous findings.24 29 34 However, a study conducted in Northwest Ethiopia reported slightly higher overall accuracy of FL in identifying preterm infants.18 Unlike ours, this study excluded extremely preterm (gestational age <28 weeks old) newborns and took repeated FL measurements to reduce measurement errors. Differences in inclusion criteria and the approach used to improve reproducibility of FL calliper measurement could explain these discrepancies. This study used a standard digital calliper for FL measurement and incorporated different parameters to assess diagnostic performance and the linear relationship between FL measurement and birth weight and gestational age. One limitation of our study is the accuracy with which we could determine gestational age in order to assess its correlation with FL. The majority of newborns’ gestational age was calculated based on early ultrasound, a highly accurate method of estimating the length of a woman’s pregnancy. This was followed by LMP, which is a widely accepted method. Ballard score and fundal height were applied to less than 2% of newborns and excluded from the ROC analysis. While this means that we cannot calculate the exact gestation age for a few of the newborns in our study, we still have a good indication of gestational age and its correlation with FL based on the widely accepted methods. In addition, our study was conducted in a health facility setting, though the intended use of this screening tool is at community level where babies are born at home, or in low-level health facilities without access to a standard infant weighing scale. Finally, our study considered the diagnostic performance of the FL measurement only within 24 hours after birth. In conclusion, FL measurement showed excellent diagnostic accuracy in identifying LBW and preterm newborns. To design effective FL-based screening tools for identification of small babies in the study area, we suggest that specific cut-off points should be considered based on the required level of sensitivity and specificity, as well as the target end-point users such as health professionals or lay mothers and family members. Optimal accuracy cut-off points for FL of 70.65 mm for LBW and 69.95 mm for preterm infants can be used as a reference point to modify from. These thresholds serve as critical markers, below which newborns are likely to be classified as LBW or preterm, respectively. Future studies should focus on implementing and evaluating these FL-based screening tools at the community level to assess their effectiveness and feasibility. Further research is also needed to explore the extent to which cut-off points can be generalised and applicable across different populations, while retaining adequate sensitivity and specificity to be effective as screening tools. Acknowledgements We would like to aknowledge Save the Children International for providing funds and Hawassa University College of Medicine and Health Sciences for facilitating the collaboration for conducting this research. We would also like to aknowledge Mr. Tony Anderton and Marlin Engineering for their technical support in the development of the prototype tools. Footnotes Funding: This study was supported by Save the Children International. Patient consent for publication: Not applicable. Ethics approval: This study involves human participants. Ethical clearance for this study was obtained from the Institutional Review Board (IRB) of College of Medicine and Health Sciences, Hawassa University (ref. #: IRB/084/16). Participants gave informed consent to participate in the study before taking part. Provenance and peer review: Not commissioned; externally peer reviewed. Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting or dissemination plans of this research. Data availability statement Data are available upon reasonable request. References Data Availability Statement Data are available upon reasonable request.