Anthropometric measurements can identify small for gestational age newborns: A cohort study in rural Tanzania

Background: Small-for-gestational-age (SGA) is associated with increased neonatal mortality and morbidity. In low and middle income countries an accurate gestational age is often not known, making the identification of SGA newborns difficult. Measuring foot length, chest circumference and mid upper arm circumference (MUAC) of the newborn have previously been shown to be reasonable methods for detecting low birth weight (< 2500 g) and prematurity (gestational age < 37 weeks). The aim of this study was to investigate if the three anthropometric measurements could also correctly identify SGA newborns. Methods: In the current study from a rural area of northeastern Tanzania, 376 live newborns had foot length, chest circumference, and MUAC measured within 24 h of birth. Gestational age was estimated by transabdominal ultrasound in early pregnancy and SGA was diagnosed using a sex-specific weight reference chart previously developed in the study area. Receiver operating characteristic curves were generated for each of the anthropometric measurements and the area under the curve (AUC) compared. Operational cutoffs for foot length, chest circumference, and MUAC were defined while balancing as high as possible sensitivity and specificity for identifying SGA. Positive and negative predictive values (PPV and NPV) were then calculated. Results: Of the 376 newborns, 68 (18.4%) were SGA. The AUC for detecting SGA was 0.78 for foot length, 0.88 for chest circumference, and 0.85 for MUAC. Operational cut-offs to detect SGA newborns were defined as ≤7.7 cm for foot length, ≤31.6 cm for chest circumference and ≤ 10.1 cm for MUAC. Foot length had 74% sensitivity, 69% specificity, PPV of 0.35 and NPV of 0.92 for identifying SGA. Chest circumference had 79% sensitivity, 81% specificity, PPV of 0.49 and NPV of 0.95 for identifying SGA. Finally, MUAC had 76% sensitivity, 77% specificity, PPV of 0.43 and NPV of 0.94 for identifying SGA. Conclusion: In a setting with limited availability of an accurate gestational age, all three methods had a high NPV and could be used to rule out the newborn as being SGA. Overall, chest circumference was the best method to identify SGA newborns, whereas foot length and MUAC had lower detection ability. Trial registration: Clinicaltrials.gov (NCT02191683). Registered 2 July 2014. The study was part of the FOETALforNCD study carried out in Korogwe and Handeni Districts, northeastern Tanzania, a predominantly rural area with Korogwe District Hospital as the main health facility. Women were screened for enrolment between July 2014 and March 2016, and follow-up was completed in December 2016 (Fig. 1). Flow chart of the study population The FOETALforNCD study consisted of two cohorts; a preconceptional cohort that enrolled women before pregnancy who were followed throughout pregnancy if they conceived, and a pregnancy cohort that enrolled pregnant women with a GA of ≤14 weeks. In the preconceptional cohort, the inclusion criteria were age between 18 and 40 years, no current use of modern family planning except condoms, a negative pregnancy test, not having a baby under 9 months of age, not having tried to conceive unsuccessfully for > 2 years consecutively and having consented to attend all the antenatal care and giving birth at Korogwe District Hospital if they conceived during the study period. In the pregnancy cohort women were included if having a GA ≤14 weeks and based on a 1:1 distribution of maternal anemia (hemoglobin < 11 g/dL) and no maternal anemia (hemoglobin≥11 g/dL). Pregnancy was confirmed with a urine human Chorionic Gonadotropin test (One step pregnancy test strip; Vista Care Company, Shandong, China with a sensitivity of 25 mlU/ml human Chorionic Gonadotropin) followed by transabdominal ultrasound for GA estimation (Sonosite TITAN and Turbo, US High resolution, Bothell, WA, USA) using crown-rump-length in the 1st trimester [18] and head circumference in the 2nd trimester [19]. Data collection throughout pregnancy was similar in the two cohorts and was conducted in the local language Swahili and then documented into case report forms in English. At enrolment data on demographics, socio-economic status, and medical and obstetric history were collected. After enrolment in the pregnancy part of the study scheduled visits were conducted at GA of 11–14, 20–22, 26–28, 32–34 and 37–39 weeks. At every contact, a medical examination was done including blood pressure (R-champion N, Rudolf Riester, Jungingen, Germany), anthropometric measurements, urine dipstick (Combiscreen 7, Alere), hemoglobin level (venous blood using Sysmex KX-21 N hematological analyzer, Sysmex Corporation Kobe, Japan) and malaria screening (using malaria rapid diagnostic tests (ParaHIT, Span Diagnostics, Gujarat India or CareStart® Access Bio NJ, USA), blood smear for malaria microscopy, and dried blood spot for malaria species diagnostic PCR). Hypertensive disorders were defined as systolic blood pressure ≥ 140 mmHg and/or diastolic blood pressure ≥ 90 mmHg observed on at least two measurements taken > 4 h apart or systolic blood pressure ≥ 160 mmHg and/or diastolic blood pressure ≥ 110 mmHg observed at least once. Preeclampsia was defined as hypertension and proteinuria on at least two occasions after a GA of 20 weeks. Anthropometric measurements included height in centimeters (cm) (only at enrolment) (Stadiometer, SECA GmbH & Co. KG, Hamburg, Germany), and weight without outer garment and shoes recorded to nearest 0.1 kg (Digital weighing scale, SECA GmbH & Co. KG, Hamburg, Germany) [20]. Body mass index was calculated by dividing the body weight with the square of the height. At birth a thorough anthropometric examination of the newborns was performed within 24 h. Birth weight was measured on a nude newborn using a digital baby weighing scale (M107600, ADE, Germany) and noted in grams (g) to the nearest 5 g [21]. The length of the newborn was measured from the vertex to the heel of the right foot using an infantometer (Baby Infantometer 417, SECA GmbH & Co. KG, Hamburg, Germany) [21]. The foot length was measured from the heel to the tip of the longest toe on the right foot using a hard transparent plastic ruler and noted in cm [9, 11, 13]. Chest circumference and MUAC were measured with a flexible non-stretchable tape measure. Chest circumference was measured to the nearest 0.1 cm on a calm baby (mid-expiration) by circling the chest at the level of the nipples [22]. The MUAC was measured to the nearest 0.1 cm at the right upper arm mid-point halfway between the acromion of the scapula and the olecranon of the ulna [20]. All measurements were performed twice. If the difference between two measurements exceeded 50 g for birthweight, 7 mm for length, 5 mm for chest circumference, and 2 mm for foot length or MUAC, a third measurement was done and the two measurements closest to each other were documented. APGAR score, the appearance of amniotic fluid and any congenital malformations were documented as well. A sex-specific weight reference chart previously developed in the same study area [17] was used to define SGA as a birth weight below the 10th percentile. LBW was defined as birth weight < 2500 g [7] and preterm as a GA < 37 weeks [7]. In addition, the premature newborns were defined as extremely preterm (< 28 weeks), very preterm (28 to < 32 weeks) and moderate to late preterm (32 to < 37 weeks) [23]. Ethical approval was granted by the Medical Research Coordinating Committee of the National Institute for Medical Research in Tanzania (reference number NIMR/HQ/R.8a/Vol. IX/1717). All study participants gave written informed consent (or thumbprints from illiterate women) before enrolment. All study participants were given unique identification numbers to ensure anonymity, and only authorized personnel had access to the data. All project activities were conducted in accordance with Good Clinical Practice and the Declaration of Helsinki. Participants were assisted by the project in obtaining the best local medical care available if a disease was diagnosed during the study period. All data were checked for consistency, double entered, and validated using Microsoft Office Access 2007 database (Microsoft Corporation, Redmond USA). All analyses were performed using R version 3.4.3 (2017-11-30), Copyright (C) 2017 The R Foundation for Statistical Computing. Only singleton, liveborn newborns at a GA > 22 weeks without severe congenital malformations and with neonatal examination done within 24 h were included in the analysis. The study population was described with mean and standard deviation for parametric continuous data, median and interquartile range for non-parametric data, and proportion for categorical data (number (%)). Finally, for GA at delivery and birth weight the range with minimum and maximum values were reported and the 2.5th and 97.5th percentiles for birth weight were calculated. The association between foot length, chest circumference and MUAC versus birth weight and GA at delivery was calculated with Pearson and Spearman correlations, respectively. Student’s t-test was used for comparison of mean foot length, chest circumference and MUAC for SGA, LBW, and preterm newborns compared to normal weight and term newborns as well as for comparison of mean foot length, chest circumference and MUAC among male and female newborns. Receiver operating characteristics (ROC) analysis was conducted separately for each anthropometric measurement and the area under the curve (AUC) calculated to investigate which measurement best predicted SGA, LBW and prematurity, respectively. For sensitivity analyses, ROC analysis only including newborns with GA estimated ≤14 weeks and ROC analysis stratified by the sex of the newborn were performed. Operational cut-offs for foot length, chest circumference and MUAC were selected based on obtaining the highest sensitivity and specificity for identifying SGA, and positive predictive value (PPV) and negative predictive value (NPV) for SGA were then calculated. Using the same cut-offs; sensitivity, specificity, PPV, and NPV were also calculated for LBW and prematurity. For comparison, operational cut-offs for foot length, chest circumference and MUAC were also selected based on obtaining the highest sensitivity and specificity for identifying either LBW or prematurity. The 95% confidence intervals for all the sensitivities, specificities, PPVs and NPVs were reported. In total, 538 women were enrolled in the pregnancy part of the study, and hereof 401women gave birth to a live singleton newborn without severe congenital malformations. Among these, 376 newborns had foot length, chest circumference and MUAC measured within 24 h and were included in the analyses (Fig. ​(Fig.1).1). The median GA was 40 weeks + 0 days (Interquartile range 38 + 6 to 41 + 1, range 25 + 6 to 45 + 0) and the mean birth weight 3014 g (±486, range 860 to 4360, 2.5th to 97.5th percentiles 2061 to 3968). In total, 68 (18.4%) were born SGA, 39 (10.4%) were LBW, and 17 (4.5%) were born preterm. Among the preterm newborns, 15 (88.2%) were moderate to late preterm, 1 (5.9%) was very preterm and 1 (5.9%) was extremely preterm. The characteristics of the mothers and their newborns are shown in Table 1. Characteristics of the 376 women and their newborns a371 women had age estimated at enrolment, b370 women had body mass index estimated, c316 (84.0%) women had gestational age estimated at ≤14 weeks, 54 (14.4%) women had gestational age estimated between 15 and 24 weeks and 6 (1.6%) had gestational age estimated at 25–28 weeks, d375 had BW estimated, e369 newborns were categorized as either SGA/non-SGA, fSD Standard deviation All three anthropometric measurements of the newborns correlated with birth weight and GA at delivery, showing the best correlation when using chest circumference (Pearson correlation for birth weight 0.86, Spearman correlation for GA 0.41) (Fig. 2). All three anthropometric measurements were also statistical significantly smaller among newborns born SGA, with LBW or preterm as compared to newborn with normal weight and born at term (Table 2). The anthropometric measurements were comparable for boys and girls except for foot length which was significantly shorter for girls (difference = 0.12 cm, p-value = 0.01, 95% confidence interval (CI) (0.02–0.21)) [See Additional file 1]. Association between foot length, chest circumference and MUAC vs. birth weight and gestational age at delivery Differences in foot length, chest circumference and mid-upper-arm-circumference among the 376 newborns aSGA Small for gestational age. bMUAC Mid upper arm circumference ROC curve analyses were used to assess the three different anthropometric measurements’ ability to capture SGA, LBW and prematurity. Chest circumference had the highest AUC for all three outcomes. The highest observed AUC was for chest circumference detecting prematurity (0.94) while foot length detecting SGA had the lowest (0.78) (Fig. ​(Fig.3).3). A minority (60/376, 16%) of the women had GA estimated after 14 weeks of pregnancy. Sensitivity analysis only including newborns where GA was estimated ≤14 weeks yielded similar AUC for all three anthropometric measurements [See Additional file 2]. If stratifying by the sex of the newborn, AUC’s for chest circumference and MUAC were comparable for boys and girls, whereas the AUC for foot length was slightly higher for girls [See Additional file 3 and Additional file 4]. ROC curves for detecting SGA, low birth weight and prematurity, respectively using the three anthropometric measurements The sensitivity and specificity for detecting SGA was calculated for a range of different cut-offs of foot length, chest circumference and MUAC. The operational cut-offs yielding the highest combination of sensitivity and specificity (equivalent to the place on the ROC most to the top left) for detecting SGA newborns were defined as ≤7.7 cm for foot length (sensitivity at 74%, specificity at 69%), ≤31.6 cm for chest circumference (sensitivity at 79%, specificity at 81%) and ≤ 10.1 cm for MUAC (sensitivity at 76%, specificity at 77%) (Table 3). If stratified by sex of the newborn the operational cut-off for foot length for boys yielding the highest combination of sensitivity and specificity for detecting SGA was similar to the already defined operational cut-off. However, for girls the cut-off for foot length was slightly lower (≤ 7.6 cm) [See Additional file 5]. The other anthropometric measurements did not differ significantly for boys and girls. Sensitivity, specificity, PPV and NPV for operational anthropometric cut-offs together with the 95% confidence interval aSGA Small for gestational age, bFL Foot length, cCC Chest circumference, dMUAC Mid upper arm circumference, eCI Confidence interval, fPPV Positive predictive value, gNPV Negative predictive value All three anthropometric measures show high NPVs (0.92–0.95) for identifying SGA, whereas the PPVs were below 0.50 for all (0.35–0.49). The best PPV and NPV were observed when using chest circumference (0.49 and 0.95, respectively) (Table ​(Table3).3). The main objective was to assess the ability of foot length, chest circumference and MUAC in identifying SGA. Since prematurity and LBW are closely related to SGA, sensitivity, specificity, PPV, and NPV for the same operational cutoffs, but using LBW or prematurity as outcome, were also calculated. As compared to when detecting SGA, sensitivity and NPV were higher for both preterm delivery and LBW for all three anthropometric measurements, whereas specificity and PPV were slightly lower (Table ​(Table33). For comparison operational anthropometric cut-offs balancing the sensitivity and specificity for detecting LBW and preterm delivery instead of SGA were also calculated. Cut-offs for all three anthropometric measures were slightly smaller for both LBW and preterm delivery. For both outcomes specificities and PPVs increased, whereas sensitivities and NPV were either comparable or slightly decreased as compared to when using the operational cut-offs defined based on SGA (Table 4). Sensitivity and specificity of operational anthropometric cut-offs to identify Low birth weight and preterm a PPV Positive predictive value, b NPV Negative predictive value, cFL Foot length, dCC Chest circumference, eMUAC Mid upper arm circumference