The present cross-sectional study was carried out from August 2016 to March 2017 in the MR Bangur Government Hospital, the district hospital of South 24 Paraganas, West Bengal, India. A total of 495 newborn births records and anthropometric measurements were recorded using standard anthropometric procedures [27]. The cases of non-institutional or home deliveries and mother who reported any chronic disease and/or any complications were excluded. However, the newborns born with the full gestational period was taken into consideration in this study. The calculation of sample size (N) was done using criteria of anticipated population proportion (i.e., LBW prevalence) of 34% [28], the minimum absolute precision of marginal error was 5% and the confidence interval (i.e., the maximum amount of tolerance) of 95% was taken into consideration. The standard equation was used to calculate sample size as follows:

where, p=0.34, Δ=0.05 and z=1.96.

The minimum number of subjects required for reliably estimation of the prevalence/effect in a health investigation was determined using the standard sample size estimation method [29]. Thus, the minimum sample size was estimated to be 345 for the present investigation. If the response rate considered being 75% at the sample size estimation is assured, therefore the minimum number required to be N=431.

Therefore, a total of 500 newborns were identified and initially approached for the present study and a total of 5 (1.00%) cases were excluded due to the listed selection criteria. Hence, final subject of the present research investigation comprised of 495 newborns of both cesarean and vaginal deliveries, which was found to be appreciably higher than the estimated sample size. There was no case of any macrosomic baby (≥4000 gm) in the analysed samples. The necessary permission of the present study was obtained from the Ethical Research Committee of University of Calcutta, Kolkata, and MR Bangur Government Hospital, South 24 Paraganas, West Bengal, India. The inform consent was obtained from the mothers/guardians before the collection of the anthropometric measurements among newborns. The present study was conducted in accordance with the ethical guidelines of human experiments as laid down the Helsinki Declaration [30].

The present study was a hospital-based quantitative cross-sectional study, based on both hospital records and direct anthropometric measurements collection. The caste and religious beliefs of mothers were recorded at the time of admission and as per hospital register, only the data of the Bengali speaking Hindu population were selected for an identical and homogenous ethnic composition in this investigation. Mothers who delivered a normal singleton newborn completing full gestational period (i.e., 37 weeks) were included in the study. The mothers residing in the nearby rural and urban areas, belonging to lower and middle socio-economic strata predominatly participated. The collected data comprised of newborn sex, BW, BL and HC. The anthropometric measurements were obtained within 24 hours of birth by one of the authors. The BW of the newborn was recorded to the nearest of 50 gm by using standard neonatal weighing machine within 15-30 minutes of birth. The newborns crown to heel length naming as BL was measured by infantometer, with knees extended and soles to feet held against the footboard. HC was measured by placing measuring tape on glabella and posteriorly along the most prominent point opisthocranium using non-stretchable measuring tape the nearest to 0.1 cm [27].

Data were analysed using the Statistical Package of Social Sciences (SPSS) (Version 16.0). The anthropometric variables were presented in terms of descriptive statistics {mean and standard deviation (SD)}. The sex specific-mean comparisons of the anthropometric variables were done using One-Way Analysis of Variance (ANOVA). Pearson’s correlation coefficient and linear regression analyses were done to examine the linear relationship between two or more continuous variables (e.g., BL and HC). The chi-square (χ²) analysis was done to assess the sex-specific prevalence of LBW among newborns. BLR analysis was fitted into estimating the odds of being an LBW which allows controlling the different independent variables. To create the dependency newborn BW of ≤2.49 kg was coded as ‘0’ and BW ≥2.50 kg and above were coded as ‘1’. The predictor variables used in the BLR model analysis the newborn’s BL and HC. Receiver operating characteristic curve (ROC) analysis was done to compare the overall predictability of anthropometric measurements (i.e., BL and HC) for the identification of LBW newborn and calculate the 95% confidence intervals of Areas Under the Curve (AUC) that was to determine the overall accuracy and the sensitivity of the cut-off points to identify best surrogate anthropometric measurement. Finally, the predictive performances of the cut-off points were calculated. A p-value of <0.05 and <0.01 were considered being significant and 95% Confidence Interval (CI) was used to show the strength of association between independent and dependent variables.

The descriptive statistics of newborns BW and anthropometric variables are presented in [Table/Fig-1]. A total of 495 newborns were recorded and of the 241 (48.7%) and 254 (51.3%) were male and female newborn, respectively. The sex-specific mean BW was observed to be significantly higher among boys than girls in NBW (p<0.05), but found contrary in LBW (p<0.05). Sex-specific mean BL (48.01 cm vs. 47.54 cm) and HC (32.80 cm vs. 32.74 cm) was observed to be significantly higher among male than female newborns (p<0.05). Sex-specific mean differences were observed to be statistically significant in BW, BL and HC using ANOVA (p<0.05). The mean anthropometric variables of BW, BL and HC were observed to be significantly greater in NBW than LBW newborns (p<0.01). The results indicated that the mean BW was 2.72±0.44 kg and 134 (27.07%) newborn were found to be LBW [Table/Fig-2]. The sex-specific prevalence of LBW was observed to be insignificantly higher among boys (28.23%) than the girls (25.98%) (χ²-value: 0.179; p>0.05).

Linear regression models analysis for BL and HC as independent variables and BW as the dependent variable were performed and results are depicted in [Table/Fig-3]. The anthropometric measurements were correlated with BW, the maximum correlation with BW was observed for BL (r=0.791; p<0.01) than HC (R=0.757; p<0.01). The BL had the lowest Standard error of the estimate (SEE) and it had a higher coefficient of determination (R²) when compared with HC (0.625 vs. 0.572). The results of the BLR analysis showed that anthropometric variables (e.g., BL and HC) exhibited significant influences on LBW (p<0.01) [Table/Fig-4]. The [Table/Fig-4] provides the regression coefficient (R), the Wald statistic and the Odds Ratio {(Exp (β)} for each variable. The Wald values of BL and HC are 96.39 and 95.67 and the odd ratios (β) are 0.50 and 0.14 for BL and HC, respectively (<0.001). The values of coefficient of domination were observed to be significantly higher in HC (R²=0.392) and HC (R²=0.352) with LBW using BLR analysis (p<0.01) [Table/Fig-4].

ROC analysis was performed to diagnose the accuracy of the different anthropometric measures (e.g., BL and HC) to predict the LBW showed that HC had the significantly greater than the BL predictability to identify the LBW newborns (p<0.01) [Table/Fig-5]. For each anthropometric measure, sensitivity and specificity for the ranged values were calculated and operational cut-off points determined by considering the value with the highest average of sensitivity and specificity. At the proposed cut-offs, HC has the lower sensitivity (78.4%) and higher specificity (81.0%) where BL has the higher sensitivity (81.7%) and lower specificity (80.6%) respectively for screening LBW newborns (p<0.01) [Table/Fig-6]. The best discrimination of LBW, as detected by ROC-AUC, was obtained by HC (AUC=0.91, 95% CI 0.88 to 0.94) followed by BL (AUC=0.86, 95% CI- 0.86 to 0.92). An optimum cut-off point was estimated to identifying LBW newborn were <32.5 cm and <46.5 cm for HC and BL, respectively. The Correct prediction percentage (CPP) in the determination of LBW based on ROC-AUC analysis showed that the overall CPP was observed to be higher in HC (91.0%) than BL (80.6%). Sex-specific CPP was ranging between 76.5% and 84.8% (in BL) and between 91.1% and 90.9% (in HC) among boys and girls newborn, respectively [Table/Fig-7].

[Table/Fig-6]:

ROC-AUC analysis of HC (A) and BL (B) with the birth weight or LBW among newborns.

The LBW is the single most significant reproductive outcome of IUGR and determinant of newborns mortality, morbidity and the chances of a newborn to experience healthy development and/or survival. The prevalence of LBW appears as a major public health issue in several developing countries and reported being the highest among South-Asian countries including India [5,8,11,31,32]. The present study was undertaken to identify the best suitable surrogate anthropometric measure to predict LBW, which could be useful for the health personnel of domiciliary outreach to identify the maximum number of at-risk newborns for providing the timely and needed intervention strategies in the target population [5,22]. The Bengali speaking Hindu Population is the major dominant caste ethnic group in West Bengal, India and the present investigation has reported the overall mean BW was 2.72±0.44 kg and the overall prevalence of LBW was 27.07% [Table/Fig-2]. Similar studies have shown that the lower prevalence of LBW and mean BW were 17.3% and 2.75±0.40 kg [8], and 19.1% and 2.71±0.41 kg [11] in hospital-based investigations in West Bengal, India [Table/Fig-8]. However, the mean BW comparison was observed to be higher than the reported mean BW among the newborns of Rural Karnataka [1]; Puruliya, West Bengal [2]; and Slum-area, Mumbai [33]. The sex-specific LBW was observed being higher among male than female newborns (p>0.05) [Table/Fig-2]. The comparative evaluation of the LBW prevalence in India [1,2,5,8,11,17,19,22,24,33-38] with present study was observed being similar than those reported from investigations were undertaken in Darjeeling (17.30%) [8], Indian population (28.00%) [9], West Bengal (19.19%) [11], Maharastra (44.60%) [18], Kolkata (34.00%) [28], Maharastra (47.00%) [32], Mumbai (45.20%) [33] and Haryana (28.8%) [34] [Table/Fig-8].

Moreover, several researchers have reported the magnitude of newborns LBW in different ethnic/caste populations were ranged from 17.30% to 34.00% in West Bengal [8,9,11,28]. The results of the present study showed the lower prevalence of LBW than the estimated national average {i.e., 28.00% [8]} maybe attributed to the selection of a homogenous ethnic population (i.e., Bengali Hindu population). Several researchers have reported that higher BW values could be attributed to inclusion of full-term singleton live birth babies [11,19,22]. Although, several researchers did not specify such inclusion criteria of the newborn and the newborns born before completion of the full-term gestation period (i.e., <37 weeks) may also have been considered in the above-mentioned study [39,40].

The BW is still considered being an important screening tool for detecting newborns with LBW and a high proportion of newborn LBW deliveries are performed at home especially in rural populations in India [26,41]. It is attributed to the inadequate and cultural inappropriateness of healthcare facilities, rural-urban differences, availability of the traditional birth attendant, family preferences, maternal autonomy or decision making with respect to movement and healthcare accesses and extrinsic factors (e.g., unavailability of transport, poor road conditions, financial conditions, poor awareness and ignorance of healthcare schemes or inadequate utilisation of antenatal care services and distant hospitals, health-related behaviours and decisions making) are promoting home deliveries in India [8,9,31,41]. Due to unavailability of logistic supports (e.g., inaccurate weighing scales/infant-meter) and the high prevalence of non-institutional or home deliveries, measuring newborns BW and the relative risks LBW newborn is considered being a difficult task in developing countries [5,42]. Therefore, researchers have advocated the need for alternative/proxy anthropometric measures to identify the LBW newborns. The results of the present study showed that the BW was significantly correlated with the anthropometric variables of BL and HC (p<0.01). The BL showed the best correlation with BW and HC (p<0.05) [Table/Fig-3]. The coefficient of determination model showed that BL was more precise than HC (p<0.01) to identify LBW newborns [Table/Fig-3]. However, the coefficient of determination in regression analysis (R²-value) which considered being the better means for evaluating the strength of a relationship was used to identify the best alternative measure of newborns BW. Several research studies have reported that significant correlation coefficient values in anthropometric measurements (e.g., BL and HC) and LBW among newborns [5,7,17-19,22,41]. The linear regression analysis of anthropometric variables showed that BL had the lowest and shown a higher coefficient of determination as compared with HC with BW (p<0.01) [Table/Fig-3]. Several research investigations have used linear regression analysis to identify the anthropometric measurements to determine the highest coefficient with BW in newborns [5,7,42]. A similar study has reported the significant positive correlation between BL and HC with BW among newborns in Tehran, Iran [22]. The results of the BLR regression analysis showed that both BL and HC have significant predictability of LBW among newborns (p<0.01) [Table/Fig-4]. Similar studies have reported that HC is relatively reliable surrogate anthropometric measures of LBW in newborns [5,22,38].

The AUC-ROC analyses showed that HC (AUC=0.91) is the most appropriate surrogate anthropometric measurement over BL (AUC=0.89) for identifying LBW newborns [Table/Fig-5]. A similar study has reported that HC (AUC=0.89) with cut-off ≤33.50 cm has shown optimal surrogacy with LBW among Nepalese newborns [35]. This is comparable with the cut-off points of HC is 32.5 cm, while the values of sensitivity and specificity are 78.4% and 81.0%, respectively (p<0.01) [Table/Fig-5]. Furthermore, the overall CPP was observed to be significantly higher in HC (91.0%) than BL (80.6%) to determine the LBW among newborns utilising the derived cut-offs values in present study [Table/Fig-7]. Jeyashree K et al., has confirmed that HC with the cut-off point of 33.8 cm (Sensitivity: 84.88%; Specificity: 68.14%) and BL with 45.7 cm (Sensitivity: 84.88%; Specificity: 68.14%) being a good surrogate anthropometric measure of LBW in India [41]. A similar study has reported that HC of 33.25 cm had the highest positive predictive value (AUC=0.93, specificity: 77.0%) of LBW in Ethiopia [5]. It is to be mentioned here that HC measurement may not be accurate due to the moulding of the head during birth in case of the prolonged and obstructed labour [17]. A recent study in Korea reported a high correlation coefficient between first-trimester crown-rump length and BW [43]. This is also true of findings from studies reported among newborn in Pune, India [21]. The present study has reported the cut-off point of BL is 46.5 cm, while the values of sensitivity and specificity are 81.7% and 80.6% [Table/Fig-6]. A similar study has also reported the cut-off point of BL (48.5 cm, 84.0% sensitivity and 88.0% specificity) and HC (33.25 cm, 73.0% sensitivity and 85.0% specificity) among newborns of Tehran, Iran [22]. Furthermore, this present study was conducted using a dataset from the regional hospital in Kolkata, West Bengal and the anthropometric data (i.e., BW) was collected and recorded by hospital staff, without the direct supervision, hence this situation might cause bias in data collection and management. Moreover, maternal determinant variables include anthropometric characteristics (e.g., height or BMI), socio-economic, demographic and lifestyle variables are considered to be very important determinants of newborns prenatal growth attainments or IUGR (e.g., LBW) [8,11,36,37].

It is to be mention here that the present study did not consider the effect of maternal anthropometric, body composition, socio-economic, demographic and lifestyle variables on LBW among newborn. The variations in these determinant variables could have significant effect on the attainment of BW and/or LBW prevalence among newborns.

It is clear from present study that HC may be suitable and simple surrogate anthropometric measure may be used in the domiciliary outreach when it is impossible to record the weight of the newborn at birth, while BL is the next best parameter to identify the LBW. Moreover, the easiest way to record the HC or BL at home is to use a simple non-stretchable tape measurement. These proxies anthropometric measurements (e.g., BL and HC) have shown higher specificity and sensitivity to predict LBW in newborns. With this technique, measurement errors are likely to be less than the measurement of HC and BL are taken into consideration to screen the risk of LBW. Therefore, using such an indicator would help to identify neonates requiring intervention to prevent premature deaths, diseases, disability and this will be a viable example of practical and cost-effective healthcare strategy especially in field situations.