Data for this study were collected in March to June 2014 as the follow–up household survey for evaluation of Ananya, a public health program that supported a combination of supply–side and demand generation efforts to increase maternal and child health care utilization via the public health system in Bihar [34]. Evaluation of Ananya involved a two–armed quasi–experimental design, comparing mothers of infants and children from the eight districts in which the Ananya program was implemented to those from the remaining 30 districts assigned to the standard of care control condition. Although Ananya included both a baseline and one–year follow–up household survey, only the latter assessed neonatal health and health care utilization outcomes for mothers of 0– to 11–month–olds, providing variables for this analysis. Trained female study staff, subsequent to acquisition of written informed consent, collected all data. Details on the study’s multi–stage sampling and other procedures are available elsewhere [35].

The response rate for the sample of mothers of children 0 to 11 months in the survey was 87% (n = 11 654) [35]. The sample was further restricted for this analysis to mothers of living, singleton children (n = 11 570). Ethical approval for the original study was provided by India’s Health Ministry Screening Committee. Ethical approval for this analysis was provided by the University of California, San Diego.

Female infants made up 46.3% (95% CI = 45.0–47.7) of the sample, which corresponds to an infant sex ratio of 863 females/1000 males (95% CI = 817–912) ( Table 1 ). Crude logistic regression analyses indicate that a lower share of female infants had symptoms of neonatal illness compared to males (9.8% vs 11.9%; OR = 0.7, 95% CI = 0.6–0.9) and a lower share of female infants had been hospitalized compared to males (1.2% vs 4.9%; OR = 0.4, 95% CI = 0.3–0.6) ( Table 2 ). However, compared to boys, girls had lower odds of receiving neonatal health care for identified symptoms (80.6% vs 89.1%; OR = 0.5, 95% CI = 0.3–0.8) and of receiving preventive care at a facility at 1 month (5.4% vs 7.3%; OR = 0.7, 95% CI = 0.6–0.9). Gender differences were not seen in immunizations, FLW postnatal visits or FLW advice to seek care. However, neither immunizations nor FLW visits occurred at target levels. Receipt of all immunizations (BCG, Polio 3 and DPT 3) by nine months was 59.9% for girls and 62.2% for boys. Only 16% of girls and boys had received a postnatal care home visit from an FLW in the first week post–partum. Multivariable analyses yielded similar association findings, with girls having lower odds than boys of having health concerns but also lower odds of receiving curative or preventive neonatal care ( Tables 3 and 4 Table 4). No effects of sex were seen on FLW visits.

Stratum–specific odds ratios for interactions between sex of child and birth order, as well as sex of child and wealth, are presented in Table 5 for outcomes associated with child sex in the primary analyses and demonstrating significant interaction effects (P < 0.15). Only receipt of care for neonatal illness met these criteria. In the case of birth order modifying the effect of sex on receipt of care for neonatal illness, there was no difference among firstborn children but an increasing disparity for females with higher birth order. This culminated in girls having 90% reduced odds of receiving care relative to boys when restricting to fourth or higher child (adjusted OR (aOR) = 0.1, 95% CI = 0.1–0.3). Wealth index also demonstrated significant effect measure modification with child sex in the model predicting receipt of care for neonatal symptoms. The gender inequity disfavoring females was most pronounced among the poorest quartile; similarly, the disparity between highest and lowest quartiles of wealth was more pronounced among females than males. Among the lowest wealth quartile, girls’ odds of receiving care were 80% reduced relative to boys (aOR = 0.2, 95% CI = 0.1–0.5). Similarly, among females, those in quartile 1 had lower odds of this outcome relative to those in quartile 4 (OR = 0.3; 95% CI = 0.1–0.7); this association of wealth with receiving neonatal care did not hold among males.

Results of this study document significant gender inequities in indicators of curative and preventive neonatal health care use, with girls having lower odds than boys to receive care for neonatal illness and facility–delivered postnatal wellness care. Moreover, observed gender inequity in neonatal care seeking is more pronounced among infants with a larger number of siblings and among households in the poorest wealth quartile, conditions likely associated with a relative scarcity of household resources. These findings indicate that, as seen in smaller scale and older studies from India [18,21,24], gender bias continues to affect parents’ health care seeking for the child, and these effects are exacerbated when household resources are limited or over–extended. FLW visits, although currently at low numbers, encouragingly do not exhibit the same bias.

Despite the greater odds of non–receipt of care, girls had lower odds of being reported ill during the neonatal period compared to boys. This finding corresponds with national indications of lower risk of neonatal death for girls relative to boys, suggesting greater biologic vulnerability for boys relative to girls [9]. The odds of hospitalization in the first year of life, a more rare and expensive event, differed even more starkly between boy and girl infants, and as with neonatal illness, may be attributable to biology. However, it may also, at least in part, be attributable to greater use of hospitalization when needed for boys relative to girls. As noted previously, research from India documents that higher–cost medical care, such as hospitalization, is more likely to be prioritized and used when needed for boys relative to girls [18,24,39], while discharge against medical advice is more likely for girls relative to boys [21].

Importantly, immunization coverage did not differ by infant sex at 9 months, a finding contrary to a number of studies conducted in India in the past [13–16]. These findings suggest that the growing efforts to maximize immunization coverage since 2006–2007 – when much of the data on gender disparities in immunization were collected – are not only expanding coverage but also reducing gender inequities [40]. Of note, these improvements were based on access to care broadly and not on addressing underlying causes of gender inequities in immunization uptake. Consequently, ongoing gender inequities in use of other health services persist, suggesting that improving access, while important, will be inadequate to address observed gender inequities in care. More work is needed to identify how to work more effectively with parents to increase their value for their girl children, particularly in the presence of older children and male children. Reinforcing this point is a recent analysis of infant and child mortality in neighboring Odisha state, which indicated that while infant and child mortality significantly decreased from 2006 to 2012, the trend was not observed for girls [41].

A related paradox of extending health programs to combat social inequalities in health is that boys have greater access to these facility–based services than do girls [33]. Use of FLWs via home visits may provide an opportunity to extend reach to girls more effectively, particularly for those from more vulnerable impoverished households [26]. Study findings suggest no gender bias in FLWs’ provision of a postnatal visit or advice for child health care seeking. Unfortunately, too few women received a FLW postnatal visit, only 16%. This is particularly disappointing given the promise FLWs offer to help address parents’ gender bias in care seeking. A post hoc analysis from these data showed a non–significant trend (P = 0.09) indicating that girl infants who were advised to seek care by an FLW had a higher odds of having received care compared to girls not so advised (89.4% vs 78.6%; OR = 2.3, 95% CI = 0.9–6.2). This association did not hold among boys, who had higher overall levels of care seeking. Together, these findings suggest that improving coverage of FLW services could play a role in alleviating the gender inequity in care–seeking for infants alongside efforts to improve standing of the girl child.

In addition to findings related to gender inequities and health, some findings regarding social inequities were also observed. Adolescent and Muslim mothers had higher odds of having a sick neonate, and among those with a sick neonate, Muslims had lower odds of having a front line worker (FLW) recommend that they seek care for their infant. Interestingly, there was a higher odds of FLWs recommending care for sick infants more to SC/ST and rural mothers, suggesting that FLWs do support care for more socially vulnerable groups, but perhaps not evenly. Notably, the poorest infants appear to be the most vulnerable, having lower odds than the richest infants of having a FLW recommend care when ill, to receive care when ill, to be hospitalized, and to receive vaccinations, suggesting that poverty remains a key driver for non–use of health care in Bihar. These findings may help explain the heightened risk for infant mortality among the abject poor in India [42]. Sadly, and as noted above, poverty appears to exacerbate what appears to be gender discriminatory effects on neonatal care utilization.

Some study limitations must be considered. Analyses are cross–sectional, so causality cannot be assumed. The data are self–reported by mothers and thus subject to recall and social desirability biases. Additionally, mortality data were not available for this analysis. Participation biases may exist, and we unfortunately did not collect data on those who declined participation to ascertain what these biases may be; the high participation rate (87%, as noted in the methods) reduces our vulnerability to these potential biases. A key challenge for interpretation of this study and many that investigate gender bias in health care is that some of the important outcomes represent both health need and health care utilization. Differentials in hospitalization of baby boys and girls likely reflect differences in severity of underlying illness, gender bias in parents’ use of care or referral systems’ recommendations for care. Bhan and colleagues found hospitalization for diarrheal illness in a New Delhi hospital was much lower among girls, but mortality was higher at the same time, undermining differential vulnerability as an explanation [23]. Additionally, symptoms of illness in the first month can reflect perception or actual rates of illness [13]. In our survey, the key question about actual receipt of care, restricted to those who endorsed neonatal illness, shows a marked gender inequity, as does the preventive care outcome of one–month well visits at a facility. These together suggest gender inequity even within the recognized and reported cases of illness and further imply that the same underlying gender bias could contribute to other findings such as the stark difference in hospitalizations. Assessing the presumed effect of health care use on mortality with any possible effect modification by gender would be an important future direction for this research.

This study documents significant gender inequities disadvantaging infant girls in the receipt of facility–based curative and preventive health care in Bihar, India, implying a role of gender bias and neglect on the part of parents. Notably, gender inequity in immunization coverage at 9 months of age was not observed, suggesting dramatic improvement on this issue over the past decade [13–17], likely via expansion in local access to vaccinations [40]. Front–line worker visits and services were not implicated in the gender disparity. These findings suggest that focus on gender bias and its impact on parents’ health care decision–making remains a concern in Bihar, and inadequate use of health services for girls may be contributing to the excess female mortality rate in the country. Study findings indicate that poverty and a higher number of siblings worsened observed gender inequities and suggest a role for targeting those particularly limited by resources. Broader reach of care beyond the facility, possibly through FLWs and home visits, may offer opportunity to support improvement, but such efforts must coincide with broader social change to support the value of the girl child in India and parents’ recognition of this value.