2.1 Data

For the empirical analysis we used National Family Health Survey (NFHS) conducted in 2015–16 and Public Affairs Index (PAI), 2016. NFHS employs a two-stage stratified random sampling method and interviewed women belonging to the reproductive age group (15–49 years). Total sample of this survey is 601,509 households, 699,686 eligible women, and 259,627 children (younger than 5 years). Data on Public Affairs Index (PAI) discerning the quality of governance, is extracted from the Public Affairs Centre (PAC) [21]. This centre adopted a multifaceted approach considering 10 distinct theme to construct a governance indicator. These themes were normally categorised into rule based and performance based measures (Further details are provided in S1 Data). To ensure the compatibility of PAI with NFHS (2015–16), we used PAI indices published in 2016 for all states, except Telangana and Andhra Pradesh because the information regarding these two states were collated into one (collected before territorial bifurcation). Hence, for these two states we resorted to PAI (2017).

2.2 Methodology

We defined a comprehensive list of RMNCH indicators and constructed two summative indices namely, CCI and Co-Cov. These indices provide complete information about RMNCH interventions by avoiding information overload.

CCI is constructed using four distinct themes swaying across different stages of the continuum of maternal and child healthcare services, starting from reproductive health (usage of modern contraceptives (FPS)), maternal health care services (four or more antenatal visits (ANC4), Skilled Birth Attendants (SBA)), new-born and child healthcare vaccination (3 doses of Diphtheria, Pertussis and tetanus vaccine (DPT3), measles vaccination (MSL), Bacilli Calmette- Guerin (BCG) vaccination (BCG)) and utilisation of healthcare services among children’s diagnosed with pneumonia (CPNM) and Oral rehydration therapy and other fluids for children with diarrhoea (ORT). A detailed description of these variables is provided in the [S2 Data] supplementary document.

Co-Cov indicator entails a range of preventive public health interventions—ANC4, Tetanus Toxoid dosage during pregnancy, SBA, BCG vaccination, DPT3 vaccination, Measles vaccination and household’s access to improved drinking water [19, 20]. The Co-Cov is defined as the number of interventions utilised by each woman and child pair. It is a binary indicator coded as 1 for (adequate utilisation) woman and child pair receiving 6 or more interventions of the aforementioned interventions and 0 for those availing less than 6 interventions (inadequate utilisation).

These two indices have been broadly used in the literature to examine the overall performance of different geographical regions such as, country/state/specific region in the provisioning of maternal and child health care services [9, 18, 22]. In the present study, we have adopted the following formula for the construction of CCI [Input file indicating the definition and computation of CCI is provided in S3 Data].

The above equation demonstrates a weighted average mean of eight essential interventions by assigning equal weightage for each of its component [16, 18, 22].

This paper aims to undertake an inequality analysis of RMNCH coverage across distinct wealth horizons. Wealth Index have been widely used to measure economic status of the individuals [23]. It is defined in the form of quintiles, where 1^st quintile (Q1) represents the poorest 20 percent of the population and last quintile (Q5) reflects the richest 20 percent of the population. Previous studies have predominantly used absolute and relative inequality methods to demonstrate inequality in the coverage of RMNCH. The absolute inequality (Q5-Q1) ascertained the magnitude of differences in the coverage of maternal and child health services between the richest and poorest quintile. Whereas, relative inequality (Q5/Q1) explains the ratio/relative differences in the coverage of maternal and child health services between the richest and poorest quintile.

Although these two methods hold an advantage of easy interpretability from a layman’s perspective, they suffer from certain serious drawbacks. First, when the stratification within a concerned sub-group changes, the level of inequality also varies. Second, the variations in the overall coverage levels do not necessarily represent a similar variation across wealth quintiles. In other words, the highest and lowest coverage among top and bottom quintile population do not necessarily represent highest and lowest coverage in the overall population. Finally, these two techniques do not take into account the intermediate wealth quintiles such as 2^nd quintile (Q2), 3^rd quintile (Q3) and 4^th quintile (Q4). We have adopted two advanced measurement techniques such as Slope Index of Inequality (SII) and Relative Index of Inequality (RII) to overcome the shortcomings encountered in previous techniques [9, 24].

To compute inequality for CCI and Co-Cov, we gathered indicator-wise information for each quintile by incorporating their respective weights. For each quintile, we followed a systematic approach encompassing five stages of computation. In the first stage, we used the individual level dataset and grouped it across quintiles for each of the indicator, in the second stage, quintile wise proportions, were computed for each indicators. These proportions were cumulatively added in the third stage and in the fourth stage, we divided these cumulative proportions by 2 to compute mid-point values. In the final stage, the logistic regression was conducted between the mid-point of the population in each quintile and indicators (CCI/Co-Cov) to estimate SII and RII.

The relationship between OI and RMNCH coverage inequality indices (CCI & Co- Cov) were captured by constructing Spearman’s rank correlation matrices. Detailed insights were gathered by conducting correlation analysis across each theme of governance indicator and coverage indicators. The governance indicator and its indices values are calculated by PAC and details of each of the index values across states are provided in appendix section [S4 Data].

We adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines published by the PLOS medicine and structured our manuscript accordingly [25]. We employed secondary dataset and hence ethical consideration was not required. Internal consistency of the indicators used for the construction of indices were ensured through Cronbach’s alpha test. All the statistical analyses were performed using Excel, Stata version 15.0, R version 3.5.2 and GeoDa 1.14.0 (shape file can be availed from Survey of India).

3.1 Inter-state disparity across RMNCH interventions

Table 5 captured inter-state disparity across RMNCH interventions. Evidently, interventions necessitating frequent patient-provider consultations demonstrated maximum disparity. In India, ANC4 is the most inequitable intervention with an absolute gap of 49 percent (Q5-Q1). Similarly, DPT3, a child-related vaccination entailing three frequent visits to the facility demonstrated the largest absolute inequality of 16 percent. Despite garnering enormous attention from central and respective state governments, the coverage of SBA remains unevenly distributed with a maximum utilisation undertaken richest quintile population while poorest quintile population slogging much behind.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Absolute gap in RMNCH coverage indicators across major states.

https://doi.org/10.1371/journal.pone.0258244.t005

State-level mapping of ANC4 revealed widespread disparity—Uttar Pradesh indicated an absolute disparity of 48 percent, while Karnataka, recorded an absolute disparity of 1.6 percent. For DPT3, Haryana demonstrated a highest pro-rich distribution with an absolute inequality of 47 percent, while Karnataka (-0.08) exhibited a pro-poor distribution. Absolute inequality for SBA utilisation in India was 30 percent. Amongst the states, Haryana experienced maximum disparity with an absolute gap of 52 percent, while in Kerala the disparity was negligible.

3.2 Overall governance as a contextual factor for the coverage of RMNCH interventions

This section illustrates an empirical relationship between overall governance indicator (OI) and RMNCH interventions. Further, we enumerated the relationship between RMNCH interventions and different themes of OI to capture relationship associated with each theme [Details of each of the theme are provided in S2 Data].

The results of Spearman’s correlation matrix indicate a strong and positive correlation between CCI and OI (Fig 3A) and between Co-Cov and OI (Fig 3B), with respective correlation coefficient values of 0.68 and 0.61. Although correlation analysis was conducted for each of the theme, the discussion is carried out only for themes indicating statistically significant association with CCI and Co-Cov (The correlation results for all the theme are provided in S5 Data). Theme-wise disaggregation demonstrated higher levels of positive correlation between Women and Child (WC) and Health and Education (HE) with RMNCH indicators. While Delivery of Justice (DJ) and Environmental factors (ENV) indicated a lower levels of positive correlation for CCI and Co-Cov.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3.

https://doi.org/10.1371/journal.pone.0258244.g003

State-wise analysis revealed that the states performing better in terms of governance indicators, examples of which are Kerala, Tamil Nadu, Karnataka, Himachal Pradesh, Maharashtra and Punjab also performed better in terms of the coverage in CCI. While, the states of Bihar, Madhya Pradesh and Uttar Pradesh demonstrated poor performance in terms of both governance indicator and RMNCH coverage. Andhra Pradesh and Telangana turned out to be the exceptional ones, these states showed outstanding performance in terms of the coverage indicators despite having moderate governance scores. The details are depicted in Fig 4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4.

https://doi.org/10.1371/journal.pone.0258244.g004

4.1 Inequality in coverage of RMNCH services: Geographical regions

Findings of this study demonstrates an erratic disparities across and within the regions of India in terms of the coverage of RMNCH services. Similar conditions prevailed in other developing and developed countries [14, 18, 26]. In our study, the performance of RMNCH in southern states of India, were much higher than other states. The outstanding performance of southern region could be attributed to the prevalence of strong political commitment in the social sector, better awareness of health services within the community, greater levels of literacy among females, upgraded infrastructure of public health facilities and stronger delivery mechanisms [27–29]. It can hence be conferred that states concerting greater levels of attention in terms of the provisioning and distribution of health services have been successful in obtaining higher levels of RMNCH coverage [29].

Amongst the states experiencing pro-rich inequality, the absolute and relative disparity of CCI and Co-Cov are found to be lowest in Punjab. This could be mainly because, in Punjab, the primary health system had undergone a huge transformation, with most of the PHCs being upgraded to the Block PHC. The infrastructure and quality of these centres are comparable to the Community Health Centres (CHCs). Moreover, various new health facilities were created and managed by the local Governments [30, 31]. It can be concluded that relegation of authority to local governments might be an efficient strategy to reduce health inequality.

Surprisingly, in Karnataka, the utilisation of RMNCH services amongst the richest quintile population were proportionally lesser compared to poorest quintile. This could be explained by the proactive approach adopted by them to deliver quality healthcare services. The coverage of prenatal care services, delivery services, BEmOC and CEmOC facilities has increased across the districts indicating a concomitant reduction in the inequality of maternal and child healthcare services in Karnataka [27]. Moreover, a part of Karnataka which is also known as the “Silicon valley of India” accruing to its technological innovations might have transmuted its benefits to health sector in the form of promotion of digital health initiatives. The existence of digital monitoring mechanisms at the gram panchayat levels might have turned out to be an effective way of catering to the poorest quintile population. The role of technological innovations such as Digital Life Care, e-hospital projects linking patients and facilities, mobile health clinics ensure increasing accessibility of health care services to remote areas in Karnataka [32].

Inequality in the coverage of CCI and Co-Cov was highest in Haryana reflecting an outrageously high levels of inequalities across its districts [17, 31]. The existing disparities in the districts of Haryana could be explained by the variations in demand side factors like place of residence and socio-demographic factors and also by supply side components such as health policies, programme implementation, infrastructure and governance [31].

4.2 Inequalities in RMNCH interventions: Standalone indicators

The coverage of maternal health interventions varied distinctly across wealth quintiles, whereas, child health related interventions were more or less equitably distributed across bottom and top quintiles. Our analysis revealed that, inequality in the coverage of ANC4 (Q5-Q1 = 49) is much higher compared to other interventions. However, enormous inequality was also encountered in the coverage of SBA across different wealth quintiles (Q5-Q1 = 30). Under-utilisation of ANC4 and SBA was by the poorest quintile population was mainly contributed by financial and accessibility barriers [33, 34].

The distribution of ANC4 is quite disparate across states. Highest level of inequality in ANC4 was witnessed in Uttar Pradesh (Q5-Q1 = 48), while lowest inequality was experienced in Karnataka (Q5-Q1 = 2). The lowest level of inequalities in Karnataka could be attributed by the democratic decentralisation framework adopted by this state [29]. Similarly, the distribution of the utilisation of SBA were erratic across the states, with Kerala surpassing other states and Haryana displaying a greatest level of inequality with an absolute gap of 52 percent. Evidence suggests that Kerala has better accessibility and availability of health care services [35]. The National Health Profile (2017) has shown that the availability of government healthcare facilities in Kerala is 1280, while Haryana has only 159 health facilities in place [35].

Furthermore, amongst the child health related interventions—DPT3 and Measles has the higher level of inequalities (Q5-Q1 = 16). Whereas, the utilisation of BCG vaccination demonstrated an equitable distribution across wealth horizons. This could be partly associated with the accessibility hinderances prevalent amongst the poorest quintile population when the utilisation entails multiple visits. Typically, utilisation of adequate ANC4 and DPT3 entails multiple visits to the facilities [12], whereas the utilisation of BCG shots requires a one-time visit to the facility. Furthermore, the programmatic interventions and outreach programs related to BCG vaccination have percolated to grassroot levels due to the early and strict initiatives undertaken by the Government of India. Since 1948, BCG has gathered incessant attention as it was the only preventive vaccination available to control Tuberculosis in India [36]. DPT and Measles acquired attention much later with the implementation of the National Immunisation Programme called Expanded Programme of Immunisation in 1978 which aimed at expanding the coverage of both the diseases to 80 percent. These targets were however revised to 100 percent in the year 1985 with the initiation of the Universal Immunisation Programme. Despite these efforts, the coverage of DPT3 and Measles had remained sub-optimal and inequitable. This could be due to dearth of trained personnel managing the programme at the national and state levels [36, 37].

4.3 Correlation between governance indicators and RMNCH services

We discovered a significant correlation between OI and coverage indicators. We found that governance indicators such as, HE and WC demonstrated strong and positive association with coverage indicators, our findings are comparable to the findings of sub-Saharan countries [15]. Previous studies have postulated that the effectiveness of governance indicators have strong association with coverage amongst the poorest quintile compared to the richest quintile [16].

The higher correlation of WC and coverage indicators might be due to the influence of primal components of WC such as government’s utilisation of JSY fund and percentage of women working population ratio. Greater utilisation of JSY fund would have a direct/ intended impact on delivery and post-delivery care services and indirect/ unintended impact on child immunisation services. This argument have also been supported by various researchers in the previous studies [37, 38]. It indicates a greater level of government investment in the form of cash benefit programs to ensure higher utilisation of RMNCH services. The women working population ratio also had a positive correlation with the coverage of RMNCH services, this could be explained by a potential financial freedom of women resulting from an increase in female workforce. Interestingly, both these factors increase the purchasing power of the women and their financial freedom [39].

Our analyses also showed that HE is directly associated with the coverage of RMNCH services. It might be due to indicators of HE such as educational expenditure as a percentage of Gross State Domestic Product (GSDP) and health expenditure as a percentage of GSDP. Access to education in general and women in particular is an essential building-block for increasing RMNCH coverage. Previous research studies have also reported a direct association of the contribution of Gross Domestic Product (GDP) per capita and coverage [16]. However, our findings suggest that the coverage of RMNCH is influenced by specific contribution of GSDP on health and education.