6. Inequality of Opportunity in Healthcare Services

The Human Opportunity Index (HOI) is a measure designed to assess both the level of coverage of basic opportunities and the extent to which the distribution of these opportunities depends on circumstances beyond an individual's control (Barros et al., 2009). Essentially, the HOI combines the overall coverage of opportunities and the IOp (measured by the Dissimilarity Index or D-Index) into a single metric (Vani & Madheswaran, 2018). This is particularly useful for evaluating access to health services. For instance, in the context of health services, the HOI reflects both the overall access rate to health services and how equitably this access is distributed across different socio-economic groups. By adjusting for the inequality in distribution, the HOI provides a more nuanced understanding of health opportunities in a society (Sanoussi, 2017). It shows not only how many people have access to health services but also whether this access is fairly distributed.

Estimating HOI Using the Random Forest Classification ML Model: To estimate the HOI, this chapter employs the Random Forest (RF) Classification model. This model is used to specify a functional relationship between an individual's circumstances and outcome such as their access to healthcare services. In this model, access to a healthcare service is denoted by 1, and lack of access is denoted by 0. The RF classification model helps to estimate the predicted probability of access to a particular service such as healthcare service for each individual. The overall probability of access to health care (\(\overline{p}\)) is then estimated using individual probabilities and the D-Index. The equation is as follows:

After calculating the overall probability of access to health care and the D-index, the HOI is calculated using the following formula:

The HOI along with advanced Random Forest Classification ML model, offers a deeper understanding of health IOp. The HOI not only measures overall access to health services but also assesses the fairness of their distribution, providing a comprehensive view of health opportunities within a population. This approach supports the principle of equality of opportunity and aids in developing policies that foster a more equitable health system, ensuring everyone has the chance to lead a healthy life (Barros et al., 2009; Sanoussi, 2017).

Hierarchical Clustering for Regional Analysis: Apart from supervised ML method (RF classification), hierarchical clustering, an unsupervised learning method, is also used to perform regional clustering analysis by building a hierarchy of clusters. This method groups data points based on their differences or similarities. There are two types of hierarchical clustering: agglomerative and divisive (James et al., 2023). Agglomerative clustering constructs the hierarchy from the bottom up by merging the closest or most similar data points into a desired number of clusters. In contrast, divisive clustering creates a hierarchy from the top down, splitting the most dissimilar data points until each one forms its own cluster. To link data points and create clusters, hierarchical clustering employs the following linkage criteria, including:

In this chapter, the agglomerative approach is used to perform hierarchical clustering of districts in India, specifically focusing on access to maternal and child healthcare services. The Ward linkage criterion is chosen because it minimizes the variance within clusters, leading to more homogeneous clusters with similar data points (or districts, in this case).

This study analyses health IOp with a focus on maternal and child healthcare services in India. It uses two measures: the D-Index and the Health Opportunity Index (HOI). The study also evaluates access to five key healthcare indicators: (i) place of delivery (whether a child was born in a public or private institution), (ii) immunization (whether the child received vaccinations), (iii) prenatal care (whether important health checks and tests were performed during pregnancy), (iv) prenatal care by trained professionals (whether the care was provided by a qualified doctor or health worker), and (v) antenatal care (whether the mother had at least four prenatal visits). Additionally, the study creates a composite index called ‘adequate care’, which includes all these factors to determine if a child and mother received comprehensive health care. The hierarchical clustering is also based on five key variables as discussed above that assess access to maternal and child healthcare services. Using these variables, the analysis aims to identify clusters of districts with similar levels of access to maternal and child healthcare services, providing valuable insights into regional disparities and areas needing improvement.

Data for this analysis was sourced from the National Family and Health Survey (NFHS-5) (2019–2021), conducted by the International Institute for Population Sciences (IIPS), Mumbai, and the Ministry of Health and Welfare, Government of India. The NFHS sample includes 190,355 children for whom parental information is available. The circumstance variables, such as parental education and occupation, are calculated using the children's data file, men's data file, and household member file from NFHS-5. The seven circumstance variables include: (i) Gender of the child (male or female); (ii) Sector, indicating the child's place of residence or birth (rural or urban area); (iii) Social group (Scheduled Caste (SC), Scheduled Tribe (ST), Other Backward Class (OBC), General Category (GC)); (v) Income quintile or classes ranging from Q1 to Q5, where Q1 represents the poorest and Q5 the wealthiest; (vi) Region, indicating the geographical area the child belongs to (north, south, east, west, central, or northeast); (vi) Mother's education level (no education, primary education, secondary education, or higher education); (vii) Father's education level (no education, primary education, secondary education, or higher education); This comprehensive dataset helps to understand the different factors contributing to inequality in access to healthcare services in India.

The HOI and D-Index are tools for measuring health IOp in maternal and child healthcare services in India. The D-Index shows how access to basic health services varies among children based on their circumstances, with a range from 0 to 100%, where 0 indicates perfect equality. It calculates both the average access rate and the share of services needed to ensure equal access (Barros et al., 2009; Sanoussi, 2017). The HOI evaluates two aspects: the overall coverage of basic health services and the fairness of their distribution. It combines these two factors to give a single measure of how health opportunities are shared across a population (Barros et al., 2009). The HOI reflects both the level of access to health services and the inequality in that access. The RF Classification model predicts the likelihood of individuals accessing a health service based on their circumstances. It uses a functional relationship to estimate this probability as mentioned above, where access is coded as 1 and no access as 0. The model helps estimate the overall probability of access to health services in the population.

This analysis presents a comprehensive overview of the demographic and socio-economic characteristics of the sample population, highlighting the distribution across gender, sector (rural/urban), social group, economic quintiles, regions, and parental education levels. This detailed breakdown is important for understanding the disparities in access to maternal and child healthcare services in India (Table 6.1). The sample includes 190,355 children, with a nearly equal number of boys and girls. Most of these children live in rural areas, with a smaller portion in urban regions. Socially, the largest group of children belongs to Other Backward Classes (OBC), followed by Scheduled Tribes (ST), the general caste group, and Scheduled Castes (SC). Economically, the children come from households with different income levels, with highest concentration in the poorest  households or Q1 (26%) and the lowest concentration in richest households or Q5 (14%). Geographically, the children are spread across different parts of India, with the highest concentration in central India and the lowest in the western region. In terms of parental education, a significant proportion of mothers have secondary education, while less have higher or no education. Similarly, most fathers have secondary education, with relatively less having higher or no education. This data shows the diverse backgrounds of the children, highlighting the complexity of the social and economic setting in India.

In the sample, more than half (56%) of the children had received at least one vaccination, while mothers of less than half (45%) of the children had more than four antenatal care visits (Table 6.2). The majority of mothers (86%) had institutional deliveries, and a significant portion of mothers (66%) had access to important tests during prenatal care. Nearly all mothers (88%) received prenatal care from a trained professional. However, when considering the overall access to all the selected maternal and child healthcare services, only a small portion of children or their mothers (26%) had comprehensive access to these essential services, highlighting gaps in the provision and utilization of maternal and child healthcare services in India.

Place of Delivery: The coverage rate for delivery in private or public health centres is high at 86.5% (Nair & Sadanandan, 2017). This high rate reflects increased efforts in improving institutional deliveries through programs like Janani Suraksha Yojana (JSY), which incentivizes deliveries in health facilities (Jain & Gupta, 2016). The dissimilarly index for the place of delivery is 6.4% (Srivastava & Mishra, 2020). This indicates a moderate level of dissimilarly index, suggesting that redistribution efforts could enhance access to institutional deliveries, particularly in marginalized communities (Ghosh, 2021).

Prenatal Care by Skilled Staff: Access to prenatal care by skilled staff is also high at 88.8% (Gupta et al., 2019). This high rate indicates effective outreach and availability of skilled professionals, though regional disparities persist (Balarajan et al., 2011). The dissimilarly index for prenatal care by skilled staff is 5.2% (Nair & Sadanandan, 2017). This lower dissimilarly index reflects relatively better equity in access compared to other services, though disparities remain in certain regions (Mukherjee & Sikdar, 2019).

General Prenatal Care: The general coverage rate for prenatal care is 65.8% (Singh & Kumar, 2018). Despite significant improvements, some areas still struggle with inadequate prenatal care access due to infrastructural and socio-economic barriers (Kumar et al., 2019). The dissimilarly index for general prenatal care is 8.7% (Gupta et al., 2019). This relatively higher dissimilarly index indicates presence of inequality, with some gaps in access between different socio-economic groups and regions (Roy & Gupta, 2020).

Immunization Services: Immunization services have a lower coverage rate of 55.6% (Pathak & Singh, 2021). This lower rate is attributed to issues such as vaccine supply disruptions and logistical challenges in reaching remote areas (Mishra & Sharma, 2017). Despite the lower coverage rate, the dissimilarly index for immunization services is 6% (Pathak & Singh, 2021). This suggests that even though immunization services are less accessible, the distribution of these services could be improved to reduce disparities (Kumar et al., 2019).

At Least Four Antenatal Visits: Access to at least four antenatal visits is even lower, at 45% (Desai & Sinha, 2021). This is indicative of gaps in routine check-ups and follow-ups, particularly in rural and underserved regions (Pandey & Ladusingh, 2019). The dissimilarly index for access to at least four antenatal visits is 15% (Desai & Sinha, 2021). This higher IOp reflects presence of inequality, indicating that effort is needed to ensure all women receive the recommended number of visits (Jain & Gupta, 2016).

Overall Adequate Health Care: Overall adequate health care, encompassing all necessary maternal and child health services, has the lowest coverage rate at 26% (Patel & Chatterjee, 2020). This reflects significant gaps in comprehensive care delivery, influenced by various socio-economic and regional factors (Srinivasan & Patel, 2015). The dissimilarly index for overall adequate care is the highest at 19% (Patel & Chatterjee, 2020). This higher DI figure highlights the extensive redistribution required to ensure equitable access to comprehensive maternal and child health services (Singh & Kumar, 2018).

Overall, the high coverage rates were observed for institutional deliveries and prenatal care by skilled staff. Prenatal care coverage, immunization, and at least four antenatal visits had relatively lower coverage. Only around one-fourth of children and their mothers had comprehensive access to all selected health services, reflecting gaps in antenatal care and immunization. The IOp for place of delivery, prenatal care by trained staff, immunization stands at low, while four antenatal visits stand at relatively high, which need redistribution for equal access. Overall inequality in healthcare services is at 19%, which is higher than any single service.

Place of Delivery: The HOI for the place of delivery is 0.81, indicating that 81% of the existing opportunities are equitably distributed (Srivastava & Mishra, 2020). This relatively high HOI reflects the effectiveness of interventions aimed at increasing institutional deliveries but also highlights room for improvement (Jain & Gupta, 2016).

Prenatal Care by Skilled Staff: The HOI for prenatal care is 0.83 reflecting that 83% of the prenatal care covered by skilled staff (Nair & Sadanandan, 2017). This high value also suggests that access to skilled prenatal care is relatively well-distributed, although disparities in some regions remain (Gupta & Sinha, 2019).

General Prenatal Care: The HOI for general prenatal care is 0.60, indicating that 40% of did not have access to general prenatal care (Srivastava & Mishra, 2020). This lower HOI indicates significant inequality in access to prenatal care services, with substantial disparities based on socio-economic and regional factors (Kumar et al., 2019).

Immunization Services: The HOI for immunization services is 0.52, indicating that around 48% did not have immunization services (Pathak & Singh, 2021). This reflects moderate distribution, with considerable room for improvement in ensuring equal access across different population groups ((Mishra & Sharma, 2017).

At Least Four Antenatal Visits: The HOI for access to at least four antenatal visits is 0.38, which reflect 68% did not have at least four antenatal visits (Desai & Sinha, 2021). This lower HOI indicates significant inequalities, suggesting that much more needs to be done to achieve equitable access to routine antenatal care (Jain & Gupta, 2016).

Overall Adequate Care: The HOI for overall adequate care services is 0.21, which shows that about 79% did not have adequate access to maternal and child care health services (Patel & Chatterjee, 2020). This very high inequity in access to adequate maternal and child health services, underscoring the poor healthcare services in the country (Singh & Kumar, 2018).

The HOI scores show that high inequality in access to prenatal care, immunization and at least four antenatal visits healthcare services, while the most mothers have access to institutional deliveries, and prenatal care services by skilled staff.

Understanding the factors that influence access to health services is important for creating effective policies and interventions. The variable importance metric helps to identify which factors play significant roles in determining access to various health services (Table 6.4).

Place of Delivery: For the place of delivery, region is the most important variable, accounting for 24.1% of the variation in access to public or private healthcare facilities for delivery. This is because regional disparities often reflect differences in healthcare infrastructure, availability of skilled professionals, and accessibility of medical facilities. Studies in India have shown that urban regions tend to have better healthcare facilities compared to rural areas, leading to higher institutional delivery rates in urban regions (Ghosh, 2021). The quintile group (wealth index) contributes 23%, indicating that wealthier families have better access to quality healthcare facilities for delivery. Wealth enables families to afford transportation to hospitals, pay for medical expenses, and access private health care, which is often perceived as better than public health care (Mukherjee & Sikdar, 2019). Mother's education, accounting for 18.4%, plays a critical role as educated mothers are more likely to be aware of the benefits of institutional deliveries and are better equipped to navigate the healthcare system (Singh et al. 2021). Social group and father's education also play significant roles, contributing 14.7% and 12.1%, respectively. Social group often reflects caste-based disparities in access to health care, with marginalized communities having less access to institutional deliveries (Balarajan et al., 2011). Father's education often influences household decisions regarding health expenditures and healthcare-seeking behaviour (Srivastava & Mishra, 2020). In contrast, the child's gender and sector (whether rural or urban) contribute minimally to access, with 3.8% and 3.9%, respectively. Gender biases in access to delivery care are minimal compared to other health services, and sector differences are already captured by the region variable (Mishra & Sharma, 2017).

Immunization: Access to immunization services is also heavily influenced by region, which accounts for 21.2% of the variation. The quintile group follows closely with a 20.5% contribution. Regional disparities in healthcare infrastructure and outreach programs significantly impact immunization rates. Urban areas with better healthcare facilities and outreach programs tend to have higher immunization rates (Banerjee et al., 2004). The quintile group indicates that wealthier families are more likely to access immunization services due to better awareness, affordability, and access to healthcare facilities (Shrivastava & Shrivastava, 2020). Social group is next at 17.8%, reflecting caste-based disparities where marginalized groups have lower immunization rates due to historical neglect and discrimination in health care (Subramanian & Smith, 2006). Father's education (12.5%) and mother's education (10.9%) also play significant roles. Educated parents are more likely to understand the importance of immunization and ensure their children receive the necessary vaccines (Pathak & Singh, 2021). The child's gender and sector again play minimal roles, contributing 10.1% and 7%, respectively, indicating that gender biases are less pronounced in immunization and that urban–rural differences are already accounted for by the region variable (Kumar et al., 2019).

Prenatal Care: In the case of prenatal care, the quintile group and region are the most critical factors, each contributing 21.1% to access. Mother's education follows with a 17.5% contribution, social group at 16.6%, and father's education at 12.9%. Wealth enables access to quality prenatal care services, and regional disparities reflect differences in healthcare infrastructure (Gupta et al., 2019). Mother's education significantly influences access to prenatal care, as educated mothers are more likely to seek regular check-ups and understand the importance of prenatal care (Roy & Gupta, 2020). Social group disparities highlight the impact of caste-based discrimination in accessing prenatal care services (Acharya, 2023). The child's gender and sector have marginal impacts, contributing 5.1% and 5.4%, respectively, indicating minimal gender biases in prenatal care access and that urban–rural differences are already captured by the region variable (Desai & Sinha, 2021).

Prenatal Care by Trained Staff: Access to prenatal care provided by trained staff is primarily determined by the quintile group (wealth index) at 23.9% and region at 23.1%. Mother's education contributes 16.1%, social group 15.2%, and father's education 12.3%. Wealth enables families to afford care from trained professionals, and regional disparities reflect differences in the availability of trained staff (Nair & Sadanandan, 2017). Mother's education influences the likelihood of seeking care from trained staff, as educated mothers are more aware of the benefits (Paul & Singh, 2019). Social group disparities again highlight caste-based discrimination (Ravindran & Misra, 2020). The sector and child's gender contribute the least, at 4.6% and 4.8%, respectively, indicating minimal impact from gender biases and that urban–rural differences are already accounted for by the region variable.

Antenatal Care: For antenatal care, region plays a major role, accounting for 24.7% of the variation in access. The quintile group contributes 21.3%, mother's education 16.2%, social group 15%, and father's education 11.9%. Regional disparities in healthcare infrastructure and outreach programs significantly impact antenatal care access (Pandey & Ladusingh, 2019). Mother's education influences access to antenatal care, as educated mothers are more likely to seek regular check-ups (Rao & Singh, 2020). Social group disparities reflect caste-based discrimination in accessing antenatal care services (Jain & Gupta, 2016). Sector and child's gender again have minimal impact, with contributions of 5.5% and 5.5%, respectively, indicating minimal gender biases in antenatal care access and that urban–rural differences are already accounted for by the region variable.

Overall Adequate Care: Overall access to adequate care is heavily influenced by region, which contributes 21.4%. This is followed by the quintile group at 19.3%, mother's education at 17.7%, social group at 15.6%, and father's education at 12.9%. Regional disparities in healthcare infrastructure and outreach programs significantly impact access to adequate care (Singh & Kumar, 2018). Mother's education significantly influences access to adequate care, as educated mothers are more likely to seek regular check-ups and understand the importance of adequate care (Kumar et al., 2019). Social group disparities highlight the impact of caste-based discrimination in accessing adequate care services (Srinivasan & Patel, 2015). Sector and child's gender play minor roles, contributing 5.9% and 7.4%, respectively, indicating minimal gender biases in access to adequate care and that urban–rural differences are already captured by the region variable (Patel & Chatterjee, 2020).

The analysis highlights the critical role of socio-economic factors, especially region and wealth index (quintile group), in determining access to health services. Parents’ education, particularly the mother's education and social group, play significantly role in access to health service, while the sector (rural or urban) and child's gender have minimal influence.

The average silhouette score has been used to determine the optimal number of distinct clusters for the regional analysis. This score indicates that the minimum and maximum suitable number of clusters ranges from 2 to 10, which helps in creating geographical clusters in India to understand regional differences in accessing healthcare service inequality (Fig. 6.1). However, using only two clusters does not adequately capture India's regional diversity. Therefore, for a more detailed regional analysis, three clusters have been formed to categorize districts into those with poor access, moderate access, and better access to healthcare services.

The three clusters have been formed out of 706 districts in India. Cluster 1 includes 239 districts, cluster 2 has 193 districts, and cluster 3 comprises 274 districts (Table 6.5). The average access to vaccination was roughly the same across all the districts in the three clusters (around 55–57%), However, districts in cluster 1, when compared to districts in cluster 2 and 3, had lower acess to services such as antenatal care (29%), institutional delivery (78%), prenatal care (55%), and prenatal care by trained staff (81%). Compared to districts in cluster 1, the districts in cluster 3 had higher access to these services, but were lower compared to the districts in cluster 2 (Table 6.5). Therefore based on access to maternal and child healthcare services the cluster are classified in low (Cluster 1), medium (Cluster 3), and high (Cluster 2).

Map 6.1 maps the spatial distribution of these clusters, showing where districts with low, medium, and high access to maternal and child health services are concentrated (Appendix Table 6.6). The districts with low access to these health services (Cluster 1) are primarily found in the eastern and central parts of India, including Uttar Pradesh (UP), Bihar, Jharkhand, northern Chhattisgarh, and the northeastern region of Madhya Pradesh. Conversely, the districts with high access (Cluster 2) are mostly located in southern India, particularly in states like Kerala, Tamil Nadu, and southern Karnataka. Additionally, western and eastern Maharashtra have districts with high access to health services, whereas central and northern Maharashtra have a mix of medium and low access districts. Central Odisha and West Bengal also feature districts with high access to maternal and child care services. This distribution highlights the regional disparities in healthcare access across India.

Better Access to Healthcare Services in Low Performing Cluster: In states like Uttar Pradesh (UP), Jharkhand, and Madhya Pradesh (MP), there are specific districts that perform well in terms of access to health services (Appendix Table 6.7). In UP, districts such as Lucknow, Ghaziabad, and Kanpur fall into Cluster 2, indicating a medium level of access to health services. Similarly, in Jharkhand, districts like Purbi Singhbhum, Kodarma, and Saraikela-Kharsawan show better access compared to other districts in the state. In Chhattisgarh, around 14 districts have medium access to healthcare services, including Raipur and Sukma. Additionally, districts like Uttar Bastar Kanker, Durg, Balod, and Rajnandgaon in Chhattisgarh exhibit high access, outperforming the northern parts of the state. In Madhya Pradesh, which generally has medium and low access districts, Indore and Agra stand out with high access. In Rajasthan, where most districts have medium access, Kota, Chittaurgarh, and Jhalawar are notable for their high access to health services. This highlights that even within states with generally low or medium access, there are pockets of districts that achieve better healthcare access.

Poor Access to Healthcare Services in High Performing Cluster: Though low-access districts are primarily found in Uttar Pradesh (UP), Bihar, and Jharkhand, there are also districts with low access to maternal and child health services in states that generally have medium access (see Table 6.6). In Punjab, Fazilka district struggles with low access to these health services. In Uttarakhand, the districts of Almora, Garhwal, Hardwar, Tehri Garhwal, and Chamoli face similar challenges. In Haryana, Palwal and Mewat have low access to maternal and child health services. Rajasthan has several districts with low access, including Jaisalmer, Churu, Bharatpur, Alwar, Dhaulpur, Sawai Madhopur, and Karauli. In Gujarat, Banas Kantha and Surendranagar districts have low access. In Maharashtra, several northern districts such as Parbhani, Dhule, Nandurbar, and Jalgaon have low access. Lastly, in Karnataka, the northern districts of Gulbarga and Koppal also face low access to maternal and child health services. This shows that while some states may generally have better healthcare access, there are still specific areas within these states that need significant improvements.