Data source
Data from India’s National Family Health Survey (NFHS-4) was utilized for analysis in the present study. NFHS is the Indian version of the Demographic Health Survey (DHS). NFHS-4 was conducted in 2015–16 under the stewardship of the Ministry of Health and Family Welfare (MoHFW), Government of India, and coordinated by the International Institute for Population Sciences (IIPS), Mumbai [17]. NFHS is a large-scale cross-sectional survey that provides reliable data on a range of topics like fertility, family planning, maternal and child health, domestic violence, sanitation and hygiene, morbidity, nutrition, household amenities and women empowerment, and domestic violence. For data collection, a two-stage stratified random sampling design was adopted in NFHS-4. The survey covered all 29 states, 7 union territories, and all 640 districts as per census 2011. In rural areas, villages, while in urban areas, ‘Census Enumeration Blocks’ were chosen as PSUs. Detailed information about the sampling design, the data collected, and the instruments used can be accessed from the NFHS-4 report [17].
Sample size
NFHS-4 collected data from 699,686 women aged 15–49 chosen from 28,583 primary sampling units (PSUs), with a non-response rate of 3%. NFHS-4 provides comprehensive data related to live births that happened 5 years preceding the survey. For the present study, information pertaining to only the last live birth (n = 190,898) was utilized. Due to the missing values, the sample size was slightly different for different outcome variables. Variables on ANC, tetanus injection, and iron folic consumption had missing values due to recall lapse. For example, 1854 women were unable to recall the number of times they went for ANC, resulting in 1854 missing cases for the ANC variable. Some children may have died in infancy or before becoming eligible for full immunization, which is the reason why the variables of full immunization and ‘Child received benefits from Anganwadi center (AWC)/ICDS’ had some missing cases.
Outcome Variable
|
Sample Size
|
Missing Cases
|
---|
Institutional delivery
|
190,337
|
337
|
Skilled birth attendance
|
190,898
|
0
|
Post-natal care
|
190,898
|
0
|
At least 4 ante-natal care visits
|
189,044
|
1854
|
Full immunization
|
184,304
|
6594
|
At least two tetanus injections before birth
|
189,566
|
1332
|
Given/took iron folic tablet/syrup for at least 100 days during pregnancy
|
187,578
|
3320
|
Child received benefits from AWC/ICDS in last 12 months
|
185,101
|
5797
|
Received benefits from AWC/ICDS during pregnancy
|
190,804
|
94
|
Received financial assistance for delivery cost
|
190,337
|
561
|
Received financial assistance under JSY to cover delivery cost
|
190,337
|
561
|
Outcome variables
NFHS provides information on a range of subjects related of MCH services. The present study used 11 variables pertaining to the utilization of MCH services for the analysis. The description of the outcome variables is as follows:
-
Skilled Birth Attendance (SBA): SBA is defined as a delivery conducted in either in a medical institution or at home assisted by a skilled attendant (doctor/nurse/Lady Health Visitor/Auxiliary Nurse Midwife) [31]. The variable was converted into a binary variable (0 = No skilled attendance at delivery; 1 = skilled attendance at delivery).
-
Post-natal care (PNC): It is defined as receiving ‘post-natal care/health check-up’ from a health facility or at home within 48 h of delivery for last live birth. The variable was converted into a binary variable (0 = didn’t receive PNC; 1 = received PNC).
-
Ante-natal care (ANC): ANC is defined as at least 4 ante-natal visits for a pregnant woman, as per the Government of India’s guidelines, to minimize pregnancy-related risks [32]. Data on the number of ANC visit during pregnancy was available in NFHS-4. The variable was converted into a binary variable (0 = less than 4 ANC visits; 1 = at least 4 ANC visits).
-
Full immunization: This is defined as having one dose of BCG vaccine, three injections against DPT, three doses of polio vaccine, and one vaccine against measles [17]. The variable was converted into a binary variable (0 = received partial or no immunization; 1 = received full immunization).
-
No immunization: Children who did not get any dose of BCG, polio, and DPT were categorized as having received no immunization. The variable was converted into a binary variable (0 = received partial or full immunization; 1 = received no immunization).
-
At least two tetanus injections before birth: WHO recommends at least two tetanus injections during pregnancy to avoid the risk of tetanus infection [33]. The respondents were asked about the number of tetanus injections they had received before delivery. The variable was converted into a binary variable (0 = received less than two tetanus injections before birth; 1 = received at least two tetanus injections before birth)
-
Consumed Iron Folic Acid (IFA) tablet/syrup for at least 100 days during pregnancy: The National Health Mission (Govt. of India) recommends taking iron folate tablets or syrup for at least 100 days during pregnancy to avoid anemia [34, 35]. NFHS-4 provided data on iron folate consumption by mothers during pregnancy. The variable was converted into a binary variable (0 = did not consume iron folate/ consumed iron folate for less than 100 days; 1 = consumed iron folate for at least 100 days)
-
Child received benefits from AWC/ICDS in last 12 months: Anganwadi centers and Integrated Child Development Services (ICDS) are established to improve health, nutrition, and education of children. Its beneficiaries include children up to age 6 and pregnant and lactating women. In NFHS-4, respondents were asked whether their newborn children had received benefits from AWC or ICDS scheme during 12 months prior to the survey. The variable was binary in nature (0 = No; 1 = Yes).
-
Received benefits from AWC/ICDS during pregnancy: The variable was binary in nature (0 = No; 1 = Yes).
-
Received financial assistance for delivery cost: The Government of India aims to provide financial assistance to pregnant and lactating women to cover delivery and other related costs through various schemes. A question was asked to the mothers as to whether they had got any financial assistance to cover delivery costs or not. The variable was binary in nature (0 = No; 1 = Yes).
-
Received financial assistance under the Janani Suraksha Yojana (JSY) for delivery cost: JSY is a centrally sponsored scheme of India under which cash assistance is provided to mothers for delivery and post-delivery care [36]. The scheme mainly focuses on pregnant women in rural areas. The variable was binary in nature (0 = No; 1 = Yes).
Exposure variable
The exposure variable of interest was ‘type of region of residence (plain or hilly)’. The states and UTs were divided into two categories: hilly states and plain states. All the states lying in the Himalayan range were categorized as hilly. These states are: Jammu & Kashmir (erstwhile), Himachal Pradesh, Uttarakhand, Sikkim, Arunachal Pradesh, Nagaland, Meghalaya, and Mizoram. Other states like Maharashtra, Madhya Pradesh, and Kerala also have some hilly landscape, but the proportion of hilly area is extremely low. Himalayan states have a significantly higher ecological complexity than hilly areas of the other states. Most of the population in the Himalayan states lives in a knotty hilly terrain because these states have a negligible amount of plain region to live in. Although states like Maharashtra and Madhya Pradesh also have hilly regions, most of those regions are scarcely or unpopulated because the plain region has enough accommodation for the states’ population. This is why, except the Himalayan region states, we did not consider any other state as a hilly state. The present study, therefore, shows disparity in the utilization of MCH services between the hilly states of the Himalayan range (only) and the rest of the country. On the whole, 25,712 respondents were from the hilly Himalayan region, whereas 165,186 were from the plains.
Type of Area
|
Sample Size
|
---|
Hilly
|
25,712
|
Plain
|
165,186
|
Control variables
In literature, a range of variables have been found to affect the utilization of child and maternal healthcare services in India [19,20,21, 37]. In keeping with the literature and the availability of the data, in the multivariate analysis, the association between the outcome and the exposure variables was controlled for the following variables: age of respondent (15–24, 25–34,35–49); age of husband; child marriage (0 = No, 1 = Yes; those married at age 18 or after were categorized as ‘0’, and those married before age 18 years were categorized as ‘1’); place of residence (urban/rural); wealth quintile (poorest, poorer, middle, richer, richest); respondent’s educational attainment (no education, primary, secondary, higher); religion (Hindu, Muslim, Other); Caste (Scheduled Castes (SCs), Scheduled Tribes (STs), Other Backward Class (OBC), Others); sex of head of household (male, female); parity of respondent (1 to 2, 3 to 5, ≥6); and household’s ownership of television (0 = No, 1 = Yes). A pre-calculated wealth quintile, also popularly known as wealth Index, variable is provided in the dataset of NFHS and does not require any manual calculation; it represents the socioeconomic status of the individuals. Further details regarding the generation of the wealth quintile can be accessed from the DHS manual [38].
Statistical analysis
In the present study, univariate analysis was used to summarize the characteristics of the respondents using weighted percentage and frequencies. To explore the relationship between the exposure and the outcome variables, bivariate analysis was performed using chi-square(χ2) tests. The association between the exposure and the outcome variables was analyzed by calculating the adjusted odds ratios (AOR) with 95% confidence interval (CI) using random slope multilevel logistic regression models. The NFHS data is hierarchical in nature, that is, the observations are nested in clusters. So, there are chances that the observations within the clusters may be correlated. If this is the case, a simple logistic regression will underestimate the true variance, leading to an increase in type-1 error [39]. This makes it imperative to control for intra-cluster correlation and cluster-level variations. We achieved this by fitting a two-level random intercept logistic regression model in which clusters were set as level-2 [40]. Since there are four levels in the NFHS-4 data (individual, PSU, district, state), a three-level or four-level regression model could also have been used. But the intra-class correlation for the district or the state levels had a very low value; so the authors decided to use a two-level model and avoid any unnecessary complexity. The benefit of using a multi-level model is that it not only acknowledges the intra-cluster correlation, but also controls the effect of cluster level variables like cluster size, geo-political profile of the cluster, development level of the cluster, accessibility to healthcare facilities in the cluster, etc. The ‘melogit’ command in STATA was utilized to run the models. All the statistical analyses were performed on STATA 16 [41].