ISSN (Print) - 0012-9976 | ISSN (Online) - 2349-8846

A+| A| A-

Exploring the Determinants of Childhood Immunisation

This study attempts to analyse the effects of some selected demographic and socio-economic predictor variables on the likelihood of immunisation of a child for six vaccine-preventable diseases covered under the Universal Immunisation Programme. It focuses on immunisation coverage (a) at the all India level, (b) in rural and urban areas, (c) in Bihar, Tamil Nadu and West Bengal, and (d) for three groups of states, the empowered action group, north-eastern and other states. The study applies a logistic regression model to National Family Health Survey-2 (1998-99) data. The likelihood of immunisation increases with urban residence, mother's education level, mother's exposure to mass media, mother's awareness about immunisation, antenatal care during pregnancy and other such variables. Further research with both demand- and supply-side issues and current data is critical to help policymakers make the immunisation programme universal.

SPECIAL ARTICLEEconomic & Political Weekly EPW march 22, 200897Exploring the Determinants of Childhood ImmunisationNilanjan PatraThis study attempts to analyse the effects of some selected demographic and socio-economic predictor variables on the likelihood of immunisation of a child for six vaccine-preventable diseases covered under the Universal Immunisation Programme. It focuses on immunisation coverage (a) at the all India level, (b) in rural and urban areas, (c) in Bihar, Tamil Nadu and West Bengal, and (d) for three groups of states, the empowered action group, north-eastern and other states. The study applies a logistic regression model to National Family Health Survey-2 (1998-99) data. The likelihood of immunisation increases with urban residence, mother’s education level, mother’s exposure to mass media, mother’s awareness about immunisation, antenatal care during pregnancy and other such variables. Further research with both demand-and supply-side issues and current data is critical to help policymakers make the immunisation programme universal.This paper was presented at the 42nd Annual Conference (January 5-7, 2006) of The Indian Econometric Society, at the Forum-10 of Global Forum for Health Research (October 29-November 2, 2006 in Cairo, Egypt) and also at the Platinum Jubilee Seminar on Gender Issues and Empowerment of Women at Indian Statistical Institute, Kolkata (February 1-2, 2007). Fuller version of the paper is available at http://ssrn.com/abstract=881224.I am grateful to Jean Dreze, Indrani Gupta, Arup Mitra, Ritu Priya, Sanghmitra Acharya, Lekha Chakraborty, Francis Xavier, Puspita Datta, Samik Chowdhury and Dibyendu Samanta. I am also grateful to an anonymous referee for coments. The remaining errors, if any, are solely my responsibility.Nilanjan Patra (nilanjanpatra@gmail.com) is a PhD scholar at Jawaharlal Nehru University, Delhi.In India, under the Universal Immunisation Programme (UIP), vaccines for six vaccine-preventable diseases (tuberculosis, diphtheria, pertussis (whooping cough), tetanus, poliomyeli-tis and measles) are available free of cost to all. The UIP was launched in 1985 with much dynamism to attain the target to immunise all eligible children by 1990. A lot of energy and money has been spent on the UIP but various survey results show a glar-ing gap between the targets and achievements even after several years. Given the tight budgetary allocations, one should take care of the effectiveness of the programme. This paper attempts to find the causes of poor immunisation coverage rate in India. There are some bottlenecks from both the supply- and demand-side. In a developing country like India, any programme like the UIP could be affected by supply-side financial constraints when the overall central and state budgetary allocations on healthcare are meagre. Moreover, the availability of supply-side data at dis-aggregated level is rare. Thus supply-side analysis is beyond the scope of this study. It concentrates purely in the demand-side assuming the ceteris paribus supply-side constraints. The report of the sub-committee on national health prepared for the consideration of National Planning Committee of the Indian National Congress had advocated state intervention to preserve and maintain health of the people by organising and controlling healthcare to achieve proper integration of curative and preventive services [National Planning Committee 1948: 224-25]. TheUIP was launched as a carefully planned strategy, in 1985-86, aimed at systematic districtwise expansion to cover all the districts by 1989-90 [GoIMoHFW 1985; Sokhey 1985]. More than 90 million pregnant women and 83 million infants were to be immunised over a five-year period as per the programme [Sokhey 1988]. The programme was given the status of a National Technology Mission in 1986 [GoI 1988] to provide a feel-ing of urgency and commitment to achieve the goals within the specified period. Later, theUIP became a part of the Child Survival and State Motherhood(CSSM) Programme in 1992 and Reproductive and Child Health(RCH) Programme in 1997 [MoHFW 2002-03:176]. The GoI constituted a National Technical Committee on Child Health on June 11, 2000 and launched the Immunisation Strengthening Project on recommendation of the committee [MoHFW 2002-03:173]. The department of family welfare established a national technical advisory group on immunisation on August 28, 2001 to assist the government of India in developing a nationwide policy framework for vaccines and immunisation [Annual Report, MoHFW 2002-03: 174].
SPECIAL ARTICLEmarch 22, 2008 EPW Economic & Political Weekly98The vaccine-preventable diseases have many socio-economic costs: sick children miss school and can cause parents to lose time from work. These diseases also result in doctor’s visits, hospitali-sation, poor health and even premature deaths. Vaccinations are one of the best ways to put an end to the serious effects of certain diseases. Vaccination not only protects the children of today, but it also protects the next generation. Several survey results support evidence of a glaring gap between the goals aspired for and the targets touched. To quote, “...achievement of the target of protecting 100 per cent of preg-nant women withTT (tetanus toxoid) and 85 per cent of infants with vaccines …remains a distant dream” [Gupta et al 1989: 160]. This national review mentioned some supply-side bottlenecks that may prevent the UIP from achieving its goals. But Padmana-bha (1992) argues that “…the programme suffers not so much from lack of funds as from functional isolation”. Public health should not be treated as the sole responsibility of the health sector. Policies and programmes in other sectors such as environment, education, welfare, industry, labour, information, etc, have also to be informed and influenced by public health considerations [Gopalan 1994].No matter how noble the idea of the UIP, a “non-controversial” programme of GoI, it faces severe criticism from many scholars. Banerjee (1986, 1993) pointed out that it is a part of an “ill- conceived and unimaginative global venture” and “...revealed many serious flaws in the programme itself. The most important among them was that a massive, expensive and a very compli-cated programme had been recommended for launching without even finding out what the problem was, leave alone the other im-portant epidemiological considerations, such as incidence rates under different ecological conditions and time trends of the chosen diseases.” Banerjee (1993) also mentioned that the pro-gramme reflects a totalitarian approach of the developed north to “sell” their “social” products in the vast “market” of developing south deviating from the Alma Ata Declaration [WHO 1978]. Banerjee (1990) dubsUIP as “an unholy alliance of national and international power brokers (who) could impose their will on hundreds of millions of human beings living in the poor countries of the world…”. Madhavi (2003) also noted that the immunisation policy in India, instead of being determined by disease burden and demand, is increasingly driven by the sup-ply push, generated by industry and mediated by international organisations.1 Data and MethodologyThe present study uses data from the National Family Health Survey(NFHS)-2 (1998-99). The NFHS-2 covers a representative sample of more than 90,000 ever-married women of age 15-49 years from 26 states of India that comprise more than 99 per cent of India’s population. Though it has some limitations, it is regard-ed as a “storehouse of demographic and health data in India” [Rajan and James 2004].NFHS-2 data on immunisation is based on the vaccination card for each child born since January 1995 (or since January 1996 in states in which the survey began in 1999) or on the mother’sreport in case of non-availability of the card. The 12-23 month age group was taken for the present analysis because both international and the GoI guidelines specify that childrenshould be fully immu-nised by the time they complete their first year of life. InNFHS-2, children who received the Bacillus Calmette-Guerin (BCG), measles, and three doses each of diphtheria, pertussis (whooping cough) and tetanus (DPT) and polio (excluding polio 01) are considered to be fully vaccinated. Based on the information obtained from “either source”, 42 per cent of children are fully vaccinated and 14 per cent have not received any vaccinations (IIPS: NFHS-2, India, Table-6.9, pp-204). The analysis of vaccine specific data indicates much higher coverage of all vaccines in urban (61 per cent) than rural areas (37 per cent) for children aged 12-23 months. The dropout rates for bothDPT and polio are lower in urban areas than in rural areas. The immunisation cov-erage in India had improved slightly since the time of NFHS-1 (1992-93) when the proportion of fully vaccinated children was 36 per cent (an increase by 6 percentage points in six years!). But these marginal improvements indicate that achievement is lagging far behind the goal of universal immunisation programme in India.An immunisation coverage model is used in this study to esti-mate the effects of the selected background variables on cover-age. The measure of a child’s immunisation is a binary variable that indicates whether a child has had all six vaccinations or not. The analysis uses bivariate (unadjusted) and multivariate (adjusted) binary logit regression tools. Logistic regression results are pre-sented in multiple classification analysis(MCA) form. Probability (p) is presented in percentage form (multiplying p by 100). The unadjusted values are calculated from logit regressions incorporating only one predictor variable. Adjusted values are calculated from logit regressions incorporating all predictor vari-ables simultaneously. When calculating the adjusted values for a particular predictor variable, all other predictor variables are controlled by setting them to their mean values in the underlying regression [Patra 2005; Retherford and Choe 1993].2 Determinants of Full Immunisation in IndiaChildren are the units of the present analysis. A child data file is created by merging selected household and mother’s characteri-stics from the household and women’s data files, respectively. Thus, the child data file contains selected characteristics of children aged 12-23 months, selected characteristics of their mothers and selected characteristics of the households in which the mother and child reside. The analysis focuses on the 10,076 children of 12-23 months of age during the survey.The analysis of immunisation coverage uses a number of demo-graphic and socio-economic variables. The dependent variable is full immunisation that says whether a particular child is fully immunised or not. The selected predictor variables are sex of the child (female, male), birth order of the child (1, 2, 3, 4 and above), residence (rural, urban), mother’s education (illiterate, less than middle school complete, middle school complete, high school complete and above), mother’s age (15-19, 20-24, 25-29, 30-49), antenatal care (no, yes), religion (Hindu, Muslim, Christian and other minorities), caste/tribe (general, other backward castes, scheduled caste, scheduled tribe), standard of living index (low,
SPECIAL ARTICLE

medium, high), media exposure (no, yes), mother’s awareness Figure: Percentage Distribution of Mother’s Empowerment Index by States (percentage) Goa

(no, yes), sex of household head (female, male), mother’s em-

Tamil Nadu powerment index (MEI) (low, medium, high), zone of states Meghalaya Gujarat

(central, north, east, north-east, west, south) and electricity

Arunachal Pradesh

(no, yes). Mean values (in percentage) of the variables are

Mizoram presented in Table 1. Kerala New Delhi

An attempt has been made to construct an indicator (MEI) to

Punjab see how mother’s decision-making power in the household affects Maharashtra Karnataka

the likelihood of immunisation. Such an index could vary widely

Manipur Haryana

Table 1: Mean Values* of Variables

Bihar Variable (%)

Mean

Andhra Pradesh India Rural Urban Bihar TN WB EAG NE Other

Sikkim Full Immunisation (yes) 42.0 36.6 60.4 11.1 88.9 43.8 20.1 20.2 65.7 Himachal Pradesh

Madhya Pradesh Nagaland

Sex of child (male) 51.2 51.4 50.7 51.9 52.6 50.3 51.5 56.8 50.6

Birth

order

Uttar Pradesh

1# 29.3 27.4 36.1 24.0 42.6 33.8 23.4 27.5 35.4

Rajasthan

2 26.4 25.6 29.3 25.3 35.7 31.1 22.6 25.4 30.4

Orissa

3 17.9 18.3 16.6 18.5 12.9 16.5 18.5 15.7 17.4Assam West Bengal

4+ 26.3 28.7 18.0 32.2 8.8 18.6 35.5 31.4 16.8

Jammu and Kashmir

Residence (urban) 22.6 – – 9.1 34.6 17.9 15.8 10.2 30.4

0 20 40 60 80 100

Mother’s

education Illiterate# 58.2 65.3 34.2 75.6 41.9 52.6 72.4 51.2 44.4

Low Medium High

Lit, < mid sch com 17.7 17.2 19.2 10.3 20.6 28.4 11.6 25.0 23.3

with changes in its components or their weights. Percentage distri

Middle sch comp 9.1 8.1 12.5 4.2 18.1 9.7 6.4 13.6 11.6

bution of MEI2 by states is shown in the figure. From the figure, it is

High sch comp and + 14.9 9.3 34.1 9.9 19.4 9.3 9.6 10.2 20.7

evident that excluding Bihar, the other Empowerment Action

Mother’s

age Group (EAG) of states are among the bottom eight states, but the

15-19# 12.2 13.3 8.3 13.7 8.1 14.4 11.7 11.7 12.7

three north-eastern (NE) states (Arunachal Pradesh, Mizoram and

20-24 39.6 39.6 39.5 40.1 45.4 39.5 36.8 32.4 42.9 25-29 29.8 28.7 33.4 27.6 34.3 30.1 28.5 30.0 31.1Meghalaya) are among the top six states in terms of MEI.

30-49 18.4 18.3 18.7 18.6 12.1 16.0 23.0 25.8 13.3 For the variable zone of states, central includes Madhya Antenatal care (yes) 62.3 55.5 85.5 30.3 95.1 88.2 39.4 60.2 85.7 Pradesh and Uttar Pradesh; north includes Delhi, Haryana,

Hindu# 78.8 81.0 71.4 80.9 87.5 67.1 84.3 44.4 75.6

Religion Himachal Pradesh, Jammu and Kashmir, Punjab and Rajasthan;

east includes Bihar, Orissa and West Bengal; north-east includes

Muslim 15.9 14.1 22.1 18.2 6.7 30.3 14.4 31.2 16.4

Arunachal Pradesh, Assam, Manipur, Meghalaya, Mizoram,

Christ and minorities 5.3 4.9 6.4 0.9 5.8 2.6 1.30 24.3 8.0

Nagaland and Sikkim; west includes Goa, Gujarat and Maharashtra;

Caste/tribe General# 37.9 34.3 49.8 17.5 1.1 62.8 32.4 47.7 42.7south includes Andhra Pradesh, Karnataka, Kerala and Tamil

SC 20.4 21.6 16.2 24.4 27.6 26.1 21.5 8.9 20.1Nadu. Media exposure includes whether a children’s mother ST 9.4 10.8 4.6 6.0 1.2 7.4 9.9 37.3 7.0reads newspaper once a week or watches TV every week or listens OBC 32.3 33.2 29.4 52.1 70.1 3.7 36.2 6.1 30.2 to radio every day or every week. Mother’s awareness includes

SLI whether immunisation was discussed with family planning workers

Low # 36.5 42.6 15.7 56.6 40.4 50.9 39.8 46.4 32.4

or during health facility visits.

Table 2: Hypothetical Relationship

Medium 47.0 46.7 48.0 35.0 46.1 41.0 46.5 45.5 47.6

The hypothesised direction of of Variables with Full Immunisation

High 16.5 10.7 36.3 8.4 13.6 8.1 13.7 8.0 20.0

Variable Hypothesised

relationship between dependent Sign

Media exposure (yes) 54.9 46.2 84.7 27.6 81.6 52.6 39.0 51.5 71.2

Sex of child + Mother’s awareness (yes) 35.7 33.6 42.8 23.0 75.5 61.7 21.4 27.4 50.6 variable and each of the predictor Birth order + Sex of HH-head (male) 93.5 94.0 91.6 94.8 91.3 91.9 94.4 93.4 92.6 variables are presented in Table 2.

Residence +

MEI Before going to the regression

Mother’s education + Low # 77.4 80.3 67.6 78.4 41.6 93.3 84.6 82.9 69.8

results, it is important to look at

Mother’s age + Medium 12.4 11.3 16.2 11.5 22.2 4.5 9.2 9.8 15.9

the possible collinearities among

Antenatal care + High 10.2 8.4 16.2 10.1 36.2 2.2 6.2 7.3 14.3

the predictor variables to avoid the

Religion +/-Zone

problems of multicollinearity. In Caste/tribe +/

Central# 28.1 29.9 22.1 – – – – – –

most real life observational re-Std of living index +

North 12.4 11.3 16.0 – – – – – –

East 22.0 24.9 11.9 – – – – – –search (as opposed to experimen- Media exposure +

North-east 3.3 3.8 1.5 – – – – – –tal research, where treatments can Mother’s awareness + Sex of HH-head

West 13.8 11.1 23.0 – – – – – –be randomised), a certain amount Mother’s empowerment

South 20.5 19.0 25.6 – – – – – – of multicollinearity is inevitable,

index +Electricity (yes) 55.4 45 90.8 16.4 79.5 25.6 39.0 33.1 73.5

because most of the predictor

Zone +/-Number of children 10076 7795 2281 1120 566 763 4901 332 4844

variables (such as mother’s age

Electricity +

#: Reference category; *: Mean value of a variable represents the set of proportions of children falling in each

category of that variable. Standard deviations of the variables are not reported. and birth order of children) are HH- Households.

Economic & Political Weekly EPW march 22, 2008

SPECIAL ARTICLEmarch 22, 2008 EPW Economic & Political Weekly100correlated to some extent. As a thumb rule, when two predictor variables are correlated and both are relevant for explanation from a theoretical point of view, one should not eliminate one of the variables to reduce multicollinearity, unless the correlation coefficients are higher than about 0.8 [Retherford and Choe 1993: 39-40; Hill et al 2001: 264 (threshold to be 0.9)]. But the Pearson Correlation Matrix (not reported) shows that the maxi-mum correlation coefficient is 0.6 which is much less than the threshold magnitude. Also given the huge observations in the data, the present analysis enjoys the luxury of keeping all the predictor variables.2.1 Effect on Full Immunisa-tion Coverage in IndiaThere is some gender discrimina-tion in being vaccinated in India (see India column of Table 3) though the vaccines are freely available. In India, boys are sig-nificantly more (5 per cent for un-adjusted, 10 per cent for adjusted) likely to be fully immunised than girl children. Some researchers also noted such behaviour of fami-lies to neglect and discriminate against girl children [Das Gupta 1987; Islam et al 1996; Rajeshwari 1996]. However, Hill and Up-church (1995) noted that although there are substantial mixed varia-tions in immunisation coverage by sex, the median difference across all countries is very close to zero.There is a consistently inverse relationship between immunisa-tion coverage and birth order of a child. The different likelihoods of immunisation for different birth orders are also strongly significant.One can think of two counter-vailing effects of higher-order births on likelihood of vaccina-tion. The positive one could be some kind of learning effect about immunisation which almost does not vary or may increase margin-ally with higher birth-order. The negative one could be some kind of negligence effect, and this ef-fect perhaps, increasingly increas-es with higher birth-order. Thus for higher order births, it seems that the negligence effect more than offsets the learning effect.Another variable, namely, sex-wise birth-order (an interaction variable of sex of the child and his/her birth-order), is construct-ed to see whether likelihood of vaccination decreases with in-crease in birth-order for girls only or not. Likelihood (unadjusted) of vaccination decreases with increase in birth-order irrespective of the sex of a child, and surprisingly, the rate of decrease is lower for girl children except third birth-order (Table 4, p 101). Urban children are much more (62 per cent for unadjusted) likely to be fully vaccinated than rural ones. Higher immunisa-tion coverage in urban areas is supported by many researchers [Padhi 2001; Pebley et al 1996]. But the adjusted effects are almost same and the rural-urban disparity is not statistically significant. It suggests that the unadjusted effect of rural-urban residence is actually due to the other predictor variables correlat-ed with residence. There is a strong positive rela-tionship between mother’s educa-tion and children’s immunisation coverage. The likelihood is almost three times higher for the children of mothers with high school or above education than the children of illiterate mothers. The adjusted effects are lower than unadjusted ones but still strongly significant and the effect levels off at higher level of education. Such a positive effect of maternal education is also hypothesised by Desai and Alva (1998), Gage et al (1997), Islam et al (1996), Mosley and Chen (1984), Padhi (2001) and Pebley et al (1996) though Gauri and Khaleghian (2002) finds a spurious effect.The variable, father’s education was also tried to examine how the likelihood of vaccination is affect-ed by it as around 60 per cent of Indian mothers are illiterate. The effect of father’s education (unad-justed) is significantly positive but its extent is less than that of moth-er’s education (Table 4).The chances of immunisation of children increase with their mother’s age only up to the age group of 25-29 and then decreas-es. A positive relationship is also noted by Steele et al (1996). In the context of rural Bangladesh, Islam et al (1996) shows that likelihood of vaccination decreas-es for the mothers older than 28 years.Table 3: Summary of Effects (P in %) on Full Immunisation Coverage IndiaRuralUrbanBackgroundVariables Unadjusted Adjusted Unadjusted Adjusted Unadjusted AdjustedSex of child Female# 41* 39 35* 31 61* 64 Male43**43**38*34*6063Birth order 1# 54* 49 48*** 42 69* 71 2 49* 43* 44* 36* 65**64** 3 39* 35* 34* 29* 58* 59* 4+24* 35* 22* 24* 38* 49*Residence Rural# 37* 41 – – – – Urban 60* 42– – – –Mother’s Illiterate#28*36 26*2939*51education Lit,<mid52* 45* 48* 36* 65* 67* Mid sch. 63* 52* 59* 43* 71* 71* Highsc+ 73* 52* 69* 44* 76* 69*Mother’s age 15-19# 37* 28 35* 22 47 45 20-24 45* 38* 40* 30* 61* 58* 25-2946**47* 40* 39* 64* 66* 30-49 33* 47* 25* 36* 60* 74*Antenatal care No# 18* 30 17* 23 28* 52 Yes 57* 48* 53* 41* 66* 65*Religion Hindu#42* 4237* 3363* 65 Muslim33* 32* 25* 25* 49* 55* Christian 64* 56* 59* 49* 77* 69Caste/tribe General#47* 42 40* 34 63* 64 OBC43*4138326365 SC40*4437***37***53*62 ST26* 31* 24* 23* 46* 51***Standard of Low# 30* 39 29* 31 43* 65living index Medium 43* 40 39* 33 57* 60 High65* 46* 58* 37**72* 66Media exposure No# 25* 38 24* 30 38* 61 Yes 56* 43* 52* 35* 65* 64Mother’s No# 33* 36 28* 28 52 56awareness Yes 58* 51* 53* 42* 72* 71*Sex of HH-head Female# 48 40 40* 32 65* 64 Male42*4136***336063Mother’s Low# 39* 41 34* 3256* 63empowerment Medium 51* 40 44* 32 68* 62index High58* 4350* 3472* 65Zone Central#22* 28 19* 2336* 42 North 43* 39* 36* 31* 58* 58* East 27* 31*** 25* 25 44** 46 North-east2021**1715*4645 West71* 66* 68* 61* 75* 76* South 70* 60* 66* 52* 79* 77*Electricity No# 24* 37 24* 3032* 46 Yes 57* 44* 52* 36* 63* 65*#: Reference category; Significance level: ***10%, **5%, *1%.
SPECIAL ARTICLEEconomic & Political Weekly EPW march 22, 2008101Antenatal care during pregnancy is positively associated with child vaccination. The chances of immunisation are three times (unadjusted) or more than one-and-a-half times (adjusted) high-er for the children of mothers’ with some antenatal care than the children of mothers’ with no antenatal care. Such a positive re-lationship is also noted by Islam et al (1996). The likelihood of immuni-sation seems to vary with reli-gion also. The likelihood of being fully immunised is 42 per cent for children from Hin-du households, 33 per cent for children from Muslim house-holds and 64 per cent for chil-dren from Christian and other minority community house-holds. The adjusted chances are 42, 32 and 56 per cent, re-spectively. Caste/tribe also seems to affect immunisation coverage. The chances of being fully vaccinated are 47 per cent for children from general category households, 43 per cent for children from other back-ward class(OBC) households, 40 per cent for children from sched-uled caste(SC) households and 26 per cent for children from scheduled tribe(ST) households. The result is, surprisingly, con-sistent with the relative hierarchy of castes/tribes. But the ad-justed chances do not vary significantly except in the case of STs. This implies that the adjusted effect ignores some important ef-fects of other variables correlated with caste/tribe and the unad-justed differences by caste/tribe stem mainly from the relatively lowersocio-economic status of families belonging to backward castes/tribes.The likelihood of immunisation increases with standard of living index (SLI) of children’s households. The unadjusted chances are 30 per cent for children from low SLI households, 43 per cent for children from mediumSLI households and 65 per cent for children from highSLI households. When all other predictor variables are controlled, the chances do not vary significantly except for children from highSLI households. It indicates that the effect ofSLI on full immunisation largely disappears, suggesting that the unadjusted likelihoods actually reflect the effects of other variables (e g, education) that are correlated withSLI. Theresult is consistent with expectation, as underUIP vaccines are available free of cost. Mosley et al 1984 also argues for house-hold income as a proximate determinant of immunisation cover-age. Islam et al 1996 also noted such positive relationship with household income.Media exposure has a significantly positive effect on immunisa-tion. The chances of full immunisation are higher (124 per cent for unadjusted, 13 per cent for adjusted) for children of mothers’ who have some media exposure compared to children whose mothers’ are not exposed to mass media. But Gauri et al 2002 does not find any significant effect of media.Mother’s awareness about immunisation also has a strong posi-tive effect on vaccination. The chances of full immunisation are higher (75 per cent for unadjusted, 42 per cent for adjusted) for children of mothers’ with some awareness than children of una-ware mothers’. The unadjusted chance of being fully immunised is 48 per cent for children from households with female headship and 42 per cent for children from households with male headship. But the adjusted chances do not vary significantly. It implies that the sex of household headship affects immunisation mainly through other predictor variables (e g, SLI) correlated with sex of house-hold headship. However, in the context of rural Orissa, Panda 1997 shows that children from male-headed households are more likely to be immunised than those from female-headed house-holds. Moreover, he shows that the gender inequality (boys are more likely than girls) in preventive healthcare persists regard-less of the gender of the household headship.Both unadjusted and adjusted effects of mother’s empower-ment index are almost positively related to immunisation cover-age. The unadjusted chances of being immunised are 39 per cent for children of mothers with low MEI, 51 per cent for children of mothers with mediumMEI and 58 per cent for children of moth-ers with highMEI, but the adjusted chances do not vary signifi-cantly. It indicates that the effect of MEI on full immunisation largely disappears, suggesting that the unadjusted likelihoods actually reflect the effects of other variables (e g, mother’s employment type) that are correlated withMEI.The variable, mother’s employment type, has also tried to see how likeli-hood of vaccination affected by it, as most Indian mothers do not contri-bute to total family earnings. Likeli-hood (unadjusted) decreases for a child whose mother is a non-wage employee, but increases (though not significant) for children whose mother is a wage employee compared to the non-working mothers (Table 4). The immunisation rate varies widely across different zones as well as within the same zone. For example, in north zone, coverage varies from 83 per cent (Himachal Pradesh) to 17 per cent (Rajasthan). The low likeli-hood in north-east is mainly due to high weight given to Assam (with 233 observations out of a total of 332 observations for north-east or 70 per cent of the total observations) that has only 17 per cent coverage rate of full vaccination.Availability of electricity has also a significantly strong positive role (ele-vates by 138 per cent for unadjusted and 19 per cent for adjusted) on full Table 4: Unadjusted Effects on Full Immunisation Coverage in IndiaBackground Variables P (in %)Sex-wise Female,birth-1#53**birth-order Female,birth-2 49** Female,birth-336* Female, birth-4 + 23* Male,birth-155 Male,birth-249*** Male,birth-342* Male, birth-4 + 25*Father’s Illiterate# 27*education Lit, < mid sch 40* Middle sch comp 47* High sch + 56*Mother’s Not working# 43*employment Work,non-wage 36* Work,wage44#: Reference category; Significance level: *** 10%, ** 5%, * 1%, P is probability.Table 5: Adjusted Effects of Demographic Factors in IndiaBackground Variables P (in %)Sex of child Female# 38 Male41**Birth order 1# 53 2 44* 3 34* 4 + 25*Residence Rural# 38 Urban46*Mother’s age 15-19# 23 20-2436* 25-2947* 30-4947*Antenatal care No# 25 Yes49*Religion Hindu#40 Muslim29* Christian58*Caste/tribe General#43 OBC39* SC40** ST27*Sex of HH-head Female# 40 Male39Zone Central#25 North39* East28 North-east19** West67* South61*#: Reference category; Significance level: ** 5%, * 1%.
SPECIAL ARTICLEmarch 22, 2008 EPW Economic & Political Weekly102immunisation in India. Such a positive effect possibly works through electronic mass media, establishment of an institutional health facility in the vicinity, higher SLI, etc. Islam et al 1996 also noted such a positive relationship.2.2 Effect on Full Immunisation CoverageSeparate regressions for rural and urban areas have tried to show clearly how the effects vary due to change in place of residence in lieu of a residence dummy. These regression results are compared with the all-India “reference” regressions. Unadjusted and adjust-ed effects on full immunisation coverage for rural (sample size 7,795) and urban India (sample size 2,281) are presented in Table 3. The likelihood of being fully immunised is slightly favourable to girl children in urban India in contrast to their counterparts in rural India or all-India. The relationship between the child’s birth order and likelihood of immu-nisation becomes strictly negative in both cases. Mother’s education has a strictly posi-tive impact on immunisation in rural areas but in urban areas, after controls, it increases only up to the middle school level of mother’s education. In rural areas, the likelihood of vaccination increases with the mother’s age only up to the 25-29 year age group but in urban India, after controls, it increases directly with mother’s age. Effects of religion and caste/tribe are much weaker in urban area. Effects of other variables remain same as the baseline regression.2.3 Adjusted Effect of Demographic FactorsHere a separate regression is tried, incorporating only the demo-graphic factors to see their independent effect. The adjusted ef-fects of demographic factors on full immunisation coverage in India are shown in Table 5 (p 101).Urban children are significantly more likely to be vaccinated even if the rural-urban gap vanished after controls in all-India regression. It implies that the unadjusted likelihood for residence in all-India regression capture mainly the effects of the selected socio-economic variables. Hence, it can be assumed that the rural-urban disparity is not due to the demographic factors but the socio-economic factors. 2.4 Adjusted Effect of Socio-economic FactorsHere another regression has been attempted, incorporating only the socio-economic factors to see their independent effect. The adjusted effects of socio-economic factors on full immunisation coverage in India are shown in Table 6.The relationship between mother’s education and immunisa-tion becomes strictly positive here. Effect of SLI of children’s family isU-shaped as in the case of urban India. Effect of MEI on immunisa-tion becomes strictly positive. It implies that the unadjusted likeli-hoods for MEI in all-India regression capture mainly the effects of the selected demographic variables. 3 Extension:3 State-specific Pattern Three states of India, namely, Bihar, Tamil Nadu and West Ben-gal are selected for state-level analysis. These states were select-ed because Bihar (11 per cent) and Tamil Nadu (89 per cent) are two extreme cases and West Bengal (44 per cent) is one with just above the national average (42 per cent) in terms of coverage of full vaccination. Adjusted Effect on Full Immunisation in Bihar: The adjusted effects on full immunisation coverage in Bihar (including Jharkhand) are presented in Table 7 (p 103) for 879 children. Higherbirth-orderchildrenare less likely to be vaccinatedexcludingsecond order (not signifi-cant) births. Residence has a significantly posi-tive effect favouring urban children. Relation-ship between mother’s education and immuni-sation becomes inverted-U shaped. Immunisa-tion chance does not affect significantly by mother’s age or antenatal care or caste/tribe or media exposure or mother’s awareness or MEI or electricity. The chances of immunisa-tion significantly decrease for children from male-headed households compared to those from female-headed households.Adjusted Effect on Full Immunisation in Tamil Nadu: The likelihood of immunisation is not significantly much affected by almost all the predictor variables. Chances of vaccination are almost certain for the children of Tamil Nadu. Herd immunity is already achieved by Tamil Nadu and in near future hopefully it will achieve universal immunisation. The programme managers of UIP could cite Tamil Nadu as a model as far as the performance of vaccination is concerned. The adjusted effects on full immunisation coverage in Tamil Nadu are presented in Table 7 for 430 children. Gender discrimi-nation in immunisation is not significant. But residence has a sig-nificant positive impact favouring urban children. The likelihood of Immunisation does not get affected significantly by birth order or mother’s education or religion or caste/tribe or SLI or media exposure or mother’s awareness or sex of household head or MEI.Adjusted Effect on Full Immunisation in West Bengal: The ad-justed effects on full immunisation coverage in West Bengal are presented in Table 7 for 398 children. Gender discrimination in immunisation is also not significant here. Higher birth-order chil-dren are less likely to be vaccinated except in the last category. Immunisation likelihood does not get affected significantly by res-idence or antenatal care or SLI or sex of household head. Chances of immunisation increased significantly for children of at least middle school educated mothers. The likelihood increases with mother’s age up to 25-29 year age group and then decreases. OBC children are least likely to be vaccinated. Adjusted Effects on Full Immunisation: A group of backward states with poor socio-demographic indicators was formed as Table 6: Adjusted Effects of Socio-economic Factors in IndiaBackground Variables P (in %)Mother’s education Illiterate# 31 Lit, < mid sch com 47* Middle sch comp 55* High sch comp + 61*Standard of living Low# 43index Medium38* High39**Media exposure No# 34 Yes46*Mother’s awareness No# 33 Yes54*Mother’s Low# 39empowerment Medium 45*index High 49*Electricity No# 29 Yes50*#: Reference category; Significance level: ** 5%, * 1%.
SPECIAL ARTICLEEconomic & Political Weekly EPW march 22, 2008103Empowered Action Group (EAG), consisting of Bihar (including Jharkhand), Madhya Pradesh (including Chhattisgarh), Orissa, Rajasthan and Uttar Pradesh (including Uttaranchal). The group was formed in 2001 under the MoHFW to design and implement area-specific programmes to strengthen the primary healthcare infrastructure. The group of north-eastern(NE) states consists of Arunachal Pradesh, Assam, Manipur, Meghalaya, Mizoram, Nagaland and Sikkim (excluding Tripura). The remaining 13 states (Andhra Pradesh, Goa, Gujarat, Haryana, Himachal Pradesh, Jammu and Kashmir, Karnataka, Kerala, Maharashtra, Punjab, Tamil Nadu, West Bengal and Delhi) are clubbed as other states. Immunisation coverage rates are 20.1, 20.2 and 65.7 per cent and the sample sizes are 4,244, 332 and 4,359 for EAG, NE and Other group of states, respectively. Effects on full immunisation for EAG, NE and the Other group of states are given in Table 7 and these are compared with the na-tional level effects. Male children are more likely to be vaccinated in each case. Children of higher birth-order are less likely to be vac-cinated except the north-eastern children of fourth or higher birth-order (not significant). Urban chil-dren are more likely to be immu-nised (not significant) in each case. Children of more educated moth-ers are more likely to be immu-nised except the children of moth-ers with at least high school educa-tion in EAG states. The likelihood of immunisation increases with mother’s age up to 25-29 year age group everywhere except NE states (not significant). Children of moth-ers with some antenatal care are more likely to be vaccinated. Mus-lim children are least likely to be immunised and Christian and other minority community children are most likely to be vaccinated in each case. ST children are least likely to be vaccinated in each case except in NE states (not signifi-cant). The effect of household SLI is almost positive everywhere, but the NE states (not significant). Ef-fects of media and mother’s aware-ness are both positive. Likelihood decreases for children from male-household headship in EAG and NE states (none significant). Likeli-hood increases with MEI in NE states and Other states but the relationship becomes U-shaped for EAG states and India as a whole. Electricity has a positive effect in each case.4 ConclusionsSix vaccine-preventable diseases are covered under the UIP, and vaccination is given free of cost to every child in India. Though vaccines are available for free, the goals of UIP are far from being achieved after almost one and a half decades since its inception. The present study attempts to investigate the demographic and socio-economic determinants of immunisation in India. It is pos-sible to give a big push to the immunisation uptake, only when one understands the demand-side factors well, to achieve the chartered goals of UIP.Robust Results: (i) Boys are more likely to be immunised than girl children. (ii) Children of higher-order births are less likely to be vaccinated. This is true irrespective of the sex of a child, but the rate of decrease is higher for girl children, except third birth-order. It seems that the negligence effect more than offset the learning effect. The result perhaps shows the apathy on part of the parents to immu-nise their children of higher- order births. (iii) The likelihood of immunisation is higher for children from urban areas. Rural-urban disparity in vaccina-tion is not due to the demograph-ic factors but by the socio- economic factors.(iv) Likelihood of vaccination increases with mother’s education level, mother’s age up to 29 years, mother’s exposure to mass media and mother’s awareness about immunisation. (v) Some antena-tal care during pregnancy raises immunisation chances significantly. This increases the possibility to meet health personnel who help mothers, to raise awareness by disseminating information re-garding immunisation. (vi) Among the religious groups, Muslim children are least likely to be immunised whereas children from Christian and other reli-gious minority communities are most likely to be immunised. (vii) The chances of immunisa-tion increase with the standard of living index of children’s household. (viii) Children from the west zone are most likely to be immunised, followed by south, north, east, central and north-east, respectively. (ix) Children from the house-holds with electricity are more likely to be immunised. Table 7: Adjusted Effects (P in %) on Full Immunisation CoverageVariables BiharTNWBIndiaEAGN-EOther StatesStatesStatesSex of child Female# 6 93 40 39 15 11 68 Male9**944443**18*19***69Birthorder 1# 10 95524921 18 76 2 11924443* 201569* 3 9 9530* 35* 17**8 63* 4+4** 8832**35* 12* 1757*Residence Rural# 7 904341161468 Urban13**97**3542192269Mother’s Illiterate# 7 92 37 36 15 11 62education Lit, < mid sch com 11 93 37 45* 19** 16 71* Middle sch comp 13*** 89 63* 52* 28* 19 74* High sch comp+ 12*** 97 65* 52* 25* 39* 76*Mother’s age 15-19# 7 86 36 28 11 23 56 20-24 6 93***4538* 14***1567* 25-299 95**50***47* 20* 1673* 30-49 10932547* 21* 1171*Antenatal No# 7 853430149 56care Yes 9 94***4348* 23* 20**70*Religion Hindu# 9934842181670 Muslim3* 97 29* 32* 11* 1059* Christians50***845656*39*2175***Caste/tribe General# 8994142181967 OBC8 9416* 41161073* SC 9924644191569 ST 2 935131* 12**1260**Standard of Low# 6 95 43 39 14 17 69living index Medium 11**91414018*1466 High11964146*19**1374***Media No# 7 92 37 38 16 12 67exposure Yes 9 9446***43* 181969Mother’s No# 7 91 27 36 15 14 62awareness Yes 109452* 51* 22* 2074*Sex of Female# 15 97 43 40 19 19 65HH-head Male 7***934241171569Mother’s Low# 8 93 42 41 17 14 67empowerment Medium 5 96 26***40 13** 16 72**Index High 7925543152873*Electricity No# 8 8839 37151459 Yes 7 94**51**44* 20* 1672*#: Reference category; Significance level: ***10%, **5%, *1%.

To read the full text Login

Get instant access

New 3 Month Subscription
to Digital Archives at

₹826for India

$50for overseas users

Comments

(-) Hide

EPW looks forward to your comments. Please note that comments are moderated as per our comments policy. They may take some time to appear. A comment, if suitable, may be selected for publication in the Letters pages of EPW.

Back to Top