Determinants of child immunization status
Table
2 presents the marginal effects of the logit regressions. The findings show the presence of significant rural-urban differences in the probability of a child being fully immunized in both survey rounds. Specifically, the results show that compared to children in urban areas, children residing in rural households are about 8% and 4% more likely to have received the basic required immunizations in 2008 and 2014 respectively. The finding departs from earlier studies in developing countries that report statistically significant rural disadvantage in child immunization coverage [
1,
5,
13]. On the other hand, studies such as Awasthi et al. [
7], Babirye et al. [
8], and Egondi et al. [
19] report that in spite of the physical access to health facilities and immunization centres in urban areas, there exist substantial barriers to immunization coverage in urban settings, with large underserved populations in slum and informal settings.
Table 2
Determinants of Full Childhood Immunizations, 2008–2014
Rural | 0.0770** | | | 0.0448** | | |
(0.0306) | | | (0.0191) | | |
23–35 months | −0.0137 | 0.0418 | −0.0388 | − 0.0207 | − 0.0469 | 0.0039 |
(0.0249) | (0.0414) | (0.0312) | (0.0177) | (0.0286) | (0.0224) |
36–47 months | −0.1208*** | − 0.0644 | − 0.1489*** | −0.0667*** | − 0.0848*** | −0.0516** |
(0.0261) | (0.0454) | (0.0323) | (0.0186) | (0.0303) | (0.0236) |
48–59 months | − 0.1392*** | − 0.1195*** | − 0.1513*** | − 0.1170*** | − 0.1306*** | − 0.1065*** |
(0.0257) | (0.0457) | (0.0311) | (0.0191) | (0.0309) | (0.0244) |
Female | −0.0081 | −0.0060 | − 0.0076 | 0.0049 | 0.0093 | 0.0037 |
(0.0187) | (0.0315) | (0.0232) | (0.0132) | (0.0214) | (0.0168) |
Birth order | −0.0056 | − 0.0060 | − 0.0028 | − 0.0095* | −0.0067 | − 0.0101 |
(0.0072) | (0.0159) | (0.0082) | (0.0054) | (0.0094) | (0.0067) |
Delivered at health facility | −0.0024 | 0.0423 | −0.0136 | 0.0356** | 0.0765** | 0.0241 |
(0.0225) | (0.0448) | (0.0263) | (0.0169) | (0.0379) | (0.0187) |
Mother’s age at birth | 0.0048** | 0.0073* | 0.0029 | 0.0056*** | 0.0070*** | 0.0045** |
(0.0022) | (0.0042) | (0.0027) | (0.0016) | (0.0026) | (0.0020) |
Years of schooling | 0.0038*** | 0.0041 | 0.0033** | 0.0054*** | 0.0042 | 0.0056** |
(0.0015) | (0.0029) | (0.0016) | (0.0020) | (0.0030) | (0.0027) |
Number of co-resident women | −0.0181 | −0.0136 | −0.0179 | −0.0179* | −0.0344* | −0.0063 |
(0.0131) | (0.0203) | (0.0169) | (0.0101) | (0.0179) | (0.0122) |
Married/Consensual union | 0.0254 | 0.0318 | 0.0181 | 0.0716*** | 0.0809** | 0.0568** |
(0.0328) | (0.0534) | (0.0429) | (0.0215) | (0.0333) | (0.0282) |
Christian | 0.0741** | 0.0833 | 0.0790** | 0.0320 | 0.0266 | 0.0284 |
(0.0315) | (0.0952) | (0.0351) | (0.0262) | (0.0567) | (0.0293) |
Moslem | 0.0806** | 0.0610 | 0.0962** | 0.0807*** | 0.0198 | 0.1044*** |
(0.0349) | (0.1010) | (0.0402) | (0.0284) | (0.0605) | (0.0328) |
Media exposure | 0.0162 | 0.0265 | 0.0152 | 0.0303 | 0.0327 | 0.0274 |
(0.0268) | (0.0663) | (0.0302) | (0.0214) | (0.0506) | (0.0232) |
Employed – family/others | 0.1169*** | 0.0449 | 0.1439*** | −0.0097 | −0.0338 | 0.0116 |
(0.0386) | (0.0623) | (0.0502) | (0.0230) | (0.0356) | (0.0306) |
Employed - self | 0.1068*** | 0.0587 | 0.1228*** | 0.0306 | 0.0122 | 0.0493* |
(0.0349) | (0.0513) | (0.0469) | (0.0197) | (0.0299) | (0.0264) |
Has valid health insurance | 0.0585*** | 0.0955*** | 0.0449* | 0.0458*** | 0.0617** | 0.0343* |
(0.0210) | (0.0359) | (0.0265) | (0.0148) | (0.0249) | (0.0185) |
Distance to facility is a problem | −0.0174 | −0.0801* | −0.0037 | −0.0213 | −0.0289 | − 0.0160 |
(0.0212) | (0.0474) | (0.0242) | (0.0156) | (0.0290) | (0.0183) |
Middle | 0.0612** | 0.0549 | 0.0600* | −0.0337 | 0.0003 | −0.0410 |
(0.0301) | (0.0662) | (0.0354) | (0.0208) | (0.0362) | (0.0267) |
High | 0.0926*** | 0.0638 | 0.1055** | −0.0878*** | −0.0433 | − 0.1313*** |
(0.0328) | (0.0668) | (0.0424) | (0.0263) | (0.0377) | (0.0444) |
Regional dummies | Yes | Yes | Yes | Yes | Yes | Yes |
Observations | 2147 | 709 | 1438 | 4386 | 1743 | 2643 |
Demographic and health systems development may have contributed to the urban immunization disadvantage in Ghana. Over the last two decades, Ghana has witnessed a rapid pace of urbanisation, with substantial growth in slum and informal settlements. Thus, the growing number of underserved children in slum and informal settlements in urban areas in Ghana have contributed to the observed urban disadvantage in the coverage of child immunizations. Given the existence of large spatial disparities in access and utilization of health services and facilities, the child immunization strategy in Ghana appears to have place emphasis on rural areas, much to the neglect of underserved populations in urban areas. In line with this, there has been much emphasis on expanding basic health services in rural areas through the Community-based Health Planning and Services (CHPS) program. The urban-based CHPS has turned out to be much more challenging to implement as the population is more volatile and less well demarcated [
4].
In terms of child-specific characteristics, there are no gender disparities in the immunization status of children in Ghana. Indeed, evidence of gender differences in child immunization status have largely been reported by studies from southern Asian countries where son preference are endemic [
16,
37]. However, there exist statistically significant age differentials in the immunization status of children. Compared to children 12–23 months old, children between 36 and 47 months and 48–59 months olds are less likely to receive fully the basic immunizations. Age differentials in childhood immunization coverage have been reported in earlier studies. Such age disparities reflect the increasing trend in coverage with recent cohorts of children being fully immunized.
The results indicate a negative effect of the birth order of a child on the probability of a child receiving immunization in 2014. Adedokun et al. [
1] and Antai [
5] report that children of higher birth order are less likely to complete the required immunizations in Nigeria, and Corsi et al. [
16] report same for India. The results suggest “
immunization fatigue” of mothers as their interest to immunize their children as the number of children increases wanes [
3]. Antai [
5] on the other hand posits that the reduced likelihood of full immunization coverage of children of higher birth order reflects inter-sibling competition for parental care and limited household resources, leading to neglect.
There may be alternative explanations to the negative effect of child’s birth order on the probability of complete basic immunizations. First, the result may reflect the increased opportunity cost of time. The indirect wage of mothers in home production increases as the number of children increases. As such, as the high opportunity cost of time spent at a health centre to vaccinate a child of higher birth order reduces the probability of the child being fully immunized. Secondly, younger mothers may be better educated and informed on family planning and child health. The improved education thus reflects in fewer number of children and increased probability of complete immunization. Lastly, the results may be a reflection of a cohort effect – children with siblings have older mothers and the probability of taking children to vaccination decreases over time. The effect may be higher among old mothers than new mothers.
The place of delivery of a child shows a significant positive relationship with the probability of full immunization in 2014. The likelihood of full immunization is about 4% higher for children delivered in health facilities compared to children delivered at home. The importance of place of delivery to child immunization status is consistent to earlier studies such as Adedokun et al. [
3] and Bugvi et al. [
13]. Delivery at a health facility enables a child to receive the immunizations required at birth and provides an opportunity for the mother to be informed of the immediate schedules. The emergence of a significant relationship between the place of delivery of the child and probability of full immunization reflects the effects of improved access and utilisation of health services on child health. The reduction of the financial barriers to the demand for maternal and child health through the fee exemption policy under the National Health Insurance Scheme has contributed to achieving equity in child vaccinations in Ghana. Disaggregated by the place of residence, we find that the effect of the place of delivery is significant for urban areas only. This finding suggests the relative success of outreach programmes in rural areas where home vaccinations have been promoted through CHPS.
Religion is an important determinant of health seeking behaviours as well as health outcomes [
28,
36]. The results indicate the presence of significant religious effects on the probability of full immunization status in 2008. Compared to mother with no or other religious backgrounds, children with Christian or Moslem mothers are more likely to be fully immunized. The effects of religion, however, appear to have been attenuated in 2014. The significant difference between the immunization of children born to Christian mothers and mothers with no and other religious affiliations disappears, whilst, Moslem mothers are more likely to complete the required immunization routine for their children.
The significant relationships between maternal characteristics and the probability of full childhood immunization indicate the importance of maternal characteristics to child immunization status. Similar to Adedokun et al. [
3] and Antai [
5], the results indicate a positive relationship between the age at birth of the mother and the probability of the child being fully immunized in 2008 and 2014. Given pervasive cultural disapproval of early childbirth, the results may reflect the attitude of health workers and community members towards teenage mothers leading such mothers to drop-out of the routine child immunization [
5]. The effects of mother’s age at childbirth, thus, sheds light on the possible barriers that cultural values and belief systems impose on child immunization attitudes in Ghana. In addition, younger mothers have been found to have limited bargaining power in intra-household decision making, as young mothers may be required to seek the permission of other household members regarding decision concerning the health of the child [
30,
39]. Thus, delays in household decision making may have negative consequences for child health outcomes.
Improving female and maternal education has been promoted globally as a mechanism to enhance child health outcomes, especially in developing countries. Indeed, there is a growing body of empirical literature that show significant effects of increased maternal education on improved child health outcomes [
25,
32]. The results from the logit estimations are consistent with the findings of these previous studies such as Abadura et al. [
1], Adedokun et al. [
3] and Ataguba, Ojo and Ichoku [
6]. An additional year of completed schooling of a mother exerts a positive effect of the probability of a child receiving the basic immunizations. Educational attainment of the mother enhances the access and reception of information as well as facilitates communication between the mother and health workers, leading to better understanding of immunization schedules and practices [
28].
The labour market status of the mother is included as a proxy for the mothers’ time use as well as the opportunity cost of waiting at health and immunization centres. The variable may also capture the effect of the socioeconomic status of a woman on the immunization status of the child. Uthman et al. [
43] finds that the probability that a child receives the full basic required immunizations is lower for children with unemployed mothers. Bugvi et al. [
13] suggest that unemployed mothers and those in low paying occupations do not find the time and resources to travel to health centres for the immunization of their children, leading to incomplete immunization routines. On the other hand, employed mothers may face high opportunity cost of time foregone, thus causing them to forego routine visits to complete the basic immunization requirements. Our results in this paper are consistent with Uthman et al. [
43]. The probability of full immunization coverage is higher for children with employed mothers in 2008. However, there exist no significant effect of maternal employment on immunization status of children in 2014 except for children in rural areas with self-employed mothers. The changes observed in the labour market status of mothers between 2008 and 2014 appear to have shifted the effects on the probability of full vaccination of children. An increase in unemployment increases the opportunity cost of foregone wages. As such, employed mothers faced higher opportunity cost for waiting time at vaccination centres in 2014. Other things equal, this would reduce the propensity of employed mothers to spend time at vaccination centres, thereby eroding the higher likelilhood of having their children fully immunized observed in 2008.
Enrolment in a health insurance scheme has been found to increase the utilisation of outpatient services and reduction of out-of-pocket expenditures [
21]. Brugiavini and Pace [
11] finds a positive effect of health insurance membership on maternal health in Ghana. Our results show that mothers with valid health insurance policy are more likely to complete the basic immunizations for their children. Thus, there exists a spill-over effect of mother’s attitude to their own health to the health outcomes of their children. Mothers with health insurance may generally seek care more often when ill or when their child is ill. Health workers may use this opportunity to catch up on missing immunizations, thus resulting in the higher coverage in this group. Equally, the positive relationship between health insurance policy and completed basic child immunization may reflect possible financial barriers to child immunization. Child immunizations are offered for free in Ghana. However, mothers without valid health insurance coverage may be deterred by the fear of financial payments for immunization services.
Previous studies have reported significant socioeconomic inequalities in the coverage of child immunization across developing countries. The socioeconomic status of the household is an indicator of standard of living and has been adopted as a proxy for opportunity cost of women’s time as well as financial access to healthcare. Ataguba, Ojo and Ichoku [
6], Olorunsaiye and Degge, [
36] and Singh and Parasuraman [
40] report a disadvantage against children residing in households with low socioeconomic status. The present results from logit regressions reveal changes in the socioeconomic inequalities in child immunization coverage in Ghana between 2008 and 2014. Whilst children in high socioeconomic households are more likely to receive the full basic required immunizations compared to children from low socioeconomic households.
The results reveal a reversal of the disadvantage faced by households with low socioeconomic status in 2014 as there exist a significant negative effect of high household socioeconomic status and the probability of full childhood immunization. The presence of inequalities in child immunization against high socioeconomic households departs from a majority of studies on socioeconomic inequalities in child health outcomes. The results, however, reflect changing economic conditions witnessed between 2008 and 2014. A slowdown of economic activity exerted pressure on the time of women as such women may be required to engage in market activities to smoothen household standards of living. The increased demand for mother’s time in market activities may have led to an increase in the opportunity cost of time for mothers potentially leading mothers to pursue income-generating activities instead of completing the immunization routines for their children. Equally, the expansion of services and coverage through the CHPS initiative have also contributed to better access for least wealthy populations especially in rural areas.
Decomposition of rural-urban inequalities in child immunization, 2008/2014
A summary of sources of rural-urban inequalities in child immunization coverage in Ghana in 2008 and 2014 obtained from the decomposition analysis are presented in Table
3. The rural subsample is the reference category for the decomposition of the gaps in immunization coverage. The findings reveal the existence of rural-urban disparities in the probability of a child receiving the basic immunizations. Unlike logit estimates, the decomposition analysis reveals that the direction of disparities in child immunization coverage changes between 2008 and 2014.
Table 3
Summary of Oaxaca-Blinder Decomposition Results
Rural | 0.6940*** | 0.7276*** |
(0.0121) | (0.0086) |
Urban | 0.7348*** | 0.7011*** |
(0.0165) | (0.0109) |
Difference | −0.0408** | 0.0265* |
(0.0205) | (0.0139) |
Endowment / Explained | −0.1335*** | −0.0196*** |
(0.0286) | (0.0154) |
Coefficient / Unexplained | 0.0926** | 0.0461** |
(0.0362) | (0.0204) |
Coefficient / Unexplained (%) | 41.0 | 70.2 |
Observations | 2147 | 4386 |
In 2008, the average probability of a child in a rural household to receive the full basic immunization is 0.6940 compared to an average probability of 0.735 for a child in an urban area. The rural-urban differential in the average probability of a children being fully immunized in 2008 is statistically significant at 5%. In terms of the source of the disparities in the probability of full immunization, the results further indicates that the endowment or explained effect contributes about 59% of the gap. Indeed, differences in the endowments or characteristics favours urban resident children. This suggests that on average, children residing in urban areas possess higher levels of endowments compared to their counterparts in rural areas. The coefficient or unexplained effect, however, favours children in rural areas. This result therefore, implies that the health system in Ghana, in terms of child immunization services reward rural households higher for the same level of characteristics compared to urban households. Given the disparities in access to health care services and facilities, the child immunization program in Ghana has adopted strategies that aim at reaching the most vulnerable households, especially in rural areas. This strategy, however, appear to neglect vulnerable children in fast growing slums and informal settlements in urban areas.
Table
4 shows the contribution of each group of characteristics to the rural-urban gap in child immunization coverage. The decomposition reveals that maternal and household characteristics are the significant contributors to the endowment effect. The locational differences in both group of characteristics contribute to widening the gap in the probability of a child being fully immunized between rural and urban areas.
Table 4
Sources of contributions to rural-urban inequalities in child immunization coverage
Child | −0.0075 | −0.1042 | −0.0260*** | − 0.0254 |
(0.0139) | (0.1060) | (0.0089) | (0.0564) |
Maternal | −0.0379*** | −0.1687 | − 0.0398*** | −0.0276 |
(0.0117) | (0.2629) | (0.0101) | (0.1133) |
Household | −0.0701*** | 0.0274 | 0.0602*** | −0.0295 |
(0.0262) | (0.0687) | (0.0231) | (0.0268) |
Region dummies | −0.0180 | − 0.0454 | −0.0140** | − 0.1415** |
(0.0141) | (0.1416) | (0.0070) | (0.0701) |
Constant | | 0.3835 | | 0.2701** |
| (0.2928) | | (0.1339) |
Observations | 2147 | 4386 |
In 2014 however, the average probability of a child in a rural area to be immunized was 0.728 compared to 0.701 for a child in an urban area. The difference in the average probability of fully immunization between rural and urban areas is statistically significant at 10%. However, the coefficient or unexplained component of the gap dominates in 2014, accounting for 70.2% of the estimated rural-urban differentials in the probability of a full immunization of a child. The dominance of the coefficient effect may be attributed to the expansion of primary healthcare in rural Ghana between the periods though the CHPS programme The CHPS programme aims at improving access to primary healthcare and family planning services through community participation and mobilization. The number of functional CHPS facilities increased from 409 in 2008 to 2948 in 2014 [
23]. The programme has achieved considerable success in rural areas whilst the implementation in urban areas has been challenging.
The endowment effects in 2014 arises from significant differences in child, maternal, household and regional characteristics. Child, maternal and regional differences favor children in urban households, thus, contributing to widening the gap in immunization coverage between rural and urban areas. Household characteristics, which includes the number of resident women of reproductive age and household socioeconomic status, contribute to the endowment effect in favour of rural-resident children. On the other hand, returns to region of residence – contribution to the unexplained or coefficient component – is positive in favour of children in urban areas. This implies that in each region, the health system rewards children in urban areas higher in terms of probability of full immunization than children in rural areas.