Equity and vaccine uptake: a cross-sectional study of measles vaccination in Lasbela District, Pakistan

In health planning, a pro-equity approach requires the removal of obstacles to accessing services. However, inequities in many countries are increasing, leaving an ever widening gap between the rich and poor, and even dividing the poor into further gradients of vulnerability [1, 2]. Expenditure on services is notoriously unbalanced, with the least vulnerable receiving the majority of investment [3]. This "poverty trap" means that the poorest and most vulnerable populations are less able to take up health care offers; this in turn worsens their socio-economic situation and health status [4, 5].

Although there is debate about the definition of equity, there is a general consensus that health inequity constitutes inequalities in health that are unfair or unjust [6, 7]. Rates of childhood vaccination are a good example. Vaccination is officially free in most countries, and in many developed countries its uptake is so close to universal as to have it considered "an indicator of how well children's rights are being respected" [8]. But vaccination coverage is lower in most developing countries, particularly in the poorest segments of these countries. Although vaccination is theoretically free, this does not account for costs of travel to the facilities and time away from work or the home. Poor access to facilities providing vaccination is a common reason for low uptake [9–11]. Other factors associated with reduced vaccination uptake include lack of maternal education [12], large family size [13], lack of household visits from health workers [14], and service provision issues such as a poor relationship between staff and clients and lack of trust that the vaccine is safe [15]. These disadvantages may be increased in vulnerable areas by, for example, water shortage, in comparison with which vaccination may not seem a pressing need [16]. Compounding reduced vaccination uptake, children from vulnerable households may have weaker immune systems and therefore be at increased risk of suffering severe consequences from measles [17]. In Bangladesh, for example, an unvaccinated child from a poor family in 2001 was more than twice as likely to die as an unvaccinated child from a family of higher economic status [18]. The costs of not vaccinating against measles have different implications for wealthier households for whom, with much less malnutrition and concomitant illness, childhood measles presents little more than an inconvenience [19].

Measles, a disease preventable by vaccination, primarily affects children in developing countries. According to the World Health Organisation (WHO) in 2001 there were over 30 million cases of measles and 777,000 deaths worldwide [20]. In Pakistan, estimates show that 20,000 children die from measles annually [21, 22]. This is despite the Pakistan Expanded Programme on Immunization (EPI) which provides BCG, DPT, polio vaccine, and measles vaccine during the first year of a child's life. According to the Pakistan Ministry of Health, the programme is aiming for "90% routine immunization coverage of all EPI antigens with at least 80% coverage in every district by 2012" [23]. There is evidence suggesting that measles vaccination coverage has increased only slightly or even stagnated in some provinces in the last few years [24–26]. According to the Pakistan Social and Living Standards Measurement Survey (2006-07), for example, measles vaccination in Balochistan province fell from 70% in 2006 to 54% in 2007 [27].

To examine uptake of measles vaccination, we conducted a household survey in Lasbela, an impoverished district in the south-east corner of Balochistan province of Pakistan, in 2005. Using data from this survey, we have examined the factors related to measles vaccination uptake and in particular the role of inequities in determining this uptake.

The survey covered 3366 households in total, and reached 2479 mothers who provided information about 4007 children between 10 and 59 months of age.

Inequity and measles vaccination uptake

Among children aged 12-23 months, slightly more than half (51%, 477/904) had received measles vaccine. Similarly, 51% (2103/3964) of the children aged 10-59 months had received the measles vaccine.

Tables 3 and 4 show the percentage of children aged 10-59 months vaccinated in urban and rural areas by different equity indicators. Notable is the fact that even among those in better socio-economic situations, in most cases only two-thirds of children are immunised. For example, in urban areas located within 5 km of a government vaccination facility, 68% (692/1023) of children aged 10-59 months are vaccinated. Similarly, among children in urban areas where the main breadwinner has a good job, 67% (433/643) are vaccinated. One exception to this is children from urban areas whose mothers' are educated (87% - 173/199).

	Percent (number vaccinated)	p value
Males	64% (410/636)	p = 0.724
Females	63% (374/589)
Live within 5 km of vaccination facility	68% (692/1023)	p < 0.001
Live further than 5 km of vaccination facility	44% (91/190)
Visited by vaccination team	66% (182/277)	p = 0.575
Never visited by vaccination team	63% (617/966)
Better roofs	66% (522/779)	p = 0.007
Poor roofs	59% (281/473)
Better job	67% (433/643)	p = 0.007
Poor job	59% (356/594)
Educated mother	87% (173/199)	p < 0.001
Non-educated mother	59% (636/1065)

	Percent (number vaccinated)	p value
Males	47% (672/1350)	p = 0.053
Females	43% (617/1340)
Live within 5 km of vaccination facility	66% (551/811)	p < 0.001
Live further than 5 km of vaccination facility	38% (668/1674)
Visited by vaccination team	51% (552/1064)	p < 0.001
Never visited by vaccination team	38% (521/1232)
Better roofs	57% (383/660)	p < 0.001
Poor roofs	41% (908/2035)
Better job	49% (706/1374)	p < 0.001
Poor job	41% (575/1304)
Educated mother	62% (59/92)	p = 0.002
Non-educated mother	44% (1234/2605)

Table 5 shows the variables included in the multivariate analysis, to further investigate the role of equity and other behavioural indicators in vaccination uptake of children aged 10-59 months. Table 6 shows the final model of the multivariate analysis for children living in urban areas. Having a mother with some education, who had discussed vaccinations, who knew of at least one vaccine-preventable illness, living in a community within 5 km of a government vaccination facility, being visited by a LHW who talked about vaccinations, and living in a house with a good roof were associated with vaccination uptake.

Outcome:	Measles vaccination
Covariants:	Roof type
	Occupation
	Crowding
	Education of mother
	Rural/urban setting
	Sex of the child
	Mother could correctly identify an illness preventable by
	vaccination
	Mother discussed vaccination within the family
	Mother involved in decisions about child's vaccinations
	Mother visited by LHW and told about vaccinations
	Government vaccination facility within 5 km of area
	Community visited by a mobile vaccination team

Variable	Unadjusted OR	Adjusted OR	95%CI for adjusted OR	Cluster adjusted 95% CI for OR
Education of mother	4.46	3.59	2.33-5.55	2.14-6.03
Discussed vaccinations	3.80	3.13	2.04-4.82	1.47-6.69
Knowledge about vaccinations	3.71	2.57	1.74-3.82	1.67-3.97
Govt vaccination facility <5 km	2.45	1.95	1.37-2.78	1.63-2.34
Good roof	1.41	1.47	1.12-1.93	1.08-2.01
Access to LHW	1.79	1.76	1.22-2.53	0.90-3.43

Table 7 shows the final model of factors for children living in rural areas. The model included many of the same variables as in urban areas, with access to vaccination and roof of dwelling having a combined effect, where the individual effects were statistically insignificant.

Variable	Unadjusted OR	Adjusted OR	95%CI for adjusted OR	Cluster adjusted 95% CI for OR
Discussed vaccinations	6.42	4.17	3.10-5.61	1.19-14.64
Knowledge about vaccinations	3.50	2.22	1.78-2.78	1.36-3.63
Government vaccination <5 km and good roof	3.19	3.7	2.41-5.19	1.85-5.64

Figure 1 illustrates the compounding effects of inequities in urban areas, showing the two most prominent equity factors that resulted from the urban multivariate analysis model - education and access. There is a significant trend for increased vaccination as inequities are removed (Chi square for linear trend 72.510, p < 0.000). Among children living in households more than 5 km from a government facility providing vaccinations (poor vaccination access) and whose mother had no education, just 41% (75/173) had received measles vaccine. Among children with poor vaccination access and with mothers with some education, 64% (541/850) had received measles vaccine. Among children with the advantages of both better access to a vaccination facility and a mother with some education, 87% (151/173) had received measles vaccine. There were too few children with poor vaccination access whose mothers had any education to include this category in the figure.

Figure 2 illustrates the compounding effects of inequities in rural areas, showing two prominent equity factors that resulted from the rural multivariate analysis model - access and roof type. Again, there is a significant trend for increased vaccination as inequities are removed (Chi square for linear trend 204.001, p < 0.000). Among children with poor vaccination access, those living in houses with a better roof were more likely to be vaccinated than those in houses with a poor roof (57% - 207/354 vs. 33% - 461/1320). Children with better vaccination access had higher rates of vaccination, and in this group there was little difference between those with a poor roof (66% (422/611) vaccinated) and those with a better roof (64% (129/200) vaccinated).

Measles coverage in Lasbela is unacceptably low, with only one half of the children aged 12-23 months old vaccinated. This is despite most mothers knowing of a vaccine preventable illness, considering vaccination worthwhile, and discussing vaccinations within the family. We found a strong association between discussion of vaccination and vaccination status in both urban and rural areas. Yet discussion is clearly not enough; vaccination rates in Lasbela are low even among the high proportion of mothers who had discussed vaccination in the family.

Among the limitations of this study is the reliance on vaccination as reported by the mother, as an indicator of vaccination uptake. Some developed country authors have suggested that maternal recall is not a good enough indicator of vaccination status compared with health facility records [20, 21]. However, a study from Italy found that parental recall alone was similar to other measures of vaccination status and concluded that "verbal recall should be accepted as reasonably reliable in the absence of cards" [22], while in Australia parental recall of measles vaccination coincided as well as vaccination cards with the presence of antibodies [23]. In Turkey, taking polio antibodies as the 'gold standard', a study found that parental recall was more sensitive but less specific than official records [24]. A study in India found that maternal recall underestimated children's vaccination status but using vaccination cards was not helpful because less than half the mothers had cards and the cards were often incomplete or grossly inaccurate [25]. Our own experience in Pakistan is that vaccination cards are frequently missing or highly inaccurate. Valadez et al in Costa Rica concluded that maternal recall could be used for estimating vaccination status, especially for younger children and for single dose vaccines [26]. Langsten and Hill in rural Egypt found mothers' reports were later confirmed by card data for at least 83% of children aged 12-23 months [27]. Gareaballah and Loevinsohn found that mothers' reports in the Sudan were accurate and concluded that for both DPT and measles vaccination, reliance on mothers' reports alone gave accurate estimates of vaccination coverage [28]. Goldman and Pebley in Guatemala highlighted the serious problems with service-based data (including vaccination cards) and recommended using mothers' reports to improve estimates of vaccination coverage [29]. Importantly, authors have reported that even if maternal recall may under- or over-estimate vaccination status, this was not related to factors such as maternal education level or poverty status [25, 30]. We therefore believe that our reliance on maternal recall of vaccination status is reasonable and is not likely to have introduced bias into the analysis of factors related to vaccination in the Lasbela district.

Our findings illustrate the role of equity in determining vaccination uptake in Lasbela. In both urban and rural areas, access to a government facility providing vaccinations, a key equity factor, was a determining factor for uptake. This is consistent with other reports that identify poor access to vaccination services as an obstacle to uptake [9, 10]. This was also confirmed during community feedback focus groups where many participants claimed their main obstacle to vaccinating their children was access to the facilities. In Lasbela, the proportion of children in rural areas with access to government facilities providing vaccination services is much lower than in urban areas and it is not universal even in urban areas.

In Lasbela we did not find visits by mobile vaccination teams associated with increased vaccination in either urban or rural areas. This is in contrast with the findings in other districts of Pakistan, where visits by vaccination teams were associated with increased measles vaccination, particularly in rural areas [35]. When asked how vaccinations might be increased, nearly all of the focus groups (29/32 male and 25/32 female) suggested door-to-door visits by the vaccination teams. It is likely that in Lasbela, where the terrain is mountainous and difficult to traverse, even when vaccination teams supposedly visit the community they are not reaching the most remote households in the communities. There could also be issues around service delivery (such as service provider attitudes, unofficial payments) that restrict the effectiveness of these initiatives.

In urban areas of Lasbela, where access to vaccination services is better than in rural areas, maternal education also played an important role in determining vaccine uptake, consistent with the findings of other authors [12, 35]. However, we did not find that mother's education was related to vaccine uptake in rural areas, although the small number of mothers in these areas with formal education might explain this. We did find that indicators of better socio-economic status, such as good roof type and having a better job were important determinants of vaccine uptake. The focus groups confirmed the importance of costs (such as travel and time away from work) as obstacles to vaccination for poor families. These costs get higher as the distance to the facility increases, compounding inequity for those who are poorest.

Figures 1 and 2 illustrate the importance of different aspects of equity in determining vaccine uptake in urban and rural areas. In urban areas (Figure 1) better access to vaccination and maternal education seem to increase vaccination uptake by about the same amount, and maternal education compounds with access when it is present. This illustrates how, in urban areas with good access, inequities between households still exist, in this case in terms of maternal education. This is consistent with authors who believe that inequities are not limited only to rural and marginal areas, and can exist within all socioeconomic groups [2, 36]. It also demonstrates that in urban areas at least, access alone is not the only equity factor involved in uptake of vaccination.

In rural areas, however, (Figure 2) there is a clear advantage for those living in areas with better access. The advantage for better off households (those with a better roof) seems to be confined to areas with poor access to services. It is probable that in areas with poor access the costs of taking a child for vaccination are much higher and therefore the disadvantage of poor households is more apparent. This compounding of inequity results in the very low vaccination rate of only 33% among poor rural households with poor access to services. This supports existing research that shows, while inequity is not limited to the most marginalised communities, the equity gap is increasing between the rich and the poor and that the poorest continue to receive the poorest service [1, 37, 38]. Overall rates of vaccination are lower in rural areas than in urban areas, and even lower still in the most marginalised rural areas.

Measles vaccination coverage is stagnating, or even decreasing, in some parts of Pakistan and this could be related to increasing inequities. Even when overall vaccination coverage in a country is increasing, this may mask considerable and even increasing inequities in coverage, particularly among the most vulnerable households [39]. Measures of vaccination coverage should include an assessment of inequities. The importance of different measures of inequity will vary from place to place, and even within different regions of one district. Understanding the particular dynamics of inequity and how it interacts with other factors related to vaccine uptake is a step towards increasing equity in vaccine coverage.

Inequities of access, maternal education and household socio-economic status are important determinants of childhood measles vaccination uptake in a poor district of Pakistan with limited provision of vaccination services. These inequities compound one another, so children from families with multiple disadvantages are very unlikely to be vaccinated, marginalising them even further with higher risk of poor health.

A hopeful finding is that discussion about vaccines and knowledge about vaccines had a positive effect that was independent of the negative effect of inequity - in both urban and rural areas. At least as a short-term strategy, there seems to be reason to expect an intervention increasing knowledge and discussion about vaccination in this district might increase uptake.