Determinants Of Immunization Uptake Health And Social Care Essay
The World Health Organization recommends that all children under one year receive one dose of Bacille Calmette-Guérin vaccine (BCG), three doses of Diphtheria-Pertussis-Tetanus vaccine (DPT), three doses of oral polio vaccine (OPV), three doses of hepatitis B vaccine (HebB3), three doses of pneumococcal conjugate vaccine (PCV), three doses of Hemophilus influenza type B vaccine and one dose of the measles vaccine (WHO & UNICEF, 2012). The third dose of DPT (DPT3) is considered the proxy indicator for assessment of immunization coverage. The global DPT3 immunization coverage rose from 74% in 2000 to 85% in 2010 (Brown, Burton, Gacic-Dobo, & Karimov, 2011). Despite this, disparities in vaccination coverage exist between developed and developing countries (Brown et al., 2011).
Immunization coverage in Kenya, as measured by the third dose of DPT is 88% (WHO & UNICEF, 2012). This national coverage tends to hide regional inequalities in access to immunization. In 2009, Central province recorded the highest immunization coverage of 85.8% while the lowest was North Eastern province with 48.3% (KNBS & ICF Macro, 2010). These disparate proportions are an indication that some regions in Kenya enjoy access to immunization services more than others. There is need therefore to understand why some regions immunize children more than others and factors associated with immunization uptake. Literature review will focus on research findings on the effects of maternal, paternal, child and household factors on immunization, followed by analysis of methodological issues arising from these studies which will be relevant to the research proposal.
Search strategy
Information on factors affecting immunization uptake was obtained from Medline, Summon, Google scholar databases and relevant peer reviewed journals. Additional information was obtained from reference lists from articles published between 2000 and 2012 in which one or several factors had been reported to affect immunization uptake in children. Key words used in the search are indicated below, with their corresponding success rates after narrowing the search criteria:
Immunization 208,550
Determinants/factors 33,878
Children aged 12-59 months 2757
2000-2012 382
After screening the titles and abstracts of the final articles for relevance to the study problem, those that qualified were included in the literature review.
Immunization determinants
Research on factors affecting immunization uptake has been undertaken in both developed and developing countries. In developed countries, low socioeconomic status (Babatsikou, Vorou, Galani, Ktenas, & Koutis, 2010), low maternal education (Samad et al., 2006) and higher birth order (Haynes & Stone, 2004) have been associated with low immunization uptake. In developing countries, education level of both parents (Sullivan et al., 2010), maternal age (Kamau & Esamai, 2001), distance to the nearby health facility (Phukan, Barman, & Mahanta, 2009) and maternal knowledge affect immunization (Phimmasane, Douangmala, Koffi, Reinharz, & Buisson, 2010).
Maternal factors
Studies have shown that maternal education or literacy is a strong and consistent predictor of child immunization outcome (Kamau & Esamai, 2001; Kumar, Aggarwal, & Gomber, 2010). In both the US and Kenya, studies have shown that higher maternal education is associated with higher immunization uptake (Luman, McCauley, Shefer and Chu, 2003; Abuya et al., 2011). Despite most studies associating maternal education with immunization uptake, this relationship is not so clear. Some researchers have demonstrated a U-shaped association where children born to mothers with no formal education or higher education levels being likely to be immunized than those falling in between the two groups (Streatfield, Singarimbun, & Diamond, 1990). Others studies have shown that the association between maternal education level and immunization uptake disappears when other variables are considered, including the father’s education level and socioeconomic status (Steele, Diamond, & Amin, 1996).
Maternal age has been associated with immunization uptake. In Kenya, Mutua, Kimani-Murage, and Ettarh (2011) found that children belonging to older mothers were more likely to be vaccinated than those whose mothers were less than 20 years. Similarly, Fatiregun and Okoro (2012) were able to establish an association between maternal age and immunization completion. In this study however, young mothers were more likely to immunize their children than older mothers, perhaps due to the fact that young mothers have fewer children and are motivated to care for them. A similar study carried out in the Philippines failed to identify maternal age as a significant predictor of full immunization status in children though there was a trend approaching significance (p=.053), (Bondy, Thind, Koval, & Speechley, 2009).
Other factors linked to child immunization include maternal attendance of antenatal classes and place of delivery. In their study in Ambo Woreda, Ethiopia, Etana and Deressa (2012) were able to show that children belonging to mothers who had fully attended antenatal care were likely to be fully immunized. In the Philippines, children whose mothers attended at least four antenatal visits were one and a half times more likely to be vaccinated than those whose mothers attended less visits or none (Bondy et al., 2009). Closely linked to antenatal care is the choice of the place of delivery. Studies have been able to demonstrate a strong correlation between place of delivery and the ability of the child to complete immunization. An example is the study by Nath et al. (2007) in which children born at home were less likely to be vaccinated than those born in a health facility. Similar findings were reported in a study by Antai (2009) in Nigeria. Children born in a health facility receive the first vaccine (BCG) before they leave for home and mothers are reminded to bring their children for follow up vaccines.
Paternal factors
Most studies on determinants of immunization tend to focus more on maternal than paternal variables. This is perhaps due to the fact that it is mostly mothers who take children to immunization clinics. A study in rural Ethiopia by Sullivan et al (2010) was able to identify the effect of paternal age on immunization. Children belonging to older fathers were more likely to be immunized than those whose fathers were young. Fathers with a higher education level are also likely to have their children immunized (Phimmasane et al., 2010). Higher literacy level of the father (and mother) was also found to be associated with full immunization of the child in studies by Chhabra et al. (2007) and Bondy et al. (2009).
Characteristics of the child to be immunized
The child’s gender and birth order are important characteristics that have been included in a number of studies. In the 2009 Kenya Demographic and Health Indicator Survey (KDHS), there was no significant difference by sex in vaccination status of the child. Birth order was however an important factor with first born children being more likely to be immunized than those of the sixth or higher birth order (KNBS & ICF Macro, 2010). Studies conducted elsewhere in Kenya however failed to show any significant differences in immunization uptake based on gender of the children being immunized (Mutua et al., 2011; Owino, Irimu, Olenja, & Meme, 2009). It is possible that both male and female children have an equal chance of accessing health services due to gender mainstreaming campaigns by both the Kenyan government and non-governmental organizations. In rural Bangladesh, Rahman and Obaida-Nasrin (2010) discovered that male children were more likely to be immunized. This could be explained by cultural differences between the two countries.
A child’s birth order has been shown to be associated with various health outcomes including growth and development, accidents, morbidity and mortality due to some diseases (Elliott, 1992). Two studies in Kenya and the Philippines have been able to associate birth order with immunization uptake (Owino et al., 2009; Bondy et al., 2009). First born children are more likely to be immunized due to the excitement associated with the first child. As mothers get more children, resources might get constrained with parental excitement waning (Brenner, Simons-Morton, Bhaskar, Das, & Clemens, 2001). A study in northern India however showed no association between birth order and immunization uptake (Kumar et al., 2010). This study had two limitations; the sample size was small (n=325) and participants were selected from children admitted into one tertiary level hospital. The hospital attends to children predominantly from surrounding slum areas and so use of admitted children as participants may limit generalisation of research findings.
Household factors
Some of the household factors that determine child immunization include socioeconomic status, number of children in the household and distance between the household and the nearby health facility. Family socioeconomic status is associated with child immunization uptake with children from higher socioeconomic status households being likely to be immunized than those from low socioeconomic status (Hu, Chen, Li, Chen, & Qi, 2011; Kusuma, Kumari, Pandav, & Gupta, 2010; Topuzoglu et al., 2005). This was indeed noted by the 2009 KDHS in which low immunization rates were noted in children from households in the lowest wealth quintile. A study conducted earlier on in Mathare Valley, Kenya by Kamau and Esamai (2001) had failed to show any association between socioeconomic status and child immunization. This study, however, had a small sample size (n=360) compared to the KDHS (n=1096) and researchers did not clarify how they analysed association between socioeconomic status and immunization uptake.
Distance from the household to the nearby health facility is an important indicator because it affects access to health services. In coastal Kenya, Ndiritu et al. (2006) were able to accurately measure the distance between each household and the nearby vaccination clinic. Using Geographical Information System (GIS) software ArcGIS v9.0, they reported that distance to the nearest clinic affects immunization, with those closer to clinics more likely to take children for immunization. In Lasbella district, Pakistan, staying closer to a health facility (less than 5 kilometres) was associated with more immunization (Mitchell et al., 2009). These findings highlight the importance of ensuring that health facilities are adequate and accessible to all households.
Methodological issues arising from these studies
Several studies have been conducted in various parts of the world to understand the effect of sociodemographic factors on immunization uptake. These have resulted in varied findings, perhaps due to methodological issues including research design, participants, measurement tools and data analysis. Each of these is discussed below.
Majority of the studies reviewed made use of cross sectional study design. One study used a longitudinal design with baseline immunization data being collected at four months after birth, there after every four months till the child’s age reached 24 months (Mutua et al., 2011). Another study used mixed methods design where cross sectional study was combined with qualitative research methods (Owino et al., 2009).
Participants in most of the studies were children aged below five years. It is expected that most children complete immunization between 12 and 60 months. This wide range is due to variations in immunization schedules in different countries. While looking at the prevalence and determinants of full immunization in Turkey, Babatsikou et al. (2010) recruited children aged between 0 to 12 years old. This is likely to affect the quality of research findings because children below one year are still receiving immunization and might be misclassified as unvaccinated. While most studies described how they made use of the WHO (2005) cluster sampling technique for immunization and explained the rationale behind its use, one study only mentioned it without details of the sampling procedure (Kamau & Esamai, 2001).
Two measures used to identify immunization history include records on vaccination cards and maternal recall. Despite the possibility of recall bias, asking mothers questions on routes of administration and dosage schedule helps to improve accuracy of maternal recall. Studies have shown that maternal recall can be relied upon to assess immunization with a 98% parental accuracy (AbdelSalam & Sokal, 2004) and high correlation between maternal recall and vaccine card information, Spearman’s r=.71 (Valadez & Weld, 1992). It is necessary to involve mothers without immunization cards to avoid missing out on vital information.
Studies varied in their approach to measurement of socioeconomic status within households. Two studies used family income as a proxy indicator of socioeconomic status (Kamau & Esamai, 2001; Kusuma et al., 2010). This can introduce measurement bias especially in developing countries where majority of families do not have stable income. It is not possible to get accurate records of monthly income from such families. Sullivan et al. (2009) used a picture of a ladder with ten rungs to represent varying levels of socioeconomic status. Participants were asked to use this to rank their socioeconomic status relative to others in the community. This is a subjective measurement tool prone to bias and has not been tested for validity and reliability. A study by Chhabra et al. (2007) did not consider the effect of socioeconomic status on immunization uptake, while Hu et al. (2011) did not explain in their methods how they estimated socioeconomic development.
The process of determining socioeconomic status needs to be considered before analysis of the relationship between socioeconomic status and immunization uptake. A good measure of economic status would be one in which several indicators are combined before establishing of the final score. Topuzoglu et al. (2005) used the Socioeconomic Status Index for Turkey, a test that utilizes variables such as occupation, household assets and education level. Kusuma et al. (2010) used similar measures to rank socioeconomic status in India. This is more applicable in developing countries where use of fixed assets and house hold utilities would useful in estimating socioeconomic status. This is illustrated in the KDHS (2009) and a study in Kenya by Mutua et al. (2011) who used Principal Component Analysis to construct wealth indices for socioeconomic status, based on ownership fixed assets. Most studies did not explain how they measure distance between household and health centres except for Ndiritu et al (2006) who used Geographical Information System (GIS) software.
Despite most of the studies reporting association between sociodemographic variables and immunization uptake, some reported negative association or none at all. This variation could be due to study design and nature of data analysis. Some studies focused on the effect of one variable for example maternal age (Salmon et al., 2009) or education (Abuya et al., 2011) on immunization uptake. These two studies reported significant association between the stated variables. Consideration of other variables including paternal education and socioeconomic status led to contradicting results (Steele, 1996). This highlights the importance of considering all potential factors and confounders in such studies.
Choice of a longitudinal study design or larger sample size (more than 3000 participants) led to robust results and consistency in research findings (Mutua et al., 2011; Topuzoglu et al., 2005; Mitchel et al., 2009). Studies with contradictory findings had relatively small sample sizes with less than 600 participants (Fatiregun & Okoro, 2012; Kumar et al., 2010; Kamau & Esamai, 2001). Studies need to have larger sample sizes to minimize selection and information bias hence consistency in research findings. Use of multivariate data analysis methods led to disappearance of association between variables in studies by Chhabra et al. (2007), Bondy et al. (2009) and Rahman and Obadia (2010). In these studies, multivariate methods revealed significant predictors of immunization after ruling out interaction between individual variables, hence revealing true association. This informs researchers on the importance of analyzing association between variables in both bivariate and multivariate models.
Summary
A number of studies have been conducted on the effect of sociodemographic factors on immunization uptake. Majority of the studies reviewed seem to be in agreement that sociodemographic factors affect immunization uptake. Varied findings could be attributed to use of different sample sizes, measurement tools and data analysis methods. Use of larger sample sizes and validated measurement tools would help in ensuring consistency in findings of future research.
Sociodemographic factors are likely to have an interaction with each other for example maternal age, education level, paternal age and socioeconomic status. Use of multivariate methods, for instance, logistic regression, is important in ruling out interaction between variables thereby revealing significant predictors of immunization uptake. Informed by these findings, the proposed study will seek to improve quality of findings by recruiting a large sample size, use validated measurement tools and analyse factors on both bivariate and multivariate levels. The literature review also revealed that few studies have been conducted in Kenya on determinants of immunization uptake and none in Western Kenya. This highlights the need for more studies to be conducted in Kenya to add to already existing knowledge. A clear understanding of these factors will direct decision makers in understanding the entry levels at which to initiate community based interventions that can increase immunization access and utilization.
Aims and objectives
Research questions
What is the prevalence of complete immunization among children aged between 12 and 59 months in Kakamega central district, western Kenya?
What is the association between maternal, paternal, child and household sociodemographic characteristics and immunization uptake?
Objectives
To assess the prevalence of complete immunization among children aged between 12 and 59 months in Kakamega central district, western Kenya.
To explore the nature of association between maternal, paternal, child and household sociodemographic characteristics and immunization uptake.
Methodology
Study design
A cross sectional study will be utilized to collect information on immunization coverage and family sociodemographic variables. The variables of interest include maternal characteristics (age at delivery of the child of interest, education level, antenatal visits and place of delivery), paternal characteristics (age and education level), characteristics of the child to be immunized (birth order and gender) and household characteristics (socioeconomic status and distance between the household and the nearest health facility). Also included is Immunization status of the child, classified as either fully immunized or not.
Kakamega central district consists of 28 administrative units, called sub-locations. Each sub-location will constitute a stratum from which households will be drawn for the survey. The ministry of provincial administration maintains an updated list of all households per sub-location with names of household heads. Households to be surveyed per stratum will be selected randomly using a computer assisted generation of random numbers. The first household will be selected randomly. The next one will be the nearest household that has been listed on the computerized random ranking and meets the inclusion criteria. The process will continue until 19 households have been visited per stratum.
Participants
The target population in this survey will be children aged between 12 and 59 months. To be considered for the study, children must belong to the household and must have been present in the household for more than six months. Children from neighbouring households and those visiting will not be considered. In households with two or more children qualifying for inclusion, the youngest will be selected. In case of twins, only one child will be considered after tossing of a coin.
A sample of 541 children will be included in the study, based on the WHO (2005) recommended single proportion formula for immunization; with a 95% confidence level, 5% margin of error and 80% estimated immunization coverage rate in Kakamega central. A 5% non-response rate and a design effect of 2 will be considered. This method has been used in a number of immunization surveys (Etana & Deressa, 2012; Fatiregun & Okoro, 2012; KNBS & ICF Macro, 2010).
Measures
A structured questionnaire will be used to collect information on sociodemographic characteristics of interest. Some of the questions will be adopted from those used in the Kenya Demographic and Health Survey (KNBS & ICF Macro, 2010). The questionnaire will initially be developed in English then translated to Swahili. It will include sections on sociodemographic characteristics of the child and parents, antenatal visits, place of delivery, household assets and child’s immunization history. Socioeconomic status will be determined by use of asset indicators to develop the wealth index.
Immunization history for the child of interest will be obtained from the vaccination card. This includes types of vaccines, doses and timeliness of these vaccines. In the absence of vaccination cards, mothers will be asked to give verbal reports based on recall. Routes of vaccine administration and the dosage schedule will help in identifying vaccines based on maternal history. Research has shown that maternal recall is just as reliable as use of vaccination cards in collecting immunization data (Bondy et al., 2009). Information on sociodemographic profiles will be obtained verbally then corroborated with available documents including national identification cards, birth certificates and academic certificates.
Procedure
Ethical approval for the study will be sought and obtained from Auckland University of Technology Ethics Committee (AUTEC). Permission to proceed with the study will also be sought from the Kenyan Ministry of Health. Before beginning the study, five research assistants will undergo a five day training to be familiarized with the research process and data collection tools. On the last day, administrative issues and logistics will be addressed. Each research assistant will be allocated five sub-locations which include selected households. Prior to the research exercise, announcements will be made in local schools, churches and market places to ensure availability of participants during field days.
On arriving at each household, researchers will check whether the household fits the inclusion criteria. Mothers from qualifying households will then give both verbal and signed consent before proceeding with the study. Literate mothers will then be given the questionnaire and allowed to fill under guidance from the researchers. Those who do not understand English will be given the Swahili version. At the end of the exercise, the researcher will thank the respondent then move on to the next household as per the schedule. Village elders will assist researchers in accessing households since have local knowledge on location of various households within the study area.
Data analysis
Statistical analysis will be done using the software SPSS v19 for Windows, with the alpha value set at .05. Data will initially be checked for extreme outliers and normality in distribution through use of histograms and box plots. Mean, median and standard deviation will be used to describe continuous data while frequencies will be used for categorical data.
Purpose 1: Assess the prevalence of complete immunization among children aged between 12 and 59 months in Kakamega central district, western Kenya.
The total number of fully immunized children aged between 12 and 59 years will be divided by the total number of children interviewed. This will then be expressed as a percentage.
Purpose 2: Determine the nature of association between maternal, paternal, child and household sociodemographic characteristics and immunization uptake.
To investigate continuous variables, data will be split into two groups; fully immunized versus not fully immunized. In each subset, data on continuous variables will be correlated with each other using Pearson’s product moment correlation. Differences in correlation between the two groups will then be assessed through the use of independent sample t-test.
Chi square test will be used to investigate association between categorical variables. In cases where values for categorical variables are less than five per cell, Fisher’s exact test will be used to test for association. Odds ratios and their associated 95% confidence intervals will be calculated.
Based on results from Pearson’s correlation and Chi square test, all variables with significant association will be entered into logistic regression. This will assess independent contribution of each variable into the model hence significant contributors to estimating the probability that a child will be fully vaccinated.
Ethical and cultural considerations
Informed consent: Participants will be informed about the nature of the study and given a chance to withdraw from the study any time up to the start of data analysis. Both verbal and written consent will be obtained voluntarily from the participants.
Confidentiality: Information from participants, including raw data, will be locked up at AUT University in a locker for a maximum of 10 years. The principal investigator will be the sole custodian of participants’ information and will keep it confidential. Data files will be stored under a password protected folder.
Risk minimization: Before the study, participants will be informed about the nature of questions to be asked; some which may be sensitive for instance pregnancy related information. Participants who might become distressed by the study can then choose not to participate, thereby preventing emotional discomfort.
Cultural considerations: The people of western Kenya value their cultural beliefs and practices. Participants will be encouraged to point out any questions or behaviour that appears to be culturally inappropriate and suggest ways of addressing these. Partnership, participation and protection will guide the research process. This is explained below:
Partnership: Village elders will be consulted about the study since they represent the people.
Participation: Social mobilization will be done before the study in churches, schools and markets. Everyone will be encouraged to participate in the study.
Protection: The local administration (chiefs), together with researchers will ensure that participants are protected from physical or psychological harm.
Significance of the study
Immunization in children is important in preventing morbidity and mortality due to vaccine-preventable illnesses. Despite several studies being published on determinants of immunization uptake in children, fewer studies have been conducted in Kenya and none in Kakamega central district. Should this study identify significant sociodemographic factors associated with immunization uptake, policy makers will be in a position to identify families at risk of failing to immunize. This will help in ensuring that interventions are targeted to vulnerable groups. Such interventions might include outreach activities and community health education. If policy efforts were to focus on ensuring that all children receive full vaccination, the country stands a chance at attaining more than 90% immunization coverage. Control of infectious diseases through vaccination will lead to reduced mortality in children aged below five years. Subsequently, government expenditure on management of disease outbreaks will reduce with funds being redirected to scaling up of other healthcare service delivery programmes.
Timeline
2012
2013
2014
Aug
Sept
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Draft proposal and ethics
Submit proposal and ethics
Literature review
Prepare for data collection
Data collection
Data analysis
Draft report
Final report submission
Budget
Item
Unit Cost (NZD)
Total cost (NZD)
Return flights to Kenya for data collection
1500.00 per flight
3000.00
Cost of transport within Kenya for one month
20.00 per day
600.00
Accommodation costs in Kenya for one month
30.00 per day
900.00
Food
20.00 per day
600.00
Stationary: Printing and photocopying of the questionnaire (600 copies)
Notebooks and pens for 6 research assistants
10 cents
10.00
60.00
60.00
Token for participants. Each respondent will be given 1 pen for one child
50 cents
300.00
Lunch for 6 researchers for 5 days
20
600.00