Article Text
PDF
Abstract
Background This study assessed the extent of delays in childhood vaccinations and examined sociodemographic correlates of delayed and missing vaccinations.
Methods Datasets from the 2005–7 Multiple Indicator Cluster Surveys from 31 countries were used. Information on vaccinations was based on vaccination cards. Survival analysis was applied to assess age-specific vaccination rates, and multilevel logistic regression analysis was used to assess factors associated with delayed and missing vaccinations.
Results The median vaccination coverage across all countries varied from 91% measles-containing vaccine (MCV) to 98% bacille Calmette–Guérin vaccine (BCG). The median fraction of timely administered vaccinations was 65% (range 14.5–97.2%) for BCG, 67% (11.6–89.3%) for the first dose of vaccine against diphtheria, tetanus and pertussis (DTP1), 41% (10.8–82.1%) for DTP3, 68% (29.7–90.3%) for the first dose of polio vaccine (polio1), 38% (10.5–81.0%) for polio3 and 51% (22.3–91.1%) for MCV. The median of the median delays across all countries was 2.1 weeks (IQR 0.9–3.0) for BCG, 2.4 weeks (1.5–3.1) for DTP1; 6.3 weeks (3.3–9.0) for DTP3; 2.0 weeks (1.3–3.1) for polio1, 6.6 weeks (4.3–9.3) for polio3 and 4.1 weeks (2.5–5.8) for MCV. A higher number of children in households and lower socioeconomic status were associated with delayed and missing vaccinations; however, the effects of socioeconomic gradient varied by country.
Conclusion Most countries achieved high up-to-date vaccination coverage. However, there were substantial vaccination delays. Collecting information on the timeliness of vaccination in national surveillance systems will provide a more complete view of vaccination coverage. Missing and delayed vaccinations can be addressed jointly in prevention programmes.
- Age-appropriate vaccination
- children
- delayed vaccination
- low and middle-income countries
- vaccination
Statistics from Altmetric.com
Achieving high vaccination coverage is crucial in the control and prevention of childhood infections.1 Currently, standard estimates of vaccination coverage are based on vaccination status at a predefined age, for example, from 12 to 23 months of age,2 3 but this does not take into account possible delays in the administration of vaccines. If the delay in vaccination is substantial, it presents children with late administration a longer time at risk to acquire infection and also negatively affects herd immunity, which protects the population from the spread of the disease.4 A recent analysis of 45 low and middle-income countries using data from the Demographic and Health Surveys (DHS) demonstrated a substantial variation in the timeliness of vaccination, with considerable delays found in some countries.5 Some studies have recently proposed including age-appropriate vaccination as an additional indicator of the quality of immunisation services.6 7
Factors associated with a lack of vaccination were studied extensively in developed and developing countries and specific patterns were identified.8–12 In contrast, less is known about factors associated with vaccination delays and whether they follow the same patterns in different societies. Few studies exist for the US population: Bobo et al13 found that first-born children and those whose mothers have a higher level of education were less likely to be vaccinated with delays. In addition, Dombkowski et al14 demonstrated that factors associated with healthcare utilisation (eg, no insurance and no medical enrolment) were associated with a higher risk of delayed vaccinations. In two further large US studies comprising data of approximately 15 000 and 21 000 children, Luman and colleagues15 16 showed that parental factors were strongly associated with delayed vaccinations; for example, children from migrant families (eg, non-Hispanic blacks), children whose mothers had lower education and children from larger families were more likely to be vaccinated with delay. At the same time, provider characteristics (eg, having public vaccination provider) also played an important role in the timely administration of childhood vaccines. Danis et al17 compared both socioeconomic factors and parental attitudes towards vaccinations in Greece and found that the latter played a less important role in timely vaccination than the former. An analysis of four former Soviet republic countries in central Asia showed that delays in vaccination and lack of vaccination were associated with similar individual-level characteristics across these countries, but the largest was the country effect.4 The latter finding in particular was surprising because the medical care systems in these countries with a joint past as Soviet republics are based on the same tradition of organisation.18
We used data from the third round of the Multiple Indicator Cluster Surveys (MICS3). The MICS3 data are in most cases more recent than DHS data used in the latest cross-country comparison of vaccination delay5 and focus more on former Soviet and former Yugoslavian countries. The aims of our analysis are to assess patterns of factors associated with delayed and missing vaccinations and their variation across countries, and to provide data on vaccination delays for countries not covered in the previous analysis using DHS data5 (22 of the 31 countries covered in the current analysis were not included in the previous publication using DHS data).
Methods
MICS3 was conducted in 52 countries in the years 2005 and 2006, and in one country (Mauritania) in 2007. The surveys were undertaken to provide nationally representative data regarding women’s reproductive health and the health of children. Technical and financial support was provided by the United Nations Children’s Fund.19 Datasets from 38 countries were available for analysis.
MICS3 used a multistage cluster sampling design. First, clusters (usually census enumeration areas) were selected with the probability proportional to population size. Furthermore, in each cluster a sample of households was selected randomly, and in each household women of eligible age (between 15 and 49 years) were interviewed using a standardised questionnaire. Information on vaccinations was available for children born in the 5 years preceding the surveys. As the exact vaccination date was necessary for the calculation of age-specific vaccination, the dataset was restricted to children with available vaccination charts. In five countries (Georgia, Kyrgyzstan, Kazakhstan, Somalia, Tajikistan and Uzbekistan) less than 10% of vaccination charts were available. We excluded these countries from further analysis. In Kyrgyzstan and Uzbekistan the proportion of vaccination charts was less than 1%. We excluded one survey (Ukraine) because data on vaccinations were missing. One further country (Cuba) differed from the others in terms of the included population (only children up to the age of 3 years) and information about vaccination (only for 2 years) and was excluded from the analysis. All remaining datasets were checked for consistency and merged (n=202 944). The median sample size was 4204 (IQR 2866–6788, range 796–34 710).
For presentation of the results, we grouped countries into the six following groups: (1) African countries; (2) former Soviet countries; (3) former Yugoslavian countries; (4) central American countries; (5) south-east and east Asian countries, and (6) Middle East countries.
The following vaccinations were administered at the time of the surveys in all countries: bacille Calmette–Guérin (BCG) vaccine (except in Trinidad and Tobago), a vaccine against diphtheria, tetanus and pertussis (DTP), a vaccine against poliomyelitis (polio), and a measles-containing vaccine (MCV). Delayed vaccination was defined as vaccination after 1 month (4.3 weeks) of the age specified in the national immunisation schedules of each country. Up-to-date vaccination was defined as vaccination at any time (including delayed vaccination), and missing vaccination—if the child did not receive vaccination at all after the age specified in the national immunisation schedules. There was a substantial variation in the recommended age for vaccination according to national guidelines across countries (table 1).
The recommended age at vaccination for vaccine doses included in the analysis according to national immunisation schedules by country*
For the analysis solely information available in the children dataset was used: children's sex, place of residence, number of children in a family and the wealth index for the economic status of the family. The wealth index was developed by the World Bank and the DHS to allow comparisons across countries.20 It is constructed using several—including country-specific—indicators, comprising household assets (eg, having television, refrigerator, type of vehicle, etc) and utility facilities (eg, having electricity, type of floor, toilet and water supply, number of persons per sleeping room, etc). A score of the wealth index is created using principal component analysis and divided into quintiles: from the poorest 20% to the richest 20%. The scores were already available in the datasets.
Statistical analysis
We first calculated the up-to-date vaccination coverage in each country. For the BCG, DTP and polio vaccines, we used a subsample of children aged between 12 and 59 months, also to include children who received vaccination with a delay of up to 6 months. This resulted in a sample of 83 001 children. Similarly, for the MCV, which is usually administered between 9 and 13 months of age, we used a subsample of children of 18–59 months; the resulting sample size was 68 521 children. Subsequently, we used the Kaplan–Meier method to estimate age-specific vaccination coverage.21 For this analysis, we included all children aged between 0 and 59 months at the time of interview. We calculated age at vaccination in weeks using information on the date of birth and date of vaccinations. Finally, we calculated OR and 95% CI for the following factors potentially associated with missing and delayed vaccinations: children's sex; place of residence (urban vs rural); number of children in the family and the wealth index using multilevel binary logistic regression analysis. Country was treated as a random effect; this was supported by the two times higher variance of the intercept (country effect) than the corresponding SE in all models.22 In total, eight separate models were estimated, for each BCG, DTP3 (third dose of DTP vaccine), polio3 (third dose of polio vaccine) and MCV vaccines, one for missing and one for delayed vaccination compared with timely vaccination. We chose the third doses of DTP and polio vaccines because the final doses of vaccines are more likely to be administered with delays, and thus the effects can be best studied.21 For this analyses, we selected only one child per household (the youngest) as the vaccination status of siblings may correlate and therefore lead to biased estimates,23 and used a subsample of children aged 12–59 months for BCG, DTP3, and polio3 vaccines (n=61 010) and aged 18–59 months for MCV vaccine (n=52 128). In a separate analysis, we studied the variation in the effect of socioeconomic factors. We fitted logistic regression with the wealth index as independent, and vaccination status (missing or delayed versus timely) as dependent, variable in each country separately and displayed the country-specific effects graphically. The analysis was conducted with the statistical program SAS for Windows version 9.2. The command Proc GLIMMIX was used for multilevel analysis and Proc LIFETEST for Kaplan–Meier analysis.
Results
Availability of vaccination cards
There were substantial differences within even the same group of the countries regarding the availability of vaccination cards, for example, in African countries cards were only available for less than 30% of children in Djibouti and Mauritania, and more than 60% in Burkina Faso, Côte d'Ivoire, Guinea-Bissau and Malawi. In other groups of countries the figures were more homogeneous. In total, for approximately 58% of the sample vaccination cards were available and were seen by the interviewer. In the remaining group, for approximately 28% of children cards were only marked as available, but were not seen by study interviewers for some (unknown) reasons (and therefore data on the exact dates of vaccination were not available), and for 14% of the children vaccination cards were not available at all. In the further analysis, we used only vaccination data based on the vaccination cards containing the exact date of vaccination, and excluded surveys for which the proportion of vaccination cards was 10% or less (Georgia, Kyrgyzstan, Kazakhstan, Somalia, Tajikistan and Uzbekistan) (113 822 of 202 944, 58.0%).
Up-to-date vaccination coverage
In most countries, the up-to-date vaccination coverage was at high levels (table 2). The highest coverage was observed for BCG and DTP1 vaccines. The median vaccination coverage across all countries for BCG vaccine was 98.1% (range 55.9–100%), for DTP1, 97.0% (range 40.9–100%), for polio1, 96.8% (range 67.8–99.8%). Coverage was lower for DTP3, polio3 and MCV; the median for DTP3 was 91.4% (range 29.2–99.6%), for polio3, 91.6% (range 48.4–99.4%) and MCV, 89.7% (range 60.5–97.0%). Overall, the lowest coverage for DTP vaccines was observed in Ghana (<60% of children received at least two doses), Trinidad and Tobago (<40%) and Yemen (<70%). Vaccination coverage varied not only across countries from different groups, but also within groups. For example, in Middle East countries, the coverage was at high levels in Syria (over 90% for all vaccines), but was much lower in Iraq and Yemen (table 2).
Year of surveys, sample size and up-to-date vaccination coverage by countries and country groups, % (vaccination coverage based on vaccination cards for children aged between 12 and 59 months)
Age-appropriate vaccination and delay in vaccinations
Age-appropriate vaccinations varied substantially between countries and vaccines (table 3). The median of age-appropriate vaccination across all countries was 65% (range 14.5–97.2%) for BCG vaccine, 67% (range 11.6–89.3%) for DTP1, 68% (range 29.7–90.3%) for polio1, 41% (range 10.8–82.1%) for DTP3, 38% (range 10.5–81.0%) for polio3 and 51% (range 22.3–91.1%) for MCV. The median of the median delays across all countries for BCG was 2.1 weeks (IQR 0.9–3.0), 2.4 weeks (1.5–3.1) for DTP1, 2.0 weeks (1.3–3.1) for polio1, 6.3 weeks (3.3–9.0) for DTP3, 6.6 weeks (4.3–9.3) for polio3 and 4.1 weeks (2.5–5.8) for MCV.
Age-appropriate vaccination and delays after the specified ages by country*
In order to obtain a more complete picture of the vaccination delays, we displayed the fraction of children who received vaccination by age and by country for BCG, DTP1, DTP3, polio1, polio3 and MCV (figure 1). BCG was in general administered very timely, and high coverage at 3 years was reached. Few countries departed from this pattern: Serbia, Montenegro and Yemen had a substantially higher fraction of missing vaccinations. In Montenegro and Yemen, there was additionally a substantial amount of delay. In Lao People's Democratic Republic and Guinea-Bissau, there was a substantial amount of delay, but a low fraction of missing vaccinations at the end of the studied age range. On the other hand, there were few countries (Belarus, Mongolia and Thailand) in which the delay was exceptionally small. Missing and delayed vaccination showed similar patterns for DTP1 and DTP3, and for polio1 and polio3, with an increase in heterogeneity across the countries for the latter doses. Also there were countries (eg, Mongolia) where delay played a little role, but vaccination was missing in a substantial fraction. Also the initial fraction of vaccination was quite similar in most countries; what made a difference was the delay in vaccinating the remaining fraction of those ultimately vaccinated. This initial fraction of timely vaccinated children was substantially lower in most countries for DTP3 and polio3 than for DTP1 and polio1, creating the picture of more diversity across countries. Finally, for MCV there were two different vaccination ages recommended by immunisation schedules in the studied countries (table 2). In general, higher delays were the dominating pattern for this vaccine. The few exceptions were Bangladesh, Belarus, Thailand and Vietnam (not denoted in the figure, but can be identified in the left upper corner). Exceptionally low was the final vaccination coverage in Lao People's Democratic Republic. Of interest is the fact that DTP and MCV vaccinations in most countries are often administered at earlier ages than recommended by the immunisation schedules.
Age-specific coverage of bacille Calmette–Guérin (BCG) vaccine, the first and third doses of diphtheria, tetanus and pertussis vaccine (DTP1 and DTP3), the first and third doses of polio vaccine (polio1 and polio3) and measles-containing vaccine (MCV) by countries (based on Kaplan–Meier estimates; y-axis, proportion of vaccinated children at given age). The vertical lines represent most common observed age at vaccination recommended by national immunisation schedules in different countries. Country abbreviations: AL (Albania), BY (Belarus), GH (Ghana), GY (Guyana), LA (Lao People's Democratic Republic), ME (Montenegro), MK (Macedonia), SP (Serbia), TT (Trinidad and Tobago), YE (Yemen).
Factors associated with delayed and missing vaccinations
The results of the multilevel multiple logistic regression analyses of factors associated with delayed (columns 2–4) and missing (columns 5–7) vaccination as compared with the reference group of timely vaccinated children are presented in table 4. In general, we found a similar pattern of association for the analysed vaccines (BCG, DTP3, polio3 (data for polio3 not shown) and MCV). The risk of delayed vaccination was slightly higher among boys compared with girls (significant for DTP3 vaccine). A higher number of children in household was associated with delayed BCG, DTP3, polio3 and MCV vaccinations. Children living in urban areas were less likely to be vaccinated with delay against BCG, DTP3 and polio3. Children from poorer families were more likely to be vaccinated with delay with BCG, DTP3, polio3 and MCV vaccines, with a more pronounced effect for DTP3 and polio3 vaccinations.
Factors associated with delayed and missing vaccination compared with timely vaccination (results of six multilevel logistic regression analyses with country as a random effect)*
A slightly different pattern of association was observed for missing BCG, DTP3, polio3 and MCV vaccinations. There were no gender and urban/rural differences in missing DTP3 and MCV vaccinations; however, children living in urban areas were less likely to get delayed BCG and polio3 vaccinations. A similar association was observed for the wealth index and all vaccinations. This effect was stronger for missing vaccinations compared with delayed vaccinations.
Variation in the impact of economic factors across the countries
In a separate (univariable) analysis, we examined the association between delayed and missing DTP3 and MCV vaccinations and the wealth index for each country separately (figure 2). A heterogeneous picture appeared: there were countries in which no association existed (eg, Belarus, Bosnia and Herzegovina and Gambia) and countries with an association of a various degree; the strongest association was seen in Bangladesh, Cameroon, Ghana, Sierra Leone, and Togo. On average, the association was stronger for DTP3 than for MCV, and for each of the vaccines the association was similar for missing and delayed vaccination.
Associations between delayed and missing vaccinations and the wealth index. (A) third dose of diphtheria, tetanus and pertussis vaccine (DTP3) delayed vaccination; (B) measles-containing vaccine (MCV) delayed vaccinations; (C) DTP3 missing vaccination; and (D) MCV missing vaccination; reference group for all—timely vaccinations in very rich group.
Discussion
We examined up-to-date coverage, age-appropriate vaccination and delay in vaccination in 31 low and middle-income countries. Most countries achieved high up-to-date vaccination coverage, except Trinidad and Tobago (<40% for DTP vaccines), Ghana (<60%) and Yemen (<70%). Up-to-date coverage was above 90% for all vaccines in 13 countries. However, in some countries the coverage was not sufficiently high to eliminate childhood infections. For example, to maintain herd immunity and eliminate measles infection a coverage of at least 95% is necessary.1 Such coverage was only achieved in 10 countries. Also, some of the vaccinations administered well ahead of schedule could have been too early to generate a full immunological response,16 24 in consequence the real number of children protected by the vaccine was further reduced.
We also found that age-appropriate vaccination was considerably lower compared with up-to-date vaccination coverage, indicating a substantial amount of delay in the administration of vaccination. For example, only approximately half of the children were vaccinated timely with MCV and approximately 40% with DTP3. The shortest delay was observed for BCG vaccine (median 2.1 weeks, IQR 0.9–3.0). The longest delays were for the third doses of DTP and polio vaccines (6.3 weeks (3.3–9.0) and 6.6 weeks (4.3–9.3), respectively). In a recent analysis of the DHS data for 45 countries, Clark and Sanderson5 have shown similar findings; the median delay for BCG was 2.3 (1.4–3.3) and for DTP3 6.2 (3.5–8.5). However, the delay for MCV reported by Clark and Sanderson5 was lower (median 2.7 weeks, IQR 1.7–3.1 vs 4.1 weeks, 2.5–5.8 in our analysis). Our analysis included 10 countries for which data were also presented by Clark and Sanderson.5 We found only slightly different results with respect to delayed vaccination. For example, the median of the median delays across the 10 countries for BCG vaccine in our analysis was 2.7 weeks (vs 3.0 weeks in the analysis by Clark and Sanderson),5 2.3 weeks for DTP1 (vs 2.8 weeks), and 7.0 weeks for DTP3 (vs 7.2 weeks). The median of the median delays was higher in our analysis for MCV (4.2 weeks vs 2.6 weeks). A possible explanation for the larger difference is that in contrast to the previous analysis, which included all children, our analysis included only the youngest child per household. Having siblings increased the risk of delayed vaccination, and the effect was the strongest for MCV—in previous analysis all siblings were included, and this probably increased the average delay.
In general, we observed similar patterns of factors associated with missing and delayed vaccinations; there were socioeconomic differences regarding all three outcomes, namely the risk of the outcomes was higher among children from poorer families (based on the wealth index). This finding was mainly observed in some low and middle-income countries (eg, Bangladesh, Cameroon, Ghana, Sierra Leone and Togo). These findings point to socioeconomic inequalities and can be explained by, for example, inadequate access to healthcare services or poor utilisation of primary healthcare services among individuals with a lower socioeconomic status. Parental attitudes towards childhood vaccinations, which were found to differ among different socioeconomic groups,25 26 may also explain socioeconomic differences in vaccination uptake and negatively affect age-appropriate vaccination. A higher number of children in households also leads to the increased risk of delayed and missing vaccinations. The risk of delayed BCG and DTP3 vaccinations was also higher in rural areas compared with urban areas. Rural/urban differences in primary healthcare utilisation, including the utilisation of childhood vaccinations, still remain a problem in low and middle-income countries.27
Our study has several strengths and limitations. First, MICS3 generated nationally representative samples. MICS and DHS are both data sources recommended in the WHO manual ‘Immunisation coverage cluster survey’ for the assessment of vaccination coverage if no survey solely dedicated to assess this question is conducted.28 Second, the surveys applied standardised sampling methods and standardised questionnaires. Third, all surveys used in the current analysis were conducted in a short period of time (2005–7). All these three conditions allow the appropriate comparison of up-to-date vaccination coverage and age-appropriate vaccination across the studied countries. Information on vaccination was based on health cards and therefore has to be considered as more valid than if it was obtained from parental reports only.29 At the same time, health cards were available only for half of the sample, and there was a substantial variation in this fraction across countries. Children for whom health cards were not available might have a poorer vaccination status and possibly received less timely vaccination, which would result in overestimated vaccination coverage and better age-appropriate vaccination in our analysis than in the whole population. The reliability of information based on health cards is also sometimes questionable; this was discussed especially for former communist countries, but might persist in societal structures despite political transitions.30 Furthermore, our analysis was restricted only to variables available in the children datasets. Other predictors, primarily maternal age and education, parental attitudes towards vaccinations, might influence vaccination status and timely vaccinations.11 Some criticism can also be directed towards the use of the wealth index as a measure of socioeconomic status across different countries. To overcome differences in economic standards, the wealth index uses country-specific variables in addition to variables used in all countries. Nevertheless, it might not establish a full comparability across countries and therefore different effects of the wealth index on vaccination status can also partly result from a different performance of the wealth index in differentiating within the populations.
Conclusions
The overwhelming majority of the studied countries achieved high childhood vaccination coverage. However, the coverage in several countries was often not sufficiently high to maintain herd immunity and was further affected by a delayed administration of vaccines. Both missing and delayed vaccinations cause considerable variations in age-appropriate vaccination across countries. Therefore, the commonly used indicator ‘up-to-date vaccination coverage’ should be supplemented by ‘age-appropriate vaccination’. This may help to identify gaps in timely vaccinations and stimulate interventions. A socioeconomic gradient is not present in all countries but where it is present its effects are stronger than those of other variables included in our study. Understanding these differences may help resolve some of the problems. Similarity between factors associated with missing and delayed vaccination suggests that the same interventions can address both. Such interventions could be targeted information or even active invitations for parents. Our analysis provided information about ecological variation, studying its causes and possible associations between the organisation of healthcare systems and vaccination coverage needs further, more in-depth, studies.
What is already known on this subject
Not only missing but also delayed vaccination can substantially affect the vaccination status of the population. Socioeconomic characteristics are risk factors for missing vaccination. The extent of delayed vaccinations was not studied for many countries, and there is a limited knowledge of factors associated with delayed vaccination and with their variation across countries.
What this study adds
We found a substantial amount but also different patterns of delayed vaccination in an analysis of 31 low and middle-income countries. Given the similar factors associated with delayed and missing vaccination, they appear to be parts of the same continuum. The effect of the socioeconomic status varied across countries, being more pronounced in low-income countries (eg, African and south-east Asian countries).
Acknowledgments
The authors would like to thank UNICEF for providing them with the MICS datasets.
References
Footnotes
MKA and RTM contributed equally to this work.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.