Professional Documents
Culture Documents
Narváez Et Al. - 2017 - Immunization Coverage PDF
Narváez Et Al. - 2017 - Immunization Coverage PDF
Vaccine
journal homepage: www.elsevier.com/locate/vaccine
a r t i c l e i n f o a b s t r a c t
Article history: Objectives: This study aimed to evaluate the coverage of the Colombian Expanded Program on
Received 18 October 2016 Immunization among children less than 6 years old, to evaluate the timeliness of immunization, to assess
Received in revised form 19 January 2017 the coverage of newly introduced vaccines, and to identify factors associated with lack of immunization.
Accepted 30 January 2017
Methods: We conducted a cross-sectional survey in 80 municipalities of Colombia, using a two-stage
Available online 13 February 2017
cluster random sampling. We attempted to contact all children less than 6 years old living in the sampled
blocks, and asked their caregivers to provide immunization record cards. We also collected basic sociode-
Keywords:
mographic information.
Colombia
Vaccination
Results: We reached 81% of the attempted household contacts, identifying 18,232 children; of them,
Immunization schedule 14,805 (83%) had an immunization record card. Coverage for traditional vaccines was above 90%: BCG
Immunization programs (tuberculosis) 95.7% (95%CI: 95.1–96.4), pentavalent vaccine 93.3% (92.4–94.3), MMR (measles, mumps,
Risk factors rubella) initial dose 94.5% (93.5–95.6); but it was lower for recently introduced vaccines: rotavirus 80%
Immunization coverage (77.8–82.1), influenza 48.4% (45.9–50.8). Results for timely vaccination were not equally successful: pen-
Timely immunization tavalent vaccine 44.2% (41.4–47.1), MMR initial dose 71.2% (68.9–73.4). Mother’s education was signifi-
Underutilized vaccines cantly associated with higher immunization odds. Older age, a greater number of siblings, low
socioeconomic status, and not having health insurance were significantly associated with lower immu-
nization odds. There was significant heterogeneity in immunization rates by municipality across the
country.
Conclusions: Although absolute immunization coverage for traditional vaccines met the goal of 90% for
the 80 municipalities combined, disparities in coverage across municipalities, delayed immunization,
and decline of coverage with age, are common problems in Colombia that may result in reduced protec-
tion. Newly introduced vaccines require additional efforts to reach the goal. These results highlight the
association of health inequities with low immunization coverage and delayed immunization.
Identification of vulnerable populations and their missed opportunities for vaccination may help to
improve the reach of immunization programs.
Ó 2017 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.vaccine.2017.01.073
0264-410X/Ó 2017 Elsevier Ltd. All rights reserved.
1502 J. Narváez et al. / Vaccine 35 (2017) 1501–1508
summary measure may hide sub-national disparities. In 2012, and the range of administrative coverage reports, including munic-
about 50% of the municipalities in Latin America and the Caribbean ipalities with high (>95%) and low (<70%) coverage reports from
were below 95% coverage, and 23% were below 80% [4], thus creat- the previous year. Within those municipalities, which can be
ing pockets of low coverage that represent a risk for reemergence deemed as strata, we used a two-stage cluster sampling strategy,
of vaccine-preventable diseases. with census block groups as primary sampling units, and the
As middle-income countries strengthen their immunization blocks within them as secondary units. At both stages, the sam-
programs, the national coverage may meet the goal proposed by pling units were selected using probabilities of inclusion propor-
the WHO, especially for vaccines that have been in use for decades tional to the size of the total population living in each unit, as
[5]. However, three key problems may be missed when using an described by Särndal et al. [10]. Once a block was selected in the
evaluation approach centered on national coverage: (1) the impor- sample, all the households in that block were surveyed to identify
tance of timely vaccination may be overlooked; (2) national aver- all children in the targeted age interval. The sampling frame and
ages may hide within-country heterogeneities in immunization population sizes were supplied by the Colombian National Admin-
coverage; and (3) coverage may be lower for newly introduced istrative Department of Statistics (i.e. the cartographic statistical
vaccines. frame referenced in the Methodological Summary Sheet of the
Colombia reports a high national coverage [6], but those three Colombian census of 2005 [11]).
issues are a concern. First, in spite of the paucity of data regarding
timely vaccination, at least two Colombian studies [7,8] have iden- 2.2. Participants
tified delayed vaccination as a potential barrier for the success of
immunization programs. Second, Colombia is a very heterogeneous Children under 6 years of age residing in the selected blocks
country comprising 1102 municipalities (i.e., districts), the political were considered eligible for the survey if: (1) there was at least
and administrative subdivisions ultimately responsible for imple- one adult in the household during the interview, (2) the adult con-
menting, administering and evaluating the immunization program sented to serve as informant for the survey, and (3) the informant
at the local level. Since their cultural, social and economic charac- reported that the visited household was the primary place of resi-
teristics are very diverse, pronounced inequalities in immunization dence of the child. When it was not possible to locate an informant
rates across municipalities are expected. In addition, it is estimated at the first attempt, we did up to two additional visits to maximize
that 76% of the total population lived in urban areas as of 2012, and the chance of including all eligible children.
nearly 50% lived in the 10 major cities [9]. As a consequence, an
average national immunization rate may be easily driven by the 2.3. Data sources and data collection
major urban centers, which are likely to have better coverage than
other territories. Third, three vaccines have been added to the The CEPI uses immunization record cards to keep track of the
Colombian Expanded Program on Immunization (CEPI): seasonal immunization history of all children in the country. Each time a
influenza (added in 2007), oral rotavirus vaccine (2009), and con- vaccine is administered, it is recorded on the immunization card,
jugate pneumococcal vaccine (2010). However, adherence to this which is kept by the child’s caregivers. In this survey, the infor-
new expanded immunization schedule has not been thoroughly mant at each household was asked to provide immunization record
evaluated. cards for all eligible children living in the residence. When record
Following these considerations, the aims of the present study cards were not available, this was annotated along with the reason
were: (1) to evaluate the differences between the absolute immu- reported for not having it. Information regarding immunization
nization coverage and timely immunization coverage in children status and dates was directly taken from the immunization record
less than 6 years old in a sample of municipalities in Colombia in cards. In addition, the informants were asked to respond a ques-
2012; (2) to evaluate the heterogeneity in immunization rates tionnaire collecting socio-demographic information about the
across municipalities; (3) to evaluate the coverage of the vaccines household, the respondent, and the child.
recently added to the CEPI; and (4) to identify the sociodemo- All field procedures adhered to international guidelines on eth-
graphic factors associated with lack of vaccination and late admin- ical conduct of research [12], and were approved by the institu-
istration of vaccines among children less than 6 years in Colombia tional review board at the National University of Colombia
in 2012. School of Medicine.
Improved understanding of the factors associated with immu-
nization coverage and timely vaccination will help to implement 2.4. Variables and measures
appropriate interventions, or redirect the program efforts to secure
the coverage of population groups with non-modifiable Vaccines included in the CEPI as of 2012 were BCG (tuberculo-
vulnerabilities. sis), hepatitis B at birth, oral polio, pentavalent (DTP, Haemophilus
influenzae type b, Hepatitis B), oral rotavirus, pneumococcal conju-
gate, seasonal influenza virus, MMR, DTP booster doses, and yellow
2. Materials and methods fever. Several of these are administered simultaneously according
to a schedule based on child’s age. For the purpose of this paper
2.1. Study design and setting we focused on two main subsets of vaccines: (1) an indicator sub-
set of the basic schedule (BCG, pentavalent, MMR, DTP), selected to
Between June and December 2012, we conducted a cross- reflect the different ages at which vaccines should be administered,
sectional survey of children less than 6 years of age residing in and (2) the subset of those vaccines more recently introduced to
80 municipalities of Colombia. These municipalities were selected the CEPI (rotavirus, pneumococcal conjugate, and influenza virus).
by the Colombian Ministry of Health (CMH) in a non-probabilistic For each subset, we evaluated absolute and timely coverages.
manner to cover the spectrum of the cultural, geographical and Absolute coverage was defined as the proportion of children having
demographic heterogeneity of the country; thus, the municipalities the number of doses of each vaccine required for their age, even if
were distributed across the five geographical regions of Colombia attained through the use of a catch-up schedule. Timely coverage
(the Andean region, the Amazon basin, the Caribbean coast, the was defined as the proportion of children receiving every single
Pacific coast, and the Eastern plains), the spectrum of urbanicity, dose within the time frame specified in the immunization schedule
including large urban centers as well as predominantly rural areas, and guidelines published by the CMH [13]. For example, if a child
J. Narváez et al. / Vaccine 35 (2017) 1501–1508 1503
who was 9 months old at the time of the survey had a BCG vaccine uate between-municipalities differences in vaccination coverage,
administered at birth, and 3 doses of the pentavalent vaccine and to allow for random intercepts accounting for between-
administered at 5, 7, and 9 months old, then the child was consid- cluster differences in the baseline odds of immunization that
ered completely vaccinated according to the absolute coverage remained unexplained by the variables in the models at the munic-
indicator, but was not counted as timely vaccinated since the ipality, census block group, and block levels.
immunization schedule specifies that pentavalent vaccine should All the analytical procedures were conducted using Stata Statis-
be administered at 2, 4 and 6 months old. Details about the accept- tical Software release 12.1 [15], along with the Stata program
able Colombian immunization timeframe are provided later in this gllamm, documented by Rabe-Hesketh and Skrondal [16], for the
paper. Children still lacking at least one dose of a vaccine that they generalized linear latent and mixed models. All procedures used
should have already received according to their age by the date of a statistical significance level a = 0.05.
interview were deemed as incompletely immunized.
We also collected information about socioeconomic position
(SEP), defined according to Colombian law [14] as an ordinal vari- 3. Results
able with 6 strata, where 1 is the more deprived category and 6 is
the wealthiest; this measure is assigned by municipal governments In total, we identified 18,232 children living in 12,846 house-
to geographically defined groups of households and is linked to the holds across 80 municipalities (successful contact rate at the
cost of utilities, but also to rent, real estate prices, and taxes. Addi- household level = 81%). Demographic data were obtained for
tional information collected during the interview included status 17,860 children whose parents agreed to participate in the survey
as internally displaced by the armed conflict (yes vs. no), time (participation rate among successfully contacted = 98%). Of them,
residing in the municipality, maximum education level attained 14,805 (82.9%) had an immunization card available at the time of
by the informant (complete high-school or above vs. less than the interview. Table 1 provides a summary description of the
high-school), employment status of the mother (paid employment 14,805 children with an immunization card. Among the 3055 with-
vs. unpaid/unemployed), and self-identification as belonging to a out an immunization card, only 0.72% reported not having it due to
racial/ethnic minority (yes vs. no). Finally, the child’s gender, birth not being vaccinated. According to the informants, other reasons
date, birth order, number of siblings, and affiliation to the health for not having the immunization record card were that it was kept
system (insured in the regular contributive regime, in any of the in a relative’s house (45%), it was lost (16%), or it was archived in a
special regimes, in the subsidized regime, or not insured) were daycare facility (16%).
collected. The overall estimated averages for absolute and timely immu-
nization coverage with different vaccines in the 80 municipalities
2.5. Statistical analyses are shown in Tables 2 and 3, respectively. The above defined indi-
cator subset of vaccines had an overall absolute coverage of 78.1%
We estimated ratios of vaccinated and timely vaccinated chil- (95% CI: 76.6–79.5), whereas timely coverage for this subset was
dren to the total number of children with an immunization record 32.7% (95% CI: 31.0–34.4). Only 13 of the 80 evaluated municipal-
card, at the municipality level and for the 80 municipalities com- ities had an absolute immunization coverage above the 90% goal,
bined. We estimated totals for the numerator and the denominator while 21 had an absolute coverage below 70%. Regarding timely
of each ratio using expansion factors inverse to the inclusion prob- coverage, no single municipality achieved the proposed goal of
abilities of the sampled units. Then, we calculated the ratios along 90% for the same indicator subset of vaccines, with point estimates
with their bounded 95% confidence intervals (CIs). The standard ranging from 8% to 52%. Mean delays in immunization with the dif-
errors used in the CIs calculation took into account the intra- ferent vaccines are presented in Table 4, along with the acceptable
class correlation coefficients and sampling weights to adjust for immunization timeframe according to the guidelines of the CMH.
the complex design effect of the survey. Multivariable analyses using generalized linear latent and
Finally, we calculated odds ratios (ORs) to estimate the associa- mixed models showed that child’s age, number of siblings, belong-
tion of sociodemographic factors with absolute immunization cov- ing to the lowest SEP category, identifying as racial/ethnic minor-
erage and timeliness of vaccine administration. We used ity, and lack of health insurance, were all inversely and
generalized linear latent and mixed models to simultaneously significantly associated with immunization status in terms of abso-
adjust for the correlation induced by the sampling scheme, to eval- lute coverage (Table 5); by contrast, a mother completing high
Table 1
Sociodemographic characteristics of study participants in a survey of immunization coverage with an indicator vaccine schedule (BCG, pentavalent, MMR, DTP) among children
less than 6 years of age living in 80 municipalities of Colombia in 2012.
Table 2
Average immunization coverage for different vaccines in Colombia in 2012, estimated from a survey of children less than 6 years of age living in 80 municipalities
Type of vaccine Average coverage estimate 95% CI for the average Range of coverage across municipalities
BCG at birth 95.7% 95.1–96.4 77.2–100
Hepatitis B at birth 91.9% 91–92.8 64.6–100
Pentavalent, 3 doses 93.3% 92.4–94.3 72.8–100
DPT, 1st booster dose 87.9% 86.4–89.4 59.5–100
DPT, 2nd booster dose 66.4% 62.6–70.1 7.0–100
Yellow fever, single dose 91.6% 90.2–93.1 67.6–100
MMR, initial dose 94.5% 93.5–95.6 62.5–100
MMR, booster dose 57.1% 52.7–61.5 8.0–98.9
Oral polio, three initial doses 93.8% 92.8–94.8 31.1–100
Oral polio, first booster dose 89.8% 88.7–91 60.0–100
Oral polio, second booster dose 69.8% 65.9–73.6 7.0–100
Influenza, two doses 48.4% 45.9–50.8 13.5–86.4
Rotavirus, two doses 80.0% 77.8–82.1 41.8–100
Pneumococcal conjugate, two initial doses 77.9% 75.5–80.3 37.0–100
Pneumococcal conjugate, booster dose 14.6% 13.1–16.1 1.4–32.1
Indicator subset (BCG, pentavalent, MMR, DTP) 78.1% 76.6–79.5 46.3–96.8
Table 3
Average timely immunization coverage for different vaccines in Colombia in 2012, estimated from a survey of children less than 6 years of age living in 80 municipalities.
Type of vaccine Average timely coverage estimate 95% CI for the average Range of coverage across municipalities
BCG at birth 87.7% 86.6–88.8 63.5–98.2
Hepatitis B at birth 86.9% 85.8–88 55.1–97.4
Pentavalent, 3 doses 48.9% 47.1–50.7 19.0–68.8
DPT, 1st booster dose 44.4% 42.1–46.7 9.7–69.0
DPT, 2nd booster dose 61.8% 58.1–65.4 7.0–93.9
Yellow fever, single dose 71.7% 69.6–73.7 40.4–90.1
MMR, initial dose 75.0% 73.1–77 38.2–90.8
MMR, booster dose 48.3% 43.5–53.2 8.0–93.4
Oral polio, three initial doses 49.6% 47.9–51.3 17.2–71.4
Oral polio, first booster dose 45.9% 43.6–48.2 17.6–70.8
Oral polio, second booster dose 66.1% 62.3–69.8 7.0–100
Influenza, two doses 7.4% 6.6–8.1 1.0–18.3
Rotavirus, two doses 66.2% 64.2–68.3 32.0–88.9
Pneumococcal conjugate, two initial doses 47.3% 44.8–49.8 11.8–95.1
Pneumococcal conjugate, booster dose 11.2% 10–12.4 0.9–31.0
Indicator subset (BCG, pentavalent, MMR, DTP) 32.7% 31–34.4 7.9–51.7
school or above was associated with a 16% higher odds of being tinely estimated by countries and reported by the WHO and UNI-
immunized. CEF as part of the surveillance of immunization programs, thus
Regarding timely vaccination coverage, we also found that allowing comparisons between countries and along time [3], it
child’s age, number of siblings, belonging to the lowest SEP cate- may be insufficient to describe some of the more recent challenges
gory, lack of health insurance, and having subsidized health insur- experienced by immunization programs, such as the persistence of
ance, were inversely associated with timely immunization status. low coverage pockets, and the missed opportunities for vaccination
On the other hand, children whose mother completed high school, resulting in untimely immunization [4]. In this study, the observed
children with older siblings, and those whose families were living gap between absolute and timely coverage was as large as 40 per-
in the town for more than one year had higher odds of timely cent points for pentavalent (three doses) and oral polio (three ini-
immunization. In contrast to what was observed for absolute cov- tial doses) vaccines, suggesting that absolute immunization
erage, identifying as a racial/ethnic minority was not associated coverage may importantly overestimate the success of the CEPI,
with timely immunization (Table 5). A graphical summary of the and that indicators of timeliness of immunization, in addition to
main findings described above is presented in Fig. 1. indicators of absolute coverage, may help to depict a more compre-
In the absolute coverage analysis, we found statistically signifi- hensive evaluation of the performance of immunization programs.
cant and sizable heterogeneities in the baseline odds of immuniza- The observed discrepancies are consistent with studies in other
tion at the municipality level, the census block group level, and the countries [17–20] that have raised questions about the utility of
block level (p-values < 0.0001 from likelihood ratio tests for the absolute coverage as an indicator of success of the immunization
contribution of random intercepts). Random effects for hetero- programs. Since absolute coverage does not necessarily reflect
geneity in the baseline odds of timely immunization were also sig- timely administration of vaccines, it may be a biased estimator of
nificant at the municipality and census block group levels, but not the fraction of the population truly protected [19], and relying
at the block level. exclusively on it may place the population at risk of outbreaks
[20]. On the other hand, routine estimation of indicators of timeli-
4. Discussion ness of immunization may suffer from challenges related to data
quality and availability. In that regard, the relative advantages
This survey evaluated the differences between absolute and and disadvantages of different data sources for the estimation of
timely immunization rates among children less than 6 years in absolute coverage have been previously described [21,22], and
Colombia in 2012. While absolute immunization coverage is rou- the same strengths and limitations apply to estimators of timely
J. Narváez et al. / Vaccine 35 (2017) 1501–1508 1505
Table 4
Ranges of age for timely immunization as defined in the Colombian Expanded Program of Immunization, and mean delay in immunization (days) among children less than 6 years
of age living in 80 municipalities of Colombia in 2012.
Type of vaccine Range of timely administration as defined Mean (SD) delay in days counted n delayed
in the Colombian Expanded Program from last possible day of timely
of Immunization administration
BCG at birth From 0 to 28 days of life 128 (244) 1369
Hepatitis B at birth From 0 to 28 days of life 135 (271) 876
Pentavalent, 3 doses At 2, 4 and 6 months; each dose with a 1 month leeway 117 (191) 4547
DPT, 1st booster dose Between 15 and 18 months 233 (264) 1903
Yellow fever, single dose Between 12 and 15 months 211 (265) 1614
MMR, initial dose Between 12 and 15 months 232 (289) 1461
Oral polio, three initial doses At 2, 4 and 6 months; each dose with a 1 month leeway 112 (185) 4524
Oral polio, first booster dose Between 15 and 18 months 222 (244) 1934
Influenza, two doses Initial dose at 6 months, second at 7 months; each dose with 275 (287) 5335
a 1 month leeway
Rotavirus, two doses First dose between weeks 8 and 15; second dose between 4 130 (153) 241
and 7 months
Pneumococcal conjugate, two initial doses At 2 and 4 months; each dose with a 1 month leeway 87 (108) 1567
Pneumococcal conjugate, booster dose Between 12 and 15 months 130 (112) 225
Table 5
Factors associated with absolute immunization and timely immunization among children less than 6 years of age living in 80 municipalities of Colombia in 2012, using an
indicator subset of vaccines as a proxy for total coverage
coverage; however, there may be additional difficulties related to was predominantly taken from rural settings, we identified a sig-
the lack of standardization of the definitions of timely immuniza- nificantly lower timely coverage in the rural area (32% vs 16%,
tion. For instance, the definitions used along this paper and p = 0.001), although this was not the case for absolute coverage
detailed in Table 4 are not identical to those described in prior (77% vs. 86%, p = 0.14). This suggests that urbanicity may be
studies of missing opportunities for vaccination and timeliness of another factor driving within-country heterogeneity, insufficiently
immunization [4]. The large difference between absolute and explored in the present survey, but worth to examine in future
timely coverage observed in our study is not unprecedented: in studies.
Uganda in 2008, this difference was reported to be as large as Heterogeneities in vaccine access have been linked to material
75% vs. 18% [18]. deprivation/wealth [24–26], lack of health insurance [27], lack of
This survey also demonstrated important disparities between education [22,24,25], distance to health centers [25], and racial/
municipalities: the absolute coverage for the indicator subset of ethnic inequities [28]. In the present study too, sociodemographic
vaccines varied between 46% and 97%, while for timely coverage inequalities contributed to explain immunization coverage.
the range was from 8% to 52%. This shows immunization coverage Mother’s education at the high school level or beyond was associ-
in Colombia is far from homogeneous, and a summary measure like ated with higher odds of absolute and timely immunization. In
the mean coverage across the entire study population may be comparison, older age, a greater number of siblings, low SEP, and
insufficient to capture and describe within-country disparities. lack of health insurance contributed to lower absolute and timely
This is in agreement with a 2005 study in Papua New Guinea, immunization. Analogous findings have been observed in the Uni-
where sub-national areas had much lower immunization rates ted States, where racial/ethnic minorities and groups served by
than the national average [23]. Although the present analysis public vaccination providers remain persistently under-reached
was not specifically aimed to contrast the coverage between rural by immunization programs [29]. However, as opposed to previous
and urban areas, a secondary analysis of our data showed that reports [27], internal displacement was not linked to lower immu-
absolute coverage and timely coverage were not different compar- nization coverage in this analysis. We hypothesize this may be a
ing 21 municipalities with more than 50% rural population vs. the result of the efforts of governmental and non-governmental agen-
other 59 (77% vs. 78% for absolute coverage, and 30 vs. 34% for cies to provide internally displaced families with basic health and
timely coverage). However, in 4 municipalities were the sample social assistance.
1506 J. Narváez et al. / Vaccine 35 (2017) 1501–1508
(a) (b)
80
90
70
Proportion immunized
Absolute coverage
80
60
70
50
60
40
30
50
Absolute coverage Timely coverage Below median Above median
(c)
100
(d)
100
90
90
Absolute coverage
Absolute coverage
80
80
70
70
60
60
50
50
Fig. 1. Immunization coverage, timely immunization, variation with age, and coverage of new or underutilized vaccines in Colombia, 2012. The vertical bars represent 95%
confidence intervals for immunization coverage and the central marks are the point estimates. (a) The red arrow illustrates the gap between national absolute immunization
coverage and timely coverage. (b) The red arrow illustrates the difference in mean absolute coverage between the top half and bottom half of the studied municipalities,
classified according to coverage levels. (c) The red line illustrates the stated goal for national immunization coverage. (d) The red line illustrates the decay in immunization
coverage through age. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Most factors associated with immunization coverage and timely absolute coverage greater than 90%. These low coverage levels may
coverage in our models are to some degree inherent to the studied leave room for outbreaks among the unprotected population.
individuals (e.g. age, race, birth order), their families (e.g. mother’s Specific reasons for the low uptake of vaccines were not ascer-
education, number of siblings), or are structurally related to the tained in this study, but other studies suggest this may be due to
organization of the society they live in (e.g. types of health insur- socio-economic and geographic barriers [25], barriers related to
ance, low SEP). These constitute barriers for immunization that the organization of the immunization program [31], patterns of
can and should be addressed by public health systems. In order utilization of health care services [32], and beliefs and attitudes
to do so, associations of these factors with immunization coverage from parents and guardians [33].
must be translated into interventions on modifiable events that One important limitation of this study is related to the use of
contribute to explain such associations [30]. For instance, identify- immunization record cards as the primary source of information.
ing as a racial minority is not in itself modifiable, but racial dispar- It has been shown in countries with robust health information sys-
ities in immunization can be interpreted as the result of a complex tems that methods based on record cards tend to underestimate
set of social disadvantages experienced by persons who belong to a immunization rates when compared to health care provider
racial minority, and policies can be put into effect to ameliorate sources, because a substantial proportion of children lacking
these disadvantages. For instance, incentives to households and immunizations on the card may actually be vaccinated [34]. How-
health care workers have been suggested by the WHO as strategies ever, the limitations of health information systems in Colombia
to reach underserved populations, as long as the autonomy of the [35] do not allow to use that approach. A nominal electronic EPI
beneficiaries is respected [2]. registry is currently under implementation and may serve to this
Another relevant finding concerns the differences between purpose in the future. Additionally, the absolute coverage indicator
recently introduced vaccines and traditional ones. Absolute cover- for many of the individual vaccines in this study was above 90%,
age for rotavirus and pneumococcal vaccines was only 80%, and for leaving little room for a substantial underestimation.
seasonal influenza vaccine was less than 50%, while traditional Although this study was initially conceived as a national survey,
vaccines targeted towards children of the same age range reported generalizability of the results to specific municipalities not
J. Narváez et al. / Vaccine 35 (2017) 1501–1508 1507
included in the sample may be limited by the profound hetero- [6] Ministerio de Salud y Protección Social – República de Colombia. [Vaccination:
Achievements of the Government] [Spanish]. 2012 <https://www.
geneity revealed by this study, and the non-probabilistic nature
minsalud.gov.co/DocumentosyPublicaciones/Logros+del+Gobierno+en
of the initial selection of the municipalities provided by the Colom- +vacunacion.pdf> [accessed September 27, 2016].
bian Ministry of Health. Thus, the conclusions of this study apply to [7] de La Hoz F, Perez L, de Neira M, Hall AJ. Eight years of hepatitis B vaccination
the 80 municipalities included in the survey, and possibly to others in Colombia with a recombinant vaccine: factors influencing hepatitis B virus
infection and effectiveness. Int J Infect Dis 2008;12:183–9. http://dx.doi.org/
within a similar spectrum of geographic, cultural, and demographic 10.1016/j.ijid.2007.06.010.
characteristics, but the results should not be regarded as a quanti- [8] Morón L, Espitia M. A Rapid Evaluation of Vaccination Coverage in Bogotá,
tative summary of the national immunization coverage. Addition- 2006. Rev Salud Pública 2009;11:237–46. http://dx.doi.org/10.1590/S0124-
00642009000200008 (Spanish).
ally, direct comparison of these results to the routine surveillance [9] National Administrative Department of Statistics – DANE (Colombia).
of the CEPI may be limited by the methodological differences Estimates and projections of the total national, departmental and municipal
between these two assessment approaches. Routine surveillance population by area, 1985-2020; 2009 <https://www.dane.gov.co/files/
investigaciones/poblacion/proyepobla06_20/Municipal_area_1985-2020.xls>
is based on administrative coverage and rapid monitoring surveys, [accessed September 27, 2016].
but these methodologies may be affected by inaccuracies in the [10] Särndal E, Swensson B, Wretman J. Chapter 3.6. Probability proportional-to-
administrative population denominators and disparities in immu- size sampling. Model Assist. Surv. Sampl.. New York: Springer-Verlag; 1992. p.
87–99.
nization practices between and within municipalities [21]. [11] National Administrative Department of Statistics – DANE (Colombia).
[Methodological Summary Sheet of the General Census, 2005][Spanish];
4.1. Conclusions 2006 <http://www.dane.gov.co/files/censo2005/ficha_metodologica.pdf>
[accessed September 27, 2016].
[12] WMA General Assembly. World Medical Association Declaration of Helsinki:
Immunization of children with the basic schedule of traditional Ethical Principles for Medical Research Involving Human Subjects (as amended
vaccines included in the CEPI in Colombia in 2012 met the pro- by the 59th WMA General Assembly, Seoul, October 2008). World Med Assoc
2008.
posed WHO goal of 90% coverage, although coverage of booster [13] Ministerio de Salud y Protección Social - República de Colombia. [Vaccination
doses in older children fell far below that goal. In addition, cover- Week in the Americas 2012: a framework document][Spanish]; 2012:28–30
age for the more recently introduced vaccines was low. Delayed <https://www.minsalud.gov.co/Lineamientos/Documento_Marco_Jornada_
de_Vacunacion_de_las_Américas_abril2012.pdf> [accessed September 27,
immunization is a common problem in Colombia, which may 2016].
result in reduced levels of individual protection and create a pool [14] Congreso de la República de Colombia. Ley 142 de 1994. The Congress of the
of susceptible individuals who can spread infection throughout Republic of Colombia; 1994.
[15] StataCorp LP. Stata Data Analysis and Statistical Software: release 12.1; 2014.
the population. This study showed that socioeconomic inequities
[16] Rabe-Hesketh S, Skrondal A. Multilevel and Longitudinal Model Using Stata.
are important determinants of immunization coverage and 2nd ed. College Station, TX: Stata Press; 2008.
delayed immunization in Colombia. The identification of vulnera- [17] O’Grady K-A, Krause V, Andrews R. Immunisation coverage in Australian
Indigenous children: Time to move the goal posts. Vaccine 2009;27:307–12.
ble sub-populations may help to focus interventions aimed at cre-
http://dx.doi.org/10.1016/j.vaccine.2008.09.096.
ating incentives and addressing barriers for immunization, as well [18] Fadnes LT, Nankabirwa V, Sommerfelt H, Tylleskär T, Tumwine JK, Engebretsen
as to improve the efficacy and reaching capacity of immunization IMS, et al. Is vaccination coverage a good indicator of age-appropriate
programs until a more comprehensive social response is put in vaccination? A prospective study from Uganda. Vaccine 2011;29:3564–70.
http://dx.doi.org/10.1016/j.vaccine.2011.02.093.
effect in order to tackle health inequities. [19] Akmatov MK, Kretzschmar M, Krämer A, Mikolajczyk RT. Timeliness of
vaccination and its effects on fraction of vaccinated population. Vaccine
2008;26:3805–11. http://dx.doi.org/10.1016/j.vaccine.2008.05.031.
5. Source of funding [20] Dannetun E, Tegnell A, Hermansson G, Törner A, Giesecke J. Timeliness of MMR
vaccination–influence on vaccination coverage. Vaccine 2004;22:4228–32.
This work was supported by the Ministry of Health and Social http://dx.doi.org/10.1016/j.vaccine.2004.04.018.
[21] Dietz V, Venczel L, Izurieta H, Stroh G, Zell ER, Monterroso E, et al. Assessing
Protection of the Republic of Colombia. and monitoring vaccination coverage levels: lessons from the Americas. Rev
Panam Salud Pública = Pan Am J Public Heal 2004;16:432–42. http://dx.doi.
org/10.1590/S1020-49892004001200013.
6. Conflicts of interest [22] Casey RM, Dumolard L, Danovaro-Holliday MC, Gacic-Dobo M, Diallo MS,
Hampton LM, et al. Global Routine Vaccination Coverage, 2015. MMWR Morb
None. Mortal Wkly Rep 2016;65:1270–3. http://dx.doi.org/10.15585/mmwr.
mm6545a5.
[23] Toikilik S, Tuges G, Lagani J, Wafiware E, Posanai E, Coghlan B, et al. Are hard-
Acknowledgements to-reach populations being reached with immunization services? Findings
from the 2005 Papua New Guinea national immunization coverage survey.
Vaccine 2010;28:4673–9. http://dx.doi.org/10.1016/j.vaccine.2010.04.063.
We are thankful to the Ministry of Health and to the National [24] Kawakatsu Y, Honda S. Individual-, family- and community-level
Administrative Department of Statistics for their assistance provid- determinants of full vaccination coverage among children aged 12–23
ing the sampling frame and other key information required for plan- months in western Kenya. Vaccine 2012;30:7588–93. http://dx.doi.org/
10.1016/j.vaccine.2012.10.037.
ning this study. We are also thankful to Dr. S. Hoffman, K. Alvarez, S. [25] Rahman M, Obaida-Nasrin S. Factors affecting acceptance of complete
Vaidya, E. Kulick, J. Blachman-Forshay and M. Goldberg for their immunization coverage of children under five years in rural Bangladesh.
invaluable feedback on preliminary versions of this manuscript. Salud Pública de México 2010;52:134–40. http://dx.doi.org/10.1590/S0036-
36342010000200005.
[26] Konstantyner T, Taddei JA de AC, Rodrigues LC. Risk factors for incomplete
References vaccination in children less than 18 months of age attending the nurseries of
day-care centres in Sao Paulo, Brazil. Vaccine 2011;29:9298–302. http://dx.
doi.org/10.1016/j.vaccine.2011.10.02.
[1] WHO, UNICEF, World Bank. State of the world’s vaccines and immunization.
[27] Ruiz-Rodríguez M, Vera-Cala LM, López-Barbosa N. [Health insurance for
3rd ed. Geneva: World Health Organization; 2009.
infants and infant vaccination related to forced-displacement in Colombia].
[2] WHO. Global vaccine action plan 2011-2020. World Health Organization;
Rev Salud Pública 2008;10:49–61. http://dx.doi.org/10.1590/S0124-
2013.
00642008000100005 (Spanish).
[3] Burton A, Monasch R, Lautenbach B, Gacic-dobo M, Neill M, Karimov R. WHO
[28] Bárcenas-López RM, Curiel-Reyes R, Caballero-Hoyos R, Villaseñor Sierra A.
and UNICEF estimates of national infant immunization coverage: methods and
Low vaccine coverage in indigenous migrant boys and girls. Salud Pública de
processes 2009:535–41. http://dx.doi.org/10.2471/BLT.08.053819.
México 2012;54:361–2. http://dx.doi.org/10.1590/S0036-
[4] Pan American Health Organization. Methodology for the evaluation of missed
36342012000400001.
opportunities for vaccination. Washington DC: PAHO; 2013.
[29] Zhao Z, Smith PJ. Trends in vaccination coverage disparities among children,
[5] Pan American Health Organization. Coverage in the Americas by Country 2016
United States, 2001–2010. Vaccine 2013;31:2324–7. http://dx.doi.org/
<http://ais.paho.org/phip/viz/im_coveragebycountry.asp> [accessed
10.1016/j.vaccine.2013.03.018.
September 27, 2016].
1508 J. Narváez et al. / Vaccine 35 (2017) 1501–1508
[30] Greenland S. Epidemiologic measures and policy formulation: lessons from [33] Favin M, Steinglass R, Fields R, Banerjee K, Sawhney M. Why children are not
potential outcomes. Emerg Themes Epidemiol 2005;2:5. http://dx.doi.org/ vaccinated: a review of the grey literature. Int Health 2012;4:229–38. http://
10.1186/1742-7622-2-5. dx.doi.org/10.1016/j.inhe.2012.07.004.
[31] Ransom J, Schaff K, Kan L. Is there an association between local health [34] Miles M, Ryman TK, Dietz V, Zell E, Luman ET. Validity of vaccination cards and
department organizational and administrative factors and childhood parental recall to estimate vaccination coverage: a systematic review of the
immunization coverage rates? J Health Hum Serv Adm 2012;34:418–55. literature. Vaccine 2013;31:1560–8. http://dx.doi.org/10.1016/
[32] Glanz JM, Newcomer SR, Narwaney KJ, Hambidge SJ, Daley MF, Wagner NM, j.vaccine.2012.10.089.
et al. A population-based cohort study of undervaccination in 8 managed care [35] Pan American Health Organization. Health in the Americas: 2012 Edition.
organizations across the United States. JAMA Pediatr 2013;167:274–81. http:// Regional outlook and country profiles. 2012 ed. Washington DC: PAHO; 2012.
dx.doi.org/10.1001/jamapediatrics.2013.502.