To evaluate progress in timely vaccination coverage associated with low-income households.
The US National Immunization Survey.
Children aged 19 to 35 months living in low-income households who were sampled between 1995 and 2007 (N = 232 318). Low-income households had an annual income that was 133% or less of the federal poverty level, and high-income households had an annual income of 400% or more of the federal poverty level.
Main Outcome Measures
Administration of 4 or more doses of diphtheria, tetanus, pertussis (DTaP-DTP) vaccine; 3 or more doses of polio; 1 or more doses of measles, mumps, rubella (MMR); 3 or more doses of Haemophilus influenzae type b (Hib); 3 or more doses of hepatitis B; and 1 or more doses of varicella vaccines by age 19 months as reported by the children's vaccination providers. Progress in timely coverage was evaluated by tracking changes between consecutive annual birth cohorts born between 1994 and 2004.
Among low-income children, timely vaccination coverage increased significantly between consecutive birth cohorts by an estimated 0.5% for DTaP-DTP, 0.3% for polio, 0.6% for MMR, 1.2% for hepatitis B, and 5.3% for varicella vaccines but did not change significantly for the Hib vaccine. Disparities in timely coverage for low- vs high-income children declined significantly between consecutive birth cohorts by an estimated −0.3% for MMR, −0.3% for hepatitis B, and −0.5% for varicella vaccines, did not change significantly for the polio vaccine, and increased significantly by 0.4% for the DTaP-DTP vaccine.
Disparities in vaccination coverage associated with low household income persist. Further progress in timely vaccination may be achieved by improving health care providers' reminder/recall systems, implementing educational interventions that address barriers to vaccination, and increasing parents' awareness of the Vaccines for Children Program.
In the United States, there was a resurgence in the incidence of measles between 1989 and 1991.1 Research conducted by the Centers for Disease Control and Prevention (CDC) revealed that cases observed during the measles resurgence were disproportionately inner-city, preschool-aged, American Indian, Hispanic, or black children younger than 5 years who had not been vaccinated2-7 and who were living in low-income households.8 In response to the outbreaks, in 1993 the Childhood Immunization Initiative9 was proposed to address significant disparities in vaccination coverage among young children in America. Among the strategies for achieving this goal was to increase access to vaccines by eliminating cost as a barrier. In October 1994, the Vaccines for Children (VFC) Program10 was established to achieve this goal by providing publicly purchased vaccines at no cost to VFC-enrolled vaccination providers to be administered to children who were eligible for the program.
An important conclusion of CDC research conducted during the US measles resurgence was that vaccines need to be administered on time because delays indicate inadequate protection against vaccine-preventable disease.
The purpose of this study is to describe progress in timely coverage for the diphtheria, tetanus, pertussis (DTaP-DTP); polio; measles, mumps, rubella (MMR); hepatitis B; Haemophilus influenzae type b (Hib); and varicella vaccines among 19- to 35-month-old children living in low-income households. Progress in timely pneumococcal conjugate vaccination coverage among low-income children has been described previously.11,12 Specifically, in this study, we describe the estimated percentage of children living in low-income households; evaluate progress in timely vaccination coverage for consecutive annual birth cohorts born between 1994 and 2004 among children living in low-income households; and evaluate whether disparities in timely vaccination coverage between children living in low- and high-income households have increased or decreased between these consecutive birth cohorts.
National immunization study
We analyzed data from 232 318 children sampled by the 12 annual surveys of the National Immunization Study (NIS) conducted between 1995 and 2007 who had provider-reported vaccination histories. In the first phase of NIS data collection, a list-assisted random digit dialing survey is conducted to identify households that include a 19- to 35-month-old child. Among households with age-eligible children, sociodemographic information about the children, their mothers, and the household is collected. If consent is obtained from the NIS respondent, age-eligible children's vaccination providers are mailed a questionnaire to obtain the vaccination history. Age-eligible children's provider-reported vaccination histories were used to evaluate the vaccination status of children sampled in the NIS. Between 1995 and 2007, the NIS CASRO (Council of American Survey Research Organizations)13 response rates (response rate number 3; American Association for Public Opinion Research, Lenexa, Kansas) for the random digit dialing portion of the NIS ranged from 65% to 76%, and the percentages of sampled children that had a sufficiently detailed vaccination history returned from vaccination providers to accept as a complete report ranged from 62% to 70%. More detailed descriptions of the sampling design and methods used by the NIS are available.14-17 In all our analyses, we used the survey library18 in the R statistical software package,19 which took into account the NIS sampling weights and design, independence of sampling from year-to-year, and clustering of age-eligible children within households. The use of human subjects in the NIS was last approved by the National Center for Health Statistics on October 14, 2008.
Children belonging to an annual birth cohort are between 19 and 35 months of age in the 3 consecutive years that follow their birth year and were eligible to be sampled by the NIS in those consecutive years. We restricted our analyses to cohorts born between 1994 and 2004 because complete information was obtained from each of them from 3 consecutive annual survey years of the NIS. Because the varicella vaccine had wide geographic distribution after 1996, we tracked progress for that vaccine using cohorts born between 1996 and 2004. In our study, estimated percentages and differences of percentages are reported with their 95% confidence intervals (CIs).
Timely vaccination coverage
The current vaccination schedule20 recommends the administration of 4 doses of DTaP-DTP, 3 doses of polio, 1 dose of MMR, 3 doses of hepatitis B, 3 doses of Hib, and 1 dose of the varicella vaccines by age 19 months. Therefore, we defined timely vaccination coverage as receipt of at least the recommended number of doses of each vaccine by age 19 months.
States receiving federal funds to support Medicaid programs are mandated to serve families that meet a variety of criteria, including annual income 133% or less of the federal poverty level (FPL).21 For this study, we defined a low-income household as one with an annual income of 133% or less of the FPL. Children living in a household with an annual income between 134% and 399% of the FPL and 400% or more of the FPL are referred to as middle-income and high-income, respectively. In our analyses, FPL was determined by comparing sampled households' reported annual income to the current poverty thresholds published by the US Census Bureau.22 In 1997, for a family with 2 adults and 3 children younger than 18 years, the FPL was $24 744. Low-income children had a reported annual household income that was $32 909 or less, and high-income children lived in households with a reported annual income that was more than $98 976. All our analyses accounted for the change in official poverty thresholds over time.22
Progress in timely vaccination coverage
To evaluate progress in timely vaccination coverage across successive birth cohorts, we estimated vaccination rates for every annual birth cohort born between 1994 and 2004 and then used statistical regression analyses to evaluate whether those estimates increased, decreased, or remained the same across consecutive birth cohorts. Bootstrap methods23 were used to account for the fact that the coverage rates used in the regression analyses were estimates. The median of the estimated slopes obtained from 1000 bootstrap replications of the regression analyses provides an estimate of the mean percentage change in timely vaccination coverage between consecutive birth cohorts. The upper and lower 2.5 percentiles calculated from the 1000 bootstrap replicate estimates of the slope provide a 95% CI. There was a statistically significant increasing trend in progress of timely vaccination coverage if the 95% CI contained only positive values and a decreasing trend if the 95% CI contained only negative values.
As a first step in evaluating disparities in timely vaccination coverage between low- and high-income children, we computed the differences in vaccination rates for low- vs high-income children for each annual birth cohort. Differences in estimated percentage values were declared to be statistically significant if their associated z scores had a P value of less than .05.
To evaluate whether there was a widening (or narrowing) disparity in timely vaccination coverage between low- and high-income children across consecutive birth cohorts born between 1994 and 2004, we used statistical regression analyses to evaluate whether the trend in estimated disparities had increased, decreased, or remained the same across consecutive birth cohorts. Bootstrap methods were used to account for the fact that the differences in timely vaccination rates used in those analyses were estimates. The median of the estimated slopes calculated from 1000 bootstrap replicates in this regression analysis provides an estimate of the mean percentage change in the disparities in vaccination rates between consecutive birth cohorts. We declared that there was a statistically significant increasing (decreasing) trend in the disparity if the 95% CI contained only positive (negative) values.
Percentage of low-income children
Among the 6 025 000 children who were aged 19 to 35 months in 2007, 2 519 000 (41.8%; 95% CI, 40.3%-43.3%) lived in a low-income household. The estimated percentage of low-income children aged 19 to 35 months varied widely among states, from 16.3% (95% CI, 9.0%-23.6%) to 56.6% (95% CI, 48.7%-64.5%). Approximately 1 075 000 children living in low-income households were Hispanic (42.7%), 826 000 (32.8%) were non-Hispanic white, 457 000 (18.1%) were non-Hispanic black, 66 000 (2.6%) were Asian, and 95 000 (3.8%) belonged to other racial/ethnic groups, including American Indian/Alaskan Native. Compared with non-Hispanic white children (mean, 25.0%; 95% CI, 23.5%-26.5%), the estimated percentage of low-income children was significantly higher among the Hispanic (68.1%; 95% CI, 65.1%-71.1%), non-Hispanic black (64.0%; 95% CI, 60.3%-67.7%), and other racial/ethnic groups (51.2%; 95% CI, 41.9%-61.5%) (P < .05) but was not significantly different for Asian children (26.6%; 95% CI, 18.2%-35.0%; P = .71).
Timely vaccination coverage
Among low-income children, timely vaccination coverage increased significantly between consecutive birth cohorts from 1994 through 2004 for the DTaP-DTP, polio, MMR, hepatitis B, and varicella vaccines (Table and Figure 1 and Figure 2). However, the mean increase in timely vaccination coverage between consecutive birth cohorts was not statistically significant for the Hib vaccine.
Estimated vaccination coverage by age 19 months for 4 or more doses of the diphtheria, tetanus, pertussis (DTap-DTP) vaccine (A), 3 or more doses of the polio vaccine (B), and 1 or more doses of measles, mumps, rubella (MMR) vaccine (C) by annual birth cohort. *A significantly lower estimated timely vaccination coverage rate for low-income (≤133% federal poverty level [FPL]) or middle-income (134%-399% FPL) children compared with high-income (≥400% FPL) children for the annual birth cohort designated on the x-axis.
Estimated vaccination coverage by age 19 months for 3 or more doses of Haemophilus influenzae type b (Hib) vaccine (A), 3 or more doses of hepatitis B vaccine (B), and 1 or more doses of varicella vaccine (C) administered after 12 months of age. *A significantly lower estimated timely vaccination coverage rate for low-income (≤133% federal poverty level [FPL]) or middle-income (134%-399% FPL) children compared with high-income (≥400% FPL) children for the annual birth cohort designated on the x-axis.
Estimated Average Increase (in Percentage) in Timely Vaccination Coverage Between Consecutive Birth Cohorts Born Between 1994 and 2004 by Vaccine
Low- vs High-Income Children
For every annual birth cohort born between 1994 and 2004, estimated timely vaccination rates for the DTaP-DTP, MMR, Hib, and varicella vaccines were significantly lower among low- vs high-income children (Figures 1 and 2). Also, estimated timely vaccination rates for the polio and hepatitis B vaccines were significantly lower among low- vs high-income children for every birth cohort born between 1994 and 2003. However, timely vaccination rates for high- vs low-income children were not significantly different for the 2004 birth cohort for the polio and hepatitis B vaccines.
Consecutive Birth Cohorts
The disparity in timely vaccination coverage for low- vs high-income children increased significantly between consecutive birth cohorts by approximately 0.4% for the DTaP-DTP vaccine (Figure 3) and decreased significantly between consecutive birth cohorts by approximately −0.3% for MMR, −0.3% for hepatitis B, and −0.5% for varicella vaccines (Figure 3 and Figure 4).
Trends in disparities in timely vaccination coverage for low- vs high-income children for diphtheria, tetanus, pertussis (DTap-DTP) (A), polio (B), and measles, mumps, rubella (MMR) (C) vaccines. Vertical blue lines are 95% confidence intervals (CIs).
Trends in disparities in timely vaccination coverage for low- and high-income children for Haemophilus influenzae type b (Hib) (A), hepatitis B (B), and varicella (C) vaccines. Vertical blue lines are 95% confidence intervals (CIs).
For the polio and Hib vaccines, the disparity between low- vs high-income children did not change significantly across consecutive birth cohorts (Figures 3 and 4). For the birth cohorts born between 1994 and 2004, children in high-income households had a timely polio vaccination rate that was 3.6% (95% CI, 2.9%-4.3%) higher and a timely Hib vaccination rate that was 6.9% (95% CI, 6.4%-7.4%) higher than that of children in low-income households.
Among the problems that contributed to the measles resurgence in the late 1980s and early 1990s was the low level of timely vaccination coverage throughout the population, particularly among low-income children. In our analyses, we found that among low-income children, timely vaccination coverage rates for all vaccines except Hib have increased significantly between consecutive cohorts born after the measles resurgence. However, estimates of timely vaccination coverage for Hib have also been high among low-income children and have exceeded 85% for every birth cohort born between 1994 and 2004.
Also, significant disparities in timely vaccination coverage were found between low- and high-income children for all childhood vaccines and nearly every birth cohort born between 1994 and 2004. However, those disparities have been declining significantly for the MMR, hepatitis B, and varicella vaccines. For the 2004 birth cohort, the disparity in estimated timely vaccination coverage for hepatitis B is not statistically significant, offering hope that progress in closing gaps in timely coverage between low- and high-income children can be achieved. Although the mean disparity in timely coverage did not change significantly for the polio vaccine from 1994 through 2004, the estimated timely coverage rate for low-income children is not significantly different from that of high-income children for the 2004 birth cohort, which provides hope that further progress in disparity reduction may be forthcoming.
Of potential concern is the widening disparity between low- vs high-income children in timely coverage for the DTaP-DTP vaccine. However, our results show that timely coverage for the DTaP-DTP vaccine has increased significantly across consecutive birth cohorts for low-income children. Therefore, the disparity between low- and high-income children for DTaP-DTP is attributable to a more rapid increase in timely coverage for DTaP-DTP among high-income children. Because the fourth dose of DTaP-DTP is recommended to be administered no sooner than age 6 months after the administration of the third dose, further progress in timely coverage of DTaP-DTP by age 19 months may be achieved by ensuring that the third dose is administered no later than age 12 months, and, preferably, soon after it is recommended at age 6 months. The recommended vaccination schedule permits the fourth dose of DTaP-DTP to be administered as early as age 12 months. Arranging the administration of vaccines so that completion of the entire childhood vaccination schedule can be completed by age 12 months may be more convenient and less costly for low-income families. These adjustments might lead to further progress in timely vaccination coverage for low-income children.
Our work has several strengths. First, data from our study were drawn from a large, nationally representative sample of 19- to 35-month-olds that has been conducted annually for the past 14 years using the same methods. Also, estimates of vaccination coverage are based on health care provider–reported vaccination histories that are known to be more accurate than parental reports from memory. These features allow us to obtain reliable estimates of trends in vaccination coverage over time.
However, the findings in this report are subject to potential limitations. Because the NIS is a telephone survey, results are weighted to be representative of all children aged 19 to 35 months in the United States. Separate statistical adjustments have been made to the survey weights to account for noncoverage of households with no telephones,24 nonresponse to the NIS telephone interview,16 nonresponse to the NIS survey mailed to vaccination providers,16 underascertainment of vaccination histories among children with more than 1 vaccination provider,17 and other effects that could bias estimates from the NIS.16 When data used in this study were collected, the percentage of children who lived in a household with no telephone service was approximately 1.5%. However, in recent years, the percentage of households with children that have cellular phone service only has grown from 2.3% in 2004 to 11.6% in 2007.25 Bias in complex sample surveys such as the NIS is equal to the product of the percentage of the target population not covered by the survey multiplied by the difference in vaccination coverage between the portion of the target population that is covered by the survey and the portion that is not covered by the survey.26,27 Because the percentage of children in nontelephone and cellular-only households is moderate (13.1%) and the difference in vaccination coverage between covered and noncovered portions of the target population is expected to be no more than moderate (<10%), the maximum bias in national estimates that can be attributable to noncoverage of nontelephone and cellular-only households is expected to be small (<2%). Recent work suggests that adjustments made to NIS survey weights to reduce bias are effective and that the bias in national-level estimates from the NIS that can be attributed to noncoverage of portions of the population28,29 and nonresponse to the survey mailed to health care providers is small.30-32
An important conclusion reached from the CDC research conducted during the measles resurgence in the United States was that vaccines need to be administered on time because delay of timely vaccine administration represents inadequate protection against vaccine-preventable disease. Subsequent research found that many children in the United States experienced substantial delays in receiving recommended vaccinations and that these delays were especially prevalent among socioeconomically disadvantaged subpopulations.33-37 Other literature has evaluated the association between poverty and low vaccination coverage12,38-40 and found that poverty continues to be a persistent barrier to receiving the recommended number of vaccinations. The results of our study are concordant with those previous findings. However, our study shows that there has been progress in timely vaccination coverage among low-income children for selected vaccines and suggests that further progress is possible.
How could further progress be realized? First, higher levels of timely vaccination coverage and reduction of disparities may be achieved by increasing awareness of the VFC Program. For eligible children, VFC eliminates cost as a barrier to being vaccinated.41 Next, educational interventions that address health concerns and barriers to vaccinations have been strongly associated with increased coverage42 and could be implemented to achieve higher timely coverage levels. Also, reminder/recall systems have been strongly associated with higher vaccination coverage.42 Therefore, higher levels of timely vaccination coverage and reduction of disparities attributable to family income may be achieved by ensuring the effective implementation of these strategies. Furthermore, clinic-based assessment of timely vaccination43 and other systems such as state registries may offer information needed by vaccination providers to administer vaccines in a timely manner. Finally, inadequate provider reimbursement has been identified as a potential barrier to being vaccinated for low-income children eligible for Medicaid and VFC.44 To overcome this barrier, the Vaccine Financing Working Group of the National Vaccine Program Office has recommended that VFC should be expanded to cover vaccine reimbursement for all VFC-entitled children and adolescents; that the maximum allowable Medicaid reimbursement amounts be updated for each state to include all appropriate non–vaccine-related costs; and that the federal match for vaccine administration reimbursement in Medicaid be increased to levels in line with other services of public health importance.44
Correspondence: Philip J. Smith, PhD, National Center for Immunizations and Respiratory Disease, Centers for Disease Control and Prevention, Mail Stop MD E-32, 1600 Clifton Rd NE, Atlanta, GA 30333 (firstname.lastname@example.org).
Accepted for Publication: December 3, 2008.
Author Contributions: Dr Smith had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Smith. Acquisition of data: Smith. Analysis and interpretation of data: Smith, Jain, Stevenson, Männikkö, and Molinari. Drafting of the manuscript: Smith and Jain. Critical revision of the manuscript for important intellectual content: Smith, Stevenson, Männikkö, and Molinari. Statistical analysis: Smith, Stevenson, and Molinari. Obtained funding: Smith. Administrative, technical, and material support: Smith and Männikkö. Study supervision: Smith.
Financial Disclosure: None reported.
Disclaimer: The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.
Centers for Disease Control and Prevention, Measles—United States 1992. MMWR Morb Mortal Wkly Rep
378- 381PubMedGoogle Scholar
LE Measles and measles vaccine. Semin Pediatr Infect Dis
1991;2100- 107Google Scholar
S The resurgence of measles in the United States, 1989-1990. Annu Rev Med
1992;43451- 463PubMedGoogle ScholarCrossref
R Causes of low preschool immunization coverage in the United States. Annu Rev Public Health
1992;13385- 398PubMedGoogle ScholarCrossref
et al. Measles outbreaks among unvaccinated preschool-aged children: opportunities missed by health care providers to administer measles vaccine. Pediatrics
369- 374PubMedGoogle Scholar
SS Epidemiology of measles in the United States in 1989 and 1990. Pediatr Infect Dis J
841- 846PubMedGoogle ScholarCrossref
WA Measles surveillance—United States, 1991. MMWR CDC Surveill Summ
1- 12PubMedGoogle Scholar
The National Vaccine Advisory Committee, The measles epidemic: the problems, barriers, and recommendations. JAMA
1547- 1552PubMedGoogle ScholarCrossref
et al. Using registry data to evaluate the 2004 pneumococcal conjugate vaccine shortage. Am J Prev Med
347- 350PubMedGoogle ScholarCrossref
KM The effect of vaccine shortages on timeliness of pneumococcal conjugate vaccination: results from the 2001-2005 National Immunization Survey. Pediatrics
et al. Overview of the sampling design and statistical methods used in the National Immunization Survey. Am J Prev Med
17- 24PubMedGoogle ScholarCrossref
J Racial/ethnic disparities in vaccination coverage by 19 months of age: an evaluation of the impact of missing data resulting from record scattering. Stat Med
4107- 4118PubMedGoogle ScholarCrossref
DMR Development Core Team, An introduction to R—notes on R: a programming environment for data analysis and graphics, version 2.8.0. http://www.r-project.org
. Published October 20, 2008. Accessed November 29, 2008
Centers for Disease Control and Prevention, Recommended immunization schedules for persons aged 0-18 years—United States, 2007. MMWR Morb Mortal Wkly Rep
Q1- Q4Google Scholar
DR Computing methods for variance estimation in complex surveys. J Off Stat
1985;1323- 329Google Scholar
et al. Adjustments for non-telephone bias in random-digit-dialing surveys. Stat Med
1611- 1626PubMedGoogle ScholarCrossref
WG Sampling Techniques. 2nd ed. New York, NY John Wiley & Sons1963;
DB Statistical Analysis With Missing Data. New York, NY John Wiley & Sons1987;
S An evaluation of methods to compensate for noncoverage of nontelephone households using information on interruptions in telephone service and presence of wireless phones. Proceedings of the Sections on Survey Research Methods 2006 [CD-ROM].
Alexandria, VA American Statistical Association2006;Google Scholar
J An assessment of bias due to noncoverage of wireless only households in RDD surveys. Proceedings of the Sections on Survey Research Methods 2007 [CD-ROM].
Alexandria, VA American Statistical Association2007;Google Scholar
LC Evaluating assumptions of weighting class methods for partial response using a selection model. Stat Med
4569- 4580PubMedGoogle ScholarCrossref
R Alternative methods to compensate for provider nonresponse in the National Immunization Survey. Proceedings of the Sections on Survey Research Methods 2006 [CD-ROM].
Alexandria, VA American Statistical Association2006;2792- 2796Google Scholar
P Evaluation of adjustments for nonresponse bias applied to provider nonresponse in the National Immunization Survey. Proceedings of the Sections on Survey Research Methods 2000.
Alexandria, VA American Statistical Association2000;715- 720Google Scholar
LK Timeliness of childhood vaccinations in the United States: days undervaccinated and number of vaccines delayed. JAMA
1204- 1211PubMedGoogle ScholarCrossref
et al. Assessment of delay in age-appropriate vaccination using survival analysis. Am J Epidemiol
561- 570PubMedGoogle ScholarCrossref
Centers for Disease Control and Prevention, Vaccination coverage by race/ethnicity and poverty level among children aged 19-35 months—United States, 1996. MMWR Morb Mortal Wkly Rep
963- 969PubMedGoogle Scholar
ET US children living in and near poverty: risk of vaccine-preventable diseases. Am J Prev Med
41- 46PubMedGoogle ScholarCrossref
LE Socioeconomic factors and persistent racial disparities in childhood vaccination. Am J Health Behav
434- 445PubMedGoogle ScholarCrossref
et al. the Task Force on Community Preventive Services, Reviews of evidence regarding interventions to improve vaccination coverage in children, adolescents, and adults. Am J Prev Med
97- 140PubMedGoogle ScholarCrossref
Department of Health and Human Services, The National Vaccine Advisory Committee Financing Working Group: recommendations for September 2008 provisional voting. http://www.hhs.gov/nvpo/nvac/reports.html
. Accessed November 29, 2008