OBJECTIVES. This study explored how vaccination coverage is associated with not being insured and with insurance type among children who are insured and to show how these associations are modified by race/ethnicity.
METHODS. We determined whether 8324 children sampled in the National Immunization Survey in 2001 and 2002 were covered by private insurance only, Medicaid/State Children's Health Insurance Program, or another insurance type or were uninsured at the time of the National Immunization Survey interview or were uninsured at some time before the interview. Children were up to date if, by the date of the interview, their vaccination providers had administered ≥4 doses of diphtheria-tetanus toxoids-acellular pertussis vaccine, ≥3 doses of polio vaccine, ≥1 dose of measles-mumps-rubella vaccine, ≥3 doses of Haemophilus influenzae type b vaccine, and ≥3 doses of hepatitis B vaccine. To evaluate the association between insurance type and breaks in insurance with timely completion of the recommended vaccination schedule soon after 19 months of age, we restricted our analyses to children 19 to 24 months of age.
RESULTS. Nationally, 12.6 ± 1.6% of all children 19 to 24 months of age were uninsured at some time. Children who were uninsured at the time of the National Immunization Survey interview had significantly lower vaccination coverage than did children with Medicaid/State Children's Health Insurance Program coverage or children with private insurance only (52.6% vs 70.0% and 75.6%). Children who had never been insured and children who were insured but had a break in insurance coverage in the 12 months immediately preceding the National Immunization Survey interview had significantly lower vaccination coverage than did children who had been insured continuously (47.4% and 64.8% vs 73.5%).
CONCLUSIONS. Approximately 1 of 8 children were uninsured at some time, and those children were at greater risk of not being vaccinated on time as recommended.
Cost has always been a barrier to receiving and/or providing timely preventive medical care to children living in poverty. Under the Social Security Amendments of 1965, the Medicaid program was established as a federally funded and state-administered health insurance program for people living in poverty. Current federal rules specify that state Medicaid programs must cover children <6 years of age from families with incomes ≤133% of the federal poverty level. In 1997, Congress enacted the State Children's Health Insurance Program (SCHIP), which afforded states the opportunity to expand insurance coverage to children from families with incomes up to 200% of the federal poverty level.
Among infants, a desirable consequence of being covered by any type of health insurance is the receipt of vaccinations according to a recommended schedule1 that completes the administration of vaccines by 19 months of age. Among children who had become 19 months of age recently, we explored how vaccination coverage was associated with being covered by private health insurance, Medicaid, or SCHIP or being uninsured and with breaks in insurance coverage and how these associations were modified by a child's race/ethnicity.
National Immunization Survey
In our analyses, we used data collected between the third quarter of 2001 and the fourth quarter of 2002 in the National Immunization Survey (NIS). The NIS is a telephone survey conducted by the Centers for Disease Control and Prevention for the purpose of monitoring vaccination coverage rates among children 19 to 35 months of age in the 50 states, the District of Columbia, and other large metropolitan areas.
Data collection in the NIS occurs in 2 phases, namely, a list-assisted, random-digit-dialing (RDD), telephone survey of households with children 19 to 35 months of age at the time of the telephone interview, followed by a vaccination provider record check (PRC). Cellular telephone numbers are not included in the list of telephone numbers from which the RDD telephone survey sample is obtained. When a household with an age-eligible child is identified in the RDD phase of the survey, the person in the household who is most knowledgeable about the child's vaccination history is asked to serve as the NIS respondent. The RDD portion of the NIS is then conducted, collecting demographic information about each age-eligible child in the household, demographic information about the age-eligible child's mother, and sociodemographic information about the household. At the end of the RDD telephone interview, consent to contact the age-eligible children's vaccination providers is requested. If consent is given, then the PRC phase of the NIS is conducted. In the PRC phase, all of the vaccination providers named by the NIS respondent are contacted by mail, to obtain the provider-reported vaccination histories for the household's age-eligible children. In the NIS, vaccination histories reported by sampled children's vaccination providers are used to determine whether the children had received at least the recommended number of doses of each recommended vaccine1 by the date of the NIS telephone interview.
The goal of our analyses was to evaluate the extent to which insurance history, insurance type, and breaks in health insurance coverage in the 12 months preceding the NIS telephone interview were associated with the receipt of vaccinations according to a recommended schedule1 that completes the administration of all recommended vaccines by 19 months of age. Because the NIS is a survey of children who are 19 to 35 months of age as of the date of the NIS interview, a break in health insurance coverage in the 12 months preceding the NIS telephone interview might have occurred long after 19 months of age for the older children sampled by the NIS. Therefore, to evaluate the extent to which insurance history, insurance type, and breaks in health insurance coverage were associated with timely completion of the vaccination schedule at ∼19 months of age, we restricted the analyses to 8324 children who were 19 to 24 months of age at the time of the NIS telephone interview. For the race/ethnicity comparisons, however, this sample size was too small to yield reliable estimates of vaccination coverage, especially among uninsured children. Therefore, for those analyses, we included all 23636 children 19 to 35 months of age between the third quarter of 2001 and the fourth quarter of 2002.
The response rate of the NIS is the product of 3 rates, namely, (1) the estimated percentage of households that reported having a 19- to 35-month-old child among those that actually had a 19- to 35-month-old child, (2) the Council of American Survey Research Organizations rate2 for the RDD portion of the NIS, and (3) the proportion of children for whom a provider-reported vaccination history is obtained that is sufficiently detailed to be accepted as a complete report. In 2001 and 2002, among households with children 19 to 35 months of age, the percentage with a wire-to-the-wall telephone that could be sampled by the NIS was ∼92%. For those years, the estimated percentages of households that reported having a 19- to 35-month-old child among those that actually had a 19- to 35-month-old child were 74.2% and 69.5%, respectively, and the Council of American Survey Research Organizations rates were 76.1% and 74.2%. Among children with completed NIS RDD telephone interviews, the percentages with sufficiently detailed, provider-reported, vaccination histories obtained in the NIS PRC were 67.8% and 65.0% in 2001 and 2002, respectively.
Because all of the children in our study were born in 1999 or thereafter, we evaluated the vaccination status of the children according to the 1999 recommended vaccination schedule.1 Sampled children were determined to be 4:3:1:3:3 up to date if their vaccination providers reported administering ≥4 doses of diphtheria-tetanus toxoids-acellular pertussis vaccine, ≥3 doses of polio vaccine, ≥1 dose of measles-mumps-rubella vaccine, ≥3 doses of Haemophilus influenzae type b vaccine, and ≥3 doses of hepatitis B vaccine by the date of the NIS RDD telephone interview.
Insurance status was collected in the health insurance module of the NIS.3,4 For our analyses, we defined a sampled child to be currently insured if, as of the date of the NIS RDD telephone interview, the NIS respondent reported that the child was covered by private health insurance, Medicaid, SCHIP, military health care, the title V program, the Indian Health Service, or any other health insurance that paid for both doctor visits and hospital stays. Among children who were currently insured, we defined a child to be continuously insured if the respondent reported that the child had continuous health insurance coverage for the 12 months immediately preceding the NIS RDD telephone interview.
Among children who were currently insured, we defined a child as having experienced a break in health insurance coverage if there was a time in the 12 months immediately preceding the NIS RDD telephone interview when the child was not covered by any health insurance. Children who were not currently insured were defined as being currently uninsured. Among children who were currently uninsured, those who had been uninsured since birth were referred to as continuously uninsured since birth. Children who were currently uninsured but had been insured previously were referred to as being previously insured. Table 1 lists the insurance histories discussed in this article.
Insurance Type and Eligibility for Public Insurance
Sampled children were defined as having private health insurance only if they were covered by private health insurance and no other insurance type and as being covered by Medicaid or SCHIP if they were covered by Medicaid or SCHIP. We defined children as being covered by other health insurance if they were covered by health insurance but that health insurance was not private health insurance only and did not include Medicaid or SCHIP. A sampled child was defined to be eligible for Medicaid or SCHIP if the reported annual income of the child's household was less than the income cutoff that defined eligibility for the Medicaid or SCHIP program in 2000 in the state in which the child lived. Table 2 lists the insurance types discussed in this article.
In our article, estimated coverage rates are provided with the half-widths of 95% confidence intervals (CIs). To evaluate the statistical significance of the estimated differences in 4:3:1:3:3 coverage rates and differences in other estimated percentages and means, we evaluated the 95% CI for each estimated difference. If the 95% CI did not include 0, then we declared the difference to be statistically significant. To evaluate whether the 4:3:1:3:3 coverage rates differed between children who were covered by private insurance only and children who were covered by Medicaid or SCHIP who had continuous insurance coverage for at least the past 12 months, the distributions of selected sociodemographic factors were compared for these 2 groups. Factors whose distributions differed significantly between the 2 groups were noted. These factors were included as covariates for adjustment in a multivariate regression analysis of 4:3:1:3:3 up-to-date status and whether a child was covered by private insurance only or was covered by Medicaid or SCHIP continuously for at least the past 12 months. The coefficient associated with being covered by Medicaid or SCHIP from the multivariate regression analysis summarized the estimated difference in vaccination coverage between the 2 groups, after accounting for the significant differences in the distributions of sociodemographic factors between the groups. We referred to the coefficient as the adjusted difference and declared this difference to be statistically significant when the 95% CI of the adjusted difference did not contain 0. All analyses were conducted with the statistical software package SUDAAN,5 a software package that allows the sampling weights and complex survey design of the NIS to be taken into account in statistical analyses. NIS survey weights used in our analyses account for the probability of a telephone number being sampled by the RDD survey. Adjustments to these survey weights used in our analyses include adjustments for households with multiple nonbusiness telephones, households that do not have a noncellular “wire-to-the-wall” telephone reachable by the NIS, nonresponse to the NIS RDD survey, failure to obtain a provider-reported vaccination history in the NIS PRC for children in sampled households that completed the NIS RDD telephone interview, and failure to ascertain completely the provider-reported vaccination histories for those for whom a provider-reported vaccination history was obtained in the NIS PRC.4 Zell et al6 and Smith et al7 summarized the design of the NIS, Smith et al4,8 provided a detailed description of the statistical methods used in the NIS, and Smith et al3,4 provided additional details about the purpose and design of the NIS health insurance module. The NIS was reviewed and approved by an institutional review board at the Centers for Disease Control and Prevention in 2001.
Association Between Vaccination Coverage and Recentness of Being Uninsured
Among children 19 to 24 months of age in 2001 and 2002, 12.6 ± 1.6% had been uninsured at some time, that is, they were either currently uninsured or had a break in health insurance coverage. The remaining 87.4 ± 1.6% had continuous health insurance (Table 1). Among the 50 states and the District of Columbia, the estimated percentages of children who had been uninsured at some time ranged from 3.5 ± 3.4% to 29.7 ± 10.2%. Children who were continuously uninsured since birth, children who were currently uninsured but previously insured, and children who were currently insured but had experienced a break in insurance coverage had significantly lower 4:3:1:3:3 vaccination coverage rates than did children who were continuously insured (Table 1). Moreover, children who had the longest and most current history of being uninsured were least likely to be 4:3:1:3:3 up to date soon after 19 months of age, when the recommended vaccination schedule should have been completed (Table 1).
Associations Between Vaccination Coverage and Insurance Type
In 2001 and 2002, 5.9 ± 1.4% of children were uninsured at the time of the interview, 43.2 ± 1.8% had private health insurance only, 36.6 ± 1.8% were insured by Medicaid or SCHIP, and 14.3 ± 1.2% were covered by other insurance types (Table 2). Children who were currently uninsured at the time of the NIS telephone interview were significantly less likely to be 4:3:1:3:3 up to date soon after 19 months of age, when the recommended vaccination schedule should have been completed, compared with children who were covered by private insurance only or by Medicaid or SCHIP (Table 2).
Association Between Breaks in Insurance and Vaccination Coverage
Compared with children who had continuous health insurance coverage, those who had a break in health insurance coverage had significantly lower 4:3:1:3:3 vaccination coverage; this was true both for children covered by private health insurance only and for children covered by Medicaid or SCHIP (Table 3). Children who were covered by Medicaid or SCHIP were significantly more likely to have experienced a break in insurance coverage, compared with those who were covered by private insurance only. Furthermore, among children who had a break in insurance coverage, the number of months of being uninsured was significantly greater for children who were covered by Medicaid or SCHIP, compared with those with private insurance only.
Although the 4:3:1:3:3 vaccination coverage rate among children who were covered by private insurance only was greater than that among children who were covered by Medicaid or SCHIP and had continuous insurance coverage (75.6% vs 70.9%; estimated difference: 4.7%; 95% CI: 1.0–7.1%), these groups were different with respect to important sociodemographic factors. Specifically, compared with children covered by private health insurance only, children covered by Medicaid or SCHIP continuously were significantly more likely to be Hispanic or non-Hispanic black than non-Hispanic white; more likely to have a mother who was not married, who spoke Spanish, or who was younger; less likely to have a mother who had a college education; more likely to live in a household with an annual income of less than $75000; more likely to live in a household with ≥4 children than in a household with only 1 child; more likely to live in a central city metropolitan statistical area or in a non-metropolitan statistical area than in a noncentral city metropolitan statistical area; and more likely to have received all vaccine doses from a public provider than a private provider. However, after accounting for these differences, the 4:3:1:3:3 vaccination coverage rate of children covered by Medicaid or SCHIP who had continuous insurance coverage was not significantly different from the coverage rate of children covered by private insurance only (adjusted estimated difference: 0.0%; 95% CI: −0.6% to 0.6%).
Association Between Children's Race/Ethnicity and 4:3:1:3:3 Coverage According to Insurance Type Among Children 19 to 35 Months of Age
Among children who were uninsured at the time of the NIS interview, children were significantly more likely to be Hispanic (11.6 ± 2.0%) than non-Hispanic white (3.3 ± 0.6%) or non-Hispanic black (4.8 ± 2.2%). Among children who were insured by Medicaid or SCHIP, children were significantly more likely to be Hispanic (54.3 ± 2.6%), non-Hispanic black (59.7 ± 3.1%), or American Indian (58.0 ± 11.2%) than non-Hispanic white (23.1 ± 1.2%). Among children who were insured with private health insurance only, children were significantly more likely to be non-Hispanic white (56.5 ± 1.3%) than Hispanic (23.5 ± 2.1%) or non-Hispanic black (24.3 ± 2.4%).
For each insurance type, the 4:3:1:3:3 vaccination coverage rates for Hispanic and Asian children were not significantly different from those for non-Hispanic white children (Table 4). Also, the vaccination coverage rate for American Indian children was not significantly different from that of non-Hispanic white children except among those insured by Medicaid or SCHIP. Although there were significant disparities in the estimated vaccination coverage rates between non-Hispanic black children and non-Hispanic white children for each insurance type, the disparity was significantly lower for children insured by Medicaid or SCHIP than for uninsured children (5.4% vs 24.0%; P = .05).
Our research showed that US children 19 to 24 months of age who had been covered continuously by private or public health insurance were significantly more likely than uninsured children to have been administered all of the recommended doses in the 4:3:1:3:3 vaccination series soon after the age at which this series is recommended to be completed. The positive effect of continuous insurance on health care access is not limited to immunizations; for example, Lave et al9 and Olson et al10 reported that children with continuous insurance coverage were significantly less likely to have an unmet health care need, compared with children who had a break in insurance coverage or were uninsured. Because underimmunization has been found to be a marker for insufficient use of preventive and primary care,11 monitoring of disparities in vaccination rates is critical.
Not only did children with continuous public health insurance coverage have immunization coverage levels nearly equal to levels found among children with private insurance, but also a large disparity in coverage between non-Hispanic white and non-Hispanic black uninsured children was attenuated notably as a result of coverage by public health insurance. State programs that increase enrollment and retention of eligible children in public insurance programs have proved helpful in the past. In 3 successive surveys of state Medicaid and SCHIP officials, conducted between 2000 and 2004, Ross and Cox12–14 documented changes in enrollment processes that had both positive and negative effects. In 2000, state officials identified a number of important barriers to Medicaid or SCHIP enrollment, including public lack of information about the programs and confusion about eligibility. By 2002, a number of states had made changes that decreased enrollment barriers, by increasing community-based outreach to enroll uninsured families, shortening applications for public health insurance, allowing mailed rather than in-person applications, and guaranteeing 12 months of public health insurance coverage for children regardless of changes in family circumstances. However, in the 2004 survey, some states reported that cutbacks in health care spending were affecting outreach efforts. Furthermore, some of these cutbacks resulted in changes that affected enrollment and retention negatively, including increases in required insurance premiums, additional recertification requirements, and enrollment freezes. The impact of increasing insurance coverage varies widely by state, given the tremendous state-to-state variations in the percentages of children who are uninsured and the particular barriers they face. However, as our results demonstrate for childhood vaccinations, continuous insurance coverage, whether private or public, ensures better access to health services overall.
The strengths of this study include the use of data from a survey with large samples from each state. A limitation of our study is that sampled children's insurance histories were ascertained at the time of the NIS RDD interview and not continuously between the child's birth and 19 to 24 months of age, the period in which children are recommended to receive their vaccinations. Insurance history during this period might be measured with error,15 and gaps in vaccination coverage could be underestimated. As a result, the impact of having insurance and being continuously insured may be more beneficial than we reported here. Also, data from the NIS are not able to indicate accurately whether the private insurance that respondents reported covered all costs of childhood vaccines. Children covered by private health insurance that does not cover all costs may not be as well vaccinated as children covered by other types of private insurance that cover all costs of childhood vaccines. Lacking the ability to distinguish between these 2 types of private insurance, we categorized children who were covered by private insurance into only 1 insurance category.
Another limitation is the somewhat low response rate of the NIS. However, our estimates of the percentages of children with each of the different insurance histories (Table 1) agree closely with the percentages reported by Olson et al16 in their analysis of National Health Interview Survey data collected in 2000 and 2001. This is reassuring, because the National Health Interview Survey is a nationally representative survey, with face-to-face interviews, conducted by the National Center for Health Statistics, Centers for Disease Control and Prevention,17 with household and adult response rates of 88.9% and 90.0%, respectively, in 2000.
Our results showed that ∼1 of 8 children were uninsured at some time and these children were at greater risk of not being vaccinated on time as recommended. All of these children were eligible to receive publicly purchased vaccines at no cost, through the Vaccines for Children program. The low vaccination coverage rates for these children underscore the need to integrate uninsured children into the health care system more effectively. Also, our results suggest that efforts to enroll children and to maintain continuous enrollment in either public or private health insurance may result in increased vaccination coverage rates at 19 months of age and may decrease disparities in vaccination coverage rates experienced currently by children at greatest risk of low vaccination coverage. Among those who are privately insured, vaccination coverage rates could be increased if all health insurance policies were mandated to cover the costs of all vaccines recommended by the Advisory Committee on Immunization Practices.18 For children who are not eligible for the Vaccines for Children program and are underinsured or whose parents are working poor who cannot afford the high deductibles required to vaccinate their children fully, financial barriers to being fully vaccinated might be reduced by replacing the current discretionary section 317 grants with a more stable approach to providing these children with all vaccines recommended by the Advisory Committee on Immunization Practices.19
- Accepted December 21, 2005.
- Address correspondence to Philip J. Smith, PhD, National Immunization Program, Centers for Disease Control and Prevention, MS E-32, 1600 Clifton Rd, NE, Atlanta, GA 30333. E-mail:
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.
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵American Association for Public Opinion Research. Standard Definitions. Ann Arbor, MI: American Association for Public Opinion Research; 1998. Available at: www.aapor.org/pdfs/standarddefs_3.1.pdf. Accessed December 5, 2005
- ↵Smith PJ, Simpson D, Battaglia, MP, et al. Split sampling design for topical modules in the National Immunization Survey. In: 2000 Proceedings of the Section on Survey Research Methods. Alexandria, VA: American Statistical Association; 2000:653– 658. Available at: www.amstat.org/sections/srms/Proceedings/papers/2000_109.pdf. Accessed August 15, 2005
- ↵Smith PJ, Hoaglin DC, Battaglia MP, et al. Statistical methodology of the National Immunization Survey, 1994–2002. Vital Health Stat 2.2005;(138):1– 55. Available at: www.cdc.gov/nchs/data/series/sr_02/sr02_138.pdf. Accessed August 15, 2005
- ↵Research Triangle Institute. SUDAAN User's Manual, Release 8.0. Research Triangle Park, NC: Research Triangle Institute; 2002
- ↵Smith PJ, Battaglia MP, Huggins VJ, et al. Overview of the sampling design and statistical methods used in the National Immunization Survey. Am J Prev Med.2001;20(suppl) :17– 24
- ↵Smith PJ, Rao JNK, Battaglia MP, et al. Compensating for provider nonresponse using response propensities to form adjustment cells: the National Immunization Survey. Vital Health Stat 2.2001;(133):1– 17. Available at: www.cdc.gov/nchs/data/series/sr_02/sr02_133.pdf. Accessed August 15, 2005
- ↵Ross DC, Cox L. Making It Simple: Medicaid for Children and CHIP Income Eligibility Guidelines and Enrollment Procedures. Washington, DC: The Kaiser Commission on Medicaid and the Uninsured, Center on Budget and Policy Priorities; 2000. Available at: www.kff.org/medicaid/loader.cfm?url=/commonspot/security/getfile.cfm&PageID=13443. Accessed July 8, 2004
- Ross DC, Cox L. Enrolling Children and Families: The Promise of Doing More. Washington, DC: The Kaiser Commission on Medicaid and the Uninsured, Center on Budget and Policy Priorities; 2002. Available at: www.kff.org/medicaid/loader.cfm?url=/commonspot/security/getfile.cfm&PageID=14125. Accessed July 8, 2004
- ↵Ross DC, Cox L. Beneath the Surface: Barriers Threaten to Slow Progress on Expanding Health Coverage of Children and Families. Washington, DC: The Kaiser Commission on Medicaid and the Uninsured, Center on Budget and Policy Priorities; 2004. Available at: http://coveringkidsandfamilies.org/resources/docs/policygroup/PolicyReport.pdf. Accessed July 8, 2004
- ↵Biemer PP, Groves RM, Lyberg LE, et al. Measurement Errors in Surveys. New York, NY: Wiley; 2004
- ↵Gentleman JF, Pleis JR. The National Health Interview Survey: an overview. Eff Clin Pract.2002;5(suppl) :e2 . Available at: www.acponline.org/journals/ecp/mayjun02/gentleman.htm. Accessed August 12, 2005
- Copyright © 2006 by the American Academy of Pediatrics