Objective. To evaluate personal, financial, and structural barriers to vaccination in socioeconomically disadvantaged urban children in the first 2 years of life.
Design. Prospective cohort study.
Setting. A large municipal teaching hospital in the Midwest.
Participants. Healthy term newborns discharged to the care of their mothers. Mothers were interviewed 24 to 72 hours postpartum regarding personal and financial barriers, and 2 years later regarding personal, financial, and structural barriers to care.
Main Outcome Measure. Vaccination status at age 2 years.
Results. Of 399 children with documented vaccination status, 47% had not received all recommended vaccinations by 2 years of age. After adjusting for mother's age, race, and education, mothers who were unmarried (adjusted odds ratio [AOR] 1.74; 95% confidence interval [CI]: 1.05, 2.90), multiparous (AOR 2.10; 95% CI: 1.26, 3.52), not coresident with the child's grandmother (AOR 1.75; 95% CI: 1.01, 3.03), had not received adequate prenatal care (AOR 1.78; 95% CI: 1.12, 2.84), or lived in poverty (AOR 2.62; 95% CI: 1.44, 4.75) were more likely to have undervaccinated children, as were mothers who perceived less satisfaction with their child's health care (AOR 1.63; 95% CI: 1.01, 2.61), less control over their lives (AOR 2.01; 95% CI: 1.03, 3.94), or more benefit of medical care to prevent vaccine-related diseases (AOR 1.76; 95% CI: 1.25, 2.48).
Conclusions. Family environment, a mother's history of prenatal care use, and financial barriers are important factors related to vaccination receipt among socioeconomically disadvantaged children at age 2 years. These factors, however, do not fully explain the variation in vaccination status.
- RMRS =
- Regenstrief Medical Record System •
- NIS =
- National Immunization Survey
It is widely agreed that childhood vaccinations are the most cost-effective way to prevent certain infectious diseases in children.1 Estimates indicate that the prevalence of adequate vaccination has increased recently, with 75% of 19- to 35-month-olds up-to-date in 1994.2 This level, however, is still well below the Healthy People 2000 objective for 90% completion of the primary vaccination series among the nation's 2-year-olds.3 Moreover, use rates of pediatric preventive health services in general, and vaccinations in particular, are significantly lower in socioeconomically disadvantaged children, despite their poorer health status.4 Indeed, one of the most important problems in delivering high-quality health care to socioeconomically disadvantaged children is that their parents are more likely to bring them in for sick care than for preventive services. Although Medicaid has lessened some of these differences, it has not eliminated them.5
Many studies have examined the sociodemographic risk factors and personal barriers to the use of preventive services. Older children, those further along in the birth order, those from large families, and those whose mothers are nonwhite and poorly educated receive fewer health care services.6 7 Mothers' health beliefs also are important personal barriers to use of preventive services. Financial barriers, such as low income and lack of insurance, and structural barriers such as lack of a routine source of health care or decreased availability of physician services, are associated with use rates.5 8-13 However, the relative contribution of these three types of barriers to undervaccination has never been prospectively evaluated, despite having been identified by the Institute of Medicine as an important area for research.1The purpose of this study is threefold. First, we prospectively evaluate the relationship between personal, financial, and structural barriers to care and undervaccination at age 2 years in socioeconomically disadvantaged urban children. Second, we further explore the roles of perceived barriers to care and social support in undervaccination at age 2 years. Third, we compare the risk factors for undervaccination at age 2 years with those identified previously at ages 3 and 7 months.14
We have shown previously that lack of family support, perceived barriers to receipt of well-child care and vaccinations, lack of adequate prenatal care, and living in poverty were associated with undervaccination at ages 3 and 7 months in a cohort of 500 socioeconomically disadvantaged infants and mothers.14 We now evaluate receipt of vaccination by these children at age 2 years, which is widely considered the most important juncture at which to judge the adequacy of vaccination delivery.
Study Sample and Design
Mothers were recruited to participate in the study from the postpartum unit of a large municipal teaching hospital. Enrollment of the mothers has been described previously.14 Women delivering infants between April 4, 1992, and September 15, 1992, were eligible for inclusion (N = 1242). Mothers were excluded if they were not 24 to 72 hours postpartum on a Monday through Friday (n = 294), if their infants were admitted to the intensive care nursery (n = 204), if they were non-English-speaking (n = 10), if they were placing their infants for adoption (n = 14), or if they were residents of a penal institution (n= 9). One of each pair of eligible twins was included (n = 3). Of the 708 mothers eligible to participate, 500 (71%) gave informed consent and completed a face-to-face baseline interview 24 to 72 hours postpartum. Data from the Regenstrief Medical Record System (RMRS), a computerized medical record,15 indicate that mothers who consented to participate were more likely to be white (49% vs 38%,P < .001) and have higher educational attainment (11.2 years vs 10.7 years, P < .05) than those who refused. No other significant differences were observed between respondents and nonrespondents to the baseline interview.
The baseline interviews, conducted by a trained interviewer, included measures of sociodemographics, personal variables, health beliefs, and financial variables. We also asked mothers where they intended to take their infants for well-child care. Follow-up telephone interviews were conducted at 9 to 12 and 24 to 30 months after birth. Our previous experience with this extremely hard to reach population indicated that telephone interviews resulted in a much higher response rate than mail interviews. Mothers who could not be reached after at least 10 attempts at telephone contact were mailed the survey instrument. The results of the first follow-up data collection have been reported previously.14 Data collected at the second follow-up included updated financial variables, additional measures of personal variables, and health care system variables. We also asked whether the child was still living with the mother, and the dates and sites of vaccination receipt. This study was approved by the institutional review board of the Indiana University School of Medicine.
The main outcome variable is vaccination status at age 2 years. Consistent with the Centers for Disease Control and Prevention recommendations,16 children were considered up to date at age 2 years if they had received at least four doses of diphtheria and tetanus toxoids and pertussis vaccine, three doses of poliovirus vaccine, and one dose of measles–mumps–rubella vaccine (4:3:1 series). Vaccination dates were obtained by medical record review for all children. To facilitate that review, we asked mothers as part of each interview whether, where, and when their children had received vaccinations. We reviewed three sources of medical records for each child. The county public health vaccination registry contains the dates of all vaccinations received at any county public health delivery site. The RMRS is a computerized medical record linking the hospital of delivery, its large pediatric clinic, and dozens of affiliated community health care sites. When the RMRS indicated incomplete vaccination status, we reviewed the paper chart for additional available vaccination data. We also reviewed private physician records for children whose mothers identified private sources of vaccination.
Of the 500 children participating in the study, 7 were excluded at the time of follow-up because 2 had died, 2 were no longer living with their mother, and 3 had moved out of state (Figure). We were able to reinterview 290 of the remaining 493 mothers 24 to 30 months after birth. We documented the vaccination status for 269 of these 290 children by review of medical records. Of these, 31% had dates recorded in the county registry, 74% had dates recorded in the RMRS, and 16% had dates obtained from private physician records. For 21 children, private records were not available to review, and those children were excluded from further analysis. For the 203 children whose mothers could not be reached for reinterview, we considered vaccination status complete if we had recorded receipt of all recommended vaccinations by age 2 years. If records were available that documented incomplete vaccination status or receipt of vaccinations (such as a measles–mumps–rubella vaccine or fourth diphtheria and tetanus toxoids and pertussis vaccine) at or after age 2 years, then vaccination status was considered incomplete. Using these criteria, we documented vaccination status for 130 of the 203 children whose mothers could not be reached for reinterview. Of these, 31% had dates recorded in the county registry, 66% had dates recorded in the RMRS, and 5% had dates obtained from private physician records. We could not document vaccination status for 73 children who were excluded from further analysis. Therefore, of the 493 eligible children, we documented vaccination status for 399, or 81%.
Sociodemographic data obtained from all mothers at the baseline interview included mother's age, race, and education. Personal variables included marital status, household composition, parity, employment, and ownership of a car or telephone. Health beliefs were assessed using the Maternal Health Belief Questionnaire,17 which is based on the Health Belief Model.18 Eight scales measure mothers' perceptions of severity of childhood illnesses, susceptibility to disease, susceptibility to symptoms, benefit of medical care to prevent disease, benefit of medical care to prevent symptoms, logistical barriers, cost barriers, and general health motivation. The mother's locus of control toward the child's health was assessed using the seven-item externality scale from the Parent Health Belief Scales.19Data on the use of prenatal care were taken from the RMRS, and were classified as adequate, intermediate, or inadequate based on the number and timing of prenatal visits using the Kessner Index.20(Mothers with adequate prenatal care were compared with those with intermediate or inadequate prenatal care.) Financial variables included insurance coverage and income. The child's insurance status was obtained from the RMRS and classified as Medicaid, private insurance, or uninsured. Poverty was assessed by comparing reported income and family size with the 1992 US Bureau of the Census guidelines.21 Families of two persons with household incomes <$10 000 and families of three or more persons with household incomes <$15 000 were considered impoverished. Because 10% of the mothers did not report income, poverty status was classified as poverty, nonpoverty, or missing. Dummy variables were constructed to compare the poverty and missing data groups with the nonpoverty group.
Personal variables assessed at the 2-year telephone follow-up included current marital status, employment, car ownership, and coresidence with the child's father or grandmother. We also assessed the perceived health status of the child, the mother's perceived social support,22-24 satisfaction with the child's health care,24 and whether the mother had had subsequent children. The Maternal Health Belief Questionnaire was readministered at follow-up. Locus of control was assessed using a balanced measure of control over one's life.25 Financial variables included perceived adequacy of income and the child's current health insurance. Reported health care system variables included accessibility of health care as measured by having a usual source of routine health care for well and sick visits, travel time, office wait time, site of routine health care, and continuity of physician care, measured by whether the child was usually examined by the same physician.
χ2, Mantel–Haenzel, and t tests were used to examine the univariate relationships between vaccination status and the personal, financial, and structural variables. To assess whether there were any differences between the 94 children who were excluded because we could not document their vaccination status and the 399 children included in the study, all independent variables were compared between the two groups. No significant differences were found. Next, undocumented vaccination status was regressed (using multivariable logistic regression) against all of the baseline variables. That model had a very low χ2 (χ2= 14.05; P = .21), indicating that undocumented vaccination status was not explained by the variables under investigation. Finally, a dichotomous marker was added to the final model to indicate whether the mother had been contacted in follow-up; this, too, was not significant. Therefore, our exclusion of those with undocumented vaccination status does not introduce detectable selection bias into the analysis.
The analysis was conducted in three phases. First, analyses were performed using data obtained from the baseline (postpartum) interview. Logistic regression was used to examine the independent effects of each variable on vaccination status. Variables bivariately significant atP < .05 were entered into the multivariable model. Mother's age, race, and education were controlled in all multivariable models. Multicollinearity was evaluated using standard procedures. All clinically meaningful interactions were assessed between all variables in the model. The second phase of the analysis added variables related to undervaccination to the model that were measured at the 2-year follow-up interview, including perceived social support, satisfaction with care, and health care system factors. These analyses were conducted only on the 269 children with documented vaccination status whose mothers were interviewed at the 2-year follow-up. In the final phase of the analysis, the relationship between failure to initiate the vaccination series by age 3 months, a significant predictor of incomplete vaccination status at age 2 years,6 26 was explored. This involved repeating the analysis from the first phase to determine which factors were predictive of undervaccination at age 2 years among those children who had not initiated by age 3 months.
Description of the Sample
Of the mothers included in the study, 49% were were white and 49% were black, 71% were unmarried, 64% were multiparous, 47% had less than 12 years of education, and 72% had total household incomes less than the federal poverty level. Although 72% of the mothers were unmarried, 48% were living with the child's father and 26% were living with the child's grandmother at the time of delivery. The mothers' average age at the time of delivery was ∼23 years old.
Forty-seven percent (186) of the 399 children for whom vaccination status was documented were not up to date on vaccinations at age 2 years. If the children who were excluded because we could not document their vaccination status were assumed to be up to date, the incomplete vaccination rate would fall to 38% (186/493). If these same children were assumed to be not up to date, the incomplete vaccination rate would rise to 57% (280/493). Thus, the overall rate of not completing the 4:3:1 vaccination series falls in the range of 38% to 57%.
Bivariate Results: Baseline Data
Mothers who were unmarried, multiparous, not living with the child's grandmother, and did not have adequate prenatal care were more likely to have children who were not up to date at age 2 years. Maternal age, race, coresidence with the child's father at baseline, education, employment, and possession of a car were not related to vaccination status (Table 1). No baseline health belief measures were significantly related to vaccination status at age 2 years.
Mothers living in poverty or with missing income information at baseline were more likely to have children who were not up to date at age 2 years than those living above the poverty line. Insurance coverage was not related to vaccination status, although very few children had private insurance (Table 1).
Bivariate Results: Follow-up Data
Mothers' marital status, parity, employment, and coresidence status at follow-up were not related to vaccination status nor was possession of a car, birth of subsequent children, or the perceived health status of the child (Table 2). Satisfaction with health care was a significant predictor, in that mothers who were less satisfied with their children's health care were more likely to have children who were not up to date. Mothers who had scored lower on perceived control or higher on perceived benefit of medical care to prevent vaccine-preventable diseases also were more likely to have children who were not up to date (Table3).
Perceived comfort with income and insurance coverage at follow-up were not related to vaccination status. Again, few children had private insurance (Table 2).
Having a source of routine health care, travel time, and office wait time were not associated with completion of vaccinations. However, children of mothers who reported at follow-up that their child's primary health care provider was a private physician were less likely not to be up to date. Children who were usually examined by the same physician also were less likely not to be up to date (Table 2).
Multivariable Results: Baseline Data
The results of the multivariable model of baseline risk factors for undervaccination are shown in column 1 of Table4 for the 399 children with documented vaccination status. All of the risk factors that met the bivariate screening criterion were significant independent predictors of vaccination status at age 2 years. Mothers who were multiparous, had not received adequate prenatal care, were not coresident with the child's grandmother, or were unmarried were more likely to have children who were not up to date. Mothers living in poverty and those with missing income information were more likely to have children not up to date than those living above the poverty line. No significant multicollinearity or interactions were identified.
Multivariable Analyses: Follow-up Data
Analyses of follow-up data were performed two ways. First, we replicated the model including baseline data on the subset of children whose mothers had been reinterviewed. The resultant significant variables and odds ratios (OR) were not materially different from the results shown in Table 4, column 1 (data not shown). Next, all significant risk factors from the baseline and follow-up interviews were included, controlling for the mother's age, race, and education. Results are shown in Table 4, column 2. Mothers who were multiparous, not living with the child's grandmother, perceived less personal control, perceived more benefit of medical care for disease, or were less satisfied with their child's health care were more likely to have children who were not up to date. In this model, adequacy of prenatal care, marital status, poverty, continuity of physician, and private source of routine care were no longer significant. Once again, all clinically relevant interactions were tested. The interaction between race and satisfaction with care was significant (data not shown), indicating that only for whites was satisfaction with care a significant risk factor for vaccination receipt. For whites, each point scored lower on the scale of satisfaction with care is associated with a threefold increase in the odds of not being up to date. No significant multicollinearity was identified.
Failure to initiate the vaccination series by age 3 months was highly associated with failure to complete the vaccination series by age 2 years. Of the 312 children whose vaccinations had been initiated by 3 months, 60% were up to date at age 2 years, compared with 30% of those whose vaccinations had not been initiated by age 3 months (P < .0001). Among those who had initiated the series by age 3 months, mothers who were multiparous, living below the poverty line or with missing income data, and who had not received adequate prenatal care were more likely to have children who were not up to date at age 2 years.
Although vaccination coverage of US preschool children has improved over the past 2 years,2 38% to 57% of the children in our sample had not completed the 4:3:1 series by age 2 years. Although our study was contemporaneous with the most recent National Immunization Survey, far fewer of the children in our sample had completed the series than those reported in the National Immunization Survey for the United States in general or in Indiana in particular.2 Socioeconomically disadvantaged urban children continue to be at high risk for undervaccination.
Risk factors for undervaccination at age 2 years were very similar to those reported previously at ages 3 and 7 months. Consistent independent predictors of undervaccination at 3, 7, and 24 months of age were unmarried status, lack of coresidency with the child's grandmother, and inadequacy of prenatal care. We also identified several other independent predictors of undervaccination status at age 2 years. Mothers who scored lower on perceived control, indicating that they felt less control over their life and were less responsible for life events, were more likely to have their children's vaccinations not up to date. Surprisingly, perceived benefit of medical care to prevent diseases (anemia, measles, mumps, whooping cough, and polio) as measured at follow-up also was related directly to undervaccination at age 2 years.
This last finding contradicts the Health Belief Model.15 It is possible that those mothers who expected to see a benefit to their child's health after vaccination thought that other illnesses that their children had suffered should have been prevented by the vaccination. Therefore, they may then have become disillusioned and less likely to complete the vaccination series. It is also possible that those mothers who completed their child's vaccination series by age 2 years, but whose children had had minor illnesses, indicated less perceived benefit from vaccinations at follow-up. Either possibility suggests misperceptions on the part of mothers about benefits that should be expected from vaccination. Therefore, education about the true benefits of vaccination may lead to increased vaccination compliance.
We also sought to study the effects of system barriers to care on undervaccination, and the relation between system barriers and personal characteristics. Mothers in this sample who reported less physician continuity and did not have private sources of routine care were more likely not to have their children fully vaccinated. However, after controlling for the other variables in the model, these system barriers were not significant independent risk factors for vaccination status (although the OR for private source of routine care was very close to statistical significance). Thus, the effects of these system barriers to care are likely partly indirect.
One limitation of our study is that the sample is composed primarily of poor women and children from one urban medical center. Because of the relative homogeneity of our sample, we probably have underestimated the effect of sociodemographic characteristics such as education, income, insurance status, and source of health care. If the beliefs and attitudes of this group are markedly different from higher socioeconomic and more fully vaccinated groups, then the effects of these beliefs also may be underestimated. However, this study does reflect effectively the variation inherent in this population, in that some mothers with similar socioeconomic status and access to care are more likely to vaccinate their children than are others.
A second limitation of our study arises from our limited follow-up data. We were unable to contact 40% of the mothers who were interviewed at baseline. However, we were able to document the vaccination status of two thirds of their children. Thus, vaccination status was confirmed on 81% of those enrolled at baseline. Analyses of the two groups (ie, vaccination status documented vs not documented) revealed no differences in baseline characteristics. Although many of the children who did not have documented vaccination status undoubtedly had received only the vaccinations that we found, we made no assumptions in this regard. We felt it was important that determination of vaccination status was accurate and verified, and are confident that it has been.
In conclusion, family environment, the mother's history of prenatal care use, and financial variables are important determinants of vaccination status among socioeconomically disadvantaged children in the first 2 years of life. After controlling for personal and financial characteristics, health care system variables were not significant independent predictors of vaccination status. We have identified which urban poor children are highest risk for undervaccination and therefore merit increased attention in future interventions designed to improve vaccination rates: those whose mothers are unmarried, not living with the child's grandmother, are multiparous, living in poverty, and did not receive adequate prenatal care. Mothers also may need a more realistic expectation of the health benefits of vaccination to prevent them from becoming disillusioned and discontinuing the series. Finally, additional investigation of the determinants of satisfaction with care is warranted. Improved perceptions of health care delivery systems may ultimately increase the use of preventive health services.
This work was supported in part by grants from the James Whitcomb Riley Memorial Association and the Robert Wood Johnson Foundation Generalist Physician Faculty Scholars Program (A.S.B.), and by National Institutes of Health Grant R37AG09692 (F.D.W.). This work reflects the opinions of the authors and not those of any funding agency.
We thank Laurie Adams for her assistance in data collection and preparation of this manuscript and Drs John Fitzgerald, Donald Orr, and Robert Dittus for helpful comments throughout this project.
- Received May 16, 1997.
- Accepted August 27, 1997.
Reprint requests to (A.S.B.) 14054 Old Mill Circle, Carmel, IN 46032.
This work was presented in part at the annual meeting of the Ambulatory Pediatrics Association, Washington, DC, May 6–10, 1996.
- ↵Institute of Medicine. Access to Health Care in America. Washington, DC: National Academy Press; 1993
- ↵US Department of Health and Human Services, Public Health Service. Healthy People 2000: National Health Promotion and Disease Prevention Objectives. Washington, DC: US Government Printing Office; 1990; PHS 91-50212
- Centers for Disease Control and Prevention
- Bobo JK,
- Gale JL,
- Thapa PB,
- Wassilak SG
- Marks JS,
- Halpin TJ,
- Irvin JJ,
- Johnson DA,
- Keller JR
- Szilagyi PG,
- Rodewald LE,
- Humiston SG,
- et al.
- Wood D,
- Donald-Sherbourne C,
- Halfon N,
- et al.
- Guyer B,
- Hughart N,
- Holt E,
- et al.
- ↵Hochbaum GM. Public participation in medical screening programs: a socio-psychological study. Washington, DC: Government Printing Office, PHS pub 572; 1958
- ↵Kessner DM, Singer J, Kalk CE, Schlesinger ER. Infant death: an analysis by maternal risk and health care. In: Contrasts in Health Status. Vol I. Washington, DC: Institute of Medicine, National Academy of Sciences; 1973
- ↵Davies AR, Ware JE. GHAA's Consumer Satisfaction Survey and User's Manual. GHAA. 1991
- Copyright © 1998 American Academy of Pediatrics