The Influence of Provider Behavior, Parental Characteristics, and a Public Policy Initiative on the Immunization Status of Children Followed by Private Pediatricians: A Study From Pediatric Research in Office Settings
Objectives. To determine the relative impact of parental characteristics, provider behavior, and the provision of free vaccines through state-sponsored vaccine volume programs (VVPs) on the immunization status of children followed by private pediatricians.
Study Design. Retrospective and cross-sectional surveys of immunization data.
Setting. The offices of 15 private pediatricians, from 11 states, who were members of the Pediatric Research in Office Settings network. Seven of these physicians used vaccines provided through VVPs.
Patients. Children 2 to 3 years old followed by the participating physicians.
Methods. The immunization status of children was assessed from two separate samples. For sample 1, immunization data were abstracted from the medical records of 60 consecutive eligible children seen in each office. Parents of the selected children indicated the method of payment for immunizations and the education levels of the mothers. Because this cross-sectional survey might have oversampled frequent health care users, a retrospective chart review of up to 75 randomly selected children in each pediatrician's practice was also conducted (sample 2). Additional data were collected from the parents of children in sample 2 by telephone interviews. For both samples, patients were considered to be fully immunized if they had received four diphtheria-tetanus-pertussis/diphtheria-tetanus vaccines, three oral poliovirus/inactivated poliovirus vaccines, and one measles-mumps-rubella vaccine before their second birthdays. Before collecting vaccination data, pediatricians completed a survey detailing their immunization beliefs and practices. Logistic regression was used to identify factors that were independently associated with a child being fully immunized.
Results. For sample 1, 81.7% of the 857 children surveyed were fully immunized. Practitioner-specific immunization rates varied widely, ranging from 51% to 97%. The immunization rate of children who received vaccines provided by VVPs was similar to that of children whose immunizations were not provided by VVPs (81.2% vs 82.2%; odds ratio [OR] for a VVP as a predictor for being fully immunized, 0.94, 95% confidence interval [CI], 0.66 to 1.32). In addition, parents who paid for immunizations out of pocket were as likely to have fully immunized children as those who had little or no out-of-pocket expenditures for vaccines (OR, 1.13; 95% CI, 0.75 to 1.13). In the logistic model, only individual pediatrician and size of the metropolitan area in which the pediatrician's practice was located were significant predictors of a child's immunization status. The results from sample 2 were similar; 82.1% of the 772 surveyed patients were fully immunized. With sample 2, individual pediatrician and age of the child at the time of the survey were the only predictors of immunization status. The OR of a VVP as a predictor of a child being fully immunized was 1.37 (95% CI, 0.65 to 2.90).
Conclusions. Individual provider behavior may be the most important determinant of the immunization status of children followed by private pediatricians. In our samples, the effect of parental characteristics was limited. State-sponsored VVPs were not associated with higher immunization rates, perhaps because cost of vaccines did not seem to be a significant barrier to immunization in this population.
- PROS =
- Pediatric Research in Office Settings •
- AAP =
- American Academy of Pediatrics •
- VVP =
- vaccine volume program •
- DTP =
- diphtheria-tetanus-pertussis •
- OR =
- odds ratio •
- CI =
- confidence interval
Immunization ranks after only clean water and sewage removal in its positive effect on the health of the people.1Despite this public health importance, the immunization rate of US preschool children is disappointingly low. Although exact figures are difficult to obtain, large surveys indicate that as few as 44% of children have received all recommended vaccinations by their second birthdays,2 with the overall immunization rate of American preschool children recently estimated at 65%.3 Efforts to improve the immunization status of preschool children have included various public policy initiatives,4,5 attempts at changing provider behavior,6-9 and endeavors to understand parental characteristics associated with underimmunization.10-12Unfortunately, few data are available to demonstrate the efficacy of these approaches.
Most previous research on immunizations has concentrated on children seen in public health settings. However, it has been estimated that 50% of all vaccinations occur in private offices,13 a proportion that is likely to increase as health maintenance organizations bid for Medicaid managed care contracts. If the goal of a 90% immunization rate in preschool children by 2000 is to be realized,14 it is important to determine the immunization status of patients followed by private pediatricians and to evaluate critically the effectiveness of programs designed to increase vaccinations in this population.
Previous studies on the immunization status of children followed by private physicians have been surveys in geographically confined regions with limited generalizability.7,8,15,16 However, the formation of national practice-based research networks such as Pediatric Research in Office Settings (PROS), a program of the American Academy of Pediatrics (AAP), provides the opportunity for projects that could produce more generalizable results.
In 1992, members of an AAP policy-making group requested that PROS conduct a study to assess the impact of public policy initiatives, provider behavior, and parental characteristics on the immunization status of children followed by private pediatricians.17Three specific issues were addressed in the project. First, we evaluated the impact of providing vaccines free to practitioners through state-sponsored vaccine volume programs (VVPs) in increasing immunization rates. With VVPs (also called universal purchase programs) there are no costs to the patient's family for vaccines, although providers may charge a small administration fee.4,5,18Second, we examined the independent effect of private pediatricians on the immunization status of their patients. Finally, we assessed the influence of two parental characteristics—socioeconomic status (as estimated by maternal education level) and method of payment for vac-cinations—on the immunization status of their children.
The study was conducted in the private practices of 15 PROS pediatricians who routinely provided immunizations in their offices. Seven pediatricians practiced in states where VVPs had been operational since at least 1989, and each of these pediatricians made full use of vaccines provided by VVPs. The remaining 8 physicians practiced in states where free vaccines were not provided to pediatricians for use in all patients. Efforts were made to choose non-VVP pediatricians from locations geographically close to the VVP group and to survey as geographically diverse a population as possible. At the time that study children received vaccinations, 11 states had VVPs for all patients. These VVP states were predominately located in either the northwest or northeast sections of the United States. There was at least 1 PROS private practice in 7 of the 11 VVP states; practices in 5 of the 7 states contributed data to the study. Participating physicians who used vaccines provided by VVPs practiced in Idaho, Massachusetts, Rhode Island (2 pediatricians from different practices), Vermont (2 pediatricians from different practices), and Washington. Study pediatricians who practiced in states without VVPs for all patients were located in Colorado, New Jersey, New York (2 pediatricians from different practices), Oregon, South Carolina (2 pediatricians from different practices), and Utah. In each of these states, the PROS coordinator for that location was asked to recommend a PROS practitioner for participation in the study. Initially, 17 designated pediatricians agreed to be involved in the project; however, 2 practitioners dropped out of the study before collecting data for sample 1.
Participating practitioners were compared with other PROS practitioners and a random sample of AAP providers drawn from a 1994 AAP Research Department periodic survey who identified direct patient care as their primary professional responsibility (AAP Division of Child Health Research, periodic survey 25, unpublished data, 1994). Those pediatricians who contributed data to the study were similar to other office-based PROS providers and to randomly selected AAP practitioners with respect to age (66.7%, 58.6%, and 55.7% <45 years old, respectively), gender (73%, 56.9%, and 59.7% male), and size of practice (80%, 87.0%, and 65% working in groups of three or more providers).
Before collecting data on their patients, the participating pediatricians completed a physician survey on their policies and attitudes about immunizations, especially concerning adherence to the Standards for Pediatric Immunization Practices.9 The providers were asked whether they provided vaccinations at acute visits for children who needed them, if they referred patients to other facilities for immunizations, about their perception of the profitability of providing immunizations in their offices, and whether they considered the following scenarios—gastroenteritis without dehydration, otitis media without fever, family history of severe reaction to diphtheria-tetanus-pertussis (DTP) vaccination, upper respiratory tract infection without fever, bronchiolitis without fever, fever of less than 39°C, and fever of 39°C to 40°C—to be contraindications to DTP vaccination in a hypothetical 4-month-old infant seen at a health supervision appointment.
Two separate samples were collected and analyzed to assess the immunization status of children followed by the participating pediatricians. Sample 1 was a prospective, cross-sectional survey of 60 consecutive children, 2 and 3 years of age, seen for visits by pediatricians in their offices for any reason. This sampling technique allowed parents to complete a detailed questionnaire about usual mechanisms of payment for immunizations (public assistance, private insurance, obtained for free or small fee at the office or other facility, and out-of-pocket), as well as the child's date of birth and education level of the child's mother. For each study patient, immunization data were abstracted from the practice medical record. A disadvantage of this technique, however, was the increased likelihood of oversampling frequent users of health care services, who might also have higher immunization rates or differ systematically in other ways from a random sample of practice patients.
For this reason, sample 2 was also collected, consisting of a group of up to 75 patients in each practice, 2 and 3 years of age, randomly selected from practice medical records. For practices with computerized records, names were randomly selected from the list of all eligible children. For practices without computers, the first three medical records of eligible patients from each letter of the alphabet were selected. Parents of the selected children were contacted by a member of the pediatrician's staff, and a brief telephone interview was conducted. Seven attempts were made to contact the parents of each child in the sample before they were deemed unreachable. During the telephone interview, the parent was asked to read the dates of all of the child's immunizations from any written records in their possession, and the child's status as an active patient in the participating pediatrician's practice was determined. The education level of the mother was also obtained, but because of time constraints, no information on usual payment for immunizations was collected. After the telephone interview, immunization data from the child's medical record were abstracted.
For sample 2, only those children whose parents were contacted for the telephone follow-up interview and who were active patients of the participating pediatricians were included in the analysis. Immunization data on each child obtained from the practice medical record were supplemented with additional dates of immunizations from the telephone interview. For each antigen, dates of vaccinations that occurred within 28 days of each other were counted as only one immunization.
For both surveys, study children were considered fully immunized if they had received four DTP/diphtheria-tetanus vaccines, three oral poliovirus/inactivated poliovirus vaccines, and 1 measles-mumps-rubella vaccine before their second birthdays. Immunization rates were calculated by dividing the number of fully immunized children in a particular group by the total of patients in that group. Study patients were assigned to either a VVP or non-VVP group on the basis of their pediatrician's participation in these programs. Maternal education level was categorized as less than high school graduate, high school graduate, some college, or college graduate. Study children were classified as residing in either a large city (population >250 000), medium-size city (population 50 000 to 250 000) or small city (population <50 000) on the basis of the location of their pediatricians' offices. The method of payment for immunizations was categorized as out-of-pocket or free (ie, vaccines provided by VVPs or covered by insurance). Children whose parents may have paid a small “administration fee” for vaccinations were classified as receiving free immunizations, because such charges are allowed in VVPs. Although no information was collected regarding the size of administration fees in VVP practices, a 1989 survey of 25 pediatricians participating in Washington State's VVP found that the average administration charge for a DTP vaccine was $8.06.18
For the analysis, categorical data were analyzed with χ2or Fisher's exact tests, and continuous variables were assessed witht tests. The odds ratio (OR) and 95% confidence interval (CI) for free vaccine as a predictor of a child being fully immunized was calculated with logistic regression after controlling for maternal education level, age, and size of the metropolitan area. Logistic regression was also used to assess the independent effect of each of the following variables on the immunization status of study children: vaccines provided by VVPs, maternal education, size of metropolitan area, and age of the child (2 or 3 years old). To adjust for the effect of pediatrician and practice setting on the immunization status of their patients, logistic analysis was repeated with pediatricians and practices included as random effects terms. For this analysis, a logistic-binomial model for distinguishable data was used, with matching on the study patients' pediatricians. This regression technique has been used previously to explain variations in care.19 With this technique, ORs for each of the other variables were adjusted to account for the effects of the providers, and the overall impact of providers on immunization status was assessed. For all statistical tests, differences were considered significant at P < .05 or when the 95% CI did not include 1.0.
The study was approved by the Institutional Review Board of the AAP.
For the cross-sectional survey (sample 1), immunization data were collected on 857 patients. The overall immunization rate of these children was 81.7%. Practitioner-specific immunization rates for each of the 15 participating pediatricians are displayed in the Figure. As can be seen in the Figure, immunization rates varied dramatically, ranging from 51% to 97%. In Table1 physician policies are compared with immunization rates. Because of the modest number of pediatricians surveyed, the power to detect statistically significant differences was limited. However, the mean immunization rate for physicians who indicated that three or fewer of the seven scenarios listed were contraindications to vaccination was significantly higher than for those who indicated that four or more of the scenarios were contraindications (89.7% vs 74.4%;P = .037). Immunization rates for pediatricians who referred at least some children for vaccinations were similar to those who always provided immunizations in the office.
The effect of parent and child characteristics on immunization rates is displayed in Table 2. As shown, children who resided in medium-size cities (population 50 000 to 250 000) were significantly more likely to be fully immunized than those from large cities and smaller towns. The effect of maternal education on immunization status was modest but significant; children of mothers with higher education levels were more likely to be fully immunized than those of mothers with less education. Overall, 76.8% of parents indicated that their children received immunizations for little or no out-of-pocket expenditures; parents of children immunized by VVP pediatricians were more likely to indicate that vaccines were free than those of children followed by non-VVP providers (90.5% vs 62.8%; OR, 5.66; 95% CI, 3.84 to 8.33). However, the immunization rate for children whose parents paid for vaccinations entirely out of pocket was similar to that of those who received free immunizations (80.4% vs 82.2%; OR, 1.13; 95% CI, 0.75 to 1.71). After adjusting for the size of the metropolitan area, maternal education, and the age of the child, the OR for free vaccine as a predictor of being fully immunized was 1.31 (95% CI, 0.84 to 2.04).
The immunization rates of children whose vaccines were provided by VVPs and those whose pediatricians did not participate in VVPs were similar (81.2% vs 82.2%; P = .774). The unadjusted OR for a VVP as a predictor of a child being fully immunized was 0.94 (95% CI, 0.66 to 1.32). However, as shown in Table 3, there were differences between children immunized with vaccines provided by VVPs and those receiving immunizations from non-VVP providers. There were also differences between VVP and non-VVP pediatricians. Children immunized by non-VVP providers tended to have more highly educated mothers and to reside in larger metropolitan areas than those immunized by physicians participating in VVPs. Non-VVP pediatricians were more likely to refer at least some patients for vaccinations and were more likely than VVP providers to indicate that the provision of immunizations was profitable.
To adjust for patient characteristics, logistic regression was performed after controlling for maternal education, the size of the metropolitan area, and the age of the child (2 or 3 years). With this analysis, the adjusted OR for a VVP as a predictor for being fully immunized was 1.08 (95% CI, 0.71 to 1.64). Children from medium-size cities were more likely to be fully immunized than those in large cities (OR, 3.74; 95% CI, 1.79 to 7.81), whereas those in small cities and towns were significantly less likely to be fully immunized than those in large cities (OR, 0.52; 95% CI, 0.34 to 0.81).
To adjust for the influence of the pediatrician and practice setting on the likelihood that a patient would be fully immunized, individual pediatricians were controlled for by using a random effects model. An individual pediatrician was significantly associated with a child's immunization status (P < .001); in this analysis, the OR for VVPs was 1.29 (95% CI, 0.59 to 2.81). The only variable other than individual provider that was significantly associated with immunization status was residing in a medium-size city (OR, 3.46; 95% CI, 1.07 to 11.17).
Because the cross-sectional survey (sample 1) might contain a disproportionate number of frequent health care users, we repeated our analysis on sample 2, data collected by a random review of medical records supplemented with information obtained during telephone interviews with parents. The medical records of 977 children were reviewed; parents of 860 (88%) of these patients were contacted for telephone interviews. Of these 860, data on 54 children (6.2%) were excluded because they were no longer active practice patients, and 34 (4.2%) were excluded because they were either younger than 2 or older than 3 years at the time of the survey. Thus, data on 772 eligible patients were analyzed; 82.1% of these children were fully immunized by their second birthdays. To determine which characteristics were independently associated with immunization status, logistic analysis was performed using a random effects model. In this analysis children's individual providers were again significantly associated with the likelihood that they would be fully immunized (P < .001). ORs for the size of the metropolitan area and maternal education were not significantly different from those of sample 1. Children who were 2 years old were significantly more likely to be fully immunized than those who were 3 years old (OR, 1.81; 95% CI, 1.22 to 2.69). In this analysis, after adjusting for potentially confounding variables, the OR for a VVP as a predictor of being fully immunized was 1.37 (95% CI, 0.65 to 2.90).
The data in this study, collected from a large, geographically disparate sample of children from 11 states, offer insight into the relative impact of public policy, provider characteristics, and parental factors on immunization rates in the private sector. Our results suggest that provider- and/or practice-associated characteristics may be the most important determinants of the immunization status of children followed by private pediatricians. Among patients seen in private offices, the effect of the parental characteristics measured in this study had limited impact on immunization status. Finally, in this population of children, the use of state-provided free vaccines was not associated with higher immunization rates. These findings were consistent whether data from the cross-sectional survey or retrospective medical record review were analyzed.
The most striking finding of our study was the wide variation in provider-specific immunization rates, a difference not explained by patient population or method of payment for vaccines. A child's individual provider was the only variable tested that was significantly associated with the child's immunization status in all analyses. These results suggest that specific provider beliefs and practices could be associated with high levels of fully immunized children. Because only 15 pediatricians were involved in the project, it is difficult to critically compare the effect of specific provider and/or practice policies on the immunization status of their patients. However, some of the results of our physician survey were provocative, especially when compared with the Standards for Pediatric Immunization Practices.9 One of the standards states that providers should follow only true contraindications to vaccination so that immunizations are not needlessly deferred. When asked to indicate which of seven clinical scenarios were contraindications to vaccination in their practice, those pediatricians who responded that three or fewer were contraindications had a significantly higher proportion of fully immunized patients than those who indicated that four or more of the scenarios would cause them to defer immunization. However, adherence by study pediatricians to the standard that states that immunizations be provided at all clinical encounters, including visits for acute care, was not associated with increased levels of fully immunized children. The mean immunization rate for those who provide vaccines at visits for acute care was actually lower than for those who do not vaccinate at such visits, but this difference did not achieve statistical significance. Szilagyi et al8 reported the same counterintuitive relationship among primary care physicians in Rochester, NY.
The lack of evidence for a VVP effect was surprising. Immunization rates for children followed by VVP and non-VVP pediatricians were virtually identical. Even after controlling for several potentially confounding variables, a VVP was not a significant predictor of a child being fully immunized. There are several possible explanations for this finding. It is conceivable that, in our sample of pediatricians, the group of non-VVP pediatricians was more effective at providing immunizations than the those in the VVP group. There were differences in responses to the physician survey between VVP and non-VVP pediatricians. Those not participating in such programs were more likely to refer some children for vaccinations and were also more likely to think that the provision of immunizations was profitable. It may be that the differences noted between VVP and non-VVP providers were not inherent but actually the result of participation in VVPs and that VVPs, by eliminating the profit potential for practicing pediatricians, remove an effective incentive for practitioners to immunize their patients. There may have been other, unmeasured, differences between the two groups of physicians that biased the results. However, controlling for individual pediatricians in the logistic model did not significantly alter the OR for a VVP as a predictor of immunization status. It is possible that our study was not large enough to detect a small positive effect of VVPs. Our sample size had a power of greater than 90% to detect a 10% difference in immunization rates between children vaccinated by VVP pediatricians and non-VVP providers. Smaller differences might have not been detected.
It is possible that the impact of VVPs was limited because of other sources of payment for immunizations among children followed by private pediatricians. In our population of children immunized by non-VVP providers, 62.8% of the parents indicated that they had little or no out-of-pocket expenditures for immunizations. Thus, the institution of VVPs in these practices would remove the financial burden of vaccines for only 37.2% of the patients. The high degree of insurance coverage for immunizations found in this study is reflective of a national trend. A 1992 survey found that immunizations were covered for 54% of individuals participating in conventional employer-sponsored health insurance plans; when those enrolled in health maintenance organization plans were included, this figure rose to 66%.20 Finally, cost may not be a significant barrier to immunization in this population; in our study, immunization rates of children whose parents paid for vaccines out-of- pocket were similar to those of children whose parents who had little out-of-pocket expenses.
Unlike other investigators,21,22 we found that the maternal education level, as a proxy for socioeconomic status, had little association with immunization rates. Although children of mothers with more education were more likely to be fully immunized than those of mothers with less education, the difference was not significant after controlling for other variables. This finding may be attributed to a generally high level of education among mothers of children included in the study; 38% of mothers were college graduates. Alternatively, mothers with lower education who have sought care from private practitioners who provide immunizations may be different in other critical ways from other mothers with lower education. That the immunization status of children followed by private pediatricians is significantly influenced by individual providers but minimally affected by maternal education level is an encouraging finding; physician attitudes are more easily modified than socioeconomic status.
In the cross-sectional survey (sample 1), but not sample 2, the size of the metropolitan area was significantly associated with immunization status. Conversely, among children included in sample 2, those who were 2 years old at the time of the survey were significantly more likely to have been fully immunized by their second birthdays than those who were 3 years old, indicating a secular trend in immunization rates. Neither of these characteristics were significantly associated with immunization status in both surveys, suggesting that the effect of the size of a metropolitan area or age on whether a child was fully immunized was limited.
The proportion of fully immunized children found in this study (81.7% in sample 1 and 82.1% in sample 2) was greater than what has been reported in population-based surveys.2,3 However, recent studies on the immunization status of patients followed by private pediatricians have had results similar to ours. In a 1993 study in Rochester, NY, 86% of children seen by private pediatricians were fully immunized before their second birthdays.7 A survey of two private practices in the Seattle, WA, area in 1992 found that 82.7% of children were fully immunized.16
Because our sample of practicing pediatricians was neither large nor randomly selected, it would be imprudent to make definitive conclusions about the relative impact of various provider and practice, parental, and public policy factors on the immunization status of children. More study is needed to determine optimal strategies for increasing immunization rates. However, in an era of limited resources for preventive services, wide-scale interventions designed to increase immunizations should be based on demonstrated effectiveness rather than theoretical assumption. Our findings suggest that factors particular to individual pediatricians or practices seem to have a major role in affecting the immunization status of their patients. Future research is needed to elucidate which provider and practice characteristics are associated with high levels of fully immunized patients so that appropriate interventions can be tested and implemented.
This study was supported by a grant from the Merck Vaccine Division and by grant MCJ-17702 from the US Maternal and Child Health Bureau.
- Received October 3, 1995.
- Accepted March 19, 1996.
Reprint requests to (J.A.T.) Department of Pediatrics, University of Washington, Box 356320, Seattle, WA 98195.
Presented in part at the 34th Annual Meeting of the Ambulatory Pediatric Association, Seattle, WA, May 10, 1994.
- ↵Plotkin SL, Plotkin SA. A short history of vaccinations. In: Plotkin SA, Mortimer EA, eds. Vaccines. 2nd ed. Philadelphia, PA: WB Saunders Co; 1994:1–11
- McConnochie KM,
- Roghmann KJ
- Szilagyi PR,
- Rodewald LE,
- Humiston SG,
- et al.
- Keane V,
- Stanton B,
- Horton L,
- Aronson R,
- Galbraith J,
- Hughart N
- Hinman AR
- ↵Public Health Service. Healthy People 2000: National Health Promotion and Disease Prevention Objectives. Washington, DC: US Department of Health and Human Services, Public Health Service; 1990
- McDaniel DB,
- Patton EW,
- Mather JA
- Taylor JA,
- Quinn H,
- Dassel S
- American Academy of Pediatrics 1992 Chapter Forum
- ↵Miller M, Dial TH. Employer-Sponsored Health Insurance in Private Sector Firms in 1992. Washington, DC: Health Insurance Association of America; 1993
- Marks JS,
- Halpin TJ,
- Irvin JJ,
- Johnson DA,
- Keller JR
- Copyright © 1997 American Academy of Pediatrics