Effect of the Vaccines for Children Program on Physician Referral of Children to Public Vaccine Clinics: A Pre-Post Comparison
Objective. Started in late 1994, the Vaccines for Children (VFC) program is a major entitlement program that provides states with free vaccines for disadvantaged children. Some evaluation studies have been conducted, but they do not include individually matched pre-post comparisons of physician responses. This project studied the effect of the VFC on the physician likelihood of referring children to public vaccine clinics for immunizations.
Design. In 1999, trained personnel conducted a survey of a cohort of physicians who previously participated in surveys on barriers to childhood vaccination conducted before VFC implementation. Responses were matched, and pre- versus post-VFC comparisons were made.
Setting and Participants. Minnesota and Pennsylvania primary care physicians selected by stratified random sampling and initially studied in 1990 to 1991 and 1993, respectively.
Main Outcome Measures. Likelihood of referral of a child to a public vaccine clinic.
Results. On a scale of 0 to 10, physician likelihood of referring an uninsured child decreased by a mean of 1.9 (95% confidence interval: 1.2–2.5) from pre- to post-VFC. Two fifths (45%) of physicians reported that the VFC decreased the number of referrals from their practice to public vaccine clinics and 50% gave intermediate responses. Among physicians who participate in VFC, only 9% were likely to refer a Medicaid-insured child in contrast to 44% of those not participating.
Conclusions. Physicians' reported referral and likelihood of referring Medicaid-insured and uninsured children has decreased because of VFC in Minnesota and Pennsylvania.vaccination/economics, vaccination/legislation and jurisprudence, immunization programs/economics, immunization programs/utilization, vaccines/economics, Medicaid/economics, national health programs United States, child health services.
Low vaccination rates in impoverished areas of the United States raised the issue of economic barriers to vaccination, particularly in light of the 1989 to 1991 measles epidemic. Studies in the 1980s and early 1990s found that many physicians reported referring uninsured and Medicaid-insured children to public health vaccine clinics instead of vaccinating them in their medical home.1–7
This issue of economic barriers to vaccination led to the development of the federal Vaccines for Children (VFC) program. Begun in October 1994, the VFC provides states with free vaccines for children who are Medicaid-eligible, have no health insurance, or are Native Americans or Alaskan natives. In addition, children whose insurance does not cover vaccines are eligible for VFC if they are vaccinated at a federally qualified health center or rural health clinic. VFC vaccines are distributed through 61 immunization projects to both private providers and public clinics,8 and the program costs about $560 million annually, a figure that is expected to increase as new vaccines are added. For example, as of September 27, 2000, the prices forHaemophilus influenzae type b vaccines were $5.20 to $7.75 per dose under the federal contract and $15.25 to $18.12 in the private sector, compared with the newer 7-valent pneumococcal conjugate vaccine at $44.25 per dose under the federal contract and $58 in the private sector.9 A recent evaluation of the VFC penetration rate has reported that 74% of the surveyed population received all or some of their immunizations from a VFC-enrolled provider.10
Although the VFC was a major change in vaccine financing, there was some controversy. The General Accounting Office, in a 1995 report to the United States Congress on the VFC, wrote that the “cost for at least some major vaccines has not been a major barrier to immunization.”11 The General Accounting Office had a number of concerns about the VFC and its implementation.
Another program that has had strong economic impact on vaccine service delivery is the new state Child Health Insurance Program (CHIP), enacted in 1997 as part of the Balanced Budget Act and codified at Title XXI of the Social Security Act. The main goal of the CHIP is to expand insurance coverage for children and to improve access to care. Under this program, states are given the option to provide health insurance to uninsured children, either through an expansion of existing Medicaid programs, through a separate non-Medicaid program, or through a combination plan. To date, of the 56 states and territories, 15 (including Pennsylvania) have approved a separate state child health plan, 23 (including Minnesota) have at least the initial phases of Medicaid expansion, and the remaining 18 have selected a combination plan.12,,13
The literature on the effects of VFC is limited, and only 1 previous study has compared physician-reported out-referral that was measured before and after the implementation of VFC.14 This comparison, in which physician responses were not individually matched, revealed that out-referral decreased substantially in New York.
The main purpose of our study was to compare physician reported out-referral to public vaccine clinics pre- versus post-VFC implementation in both Minnesota and Pennsylvania. Other purposes included identification of the impact of VFC, CHIP, insurance status, and a practice's primary payment source on reported out-referral. We also evaluated perceptions of the adequacy of Medicaid vaccine administration fees and capitation.
We recruited Minnesota and Pennsylvania primary care physicians who participated in previous surveys on immunization barriers in 1990 to 1991 and 1993, respectively.4,5 The Minnesota sample came from systematic samples based on a random start point of the Minnesota Medical Association master list using 4 strata: general practitioners, board-certified family physicians (FPs) in urban and suburban areas, FPs in rural areas, and pediatricians (Peds). The Minnesota Medical Association list includes nonmembers and incorporates the American Medical Association's list of physicians and data from the Minnesota Board of Medical Examiners. The Pennsylvania sample came from a random sample of FPs, GPs, and Peds from the combined listings of the American Medical Association and the American Osteopathic Association.
Minnesota has purchased additional vaccine and distributed it through its VFC program so that the underinsured can be vaccinated in their medical home instead of having to go to federally qualified health centers; Pennsylvania has not done this.
The target population was physicians seeing at least moderate numbers of children and who thereby could influence immunizations. Hence, physicians seeing ≥5 children per week <6 years old and having ≥50% primary care patients were eligible. Participants were recruited by letter and then by phone call and offered a $40 honorarium. This project was approved by the Institutional Review Board of the University of Pittsburgh.
Several questions were used essentially unchanged from the previous surveys; other questions were written to assess changes in referrals. Of the 57 questions, data from 25 are presented in this study. For many questions, respondents were asked to rate on a scale of 0 (very unlikely) to 10 (very likely) how likely they were to recommend immunization for a child in a particular clinical situation. The questionnaire was pilot-tested and revised.
Data collection was conducted by trained interviewers using Computer Assisted Telephone Interviewing (CATI). The CATI system permitted data entry during the interview,15 eliminated unintentionally skipped questions, and provided automatic range and logic checks. Interviewing was started in June 1999, and continued through November 1999.
Physicians who were unwilling to do a telephone interview were given the options of completing a self-administered version or an Internet Web site version. Of the 281 respondents, 165 completed the interview by telephone, 96 completed and returned a paper copy of the questionnaire by mail or fax, and 20 completed the Web site version.
Paired t tests were used to test for pre- versus post-VFC differences in physician ratings. Before the study, power calculations determined that the sample size required for the matched pair analyses was at least 93 respondents to have statistical power of 0.80 with a type I error of 0.05.
Responses were weighted by the estimated fraction of eligible physicians in each specialty in the target population. In addition, we weighted by state (except for analyses that included the variable state) to account for there being a greater number of Pennsylvania physicians than Minnesota physicians in the target population.
Likert scale responses were collapsed into the categories: unlikely/little concern (ratings of 0–3), intermediate (4–6), and likely/much concern (7–10), and we performed χ2 tests for association on the resulting contingency tables. Multiple linear regression analyses were performed to determine characteristics of physicians most likely to refer to public vaccine clinics children who are 1) uninsured, 2) Medicaid-insured, or 3) insured without coverage for vaccination. The dependent variable in each regression was physicians' response on the scale of 0 (very unlikely) to 10 (very likely). From demographics, practice characteristics, and physician-rated likelihood of referring an insured child to a public vaccine clinic, a forward selection regression analysis identified those variables that significantly (P < .05) contributed to the prediction of the dependent variable. Generally, P values ≤.05 were considered statistically significant, and all analyses were performed with SAS 6.12 statistical software (SAS, Cary, NC).
Of the initial 640 physicians interviewed between 1990 and 1993, we found 112 to be ineligible, and we were either unable to make contact or determine eligibility for 140. Of the remaining 388, 281 (72%) responded; in 42 cases a member of the physician's office staff refused the interview without direct communication occurring between the physician and interviewer, and 65 physicians refused. Of the original sample, including retired, deceased, and ineligible physicians in the denominator, 44% (281/640) completed interviews.
Physicians who completed the 1999 survey did not differ from those who refused to be resurveyed in mean year of graduation from medical school (1975 ± 0.5 vs 1977 ± 0.81, P = .27), or in the gender (P = .26) or state (Pennsylvania or Minnesota; P = .21) proportions. Comparisons between responders versus those who refused on items asked on the original surveys revealed no differences in the mean number of patients seen per week (120 ± 3 vs 113 ± 4, P = .16), or the percent primary care patients (95 ± 1 vs 94 ± 2,P = .75). Those who completed the 1999 survey reported on the original survey seeing a larger percentage of preschool-aged children than those who refused (37 ± 2 vs 27 ± 2,P = .0002). On the basis of the original survey, the 2 groups did not differ in their likelihood to refer to the health department for vaccination a Medicaid-insured child (3 ± 0.2 vs 3 ± 0.4, P = .21), an uninsured child (7 ± 0.2 vs 8 ± 0.4, P = .65), or an insured child (2 ± 0.2 vs 2 ± 0.3, P = .83).
Minnesota and Pennsylvania respondents differed in the number of physicians in the practice, with Minnesota physicians practicing in larger groups (P = .001; Table 1). There was a specialty difference as more Minnesota FPs responded than Peds, whereas in Pennsylvania, more Peds responded (P = .046, Table 1).
Impact of Insurance Status on Referral to Health Department Clinics for Immunizations
Physicians were asked in 1999 to rate their likelihood on a scale of 0 (very unlikely) to 10 (very likely) of referring a child to the health department for immunization in a series of survey questions in which only the insurance status of the patient changed. The percentage of respondents likely to refer (ie, 7–10 on the scale) was 61% for an uninsured child whose parents are unable to pay, 54% for a child with insurance that did not cover vaccines, 14% for a child with Medicaid, 25% for a child with insurance that pays for immunization but requires a large copayment, and 1% for a child with insurance that covers vaccines. Although there were no state differences in referring insured or uninsured children, Pennsylvania physicians were more likely to refer a Medicaid-insured child (P = .004).
The majority of physicians (60%) reported a higher likelihood of referring an uninsured child as compared with a child with insurance coverage to a health department vaccine clinic; 74% of these physicians participated in VFC. Of those physicians who would not refer either child (39%), the vast majority (92%) participated in VFC. Finally, a few (1%) reported that they would refer both insured and uninsured children for vaccination.
Participation in VFC and Impact on Referral to Health Department for Immunization
Most physicians participated in VFC; the proportions of Minnesota and Pennsylvania physicians, respectively, participating was 67% and 77%, nonparticipating was 14% and 19%, and did not know was 19% and 4%.
More physicians who did not participate in VFC were likely to refer children to the health department for immunizations than those who participated (Fig 1). The difference was especially marked with regard to Medicaid children, for whom 44% of the physicians who did not participate were likely to refer as compared with 9% of those who participated in VFC (P = .001). When state comparisons were made, similar results were found for referral of insured, insured without vaccine coverage, and insured with large copayments; however, state differences were apparent for Medicaid-insured and uninsured children. Comparing physicians who participated and did not participate in VFC, the likelihood of out-referring a Medicaid-insured child was 10% versus 57% (P = .001) in Pennsylvania and 6% versus 21% (P = .008) in Minnesota. Comparing physicians who participated and did not participate in VFC, the likelihood of out-referring an uninsured child was 53% versus 86% (P = .008) in Pennsylvania and 61% versus 77% (P = not significant) in Minnesota.
Impact of Primary Practice Payment Source on Referral to Public Vaccine Clinics
Physicians selected the 2 payment sources that cover the largest number of children in their practice. The largest sources, in descending order, were commercial health maintenance organization (HMO; 58%), fee for service (20%), Medicaid HMO (13%), traditional Medicaid (7%), and other (2%); there were no significant state differences. The second largest sources, in descending order, were fee for service (33%), commercial HMO (24%), Medicaid HMO (23%), traditional Medicaid (15%), and other (5%). This differed by state: the percentage of physicians reporting fee for service was 26% in Minnesota versus 36% in Pennsylvania, Medicaid HMO was 31% in Minnesota versus 20% in Pennsylvania, commercial HMO was 17% in Minnesota versus 27% in Pennsylvania, traditional Medicaid was 20% in Minnesota versus 13% in Pennsylvania, and other 6% in Minnesota versus 5% in Pennsylvania (P = .03).
The previous finding that physicians were influenced in vaccine referral decisions by insurance status continued to be found after considering the potential confounder of the primary payment arrangement for children (data not shown). Of note, Medicaid-insured children were almost never (0%–1%) referred by physicians whose primary payment source was a Medicaid HMO, whereas referral of Medicaid-insured children by physicians in other payment situations differed by whether or not the physician participated in VFC.
Exploratory multiple regression analyses were performed to determine which variables predict physician likelihood of referral to public vaccine clinics. Forward selection was performed among the following independent variables for all 3 regression analyses: physician specialty, gender, state, number of physicians in the practice, likelihood to out-refer an insured child for immunization, participation in the VFC program, practice administrative arrangement, practice majority owner, payment arrangement of the majority of patients, and primary practice goal. The only 2 statistically significant variables entered into the model to predict an increased likelihood of referral of an uninsured child was a higher likelihood of referring an insured child, which explained 5% of the variance in the dependent variable and not participating in the VFC program also explained 5% of the variance; thus only 10% of the total variance was explained.
Similarly, the most important predictor of increased physician likelihood of referring a Medicaid-insured child was not participating in the VFC program, explaining 16% of the variance. A higher likelihood of referring an insured child explained an additional 13% of the variance in the dependent variable. The only other variables that were found to be statistically significant by the forward selection were state and payment arrangement of the children in the practice, which together explained an additional 4% of the variance; thus a total of 34% of the variance in the dependent variable was explained.
The only important predictor of increased physician likelihood to refer a child with insurance that does not cover vaccines was a higher likelihood to refer an insured child; this explained 4% of the variance.
Historical Comparisons of Out-Referral Before and After VFC
Comparisons of physician likelihood to out-refer using data from surveys conducted both before VFC and after VFC implementation revealed a significant drop in the likelihood to out-refer. Physician likelihood to refer an uninsured child for vaccination, measured on a scale from 0 to 10 where 10 is very likely, decreased by a mean difference of 1.9 (95% confidence interval [CI]: 1.2–2.5, P = .001) from pre- to post-VFC. Stating this another way, the percentage likely to refer an uninsured child decreased from 83% to 61%. Likelihood to refer a Medicaid-insured child decreased by a mean of 0.9 (95% CI: 0.38–1.43, P = .0008) representing a decrease from 29% to 14% of physicians likely to refer. For an insured child, there was a mean decrease of 1.17 (95% CI: 0.78–1.56, P = .0001) representing a decrease from 10% to 1% likely to refer. These data are represented stratified by state in Fig 2; the differences were most pronounced for Pennsylvania physicians.
Determinants of Changing Out-Referral
More than one third (37%) of physicians reported decreased referrals of children to public vaccine clinics in a question about changes in the number of referrals since 1993; the impact was greater in Pennsylvania (P = .001; Table 2). Two fifths (45%) of physicians reported that the VFC program decreased the number of referrals from their practice to public vaccine clinics, 5% reported increased referrals, and 50% gave intermediate responses. The impact differed by state: 32% of Minnesota physicians reported decreased referrals because of VFC compared with 50% for Pennsylvania physicians (P = .015). More than one quarter (29%) of physicians reported that the CHIP decreased the number of such referrals, 10% reported increased referrals, and 61% gave intermediate responses. This also varied by state: 16% of Minnesota physicians reported decreased referrals attributable to CHIP compared with 34% for Pennsylvania physicians (P = .001). Both VFC and CHIP are associated with changes in reported referral since 1993 (Table 2).
Estimates of Prevalence of Economic Barriers Within the Physicians' Practice
Physicians estimated that vaccination costs hindered timely vaccinations for a median of 4% of the children in their practice (range: 0–100). Although costs were not commonly seen as a barrier, they seem to be a sizable barrier for a small subgroup of physicians; 10% of physicians said that costs hindered timely vaccination for 10% or greater of children in their practice. Pennsylvania physicians reported a significant decrease in the estimated percentage of children for whom vaccination costs hindered timely vaccinations; the mean estimate in 1993 was 16% and in 1999 it was 6% (mean difference of −11.7 percentage points; 95% CI: −15.0 to −8.4; P = .0001).
Almost one-third (31%) thought that the reimbursement by traditional Medicaid for the vaccine administration fee was adequate; 31% thought it was inadequate, and 38% gave intermediate answers. This administration fee varied by state ($10 in Pennsylvania and $8.50 in Minnesota); thus, it was not surprising to find that perceptions of adequacy varied, with more Pennsylvania physicians reporting that it was adequate than Minnesota physicians (36% vs 18%, P= .013).
Because of a change in Medicaid coverage in 1999, the full administration fee was paid only for the first immunization; a partial payment was made for the second, and none for any additional immunizations. Many physicians reported being affected by the cap (35%) but the majority did not know (48%), and some reported not being affected (17%); there were no state differences. Among those reporting being affected by the cap on administration fees, one-third (33%) would not give >2 vaccines simultaneously, although the majority (53%) would vaccinate simultaneously; some (14%) gave intermediate answers. More Minnesota physicians were likely to simultaneously vaccinate than Pennsylvania physicians (70% vs 50%,P = .004). Of course, simultaneous immunization is affected by other factors that we did not assess in this study.
Few (20%) physicians felt that the monthly capitation rate from Medicaid HMOs was adequate; over half (51%) rated it as very inadequate, and over one-quarter (29%) gave intermediate responses.
Perceived adequacy of the Medicaid vaccine administration fee was not related to the likelihood of referral of a Medicaid-insured child, but the adequacy of the monthly capitation rate for Medicaid was significantly related (Table 3).
We found that physicians' reported likelihood of referring children to public vaccine clinics decreased substantially from before VFC to after its implementation, and that VFC and CHIP were credited for this decrease. Differential referral based on insurance status continues to occur, among clinicians who are not participating in VFC. Clinicians who are participating are much more likely to vaccinate Medicaid-insured and uninsured children in their office.
Are Physician Responses to Questions About Vaccine Economics Valid?
In assessing the validity of physician responses, we point to 2 previous studies that used similar questions; in addition, 2 ecological sources offer supportive data. In the first 2, the questions used in the present historical comparisons were compared with data from medical records. In the first, a study of Minnesota private practices, the mean age at receipt of diphtheria and tetanus, toxoids and pertussis vaccine #3 (DTP3) was 11.6 months, 10.7 months, and 7.8 months, respectively, for uninsured, Medicaid-insured, and insured children (P = .0001).16 The study found that significant correlates of earlier DTP immunization include physician residency training, suburban/urban practice locale, and the likelihood that the physician would refer children based on insurance coverage to health departments for immunization. The second study in a Minnesota multispecialty group compared responses from physicians and nurse practitioners with immunization data from children that they had vaccinated.17 Vaccination rates were higher (P < .01) for measles-mumps and rubella vaccine 1 (77% vs 48%), DTP3 (84% vs 71%), and DTP4 (82% vs 66%) among providers that received free vaccine as compared with those providers who did not receive free vaccine.
Ecological data are based on doses administered data. Data reported to us from the Minnesota Department of Health and the Pennsylvania Department of Health reveal that the number of measles-mumps-rubella doses administered by public health agencies declined from 69 269 in 1993 to 40 681 in 1998 in Minnesota (personal communication, Diane Peterson, Minnesota Department of Health, 2000) and from 58 997 in 1993 to 24 726 in 1998 in Pennsylvania (personal communication, Nancy Mumper, Pennsylvania Department of Health, 2000). Because overall immunization rates have been increasing, the decline in vaccinations given by public health agencies is consistent with increased vaccinations given by private clinicians. Szilagyi et al18found that the number of childhood vaccinations delivered at health department clinics in Pennsylvania declined 56% between 1993 and 1997.
Do Other Studies Reveal That Economic Barriers Affect Vaccination?
In addition to the 2 aforementioned studies on immunization rates or timing of vaccination in a set of practices, we know of several other studies on the issue. One study found similar vaccination rates for private practices in a research network that did and did not receive free vaccines (81% vs 82%).19 Parents who paid for immunizations out of pocket were as likely to be fully vaccinated as those with little out-of-pocket expenses. Reasons suggested by that study's authors are the high degree of insurance coverage in practices that did not receive free vaccines and that cost may not be a barrier for certain subpopulations.19 Another study in inner-city New York found that immunization rates increased from 17.9% pre-VFC to 42.2% post-VFC.20 Physicians generally attributed the increased rates to the VFC.
Szilagyi et al6 found that the percentage of New York physicians reporting referral of some or all children out for immunizations decreased from 51% (1993 data) to 18% (1997 data); changes in referrals were primarily attributed to changes in insurance coverage due to state laws and to VFC.14 In another study, Szilagyi et al18 found that the number of childhood vaccination delivered at New York health department clinics declined markedly (53%–56%) between 1993 and 1996; two thirds of the health department clinics reported the primary cause for the decline was vaccine financing changes. The combination of these studies in New York suggest a causal role for the combination of VFC and changes in New York's insurance law requiring indemnity insurers to cover childhood immunizations.
Other evidence suggests that finances can be a barrier. First, experimental evidence was found in an older study, the Rand Health Insurance Experiment, in which the immunization rate for children <7 years old in cost sharing plans was 49% compared with 59% in free plans.21 Second, before VFC, states with universal vaccine purchase programs (ie, state purchase of vaccines for all children) have immunization rates that are about 10% higher, all else being equal.22 Third, provision of health insurance to low-income working families results in decreased immunization visits to public health department clinics, increases immunization visits to private providers, and increases immunization rates by 7%.23 Fourth, universal vaccine purchase program can substantially reduce reported immunization charges to parents and referrals from the child's medical home to public vaccine clinics, although charges for well-child visits may increase.24
Thus, the evidence shows that provision of free vaccines to providers who care for disadvantaged children seems to increase immunization rates, but that free vaccines may not increase rates in children who have insurance coverage for vaccines. Such data supports the concepts behind the VFC.
Outcome of Referral of Disadvantaged Children to Public Vaccine Clinics
Although referral of children from primary care physicians offices to public vaccine clinics is preferable to not vaccinating, there are disadvantages. First, the child may visit the public vaccine clinic late and therefore have a window of time when they are not age-appropriately vaccinated; therefore, they may be susceptible to vaccine-preventable diseases. Second, fragmentation of care occurs with an increased burden and expense for the parents of taking the child to one site for vaccines and another site for well-child care and other services. A recent study after VFC implementation found that 73% of children received all or some immunizations within a medical home.10 Third, until wider use of immunization registries, medical records are not easily transferred from one site to another and therefore, each site may lack important information available at the other.
Vaccine Financing in Minnesota and Pennsylvania
Generally, Minnesota and Pennsylvania have progressive immunization financing systems. A first dollar immunization coverage law was implemented in 1988 in Minnesota and in 1992 in Pennsylvania that enables insured children to receive vaccines at their private physicians' offices. The VFC program was implemented on October 1, 1994, in Minnesota, on January 1, 1995, in Philadelphia, Pennsylvania, and April 1, 1996, in Pennsylvania, generally. The first dollar laws greatly reduce underinsurance for immunization; however, certain plans are exempt (ie, Employee Retirement Income Security Act plans). The children who remain underinsured can be vaccinated in their medical home in Minnesota, using vaccine typically purchased by the state; in Pennsylvania, the underinsured can be referred for vaccination to public health vaccine clinics or federally qualified health centers where VFC vaccine can be used for the underinsured.
In our 1999 data, the likelihood of out-referral was fairly similar between the states (Fig 2). Historically, Minnesota physicians reported less out-referral in our 1991 study than Pennsylvania physicians in our 1993 study, perhaps reflecting Minnesota's 1988 first dollar coverage law. Comparisons between our earlier and current results reveal greater changes in Pennsylvania, reflecting the more recent changes in vaccine financing in the state.
In addition to vaccine financing, provider education programs are ongoing in both states that may have contributed to increasing immunization rates; the Pennsylvania program, Educating Physicians in their Community, offers education in the office on practical immunization issues and ways to raise rates. The 4:3:1 immunization rates in both states have risen substantially and are now 87% in Minnesota and 86.6% in Pennsylvania.
Medicaid Vaccine Administration Fees and Adequacy of Medicaid HMO Capitation
At first glance, our finding that perceived adequacy of the Medicaid vaccine administration fees was not associated with the likelihood of referral while the perceived adequacy of the monthly capitation rate was associated may seem puzzling. The 1999 cap meant that the full administration fee was paid for the first dose, half for the second dose, and none for any other vaccines. Thus, for Pennsylvania with a $10 fee, a total of $15 was available even if 4 vaccines were administered. Thus, if a 1-year-old was given 2 sets of injections during the second year of life, only $30 would be paid. The monthly capitation rate of about $12 for Medicaid HMOs in western Pennsylvania totals about $144 per year, which is a higher figure. Wood and Halfon25 suggested several years ago that the impact of the VFC in each state will depend on how that state implements it. Federal purchase of childhood vaccines resulted in savings to state Medicaid budgets. If the state chose to invest the savings in higher reimbursement for vaccine administration, as has been done in New York, Florida, Oklahoma, and several other states, then higher vaccination rates could occur.25 We wonder if the 1999 cap on administration fees may have removed an incentive for some physicians to vaccinate a Medicaid-insured child.
Strengths and Limitations
To our knowledge, this is 1 of only 2 studies that allow direct comparison on physician reported out-referral of data collected before the VFC with data after VFC implementation and the only study to do so with individually matched data. Furthermore, we have data from other studies to support the validity of the main comparison variables.16,,17 Another strength is the use of computer-assisted telephone interviewing for most participants. The actual sample size exceeded the calculated sample size required for the matched-pair analyses.
The study has certain limitations. The response rate was moderate. Tracking physicians previously interviewed, in some cases 8 years ago, is challenging and a number had retired, left primary care, or left the state in which they had initially been interviewed. Little difference in demographics was found between those who refused and participated and they did not differ in the baseline likelihood of out-referral. It is, of course, impossible to know to what extent, if any, the nonrespondents differ with regard to their current attitudes and practices concerning vaccine economics.
Historically, mixed modes for data collection have been discouraged in survey research; however, there is evidence to suggests that, under certain conditions, mixed modes can yield similar results. These conditions include the following: question sequence is not a consideration, the number of open-ended questions is limited, and the questionnaire structure is uncomplicated (ie, contains few skip patterns, complex response structures, etc). Groves26reported that the most significant finding in comparing face-to-face and telephone interviews was the lack of major differences. In another study, Fowler concluded that mode of data collection had little effect on key results.27 Given the busy schedules of primary care physicians and the difficulty we experienced in making arrangements for telephone interviews in many cases, the decision to provide alternatives to the CATI interview proved to be an effective way to minimize refusals. Finally, this is self-reported behavior, although we have data to support the validity of physician responses to the major outcomes.
Referral of children to public vaccine clinics as reported by physicians in Minnesota and Pennsylvania has decreased substantially since the VFC program was instituted. Most physicians who do not receive free vaccine supplies are likely to refer uninsured children to public vaccine clinics. Physicians who receive free vaccine supplies are much more likely to vaccinate children in their offices; this effect is most apparent for uninsured and Medicaid-insured children. Decreased out-referral was attributed to VFC, and to a lesser extent to CHIP.
This publication/project was made possible through a cooperative agreement between the Centers for Disease Control and Prevention and the Association of Teachers of Preventive Medicine (Award Number TS 247–14/15); its contents are the responsibility of the authors and do not necessarily reflect the official views of the Centers for Disease Control and Prevention or the Association of Teachers of Preventive Medicine.
- Received September 6, 2000.
- Accepted November 30, 2000.
Reprint requests to (R.K.Z.) Department of Family Medicine and Clinical Epidemiology; University of Pittsburgh School of Medicine; M-200 Scaife Hall; Pittsburgh, PA 15261. E-mail:
This paper won first place in a presentation at the American Academy of Family Physicians Scientific Assembly, Dallas, Texas in September 2000 and was also presented in part at the 34th National Immunization Conference in Washington DC in July 2000.
- VFC =
- Vaccines for Children Program •
- CHIP =
- Child Health Insurance Program •
- FP =
- family physicians •
- Peds =
- pediatricians •
- CATI =
- Computer Assisted Telephone Interviewing •
- HMO =
- health maintenance organization •
- CI =
- confidence interval •
- DTP =
- diphtheria and tetanus toxoids, and pertussis vaccine
- Schulte JM,
- Bown GR,
- Zetzman MR,
- et al.
- Bordley WC,
- Freed GL,
- Garrett JM,
- Byrd CA,
- Meriwether R
- Ruch-Ross HS,
- O'Connor KG
- Arnold PJ,
- Schlenker TL
- ↵Centers for Disease Control and Prevention. Vaccines for Children (VFC) program overview. Available at: www.cdc.gov/nip/vfc/about.htm Accessed June 1, 2001
- ↵Centers for Disease Control and Prevention. Vaccine Price List. National Immunization Program. Centers for Disease Control and Prevention, Atlanta. Available at:http://www.cdc.gov/nip/vfc/cdc_vaccine_price_list.htm Accessed September 7, 2001
- Santoli JM,
- Rodewald LE,
- Maes EF,
- Battaglia MP,
- Coronado VG
- ↵United States General Accounting Office. Vaccines for Children: Reexamination of Program Goals and Implementation Needed to Ensure Vaccination. Washington, DC: United States General Accounting Office; 1995
- ↵Health Care Financing Administration. State child health insurance program plan activity map. Available at:http://www.hcfa.gov/init/chip-map.htm Accessed January 1, 2000
- ↵Markus AM, Rosenbaum S, Roby D. CHIP, Health Insurance Premiums and Cost Sharing Lessons from the Literature. Technical publication prepared for Health Resources and Services Administration, Department of Health and Human Services. October 1998. Available at:www.hcfa.gov/init/lesslit.pdf Accessed June 1, 2001
- ↵Aday LA. Designing and Conducting Health Surveys. San Francisco, CA: Jossey-Bass Inc; 1989
- Taylor JA,
- Darden PM,
- Slora E,
- Hasemeier CM,
- Asmussen L,
- Wasserman R
- ↵Holtmann AG. The economics of US immunization policy. In: Pauly MV, ed.Supplying Vaccine: An Economic Analysis of Critical Issues. Amsterdam, Netherlands: IOS Press; 1996:153–173
- ↵Groves RM. Survey Errors and Survey Costs. New York, NY: John Wiley; 1989
- Fowler FJ,
- Gallagher PM,
- Nederend S
- Copyright © 2001 American Academy of Pediatrics