Influence of Insurance Status and Vaccine Cost on Physicians’ Administration of Pneumococcal Conjugate Vaccine

Design

The study was a cross-sectional mail survey of pediatricians and family physicians who provide primary care to children ≤5 years of age. The protocol was approved by the Institutional Review Boards of the University of Michigan Medical School and the CDC.

Sample

We drew random samples of 833 pediatricians and 788 family physicians from the American Medical Association Physician Masterfile through a contracted vendor (Medical Marketing Services). The Masterfile is the most comprehensive listing of physicians in the United States and contains both members and nonmembers of the American Medical Association. Physicians’ specialty, gender, year of graduation from medical school, and board certification status were drawn from the Masterfile.

We sampled physicians from 24 states, selected for their geographic diversity and vaccine financing strategy regarding PCV7. Twelve states had a VFC-only strategy (AR, CA, IN, MS, NJ, NV, PA, UT, VA, VT, WV, and WI), 7 had an enhanced-VFC strategy (GA, MD, MI, MN, NY, OK, and TX), and 5 had a universal purchase policy regarding PCV7 (ME, NH, NM, NC, and WA), chosen to represent in relative quantities the number of states overall with each of the 3 different vaccine financing strategies regarding PCV7 in spring 2001, as identified by the CDC (A. Vargas-Rosales, personal communication, 2001). Physician sampling was conducted from each state proportional to physician representation in the Masterfile, ie, states with relatively larger numbers of physicians in the Masterfile had proportionally greater numbers of physicians in the sample.

Excluded from the sampling frame were physicians ≥70 years old and resident physicians. Physicians were ineligible when they did not provide primary care to children ≤5 years old, served in the armed forces medical services, or were retired or otherwise no longer engaged in routine primary care practice.

Survey Instrument

Physicians who currently recommend PCV7 for children ≤23 months old (for whom the vaccine is universally recommended by ACIP²¹) were asked to estimate the percentage of children in their practice for whom purchase of PCV7 is paid for by private health insurance, VFC, or a state vaccination program (“insurance coverage”). They were then asked to specify the site at which they administer PCV7 for children with and without insurance coverage: to be given in their own practices, to be given at a local public health clinic, or not recommended at all. Physicians were also asked whether and how the cost of PCV7 has influenced their purchase and administration of the vaccine and how their overall effort to implement PCV7 recommendations compares with their efforts for previously recommended vaccines. Physicians’ attitudes about vaccination and their expectations about the effectiveness of PCV7 and their practice characteristics (size of practice, ownership/affiliation, annual newborn enrollment, and proportions of Medicaid beneficiaries and black patients) were also collected through the survey.

Survey Administration

Before mailing, survey items were pilot-tested with family physicians and pediatricians. The survey was initially mailed with a personalized cover letter in April 2001. Nonrespondents were contacted with additional letters and surveys in May and June 2001.

Data Analysis

All responses received by the study closure date of July 15, 2001, were included in the analysis. All analyses were conducted using SAS (version 6.12; SAS Institute, Cary, NC). Physicians’ responses regarding the effects of children’s insurance coverage on their PCV7 recommendations and their perceptions of cost were analyzed with respect to demographic and practice variables, state vaccine financing strategy, and physician attitudes regarding vaccinations. Tests of significance in bivariate comparisons were computed using likelihood ratio χ², except in cases of 0 values in 1 or more cells, when Fisher exact test was used.

Characteristics of Respondents

Among 1621 physicians in the original sample, 507 of 833 pediatricians ([PD]; 61%) and 458 of 788 family physicians ([FP]; 58%) responded, for an overall response rate of 60%. Respondents and nonrespondents did not differ significantly with respect to gender, age, year of graduation from medical school, board certification, state in which they practice, or state vaccine financing strategy. Some respondents were ineligible for analysis because they did not see children younger than 5 years (53 PD, 157 FP), because they had retired (5 PD, 16 FP), or for other miscellaneous reasons, such as having a practice in the uniformed services or a suspended or revoked license (14 PD, 26 FP). Thus, our sample contained 435 PD and 259 FP who were eligible for analysis. Of these physicians, 40% are women. Although 26% graduated from medical school within the past 10 years, 11% graduated >30 years ago. The vast majority (85%) are board certified in their clinical specialties. Approximately 45% practice in states with a VFC-only strategy for PCV7 purchase, 44% in states with enhanced VFC financing, and 11% in states with universal purchase regarding PCV7.

At the time of the survey, 605 (87%) of respondents had adopted PCV7 recommendations for children younger than 2 years. Characteristics of FP who had adopted PCV7 recommendations versus those who had not are reported elsewhere.²⁰ These physicians who routinely recommend PCV7 compose the analytic sample for this study, to which all of the subsequent findings pertain. Many practice characteristics of physicians who recommend PCV7 to their patients were similar across states with different vaccine financing strategies (Table 1), with the exception that physicians who practice in universal purchase states reported higher proportions of patients on Medicaid.

Effects of Insurance and Vaccine Financing Strategy on PCV7 Administration

More than half (56%) of the physicians who recommend PCV7 reported that all children in their practices had insurance coverage for PCV7. The proportion of physicians who reported complete coverage for PCV7 in their practices was higher in universal purchase states than in enhanced VFC and VFC-only states but not to a statistically significant extent (Table 2).

For children who have PCV7 insurance coverage, 98% of physicians said that they would administer the vaccine in their own practices. For children without PCV7 insurance coverage, physician responses differed significantly by proportion of children with coverage for PCV7 in their practices, as well as by the PCV7 vaccine financing strategy of the state in which the physician practices (Table 3). Physicians who reported higher rates of PCV7 insurance coverage overall were significantly more likely to administer PCV7 in their own practices rather than refer children to public health clinics, as were physicians who practice in either enhanced VFC or universal purchase states. Moreover, 10% of physicians whose practices have the lowest proportion of children with PCV7 insurance coverage indicated that they would not recommend the vaccine for children without coverage.

There were no significant differences in referral patterns among physicians of different specialties, those practicing in practices of differing sizes or ownership, or physicians with differing proportions of black patients or Medicaid beneficiaries. However, physicians who annually enroll >40 newborns were significantly more likely to administer PCV7 in their own practices to children without PCV7 insurance coverage than were physicians who enroll fewer newborns (63% vs 48%; P < .05).

Effects of PCV7 Cost Concerns on Physician Behavior

Almost one third (29%) of physicians who had adopted PCV7 recommendations reported that the cost of PCV7 influences their purchase and administration of the vaccine. The influence of PCV7 cost on physician behavior varied with several physician practice characteristics (Table 4). FP were significantly more likely than PD to report cost concerns, as were physicians who practice in VFC-only states and physicians who reported the lowest levels of insurance coverage for PCV7 in their practices. In addition, respondents who work in practices with fewer than 4 physicians, physicians in independent private practice, and physicians who enroll fewer than 40 newborns in their practices each year were significantly more likely to be concerned about the cost of PCV7. Proportions of patients who are black or Medicaid beneficiaries in physicians’ practices were not associated with PCV7 cost concerns.

Nearly all (96%) physicians who reported PCV7 cost as a concern said that they would administer PCV7 in their practices to children who have insurance coverage for the vaccine. In contrast, only 37% would administer the vaccine in their practices to children without PCV7 insurance coverage, compared with 55% of physicians not concerned about the cost of PCV7 (P < .001).

Physicians with PCV7 cost concerns were more likely to have made adjustments to their practice to compensate. Compared with previously recommended vaccines, physicians who are concerned about PCV7 cost were significantly more likely than those without cost concerns to report that they now more often discuss reimbursement with insurers (43% vs 19%), screen patients for insurance coverage (52% vs 21%), take a financial loss by administering the vaccine (31% vs 11%), and have increased overhead expenditures to stock adequate doses of the vaccine (48% vs 19%; all comparisons significant at P < .001).

Of note, physicians’ expectations of the effectiveness of PCV7 in preventing severe pneumococcal disease and otitis media were similar among physicians with and without cost concerns. Physicians who had experienced a patient death or severe sequelae attributable to pneumococcal meningitis or sepsis were significantly less likely to report PCV7 cost concerns (24% vs 35%; P < .05).

A Step Backward?

These findings indicate a possible step backward from the progress made in the years immediately after the implementation of the VFC program in 1994. A study of primary care physicians in New York State found that physician referral rates for immunizations declined from 51% in 1993 to 18% in 1997.¹¹ Analysis of National Immunization Survey data from 1997 indicated that only 25% of US children had >1 immunization provider for the primary immunization series.²²

In this 24-state sample, we found that nearly all physicians who had adopted national PCV7 recommendations were administering PCV7 in their own practices to children who have insurance coverage for the vaccine. In contrast, referral rates for children without coverage to public health departments varied by proportion of patients with coverage for PCV7 in physicians’ practices and ranged from 32% to 41%. Although this is the first study to characterize referral rates for PCV7 specifically, referral rates that affect 1 in 3 children raise substantial concern that lack of insurance coverage for PCV7 is serving as a barrier to immunization in the medical home for many children.

Our analysis offers additional insight into physician referral patterns for children without insurance coverage for immunizations: physicians who had lower proportions of children with insurance coverage were more likely to refer than physicians with higher proportions of insurance coverage. This suggests that a child without PCV7 insurance coverage has a better chance of receiving the vaccine at her or his medical home when most of the other children at that same practice have coverage for the vaccine. Conversely, a child who does not have PCV7 insurance coverage and is seen at a practice where many other children also lack insurance coverage is more likely to be referred to a public health clinic or may not be recommended the vaccine at all. Generalization of physician behavior regarding referral for immunizations in this manner has not previously been described and merits additional investigation.

Referrals for PCV7: A 2-Tiered System

Referral rates are just 1 indication that physicians—particularly in family practice—seem to be grappling with several issues pertaining to the comparatively high cost of PCV7. Physicians with fewer newborns enrolled annually, with fewer practice partners, and in independent private practice were more likely to say that PCV7 cost had adversely affected their acquisition and administration of this vaccine compared with previous vaccines. This suggests that economies of scale in larger practices that serve younger patients and in practices that are owned by or affiliated with larger institutions facilitate the incorporation of PCV7 into practice.

The ultimate implication of physicians’ cost concerns and referrals for children without insurance coverage is a 2-tiered system for PCV7 administration. Children with insurance coverage are more likely to receive the vaccine at their usual sites of primary care, and children without insurance coverage may remain unimmunized because they are more likely to be referred to public health clinics that they may not subsequently visit.²² If a 2-tiered system for PCV7 persists in the years ahead, then disparate rates of morbidity and mortality from pneumococcus may emerge among different communities and potentially among states with different vaccine financing strategies as well. That said, it is reassuring in this sample that referral rates to public health clinics for immunizations were not associated with the proportions of black children or Medicaid beneficiaries in physicians’ practices. These are 2 groups who often face disparate access to health care but whose needs may be more equally met for PCV7 because of the identification of black children as a high-risk group for pneumococcal disease in ACIP recommendations¹⁹ and because the purchase of all ACIP-recommended vaccines for all child Medicaid beneficiaries is funded through the federal VFC program.⁷

It is concerning that only approximately two thirds of physicians in universal purchase states—in which the intention is to purchase enough PCV7 to supply the vaccine to all children, regardless of insurance status—reported that all children in their practices had insurance coverage for PCV7. This lower-than-expected proportion may reflect delays in establishing the universal purchase mechanism for PCV7 in the sample states within the first year that the vaccine was recommended by ACIP.

Limitations

Our analysis should be interpreted with some caveats. Our response rate was similar to those in other national mail surveys of physicians.^23,24 Although our respondents and nonrespondents had similar demographic characteristics, we do not know how they differed with respect to their concerns about PCV7. In addition, we did not ask physicians about their referral patterns for children without insurance coverage for other ACIP-recommended vaccines, to compare with their PCV7 referral rates. It is possible that physicians who care for children without insurance coverage for other vaccines refer those children elsewhere at the same rates as for PCV7. Physicians’ concerns about PCV7 costs in comparison with previously recommended vaccines, however, suggest that PCV7 referral rates are likely as high as or higher than those for other vaccines. Over time, increasing referral rates to public sites will strain limited federal and state funds for vaccine purchase for underinsured children and may jeopardize continued provision of vaccines to underinsured children through public programs.

Another caveat is that physician PCV7 referral rates for children without coverage may be overestimates of actual referral patterns, because parents may opt to pay out of pocket for the vaccine rather than have their children receive vaccines in >1 site. Nevertheless, the comparatively high price per dose of PCV7 makes this possibility less likely than for other ACIP-recommended vaccines.

An additional limitation is that physicians were asked to estimate the PCV7 insurance coverage of children in their practices. Although it is possible that physicians may have over- or underestimated the true rate of PCV7 coverage, the focus of our analysis is their behavior based on their perceptions of PCV7 coverage. There may be opportunities to improve physicians’ knowledge about their patients’ insurance coverage for vaccines that would influence physicians’ referral patterns.