Our assessment of New Mexico's immunization program suggests that Medicaid enrollment by families under the TANF program may have driven much of the change in childhood immunization rates. Policy changes in 2004 included automatic closure of Medicaid cases not completing the eligibility certification process and required recertification every 6 months. These system-level policy changes affected Medicaid enrollment levels adversely.
Several considerations lead us to believe that changes in TANF Medicaid enrollment played a causal role in changing immunization rates, even though analyses of statistical correlation cannot conclusively demonstrate causation: a) The results identify a compelling temporal correspondence between changes in TANF Medicaid enrollment and changes in childhood immunization levels. b) The causal direction could not plausibly have operated in the opposite direction, that is, by changes in immunization rates leading to changes in Medicaid enrollment. c) We know of no intervening or confounding variables that could account for the very high correlation between changing Medicaid enrollment and changing immunization rates.
Although the various initiatives associated with New Mexico's revitalized immunization program, such as the accelerated “Done-by-one” immunization schedule, represent important changes that aimed to achieve the state's goal of 90% coverage by 2010, these initiatives were implemented too late to be responsible for the dramatic increase in immunization coverage rates between 2001 and 2004. For example, the implementation of both the NMIC and the accelerated “Done-by-one” immunization schedule occured in the spring of 2003, too late to affect substantially the improved immunization rates reported for earlier years. Similarly, the Shot team nurse intervention program was implemented in 2004, and would not have affected rates until the following year.
We suggest, therefore, that expanded Medicaid enrollment levels for needy families played an important role for the increase in New Mexico's immunization coverage levels between 2002 and 2004. Similarly, recent dramatic decreases in TANF Medicaid enrollments contributed to the decrease in immunization rates since 2004.
Because childhood immunizations for the most part are tied to well-child visits, a change in immunization coverage levels may serve as a sentinel for the level of early childhood preventive care. Since the implementation of Medicaid managed care in New Mexico in 1997, primary care practitioners (PCPs) have provided the vast majority of immunizations during the 5 recommended well-child visits before 24 months of age, rather than public health clinics, as was the common practice prior to 1997. For example, in 1997, 39.7% of the nearly 24,000 children in Bernalillo County (New Mexico's largest county) were vaccinated at 1 of 5 public health offices operated by the state's Department of Health (DOH); by 2000 this percentage had dropped to 4.9%. Currently, approximately 2.5% of Bernalillo County's children less than 3 years of age are immunized at public health offices (unpublished data, New Mexico DOH). Consequently, PCPs employed by the managed care organizations that provide preventive care for Medicaid children play an important role in maintaining or improving immunization rates for Medicaid children. Because of New Mexico's high proportion of children enrolled in Medicaid, policy changes that reduce Medicaid enrollment also reduce the likelihood that PCPs will provide adequate preventive care, including immunizations.
Based on our analysis, we believe that increasing the proportion of Medicaid-eligible children who are enrolled in Medicaid and, importantly, who are assigned a PCP, likely affects overall immunization rates, and other aspects of preventive care 
, in a positive way. Our previous study examining the effects of Medicaid managed care on immunization rates did not recognize this relationship 
. The previous study identified 3 probable causes for declining immunization rates between 1997 and 2002: 1) reduced funding for immunizations at public health offices, 2) informal referrals by Medicaid providers to community health centers and public health offices, and 3) increased workloads at community health centers. That study also suggested that unanticipated and adverse consequences can result from health policy interventions in a complex health system. Our current findings suggest that declining enrollment in the TANF component of Medicaid also may have led to an unanticipated consequence of reduced immunizations.
Several researchers have studied the impacts of decreased TANF enrollment in other states 
. Although people who leave TANF usually remain eligible for Medicaid, confusion regarding eligibility and time limits for cash benefits led to a high proportion of uninsurance 
. In Oregon, for example, 40% of TANF recipients who were disenrolled, including approximately 15–30% of their children, became uninsured after a 1-year transitional Medicaid program ended 
. Restricting TANF enrollment, therefore, likely represents a barrier to childhood preventive care because children and their families lose their PCPs and medical homes for primary care; as a result, up-to-date status for childhood immunizations declines 
Limitations of the study
The results of our analysis are subject to several limitations. Because individual-level or aggregated data on immunization coverage by insurance status are not available, we were not able to assess directly the affects of Medicaid enrollment on immunization coverage. In addition, research has indicated that the correlation coefficient based on aggregated data can sometimes produce a biased estimate of individual level correlation 
. It is important to consider, however, that independent of purely quantitative measures, the visible correspondence between immunization rates and Medicaid enrollments is compelling. It seems unlikely that this correspondence is coincidental.
As reviewed by Burns et al.,
there are a number of important potential barriers faced by those needing immunizations in addition to insurance status. These barriers include confusion about vaccination schedules, fears about vaccine safety, transportation problems, and inconvenience of the immunization process (e.g., inconvenient clinic hours and long wait times)
. In their review, Burns et al.
cite an example from Pennsylvania, where after an outbreak of Hib disease, the most commonly cited reason by parents for not having their children immunized was that immunizations were simply not a priority compared to the other activities of life 
. In that same study, 73% of the parents also indicated they would immunize their children if vaccinations were offered locally 
. Although our study did not consider these other barriers, it is important to consider that these factors would likely not generate the temporal pattern in immunization rates that we observed. For example, it seems unlikely that there were temporal trends in transportation problems or parental concerns regarding vaccine safety similar to those observed for immunization coverage and TANF Medicaid enrollment between 1998 and 2005.
In addition to the barriers listed above, shortages in vaccine supply can also affect immunization rates 
. National vaccine shortages for at least one vaccine (pneumococcal conjugate vaccine [PCV]) in 2004 
conceivably may have corresponded with the drop in immunization coverage rates in New Mexico during 2004. However, PCV was not among the vaccines comprising the 4
3 or 4
3 series examined by our study.
Future research on Medicaid and immunization coverage in poor and rural states such as New Mexico would benefit from the collection of additional survey data at the individual level. Survey instruments should include questions to parents regarding insurance status and other socio-economic information, as well as perceived barriers to immunizing their children such as problems of Medicaid eligibility.
Subject to certain limitations, our study suggests that deteriorating childhood immunization coverage accompanied state-level changes in welfare policy in New Mexico. As in our previous study 
, the current findings illustrate how unanticipated consequences can follow policy changes in a complex health care system. The prior improvement in immunization coverage in New Mexico, attributed by CDC and others to initiatives by state government to improve immunization practices, probably reflected expansion in Medicaid enrollment. Our research underscores the importance of expanding and maintaining Medicaid enrollment as a key component of efforts to improve immunization coverage as an indicator of public health standards.