Protecting Low-Income Children's Access to Care: Are Physician Visits Associated With Reduced Patient Dropout From Medicaid and the Children
http://www.100md.com
《中华首席医学网》
ABSTRACT
OBJECTIVE. Dropout among patients who are enrolled in Medicaid and the Children's Health Insurance Program contributes to a lack of health care access among millions of Americans. The purpose of this study was to determine which, if any, types of clinical contact with physicians are associated with reduced dropout among children who are enrolled in Medicaid and the Children's Health Insurance Program.
METHODS. The data are from the nationally representative Medical Expenditure Panel Survey, 1998–2002. The sample is composed of all children (n = 3043) who were reported to have Medicaid or Children's Health Insurance Program coverage throughout their first year in the survey and who did not acquire other insurance during the study period. The outcome measure is whether an individual remained enrolled in Medicaid or the Children's Health Insurance Program by the end of the following year. Exposure variables were clinical contact during an individual's first year in the survey: numbers of office visits, hospital outpatient department visits, emergency department visits, inpatient hospital stays, and dental visits. The analysis uses multivariate logistic regression to control for patient and family characteristics—most important, health status, functional status, and overall health care expenditures.
RESULTS. Eight percent of the children in the sample had left Medicaid/the Children's Health Insurance Program by the end of the second year in the survey. More frequent contact with clinicians in an office setting was associated with a significantly lower risk for dropping out of Medicaid/the Children's Health Insurance Program among children, even controlling for demographics, health and functional status, and overall health care expenditures. After multivariate adjustment, more frequent contact in hospital outpatient departments also was associated with reduced dropout, with a borderline statistically significant odds ratio. Notably, emergency visits and inpatient stays were not associated with any significant change in the risk of Medicaid/Children's Health Insurance Program dropout.
CONCLUSIONS. These results suggest that some but not all types of clinician visits are serving an important function in maintaining Medicaid and the Children's Health Insurance Program coverage among low-income patients. Two possible approaches to improve access to care among low-income children therefore would be (1) increased awareness among clinicians, especially in hospitals and emergency departments, regarding Medicaid/Children's Health Insurance Program retention as an issue in the ongoing care of their patients and (2) Medicaid/Children's Health Insurance Program reimbursement of clinicians and their staff for assisting patients with the public insurance renewal process.
Key Words: Medicaid ? CHIP ? health insurance ? access to health care ? health policy
Abbreviations: CHIP—Children's Health Insurance Program ? MEPS—Medical Expenditure Panel Survey ? ED—emergency department ? SSI—Supplemental Security Income
At the reception desk of an outpatient clinic in a Boston teaching hospital, a sign is written in large bold letters: "Reminder for Patients on MassHealth," it calls out to families who are enrolled in Massachusetts's version of Medicaid and the Children's Health Insurance Program (CHIP). "Every year, you need to tell the State you still want MassHealth. The state mails you a short form that you need to fill out and send back in. If you need help, bring the form to your health center or hospital as soon as possible."
As this sign suggests, keeping eligible patients enrolled in Medicaid and CHIP is an important step in maintaining health care access for millions of Americans, yet nearly 45 million Americans (15.6% of the population) did not have any health insurance in 2003, including 8.4 million children, roughly 1 child in 9.1 Millions of these individuals in fact are eligible for public insurance through Medicaid or CHIP.2 Whereas some of these individuals may never enroll in these programs at all, others indeed have enrolled but then drop out, typically by failing to complete the eligibility review process that is required at least once annually in every state. One recent study indicated that as many as 3 million children leave these public programs each year and become uninsured, despite continuing eligibility.3 As the sign at the hospital's reception desk makes clear, patient disenrollment from public insurance is a key concern in providing quality care to needy patients.
A large body of research indicates that health insurance coverage is critical to children's health. For instance, children without insurance or with discontinuous coverage experience lower rates of checkups, vaccination, and follow-up care; face more frequent restrictions on activities as a result of poor health; and are more likely to go without needed care when ill.4–6 Clearly, retention of eligible and needy children in Medicaid and CHIP should be a priority for both clinicians and policymakers.
Previous research has suggested a role for physicians in retaining eligible individuals in Medicaid/CHIP. County-level physician participation in CHIP has been shown to be protective against childhood disenrollment.7 Similarly, more generous state Medicaid physician reimbursement rates are associated with lower dropout rates, presumably because of either increased provider participation in Medicaid or a greater incentive for physicians to keep patients enrolled in public insurance.3 In a single-state survey of families with children who were eligible for a public insurance program similar to CHIP, parents, particularly in non–English-speaking families, reported that doctors and clinic staff are important sources of information about the enrollment process.8 Furthermore, community health centers have been found to play an important role in Medicaid and CHIP outreach to eligible children.9 But none of these studies has examined the direct impact of exposure to health care providers on Medicaid/CHIP retention. This study, using nationally representative data from the Medical Expenditure Panel Survey (MEPS), tested the hypothesis that clinical contact is associated with lower dropout from Medicaid/CHIP among children.
METHODS
The data are from the household component of the MEPS for the years 1998–2002. The MEPS is a nationally representative survey that is conducted by the Agency for Healthcare Research and Quality and features 5 rounds of interviews per household over a course of 2 years. The sample therefore contains 4 overlapping 2-year cycles: 1998–1999, 1999–2000, 2000–2001, and 2001–2002. The survey includes detailed questions on health insurance coverage, type and frequency of health care utilization, health status, and demographics.
The dependent variable is disenrollment from Medicaid/CHIP, determined by whether an individual is still enrolled in Medicaid or CHIP by the end of his or her second year in the survey. Note that the MEPS does not distinguish between Medicaid and CHIP and simply combines them in the survey. Medicaid by far is the larger program, with 24 million children compared with 4 million in CHIP, as of 2003.10,11 In roughly one third of states, Medicaid and CHIP actually are run as a single combined program, with the remaining states administering separate programs; previous research suggested that the choice of program structure can affect dropout and that there may be differences in dropout between Medicaid and CHIP,12 but the MEPS does not allow for analysis of these questions. Therefore, this study explored the associations between clinical contacts and dropout from children's public insurance in general, comprising both Medicaid and CHIP.
The key independent variables are the measures of clinical contact for the first year in the survey: number of nonhospital office-based visits, number of visits to hospital outpatient departments, number of emergency department (ED) visits, number of inpatient stays, and number of dental visits. Note that these are the visit types classified by the MEPS; the difference between "office" and "outpatient" visits is determined solely by whether the visit occurred at a hospital, without respect to the kind of services provided. The effect of these exposure variables on Medicaid/CHIP disenrollment was estimated using multivariate logistic regression.
For studying the impact of health care utilization on retention, several important threats to validity must be addressed. First, health care coverage is a strong predictor of health care utilization, so we must be careful to avoid reverse causation in measuring this relationship. To ensure that this analysis is identifying the effect of utilization on coverage, rather than the reverse, the study sample consists only of those individuals who were reported to have Medicaid/CHIP coverage for the full first year in the survey. The exposure is based only on health care utilization during this first year, and the outcome, Medicaid/CHIP retention, is based only on coverage during the second year. Therefore, we are guaranteed that the exposure predates the outcome. Furthermore, this approach ensures that all children in the sample have equal coverage during the exposure period, meaning that differential access on the basis of insurance coverage in the first year will not be an issue. With the exception of dental care, all types of health care utilization that are being used as independent variables are required by federal regulations to be covered under Medicaid and CHIP; therefore, variation in Medicaid and CHIP generosity across states also should not be a source of bias in this analysis.13
The second chief concern is that health status and preferences for health care utilization both are potential confounders. Health or functional status may confound the effect of utilization on coverage, because sicker individuals or those with disabilities will use more services in year 1, and their condition provides a strong incentive to remain enrolled in year 2. Therefore, the multivariate regression includes several distinct measures of health and functional status: self-reported health status (a 5-item scale ranging from "excellent" to "poor"); disability, based on receipt of Supplemental Security Income (SSI); whether a child has had any school- or work-related limitations as a result of medical, cognitive, or behavioral problems; and whether a child has ever received a diagnosis of asthma (the most common chronic disease of childhood) or diabetes (another increasingly common condition among children that is included in the MEPS data set).
Related to health and functional status but not identical to it is a person's overall tendency toward health care utilization. If some individuals or their parents are high-demand consumers of health care, even controlling for health status, then it is likely that such individuals would have more frequent health care contacts in year 1 and be more likely to stay enrolled in Medicaid/CHIP in year 2 as a result of this preference for health care services. To control for this effect, the regression includes an all-purpose measure of health care utilization: the total medical expenditure (from all sources, both insurance and out-of-pocket payments) for year 1. Therefore, the regression identifies the effect of clinical contacts in particular, measured in visits, controlling for the overall tendency toward health care utilization, measured in dollars.
Another factor that needs to be addressed in the study design is that disenrollment from Medicaid and CHIP is not necessarily a problem, depending on a person's reason for leaving public insurance. Previous research categorized disenrollment into 3 categories: lost eligibility, acquisition of new insurance, and "dropout" (when eligible individuals leave public insurance and become uninsured).3 The sample therefore excludes disenrollees who acquire other health insurance in year 2, and the analysis adjusts for key factors in Medicaid and CHIP eligibility (discussed next), to focus on dropout more specifically, which is the type of disenrollment on which provider outreach potentially would have the greatest impact.
The public insurance eligibility variables for children are SSI/disability, total family income, family size, and poverty status based on the percentage of the federal poverty line. Although this list of variables provides an imperfect method for adjusting for program eligibility, the publicly available MEPS data do not identify the state of residence, preventing more detailed analysis of eligibility. However, previous research indicated that among children who disenroll and do not acquire other insurance, fewer than 1 in 10 in fact have lost eligibility, indicating that this limitation should not bias the analysis significantly.3
The analysis was estimated by a survey-weighted logistic regression, using Stata 7.0, to take into account the complex survey design of the MEPS.14,15 Regression standard errors (SEs) are clustered at the household level to adjust for the nonindependence of observations within the same household. Various other factors that may affect Medicaid and CHIP retention also are included as covariates in the regressions: race, ethnicity, gender, language in which the MEPS interview was conducted, age, the highest level of parental education, the Medicaid/CHIP coverage status of family members, and the geographic region of the country. Overall, the sample consists of 3043 children (aged 0–18) who were covered by Medicaid or CHIP for the full 12 months of their first year in the survey and who did not acquire other health insurance in the second year.
RESULTS
Table 1 presents summary statistics for the study sample, stratified by enrollment behavior. Notably, 8.0% of the sample left public insurance and became uninsured by the end of the second year, the outcome of interest. This figure is slightly lower than other estimates of dropout, but this is not surprising because the sample is limited to those who were enrolled in Medicaid/CHIP throughout the entire first year, a more stable subset of enrollees than the general Medicaid/CHIP population. Not presented in Table 1 but also of note is that an additional 4.8% of children had at least temporarily gone without Medicaid/CHIP coverage at some point during the second year before reenrolling by year's end.
Table 1 indicates that the statistically significant differences between children who disenrolled and those who remained on public insurance were that disenrollees were older on average, were less likely to have diabetes, had less educated parents, were less likely to be receiving SSI, and were more likely to live in the South. In secondary analyses that are not presented in Table 1, poorer health, previous diagnosis of diabetes or asthma, and the presence of school/work limitations all were strongly associated with health care utilization, as expected.
Table 2 presents the results for the relationship between year 1 clinical contacts and year 2 dropout. The unadjusted odds ratio was statistically significant for "3 or more office visits," indicating that more frequent contact with a clinician in an office setting was associated with a reduced risk for dropout. These results essentially were unchanged when adjusting for the covariates in Table 1, most importantly, health and functional status, Medicaid/CHIP eligibility measures, and year 1 total health care expenditure. After multivariate adjustment, the only other significant result was a lower risk for dropout among children with multiple visits to a hospital outpatient department, but this result was significant only at a level of P < .10. None of the other types of clinical contact—inpatient, ED, or dental visits—was a significant predictor of dropout. Converting these adjusted odds ratios into predicted probabilities, children with >2 office visits in year 1 were 48% less likely to drop out of Medicaid/CHIP than those with no office visits, and children with >1 hospital outpatient department visit were 76% less likely to drop out than those with no outpatient visits.
DISCUSSION
Among children who were enrolled in Medicaid or CHIP for at least 12 months, exposure to clinicians in an office setting was a strong and significant predictor of decreased dropout from public insurance. Exposure to clinicians in a hospital outpatient setting also was associated with decreased dropout, although only at the P < .10 level. Meanwhile, other kinds of health care utilization and clinical contact, notably inpatient stays and ED visits, had no apparent effect on dropout. These results suggest, although they do not prove a causal effect, that clinicians can help to protect the continuity of insurance coverage for their Medicaid and CHIP patients. Given providers' own financial stake in getting reimbursed for care, as well as the chance to protect the health care access of their patients, this result is reasonable and encouraging. Furthermore, these findings are consistent with previous analyses that documented a positive association between retention and both physician participation and reimbursement rates in public insurance,3,7 but whereas previous studies could point to this association only at a county or a state level, this analysis now provides evidence at the individual patient level that clinical contact is associated with reduced dropout.
Why were only certain kinds of clinician contact, namely in the office setting and possibly in hospital outpatient departments, associated with reduced dropout? This likely reflects 2 aspects of the clinical encounter that differ across the location and the type of health care service: the extent of outreach efforts and the nature of the patient–doctor relationship. First, it simply may be that offices and outpatient departments are more familiar with the Medicaid/CHIP renewal process or devote more time to helping patients with renewal paperwork. The complexity and the severity of patient conditions in the ED and inpatient settings may make insurance issues a lower priority, despite the larger expenditures involved. These explanations are consistent with previous research in which parents reported that they are more likely to receive assistance with insurance paperwork from office-based pediatricians and clinic staff than from inpatient departments or EDs.8 In terms of dental care, many Medicaid and CHIP beneficiaries do not receive any dental coverage, which clearly would minimize the financial incentive for dentists to devote time or staff resources to Medicaid/CHIP retention.16
The second factor that likely differs across sites is the nature of the patient–doctor (or, more generally, the patient–provider) relationship. Office-based visits and, to a lesser degree, hospital outpatient visits are much more likely to be part of an ongoing relationship among a patient, a patient's family, and a clinician, compared with ED visits or inpatient stays. This kind of relationship presumably increases the likelihood that a patient's parents would feel comfortable raising issues related to insurance status and also makes it more likely that the physician and staff would know which patients need to renew their Medicaid or CHIP coverage. Furthermore, repeated visits may be beneficial simply because the renewal process involves collecting paperwork (paycheck stubs, tax returns, etc) and filling out forms, which means that 1 visit often may not be enough to complete the process successfully. Last, given the finding that nearly 5% of the children in this study who indeed were covered by the end of the year and nonetheless had experienced a temporary lapse in coverage earlier in the year, it seems likely that recurring office visits provide an excellent opportunity for providers to identify children whose coverage has already lapsed and to reenroll them promptly. Providers who do so not only protect their patients' coverage but also may be eligible for retroactive Medicaid/CHIP reimbursement for services already provided.
What are the implications of these findings? At the policy level, this study suggests that states should view clinicians and their staffs as key intermediaries in the retention of needy and eligible individuals in public insurance. To take advantage of this intermediary role, states that aim to reduce dropout from Medicaid and CHIP should consider adopting a specific billing code for staff and/or clinician assistance with the renewal process. Given the significant administrative costs to states of Medicaid and CHIP turnover, paying clinicians to keep eligible individuals enrolled actually could be a cost-saving measure, not to mention the significant benefit that improved retention would provide in terms of health care access.17 Furthermore, state outreach regarding Medicaid and CHIP renewal should target not just enrollees but also clinicians and their staff. Sending clinicians information once a year to explain the renewal process or simply to remind them that a renewal requirement exists will make it more likely that clinicians will be able to help their patients with this process.
At the level of individual physicians and other providers, the implications of the study are twofold. First, clinicians can have a major impact on Medicaid and CHIP retention. Second, not all clinicians are doing so. ED visits and inpatient stays currently are missed opportunities for promoting public insurance retention, and, for certain patients, these visits may be their primary contact with the health care system. Therefore, more proactive steps to improve outreach in these settings regarding Medicaid/CHIP retention should be strongly encouraged. What can physicians do? On the basis of this study, one can only speculate as to which specific approaches by clinicians and their staffs would be most beneficial in terms of public insurance retention, but patient surveys show that most families drop out of Medicaid and CHIP because they do not know that they need to renew or they do not understand the process.18,19 Therefore, if doctors' offices or hospital departments can afford a staff member or social worker to deal with these issues, then that undoubtedly would make a significant impact and might even be a profitable approach for practices that are losing money by providing uncompensated care to Medicaid-eligible but uninsured children. In the inpatient setting, including a question about Medicaid/CHIP renewal in standard discharge instructions to patients' parents could be an easy but effective approach. Even steps as simple as posting a Medicaid/CHIP renewal notice at the reception desk in offices and the ED, like the sign at the beginning of this article, may make a difference.
Of course, there are several limitations of this analysis that must be addressed. The most significant threats to internal validity were discussed already in "Methods": the possibility of reverse causation between insurance coverage and health care utilization, and the confounding effects of health status and a tendency to use health care services. In terms of reverse causation, the study design eliminates this problem, because all of the clinical contacts that were used as independent variables occurred in the year before the outcome variable, Medicaid/CHIP dropout, was assessed. The possibility of confounding cannot be eliminated, however, even with multivariate adjustment. Although several distinct measures of health and functional status, as well as a composite measure of the demand for health care services (total medical expenditure), were used it is possible that they may not capture fully the differences in these characteristics that could bias the results. However, if it actually were unobserved differences in health status that were driving the results, then one would expect that the clinical services that are most indicative of severe disease—ED visits and inpatient stays—would be the strongest predictors of retention. Yet, in fact, these 2 types of services had no predictive power regarding Medicaid/CHIP retention. Similarly, if unobserved preferences for consuming health care in general were the underlying factor, then there should be an effect of all types of utilization on retention, rather than just office and outpatient visits. Last, if the key confounder actually were a preference for preventive health care, rather than for all types of services, then one still would expect to see significant effects for dental visits, as well as physician office visits, but dental care was not a significant predictor.
An additional limitation of the study relates to the generalizability of the findings. Although the sample is nationally representative and therefore the findings cannot be attributed simply to the idiosyncrasies of any particular state's Medicaid or CHIP programs, the sample is limited to a subset of public insurance enrollees. The children in this study all were enrolled continuously in Medicaid or CHIP for their full first year in the survey. This approach had the advantage of ensuring that all children had the same access to care during the first year, when the key independent variables of clinical exposure were measured; but it has the disadvantage of restricting the analysis to a more stable subset of the public insurance population. It may be that the associations between dropout and provider visits in this group are significantly different from those among children with shorter and less stable periods of public insurance enrollment. However, if anything, it seems likely that there would be even more room for provider encounters to improve the retention rates of such children, who are at higher risk for dropout to begin with.
Last, the study was limited in its ability to differentiate subtypes of clinical contact on the basis of the available data. Although the MEPS reports the location of each visit, it does not distinguish between private practices and community health centers, between private and public hospitals, and between primary and specialty care. Each of these distinctions would be valuable in determining precisely which kinds of clinical encounter and setting are more strongly associated with reduced dropout.
CONCLUSIONS
Clinicians can serve a critical role in promoting the retention of Medicaid and CHIP insurance coverage among needy patients. Children with more frequent contact with office-based providers were much less likely to drop out of public insurance and become uninsured. Policymakers should craft retention strategies that encourage clinician assistance with Medicaid and CHIP renewal, and providers should take steps to help protect the public insurance coverage of their patients. It may be as simple as posting a reminder—"Medicaid and CHIP patients: Have you renewed your eligibility yet this year?"—that could make the difference between continued access to care and the loss of health insurance for patients in need.
ACKNOWLEDGMENTS
This project was completed while the author was supported by graduate fellowships from the National Science Foundation and the MD/PhD Program in the Social Sciences at Harvard Medical School.
Many thanks to Alan Zaslavsky, PhD, Joe Newhouse, PhD, and Melissa Wachterman, MD, for helpful suggestions.
The author had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
FOOTNOTES
Accepted Jan 3, 2006.
Address correspondence to Benjamin D. Sommers, PhD, 60E Glen Rd, Unit T-9, Brookline, MA 02445. E-mail: bsommers@post.harvard.edu
The author has indicated he has no financial relationships relevant to this article to disclose.
REFERENCES
DeNavas-Walt C, Proctor B, Mills R. Income, Poverty, and Health Insurance Coverage in the United States: 2003. Washington, DC: US Census Bureau; 2004
Selden T, Hudson J, Banthin J. Tracking changes in eligibility and coverage among children, 6–2002. Health Aff (Millwood). 2004;23 :39 –49
Sommers BD. From Medicaid and CHIP to uninsured: drop-out among children in public insurance programs. Health Serv Res. 2005;40 :59 –78
Olson LM, Tang SS, Newacheck PW. Children in the United States with discontinuous health insurance coverage. N Engl J Med. 2005;353 :382 –391
Lave JR, Keane CR, Lin CJ, Ricci EM, Amersbach G, LaVallee CP. Impact of a children's health insurance program on newly enrolled children. JAMA. 1998;279 :1820 –1825
Newacheck PW, Stoddard JJ, Hughes DC, Pearl M. Health insurance and access to primary care for children. N Engl J Med. 1998;338 :513 –519
Phillips JA, Miller JE, Cantor JC, Gaboda D. Context or composition: what explains variation in SCHIP disenrollment? Health Serv Res. 2004;39 :865 –885
Feinberg E, Swartz K, Zaslavsky AM, Gardner J, Walker DK. Language proficiency and the enrollment of Medicaid and CHIP-eligible children in publicly funded health insurance programs. Matern Child Health J. 2002;6 :5 –18
Davidoff AJ, Garrett B. Determinants of public and private insurance enrollment among Medicaid and CHIP-eligible children. Med Care. 2001;39 :523 –535
Kaiser Family Foundation. CHIP Program Enrollment: December 2003 Update. Menlo Park, CA: Kaiser Commission on Medicaid and CHIP and the Uninsured; 2003
Kaiser Family Foundation. The Medicaid and CHIP Program at a Glance. Menlo Park, CA: Kaiser Commission on Medicaid and CHIP and the Uninsured; 2003
Sommers BD. The impact of program structure on children's disenrollment from Medicaid & SCHIP. Health Aff (Millwood). 2004;24 :1611 –1618
Centers for Medicare and Medicaid and CHIP Services. Medicaid and CHIP: a brief summary. Available at: www.cms.hhs.gov/publications/overview-medicare-MedicaidandCHIP/default4.asp. Accessed June 16, 2005
Stata Corp. Stata 7.0 Statistical Software . College Station, TX: Stata Corp; 2001
Agency for Healthcare Research and Quality. Computing standard errors for MEPS estimates. Available at: www.meps.ahrq.gov/factsheets/FS_StandardErrors.htm. Accessed March 12, 2005
Center for Policy Alternatives. Dental coverage under Medicaid and CHIP. Available at: www.cfpa.org/issues/healthcare/dental/dentalfactsh.pdf. Accessed July 15, 2003
Ku L, Ross DC. Staying Covered: The Importance of Retaining Health Insurance for Low-Income Families. Washington, DC: Center on Budget and Policy Priorities; 2002
Perry M, Kannel S, Riley T, Pernice C. What Parents Say: Why Eligible Children Lose SCHIP. Portland, ME: National Academy for State Health Policy; 2001
Hill I, Lutzky AW. Is There a Hole in the Bucket? Understanding SCHIP Retention. Washington, DC: Urban Institute; 2003
PhD Program in Health Policy, Harvard Medical School, Boston, Massachusetts(Benjamin D. Sommers, PhD)
OBJECTIVE. Dropout among patients who are enrolled in Medicaid and the Children's Health Insurance Program contributes to a lack of health care access among millions of Americans. The purpose of this study was to determine which, if any, types of clinical contact with physicians are associated with reduced dropout among children who are enrolled in Medicaid and the Children's Health Insurance Program.
METHODS. The data are from the nationally representative Medical Expenditure Panel Survey, 1998–2002. The sample is composed of all children (n = 3043) who were reported to have Medicaid or Children's Health Insurance Program coverage throughout their first year in the survey and who did not acquire other insurance during the study period. The outcome measure is whether an individual remained enrolled in Medicaid or the Children's Health Insurance Program by the end of the following year. Exposure variables were clinical contact during an individual's first year in the survey: numbers of office visits, hospital outpatient department visits, emergency department visits, inpatient hospital stays, and dental visits. The analysis uses multivariate logistic regression to control for patient and family characteristics—most important, health status, functional status, and overall health care expenditures.
RESULTS. Eight percent of the children in the sample had left Medicaid/the Children's Health Insurance Program by the end of the second year in the survey. More frequent contact with clinicians in an office setting was associated with a significantly lower risk for dropping out of Medicaid/the Children's Health Insurance Program among children, even controlling for demographics, health and functional status, and overall health care expenditures. After multivariate adjustment, more frequent contact in hospital outpatient departments also was associated with reduced dropout, with a borderline statistically significant odds ratio. Notably, emergency visits and inpatient stays were not associated with any significant change in the risk of Medicaid/Children's Health Insurance Program dropout.
CONCLUSIONS. These results suggest that some but not all types of clinician visits are serving an important function in maintaining Medicaid and the Children's Health Insurance Program coverage among low-income patients. Two possible approaches to improve access to care among low-income children therefore would be (1) increased awareness among clinicians, especially in hospitals and emergency departments, regarding Medicaid/Children's Health Insurance Program retention as an issue in the ongoing care of their patients and (2) Medicaid/Children's Health Insurance Program reimbursement of clinicians and their staff for assisting patients with the public insurance renewal process.
Key Words: Medicaid ? CHIP ? health insurance ? access to health care ? health policy
Abbreviations: CHIP—Children's Health Insurance Program ? MEPS—Medical Expenditure Panel Survey ? ED—emergency department ? SSI—Supplemental Security Income
At the reception desk of an outpatient clinic in a Boston teaching hospital, a sign is written in large bold letters: "Reminder for Patients on MassHealth," it calls out to families who are enrolled in Massachusetts's version of Medicaid and the Children's Health Insurance Program (CHIP). "Every year, you need to tell the State you still want MassHealth. The state mails you a short form that you need to fill out and send back in. If you need help, bring the form to your health center or hospital as soon as possible."
As this sign suggests, keeping eligible patients enrolled in Medicaid and CHIP is an important step in maintaining health care access for millions of Americans, yet nearly 45 million Americans (15.6% of the population) did not have any health insurance in 2003, including 8.4 million children, roughly 1 child in 9.1 Millions of these individuals in fact are eligible for public insurance through Medicaid or CHIP.2 Whereas some of these individuals may never enroll in these programs at all, others indeed have enrolled but then drop out, typically by failing to complete the eligibility review process that is required at least once annually in every state. One recent study indicated that as many as 3 million children leave these public programs each year and become uninsured, despite continuing eligibility.3 As the sign at the hospital's reception desk makes clear, patient disenrollment from public insurance is a key concern in providing quality care to needy patients.
A large body of research indicates that health insurance coverage is critical to children's health. For instance, children without insurance or with discontinuous coverage experience lower rates of checkups, vaccination, and follow-up care; face more frequent restrictions on activities as a result of poor health; and are more likely to go without needed care when ill.4–6 Clearly, retention of eligible and needy children in Medicaid and CHIP should be a priority for both clinicians and policymakers.
Previous research has suggested a role for physicians in retaining eligible individuals in Medicaid/CHIP. County-level physician participation in CHIP has been shown to be protective against childhood disenrollment.7 Similarly, more generous state Medicaid physician reimbursement rates are associated with lower dropout rates, presumably because of either increased provider participation in Medicaid or a greater incentive for physicians to keep patients enrolled in public insurance.3 In a single-state survey of families with children who were eligible for a public insurance program similar to CHIP, parents, particularly in non–English-speaking families, reported that doctors and clinic staff are important sources of information about the enrollment process.8 Furthermore, community health centers have been found to play an important role in Medicaid and CHIP outreach to eligible children.9 But none of these studies has examined the direct impact of exposure to health care providers on Medicaid/CHIP retention. This study, using nationally representative data from the Medical Expenditure Panel Survey (MEPS), tested the hypothesis that clinical contact is associated with lower dropout from Medicaid/CHIP among children.
METHODS
The data are from the household component of the MEPS for the years 1998–2002. The MEPS is a nationally representative survey that is conducted by the Agency for Healthcare Research and Quality and features 5 rounds of interviews per household over a course of 2 years. The sample therefore contains 4 overlapping 2-year cycles: 1998–1999, 1999–2000, 2000–2001, and 2001–2002. The survey includes detailed questions on health insurance coverage, type and frequency of health care utilization, health status, and demographics.
The dependent variable is disenrollment from Medicaid/CHIP, determined by whether an individual is still enrolled in Medicaid or CHIP by the end of his or her second year in the survey. Note that the MEPS does not distinguish between Medicaid and CHIP and simply combines them in the survey. Medicaid by far is the larger program, with 24 million children compared with 4 million in CHIP, as of 2003.10,11 In roughly one third of states, Medicaid and CHIP actually are run as a single combined program, with the remaining states administering separate programs; previous research suggested that the choice of program structure can affect dropout and that there may be differences in dropout between Medicaid and CHIP,12 but the MEPS does not allow for analysis of these questions. Therefore, this study explored the associations between clinical contacts and dropout from children's public insurance in general, comprising both Medicaid and CHIP.
The key independent variables are the measures of clinical contact for the first year in the survey: number of nonhospital office-based visits, number of visits to hospital outpatient departments, number of emergency department (ED) visits, number of inpatient stays, and number of dental visits. Note that these are the visit types classified by the MEPS; the difference between "office" and "outpatient" visits is determined solely by whether the visit occurred at a hospital, without respect to the kind of services provided. The effect of these exposure variables on Medicaid/CHIP disenrollment was estimated using multivariate logistic regression.
For studying the impact of health care utilization on retention, several important threats to validity must be addressed. First, health care coverage is a strong predictor of health care utilization, so we must be careful to avoid reverse causation in measuring this relationship. To ensure that this analysis is identifying the effect of utilization on coverage, rather than the reverse, the study sample consists only of those individuals who were reported to have Medicaid/CHIP coverage for the full first year in the survey. The exposure is based only on health care utilization during this first year, and the outcome, Medicaid/CHIP retention, is based only on coverage during the second year. Therefore, we are guaranteed that the exposure predates the outcome. Furthermore, this approach ensures that all children in the sample have equal coverage during the exposure period, meaning that differential access on the basis of insurance coverage in the first year will not be an issue. With the exception of dental care, all types of health care utilization that are being used as independent variables are required by federal regulations to be covered under Medicaid and CHIP; therefore, variation in Medicaid and CHIP generosity across states also should not be a source of bias in this analysis.13
The second chief concern is that health status and preferences for health care utilization both are potential confounders. Health or functional status may confound the effect of utilization on coverage, because sicker individuals or those with disabilities will use more services in year 1, and their condition provides a strong incentive to remain enrolled in year 2. Therefore, the multivariate regression includes several distinct measures of health and functional status: self-reported health status (a 5-item scale ranging from "excellent" to "poor"); disability, based on receipt of Supplemental Security Income (SSI); whether a child has had any school- or work-related limitations as a result of medical, cognitive, or behavioral problems; and whether a child has ever received a diagnosis of asthma (the most common chronic disease of childhood) or diabetes (another increasingly common condition among children that is included in the MEPS data set).
Related to health and functional status but not identical to it is a person's overall tendency toward health care utilization. If some individuals or their parents are high-demand consumers of health care, even controlling for health status, then it is likely that such individuals would have more frequent health care contacts in year 1 and be more likely to stay enrolled in Medicaid/CHIP in year 2 as a result of this preference for health care services. To control for this effect, the regression includes an all-purpose measure of health care utilization: the total medical expenditure (from all sources, both insurance and out-of-pocket payments) for year 1. Therefore, the regression identifies the effect of clinical contacts in particular, measured in visits, controlling for the overall tendency toward health care utilization, measured in dollars.
Another factor that needs to be addressed in the study design is that disenrollment from Medicaid and CHIP is not necessarily a problem, depending on a person's reason for leaving public insurance. Previous research categorized disenrollment into 3 categories: lost eligibility, acquisition of new insurance, and "dropout" (when eligible individuals leave public insurance and become uninsured).3 The sample therefore excludes disenrollees who acquire other health insurance in year 2, and the analysis adjusts for key factors in Medicaid and CHIP eligibility (discussed next), to focus on dropout more specifically, which is the type of disenrollment on which provider outreach potentially would have the greatest impact.
The public insurance eligibility variables for children are SSI/disability, total family income, family size, and poverty status based on the percentage of the federal poverty line. Although this list of variables provides an imperfect method for adjusting for program eligibility, the publicly available MEPS data do not identify the state of residence, preventing more detailed analysis of eligibility. However, previous research indicated that among children who disenroll and do not acquire other insurance, fewer than 1 in 10 in fact have lost eligibility, indicating that this limitation should not bias the analysis significantly.3
The analysis was estimated by a survey-weighted logistic regression, using Stata 7.0, to take into account the complex survey design of the MEPS.14,15 Regression standard errors (SEs) are clustered at the household level to adjust for the nonindependence of observations within the same household. Various other factors that may affect Medicaid and CHIP retention also are included as covariates in the regressions: race, ethnicity, gender, language in which the MEPS interview was conducted, age, the highest level of parental education, the Medicaid/CHIP coverage status of family members, and the geographic region of the country. Overall, the sample consists of 3043 children (aged 0–18) who were covered by Medicaid or CHIP for the full 12 months of their first year in the survey and who did not acquire other health insurance in the second year.
RESULTS
Table 1 presents summary statistics for the study sample, stratified by enrollment behavior. Notably, 8.0% of the sample left public insurance and became uninsured by the end of the second year, the outcome of interest. This figure is slightly lower than other estimates of dropout, but this is not surprising because the sample is limited to those who were enrolled in Medicaid/CHIP throughout the entire first year, a more stable subset of enrollees than the general Medicaid/CHIP population. Not presented in Table 1 but also of note is that an additional 4.8% of children had at least temporarily gone without Medicaid/CHIP coverage at some point during the second year before reenrolling by year's end.
Table 1 indicates that the statistically significant differences between children who disenrolled and those who remained on public insurance were that disenrollees were older on average, were less likely to have diabetes, had less educated parents, were less likely to be receiving SSI, and were more likely to live in the South. In secondary analyses that are not presented in Table 1, poorer health, previous diagnosis of diabetes or asthma, and the presence of school/work limitations all were strongly associated with health care utilization, as expected.
Table 2 presents the results for the relationship between year 1 clinical contacts and year 2 dropout. The unadjusted odds ratio was statistically significant for "3 or more office visits," indicating that more frequent contact with a clinician in an office setting was associated with a reduced risk for dropout. These results essentially were unchanged when adjusting for the covariates in Table 1, most importantly, health and functional status, Medicaid/CHIP eligibility measures, and year 1 total health care expenditure. After multivariate adjustment, the only other significant result was a lower risk for dropout among children with multiple visits to a hospital outpatient department, but this result was significant only at a level of P < .10. None of the other types of clinical contact—inpatient, ED, or dental visits—was a significant predictor of dropout. Converting these adjusted odds ratios into predicted probabilities, children with >2 office visits in year 1 were 48% less likely to drop out of Medicaid/CHIP than those with no office visits, and children with >1 hospital outpatient department visit were 76% less likely to drop out than those with no outpatient visits.
DISCUSSION
Among children who were enrolled in Medicaid or CHIP for at least 12 months, exposure to clinicians in an office setting was a strong and significant predictor of decreased dropout from public insurance. Exposure to clinicians in a hospital outpatient setting also was associated with decreased dropout, although only at the P < .10 level. Meanwhile, other kinds of health care utilization and clinical contact, notably inpatient stays and ED visits, had no apparent effect on dropout. These results suggest, although they do not prove a causal effect, that clinicians can help to protect the continuity of insurance coverage for their Medicaid and CHIP patients. Given providers' own financial stake in getting reimbursed for care, as well as the chance to protect the health care access of their patients, this result is reasonable and encouraging. Furthermore, these findings are consistent with previous analyses that documented a positive association between retention and both physician participation and reimbursement rates in public insurance,3,7 but whereas previous studies could point to this association only at a county or a state level, this analysis now provides evidence at the individual patient level that clinical contact is associated with reduced dropout.
Why were only certain kinds of clinician contact, namely in the office setting and possibly in hospital outpatient departments, associated with reduced dropout? This likely reflects 2 aspects of the clinical encounter that differ across the location and the type of health care service: the extent of outreach efforts and the nature of the patient–doctor relationship. First, it simply may be that offices and outpatient departments are more familiar with the Medicaid/CHIP renewal process or devote more time to helping patients with renewal paperwork. The complexity and the severity of patient conditions in the ED and inpatient settings may make insurance issues a lower priority, despite the larger expenditures involved. These explanations are consistent with previous research in which parents reported that they are more likely to receive assistance with insurance paperwork from office-based pediatricians and clinic staff than from inpatient departments or EDs.8 In terms of dental care, many Medicaid and CHIP beneficiaries do not receive any dental coverage, which clearly would minimize the financial incentive for dentists to devote time or staff resources to Medicaid/CHIP retention.16
The second factor that likely differs across sites is the nature of the patient–doctor (or, more generally, the patient–provider) relationship. Office-based visits and, to a lesser degree, hospital outpatient visits are much more likely to be part of an ongoing relationship among a patient, a patient's family, and a clinician, compared with ED visits or inpatient stays. This kind of relationship presumably increases the likelihood that a patient's parents would feel comfortable raising issues related to insurance status and also makes it more likely that the physician and staff would know which patients need to renew their Medicaid or CHIP coverage. Furthermore, repeated visits may be beneficial simply because the renewal process involves collecting paperwork (paycheck stubs, tax returns, etc) and filling out forms, which means that 1 visit often may not be enough to complete the process successfully. Last, given the finding that nearly 5% of the children in this study who indeed were covered by the end of the year and nonetheless had experienced a temporary lapse in coverage earlier in the year, it seems likely that recurring office visits provide an excellent opportunity for providers to identify children whose coverage has already lapsed and to reenroll them promptly. Providers who do so not only protect their patients' coverage but also may be eligible for retroactive Medicaid/CHIP reimbursement for services already provided.
What are the implications of these findings? At the policy level, this study suggests that states should view clinicians and their staffs as key intermediaries in the retention of needy and eligible individuals in public insurance. To take advantage of this intermediary role, states that aim to reduce dropout from Medicaid and CHIP should consider adopting a specific billing code for staff and/or clinician assistance with the renewal process. Given the significant administrative costs to states of Medicaid and CHIP turnover, paying clinicians to keep eligible individuals enrolled actually could be a cost-saving measure, not to mention the significant benefit that improved retention would provide in terms of health care access.17 Furthermore, state outreach regarding Medicaid and CHIP renewal should target not just enrollees but also clinicians and their staff. Sending clinicians information once a year to explain the renewal process or simply to remind them that a renewal requirement exists will make it more likely that clinicians will be able to help their patients with this process.
At the level of individual physicians and other providers, the implications of the study are twofold. First, clinicians can have a major impact on Medicaid and CHIP retention. Second, not all clinicians are doing so. ED visits and inpatient stays currently are missed opportunities for promoting public insurance retention, and, for certain patients, these visits may be their primary contact with the health care system. Therefore, more proactive steps to improve outreach in these settings regarding Medicaid/CHIP retention should be strongly encouraged. What can physicians do? On the basis of this study, one can only speculate as to which specific approaches by clinicians and their staffs would be most beneficial in terms of public insurance retention, but patient surveys show that most families drop out of Medicaid and CHIP because they do not know that they need to renew or they do not understand the process.18,19 Therefore, if doctors' offices or hospital departments can afford a staff member or social worker to deal with these issues, then that undoubtedly would make a significant impact and might even be a profitable approach for practices that are losing money by providing uncompensated care to Medicaid-eligible but uninsured children. In the inpatient setting, including a question about Medicaid/CHIP renewal in standard discharge instructions to patients' parents could be an easy but effective approach. Even steps as simple as posting a Medicaid/CHIP renewal notice at the reception desk in offices and the ED, like the sign at the beginning of this article, may make a difference.
Of course, there are several limitations of this analysis that must be addressed. The most significant threats to internal validity were discussed already in "Methods": the possibility of reverse causation between insurance coverage and health care utilization, and the confounding effects of health status and a tendency to use health care services. In terms of reverse causation, the study design eliminates this problem, because all of the clinical contacts that were used as independent variables occurred in the year before the outcome variable, Medicaid/CHIP dropout, was assessed. The possibility of confounding cannot be eliminated, however, even with multivariate adjustment. Although several distinct measures of health and functional status, as well as a composite measure of the demand for health care services (total medical expenditure), were used it is possible that they may not capture fully the differences in these characteristics that could bias the results. However, if it actually were unobserved differences in health status that were driving the results, then one would expect that the clinical services that are most indicative of severe disease—ED visits and inpatient stays—would be the strongest predictors of retention. Yet, in fact, these 2 types of services had no predictive power regarding Medicaid/CHIP retention. Similarly, if unobserved preferences for consuming health care in general were the underlying factor, then there should be an effect of all types of utilization on retention, rather than just office and outpatient visits. Last, if the key confounder actually were a preference for preventive health care, rather than for all types of services, then one still would expect to see significant effects for dental visits, as well as physician office visits, but dental care was not a significant predictor.
An additional limitation of the study relates to the generalizability of the findings. Although the sample is nationally representative and therefore the findings cannot be attributed simply to the idiosyncrasies of any particular state's Medicaid or CHIP programs, the sample is limited to a subset of public insurance enrollees. The children in this study all were enrolled continuously in Medicaid or CHIP for their full first year in the survey. This approach had the advantage of ensuring that all children had the same access to care during the first year, when the key independent variables of clinical exposure were measured; but it has the disadvantage of restricting the analysis to a more stable subset of the public insurance population. It may be that the associations between dropout and provider visits in this group are significantly different from those among children with shorter and less stable periods of public insurance enrollment. However, if anything, it seems likely that there would be even more room for provider encounters to improve the retention rates of such children, who are at higher risk for dropout to begin with.
Last, the study was limited in its ability to differentiate subtypes of clinical contact on the basis of the available data. Although the MEPS reports the location of each visit, it does not distinguish between private practices and community health centers, between private and public hospitals, and between primary and specialty care. Each of these distinctions would be valuable in determining precisely which kinds of clinical encounter and setting are more strongly associated with reduced dropout.
CONCLUSIONS
Clinicians can serve a critical role in promoting the retention of Medicaid and CHIP insurance coverage among needy patients. Children with more frequent contact with office-based providers were much less likely to drop out of public insurance and become uninsured. Policymakers should craft retention strategies that encourage clinician assistance with Medicaid and CHIP renewal, and providers should take steps to help protect the public insurance coverage of their patients. It may be as simple as posting a reminder—"Medicaid and CHIP patients: Have you renewed your eligibility yet this year?"—that could make the difference between continued access to care and the loss of health insurance for patients in need.
ACKNOWLEDGMENTS
This project was completed while the author was supported by graduate fellowships from the National Science Foundation and the MD/PhD Program in the Social Sciences at Harvard Medical School.
Many thanks to Alan Zaslavsky, PhD, Joe Newhouse, PhD, and Melissa Wachterman, MD, for helpful suggestions.
The author had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
FOOTNOTES
Accepted Jan 3, 2006.
Address correspondence to Benjamin D. Sommers, PhD, 60E Glen Rd, Unit T-9, Brookline, MA 02445. E-mail: bsommers@post.harvard.edu
The author has indicated he has no financial relationships relevant to this article to disclose.
REFERENCES
DeNavas-Walt C, Proctor B, Mills R. Income, Poverty, and Health Insurance Coverage in the United States: 2003. Washington, DC: US Census Bureau; 2004
Selden T, Hudson J, Banthin J. Tracking changes in eligibility and coverage among children, 6–2002. Health Aff (Millwood). 2004;23 :39 –49
Sommers BD. From Medicaid and CHIP to uninsured: drop-out among children in public insurance programs. Health Serv Res. 2005;40 :59 –78
Olson LM, Tang SS, Newacheck PW. Children in the United States with discontinuous health insurance coverage. N Engl J Med. 2005;353 :382 –391
Lave JR, Keane CR, Lin CJ, Ricci EM, Amersbach G, LaVallee CP. Impact of a children's health insurance program on newly enrolled children. JAMA. 1998;279 :1820 –1825
Newacheck PW, Stoddard JJ, Hughes DC, Pearl M. Health insurance and access to primary care for children. N Engl J Med. 1998;338 :513 –519
Phillips JA, Miller JE, Cantor JC, Gaboda D. Context or composition: what explains variation in SCHIP disenrollment? Health Serv Res. 2004;39 :865 –885
Feinberg E, Swartz K, Zaslavsky AM, Gardner J, Walker DK. Language proficiency and the enrollment of Medicaid and CHIP-eligible children in publicly funded health insurance programs. Matern Child Health J. 2002;6 :5 –18
Davidoff AJ, Garrett B. Determinants of public and private insurance enrollment among Medicaid and CHIP-eligible children. Med Care. 2001;39 :523 –535
Kaiser Family Foundation. CHIP Program Enrollment: December 2003 Update. Menlo Park, CA: Kaiser Commission on Medicaid and CHIP and the Uninsured; 2003
Kaiser Family Foundation. The Medicaid and CHIP Program at a Glance. Menlo Park, CA: Kaiser Commission on Medicaid and CHIP and the Uninsured; 2003
Sommers BD. The impact of program structure on children's disenrollment from Medicaid & SCHIP. Health Aff (Millwood). 2004;24 :1611 –1618
Centers for Medicare and Medicaid and CHIP Services. Medicaid and CHIP: a brief summary. Available at: www.cms.hhs.gov/publications/overview-medicare-MedicaidandCHIP/default4.asp. Accessed June 16, 2005
Stata Corp. Stata 7.0 Statistical Software . College Station, TX: Stata Corp; 2001
Agency for Healthcare Research and Quality. Computing standard errors for MEPS estimates. Available at: www.meps.ahrq.gov/factsheets/FS_StandardErrors.htm. Accessed March 12, 2005
Center for Policy Alternatives. Dental coverage under Medicaid and CHIP. Available at: www.cfpa.org/issues/healthcare/dental/dentalfactsh.pdf. Accessed July 15, 2003
Ku L, Ross DC. Staying Covered: The Importance of Retaining Health Insurance for Low-Income Families. Washington, DC: Center on Budget and Policy Priorities; 2002
Perry M, Kannel S, Riley T, Pernice C. What Parents Say: Why Eligible Children Lose SCHIP. Portland, ME: National Academy for State Health Policy; 2001
Hill I, Lutzky AW. Is There a Hole in the Bucket? Understanding SCHIP Retention. Washington, DC: Urban Institute; 2003
PhD Program in Health Policy, Harvard Medical School, Boston, Massachusetts(Benjamin D. Sommers, PhD)