4Systematic Review of the Impact of Managed Care on Children With Special Health Care Needs: A Special Focus on Medicaid

Beyond specific intervention programs, efforts to enhance care coordination for CSHCN have relied on changes in the overall organization of services, primarily through some form of managed care. In a recent review of the literature regarding insurance coverage and CSHCN, Jeffrey and Newacheck detailed the published evidence to date regarding the experience of CSHCN under different insurance systems.48 This cogent review suggested that the few studies assessing the impact of managed care on CSHCN revealed mixed effects on health care utilization and satisfaction. Among their most important findings was a troubling lack of research regarding the actual health outcomes of chronically ill children in relation to different insurance coverage profiles.

This review is intended to extend the findings of the Jeffrey and Newacheck review by offering a more detailed examination of published reports regarding the impact of managed care on a particular group of CSHCN: those enrolled in Medicaid. The focus on this group is based on several considerations. First, poor children are more likely to be particularly vulnerable to any deleterious effects of structural changes in care delivery.49 Second, they also appear to be in greatest need of improvements in the coordination of care.50 Third, over the last two decades, legislative and regulatory action by state governments has resulted in US states moving large number of children from Medicaid fee for service (M-FFS) into Medicaid managed care (M-MC) programs,51 a shift whose effects have taken on a variety of forms.52, 53 Current policy trends will likely extend this shift in delivery structure to a growing number of CSHCN.54 Because Medicaid provides health insurance for the single largest group of CSHCN in the country, the prospective shift from M-FFS to M-MC for these children represents the most critical, potential change in the structure of health care delivery for CSHCN currently being deliberated at the state and national level.

Background

Despite recurring calls for data,5558 the effects of the movement from M-FFS to M-MC for CSHCN remain unclear.59, 60 The potential utility and uncertainties related to poor children served by Medicaid managed care were voiced as early as 199060, 61 and included concerns about access, utilization, and health outcomes. Over the subsequent fifteen years, others have revisited these concerns, particularly for CSHCN.6265 When articulated, the justification for the shift of CSHCN from M-FFS to M-MC usually includes one of three objectives. A central objective is to enhance access to a primary care provider as well as appropriate specialty care, with improved service utilization mix.66 A second objective is to improve expressly the coordination of services. Such coordination can occur by providing one usual place of care with an emphasis on providing preventive services.67 It also can occur by making services more readily available with improved record-keeping and without duplication. A third objective is to reduce costs through more cost-effective prescription practices,68 preventing hospitalization and emergency department use and, reducing direct and indirect overhead costs.69

In large part, this rationale for moving CSHCN into M-MC is based on experiences with adults and generally healthy children. However, CSHCN are more likely than other children to be enrolled in Medicaid70 and the implementation of managed care has varied.71, 72

Review Methods

To identify published peer-reviewed research related to the effect of M-MC on CSHCN, we conducted a literature search and sought further information from existing bibliographies and expert colleagues. For the purposes of this review, we excluded ‘gray literature’ — materials that are found in recorded, written, or electronic form that are not traditionally well indexed, readily available, or peer-reviewed (e.g., conference papers, white papers, technical reports, electronic theses and dissertations, online documents, and oral presentations/abstracts).

The PubMed database was searched (1988 – February 2006; includes HealthSTAR since 2000). For this database, a broad search was conducted using MeSH headings and terms:

  • “Disabled children” or “Chronic illness[Multi]” or “Chronic Disease” or “Catastrophic illness” or “special health care needs” or “special needs.”
  • In conjunction with: “infant” or “child” or “child, preschool” or “adolescent.”
  • In conjunction with: “Medicaid” and “Managed Care Programs” or “Medicaid” and “HMO.”

We also searched the Cochrane Controlled Trials Register database and the Cochrane Collaboration's Specialized Register of Effective Practice and Organization of Care (EPOC). For this database, a free text search strategy was applied, using the following terms (* indicates wild card symbol):

  • Special health care needs* or disabil* or chronic*.
  • In conjunction with: infant or child or preschool* or adolesc*.
  • In conjunction with: Managed care* and Medicaid* or Medicaid* and HMO*.

Finally, the reference lists of located papers were scanned for studies of children with special health care needs in Medicaid managed care and relevant articles were retrieved. We compiled the results from all searches into an EndNote bibliographic database, removing all duplicate records. Through this process, we identified 99 publications.

Inclusion and exclusion criteria were determined. Inclusion criteria stated that the article must be written in English and must: (1) be empirical; and, (2) address children, special health care needs (broadly considered), health care services as provided by Medicaid managed care, as well as patient experiences and outcomes (broadly considered). The exclusion criteria stated that articles would be excluded if: (1) representing opinions, commentaries, reviews; (2) addressing behavioral or mental health issues exclusively; or, (3) focusing on state children's insurance programs (SCHIP) — within Federal guidelines, each state establishes the design and administrative/operating procedures of its SCHIP program; thus, not all SCHIP programs are extensions of Medicaid.

Two persons (LH and GB) independently examined the titles and abstracts of the 99 publications. From this set, 59 publications were excluded (the two reviewers agreed that articles did not meet inclusion/exclusion criteria). Thus, there were 40 publications that were put forward for full review. The two reviewers independently reviewed the full text of the 40 articles. Of these, an additional 24 were excluded because of one of the following: (1) editorial, comment, or letter; (2) not relevant for specific reason (i.e., discussion of design or methodological issues, opinion/commentary/description, or review or overview).

The remaining 16 articles were the subject of our efforts concerning the quality of studies, strength of evidence, and summary of findings. For the quality of studies assessment, the two reviewers completed and compared their abstractions; disagreements were settled by discussion and additional review of disputed articles.

Results

Characteristics and Quality of Individual Studies

Sixteen articles addressed the impact of Medicaid Managed Care on health services delivery to children with special health care needs (see Table 3). There was one randomized trial.73 The remaining 15 articles represented observational studies: one was a prospective cohort design,74 seven were retrospective cohort designs,7581 one was a time series design,82 four were before-and-after designs,8386 one was a cross-sectional design,87 and one was a case series.70

Table 3. Summary of Characteristics of Reviewed Studies.

Table 3

Summary of Characteristics of Reviewed Studies.

The study quality criteria suggested by West et al. (see Appendix A) were used in the review. Table 4 presents the summary of these ratings. Of the 16 articles, two fully addressed all domains.73, 82 Three other articles addressed most of the domains, omitting only information about subject comparability,77 exposure to managed care,74 or results.86 The remaining 11 articles partially or entirely overlooked two domains,75, 7981 three domains,78, 79, 8385 or four or more domains.70, 76, 87

Table 4. Summary of Quality Elements of Reviewed Studies.

Table 4

Summary of Quality Elements of Reviewed Studies.

All of the articles described a research question that was clearly focused and appropriate, depicted the study population, and delineated the outcomes. Most articles (n+14) also included information about the exposure of the study population to Medicaid managed care (the intervention under examination). Only a few articles (n+5) included a measure of effect size (e.g., odds ratio, absolute or relative risk reduction, number needed to treat). Similarly, only six fully addressed the issue of subject comparability (e.g., defining special health care needs, noting specific inclusion/exclusion criteria, establishing comparability of groups at baseline).

Subject comparability is important in these studies as there can be notable between-group variation at the start of a new program. In the case of M-MC and children with special health care needs, such variation may include child demographics (with significant proportions of children in Medicaid who are poor, minority, or in single-parent households), child epidemiology (with low overall prevalence of disease made up of mainly rare conditions, though asthma, diabetes, epilepsy, and sickle cell anemia are exceptions) and adverse selection (children who are less healthy tending to join one plan or another).

Evidence of Program Effects

In this section, we distill the results of the 16 identified studies. We apply the strategy of Newacheck and colleagues56 for categorizing these outcomes. Briefly, there are seven outcomes categories that are salient for CSHCN. While one involves care coordination directly, the other elements also reflect elements of care coordination more generally, including satisfaction, access and family impact. Moreover, unlike specific care coordination interventions, M-MC is structural and organizational system whose coordinative effects can touch virtually all aspects of these care indicators:

1.

Access to Care, with indicators reflecting convenience and physical access, travel and waiting time, provider choice, coverage of services, and availability of services (frequently denoted in measures of unmet need).

2.

Health Care Utilization, with indicators reflecting use and volume of primary medical care, specialized medical care, specialized therapies (e.g., physical therapy, mental health therapy, home health services), family support services, equipment and supplies, and related services (e.g., early intervention, special education).

3.

Quality of Care, with indicators reflecting case finding and service coordination, provider training and supply, medical necessity, clinical quality (e.g., quality of care standards, quality improvement system), and grievance procedures.

4.

Satisfaction with Care, with indicators reflecting family and practitioner satisfaction;

5.

Health Care Expenditures, with indicators reflecting expenditures for care (including out-of-plan services), indirect costs incurred by families, degree to which other funding provides financing (e.g., Title V, EPSDT), and degree to which savings are achieved by managed care.

6.

Health Outcomes, with indicators reflecting global health as well as cognitive, physical, social, and emotional functional status.

7.

Family Impact, with indicators reflecting sibling and parent health status, parent financial burden, parent knowledge of special health care needs, and managed care plan-provided family support.

The general findings of the reviewed studies were highly varied and did not present a clear message as to the general utility of M-MC for CSHCN. Eight studies reported utilization findings; five access findings; four quality findings; four satisfaction findings; and one expenditure finding. None of the reviewed studies reported direct health outcomes measures or family functioning but did report parental survey data on child health.

Access to Care. There was no consistent set of findings regarding access to care. Fox et al.83 reported that after the introduction of M-MC, care programs experienced reduced financing for most early intervention services for CSHCN but enhanced financial support for vision and enabling services. Mitchell and Gaskin79, 81 similarly reported mixed access findings. CSHCN in M-MC had similar rates as their M-FFS counterparts of unmet needs for physician/hospital care, mental health, home health and therapy services. However, the children in M-MC had fewer unmet needs for dental care, medical equipment and prescription drugs. Roberto et al.86 reported children in M-MC had fewer problems accessing primary, specialty, emergency and hospital care. However, Valet et al.80 found that parents of CSHCN in M-MC experienced high levels of care denial and that these levels were higher than children without special health care needs. In sum, two of the examined studies reported reduced access in M-MC, 4 improved access in M-MC, and two mixed access findings.

Utilization. As was found for access to care, utilization findings were decidedly mixed. Cooper et al.84 found that the proportion of children hospitalized with chronic conditions were similar during the periods before and after the introduction of a M-MC system. Finkelstein and colleagues77 reported that children with asthma in M-MC were more likely to receive care in emergency departments and were hospitalized more often than those in commercial managed care plans. However, both groups of children had similar rates of nonurgent and urgent ambulatory visits. Similar mixed findings were reported by Gadomsky et al.82 with CSHCN in M-MC eligible for SSI having higher emergency department use, hospitalization and ambulatory-sensitive hospitalizations but also a slight increase in preventive care than non-MC counterparts. Shatin et al.75 reported that children with chronic illness in M-MC had increased service use compared with commercial-MC counterparts. Grossman et al.85 reported that after the introduction of M-MC program, CSHCN had no change in emergency department visits but fewer hospitalizations and hospital days. Lieu et al.74 found that children with asthma in M-MC had greater hospital utilization than those in commercial managed care while Mitchell et al.78 reported no differences in hospital, physician, dentist and prescription drug services among CSHCN.

In sum, the reviewed studies reported four findings that M-MC was associated with reduced utilization, three improved utilization, and five that showed no difference with the compared populations.

Quality of Care. Two studies reported improved quality of care in M-MS. Cooper et al.84 reported that there were fewer enrollment gaps among CSHCN after M-MS was instituted. Grossman et al.85 found that certain groups of CSHCN had improved quality of care indicators in M-MC; however, most of the studied children had similar quality indicators. Lieu et al.74 found no differences in quality indicators between children with asthma in M-MC and those in commercial managed care. Finkelstein et al.77 found that children with asthma in M-MC had similar prescriptions written for controller medications but fewer of these were actually dispensed to the children in the M-MC group.

In sum, one study found reduced quality indicators associated with M-MC, two reported improved quality measures in M-MC systems, and three reported no quality differences between M-MC and comparison populations.

Satisfaction. Satisfaction findings were reported in 4 of the examined studies. Mele and Flowers70 and Millar et al.87 both reported reduced parental satisfaction among CSHCN enrolled in M-MC. Mitchell and Gaskin79 however, reported improved parental satisfaction in M-MC, particularly in office visit wait time, office hours, telephone medical advice and specialist care. Grossman et al.85 and Mitchell et al.78 reported no differences in parental satisfaction between the compared groups. Thus, two reported worse satisfaction for M-MC, one improved satisfaction and one no difference.

Expenditures. Interestingly, there was no study that indicated that children in M-MC had reduced health care expenditures, although only two addressed this factor. Chan and Vanderberg76 reported that M-MC was associated with higher costs in general but reduced claims for mental health care than their M-FFS counterparts. Grossman et al.85 found no difference in costs after M-MC was introduced.

Outcomes. There were no studies that directly assessed health outcomes. However, Lieu et al.74 assessed reported asthma physical health and found no difference between the M-MC and commercial managed care groups of children with asthma.

Family Functioning. There were no studies that assessed family functioning.

Summary of Review Findings

Despite the fact that large numbers of CSHCN are being moved into managed care programs, there are a relatively small number of peer-reviewed studies documenting the impact of such programs on this vulnerable population of children. In total, the numbers of reviewed articles is 16, representing 14 separate projects. One article describes an RCT, and 15 articles describe non-randomized comparative studies. The samples of children with special health care needs ranged in size from zero (studies where the subjects were the directors of state programs for Infants and Toddlers with Disabilities (ITDP or IDEA, Part C) or other key informants) to 3,839.

Assessing the strength of the body of evidence concerning this research involves judgments of study quality; it also includes how confident one is that a finding is true and whether the same finding has been detected using a range of studies or study participants. In this regard, only two of the sixteen articles fully address all domains that characterize high-quality observational studies. In addition, this is a generally inconsistent body of evidence. This collection of scientific work consists of a small number of lesser quality studies that can be contradictory in their conclusions.

It should be noted that these studies do not explicitly evaluate care coordination directly. However, they do assess attempts to move CSHCN into a structure of care that is expressly justified, at least in part, as a means of improving the coordination of care for these children. Moreover, the outcome measures used in these studies, although far from ideal, are largely representative of the kinds of outcomes expected to respond to coordinative effects.

The basic premise here is that care coordination is sensitive to both structural initiatives and specific programmatic interventions. Particularly given the scale of the M-MC policy initiatives and the number and vulnerability of affected children, any consideration of care coordination for CSHCN in the United States today must address both programmatic and structural pathways of effect.