Churning in Medi Cal

Document Sample
Churning in Medi Cal Powered By Docstoc
					 Discontinuous Medi-Cal and Healthy Families
and the Effects on Access to Physician Services

             Shana Alex Lavarreda, MPP
            Academy Health Annual Meeting
                    Orlando, FL
                       6/4/07




                  www.healthpolicy.ucla.edu   1
                          What Is Already Known
   Discontinuous insurance breaks the link between enrollee and necessary health care
    services (Davidoff et al., 2000; Dubay and Kenney, 2001; Kasper, Giovanni, and Hoffman
    2000; Keane and Lave, 1999; Lave et al., 1998; Olson, Tang, and Newacheck 2005; and
    Sudano and Baker, 2003).
   Relevant family or household-level indicators:
      Household income, parent’s education level, parent’s employment factors (work status, access
         to employer-based insurance), urban or rural area (Byck, 2000; Dubay and Kenney, 2004;
         Gilmer, Kronick, and Rice 2005; Lin et al., 2004; Phillips et al., 2004; and Sommers, 2005),
      Sibling and parental insurance status (Keane et al., 1999; Lave, 1998; Lin et al., 2003; Phillips
         et al., 2004; and Sommers, 2005)
   Relevant child demographic factors:
      Child’s age, race, and gender have all been linked to type and continuity of insurance (Byck,
         2000; Dubay and Kenney, 2004; Gilmer, Kronick, and Rice, 2005; Lin et al., 2004; Phillips et
         al., 2004; and Sommers, 2005).
      Child’s health status (Byck, 2000; Keane et al., 1999; Kempe et al., 2004; and Lave et al.,
         1998).
   Existence of a county health insurance program, often called Healthy Kids, and amount per
    capita spent on outreach associated with increased enrollment (Kincheloe, 2004).
   Percent of the population that is uninsured or has Medi-Cal or Healthy Families in the county
    acts as a proxy for the differences in resources (Ullman and Hill, 2001).
   Community-level variables directly influence many individual-level factors, particularly a
    parent’s ability to gain employment that has health benefits and family income (Kronick,
    Gilmer and Rice, 2004).


                                                                                              2
       Determinants and Impact of Discontinuous
       Medi-Cal and Healthy Families in California
Contextual Factors
Policy Context
1) Existence of a county-level insurance
  program
2) $ per capita spent on outreach by
  county                                     Child Insurance Status       Health Outcomes
Concentration of Need
1) % nonelderly uninsured in the county      1) Child enrollee losing    1) Access to Physician
2) % children in Medi-Cal or Healthy            coverage and             Services – Usual Source
  Families in the county                        becoming uninsured.      of Care, Delays in Care,
                                             2) Uninsured child          Visits to a Doctor
Individual Factors                              enrolling in Medi-Cal    2) Health Related Quality
                                                or Healthy Families.     of Life
1) Child socio-demographic factors (age,     3) Child is uninsured all
  race, gender, household income)               year.
2) Child health status                       4) Child has continuous
3) Urban vs. rural residence                    Medi-Cal or Healthy
4) Parent education level                       Families.
5) Parent employment factors (work status,
  access to employer-based insurance)
6) Sibling in public program
7) Parent insurance status




                                                                                       3
             What will this study answer?
 Exploratory examination of how a child’s enrollment into or
  disenrollment from Medi-Cal or Healthy Families affect their
  access to physician services.
 Main hypothesis - discontinuous Medi-Cal and Healthy Families
  leads to worse outcomes in terms of basic access to physician
  services than continuous enrollment.
    Having a usual source of care over the past 12 months
    Reporting any delay in care in the past 12 months
    Having visited a doctor at least once in the past 12 months
 Secondary hypothesis - children continuously enrolled in Medi-
  Cal or Healthy Families will be more likely to utilize doctor’s
  offices for their care.



                                                            4
       Determinants and Impact of Discontinuous
       Medi-Cal and Healthy Families in California
Contextual Factors
Policy Context
1) Existence of a county-level insurance
  program
2) $ per capita spent on outreach by
  county                                     Child Insurance Status       Health Outcomes
Concentration of Need
1) % nonelderly uninsured in the county      1) Child enrollee losing    1) Access to Physician
2) % children in Medi-Cal or Healthy            coverage and             Services – Usual Source
  Families in the county                        becoming uninsured.      of Care, Delays in Care,
                                             2) Uninsured child          Visits to a Doctor
Individual Factors                              enrolling in Medi-Cal    2) Health Related Quality
                                                or Healthy Families.     of Life
1) Child socio-demographic factors (age,     3) Child has continuous
  race, gender, household income)               Medi-Cal or Healthy
2) Child health status                          Families.
3) Urban vs. rural residence                 4) Child is uninsured all
4) Parent education level                       year.
5) Parent employment factors (work status,
  access to employer-based insurance)
6) Sibling in public program
7) Parent insurance status




                                                                                       5
                    Study Design – CHIS 2003
 CHIS 2003 is the most recent dataset available from this biennial survey that
   includes all access to care measures.
     Surveyed over 42,000 households from July 2003 to February 2004 using a
       random-digit-dial CATI methodology
     Administered in six languages (English, Spanish, Mandarin, Cantonese,
       Vietnamese, and Korean) in order to capture responses from as wide a swath of
       California’s multiethnic population as possible.
     Response rates of the survey are comparable to those of other scientific telephone
       surveys in California, such as the California Behavioral Risk Factor Surveillance
       System (BRFSS).

 Interviews parents about their own and their child’s insurance statuses, as well
   as their health statuses, demographic characteristics, and employment factors.
     Every child and adolescent is linked with a parent interview.
     A child (ages 0-11) or adolescent (ages 12-17) can only be randomly chosen as a
       survey subject if an associated adult (i.e. parent or guardian) is randomly chosen.




                                                                                  6
  Study Design – Non-Equivalent Control Group
 This study focuses only on those who are either currently uninsured or enrolled
   in Medi-Cal or Healthy Families, leaving a total unweighted sample of 3,842
   children ages 0-17. Among these cases, all missing data has been imputed and
   there are no missing values.

 This is a retrospective, non-equivalent control group design.
     Children with Medi-Cal and Healthy Families all year are the control group.
     Control group is compared to:
            Children who were uninsured all year
            Children who gained Medi-Cal or Healthy Families
            Children who lost Medi-Cal or Healthy Families

 Cross-sectional survey data approximates logitudinal data with retrospective
   questions, but only examines associations.

 Dataset chosen for its size, scope, and inclusion of all relevant independent and
   dependent variables.


                                                                                7
Study Methods – Logistic Regression Analyses
 Independent variables were used as a set of constant predictors in each of four
   different models.
     Child’s age, racial and ethnic group, household income, gender, place of residence
        and health status

 Logistic regression for dichotomous outcome measures (having a usual source
   of care, any delay in care, or visiting a doctor).
     Included calculation of coefficients, odds ratios, predicted margins and relative
       risks.

 Multinomial logistic regression for non-ordinal, categorical variable (type of
   usual source of care).
     Included calculation of odds ratios and relative risks.
     Additionally, the Hausman-McFadden, Small-Hsiao, and Wald tests were
      performed to estimate if the assumption of independence of irrelevant alternatives
      held and whether or not the categories of the outcome variable should be
      collapsed together.



                                                                                  8
 Extent of Discontinuous Medi-Cal or Healthy Families

 91.7% of current Medi-Cal enrollees and over 86.8% of current
  Healthy Families enrollees were insured for the entirety of the
  previous year.

 218,000 new enrollees joined Medi-Cal and 81,000 enrolled in
  Healthy Families.

 One-fourth of uninsured children who were eligible for Medi-Cal
  or Healthy Families had actually been enrolled at some time
  during the past year (79,000 in Medi-Cal; 43,000 in “other”).
    Children who should be enrolled are likely being dropped for
     reasons other than their eligibility, such as nonpayment of
     premiums to the Healthy Families program or administrative
     difficulties.


                                                               9
      Significant Logistic Regression Results
 Usual Source of Care
    Uninsured all year (OR= 0.434, p=0.000)
    Gained Medi-Cal or Healthy Families (OR= 0.594, p=0.003)
 Delay in Care
    Uninsured all year (OR= 1.632, p=0.000)
    Lost Medi-Cal or Healthy Families (OR=2.106, p=0.001)
 Visiting a Doctor
    Uninsured all year (OR= 0.353, p=0.000)
    Lost Medi-Cal or Healthy Families (OR=0.607, p=0.015)
    Gained Medi-Cal or Healthy Families (OR= 0.511, p=0.000)




                                                             10
  Significant Multinomial Logistic Regression Results

 Doctor’s Office or HMO
    Referent group
 Community or Hospital Clinic
    Uninsured all year (OR= 1.774, p=0.000)
 ER or Other Source
    Uninsured all year (OR= 3.442, p=0.000)
 No Usual Source
    Uninsured all year (OR= 3.293, p=0.000)
    Gained Medi-Cal or Healthy Families (OR=1.592, p=0.011)




                                                               11
                            Study Limitations
 Methodologically, logistic regression examines gross access indicators; the
   details of number of doctor visits cannot be accounted for.
     This method was chosen over a Poisson or negative binomial model due to the
       research question being addressed, namely does discontinous insurance have a
       very basic negative impact?
     Since the model estimated that discontinuity did indeed have an effect at this most
       basic level, this finding has more health policy relevance than a count model
       would.

 The 2003 CHIS has the limitations of being an entirely self-reported survey, RDD
   telephone survey.
     Recall and selection biases are of course part of the drawbacks of this type of
       survey.
     No room for full picture of breaks in coverage over past year.


 Additionally, 18-year-olds are eligible for Medi-Cal and Healthy Families but have
   not been included in this analysis.


                                                                                 12
                Conclusions and Implications
 Dropping out of Medi-Cal and Healthy Families does indeed have detrimental
   effects on access to physician services.
     Medi-Cal and Healthy Families have mostly stable populations, but thousands
       leave the programs over the course of a year who in fact are still eligible for
       coverage.
     Children who gained coverage, however, had slightly better outcomes in
       comparison with those with continuous coverage than did children who were
       uninsured all year, indicating that Medi-Cal and Healthy Families have a beneficial
       effect on access to care.

 Supports claims that increased enrollment and retention of children in Medi-Cal
   and Healthy Families will improve their access to physican services
     Demonstrates the importance of concentrated outreach and retention efforts.
     Provides data-based evidence for financial support of enrollment and retention to
       improve health access for children in California.




                                                                                 13

				
DOCUMENT INFO
Shared By:
Categories:
Stats:
views:38
posted:4/12/2008
language:English
pages:13