Finance household insurance

ABBREVIATIONS. SCHIP, State Children's Health Insurance Program; CHP+, Child Health Plan Plus; FPL, federal poverty level.

In response to the growing problem of uninsured children, the Balanced Budget Act of 1997 created the State Children's Health Insurance Program (SCHIP), providing federal funding to help states expand the provision of health insurance to uninsured low-income children. Currently, 20 states have stand-alone, non-Medicaid SCHIP programs, 14 have Medicaid expansion programs, and 17 are operating programs that are combinations of both types of plans. (1) Although enrollment in the first few years was slower than hoped, (2-4) by June 2002, it had reached 3.6 million nationally. (5,6) Recently, there is also encouraging evidence that SCHIP is making a difference in levels of uninsured children. Cunningham, (7) using data from the Community Tracking Study household survey, demonstrated that almost 20% fewer children were uninsured in 2000-2001 than in 1998-1999 nationally. In addition, Elixhauser et al (8), using data from the Medical Expenditure Panel Survey, demonstrated a decline in the percentage of children who were uninsured all year from 10.3% in 1996 to 7.8% in 1999.

Since the enactment of SCHIP, most of the published literature regarding the program has focused on issues of enrollment, eligibility, and retention. (2-5, 9-17) Because SCHIP is a relatively new program, little has been published regarding the effect of SCHIP on important processes of care measures and health outcomes. The major objectives of the present study were to compare reported access to care, utilization of health care, and overall quality of care 1 year before and during the first year after enrollment into Colorado's SCHIP, the Child Health Plan Plus (CHP+). Specifically, we addressed the following questions: Did enrollment in CHP+ increase the percentage of children with a medical home and access to preventive, acute, and subspecialty health services? Did enrollment result in higher utilization of medical services after enrollment than preceding enrollment? Did enrollment result in higher overall quality of care services from the perspective of the enrolling family?

METHODS

Description of CHP+

Colorado's SCHIP, CHP+, is a stand-alone program, although it shares a common enrollment application with the state's Medicaid program. The program provides medical benefits, including inpatient and outpatient services, prescription drugs, dental care, and mental health care. Services are provided either by a health maintenance organization or by a fee-for-service network, depending on the geographic area in which an enrollee lives. Families are eligible when they earn <185% of the federal poverty level (FPL). Enrolling families pay an initial enrollment fee of $25 for 1 child or $35 for multiple children when their income is >150% of the FPL as well as copayments when their income is >100% of the FPL.

Design and Study Population

We conducted 2 telephone surveys, 1 year apart, for a cohort of families who had recently enrolled a child in CHP+ for the first time. We randomly selected families who had enrolled 2 months before and did an initial survey during September 1999 through January 2000 (N = 711). We surveyed 2 months after enrollment, because >90% of newly enrolling families have been notified of their acceptance to the program and assigned to a provider by this time. A follow-up survey was conducted I year later, during November 2000 through February 200], for the same cohort. Families with both initial enrollment and 1-year follow-up surveys (N = 480) composed the study population. The study protocol was approved by the Colorado Multiple Institutional Review Board.

Survey Method

The survey instrument and method have been previously described in more detail. (9) During both surveys, families were asked to report only on the year preceding the interview, corresponding to the year before and the year after enrollment into CHP+. The surveys incorporated standardized questions with minor modifications from the National Health Interview Survey Household survey, the Prototype Children's Health Insurance and Health Care Questionnaire from the State and Local Area Integrated Telephone Surveys of the National Center for Health Statistics, and the Consumer Assessment of Health Plans Child Core that have been previously used by the study team. (2,9) The interviews were conducted in English or Spanish, depending on the preference of the interviewee.

Interviews were conducted by Survey Units at the Colorado Department of Public Health and Environment and the AMC Cancer Research Center in Denver, Colorado. Families were called up to 15 times at different calling periods to optimize response rates, and both home and work numbers listed with the program were used. The interview was programmed for Computer Assisted Telephone Interviewing and skip patterns and acceptable range of responses, and consistency checks were programmed into the instrument. A minimum of 10% of all interviews were monitored by supervisors who randomly listened to calls and, using Local Area Network Assist Plus software, monitored interviewers' computer screens at the monitoring station.

Definition of Measures

Measures of access to care included whether there was a usual source of care or an identified primary provider; ease in seeing provider for routine, acute, or subspecialty care (on a 4-point Likert scale); and whether there were unmet prescription, mental health, vision, dental, routine, acute, or subspecialty health care needs. Utilization of care measures included the quantity of routine, acute office, subspecialty or emergency department visits, and hospitalizations. Quality of care was determined by having parents rate the quality of health care received (10-point Likert scale with 0=worst and 10=best) and by assessing whether any preventive care had occurred in the previous year. Type of previous insurance was categorized as private, Medicaid, none, or other, and length of time uninsured before enrollment in CHP+ was grouped into 3 levels: no gap in insurance, uninsured <1 year, and uninsured 1 year or more. Race/ethnicity was determined by self-report and was categorized as white, black, Hispanic, or other.

Data Analysis

For bivariate analyses, binary categorical data were analyzed using McNemar's test for paired data, and continuous data were analyzed using paired t tests. In comparing access to care for subgroups who perceived a need for preventive, acute illness or injury, or subspecialty care, paired data could be statistically compared only for those who needed these types of care in both years. For multivariate analyses, binary categorical data were analyzed using logistic regression, and continuous outcomes (number of visits) were analyzed using Poisson regression within a Generalized Linear Model. For the multivariate analysis, each child contributed 2 records to the data set, 1 for the year before CHP+ enrollment and 1 for the year during CHP+ enrollment. To account for repeated measures of the same children over time, a generalized linear models analysis was used (SAS PROC GENMOD). We did multivariate modeling using 2 methods. We first ran the model including all of the explanatory variables and, subsequently, using backward elimination, including only explanatory variables that reached a significance level of .25 or less. (18) Results are reported for variables with significance levels of <.05.

The dependent variables in our analyses were measures of access to care, utilization of care, and quality of care. The explanatory variables that initially were included in the models were period of measurement (pre-CHP+ enrollment vs post-CHP+ enrollment), type of previous insurance, length of time uninsured before enrollment in CHP+, and reported race/ethnicity. In addition, because age is known to be a significant predictor of utilization, we included age in years as a continuous variable measured at time of enrollment and age (2) to account for a possible quadratic contribution of age. The results presented therefore are adjusted for age. We also assessed interaction terms including the pre/post time period by previous insurance and by length of time uninsured. SAS Version 8.2 (SAS Institute, Cary, NC) was used in all analyses.

RESULTS