Trends in Care for Uninsured Adults and Disparities in Care by Insurance Status Lindsay Sabik AcademyHealth Annual Research Meeting June 30, 2009 This material is based upon work supported under a National Science Foundation Graduate Research Fellowship Outline • • • • Background and motivation D t and Data d methods th d Results Summary and discussion Background and motivation • Uninsured population increasing over time • With costs projected to increase, number of uninsured likelyy to continue rising g • Compared to the insured, uninsured individuals – receive fewer and less timely services – are less likely to receive preventive care – have worse outcomes • Existing research is primarily cross-sectional Trends in care for uninsured • Ri Rising i costs t are associated i t d with ith – Advances in medical technology Improvements in treatment Increasing costs per treatment episode – Increasing insurance premiums Rising number of uninsured • Tension between improved standards of care and increasing financial strain on system Research questions 1) How is care for the uninsured changing over time? – Absolute trends indicate whether system maintains a minimal level of care through safety net 2) How are the differences in care received by the privately insured and uninsured changing? – Relative trends indicate how disparities by i insurance status t t are changing h i over titime Data Access: Communityy Tracking g Study 1996-2003 Chronic Care: National Health and Nutrition Examination Survey 1999-2006 Acute Care: National Hospital Discharge Survey 1996-2006 1996 2006 Analytic approach ((1)) Restrict sample p to uninsured individuals and estimate: Pr (Yit) = Φ (β0 + β1Xi + β2t + εit) ((2)) Include insured and uninsured individuals and estimate: Pr (Yit ) = Φ (β0 + β1Xi + β2coveragei + β3t + β4coveragei·t + εit) where Yit = outcome (access/treatment/control) for person i at time t; Xi = vector t off individual i di id l covariates i t (age, sex, race, income, etc); coveragei = 1 if privately insured, 0 if uninsured; t = indicates the time period of the observation; εit = error term. Changes in uninsured population • Changing population of uninsured affects changes in care over time • Attempt to control for these changes: – Focus on measures incorporating potential demand for care – Focus F on sub-populations b l ti off uninsured i d with ith similar i il underlying health status – Include individual-level covariates that control for observable characteristics associated with health status and demand for care CTS sample demographics Year Age % Male % Nonwhite Family Income General Health (1-5) 1996 1998 2000 2003 1996 1998 2000 2003 1996 1998 2000 2003 1996 1998 2000 2003 1996 1998 2000 2003 Privately Insured (Percent or Mean) 39.9 39.7 40.1*** 40.1 40.9*** 49.8% 49.5% 49 6% 49.6% 49.1% 21.7% 23.1%*** 22 9% 22.9% 23.4% $49,611 $54,100*** $59 941*** $59,941*** $63,344*** 2.14 2.16** 2 20*** 2.20*** 2.21 Uninsured (Percent or Mean) 34.8 35.2 35.1 35.9* 53.5% 51.9% 52 7% 52.7% 55.3%** 43.8% 46.1% 46 7% 46.7% 50.7% $19,571 $19,827 $24 010*** $24,010*** $24,193 2.48 2.50 2 56** 2.56** 2.57 Data from the CTS; Includes individuals ages 18-64; *** p<0.01, ** p<0.05, * p<0.1 in tests of whether value is significantly different from previous year within insurance category; General health: 1 = Excellent, 5 = Poor. Forgo care Access ccess to o ca care e • No significant change in probability of forgoing or delaying care for uninsured Forgo or delay care • Compared to privately insured, uninsured have – 9 9.5 5 percentage point higher probability of forgoing care – 13.6 percentage point higher probability of forgoing or delaying • No significant changes in gap Cholesterol control Chronic disease control (1) • No significant change in cholesterol or glycemic gy control among uninsured • No significant difference between privately insured and uninsured on average Glycemic control • U Uninsured i db become worse off relative to privately insured – increase in gap of ~2pp per year for cholesterol control (p<0.05) – ~1 1 pp per year ffor glycemic l i control (not significant) Chronic disease control (2) Blood pressure control • Both uninsured and privately insured show improvements in blood pressure control • Gap narrows, though not significant on average • ~ 8.1 p percentage g p point gap remains at end of period Invasive care post-AMI post AMI Invasive AMI care • Both uninsured nins red and privately insured improving • No significant difference between g groups p on average • No significant change in gap between groups Summary of results • No significant improvement or decline for uninsured for most outcome measures – Flat trends in access; diabetes, cholesterol control – Exceptions: BP control and post-AMI care • Gaps between uninsured and privately insured constant or widening on most measures Limitations • Surveyy data limitations – Access measures from CTS and diagnosis measures from NHANES based on self-report – All data are pooled cross-sections • limits ability to control for changes in population over time • Number of possible mechanisms may be driving trends – e.g. privately i t l iinsured d may b benefit fit ffrom di disease management programs Implications for policy • Disparities in access to care and chronic disease control by insurance are not improving over this period • Efforts to improve care for the uninsured (e.g. expanding community health centers) have not narrowed gap for most measures • Uninsured are not worse off in an absolute sense, but are relatively worse off on some measures – Raises question of whether widening gap is acceptable