PATIENT INVOLVEMENT IN CARE MANAGEMENT PROGRAMS FOR CHRONIC CONDITIONS
Sutch, Stephen Paul
MetadataShow full item record
Background The World Health Organization sees the increase in chronic diseases as “one of the greatest challenges that will face health systems globally in the twenty-first century”. Chronic disease affects around 50% of the US adult population, with 25% having multiple chronic conditions. Patient engagement and participation in their own care, are essential for the effective control and management of their conditions. The achievement of patient engagement is multifaceted. It can be affected by access to care, belief and knowledge, provider relationships, provider actions, co-morbidities, and personal and social circumstances. A better understanding of key factors that are associated with a patient achieving engagement would enable interventions to be taken that would increase engagement and ultimately improve care quality and outcomes. Care management has been advocated as a model of care for individuals with chronic conditions to improve health outcomes and quality of care, through improving care coordination, patient support and self-management. Care management has been found to be associated with better quality and satisfaction in care, but results with respect to cost savings differ, and changes in utilization are mixed. An important aspect in care management participation is understanding how people engage with their care, and the underlying beliefs that would lead to decisions to participate in care programs. For care management to be effective, not only must patients participate but also engage with their care and the health professionals supporting them. What influences a patient’s engagement and the resulting outcomes are important to understand in order to ensure care management interventions are appropriately designed and implemented. Goals and Aims The overarching goals of this project were to develop measures of patient care management program participation for persons with chronic disease, to gain understanding of the underlying factors and consequences of this participation in order to help improve the care management process. The specific aims of the study are: 1) To develop an approach for defining and measuring the achievement of patient participation in the care management context; 2) To apply these measurement constructs to determine which individual and organizational factors are associated with patients’ active engagement in their own care; 3) To develop a model to predict participation at various stages in the disease management cycle and to estimate the independent effects of participation on care process and utilization; and 4) To create and recommend metrics for the measurement of care management participation across the study populations to enhance understanding of patient groups and sub-populations. The study population included adult patients during the period September 1, 2009 to December 31, 2012 identified for care management in three Johns Hopkins Healthcare health plans. The study focused on care provided under contract to the managed care plans including primary care, outpatient and inpatient hospital care and care management intervention programs. The study data consisted of administrative health plan data including claims, enrollment files, care management records and patient self-reported data where available. The design was retrospective, focusing on a population of patients screened for care management across a 3-year period. There were two main dependent variables, enrollment/participation in care management programs and future health expenditure. A measure of participation, whether an individual enrolled or did not enroll in care management, was derived from claims and care management data during the initial stages of the study. A predictive model was produced from the routine administrative data, utilizing the patient variables associated with participation, to predict future cost and this was validated across all three health plans. A further predictive model was produced with the addition of patient reported variables from the Personal Wellness profile, and created a Care Management Participation Likelihood (CMPL) score based on each of the available consumer reported and administrative variables. Four sub-population “care complexity” groups were created using a modified factor analysis and clustering method. The future (year 2) health expenditures were compared for the care management enrollees against the non-enrollees using two paired matching algorithms. Findings Cost reductions were shown overall for the care management enrolled populations across the three plans, with the analysis across the complexity sub-groups showing that the cost reduction was achieved across three of the four sub-groups, with the exception of increased costs for the most complex group. Patients with higher multimorbidity, and older patients, holding other effects the same, were associated with a lower propensity to enroll in care management programs. Higher enrollment in the care management programs (holding other effects the same) was shown therefore in younger and less multimorbid patients. Higher propensity to enroll was also found in black patients compared to white patients for the Medicare plan. For all plans, members who had been referred to care management also showed increased enrolment. Multi-level (random effects) models were utilized to check that these effects remained when accounting for the higher level regional and case manager effects. Using two Propensity Scoring methods all three plans showed cost reductions for care management enrollees compared to non-enrollees. The Employee plan showed cost reductions for care management enrollees from $4186.91 to $4486.86 (17.1% - 18.3%), the Medicaid plan showed reduced costs of between $1372.66 and $4074.07 (4.6% - 13.3%), and the Family Health plan showed cost reductions for those enrolled in care management from $2458.51 to $2604.29 (7.3% - 7.7%). The plan populations were further broken down into four complexity sub-groups derived from the factor and clustering analyses, with the lower costs for care management enrolled patients compared to the non-enrolled holding for the three least complex groups, but higher costs for enrolled patients in the fourth most complex group. Summation The study, while not seeking to evaluate the current care management programs, provides a measure of participation, individual factors underlying participation, predictive models, and groupings that could be utilized in future program evaluations. The study is of a managed care plan serving multiple populations in Maryland, and while not fully generalizable to all settings, could be expected to inform other managed care organizations in the US and worldwide in other organized delivery systems.