Measuring the Value of Improving Access to Community Care

点击翻译按钮获取中文摘要

Background: Providing timely access to health care has been a long-standing VA goal that has been re- emphasized by the Commission on Care. To improve access to care, VA implemented the Veterans Choice Program in August 2014, which provides eligible Veterans the option of receiving care from community providers paid for by VA. Currently, there is a substantial gap in scientific evidence on the effect of the Choice Program and other VA community care programs, particularly in regards to the degree the program has improved access to care. More generally, non-VA literature examining the effect of greater provider options has focused on changes in utilization, but has not assessed the value of improved patient choice. The development of measures that capture the value of greater provider options is methodologically challenging because patients’ preferences are not directly observed and value encompasses many dimensions of access (e.g., travel distance, appointment wait times, provider quality, etc.). To address these evidence gaps, we propose the development and examination of new measures capturing the value of provider options to Veterans using state-of-the art econometric methods. Greater scientific evidence to help VA provide enhanced choice for Veterans through the Choice Program and future VA community care programs is consistent with the VHA FY 2018-2019 Operational Plan. This study addresses the ORD-wide Learning Health Care System priority area and HSR&D’s Access and Health Care Systems Change major priority domains. Objectives: The objectives of this study are to: 1) develop new econometric method applications to quantitatively measure the value of greater access to providers from the perspective of Veterans and 2) examine the relative importance of local area and provider characteristics in determining Veterans’ value of having improved access to providers. Methods: This observational study will examine VA administrative data and existing public data characterizing outpatient providers. In Aim 1, we will use VA administrative data to identify: 1) Veterans eligible for the VA Choice Program in 2016, 2) VA and Choice outpatient providers and 3) utilization of outpatient services from VA facilities and through the Choice Program. We will analyze Veterans’ revealed preference for providers using econometric random utility models. These models assume patients select the provider that yields the greatest benefit, given all available options. We will empirically estimate Veterans’ choice of provider within the random utility framework using a nested multinomial logit model (NMNL). We will then use parameter estimates and predictive margins from the NMNL model to calculate the value of greater provider options through the Choice Program. Specifically, econometric models will calculate Veterans’ willingness to pay (WTP), which represents the maximum dollar amount an individual would theoretically pay for greater provider options. In Aim 2, we will apply econometric decomposition methods to models developed in Aim 1 to assess the influence of key provider and local area characteristics in determining value. Notably, we will leverage novel data linkages between VA administrative data and public use data capturing an extensive set of provider characteristics. Statement on Next Steps: We will develop a simulation tool designed for non-researchers that incorporates study results to estimate the value to Veterans of a specified set of provider options (i.e. a community care network). This tool will provide the ability for operational partners to assess the adequacy of community care networks and establish the business case of “what-if” scenarios. Stakeholders will be able to adapt to changing conditions through simulating the hypothetical addition and subtraction of providers within a community care network. This simulation feature will facilitate future analyses to ensure community care networks include high quality providers that best match Veterans’ preferences.