Confounders to the Focal Relationship. Structure Table 1: Illustration of Association between Specialists’ Involvement and COC
Confounders to the Focal Relationship. While value-based payment is a mechanism for quality improvement, a hospital’s response to corresponding financial incentives may vary by individual hospital characteristics and resources (Reiter, Nahra, Alexander, & Xxxxxxx, 2006). RBV suggests that hospital characteristics such as increased market concentration and urban location positively influence organizational access to and use of resources (Yeager, Zhang, & Xxxxx, 2015) Hospitals with greater or superior resources may be more likely to have well-developed care processes. Thus, I control for confounders representing aspects of a hospital’s resources that may positively influence care management practices and clinical performance. These confounding variables include market concentration, teaching status ("Hospital Value-Based Purchasing (VBP) Program: Trend Analysis," 2014) network affiliation (X. X. Xxxxxxx, Chan, Shortell, & D'Aunno, 2000), metropolitan location, small bed size (<200 beds) (Xxxxxxx et al., 2010), and ownership (not-for- profit) (Xxxxxx et al., 2003). Indicator variables were included for participation in any type of ACO and bundled payment contract (Daaleman et al., 2014; Xxxxxx et al., 2010; Xxxxxxxxxxx et al., 2010). Market concentration was determined using the Herfindhl-Xxxxxxxxx index (HHI), where high concentration (HHI>1800) was controlled for. Teaching status was determined by the ratio of full-time residents to the number of hospital beds, where a ratio greater than or equal to 0.25 denoted a teaching hospital.
Confounders to the Focal Relationship. A hospital’s ability to participate in risk-related contracts, and thereby assume more risk, may vary by individual hospital characteristics and resources (Barney et al., 2001; Xxxxxx et al., 2015). Hospitals with greater or superior resources may be more likely to have well-developed integration processes and strategies that leverage such resources. For example, hospitals that possess an electronic health records (EHR) system generally exhibit improved clinical outcomes due to increased capacity for clinical integration (Xxxxxx et al., 2006). Accordingly, I control for covariates representing aspects of a hospital’s resources that may positively influence participation in risk arrangements and revenue at risk such as teaching status and hospital size. I also control for the influence of organizational leadership and functional integration. Functional integration is defined as the “extent to which key support functions such as financial management, marketing, and quality improvement” are coordinated across the provider organization (Xxxxxx et al., 1994). In this vein, organizational learning and leadership support can influence provider capacity to manage financial risk (Xxxxxxxxxxx & Xxxxxx, 2016; Xxxxxx et al., 2006). Physician engagement within the organization and receptiveness to hospital leadership are also likely to be positively tied to hospital performance (Frølich et al., 2007). Since it is difficult to measure the specific strategies and systems put in place at the system and individual clinician level for functional management (e.g., monitoring practice guidelines), I identify these hospital leadership constructs as an unobservable confounder to the focal relationship. The directionality of all the aforementioned covariates will likely be positive.
Confounders to the Focal Relationship. I control for confounders representing aspects of a hospital’s resources that may positively influence hospital risk-related contracting and risk assumption. These confounding variables include market concentration, teaching status (Xxxxxxx et al., 2000; "Hospital Value-Based Purchasing (VBP) Program: Trend Analysis," 2014), network affiliation, metropolitan location, and hospital size, (Xxxxxxx et al., 2010) and ownership (not-for-profit)(Xxxxxx et al., 2003). Market concentration was determined using the Herfindhl-Xxxxxxxxx index (HHI), where high concentration (HHI>1800) was controlled for. Teaching status was determined by the ratio of full-time residents to the number of hospital beds, where a ratio greater than or equal to 0.25 denoted a teaching hospital. Hospital size was measured according to the number of beds. For most analyses, I used an indicator for different categories of bed sizes to exploit more variation. However, for certain models – notably those in which we use the first differencing approach – I created an indicator variable for small hospitals defined as hospitals with <200 beds.
