The dependent variables differ by study (see Appendix D). The majority of the studies examining health plan performance study multiple measures of quality (Ayanian et al., 2014; Bird et al., 2007; Brennan & Shepard, 2010; Cahow et al., 2010; Chou et al., 2007a, 2007b ; Couto et al., 2014; Fremont et al., 2005a; Harman et al., 2010; Inovalon, 2014b; Jung et al., 2016; Mahmoudi et al., 2016; Priest et al., 2012; Report to Congress: Social Risk Factors and Performance Under Medicare's Value-Based
Purchasing Program, 2016; Schneider et al., 2002; Trivedi et al., 2005, 2006; Virnig et al., 2002, 2004, 2007; Young et al., 2014). However, a few look at a single measure only.
As discussed in the background section, the Medicare stars measure set changes annually. Because this study seeks to review the impact of the full effect of the Medicare stars program on health plan performance, and the post-ACA MA stars methodology went into full effect at the end of 2014, this literature review examines articles relative to the 2014 measure set.
Several of the articles examining the impact of SES characteristics on plan performance look at measures that had been removed from the measure set by 2014 (Bird et al., 2007; Brennan & Shepard, 2010; Cahow et al., 2010; Chou et al., 2007; Schneider et al., 2002; Trivedi et al., 2005; Virnig et al., 2002, 2007), but for which similar or successor measures were included in the 2014 measure set. For example, ten articles look at whether hemoglobin A1C (blood sugar) is checked for diabetics (a process measure). This measure was not included in the 2014 star measures set, but a related intermediate outcome measure, blood sugar controlled in members with diabetes, was included in the methodology. Similarly, two articles look at follow-up after treatment for mental illness ( Virnig et al., 2004, 2007). This is no longer a star measure, but, in 2014, a measure was used that is derived from the Health Outcomes Survey entitled “improving or maintaining mental health,” which is defined as the percentage of plan members whose mental health is the same or better than expected after two years. This measure is examined in one article (Harman et al., 2010). Appendix C details the measures included in each article and whether that measure or a related measure was included in the 2014 measure set.
Some of the 2014 Medicare stars measures are either not found in the literature or found
infrequently. For example, only studies that include plan-level data (Cahow et al., 2010; Inovalon, 2013, 2014b; Report to Congress: Social Risk Factors and Performance Under Medicare's Value-Based Purchasing Programs, 2016) and the 2015 Inovalon study (Inovalon, 2015) include the measure “management of osteoporosis following fracture,” and only two articles examine colorectal cancer screening (Cahow et al., 2010; Report to Congress: Social Risk Factors and Performance Under
status (Report to Congress: Social Risk Factors and Performance Under Medicare's Value-Based
Purchasing Programs, 2016). It is possible that the data lag between the addition of these measures to the methodology and the publication of the identified articles is responsible for their exclusion.
The dependent variable studied in the hospital readmission articles varies based on the
readmission window examined and the condition studied. Eighteen articles look at readmissions within 30 days of discharge ( Barnett et al., 2015; Bernheim, 2016; Blum et al., 2014; Carey, 2016; Eapen et al., 2015; Greysen et al., 2015; Gu et al., 2014; Herrin et al., 2015; Hu et al., 2014b; Joynt & Jha, 2011; Lindenauer et al., 2013; McHugh et al., 2010; Nagasako et al., 2014; Rodriguez et al., 2011; Sheingold et al., 2016; Singh et al., 2013; Tsai et al., 2014); four studies look at hospitals receiving penalties under the hospital readmission reduction program, which uses as its measure readmissions within 30 days of discharge for certain conditions (Carey, 2016; Figueroa et al., 2016; Joynt & Jha, 2013b; Kahn et al., 2015); one looks at both patient and hospital level data under the hospital readmissions reduction program (Report to Congress: Social Risk Factors and Performance Under Medicare's Value-Based Purchasing Programs, 2016), one looks at readmission within 60 days of discharge (Arbaje et al., 2008a); two look at readmission within one year of discharge (Rathore et al., 2003a, 2006); two look at readmission within six months of initial admission (Aranda et al., 2009; Krumholz et al., 2000); and two look at readmission within six months of discharge (Krumholz et al., 1997, 2000).
Four of the articles examine the characteristics of hospitals that received penalties under the Hospital Readmissions Reduction Program and Hospital Value-Based Purchasing Program (Carey, 2016; Figueroa et al., 2016; Joynt & Jha, 2013b; Kahn et al., 2015), one looks at both patient and hospital level data under the hospital readmissions reduction program (Report to Congress: Social Risk Factors and Performance Under Medicare's Value-Based Purchasing Programs, 2016). The remaining articles look at the impact of patient characteristics on hospital readmissions. Seven studies look at readmissions for heart failure (Aranda et al., 2009; Blum et al., 2014; Eapen et al., 2015; Krumholz et al., 1997, 2000; Rathore et al., 2003a, 2006). Eleven studies explore readmissions for heart failure, pneumonia, and acute myocardial infarction (Bernheim, 2016; Greysen et al., 2015; Gu et al., 2014; Herrin et al., 2015; Hu et al., 2014b;
Joynt& Jha, 2011; Lindenauer et al., 2013; McHugh et al., 2010; Nagasako et al., 2014; Sheingold et al., 2016). Two articles study readmission measures after heart failure and myocardial infarction (Damiani et al., 2015; Rodriguez et al., 2011). One article studies readmissions after coronary artery bypass grafting, pulmonary lobectomy, endovascular abdominal aortic aneurysm repair, colectomy, and hip replacement (Tsai et al., 2014). One article examines readmissions for heart failure, pneumonia, and acute myocardial infarction, chronic obstructive pulmonary disease, and total knee or hip arthroplasty as well as all cause readmission (Report to Congress: Social Risk Factors and Performance Under Medicare's Value-Based Purchasing Programs, 2016). Four studies examine only all-cause readmission (Arbaje et al., 2008a; Barnett et al., 2015; Fischer et al., 2014; Singh et al., 2013).