Medicare data provide information that hospitals submit for billing purposes, Medpar, or Medicare Provider Analysis and Review. It is publicly available (for a fee) at http://www.cms.hhs.gov/Limited- DataSets/02_MEDPARLDSHospitalNational.asp. There are multiple forms in Medicare data and we provide the SAS code on how to “unpack” the different forms for use in analysis. We are using the 2005 version of the data. It can be provided for one or several providers. We use Medpar data from a Wound Care Center to investigate the treatment of diabetic foot ulcers. We want to determine how such patients are treated, especially those with infections. The Medicare population is at highest risk for such problems. Using a higher risk population means that there will be more patients with what are, essentially, rare occurrences of a disease. As stated in the Chapter 3, MEPS data is insufficient to examine rare occurrences, so we need to use additional data for such problems.
TopBackground
Medpar data have not been used as frequently as other datasets to investigate health outcomes. (Cornelius, Feldman, Marsteller, & Liu, 1994; Ellis & Dushman-Ellis, 2000; Isaac, Jha, Isaac, & Jha, 2008; Preminger, et al., 2008; Sharma, et al., 2009; Vitale, et al., 1999; Wei, Mark, Hartz, & Campbell, 1995; Welke, et al., 2007) It has been used to provide dollar values for comparative effectiveness analysis. (Englesbe, Dimick, Fan, Baser, & Birkmeyer, 2009; Preminger, et al., 2008; Reynolds, et al., 2006; Tong, et al., 2009)
One of the big questions is how data restricted to the Medicare population can be generalized to the rest of the population. (Ash, et al., 2003; Carey, et al., 2008; Needleman, et al., 2003; Stringham, Young, Stringham, & Young, 2005; Welke, et al., 2007) While it probably cannot be so generalized since the population within Medpar is older than the general population, it can serve to focus on medical problems that can occur in higher proportion with an older population. In particular, it can be used to examine expenditures toward the end of life, especially with diagnoses of terminal diseases such as cancer. (Anderson, et al., 2007; Ash, et al., 2003; Barnato, et al., 2004; Haller, Gessert, Haller, & Gessert, 2007) Therefore, Medpar data are frequently used to examine cost-effectiveness, or comparative-effectiveness. (Birkmeyer, Lucas, & Wennberg, 1999; S. T. Fleming, 1995; McCandless & McCandless, 2002; Preminger, et al., 2008) In particular, one such study examined the speculative cost savings by requiring surgery to be performed in regional centers. (Ellis & Dushman-Ellis, 2000) Another study examined the cost-benefits of “do not resuscitate orders”, suggesting that a lack of such orders significantly increased costs. (Haller, et al., 2007) Other studies have focused on the added cost of complications and comorbidities for patients with specific conditions. (Bond, Raehl, & Raehl, 2006; Reynolds, et al., 2006) One very interesting study examined the results of cost studies when comparing Medpar to a second database, showing inconsistencies. (N. A. Halpern, et al., 2007)