Posthospital care transitions: patterns, complications, and risk identification

Health Services Research, Oct, 2004 by Eric A. Coleman, Sung-joon Min, Alyssa Chomiak, Andrew M. Kramer

Patients with continuous complex care needs frequently require care in multiple settings and are particularly vulnerable to poorly executed transitions (Coleman 2003; Coleman and Boult 2003). The extent to which medical errors, in general, and medication errors, in particular, occur during care transitions is only recently becoming understood (Coleman 2003; Institute of Medicine 2001; Coleman et al. 2002; Forster et al. 2003; Halm et al. 2003; Boockvar et al. 2003; Moore et al. 2003; Beers, Sliwkowski, and Brooks 1992; Meredith et al. 2002; Cook, Render, and Woods 2000; Delgado-Rodriguez et al. 2001; Mitchell, Swift, and Gilbert 1999). Although researchers have examined single care transitions (e.g., returning to the hospital after being discharged or transferring from a nursing home to a hospital), to date there has been a paucity of studies that have explored entire episodes of care--that is, the multiple transitions that these patients often experience (Ashton et al. 1995; Weissman et al. 1999; Barker et al. 1994; Oddone et al. 1996; Jones et al. 1997; Lewis, Cretin, and Kane 1985; Waite et al. 1994; Sin, Manning, and Benjamin 1990; Densen 1991).

The recent Institute of Medicine report, Crossing the Quality Chasm: A New Health System for the Twenty-first Century, calls for greater integration of health care delivery across different settings to improve the quality of care transitions and reduce the threat of medical errors (Institute of Medicine 2001). However, before quality improvement approaches can be initiated, the patterns of posthospital care transitions need to be better understood. In addition, it is important to characterize the frequency and complexity of these transitions. Finally, it is necessary to develop effective targeting strategies that can be used to identify those patients who are at greatest risk for experiencing complicated care transitions and match those patients with evidence-based interventions aimed at improving the quality of their care transitions (Naylor et al. 1999; Rich et al. 1995; Philbin 1999; Townsend et al. 1988; Parry et al. 2003). This investigation aimed to: (1) describe the patterns and prevalence of care transitions among Medicare beneficiaries during a 30-day time period following acute hospitalization; (2) characterize the complexity of these care transitions; and (3) develop and test predictive indices designed to identify Medicare beneficiaries who are at greatest risk for complicated care transitions.

METHODS

Setting

This study utilized the 1997 and 1998 Medicare Current Beneficiary Survey (MCBS) Cost and Use files and accompanying Medicare claims data. The MCBS is a continuous, multipurpose survey of a representative sample of Medicare patients that was designed to aid the Centers for Medicare and Medicaid Services (CMS) in administering, monitoring, and evaluating the Medicare program. The results from the survey are combined with information from administrative data files that the CMS maintains. Additional details of the MCBS are described elsewhere (Adler 1994; Centers for Medicare and Medicaid Services 2004). (Permission to use these files and to publish these results was granted under CMS Data Use Agreement number 11941. The Institutional Review Board of the authors' academic institution approved this study, protocol number 02-035.)

Participants

The study sample included Medicare beneficiaries who participated in the MCBS and were discharged from an acute care hospital in 1997 (n = 700) or 1998 (n = 704). Medicare beneficiaries who were aged 64 years or younger, residing in a long-term care institution, or enrolled in hospice care were excluded.

Unit of Analysis

The unit of analysis for each stage of this investigation was an episode of care, defined as the 30-day time period following discharge from an acute care hospital setting. All transfers that occurred within this time period were considered part of the same episode, including any additional acute hospitalizations. A patient could contribute more than one episode to the analysis. The 1997 and 1998 MCBS Cost and Use files included 726 and 738 discrete care episodes respectively.

To ensure that each episode was unique, acute hospitalizations that occurred within a 30-day time period following another acute hospitalization were included in characterizing the episode of the first hospitalization but were not treated as discrete care episodes. Because a patient's pre-morbid functional status may influence subsequent patterns of care (and is therefore important to adjust for a current assessment in the predictive models), the study sample was restricted to episodes that occurred within four months following the MCBS' annual functional assessment that takes place in September.

Patterns and Prevalence of Posthospital Transfers

In the first stage of this study, distinct patterns of posthospital transfers and their prevalence within the 1997 sample were determined. The following types of transfers were examined: transfers to skilled nursing or rehabilitation facilities, acute care hospitals, emergency departments, and patients' noninstitutional residences. Transfers to the emergency department that resulted in admission to an acute care hospital were considered a single transfer. Episodes that resulted in death within the 30-day time period were included in this stage of the study.