Older adults with multimorbidity (MM) have greatly increased risk for adverse outcomes, whether in ambulatory or acute care hospital settings. Consequently, research intended to address problems caused by MM has recently been deemed a priority by the NIA. This proposal develops new definitions of MM in hospitalized patients that is separately defined for surgical procedures and medical conditions and examines whether hospitals vary in their ability to treat MM patients. Results from this study will be direcly actionable for both patients (as an aid in patient referral) and for hospitals (as an aid to inform hospitals how they fare with MM patients). Without development of MM groups, and without the use of methods to identify hospitals with specific problems treating MM patients, these advantages could not accrue to patients or hospitals. The project will use Medicare claims from Illinois, Pennsylvania, and Texas from 2008-2013 and apply new methods in multivariate matching to examine the interaction between MM and hospitals. Determining if a hospital is associated with good or bad outcomes is not the same as knowing if that hospital does especially well or poorly with MM patients. This requires an examination of interactions. We will examine these interactions in two new ways: (1) Construction (inside of a surgical or medical group) of a template of patients (using a form of direct standardization), half with MM and half without MM, and we will use new multivariate matching algorithms to create patient copies of this template across hospitals. (2) Using a method we title Indirect Standardization with Matching the proposal will create 10 copies (or controls selected from hospitals outside the index hospital) per patient with MM and without MM for all patients at the index hospital, allowing for a close examination of how well the hospital treats MM and non-MM patients compared to the closest matched patients selected from all other hospitals. Using both direct and indirect standardization through matching, we will study the variation in outcomes of MM and non-MM patients across hospitals. The project has 2 main aims:
AIM 1 : Identify the unique MM combinations most likely to lead to worse outcomes, specific to each condition and procedure of interest.
AIM 2 : Identify hospitals that have especially good or especially poor results for MM patients as compared to non-MM patients using Template Matching and Indirect Standardization with Matching and develop prototype MM patient outcomes reports for all hospitals studied. This knowledge will help patients by improving the process of hospital selection for this vulnerable population, which is especially at risk because a disproportionate share are minorities. This project will allow hospitals to better understand weaknesses in their care for MM patients. At present, hospitals and patients cannot define proper MM categories because such categorization requires specific analysis of large databases, and hospitals cannot identify if they specifically have problems managing such patients because such an analysis must involve asking how very similar patients fared at other hospitals-something that is ideally accomplished with multivariate matching.
Older adults with multimorbidity (MM) have greatly increased risk for death, complications, and readmissions after being admitted to the hospital for medical conditions or surgical procedures. This proposal proposes a better way to define MM, relying on large samples to identify the correct MM clusters that most influence outcomes, and applies novel techniques to produce vital information that will allow patients and hospitals to better identify those hospitals that are best and worst at caring for such patients.
| Silber, Jeffrey H; Reiter, Joseph G; Rosenbaum, Paul R et al. (2018) Defining Multimorbidity in Older Surgical Patients. Med Care 56:701-710 |
