A retrospective cohort study was derived from data compiled in the State of Kansas Minimum Data Set (MDS) version 2.0 for nursing home residents and in the Medicare database from July 1, 1997 to July 1, 1998. Patient confidentiality was maintained by strict adherence to the consent criteria set up by the Kansas MDS committee. The subjects' identities remained classified via special codes set up by an independent party.
The Minimum Data Set (MDS) is a health status assessment questionnaire for nursing home residents mandated by the Health Care Financing Administration (HCFA) regulations for all Medicare and Medicaid licensed facilities. An assigned nurse completes this data set at the time of admission, at quarterly intervals, upon transfers, and at the time of discharge on every nursing home resident in Kansas and several other states. This detailed questionnaire includes health-related information over a broad field including demographic information, physical functioning, nutritional status, cognitive functioning, disease diagnoses, and even medications. Studies assessing the validity of the MDS data components of cognition, physical functioning, nutritional status, continence, diagnoses, and drug data have shown the MDS to be a reliable and valid data set.8–11
All residents who were entered into the MDS database and present in a Kansas nursing home on July 1, 1997, and who subsequently had 2 consecutive quarterly assessments with no hospitalizations or readmissions met the inclusion criteria and were eligible for the study. We used the criteria of 2 quarterly assessments to achieve a minimum stay of 120 days in an attempt to capture the population of nursing home residents who were truly in long-term care rather than short-term. Residents with a recent or remote past history of VTE prior to entry into the study or residents on anticoagulation therapy were excluded from the study. A previous history of VTE was determined if a resident had an entry in the MDS database or in the Medicare database using the ICD-9 codes for the diagnoses as listed: deep vein thrombosis (453.8), pulmonary embolus (415.11), pulmonary embolus NOS (415.10), femoral vein thrombi (451.11), popliteal vein thrombi (451.19), iliac vein thrombi (451.81), superficial vein thrombi (451.0), lower extremity thrombus (451.2), and hypercoaguable state (289.8).
An assessment of anticoagulation therapy use was determined through an evaluation of the National Drug Codes (NDC) for anticoagulants in the MDS database. The anticoagulants assessed were coumadin, unfractionated heparin, and the low molecular weight heparins (danaparin, dalteparin, enoxaparin, and ardeparin). Patients on these medications were excluded from the study. Aspirin therapy was not an exclusion criterion, as we believed that at that age a large proportion of our studied cohort would be on this therapy.
The primary endpoint of this study was the development of a VTE during residence. This was determined by an entry into either the MDS or the Medicare database during the study period. Each resident's MDS data (demographic, etc.) were extracted at the beginning of the study period, July 1, 1997. Residents were subsequently followed through the database for their consecutive assessments up to July 1, 1998.
We used SPSS (Statistical Product and Service Solutions) 11.0 for Windows (SPSS, Inc. 1989–2001 Chicago, IL) for statistical analyses. An equal tailed 95% confidence interval for incidence of VTE was calculated using 10,000 case bootstraps of the original data, after which the 2.5% tile and the 95% tile are calculated. We generated a thrombosis-free survival curve using the Kaplan-Meier product limit function. Interpolation makes the time of occurrence of VTE midway between the last 2 times of assessment. In the cases where the last assessment is unknown, it is conservatively assumed to be 120 days before the last assessment, well within the range of quarterly assessment requirements.
The Institutional Review Board of the Kansas University School of Medicine in Wichita approved the study.