| Related Articles |
Predicting the in-patient outcomes of acute medical admissions from the nursing home: the experience of St James's Hospital, Dublin, 2002-2010.
Geriatr Gerontol Int. 2012 Oct;12(4):703-13
Authors: Romero-Ortuno R, O'Shea D, Silke B
Abstract
AIM: To identify predictors of negative in-patient outcomes (prolonged hospital stay and death) in nursing home (NH) residents admitted to the hospital as medical emergencies.
METHODS: This was a retrospective patient series set at St James's Hospital (Dublin, Ireland). The participants were all NH patients requiring acute medical admission under the on-call medical team between 1 January 2002 and 31 December 2010. Patient characteristics on admission, such as demographics, comorbidity level, major diagnostic categories, vital signs and laboratory profile, were measured. The outcomes of the study were prolonged hospital stay (? 30 days) and in-hospital mortality. The characteristics of NH patients were compared with those of non-NH patients aged ? 65 years. Multivariate analyses were based on generalized estimating equations and classification trees.
RESULTS: There were 55,763 acute medical admissions over the period, of which 1938 (3.5%) were from NH. As compared with non-NH patients aged ? 65 years, NH patients had greater acute illness severity. NH patients had a median length of stay of 7 days, and 17% had a prolonged admission. Their overall mortality rate was 23%. However, the classification analysis showed substantial patient heterogeneity; the subgroup with the highest mortality (54%, n = 100) had positive serum troponin and a respiratory major diagnosis. The lowest mortality rate (4%) was seen in those without positive troponin, urea of 12 mmol/L or less, and albumin of more than 37 mg/L (n = 226).
CONCLUSIONS: Simple serum markers, such as troponin, urea and albumin, might predict mortality in medically admitted NH patients. This might help health-care practitioners to anticipate their clinical course at an early stage.
PMID: 22469193 [PubMed - indexed for MEDLINE]