They are designed few prognostic models of early mortality in hospitalized elderly and included geriatric variables. Hypothesis: Survival after hospital admission in the elderly probably depend on individual frailty and severity of clinical situation. The main purpose was to develop and validate a predictive model stratified early mortality in elderly patients admitted to the hospital for acute medical conditions based on measurements of severity and frailty at the time admission. It was designed a retrospective cohort study with development and validation cohorts in wards specialized in geriatrics at the Hospital de Sant Pau in Barcelona from March 1, 2002, through June 30, 2002 (development cohort), and from March 1, 2007, through June 30, 2007 (validation cohort). They included 900 patients (480 development cohort and 420 validation cohort) aged 65 and older aged admitted for an acute medical condition. The dependent variable was mortality within 30 days of admission. Sociodemographic variables, illness, comorbidity, prior functional status, severity, laboratory parameters and geriatric syndromes or conditions were collected at admission. Univariate and multivariate analysis were performing using Cox regression. The stratified model used beta coefficients of significant variables. Resulting model was validated in validation cohort. Calibration, discrimination, sensivity, specificity, predictive values and likelihood coefficients in both cohorts were calculated. Results showed a mean age of 80,1 ± 6,8 years (80,2 years in validation cohort) and 60% were women (55% in validation cohort) . Mortality at 30-days was 11,2% (development cohort) and 12,9 % (validation cohort). Univariate analysis showed significant variables: age (HR 1,12), neurologic disorder (HR 4,77), hematologic disorder (HR 14,27), APACHE (HR 1,23), APS (HR 1,22), Charlson index (HR 1,25), iAVD déficits (HR 1,24), BI (HR 0,98), number of geriatric syndromes (HR 1,43), falls (HR 3,17), cognitive impairment (HR 2,71), delirium (HR 4,21), pressure sores (HR 3,40), weightloss (HR 6,65), albumin (HR 0,89), and hemoglobin (HR 0,98). Multivariate model variables were APS (HR 1,132 [IC] 95% 1,041-1,230), Number of geriatric syndromes (HR 1,553 [IC] 95% 1,290-0,976), and Hemoglobin (HR 0,971 [IC] 95% 0,966-0,976). Calibration was adequate (p=0,292) and área down curve ROC was 0,796 [IC] 95% 0,738-0,854. The stratified model included 3 variables categorized to establish 3 risk groups: low risk (0 y 1 points with mortality probability between 0 and 1,2%), intermediate risk (2 and 3 points with mortality probability between 7,9 and 11,1%) and high risk (4 to 7 points with mortality probability between 20,1% and 27%). Positive likelihood coefficients were 3 (validation cohort) and 5,8 (development cohort). It is concluded that the survival of the elderly who are hospitalized for an acute medical condition depends on the individual frailty and severity of clinical condition. The model prediction of mortality at 30-days includes Acute Physiologic Score, the number of geriatric syndromes, and plasmatic concentration of hemoglobin. Because of the complex interrelationship predictive variables is difficult to linear models predicting mortality in elderly are properly adjusted and in the future should explore the use of dynamics mathematical techniques.
| Date of Award | 4 Nov 2016 |
|---|
| Original language | Undefined/Unknown |
|---|
| Awarding Institution | - Hospital de la Santa Creu i Sant Pau
|
|---|
| Supervisor | Salvador Benito Vales (Director) |
|---|