Usefulness of simple biomarkers at admission as independent indicators and predictors of in-hospital mortality in older hip fracture patients

Abstract

Introduction: The data on predictive value of the routinely obtained preoperative biochemical parameters in hip fracture (HF) patients are limited. The aims of this study were to examine in older HF patients (1) the relationships between a broad set of routine laboratory parameters at admission and in-hospital mortality, and (2) evaluate the prognostic value the biomarkers and clinical characteristics (alone or in combination) provide to predict a fatal outcome.

Patients and methods: In 1820 consecutive patients with low-trauma osteoporotic HF aged >60 years (mean age 82.8 +/- 8.1 years; 76.4% women; 65% community-dwelling) 35 laboratory variables along with 20 clinical and socio-demographic characteristics at admission were analysed. The validation cohort included data on 455 older (>= 60 years of age) HF patients (mean age 82.1 +/- 8.0 years, 72.1% women).

Results: The mortality rate was 6% (n = 109). On univariate analysis 14 laboratory and 8 clinical parameters have been associated with in-hospital mortality. Multiple regression analyses determined 7 variables at admission as independent indicators of a fatal outcome: 4 biomarkers (albumin <33 g/L; alanine aminotransferase/gamma-glutamyl transferase ratio [GGT/ALT] >2.5; parathyroid hormone [PTH] >6.8 pmol/L; 25(OH) vitamin D < 25 nmol/L) and 3 pre-fracture clinical conditions (history of myocardial infarction, chronic kidney disease [GFR <60ml/min/1.73 m(2)] and chronic obstructive pulmonary disease); the area under the receiver operating characteristic curve (AUC) was 0.75 (95% CI 0.70-0.80). The risk of in-hospital death was 1.6-2.6 times higher in subjects with any of these risk factors (RFs), and increased by 2.6-6.0-fold in patients with any two RFs (versus no RFs). The mortality rate increased stepwise as the number of RFs increased (from 0.43% -none RF to 16.8%- >= 4RF). The prognostic value of a single RF was low (AUC <= 0.635) but combination of 2 or more RFs improved the prediction significantly; AUC reached 0.84(95% CI 0.77-0.90) when >= 4 RFs (versus 0-1RF) were present. In the validated and main cohorts the number of predicted by 1, 2, 3 or >= 4 RFs and observed deaths were practically similar.

Conclusions: In HF patients, seven easily identifiable at admission characteristics, including 4 biomarkers, are strong and independent indicators of in-hospital mortality and can be used for risk stratification and individualised management.