Abstract

Introduction

Governmental agencies recommend risk assessment of venous thrombosis (VT) for medical inpatients at admission and provision of VT prophylaxis for moderate to high risk patients. While several risk factor models for predicting hospital-acquired VT have been proposed, none have been widely accepted and few have been prospectively validated. We sought to validate the recently published MITH VT risk assessment model in an independent cohort of medical inpatients (Zakai et al, Journal of Thrombosis and Haemostasis 2013).

Methods

Hospital-acquired VT and risk factors present at admission were collected from adult inpatients between June 2009 and April 2012 admitted to the medicine, medical intensive care, hematology/oncology, or cardiology services at Fletcher Allen Hospital (500 bed teaching hospital for the University of Vermont). Hospital-acquired VT was defined using VT discharge ICD-9 codes (flagged as not present on admission) and record of an imaging study that could diagnosis VT (such as duplex ultrasound, computed tomography angiography, or ventilation perfusions scan). Inpatients with VT ICD-9 codes flagged as present on admission were excluded. The sensitivity and specificity of the definition was confirmed by chart review of 30 cases of hospital-acquired VTE and 30 non-cases. Risk factors for hospital-acquired VT were captured using ICD-9 codes from the problem list, discharge codes, vital signs, and laboratory values at admission. The MITH score was calculated for each patient based on the points for each risk factor: history of heart failure = 5 pts, history of rheumatologic disease = 4 pts, history of fracture in past 3 months = 3 pts, history of cancer in past 12 months = 1 pt, tachycardia (HR>100 at admission) = 2pt, respiratory dysfunction (SpO2<90% at admission or intubated on hospital day 1) = 1 pt, white blood cell count >11 = 1 pt, platelet count >350 = 1 pt. The absolute rates of hospital-acquired VT for different cut points of the score were calculated and compared qualitatively to those previously published for the MITH score.

Results

There were 120 hospital-acquired VT events complicating 20,334 medical admissions (5.9 cases per 1,000 hospital admissions). The sensitivity and specificity of our definition of hospital-acquired VT was 100% and 91%, respectively. The table presents the prevalence of the MITH score at various cut-offs in cases and non-cases as well as the incidence of VT. In the derivation of the MITH score, the rate of VT per 1000 admissions for a score <1, <2, or <3 was 1.0, 1.5, and 2.1 compared with 0.7, 1.8, and 2.2 VT per 1000 admissions for the validation cohort. The incidence of VTE in the derivation of the MITH score for a score ≥1, ≥2, and ≥3 was 6.0, 8.9, and 12.4 per 1000 admissions compared with 7.9, 9.0, and 10.3 per 1000 admissions in the validation cohort.

Conclusions

We have validated a previously published VT risk score for hospitalized medical patients in an independent population. Determination of a patient's risk of VT at admission using readily available clinical and laboratory data could allow physicians to make informed decisions about risks and benefits of DVT prophylaxis. Further work is required to determine at what level of risk pharmacologic VT prophylaxis is warranted in this patient population.

Table

Performance of the MITH score

MITH Scoren (%) of cases < scoren (%) of non-cases < scoreVT per 1000 admissions for < scoreVT per 1000 admissions for ≥ score
   Current Analysis Predicted (95% CI) (Previously Published) Current Analysis Predicted (95% CI) (Previously Published) 
4 (3.3%) 5,582 (27.6%) 0.7 1.0 (0.6, 1.8) 7.9 6.0 (2.0, 17.7) 
16 (13.3%) 8,710 (43.1%) 1.8 1.5 (1.0, 2.3) 9.0 8.9 (4.2, 18.8) 
26 (21.7%) 11,189 (55.4%) 2.2 2.1 (1.5, 3.0) 10.3 12.4 (6.3, 24.5) 
40 (33.3%) 12,995 (64.3%) 3.1 2.9 (2.1, 4.1) 11.0 15.5 (7.5, 31.8) 
50 (41.7%) 14,096 (69.7%) 3.5 3.3 (2.4, 4.8) 11.3 17.3 (7.5, 39.0) 
67 (55.8%) 16,447 (81.4%) 4.1 4.4 (3.2, 5.9) 13.9 34.5 (14.4, 80.4) 
79 (65.8%) 17,786 (88.0%) 4.4 4.7 (3.4, 6.3) 16.6 66.6 (21.1, 191.0) 
MITH Scoren (%) of cases < scoren (%) of non-cases < scoreVT per 1000 admissions for < scoreVT per 1000 admissions for ≥ score
   Current Analysis Predicted (95% CI) (Previously Published) Current Analysis Predicted (95% CI) (Previously Published) 
4 (3.3%) 5,582 (27.6%) 0.7 1.0 (0.6, 1.8) 7.9 6.0 (2.0, 17.7) 
16 (13.3%) 8,710 (43.1%) 1.8 1.5 (1.0, 2.3) 9.0 8.9 (4.2, 18.8) 
26 (21.7%) 11,189 (55.4%) 2.2 2.1 (1.5, 3.0) 10.3 12.4 (6.3, 24.5) 
40 (33.3%) 12,995 (64.3%) 3.1 2.9 (2.1, 4.1) 11.0 15.5 (7.5, 31.8) 
50 (41.7%) 14,096 (69.7%) 3.5 3.3 (2.4, 4.8) 11.3 17.3 (7.5, 39.0) 
67 (55.8%) 16,447 (81.4%) 4.1 4.4 (3.2, 5.9) 13.9 34.5 (14.4, 80.4) 
79 (65.8%) 17,786 (88.0%) 4.4 4.7 (3.4, 6.3) 16.6 66.6 (21.1, 191.0) 
Disclosures:

No relevant conflicts of interest to declare.

Author notes

*

Asterisk with author names denotes non-ASH members.

This icon denotes a clinically relevant abstract