Validation of a Mortality Prediction Model for Prolonged Mechanical Ventilation

Many patients who survive the first few days of critical illness do so with multiple residual organ failures. These patients become dependent on mechanical ventilation and other organ support systems initiated in the ICU. Patients requiring prolonged mechanical ventilation (PMV) account for up to 20% of mechanically ventilated patients and consume up to 40% of all ICU resources. The patients have survived the initial severe stages of their illness, yet they remain dependent on life support systems, progress is slow, and complications are frequent. Physicians are often confused and uncertain about prognosis for PMV patients, therefore they usually do not share prognostic information with patients’ families. In order to clarify prognosis for these complicated patients, a prognostic model that identifies PMV patients who are at high risk of death at 3 months and 1 year was developed and validated at a single tertiary care medical center. This model is based upon 4 easily measured variables and converts to a clinical prediction rule called the ProVent Score. External validation of the model is required before general clinical application can be considered. This study proposes to validate the PMV prognostic model in two external cohorts with the following Aims: 1. To validate the PMV prognostic model and ProVent Score in a heterogeneous group of patients from multiple medical centers. 2. To evaluate the performance of the ProVent Score and develop new models in patients who are earlier in the course of PMV. 3. To validate the ProVent Score in patients with Acute Lung Injury and the Acute Respiratory Distress Syndrome. The cohort for the first two Aims will consist of 600 consecutive PMV patients identified at 5 diverse tertiary care hospitals. Data will be obtained by review of medical records, and one-year survival will be confirmed by the National Death Index. The second cohort will consist of all patients in the NHLBI ARDS Clinical Trials Network FACTT trial who received PMV. A valid PMV prognostic model will allow investigators to standardize illness severity in future studies of interventions for PMV patients, and a clinically useful prognostic score will enhance the confidence of clinicians in communicating prognosis to patients and families.

Principal Investigator: Shannon Carson, PhD
Primary Funding Source: NIH
Total Project Period: 09/01/10-09/29/12