Murata et al developed a model for predicting the risk of hospital admission for a patient with decompensated chronic obstructive pulmonary disease (COPD) presenting to the Emergency Department (ED). This can help identify patients who may require more aggressive management. The authors are from the Veterans Affairs Medical Center and University of New Mexico in Albequerque.
Parameters:
(1) previous admission rate
(2) previous relapse rate (a relapse was a return to the ED with dyspnea within 48 hours of the visit)
(3) rate of conservative therapy
(4) FEV1 after bronchodilator therapy as percent of predicted (from 0 to 100)
(5) prebronchodilator FEV1 to vital capacity (VC) ratio in percent (from 0 to 100) with both values in liters
admission rate =
= (number of relapses after outpatient therapy) / (number of outpatient treatments for COPD)
relapse rate =
= (number of relapses after outpatient therapy) / (number of outpatient treatments for COPD)
conservative therapy rate =
= (number of relapses after outpatient therapy) / (number of outpatient treatments for COPD)
X =
= (1.22 * (admission rate)) + (1.74 * (relapse rate)) - (0.515 * (conservative therapy rate)) - (0.035 * (postbronchodilator FEV1 in percent)) + (0.022 * (FEV1 to VC ratio)) - 0.513
probability of hospital admission =
= 1 / (1 + EXP((-1) * X))
A probability > 20.8%was "high risk" for admission; otherwise the risk was "low risk".
Performance:
• The performance of the model on the training set is given in Table 1 on page 84.
• The cutoff of 20.8% has a sensitivity of 87%, specificity of 35% in the training set. The positive predictive value was 37% and negative predictive value 86%.
• In the text on page 84 the sensitivity is said to be 90% and specificity 47%.
