Academic Research Curriculum (ARC) #5 – Attending Guide

Pulmonary and Critical Care

Clinical Scenarios:

For these clinical scenarios, ask interns/residents to answer the question by vote or "guesstimate" first, then by using the appropriate established predictive model.

  1. A 72yo woman with a PMH of well-controlled HTN and DM2 is found to have paroxysmal atrial fibrillation. Does she need anticoagulation? What is her annual risk of stroke?
  1. A 22yo man comes to urgent care after twisting his ankle playing basketball a few hours earlier. He has had ankle problems in the past and this is his 3rd inversion injury to the same side. He is limping but able to bear weight for 6 steps before having to stop. His right ankle is swollen and tender but with no tenderness over the posterior edge of either malleolus. Does he need plain radiographs of the foot/ankle?
  1. A 35yo woman at 32 weeks' gestation presents to the emergency room with acute onset shortness of breath. She is tachycardic to 110 with an SpO2 of 88% on room air. She has a swollen and tender right calf. The remainder of her exam, labs, and imaging (including CXR) are unremarkable. What is her likelihood of having a pulmonary embolism?

Yang KL, Tobin MJ. A prospective study of indexes predicting the outcom of trials of weaning from mechanical ventilation. N Engl J Med 1991;324:1445-50.

  1. What was the purpose of this trial?
  1. What are some characteristics of a good or useful predictive model (other than that it accurately predicts the event)?
  1. What are the two parts to the development of any predictive model or index? How did this study approach them?
  1. How was the study designed?
  1. Did the study favor threshold values to minimize false positives or false negatives (maximize sensitivity or specificity)?
  1. What data was collected? Were physicians blinded to this?

f (RR); VT; peak pressure; plateau pressure; PEEP; PImax (maximal inspiratory pressure) = NIF. ABG was obtained for blood gas analysis and assessment of PaO2/PAO2 ratio.

After extubation, spirometers were used to measure minute ventilation and f, with VT derived from those measurements.

  1. What is the CROP index?
  1. What were the definitions of the weaning outcomes?
  1. Define true and false positives and negatives in the context of this study.
  1. Maximum inspiratory pressure (PImax) was the most sensitive test. f/VT was the most specific. How do you interpret this clinically?
  1. Since we are given the number of patients in each group and the sensitivity/specificity values, we can reverse engineer a 2x2 table (which can sometimes help visualize the information in a familiar way). Fill in the table below for f/VT:

D+ (weaned) / D- (weaning failed)
T+ (f/VT ≤ 105) / A / B / A+B
T- (f/VT > 105) / C / D / C+D
36 / 28 / 64
  1. ROC Curve Review

a)Y-Axis:

b)X-Axis:

c)What data makes up the points on the curve?

d)What does the diagonal line represent?

e)How do we evaluate which is the best test when comparing ROC curves?

f)OK, so what do receivers, operators, and characteristics have anything to do with this name?

g)Evaluate Figure 2. What can you conclude about the accuracy of f/VT, CROP, PImax, and VE?

h)How do the rest of the AUCs in table 4 compare with each other (data is in the text)?

  1. What are the conclusions of this paper? How does this compare to what we do now, almost 30 years later?
  1. Even though the AUC was similar, why is the CROP index not commonly used in practice today?