Monte Minute‎ > ‎Monte Minute‎ > ‎

09/06/2011 - CPR for TB respiratory isolation

posted Sep 28, 2011, 5:39 PM by Chief Resident   [ updated Oct 10, 2011, 11:04 AM by Purnema Madahar ]
Sam Cohen found some great papers that answer a question that has come up several times over the past weeks: is there a good prediction rule that helps us decide who should be isolated for TB?

Methodologically very crisp and developed in New York hospitals, hence applicable to our setting, is a set of papers that first derived a prediction rule in a multivariate, retrospective case control design and then validated the rule prospectively in a different population.

The prediction rule assigns points in the following fashion:

TB risk factors or chronic symptoms      4 points
Positive PPD tuberculin test result         5 points
Shortness of breath                             -3 points
Temperature, °C
            < 38.5                                     0 points
            38.5-39.0                                 3 points
            > 39.0                                     6 points
Crackles noted during examination       -3 points
Upper-lobe consolidation                       6 points

In the validation cohort, the prevalence of TB was 19 of 516 pts (3.7%). The negative predictive value of having a score of less than 1 was 99.7%. Application of the decision rule to the study population would have led to a 35% reduction in the number of patients without TB who were isolated on admission to the hospital.

In line with our clinical teaching, shortness of breath is negatively associated with a diagnosis of TB. I was not aware that crackles also seem to be a negative predictor.

The same authors also published a review of prediction rules, which identified a number of studies with similar prediction rules with similar performance.