© 2007 American Thoracic Society
One More Prediction Rule: Has Anything New Been Added?To the Editor:In their recent article, España and colleagues have made an effort to derive an easily calculable predictive score for severe community acquired pneumonia (SCAP) (1). The study population appears to be fairly general with the exclusion of immunocompromised patients. Whether diabetics and patients with chronic renal failure were included is not clear. How many patients had evidence of infection in respiratory secretions or the bloodstream is not evident. After defining septic shock as a criterion for the presence of SCAP, the authors go on to state that a systolic blood pressure less than 90 mm Hg was contributory in 32 false-positive cases. Though the rule does help triage by achieving a good negative predictive value (NPV), what rankles is the low positive predictive value (PPV). Contrary to what the authors claim, a low PPV precludes the rule from identifying patients at risk for an adverse outcome. If triaging patients at risk for SCAP means admission to intermediate or intensive care, the cost of 414 false-positive cases would be enormous. It would be interesting to see how many patients were in septic shock or needed mechanical ventilation (MV) at arrival since they were treated as endpoints rather than being logically weighted as was done in the m-ATS score (2). By using shock and MV as endpoints rather than emergency department (ED) variables, other parameters expected to be more frequent in these patients are given greater weight than they probably deserve. It would be interesting to see the R2 for the stepwise multivariate models. The basis for concluding that this rule could identify at admission patients who subsequently develop SCAP within and beyond the first 24 hours is unclear. The sensitivity, specificity, PPV, and NPV can change depending on the cutoff chosen. There is an inherent bias in comparing the scores within the population from which the rule is derived, thereby putting other scores at a disadvantage. On closer look, though, it is evident that this rule did no better than the other scores in the internal and external validation groups. In the face of such bias, a comparison of the calibration of the four rules would have been more informative, proving more useful for the individual patient, rather than the original cohort as a whole. Also, some of the other scores are simpler (3, 4). Though the authors have made a commendable effort, this rule at best adds to the plethora of derived scores.
St. Luke's Hospital, Bethlehem, Pennsylvania FOOTNOTES Conflict of Interest Statement: S.P.B. has no financial relationship with a commercial entity that has an interest in the subject of this manuscript. REFERENCES
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
