Infectious Diseases in Clinical Practice:
Clinical Prediction Rules in the Isolation of Patients With Suspected Pulmonary Tuberculosis
Sigel, Keith MD, MPH; Wisnivesky, Juan MD, PhD
From the Division of General Internal Medicine, Mount Sinai School of Medicine, New York, NY.
Reprints: Keith Sigel, Division of General Internal Medicine, Mount Sinai School of Medicine, New York, NY. E-mail: firstname.lastname@example.org.
In this issue of Infectious Diseases in Clinical Practice, Aguilar et al describe their derivation of a clinical prediction rule for isolation of suspected cases of pulmonary tuberculosis (TB).
The isolation of patients at high risk for pulmonary TB is an important measure that has shown effectiveness in decreasing nosocomial transmission and is especially important with the emergence of multidrug-resistant strains.1 The decision to isolate patients suspected of TB must balance the risk of nosocomial transmission with the low prevalence of the infection and high associated costs. Clinical prediction rules using readily available clinical indicators are useful in this situation as a means of identifying patients at low risk of active pulmonary TB. At least 4 previous prediction models have been reported in the literature with similar sensitivity as reported in the current study.2
Previously published clinical prediction rules have found the presence of TB risk factors or chronic symptoms, self-reported positive tuberculin skin test, fever, and upper lobe abnormalities on chest radiograph to be predictors of pulmonary TB. Positive tuberculin skin test and upper lobe infiltrates on chest radiograph have shown the strongest association in derivation studies, and this appears to be consistent with the current study. Factors that have been less useful in the derivation of other prediction models have included hemoptysis and white blood cell counts.2
Guidelines for evaluating the quality of clinical prediction rule studies have been proposed and applied to models regarding TB isolation.2,3 The current study appears to demonstrate many characteristics associated with a high-quality study including sufficient description of predictor variables, a detailed reference standard, succinct reporting of important patient characteristics, and a clear account of statistical methods. It is not reported, however, if predictors were ascertained with the knowledge of culture results, potentially biasing the identification of these details during the data collection process.
The findings reported in this study represent the derivation of a clinical prediction rule and thus require a validation evaluation to merit clinical applicability.4 Previously reported clinical prediction rules have demonstrated diminished precision when used in different validation populations.5 The negative prediction value of human immunodeficiency virus seropositivity is an important aspect of the model reported in this study that should be validated on a population with a higher prevalence of human immunodeficiency virus infection.
This study is the first to include computer-aided tomography (CT) scanning as a predictor variable. Computer-aided tomography scanning has greater discriminatory power than plain radiography in the diagnosis of TB and can at times identify active disease.6 Therefore, it is not surprising that its inclusion appears to improve the sensitivity of the clinical prediction model. It is unlikely, however, that CT scanning improves the prediction model to an extent that justifies its cost. In addition, in many institutions, CT scans or their interpretation may not be available at the time of isolation, further limiting use in clinical prediction models.
Clinical prediction rules can serve as a useful tool for the identification of patients who are at a low risk for pulmonary TB and do not require respiratory precautions. Future research should be directed toward evaluating the impact of these clinical tools on physician behavior and evaluating the cost-effectiveness of implementing these rules into institutional practice.
1. Maloney SA, Pearson ML, Gordon MT, et al. Efficacy of control measures in preventing nosocomial transmission of multidrug-resistant tuberculosis to patients and health care workers. Ann Intern Med. 1995;122:90-95.
2. Wisnivesky JP, Serebrisky D, Moore C, et al. Validity of clinical prediction rules for isolating inpatients with suspected tuberculosis. A systematic review. J Gen Intern Med. 2005;20:947-952.
3. Laupacis A, Sekar N, Stiell IG. Clinical prediction rules. A review and suggested modifications of methodological standards. JAMA. 1997;277:488-494.
4. McGinn TG, Guyatt GH, Wyer PC, et al. Users' guides to the medical literature: XXII: how to use articles about clinical decision rules. Evidence-Based Medicine Working Group. JAMA. 2000;284:79-84.
5. Wisnivesky JP, Henschke C, Balentine J, et al. Prospective validation of a prediction model for isolating inpatients with suspected pulmonary tuberculosis. Arch Intern Med. 2005;165:453-457.
6. Lee KS, Hwang JW, Chung MP, et al. Utility of CT in the evaluation of pulmonary tuberculosis in patients without AIDS. Chest. 1996;110:977-984.
© 2009 Lippincott Williams & Wilkins, Inc.