Microscopic screening of respiratory specimens submitted for culture: A re-evaluation


Journal Article

Clinical laboratories examine respiratory specimens microscopically to screen out poor quality collections and optimize the clinical utility of cultures. This study evaluated the accuracy and clinical impact of screening at a large, tertiary-care medical center. All respiratory specimens submitted during a 50-day period (1/97-3/97) were screened by established Gram stain criteria, including quantitation of squamous epithelial cells and bacteria. In all, 1013 specimens (360 sputum, 653 endotracheal suction [ETS]) were analyzed. Rejection rate was significantly higher for sputum than for ETS (57% vs. 34%, p<.0001). Rejected specimens, normally discarded, were cultured and all isolates characterized. Rejected specimens yielded reportable growth significantly less often than accepted specimens (16% vs. 34%, p<.0001). Based solely on culture findings, the sensitivity and specificity of the DUMCL criteria for specimen acceptance were 77% and 40% for ETS, respectively; and for sputum, 69% and 62%. While higher sensitivity would have been desirable, only 8/71 rejected specimens with reportable growth came from patients with radiographie evidence of pneumonia (1.8% of rejected specimens overall). The pathogens isolated in these cases included Staphyhcoccus aureus (4), Pseudomonas aeruginosa (3) and Klebsiella pnettmoniae (1). Use of acceptance criteria increased net laboratory workload for respiratory cultures at the logging-in station by 20% because of repeat submissions after a rejection. It decreased workload at the culture station by 31%. Microscopic screening eliminated 48% of specimens devoid of significant bacterial growth. The data suggest strategies for improving screening, but also demonstrate that many respiratory specimens derive from patients without radiographic evidence of infection. Judicious clinical ordering practices may have greater impact on laboratory efficiency than refinements in screening techniques.

Duke Authors

Cited Authors

  • Phillips, HK; Harrell, LJ; Madden, JF

Published Date

  • December 1, 1997

Published In

Volume / Issue

  • 25 / 2

Start / End Page

  • 400 -

International Standard Serial Number (ISSN)

  • 1058-4838

Citation Source

  • Scopus