Variation in the Use of Diagnostic Bronchoscopy Among ICU Patients: Implications for Suveillance

Background:
Current CDC guidelines for diagnosing ventilator-associated pneumonia (VAP) include subjective measures. Some have suggested using a microbiological definition of VAP based on lower respiratory cultures. Variation in the frequency of diagnostic bronchoscopy by intensivists could introduce ascertainment bias to VAP surveillance.

Objective:

To determine the variability in performance of diagnostic bronchoscopy on ventilated patients (pts) among ICUs and individual intensivists at an academic medical center.

Methods:

A retrospective cohort of mechanically ventilated pts admitted between June 2008 and May 2010 to six intensivist-staffed, closed ICUs (3 medical, 3 surgical; 109 beds total) was analyzed. Performance of bronchial washing, bronchoalvaeolar lavage, or bronchial brushing cultures was used as an indication of a diagnostic bronchoscopy. Diagnostic bronchoscopy rates [# diagnostic bronchoscopies/1000 ventilator (vent) days] were calculated for each ICU. Intensivist-specific bronchoscopy rates were calculated from June 2008 to August 2009 among intensivists attending for 100 vent days. Periods of increased regional influenza activity were determined by St. Louis County Health Department infection surveillance data. Chi-square analysis was used to compare rates.

Results:

A total of 38,845 vent days occurred during the study (mean=6.67 vent days/pt). 723 diagnostic bronchoscopies were performed on 606 ventilated pts, for an overall rate of 18.6 bronchoscopies per 1000 vent days. Individual ICU rates ranged from 4.8 to 31.2, and 5 were statistically different from the overall rate (p<0.01, Figure). Diagnostic bronchoscopies were more common in medical vs. surgical ICUs [24.8 vs. 14.1 bronchoscopies per 1000 vent days; Rate Ratio (RR)=1.76; p<0.01]. Surgical ICU bronchoscopy rates increased over the 24-month study period (in the first 6 months 13.7 vs. in the last 6 months 23.2 bronchoscopies per 1000 vent days; c² for trend = 15.7, p<0.01). Physician-specific bronchoscopy rates were calculated for 39 intensivists. The median rate was 17.9 bronchoscopies/1000 vent days (range 0-47.2). The monthly rate of diagnostic bronchoscopy during the influenza season was lower than non-influenza months (15.5 vs. 19.8 bronchoscopies per 1000 vent days; RR=0.78; p<0.01).

Conclusions:

The use of diagnostic bronchoscopy among mechanically ventilated ICU patients was higher in surgical vs. medical ICUs. The rate of diagnostic bronchoscopy increased over time in surgical ICUs. Considerable variation in the use of diagnostic bronchoscopy existed between intensivists.

Figure: Overall Diagnostic Bronchoscopy Rates with 95% CI by ICU (June 2008 - May 2010)