Electronic Supplement Material
Validation of an algorithm based on direct examination of specimens in suspected ventilator-associated pneumonia
Anne Veinstein, Christian Brun-Buisson, Nicolas Derrode, Antonio Alvarez, Michel Pinsard, René Robert, François Blot
Methods
The study was performed in four ICUs (two medical ICUs, and one each surgical or mixed ICU). Patients receiving mechanical ventilation for more than 48 hours were prospectively included when clinically suspected of VAP, according to usual criteria. The clinical suspicion of VAP was based on a persistent recent or worsening pre-existing pulmonary infiltrate on chest x-ray, and the occurrence of at least two of the following clinical criteria: new onset of fever (≥ 38°2 C), leukocytosis (> 104/mm3), increase in volume and purulence of tracheal secretions, and a decrease in the PaO2/FiO2 ratio of more than 20%. Only the first episode of suspected VAP in a given patient was included in the analysis. Neutropenic patients (total blood leukocyte count < 103/mm3) were excluded from the study.
The following parameters were recorded in patients: demographic data, age and gender, reason for admission to the ICU, date of intubation, severity of acute illness, assessed by the simplified acute physiologic score (SAPS) II on ICU admission and on the day of inclusion [20], sepsis-related organ failure assessment (SOFA) score on inclusion [21], presence of sepsis and/or severe sepsis [22], severity of respiratory failure (PaO2/FiO2 ratio, PEEP level), overall duration of MV, outcome at day 14 and ICU discharge. The CPIS was also calculated in three different ways on the day of inclusion (Table S1): the modified CPIS as per Singh et al. [11], the modified CPIS including EA Gram stain examination (CPIS-EA Gram) and simplified CPIS including PTC Gram stain examination (CPIS-PTC Gram) [10]. Severity criteria at time of clinical suspicion of VAP were defined as extensive lung involvement or severe hypoxemia (PaO2/FiO2 ratio <200), or occurrence of severe sepsis or shock.
Samplings Techniques and definitions
Lower respiratory tract samples were obtained on the day of clinical suspicion of pneumonia. An EA was first obtained for Gram staining only. A PTC (Figure S1) was then performed after careful endotracheal suctioning, as previously described [14], for Gram staining and quantitative culture. Sensitivity, specificity, predictive positive and negative values of EA and PTC Gram stains were calculated according to the results of positive culture of the PTC, defined as the recovery of at least 103 cfu/ml of a potential pathogen [10, 14, 19, 23]. Candida spp. were considered as potential pathogens only in immuno-compromised patients. Specific microorganisms such as Nocardia sp., Legionella sp. were considered as pathogen whichever the bacterial count. Pneumonia due to Aspergillus sp. or other yeasts were not considered in the present study.
Empiric antibiotic regimens
A recommended first-line empiric antimicrobial regimen was defined in each center, based on local microbiological environment, the time of occurrence of suspected VAP relative to initiation of MV (early-onset; i.e. before day 7, or late-onset, i.e., 7 or more days from the beginning of MV), recent hospitalisation and prior antibiotic therapy [7, 24].
For this purpose, the antimicrobial susceptibility of microorganisms recovered during the VAP episodes during the 2 years preceding the study period was examined in each center. Thus, for early-onset VAP without prior antibiotic therapy, the first-line suggested choice was amoxicillin-clavulanic acid (at Poitiers and Henri Mondor), and amoxicillin-clavulanic acid or cefotaxime (at Institut Gustave Roussy). For early-onset VAP having received prior therapy, the first-line suggested choice was cefotaxime (at Poitiers and Henri Mondor), or cefepime (at Institut Gustave Roussy). For late-onset VAP, the suggested choice was in all four centers ceftazidime or piperacillin-tazobactam or imipenem plus amikacin when Gram-negative bacteria were seen on Gram-stain examination, with the addition of vancomycin when Gram-positive cocci were observed.
The adequacy of initial antimicrobial therapy was assessed based on definitive culture results of PTC specimens and classified as follows: (a) totally adapted, when all microorganisms recovered from cultures were adequately covered by antibiotics, (including two active antibiotics in case of infection by Pseudomonas aeruginosa); (b) partially adapted, when Pseudomonas aeruginosa was covered by only one antibiotic, or one of the microorganisms was partially covered by the empiric regimen; (c) inadequate, when at least one of the microorganisms recovered was not covered by the regimen selected, (d) excessive, when the antibiotic regimen spectrum could have be narrowed according to the microorganism(s) identified and their susceptibility pattern.
Results
Patients
The 76 patients had a mean age of 59 ± 15 year. Their mean SAPS II on ICU admission was 42 ± 17. The causes of admission were acute respiratory failure (n = 38), postoperative monitoring (n = 10), severe sepsis or septic shock (n = 8), cardiac failure (n = 8), brain failure (n = 5), trauma (n = 3), hemorrhagic shock (n = 1), metabolic disorder (n = 1), malignant hyperthermia (n = 1), drug intoxication (n = 1).
Microbiological results
Among the 41 positive PTC cultures, 30 episodes (73 %) were monomicrobial and 11 (27 %) were polymicrobial (Table S1). In 16 patients (29%), the microorganisms recovered were considered as potentially resistant (Pseudomonas aeruginosa, Stenotrophomonas maltophilia, Acinetobacter baumannii, methicillin-resistant Staphylococcus aureus).
Prior antimicrobial treatments
In 57 of the 76 episodes (75%), patients had received prior antimicrobial treatments within the two-week period preceding the suspected episode of VAP. Prior antimicrobial treatment was started or modified within 48 hours before the suspected episode of VAP in 11 of these 57 episodes. PTC culture was negative in 8 of the latter 11 episodes. None of the 3 episodes with negative EA and PTC Gram stain examinations but positive PTC culture were among this subgroup with recent modification of the prior antimicrobial regimen.
Concordance between the algorithm tested and PTC culture results
In 36 (47%) episodes, the PTC Gram stain examination was positive; in all but two of these episodes, EA direct examination was also positive. Thirty of these 36 (83%) had positive PTC cultures (Figure 2), including the two with a negative direct examination of EA.
Both Gram stain examination of EA and of PTC were negative in 21 (28%) episodes. PTC culture was negative in 18 (86%) of these episodes, and positive in only three.
In the remaining 19 (25%) episodes, EA Gram stain examination was positive whereas Gram stain PTC examination was negative; 8 (42%) had positive PTC cultures, including 4 of the 7 (57%) patients with severity criteria and 4 of the remaining 12 (33%) without such criteria (Figure 2).
Therefore, the sensitivity and specificity for the diagnosis of VAP based on Gram stain examination of EA were 88% and 51%, respectively; corresponding values for PTC were 73% and 83%. The positive and negative predictive value of Gram stain examination were 68% and 78% respectively for EA, and 83% and 73% respectively for PTC. Combining EA and PTC Gram stain examinations according to the proposed algorithm resulted in a sensitivity, specificity, positive and negative predictive values of respectively 83%, 74%, 79% and 79% (Figure 2, Table 2).
Clinical prediction of VAP based on CPIS
The modified CPIS score was ≤ 6 in 51 cases, including 23 (45%) with negative PTC cultures; it was >6 in 25 cases, including 13 (52%) with positive PTC cultures. The CPIS-EA Gram was ≤ 6 in 24 cases, including 14 (58%) with negative PTC cultures, and was > 6 in 52 cases, 31 (60%) of which had positive PTC cultures. The CPIS-PTC Gram was ≤ 6 in 33 cases, including 19 (58%) with negative PTC cultures, and was > 6 in 43 cases, including 27 (63%) positive PTC cultures. Taking a modified CPIS higher than 6 as indicative of a probable VAP justifying antimicrobial therapy, the corresponding sensitivity, specificity, positive and negative predictive values of each of these different CPIS calculations are shown Table 2, by comparison to the algorithm tested.
Adequacy of empiric antimicrobial therapy
The antibiotic regimens recommended by the protocol were adequate in 36 (88%) of the 41 episodes with positive PTC cultures; unanticipated resistance was associated with the 5 episodes where the recommended regimen was not effective.
Accordingly, 26 (90%) antibiotic regimens administered empirically to the 29 episodes actually treated and eventually having confirmed pneumonia were adequate, according to the susceptibility of organisms recovered. Nine of these 26 (35%) regimens were however of larger broad spectrum than required and de-escalation was performed within 48 hours after culture results were obtained. Altogether, and taking into account the 12 episodes not initially treated, 26 of 41 (63%) episodes with positive PTC cultures actually received initially adequate antibiotic therapy.
Table S1. Variables used to calculate the modified CPIS score.
Variable / value / No. pointsTemperature (T,°C) / 36°5 T 38°4
38°5 ≤ T ≤ 38°9
T ≤ 36° or T ≥ 39° / 0
1
2
Leukocytosis (WBC per mm3) / 4000 WBC ≤ 11 000
WBC < 4000 or > 11 000
Band forms ≥ 500 / 0
1
1
Tracheal secretions / Absence of secretion
Copious secretions
Purulent secretions / 0
1
1
PaO2/FiO2 (mmHg) / > 240 or ARDS
≤ 240 / 0
2
Pulmonary radiography / No infiltrate
Diffused or patchy infiltrate
Localized infiltrate / 0
1
2
The modified CIPS score [11] uses the five variables recorded at time of suspicion of pneumonia; when calculating the CPIS-Gram stain (CPIS-EA Gram or CPIS-PTC Gram), 2 points were added for a positive Gram stain of EA or PTC [10].
Table S2. Microorganisms identified from 41 protected specimens with positive cultures.
Escherichia coli / 6
Haemophilus sp. / 4
Klebsiella sp. / 2
Proteus mirabillis / 2
Pseudomonas fluorescens / 1
Stenotrophomonas maltophilia / 1
Acinetobacter baumannii / 1
Enterobacter cloacae
Enterobacter aerogenes / 1
1
Hafnia alvei / 1
Gram-positive cocci(n=20) / MSSA / 11
MRSA / 5
Streptococcus sp. / 3
Streptococcus pneumoniae / 1
Gram-negative cocci / Neisseria sp. / 1
Anaerobes(n=2) / Prevotella sp. / 2
Gram-positive bacilli / Corynebacterium / 1
Yeast / Candida albicans / 1
Abbreviations: MSSA = methicillin susceptible S. aureus; MRSA = methicillin-resistant S. aureus. Neisseria sp. and Corynebacterium were involved in polymicrobial infections.
Figure S1. The protected (single-sheathed plugged) telescoping catheter used for sampling lower respiratory tract secretions.
The protected catheter is composed of an outer sheath, plugged at its distal end and an inner sheath for sampling. During sampling, the device is advanced into the lower respiratory tract until a resistance is felt, then pulled back a few centimeters. The inner sheath is then advanced to expel the distal plug, and advanced further distally 3 to 5 cm. Gentle aspiration via an empty 10 ml syringe is then applied through the proximal end of the catheter; the inner sheath is then retracted back into the outer sheath, and the whole device then pulled out of the airways. After transecting the distal end of the outer sheath, the 5-cm distal end of the inner sheath is collected and sent for gram stain and culture in a test tube, after rinsing the inner catheter with 1 ml saline.
Figure S2 – Alternative algorithm based on severity of presentation of suspected VAP and the modified CPIS-Gram PTC (n=76).
Legend – This approach results in a sensitivity for VAP (as confirmed by PTC culture) of 83% (34/41) and a specificity of 48% (17/35). Thus, 51 (67%) of all patients would be managed appropriately (treated or not when appropriate), 18 patients (24%) would be treated empirically “in excess”, and 7 (9% of all patients and 17% of those with VAP) would receive delayed therapy.