Introduction

Maxillary sinusitis is a frequent complication in intubated patients [1, 2, 3]. The occurrence of ventilator-associated pneumonia (VAP) in patients undergoing prolonged mechanical ventilation can be prevented by the systematic search and treatment of nosocomial sinusitis [4]. VAP significantly increases length of stay and mortality in the intensive care unit [5]. Nosocomial sinusitis in intensive care unit (ICU) is often underestimated and frequently under-diagnosed [6]. In fact, its diagnosis in ICU is difficult and based on the presence of radiologic maxillary sinusitis (RMS) obtained from computerized tomography (CT) and the isolation of organisms from culture of purulent material obtained after transnasal puncture [1, 4, 7, 8]. Transnasal puncture has a great importance in the diagnosis and the treatment of sinusitis [8]. CT is the modality of choice for evaluating the sinonasal cavity [9]. However, this procedure requires the transportation of patients outside the ICU [10, 11]. The property of ultrasound to cross a bone may appear paradoxical, but experience shows that flat bones made up only of compact tissue, such as the anterior wall of the maxillary sinus, are not a hindrance. The princeps study has been published by Lichtenstein et al. [12]. In this study, sinus ultrasounds were performed with the patient in supine position. The sensitivity and specificity of ultrasound compared with CT for the diagnosis of RMS were respectively 67% and 87%. However, the authors describe complete sinusogram and incomplete sinusogram and establish a total correlation between complete sinusogram and total opacity of the sinus on CT, i.e., a specificity of 100%, which indicates the possibility to safely perform puncture. A recent study by our group fully confirmed the basic importance of ultrasound in this field and showed that ultrasound accuracy for all lesions was improved in the semi-erect patient [13]. We showed that B-mode ultrasound, compared with CT, was effective in the diagnosis of RMS in intubated patients, with a sensitivity of 100%, and a specificity of 96.7% [13]. We concluded that B-mode ultrasound may be proposed as the first line for diagnosis in a ventilated patient with suspicion of maxillary sinusitis [13].

Our hypothesis was that the sinus echography results could be valuable to directly indicate a transnasal puncture, avoiding CT, in intubated patients with suspicion of nosocomial sinusitis. To test this hypothesis, we have compared sinus ultrasound findings to a gold standard, i.e., direct sinus aspiration.

This work has been presented, in part, at the European Society of Intensive Care Medicine (ESICM) congress meeting [14].

Material and methods

Study population

The study was performed in the medical ICU (16 beds) of a university teaching hospital and was carried out between April 2002 and May 2003. During this period, all patients admitted to ICU and mechanically ventilated for more than 2 days were potentially eligible for this investigation. The experimental protocol was approved by the institutional review board of the hospital, and all patients or the next of kin provided written informed consent. Patients were included in the study if the following two criteria were met: (1) intubation and mechanical ventilation more than 2 days; and (2) a clinical suspicion of maxillary sinusitis with purulent nasal discharge and body temperature 38°C. They were not included if the following criteria were present: facial or skull-base fracture; a history of sinusitis and coagulation disorders contraindicating transnasal puncture.

Patients receiving mechanical ventilation in our ICU are usually placed in a semi-recumbent position. There is no written protocol concerning the route for endotracheal and nasal intubation, and the choice is left to the discretion of the attending physician. In practice, the nasal route is mostly used for tracheal intubation. In nasotracheally intubated patients, the nasogastric tube is inserted in the opposite naris to the nasotracheal tube.

Data collection

The following variables were recorded from patients: age, sex, New Simplified Acute Physiologic Score (SAPS II) [15], diagnosis on admission, temperature, leukocytosis, ventilator-associated bronchopneumonia, and duration of intubation at the moment of suspicion of sinusitis. The diagnosis of ventilator-associated bronchopneumonia was made according to American Thoracic Society criteria [16].

Study protocol

B-mode ultrasound

The method was similar to the procedure previously described [13]. The ultrasonographer was blinded to the clinical examination and was not in charge of the patient. All patients were in a semi-recumbent position. As shown in Fig. 1, three different images can be observed. The image defined as normal (negative test) was an acoustic shadow arising from the front wall. Two levels of positive echography (positive test) were described for ultrasonographic maxillary scanning: (a) a partial sinusogram and (b) an important lesion or complete sinusogram.

Fig. 1
figure 1

Images obtained by sinus ultrasound: (a) a partial sinusogram (moderate lesion) defined by the visualization of the posterior wall of the sinus; (b) a complete sinusogram (important lesion) defined by the visualization of the posterior and the extension by the internal and external walls of the sinus outlining a hypoechogenic sinus cavity; (c) a normal sinus defined by the visualization of an acoustic shadow

Full size image

Transnasal puncture

When B-mode sinus ultrasound was positive, a transnasal puncture was performed the same day by an otorhinolaryngologist. The patients were adequately pre-medicated with midazolam and sedated with fentanyl. Before transnasal puncture, we performed a large disinfection of the nasal mucosa using a povidone-iodine solution (Betadine ORL, Asta Medica, Mérignac, France). The nasal secretions were washed out with saline. The nasal cavity and the surrounding facial area were thoroughly disinfected with a chlorhexidine alcoholic solution, and the nose was surgically draped. The maxillary sinus was then punctured under visual control at the inferior meatus with a sterile trocar. Any fluid present in the sinus was suctioned through a sterile catheter. A drain was inserted in all punctured patients [7, 8]. The aspirate was then immediately sent to the laboratory in sterile tubes, and the drain was left in place.

Aspirate processing

Samples were Gram-stained and examined microscopically for bacteria. Aerobes were cultured on enriched chocolate agar, anaerobes on 5% sheep blood agar under anaerobic conditions, and fungi on Sabouraud's dextrose agar. Growth density was determined by a quantitative technique that consisted in serial tenfold dilution of the original specimens. Results were expressed in colony-forming units (CFU) per milliliter. Isolated organisms were identified by standard techniques. Susceptibility to antibiotics was determined by the disk diffusion test.

Negative sinus echography

When sinus echography was negative, a transnasal puncture was not performed. However, a sinus CT was performed the same day, as previously described [13], to confirm the absence of RMS.

Definitions

The absence of RMS on CT was defined as a normal sinus or as the presence of mucosal thickening. The presence of RMS was defined by an air fluid level or a total opacity of the sinus area [1, 4, 7, 8]. The transnasal puncture was considered as positive if a fluid was obtained with sinus aspiration. The transnasal puncture was considered as negative if there was no aspirated material and only air. Criteria for nosocomial infectious sinusitis were as follows: macroscopic purulent sinus aspiration and quantitative culture of transnasal maxillary sinus punctures yielding at least one isolate with a bacterial growth more than or equal to 103 colony-forming units (CFU) per milliliter [7].

Treatment procedures

Endotracheal tubes had low-pressure, high-volume cuffs with an internal diameter of 7.0–8.0 mm. No patient received a stress ulcer prophylactic regimen. Selective digestive decontamination was not used. Care of the patient's mouth was performed three times a day, using an antiseptic solution (Givalex; Norgan, Paris, France). No special care was given to the nares. Treatment of infectious sinusitis consisted of the intravenous administration of antibiotics and sinus lavage every 8 h through the catheter using 5–10 ml of warm sterile saline [4]. Nasotracheal and nasogastric tubes were not removed when sinusitis was diagnosed, because the nasal route for intubation as risk factor for maxillary sinusitis is still controversial [4, 7].

Data analysis

A true positive (TP) result was defined as the presence of a sinusogram in a patient who had a positive transnasal puncture. A true negative (TN) result was defined as the presence of a negative sinus ultrasound in a patient who did not have RMS on CT. A false-positive (FP) result was defined as the presence of a sinusogram in a patient who had negative transnasal puncture. A false-negative (FN) result was defined as the presence of negative ultrasound in a patient who had a positive transnasal puncture. Standard formulas were used to calculate sensitivity (TP/[TP + FN]), specificity (TN/[TN + FP]), positive predictive value (TP/[TP + FP]), and negative predictive value (TN/[TN + FN]). Values are expressed by mean ± standard deviation (SD).

Results

Patients

During the study period, 62 consecutive patients presenting with suspicion of nosocomial maxillary sinusitis were enrolled in the study. Two patients were excluded because of profound coagulation disorders. Finally, 60 patients were included. Thus, 120 paranasal sinuses were examined. The characteristics of these 60 patients appear in Table 1.

Table 1 Characteristics of the 60 patients at inclusion (SAPS simplified acute physiologic score [15])
Full size table

Sinus puncture based on echographic sinusitis evidence

Global results

The time period between ultrasound and transnasal puncture was 2 ± 1 h. The feasibility of sinus ultrasound was 100%, and we did not obtain inadequate or non-interpretable ultrasound findings. The duration of sinus echography was  < 5 min per patient. Of the 120 sinuses examined, we found sinus ultrasound evidence in 84 cases (70%). A sinus puncture was performed in these 84 cases, and 78/84 transnasal punctures were positive. Sensitivity, specificity, positive predictive value and negative predictive value of sinus ultrasound compared with transnasal puncture and CT were respectively: 100%, 86%, 93% and 100%. Only six transnasal punctures were negative, with absence of aspirated material.

Complete sinusogram and partial sinusogram

Among 84 sinusitis ultrasound cases, we obtained 54 complete sinusograms and 30 partial sinusograms. In the case of complete sinusogram, the transnasal punctures were always positive. However, in the case of partial sinusogram, six cases of transnasal punctures were negative (Fig. 2).

Fig. 2
figure 2

Results of the study. 1 Among the 120 examined sinuses 48 (36 + 12) fulfilled the criteria for infectious sinusitis, i.e., 40%

Full size image

Negative sinus ultrasound

An acoustic shadow arising from the front wall was obtained for 36 sinuses. CT scan showed absence of RMS in all these cases. A mucosal thickening observed in CT was found in 15 sinuses, and the sinuses were normal with absence of air fluid level in 21 cases.

Complications of transnasal puncture

We did not observe any complications or failures of the transnasal puncture.

Characteristic of aspiration fluid

The aspiration fluid was macroscopically purulent in the large majority of cases. In other cases, the aspiration was non-purulent. The aspiration amount was between 3 cc and 10 cc.

Infectious maxillary sinusitis and echographic sinusitis evidence

Infectious sinusitis

Among the 78 positive transnasal punctures, 48 fulfilled the criteria for microbiological infectious sinusitis. Isolated microorganisms are shown in Table 2. Thirty did not fulfill the criteria for microbiological sinusitis: 19 had sterile sinus aspiration, and 11 had sinus aspiration with  < 103 CFU/ml.

Table 2 Number of microorganisms isolated from sinus
Full size table

Infectious sinusitis and sinus ultrasound

As shown in Fig. 2, in the case of complete sinusogram, we obtained infectious sinusitis in 67%. In the case of partial sinusogram, infectious sinusitis was observed in 40%. Among the 120 examined sinuses, 48 (36+12) fulfilled the criteria for infectious sinusitis, in other words, 40%.

Sinus ultrasound results, transnasal puncture and patients

Sixty patients were included in the study. We found sinusitis ultrasound evidence in 46 patients, and the transnasal puncture was positive in 42 patients. Of 60 patients, 38 had bilateral sinusograms; eight had unilateral sinusogram, and 14 had bilateral normal signal. Twenty-six of 60 patients (43%) fulfilled the criteria for infectious sinusitis. The 43% rate includes patients with clinical suspicion of maxillary sinusitis.

Infectious sinusitis and nosocomial pneumonia

Of the 26 patients with infectious sinusitis, 15 developed VAP. The diagnosis of co-infection was made within the 24 h preceding the diagnosis of nosocomial sinusitis in one patient and at the time of nosocomial-sinusitis diagnosis in the other 14 patients. In nine patients, the same organism was isolated from the lung and sinus. Ten of these 15 patients were intubated via the nasal route. The microbiological results of patients with both sinusitis and VAP are shown in Table 3.

Table 3 Microbiological results of patients with both sinusitis and nosocomial pneumonia, especially for the nine patients [in brackets] with the same pathogen isolated from sinus and lung
Full size table

Discussion

In intubated patients with suspicion of nosocomial sinusitis, a positive sinus ultrasound was associated with the presence of fluid in the sinus cavity in 93% of cases, allowing a transnasal puncture to be performed safely. On the other hand, when sinus ultrasound was negative, no sinusitis was observed. These results allow us to validate the utility of sinus ultrasound in the diagnosis of nosocomial sinusitis and confirm that the information obtained from B-mode sinus ultrasound analysis can minimize referral of critically ill patients to CT. Thus, transnasal puncture could be performed directly on the results of a sinus ultrasound.

Compelling evidence has demonstrated that the paranasal sinuses are frequent and silent sources of nosocomial infections in mechanically ventilated patients [17, 18, 19, 20, 21]. Classically, CT is the modality of choice for evaluating the sinonasal cavity [9] and for guiding the maxillary sinus puncture [7, 8]. However, this procedure requires the transportation of critically ill patients. This is the most limiting factor. Transport outside of the ICU places the critically ill patient at additional risk [10, 11]. B-mode ultrasound represents a rapid, easily reproducible means of monitoring at the bedside and does not require the transportation of critically ill patients for CT [12, 13]. The princeps study has been published by Lichtenstein et al. [12]. In this study, sinus ultrasound was performed with the patient in supine position. The sensitivity and specificity of ultrasound compared with CT for the diagnosis of RMS were respectively 67% and 87%. However, the authors describe complete sinusogram and incomplete sinusogram and establish a total correlation between complete sinusogram and total opacity of the sinus on CT, i.e., a 100% specificity, which indicates the possibility to safely perform puncture. A recent study by our group fully confirmed the basic importance of ultrasound in this field and showed that ultrasound accuracy for all lesions was improved in the semi-erect patient [13]. The patient's position during sinus ultrasound has a great importance. An important part of radiologic maxillary sinusitis was unidentifiable with the patient in the supine position. This is because small air bubbles visible at the front wall on CT may cause the degradation of the ultrasound and an ultrasound acoustic barrier cannot distinguish air-fluid level from normal sinus. This may appear logical in a protocol performed with the patient in supine position. The half-sitting position may overcome this limitation. The relevance of semi-recumbent positioning was outlined in a study performed with the A-Mode [22]

In the present study, a complete sinusogram always seems to be associated with the presence of fluid in the sinus area, and so sinus drainage is always positive. More interestingly, a complete sinusogram would appear to be associated with an infectious sinusitis in 67% of cases. Certain questions may arise regarding the partial sinusogram. It could be a low air-fluid level or a mucosal thickening. In this study, we obtained six negative transnasal punctures, always in the case of partial sinusogram. However, a partial sinusogram seems to be associated with the presence of fluid in the sinus area in 80% of cases and with an infectious sinusitis in 40% of cases. To our knowledge, the present study is the first to show a relation between moderate air-fluid level and infectious sinusitis.

Transnasal puncture has great importance in the diagnosis and the treatment of sinusitis. Souweine et al. suggest that, in mechanically ventilated patients, even in those on prior antibiotics, quantitative culture of sinus aspirate may be helpful in diagnosing infectious maxillary sinusitis [8]. In numerous cases, causative organisms are susceptible to previously administered antibiotics, and the authors therefore recommend that sinus drainage be included in the procedures for the treatment of nosocomial sinusitis [8]. Vandenbussche et al. suggest performing among all patients, without a prior radiological examination, a puncture of the maxillary sinus in search of nosocomial sinusitis [23]. The systematic use of the puncture without prior examination raises questions. A systematic examination does not appear to be very profitable—in the study by Vandenbussche, 58% of the examinations were useless. In addition to this, there is always the potential risk of a sinus agenesis.

We did not observe failure or complications with transnasal punctures performed on the sinus ultrasound results. Sinus puncture entails several potential complications, albeit infrequent [24, 25]. In the case of sinus agenesis, sinus ultrasound will be negative. The hypoplasia of the maxillary sinus is an entity not to be ignored [26, 27]. With sinus volume being reduced, it is no longer easily accessible [27]. However, to our knowledge, an atrophied maxillary sinus will be unable to give an ultrasound lesion.

Our incidence of infectious sinusitis is high (43%) but close to the incidence previously described after 9 days of mechanical ventilation [1, 3, 7, 28, 29]. In the present study, 15 patients had a concomitant diagnosis of sinusitis and pneumonia the same days. In contrast to the study by Holzapfel et al. [4], our study was not a systematic research of sinusitis by ultrasound. Ultrasound was performed in intubated patients with a suspicion of sinusitis. Some patients with suspicion of sinusitis have fever and pulmonary infiltrates. So, we have performed a bronchoalveolar lavage in these patients to diagnose a VAP. It should be interesting to perform a study on the model of that by Holzapfel et al. [4], to test the value of systematic research of sinusitis by ultrasound in order to individualize promptly the microorganisms potentially responsible of subsequent VAP.

Ultrasound examination can be operator-dependant. For this study, we had planned that all sinus ultrasound would be performed by the same operator. This is a potential limitation of the generalization of the results of this study.

Our microbiologic results do not report anaerobes as causative microorganisms. Several studies outlined the importance of anaerobic infections of the maxillary content [30, 31]. Isolation of such bacteria requires adequate transport conditions and special care in the laboratory, which may not be systematically achieved. Most of the studies reported anaerobes in less than 10% of cases. However, Le Moal et al. showed that, under appropriate transport conditions and bacteriological analysis, anaerobes were frequently isolated in nosocomial sinusitis [30]. These differences may be explained by the careful isolation procedures used in laboratory for anaerobic bacteria determination; each sample was first prepared for anaerobic culture, then aerobic culture.

Conclusion

In clinical practice, the question to ask is: does the mechanically ventilated patient present a nosocomial sinusitis? We can answer using the diagnostic strategy shown in Fig. 3. The most important problem with ultrasound is the partial sinusogram. But the concept of the fluid thickness dynamics and its relevance for indicating immediate postural change facing a partial sinusogram [32] could be interesting to improve the diagnosis of ultrasound sinusitis evidence. Sinus ultrasound could be performed in the first line in mechanically ventilated patients with suspicion of sinusitis. The sinus echography results could be useful in directly indicating a transnasal puncture, without risk, in intubated patients with suspicion of nosocomial sinusitis.

Fig. 3
figure 3

Algorithm to show how to deal with the issue of ultrasound findings. 1 Before a partial sinusogram, a transnasal puncture can be performed, in case of high suspicion of nosocomial sinusitis defined by a purulent nasal discharge and a fever of unknown origin. 2 Another possibility is to propose an immediate postural change. One can imagine that the viscosity of the fluid, if present, will be demonstrated and that genuine fluid will be differentiated from mucosal thickening. Indeed, if the partial sinusogram visualized with the patient in a half-sitting position disappears when the patient is placed in a supine position, the partial sinusogram is probably an air-fluid level (acoustic shadow secondary to small air bubbles). If the partial sinusogram does not disappear when the patient is placed in a supine position, the partial sinusogram is probably a mucosal thickening [32]. A study will be necessary to test this hypothesis

Full size image