Ventilator-Associated Sinusitis: a Review

851

REVIEW ARTICLES

Ventilator-Associated Sinusitis: A Review

Viveka Westergren, Lars Lundblad, Henrik B. Hellquist, From the Department of Ear and Skin, Unit of Otorhinolaryngology, and Urban Forsum Head and Neck Surgery, Karolinska Institute, Stockholm; the Department of Health and Environment, Division of Clinical Microbiology, Linko¨ping University, Linko¨ping, Sweden; and the Department of Pathology, The Gade Institute, Haukeland University Hospital, Bergen, Norway Downloaded from https://academic.oup.com/cid/article/27/4/851/428661 by guest on 25 September 2021 A common foreign body of the nose in intensive care, the nasotracheal tube, has for 20 years been cited as a cause of bacterial infection of the paranasal sinus. High frequencies of bacterial culture positivity have occurred in several studies. However, the state of critically ill patients has to be evaluated before conclusions about cause of infection can be made. Nosocomial colonization with intensive care unit flora, in combination with use of antibiotics, precludes the use of procedures that are standard in office practice and microbiological diagnostics. New methods of sampling and quantitative culturing for the specific purpose of intensive care antral diagnostics, in combination with endoscopic inspection, have enlarged our knowledge of sinusitis. Among patients ventilator- treated for §1 week, the occurrence of bacterial sinusitis is õ10%. For 80% of the examined antra there were similar inflammatory reactions without clinical signs of infection. Sporadically in these, cultures of antral specimens were positive for bacteria, which, by definition, would represent colonization.

Critically ill patients have one or more diseases or injuries In 1974 Arens et al. presented a retrospective ICU study in requiring intensive care. These patients often have dysfunction which four patients had sinusitis ipsilateral to a nasotracheal of one or more organ systems, with secondary impairment of tube [7]. Cases with or without growth of bacteria in cultures of host defenses [1, 2]. Critically ill patients are subjected to purulent nasal discharge, antral drainage material, or specimens multiple invasive diagnostic and therapeutic procedures. Any from an unreported site were interpreted as maxillary sinusitis. fever developing during hospital care is a nosocomial fever More case reports followed during the next decade (table 1). [3]. Such fever is most common in the intensive care unit Several retrospective studies (table 2) showed a higher inci- (ICU) setting, whether it is infectious or noninfectious [4]. It dence of sinusitis in the adjacent vs. nonadjacent antrum of is important to differentiate nosocomial infection, a clinical nasotracheal tubes [15–17, 20–22]. Head injuries and facial infectious disease, from nosocomial colonization, in which fractures also contribute to a high incidence (table 2). pathogenic microorganisms are present but do not participate Studies involving no microbiological investigations but indi- in a disease process [3]. rect examinations yielding diagnoses of ‘‘radiographic sinus- Hospitalized persons are often colonized with enteric gram- itis’’ have also been undertaken (table 3). Studies of ICU pa- negative bacilli (EGNB) [1]. The more severe the illness, when tients with CT or sinus roentgenography showed that a patients are critically ill and mechanically ventilated in an ICU, prolonged period of nasal cannulation can lead to increased the more regularly EGNB colonize the lower airways as well incidence of sinusitis [25, 26, 29–31]. The larger nasal cannu- [5]. The particular bacterial strains in one patient are usually las, i.e., nasotracheal tubes, induce radiographically evident found in both the upper and lower airways [5, 6]. sinus pathology faster and more often than do the smaller nasogastric tubes [25, 26, 29], although radiographic sinusitis does occur with nasogastric tubes [28], even in patients who Received 1 December 1997; revised 14 May 1998. are tracheotomized [32]. Grant support: Swedish Society of OtoRhinoLaryngology, Head and Neck In the late 1980s the ﬁrst results of prospective studies con- Surgery, including the Fund of Rhinologic Research; research funds of Linko¨- ping University Hospital; research and development fund of County of So¨rm- centrating on the diagnosis of bacterial sinusitis in ICU patients land, Sweden; The Nachmanson’s fund, Svenska La¨karesa¨llskapet, Swedish were published (table 4). Investigations of the incidence of Society of Medical Research (no. 960301); Mia och Erik Myrin’s Donation bacterial sinusitis comparing nasotracheally and orotracheally Fund; and the research funds of the Karolinska Institute. Reprints or correspondence: Dr. Viveka Westergren, Department of Otorhi- intubated patients have shown that patients with fewer devices nolaryngology, Head and Neck Surgery, Karolinska Hospital, SE-171 76 Stock- in the nose have the advantage [35, 37–40]. holm, Sweden. Bacteria isolated from aspirates in these studies on ventila- Clinical Infectious Diseases 1998;27:851–64 tor-associated sinusitis (VAS) are similar to those nosocomially ᭧ 1998 by the Infectious Diseases Society of America. All rights reserved. 1058–4838/98/2704–0032$03.00 colonizing the airways in general. An evaluation of the fre-

/ 9c57$$oc04 09-28-98 06:32:23 cidal UC: CID 852 Westergren et al. CID 1998;27 (October)

Table 1. Data from case reports on ventilator-associated sinusitis.

Means of diagnosis

[Reference]/year No. of Culture of of report cases Indirect indicated sample Comments

[7]/1974 4 Radiography PND or aspirate These 4 were among Ç200 ICU patients studied [8]/1976 1 Radiography PND and aspirate ... [9]/1981 2 Radiography PND (n Å 1) ... (n Å 1) [10]/1982 3 Radiography Aspirate (n Å 1) Healed in 48 h when cannulas positioned orally Downloaded from https://academic.oup.com/cid/article/27/4/851/428661 by guest on 25 September 2021 [11]/1984 7 CT Drainage Septic patients had been nasotracheally intubated for ú5d [12]/1985 2 Radiography ...... [13]/1987 3 CT Aspirate Primary pneumococcal pneumonia in all 3 cases [14]/1988 1 CT Aspirate Candidal sinusitis in combination with bacterial sepsis

NOTE. ICU Å intensive care unit; PND Å purulent nasal discharge. quency of the different bacteria observed in numerous studies 48] of the bacterial flora of antra in people without known is shown in table 5. The results indicate that Staphylococcus paranasal sinus disease yielded contradictory results. aureus is most frequent, followed by Pseudomonas aeruginosa, Direct examination by antroscopy is the ‘‘gold standard’’ for EGNB, and various Streptococcus species. assessing antral disease [49]. Provided that an antral specimen Ventilator-associated pneumonia (VAP) is associated with bacterial culture is positive and without contaminants (and has high rates of morbidity and mortality among ICU patients [44]. been done with a technique that enables contaminants to be The high risk of VAP is compounded by many factors. First, distinguished from the indigenous flora [50]), bacterial maxil- there is immunocompromise due to the underlying disease lary sinusitis may well be the correct diagnosis. The bacteria and/or its therapy. Nosocomial colonization occurs with pro- present in the nasal cavity when antral culture specimens are longed mechanical ventilation, e.g., when the local antibacterial obtained are not necessarily indicative of the cause of the infec- host defense is impaired and the bacterial inoculation is en- tious sinusitis [51, 52]. hanced by the artificial airway [2, 45]. The coronal CT view, after endoscopy, is currently consid- The signs of VAP—new or growing infiltrates on a chest ered the most reliable means of evaluating inflammatory sinus radiograph, fever, leukocytosis, and purulent tracheal secre- disease [53, 54], as it makes it possible to picture the skeletal tions—are not enough to differentiate it from other conditions anatomy and distinguish mucosal thickening from fluid. As a with similar signs, such as atelectasis, congestive heart failure, bedside alternative there is plain sinus radiography for screen- or adult respiratory distress syndrome [41]. Once more, ing; it has been used as such for mechanically ventilated pa- S. aureus and P. aeruginosa are the bacteria most commonly tients in some reported cases (table 1) and studies (tables 2– causing nosocomial pneumonia [3]. In addition, they occur 4). The other portable alternative is ultrasonography, which frequently as nosocomial colonizers. has been considered to be better than radiography when both The occurrence of the same bacteria in antral and tracheal were compared to antral puncture as the standard [55, 56]. aspirates from 13 of 17 patients with maxillary sinusitis sup- Antral puncture, either transnasally by the inferior meatus ports the notion of general colonization of the airways [37]. or through the canine fossa, with or without lavage [57, 58], Cases of maxillary sinusitis and VAP, diagnosed with use of is a well-known diagnostic and therapeutic approach. It is bronchoscopic protected-brush specimens, have been observed also commonly used as a standard procedure with other indi- in which the bacterial agents were the same [35]. This finding rect diagnostic methods in evaluations either to confirm the was not reproduced, though, when the investigation was per- presence of antral fluid or to obtain secretions for further formed in the opposite order (i.e., VAP was diagnosed first cellular and microbiological tests. However, no systematic and then maxillary sinus aspirates were investigated) [41]. data on the diagnostic limits of antral puncture have been published. Diagnostics of Paranasal Sinus Disease A considerable risk of contamination of bacterial cultures There are problems with the diagnostics of paranasal sinus has been observed with the use of lavage [51]. The difficulty disease [46]. For example, two quite similar investigations [47, in obtaining clean specimens from the nasal cavity by disin-

/ 9c57$$oc04 09-28-98 06:32:23 cidal UC: CID CID 1998;27 (October) Ventilator-Associated Sinusitis 853

Table 2. Data from retrospective studies of ventilator-associated sinusitis.

Means of diagnosis

Culture of Cultures Prior [Reference]/year No. of cases/ Period or category Adjacent medical indicated positive antibiotic of report patients of patients studied device(s) Indirect sample (%) use Comment(s)

[15]/1982 32/2,368 Admitted to trauma 6, NTT; 26, Radiography Aspirate (25) 100 Yes Facial and cranial unit NGT; 2, nasal fracture in 11; packing 26 had corticosteroid therapy [16]/1985 11/43 Mechanical NTT Radiography Drainage (10) 100 Yes Selective control in Downloaded from https://academic.oup.com/cid/article/27/4/851/428661 by guest on 25 September 2021 ventilation for 43 with NTT ú72 h [17]/1985 6 1-y period at ICU 4, NTT; 5, NGT 4, CT Aspirate (4) 80 Yes Corticosteroid therapy, 5; diabetes mellitus, 1; head trauma, 4 [18]/1987 19/111 Head-injured 16, NTT; 3, OTT; CT Aspirate 100 Yes Some had facial all had NGT fractures [19]/1987 24/208 Head-injured 19, NTT; 5, OTT CT Aspirate 100* Yes Concurrent infection in 21 [20]/1988 4/171 With sepsis NTT Radiography Aspirate 100 Yes Diabetes mellitus and renal insufﬁciency with dialysis, 2 [21]/1988 11/198 Admitted to trauma 7, NTT; 9, NGT; Radiography Aspirate (10) 100 Yes (10) Bilateral maxillary unit 5, both or CT sinusitis, 7 [22]/1988 19 Nasotracheal tube NTT Radiography Aspirate 78 Yes (16) Lesions of nasal and fever for õ6 mucosa in 10 mo after removal of NTT [23]/1995 271 From No information . . . Aspirate 100† ? Bacterial culture microbiological results were records retrospective

NOTE. ICU Å intensive care unit; NGT Å nasogastric tube; NTT Å nasotracheal tube; OTT Å orotracheal tube. * Culture-positivity was a criterion for diagnosis. † Culture-positivity was a criterion for inclusion in the study.

fecting was certified by Rouby et al. [42], for whom only 50% adjust the antibiotic therapy to match the susceptibility of the of septum swab samples were free of bacteria. These results bacteria in the antral samples. present two further considerations: (1) the anatomy of the infe- There has been no thorough study in which all contributory rior meatus is more complicated than that of the plain septal parameters have been included and the results of therapy have area, and (2) the septum-sampling swab stays on the surface been followed. Such a study would be difficult and perhaps even and thus cannot be compared to a penetrating needle. impossible to perform. Borman et al. have contributed one of the most interesting studies to date from this perspective [38]. Their study included 26 patients with occult fever, transnasal Treatment of VAS cannulas, and ICU stays of ú48 hours, of whom 19 had abnor- Removing a foreign body from a nasal cavity improves the malities on sinus CT. All 19 underwent unilateral antral puncture, local environment of that cavity, as determined in most studies and fluid was aspirated and cultured from 15. Ten of 15 cultures of ventilator-associated sinusitis. As a direct agent of obstruc- showed bacterial growth, yet the investigators decided that the tion disappears, the continuous traumatization of the mucosa occult fever or other infectious complications could confidently ceases, and with the removal of a polymer tube, a matrix and be attributed to sinusitis in only one case. shelter for P. aeruginosa and S. aureus disappears [59–62]. The removal of the foreign body diminishes the number of New Clinical and Microbiological Knowledge About VAS bacteria colonizing the nasal cavity, allows more air to enter, Given the incomplete knowledge on the etiology and diag- and improves the local host defenses. Another measure is to nostics of VAS, there are good reasons to investigate the in-

/ 9c57$$oc04 09-28-98 06:32:23 cidal UC: CID 854 Westergren et al. CID 1998;27 (October)

Table 3. Investigations involving indirect diagnostics for ventilator-associated sinusitis.

[Reference]/year No. of cases/patients Adjacent medical of report (other data) Diagnostic means Selection criteria device(s)

[24]/1988 12/12 (In 7 cases, CT Neurosurgery patients with no NTT initially involving preexisting sinusitis antrum ipsilateral to tube affected) [25]/1988 9/20 (Risk factors: Radiography Followed prospectively from intubation See previous column prolonged time, (8 oral, 12 nasal) P õ .001; NTT, P õ .02) [26]/1989 14/16 CT Admitted to ICU; no preexisting NTT Downloaded from https://academic.oup.com/cid/article/27/4/851/428661 by guest on 25 September 2021 sinusitis [27]/1990 31% of patients with A-mode ultrasonography Followed prospectively from intubation Various any transnasal tube [28]/1991 10% of tracheotomized, Radiography after NGT for ú48 h NGT 50% of extubation nontracheotomized patients [29]/1991 23/47 Radiography Mechanical ventilation for ú5 d NTT, NGT [30]/1996 44% of those with CT A-mode Comatose patients requiring CT for No information and 41% of those ultrasonography, CT diagnosis with ultrasonographic abnormalities/50

NOTE. ICU Å intensive care unit; NGT Å nasogastric tube; NTT Å nasotracheal tube.

flammatory reactions in the paranasal sinuses (primarily in the Microbiological Analysis maxillary sinuses) that commonly occur in mechanically venti- We developed a technique for antroscopy and for taking lated ICU patients, as well as their relationship to the systemic samples for bacterial cultures by entering an antrum through inflammatory response syndrome. the canine fossa [65]. We tried different ways of collecting We performed a series of studies on material and samples secretions from the antra for bacterial culture. Using lavage collected in consecutive bilateral sinoscopies of 33 ICU patients bilaterally in patient 1, we noted that both the right and left (23 men and 10 women, aged 13–87 years [mean age, 40 samples showed contamination; there were identical bacteria years]). The patients included were critically ill and had been in both gingival and antral samples, but fewer in the latter mechanically ventilated for ú7 days. They had pyrexia of (dilution effect). From patient 2, secretions were collected by unknown cause in combination with leukocytosis and/or an absorption (for 30 seconds) on a cotton swab in a protective elevated level of C-reactive protein, clinically suspected infec- tube, immediately after penetration of the antral wall, when tious sinusitis (purulent nasal discharge), and maxillary sinus the trocar was removed from the sheath. For patients 3 through disease evidenced radiologically or ultrasonographically [63]. 7, the trocar penetration was through the gingiva of the canine The specific aims were (1) to test ultrasonography as an fossa, and the contamination rate was 67% (95% CI, 35%– alternative diagnostic tool for ICU patients; (2) to validate 90%). Starting with patient 8, a free bone area was made antroscopic observations of one examiner against those of an and contact with oral mucosa was avoided, but otherwise the expert panel [64] and validate the use of antroscopy as a ‘‘gold performance was unchanged. With this arrangement the con- standard’’ [63]; (3) to test a reliable sampling technique for tamination rate decreased to 5% (95% CI, 0–23%), when eval- bacterial culturing from the antra [65] and, if contaminations uated after its use on 11 patients. No further changes to the were found, to improve the methodology until reliable diagnos- sampling techniques were made. tic culture results were obtained [65, 66]; (4) to search for We also set up a method to double-check bacterial cultures infective agents of the antral mucosa by means of anaerobe for possible contaminants [65, 66]. A gingival culture was made cultures of tissue samples [66]; (5) to determine the concentra- for each patient to identify the contaminants. All samples were tion of antibiotic in the sinus mucosa and study its relationship cultured aerobically on blood and CLED (cystein lactose electro- to occurrence of bacterial infections and colonization [67]; and lyte deficient) agars at 37ЊC and on hematin (McLeod) agar at

(6) to test the hypothesis of an existing correlation between 37ЊCin5%CO2. They were also cultured anaerobically on infectious or noninfectious sinusitis to clinical appearance, his- prereduced Brucella blood agar, with and without added antibiot- tology, and occurrence of the mRNA of the chemokine ics (penicillin G, clindamycin, metronidazole, and imipenem). RANTES, the cytokines IL-6 and IL-12, and the endothelial Starting with patient 8, quantitation was done by means of adhesion molecules ICAM-1, VCAM-1, and E- and P-selectin. a dilution series of both samples from gingiva and antra to

/ 9c57$$oc04 09-28-98 06:32:23 cidal UC: CID CID 1998;27 (October) Ventilator-Associated Sinusitis 855

Table 4. Data from prospective investigations of sinusitis in mechanically ventilated patients.

Diagnostic means

Culture of Cultures [Reference]/year indicated positive of report Inclusion criteria No. of cases/patients Note(s) Indirect sample (%) Comment(s)

[33]/1986 NTT 27 . . . Radiography or PND (10), 100 Sinusitis associated with CT aspirate (2) NTT [31]/1987 NTT for ú8 d 12/22 Transnasal culture CT Aspirate 100 Aim to study sphenoiditis sampling [34]/1989 NTT for ú8 d 49/53 Transnasal culture CT Aspirate 83 ... sampling [32]/1990 MV for ú5 d 17/20 NTT, 10 patients; CT Aspirate 71 Bacterial sinusitis in 60% tracheotomy, with NTT and 50% Downloaded from https://academic.oup.com/cid/article/27/4/851/428661 by guest on 25 September 2021 10 with tracheotomy [35]/1990 MV for ú48 h 26/111 Random OTT Radiography or Drainage (14) 79 Sinusitis in 1.8% with (53) or NTT CT OTT and 43.1% with (58); NGT in NTT all [36]/1991 Burn patients with NTT 8/22 Clinical condition CT Aspirate (5) 80 Antral puncture and determined culture performed time for CT when no clinical improvement within 48 h [37]/1992 No sinus disease on 17/68 Random OTT Radiography Aspirate (40) 43 Cases defined by culture- admission (32) or NTT positivity. Bacterial (36) sinusitis in 6% with OTT and 42% with NTT [38]/1992 MV for ú48 h, FUO, 1/19 Investigation of CT Aspirate 67 Case defined as febrile transnasal cannula fever cause sinusitis; in 9 patients with positive cultures, concurrent infection or other disease was assessed to be cause of fever [39]/1992 Prolonged intubation, MV 34/44 Random OTT Ultrasonography Aspirate 41 Daily A-mode (20) or NTT ultrasonography (14); NGT in all [40]/1993 Admitted to ICU, need for 54/300 Random OTT CT every7dor Aspirate 91 Gastric tube same route MV for ú7 d (151) or NTT as needed as endotracheal; (149) radiologic sinusitis in 78; statistically nonsignificant differences [41]/1994 Clinical ventilator-associated 12/50 Investigation CT Aspirate 71 OTT and NTT in all; pneumonia according to radiologic sinusitis in systematic 33 diagnostic protocol [42]/1994 Admitted to ICU, need for 44/162 Initial CT for CT Aspirate 33 Many parameters, MV for ú7 d 75% abnormal; difficult to analyze random NGT- OTT and NGT- OGT [43]/1996 Referred culture samples from 103 Only positive . . . Aspirate 100* . . . ICU for suspected sinusitis cultures included

NOTE. FUO Å fever of unknown origin; ICU Å intensive care unit; MV Å mechanical ventilation; NGT Å nasogastric tube(s); NTT Å nasotracheal tube(s); OGT Å orogastric tube(s); OTT Å orotracheal tube(s); PND Å purulent nasal discharge. * Culture-positivity was an inclusion criterion. improve the identiﬁcation of contaminants. The procedure for 1001,1002, and 1003 was made. Otherwise, the cultures were the quantitative culture was as follows. The cotton swab was set as previously described from the dilutions. vortexed with 1 mL of PBS buffer, which was then regarded Of the 52 antral cultures of 26 patients (numbers 8–33), 10 as the ﬁrst undiluted sample, from which a dilution series of (19%) were contaminated. Four antral secretion cultures, out

/ 9c57$$oc04 09-28-98 06:32:23 cidal UC: CID 856 Westergren et al. CID 1998;27 (October)

Table 5. Microbiological etiologies of ventilator-associated sinusitis, as determined in previous studies in which maxillary sinus aspirates were used.

Gram-positive cocci isolated

[Reference]/ No. of Current Cultures Coagulase- year of Sampling Disinfection positive antibiotic polymicrobial neg Streptococcus Streptococcus report route performed cultures use (%) S. aureus staphylococci Streptococcaceae Enterococcus pyogenes pneumoniae Other

[15]/1982 TM Y 25 Y 56 1 0 4 0 0 0 0 [17]/1983 ? N 4 Y 100 2 0 1 2 0 0 1 [18]/1987 TM Y 19 Y 89 6 0 14 0 0 0 2 [31]/1987 TM N 11 Y 64 0 0 1 0 6 0 0 [19]/1987 TM Y 24 Y 88 8 0 8 0 7 0 4

[20]/1988 ? N 4 Y 50 1 0 0 0 0 0 0 Downloaded from https://academic.oup.com/cid/article/27/4/851/428661 by guest on 25 September 2021 [21]/1988 TM N 11 Y* 46 3 0 0 1 0 0 2 [22]/1988 TM Y 21 Y* 42 5 0 1 0 1 2 0 [34]/1989 TM N 39 Y* 54 5 0 4 0 18 2 0 [32]/1990 ? N 10 ? 30 6 0 1 0 6 0 0 [36]/1991 ? N 4 Y* 50 1 0 1 0 0 1 1 [39]/1992 ? N 9 Y ? 1 0 0 2 0 0 0 [40]/1993 TM N 54 ? 50 10 0 23 0 0 0 5 [42]/1994 TM Y 44 Y* 32 18 0 13 0 0 0 0 [23]/1995 ? N 103 ? 65 35 8 8 5 0 4 0 [43]/1996 TM Y 271 ? 58 94 14 23 14 0 0 0

Total† 196 22 102 24 38 9 15

NOTE. TM Å transmeatal; ? Å not described in the report. * The patients were, with some exceptions, treated with antibiotics at the time of puncture. † Gram-positive cocci, 406 (32%); gram-negative bacilli, 686 (54%); anaerobes, 53 (4%); yeasts, 114 (9%).

of 66 (6%), were positive when contaminants were excluded concerning the reliability of the observations [64]. This was (table 6) [66]. Two contained a mixed anaerobic flora with done particularly because we were using these observations Fusobacterium and Prevotella species as the main pathogens, as a standard criterion in other comparisons. The study used one was a monoculture of Escherichia coli, and one had a videotaped material from 18 of the sinoscopies (on 9 pa- limited number of Lactobacillus colonies. tients), and an expert panel of 4 senior consultant ear-nose- The active site of infection is in the mucosa [68]. Drainage throat surgeons with particular experience with the rhinosi- might not contain viable bacteria or might miss the true nuses (both clinically and in research) made assessments on infecting organism active in the tissue margin. Therefore, a structured form. tissue specimens from antra were cultured anaerobically, The investigation included an interobserver study of the vari- starting with patient 15. The specimens were transported in ance within the expert panel, a study of background impact thioglycolate broth [69] to an anaerobic work-box, where (bias), and the value of using video to gain a second opinion. they were cut into small pieces, pounded with a pestle in a The findings were presented in structured form: redness, bleed- mortar, and then spread onto prereduced Brucella blood agar ing tendency, edema, secretions, and mucosal alterations. The plates, with and without antibiotics. results were evaluated according to k statistic methodology Of the 38 mucosa cultures, four (11%) were positive (table [70–72]. 6) [66]. One positive specimen was from the same antrum In the expert panel, the concordance was on a random yielding an anaerobic culture–positive secretion, and the bacte- level for redness and there was a fair-strength agreement ria found in the both samples correlated well to each other. for bleeding tendency and edema; for assessment of secre- The secretion cultures corresponding to the other three positive tions and mucosal alterations the agreement was better and mucosa cultures were negative. Two were monocultures, yield- reached moderate strength. The concordances on the con- ing Peptostreptococcus species in one and Lactobacillus fer- cepts of redness, bleeding tendency, edema, secretions, mu- mentum in the other. In the third, Eubacterium and Propioni- cosal alterations, and infections between the expert panel bacterium species were found. and the performer corresponded to the interobserver results. On the free-description part of the form, the experts expressed their overall assessment of each videotaped antros- Inflammatory Responses copy, and for the assessment of ‘‘infection’’ there was almost perfect concordance (median k Å 0.78 in the In the evaluation of endoscopic assessments for our com- interexpert study and median k Å 1.00 in the performer- plete patient population, a special investigation was made experts study).

/ 9c57$$oc04 09-28-98 06:32:23 cidal UC: CID CID 1998;27 (October) Ventilator-Associated Sinusitis 857

Table 5. (Continued)

Gram-negative bacilli isolated Anaerobes Yeasts Total P. Acinetobacter Klebsiella Enterobacter Escherichia Proteus Serratia Haemophilus Bacteroides Anaerobic no. of aeruginosa species species species coli species species inﬂuenzae Other species cocci Other Candida Other isolates

12 0 7 7 3 5 0 0 0 2 0 0 0 0 41 2001002011030117 7230740053210056 0121330010001019 8 3 4 0 7 0 0 0 10 3 2 1 0 0 65

410000000000006 Downloaded from https://academic.oup.com/cid/article/27/4/851/428661 by guest on 25 September 2021 1010410010001013 3023010136000028 00354150110003061 0100010000000015 100100010000006 4000000000030010 1 2 3 3 17 6 2 7 4 5 0 0 4 0 92 11 6 0 2 6 4 4 0 6 0 0 2 14 1 88 26 20 20 8 13 11 3 9 5 2 0 17 27 0 219 64 91 35 16 34 32 0 16 30 0 0 0 62 0 525

144 127 75 47 98 84 9 35 67 22 4 27 112 2 1,261

Microbial Colonization Antibiotic Availability in Host-Microorganism Interactions

In evaluation of the antroscopic findings and bacterial infec- Extracellular antibiotic efficacy in the bilateral antral muco- tions or colonizations for the entire patient population, the sal specimens from seven patients (numbers 27–33) was deter- regular endoscopic observation was an inflammatory state (of mined and compared to serum concentrations on the same occa- various degrees) in 56 of the 66 antroscopies (85%) [66]. Ten sion [67]. The microtechnique used has been previously maxillary sinuses (15%) were found to be normal. In 38 antros- described [73]. Adequate concentrations of active antibiotic copies (58%), antral secretions of various types were found. were found in the antral mucosa, in spite of inflammatory The mucosal inflammatory reactions were expressed in differ- disease [67]. An exception was patient 28, whose serum con- ent ways, in terms of the color of edema (from pallid to vermil- centration of cefuroxime was very low (0.14 mg/L); neither ion) and degree of transparency. In some the edema was partial, the right or left antral tissue specimen had a measurable concen- and then always in the medial and ostial regions. When polyps tration. occurred, they were also primarily located medially and os- In the remaining 12 antral specimens there were measurable tially. concentrations of cefuroxime, vancomycin, or ampicillin, de- Four antra were identified endoscopically as manifesting in- pending on the choice of therapy. The mucosal concentrations fection. For two of these the secretion cultures were positive, were all lower (by 10%–95%) than the serum concentrations. for one both secretion and mucosa cultures were positive, and The difference in antibiotic concentration between right and for the fourth the mucosa culture was positive. Two of the left antrum samples from the same patient may have been due bacterial cultures for the four sites of infectious sinusitis had to the 20 minutes between sampling times and/or different a mixed anaerobic flora, one had only two anaerobic bacteria ratios of surface area to volume associated with different ex- (Eubacterium and Peptostreptococcus species), and one had a pressions of disease. single aerobic pathogen (E. coli) (table 6). Three other antral specimens yielded positive monocultures (one secretion culture and two mucosal cultures) (table 6). The Pathology of Sinusitis antroscopic findings in these three antra were the same as for all the other patients who had reactive inflammation in the antra In patients 27–33, investigations for the presence in bilateral and secretions but negative cultures. No infectious adverse antral tissue specimens of the chemokine RANTES, IL-6 and reactions were recognized. Thus, these three antra were consid- IL-12 (by reverse transcriptase PCR), and the adhesion mole- ered to be in a reactive inflammatory state with bacterial coloni- cules ICAM-1, VCAM-1, and E- and P-selectin (by immunocy- zation. tochemistry) were performed, and the results were compared

/ 9c57$$oc04 09-28-98 06:32:23 cidal UC: CID 858 Westergren et al. CID 1998;27 (October)

Table 6. Data regarding seven positive cultures of antral secretions and/or mucosal specimens from 33 patients in a consecutive series of bilateral antroscopy and sampling.

Culture ﬁndings

Patient Nasal Antral imaging Antroscopic Secretions Mucosal specimens no. Side cavity ﬁndings ﬁndings (n Å 66) (n Å 38) Clinical diagnosis Antibiotic therapy

2 L NTT ND Pus, normal Lactobacillus species ND Pus, unclear Cefuroxime mucosa etiology 6 R NGT ND Foul-smelling pus, Escherichia coli ND Infectious disease None (cefuroxime pallid edema suspended for ú48 h) 19 L NTT CT 3 d before, Mucous fluid, Negative Propionebacterium Reactive Cefuroxime bilateral air- pallid edema species inflammatory Downloaded from https://academic.oup.com/cid/article/27/4/851/428661 by guest on 25 September 2021 fluid level disease 28 L NGT MRI 7 d before, Foul-smelling pus, Streptococcus group F, b Streptococcus group Infectious disease Cefuroxime opacity in L vermilion Streptococcus group G, F, Prevotella antrum edema, Fusobacterium species, melaninogenica, fibrinogenous Prevotella Prevotella species film laterally, melaninogenica, polyps medially Prevotella species 30 R NGT MRI 25 d before, Foul-smelling pus, Enterococcus species, Negative Infectious disease Ampicillin, normal antra; vermilion Fusobacterium species, vancomycin MRI 6 d edema Eubacterium lentum, before, opacity Prevotella in R antrum (L melaninogenica, normal) Prevotella distasonis, Prevotella species 30 L NTT MRI 25 d before, Little mucous Negative Eubacterium species, Infectious disease Ampicillin, normal antra; fluid, vermilion Peptostreptococcus vancomycin MRI 6 d edema species before, opacity medially, in R antrum (L normal mucosa normal) laterally 33 L None (NTT ND Yellow serous Negative Lactobacillus Sinus cyst and Vancomycin removed fluid, some fermentum reactive 11 d blood clots, big inflammatory before) medial cyst, disease pallid edema

NOTE. L Å left; ND Å not done; NGT Å nasogastric tube; NTT Å nasotracheal tube; R Å right. Table is adapted from [66]. with histopathologic and bacterial culture findings [74]. The neutrophils. In two of these (28, left; 30, right), there were also aim was to evaluate the inflammatory expression in infectious abundant eosinophils. In the third (30, left), however, involving vs. noninfectious sinus disease. a peptostreptococcus infection, there was an accumulation of Biopsy specimens of bilateral antral mucosa that were nega- inflammatory infiltrative leukocytes, with the major part of the tive in aerobic and anaerobic bacterial cultures and were from eosinophils within the vessels. four patients who had undergone surgery for exophthalmos In the reference material from the clinically normal antral were used as reference for the occurrence of RANTES, IL-6, mucosa of the four patients surgically treated for exophthalmos, and IL-12. All four antral specimens that were positive for whose tissue specimens were obtained bilaterally, PCR was RANTES also yielded positive, noncontaminated bacterial cul- negative for IL-6. Positive findings were made in two cases tures (patient 28, left antrum; 30, right and left; and 33, left). each for IL-12 and RANTES, all in separate specimens. These The correlation between the findings of no RANTES and no findings were not significantly different from the findings for infective agents was significant (P Å .005) [74]. Both IL-6- the critically ill patients. P-selectin was prominently expressed positive antra were also IL-12-positive, and one of them was in all specimens, with 100% positive vessels compared to the also RANTES-positive. The noninfectious RANTES finding amount of CD31-positive vessels used as control in the corre- was for a specimen that was also IL-12-positive. The distribu- sponding section. ICAM-1 and E-selectin regularly showed tion of cytokines and with regard to other parameters, e.g., strong expression, but VCAM-1-positive vessels were sparse polyps and histopathologic expressions, failed to reach statisti- in all specimens. cal significance. Ultrasonography (A-mode) was tried as an alternative diag- However, in three RANTES-positive antra with anaerobic nostic tool [63], with the criteria of Jannert et al. used for infectious disease (patient 28, left antrum; 30, right and left) assessment [56]. There were no signs of discomfort during there was pronounced histologic inflammation with numerous ultrasonography of the 15 patients examined. The accuracy of

/ 9c57$$oc04 09-28-98 06:32:23 cidal UC: CID CID 1998;27 (October) Ventilator-Associated Sinusitis 859 antral ultrasonography vs. antroscopy, as the ‘‘gold standard,’’ independent assessment of video-recorded antroscopies from was estimated for fluid and edema in 0Њ and 30Њ body positions. the clinical study, with the use of a structured form [79] and There was an improvement in sensitivity to fluid, from 73% free description. With use of k statistics, a double evaluation at 0Њ examination to 91% at 30Њ examination. The specificity was performed by means of an interobserver variability study for fluid, 25%, remained unchanged. For edema the sensitivity among the experts and a study of the variability of the perfor- was unchanged (81%) with repositioning, but the specificity mer in comparison with the experts [64]. increased from 67% at 0Њ to 100% at 30Њ. The use of maxillary The less fractional the assessment of the antral condition, sinus ultrasonography in the ICU setting is primarily for screen- the higher the strength of agreement, as was illustrated by the ing for any disease. almost perfect concordance on assessment of clinical infectious sinusitis. The results, interpreted as strength of agreement, were then compared between the two studies. In this comparison no Discussion significant differences were found between the interobserver Downloaded from https://academic.oup.com/cid/article/27/4/851/428661 by guest on 25 September 2021 To improve knowledge of pathological changes occurring study of the expert panel and the study comparing the experts in the upper and lower airways of critically ill mechanically and the performer. These results were interpreted as showing ventilated patients, the requirements for examination and diag- (1) that the original observations at the antroscopies had relia- nosis must be enhanced and documented in such a way that bility as good as those of the expert panels and (2) that the the results obtained may be used for scientific evaluations. quality of observations at the antroscopies were dependable Experimental animal studies with this objective are difficult enough to be used for evaluation in the full clinical study. to set up properly to represent the corresponding conditions. The difficulty is obvious in obtaining antral samples for Furthermore, animal anatomy differs from that of humans, as bacterial cultivation without contaminating the sample in either does the bacterial flora of their upper airways. The material on of the two passage routes used. The occurrence of contaminants which our research project was based was therefore obtained was estimated in our study [65]. For an isolated bacterial spe- from a consecutive study of patients mechanically ventilated cies to be considered as representative of the agent of maxillary sinusitis, we found that it has to occur in significantly higher for §7 days in an ICU. numbers than in a corresponding control culture from either The paranasal sinuses are not accessible to direct examina- meatus nasi inferior or gingiva depending on the route of pene- tion without an invasive procedure. Indirect diagnostic methods tration used. The canine fossa route was chosen because of the are valuable for guidance in the investigation of nosocomial better operating view and better chance of avoiding unneces- fever with symptoms that could arise from a maxillary sinus sary contacts with the contaminating gingival mucosa. infection. Ultrasonography also is easy to use in comparison Disinfection has been shown not to solve the contamination with a criterion standard, e.g., antral puncture or CT, in me- problem when applied on an easily accessible, plain area of chanically ventilated patients [30, 39, 63]. skin [80] and in the more difficult nasal cavity [42]. We applied Our study, in comparing ultrasonography to direct, endo- the protected-brush-specimen model, that is, a double lumen scopic examination in an ICU setting, is the first of its kind to technique reducing contamination [81], and made a protective be published. Previously, though, comparison of ultrasonogra- tube to use at antroscopies. The sampling by cotton-swab ab- phy to antroscopy has been performed in an office setting [55, sorption was selected (1) to get quantitatively comparative sam- 56]. We found that the concordance of ultrasonography with ples from the gingiva and the antrum and (2) to obtain a sample endoscopy improved when the patient was raised. The semire- from an antrum when there were no visible secretions (pus, cumbent position of 30Њ was chosen, however, in keeping with mucus, or serous fluid). routines of the neurosurgical ICU. Thus, even better results The comparisons to gingival cultures showed that there was might have been possible if the patients’ heads had been raised still too high a rate of contamination in samples. This contami- further. nation rate was successfully reduced when a free bone area The diagnostic ability to recognize any antral abnormality was prepared and contact between the gingival mucosa and the by ultrasonography proved good, but it was inadequate for instruments introduced in the antrum could thus be avoided. differentiating the type of abnormality. A possible use for max- The contaminants that still occurred were fewer in number, illary sinus ultrasonography could be in ICU monitoring. It making them easy to identify by the comparative quantitative may also replace plain sinus radiography for screening when gingival culture [66], and the infectious agents of the antral a continued antral investigation by CT or antroscopy is being cultures could then be recognized. considered [55, 56]. Among all specimens obtained from our population (33 pa- All diagnostic means should be validated. There is always tients; 66 antra examined and secretions cultured), four cultures individual variability, even in the performance of a professional of secretions were positive (table 6). When secretions are used task [75]. Thus, the reliability of endoscopic assessments by a for bacterial cultivation, it is possible to miss an infection that single rater, used as the standard, necessarily had to be con- is developing or established in the antral mucosa [82–84]. We trolled [64]. The individual variability existing also in the ex- suspected that the number of cases of infectious sinus- perts’ assessments [76, 77] contributed to the need for a second itis diagnosed through positive secretion cultures could be control [70, 78]. An expert panel of four was engaged for falsely low.

/ 9c57$$oc04 09-28-98 06:32:23 cidal UC: CID 860 Westergren et al. CID 1998;27 (October)

Mucosal specimens from patient 15 were cultured for bacte- The finding of RANTES in specimens without bacteria, both ria [66]. The specimens were prepared before the inoculation in the study of critically ill patients and in the reference mate- to find both extracellular and intracellular bacteria. Of the 38 rial, points to a normal intermittent RANTES activation. The mucosal specimens cultured, four were from culture-positive upregulation of RANTES production in the presence of bacte- antra (table 6). In one of these cases (patient 28), the findings ria, either in infectious disease or in mere colonization, indi- correlated to the positive secretion culture. In the other three, cates that more cellular elements for microbial defense are the secretion cultures were negative. called for. Our interpretation of these findings is that antra are As presented in table 6, one of the three latter cases (patient normally free of bacteria. 30; left antrum) was assessed endoscopically as infectious dis- IL-6 and IL-12 were both found in the clinical study, but in ease. The other two (patients 19 [left] and 33 [left]) manifested the reference material only IL-12 was found. The number of only inflammatory reactions, in accordance with the majority specimens expressing IL-6 was low, compared with the number

of the 66 examined antra. Hypothetically, they may represent expressing IL-12 and RANTES. This difference was not sig- Downloaded from https://academic.oup.com/cid/article/27/4/851/428661 by guest on 25 September 2021 either the beginning of an infection or a transient occurrence, nificant in our study, although it has been shown as such in which is further considered below in the discussion of the another study on nasal disease [89]. No certain correlation was RANTES findings. found between different expressions of maxillary sinus disease To investigate the therapeutic capability of antibiotics on site and IL-6 or IL-12. and to evaluate the possible antibiotic suppression of bacterial Compared to control CD31-stained sections, there was a growth in the cultures, we determined the efficacy of antibiotics general upregulation of the adhesion molecules ICAM-1, E- in diseased antral mucosa. We found that extracellular tissue selectin, and (even more prominent) P-selectin in all samples. concentrations of the different antibiotics used in the patient The VCAM-1 level was sparse in our material, which might group were comparable to the serum levels and reached or suggest a difference of activation compared with that in sea- surpassed MIC levels for several types of bacteria capable of sonal allergic rhinitis, in which an upregulation has been indi- being infective agents. The in vitro susceptibility of bacteria cated [92]. The same distribution of the same four adhesion to an antibiotic does not confirm their in vivo susceptibility. molecules in nasal polyp endothelium has been reported [93]; This depends on the on-site conditions, such as for the estab- thus, P-selectin appears to be of importance in the eosinophil lished empyemas found in three antra (patients 6 [left], 28 migration. However, histopathologic examination has not [left], and 30 [right]). shown any particularly enhanced presence of eosinophils in Antral empyema represent aged biofilm layers where bacte- the cellular infiltrate of specimens. Our findings support the ria are inaccessible to the antibiotic [85, 86]. The endoscopic, function of P-selectin in facilitating neutrophilic activation and bacteriologic, and histopathologic findings for the left antrum in recruitment of leukocytes in general [94]. of patient 30 suggested development of an anaerobic empyema. The results of our studies differ from the results of previous We knew from repeated MRI that on admission this patient’s studies. In 85% of the examined maxillary sinuses, some ex- antra were both normal; however, 1 week prior to the antros- pression of inflammatory disease was found [66]. This is similar copy, the right antrum was diseased, according to MRI, but to the frequencies of radiographic sinusitis in studies of the the left was still normal. This patient had received antibiotics same type of settings [24, 26, 29]. The rate of infectious sinus- since admission to the ICU. The suppression of bacterial growth itis in our material was 6%, where both bacteriologic and endo- is probably most prominent when the patient has recently re- scopic diagnoses concurred [66]. This is very low compared ceived a new antibiotic [87]. with the rate in previous publications (table 4). We suggest We found it interesting to investigate differences in the in- that the remaining 79% noninfected but diseased antra were flammatory reactions between the infectious and noninfectious affected by reactive inflammatory sinusitis. sources of our material and to compare our findings with those In the evaluation of material such as this, too detailed a in other investigations of allergic inflammation. The role of division cannot be made if proper statistical analysis is to be RANTES as a chemoattractant for monocytes, eosinophils, ba- performed. Therefore, the classes of inflammation were limited sophils, and T lymphocytes [88] in nasal polyposis and chronic to edema with secretion, edema only, and secretion only. In sinusitis [89] and in the bronchial mucosa of asthmatic patients both of the two main groups of patients in the study—nasotra- [90] has been shown. cheally intubated (n Å 20) and tracheotomized but previously In a recent study, RANTES instilled in nasal mucosa was nasotracheally intubated (n Å 10)—there was a concentration shown to cause an influx of eosinophils, metachromatic cells, of edema with secretion as the antral reaction. and lymphocytes [91]. When an allergic priming was per- The differences between the groups or between the kinds of formed before the RANTES challenge, basophils, monocytes, devices (or no devices) in the adjacent nasal cavity were not and neutrophils also joined the previously mentioned inflam- statistically significant (P ú .05). There were no signs of nor- matory cells in the influx. In our material a significant correla- malization of the maxillary sinus condition after a tracheotomy. tion was found between the positivity of antral culture, indicat- It is probably a coincidence that all three patients with infec- ing an infection or a colonization, and the presence of mRNA tious sinusitis in our study had an adjacent nasogastric tube for RANTES [74]. (table 6). As 3 of 4 maxillary sinus infections in the 3 patients

/ 9c57$$oc04 09-28-98 06:32:23 cidal UC: CID CID 1998;27 (October) Ventilator-Associated Sinusitis 861 were anaerobic and only 1 was due to EGNB (E. coli), there is However, our study found no bacterial sinusitis caused either no support for the hypothesis (comparable to the hypothesized by coagulase-negative staphylococci or by S. aureus, which importance of the gastropulmonary route in the pathogenesis were the two most common isolates from the ICUs where the of VAP [95, 96]) that nasogastric tubes would promote VAS. study was performed. Furthermore, the results of our study The bacteriology of infectious sinusitis in previous studies point to a high frequency in the Swedish ICU environment of (table 5) has often been likened to tracheal and bronchial cul- reactive inflammatory sinusitis (79%) and a fairly low fre- ture findings, with a predominance of aerobes (particularly quency of infectious sinusitis (6%), with the main pathogens P. aeruginosa and S. aureus), which take advantage of foreign being anaerobic bacteria. plastic bodies in nasal cavities. The infectious sinusitis cases Among the antra with reactive inflammatory sinusitis, three diagnosed in our study were predominantly (3 of 4) anaerobic were found to be colonized with anaerobic bacteria (table 6). empyemas (table 6). There are several factors in the patient Two of these were diagnosed by monocultures of Lactobacillus

group investigated that predisposed for anaerobic infections, species. These are bacteria which have not generally been con- Downloaded from https://academic.oup.com/cid/article/27/4/851/428661 by guest on 25 September 2021 e.g., immunosuppression, corticosteroids, tissue anoxia, and the sidered pathogenic [102], but new information is available. presence of a foreign body [50]. Nonspecific immunity response [103], including production of It is also important to review the bacteria identified in our IL-6 [104], has been shown to be stimulated by Lactobacillus study as infectious agents in relation to the national level and species. Clinically significant infections associated with Lacto- the level of the individual hospital [97]. A survey of recent bacillus isolates have been documented in cases of intraabdom- data (1994–1995) on species distribution among gram-negative inal abscesses and wounds [105] as well as septicemia and isolates from consecutive cultures in ICUs of 10 Swedish hospi- infective endocarditis [102]. Predisposition seems to occur tals revealed that E. coli was the most common [98]. P. aerugi- among the elderly and patients with immunocompromising dis- nosa, which is troublesome elsewhere [3], was in fourth place ease and in the presence of foreign bodies [102, 105]. These (12% of isolates) [98]. It seems most plausible to expect the reports suggest that the finding of Lactobacillus species may frequency of P. aeruginosa we noted (6%) in the gingival not be unexpected in a population of critically ill patients. cultures, in comparison with findings in the study of Hanberger The diagnostic difficulties of VAS are similar to those of et al. [98], in which all clinically indicated cultures were per- VAP [106]. The frequency of VAP, or suspected VAP [107], formed for patients admitted to ICUs. In a study of VAP with is very much higher than that of VAS. The attempts to achieve high clinical and diagnostic criteria (ú106 cfu/mL in protected- better means of diagnosing VAP have been many [87]. Invasive brush specimens or ú105 cfu/mL in bronchoalveolar lavage techniques to acquire samples, e.g., protected-brush sampling specimens), it was shown that at one ICU of the 10 Swedish and bronchoalveolar lavage, have been evaluated, and the pros hospitals in the previously mentioned study, the most com- [108] and cons [107] considered. Insufficiency in VAP diagnos- monly occurring infectious agent was S. aureus, followed by tics by means of clinical, radiological, and laboratory criteria Haemophilus influenzae and Streptococcus pneumoniae (P. Ap- has been demonstrated [41, 108]; the diagnosis has been shown pelgren et al., unpublished data). to be incorrect in up to 77% of cases. Of 75 cases of VAP over 9 months, none was caused by This also suggests that antibiotics are prescribed to many P. aeruginosa. This has been suggested to be a result of having patients who do not need them. The diminishing bacterial sus- an antimicrobial-policy team and of measures taken by health ceptibility [109] within the ICU setting [110] underlines the care workers against transmission of bacteria [97]. Specific importance of the three objectives stated by Chastre et al.: (1) measures were also taken regarding the suction apparatus and to identify patients who need treatment with antibiotics for procedure [99], together with careful aspiration of subglottic bacterial pneumonia; (2) to select optimal antimicrobial regi- secretions [100]; the latter may affect the environment nega- mens on the basis of identification of the causative microorgan- tively in terms of P. aeruginosa colonization. ism; and (3) to withhold antimicrobial treatment from patients The two ICUs participating in our study were evaluated without pneumonia [87]. The basis of therapy guided by these together with two other ICUs within the same hospital with objectives rests on widespread use of bronchoscopy and on regard to frequency of resistance among the bacteria occurring site sampling, i.e., with a protective brush or bronchoalveolar from 1993 to 1996 [101]. This evaluation was partially parallel lavage. An analogous approach with improved diagnostic accu- with our study, and thus the culture and susceptibility results racy could probably also be applied for VAS. for patient 20 and those following were included in both evalua- In our studies, for maxillary sinus diagnostics in the ICU, tions. During the 4-year period, the most common isolates the methods were improved. The results suggest that require- (Ç40% each year) were coagulase-negative staphylococci, fol- ments for the diagnosis of infectious VAS should be inspection lowed by S. aureus and enterococci. The presence of P. aerugi- by antroscopy and quantitative bacterial culture of antral speci- nosa among the isolates varied between 4% and 10%. Anaer- mens, compared with cultures of specimens from the passage obes rarely occurred. route. The canine fossa is recommended for penetration. There These reports may explain why no infectious sinusitis due is a small difference in invasiveness between a single antral to P. aeruginosa was found in our material. The other aerobic puncture and the method presented here. Since superior diag- infectious agents could have been expected to occur instead. nostic ability can be achieved with antroscopy, it should not

/ 9c57$$oc04 09-28-98 06:32:23 cidal UC: CID 862 Westergren et al. CID 1998;27 (October) be avoided in investigation of suspected maxillary sinus infec- 4. Cunha B, Shea K. Fever in the intensive care unit. Infect Dis Clin North tion in a critically ill patient. The maxillary sinus infections Am 1996;10:185–209. 5. Schwartz S, Dowling J, Benkovic C, DeQuittner-Buchanan M, Prostko identified in this study show that it is important not only to T, Yee R. Sources of gram-negative bacilli colonizing the trachea of determine the causative microorganism but to drain an antral intubated patients. J Infect Dis 1978;138:227–31. empyema’s retained secretion before antibiotic treatment can 6. Donaldsson S, Azizi S, Dal Nogare A. Characteristics of aerobic gram- be successful [111]. negative bacteria colonizing critically ill patients. Am Rev Respir Dis 1991;144:202–7. 7. Arens J, LeJeune Jr F, Webre D. Maxillary sinusitis, a complication of Summary nasotracheal intubation. Anesthesiology 1974;40:415–6. 8. Gallagher TJ, Civetta JM. Acute maxillary sinusitis complicating nasotra- Ultrasonographic examination has great value for daily mon- cheal intubation: a case report. Anesth Analg 1976;55:885–6. 9. Pope T, Stelling C, Leitner Y. Maxillary sinusitis after nasotracheal intu- itoring and screening of antral abnormalities in ICU patients, bation. South Med J 1981;74:610–2. Downloaded from https://academic.oup.com/cid/article/27/4/851/428661 by guest on 25 September 2021 but otherwise it gives little differentiating information. The 10. Knodel AR, Beekman JF. Unexplained fevers in patients with nasotra- examination should be performed in a semirecumbent position cheal intubation. JAMA 1982;248:868–70. (30Њ) or with the head raised even higher. In the overall diag- 11. O’Reilly MJ, Reddick EJ, Black W, et al. Sepsis from sinusitis in nasotra- nostic assessment (for infectious disease) at antroscopy, the cheally intubated patients: a diagnostic dilemma. Am J Surg 1984; 147:601–4. interobserver concordance is almost perfect when the quality 12. Willatts SM. Paranasal sinusitis: a complication of nasotracheal intuba- of antroscopy as the means for diagnosis is evaluated by k tion. Two case reports. Br J Anaesth 1985;57:1026–8. statistics. 13. Meyer P, Guerin JM, Habib Y, Levy C. Pseudomonas thoracic empyema A substantial reduction of bacterial contamination in antral secondary to nosocomial rhinosinusitis. Eur Respir J 1988;1:868–9. samples for bacterial cultivation is achieved when a free bone 14. Wolf M, Zillinsky I, Lieberman P. Acute mycotic sinusitis with bacterial sepsis in orotracheal intubation and nasogastric tubing: a case report area is prepared to avoid contact between the gingival mucosa and a review of the literature. Otolaryngol Head Neck Surg 1988; and the instruments introduced into the antrum and when a 98(6):615–7. protective tube is placed over the cotton swab used for the 15. Caplan ES, Hoyt NJ. Nosocomial sinusitis. JAMA 1982;247(5):639–41. sampling. The resulting decrease in the number of contaminat- 16. Deutschman CS, Wilton PB, Sinow J, Thienprasit P, Konstantinides FN, ing bacteria, together with the use of a comparative quantitative Cerra FB. Paranasal sinusitis: a common complication of nasotracheal intubation in neurosurgical patients. Neurosurgery 1985;17(2):296–9. gingival culture, enables more conclusive identification of the 17. Kronberg FG, Goodwin WJ. Sinusitis in intensive care unit patients. contaminants in antral cultures. Laryngoscope 1985;95:936–8. There is a significant occurrence of mRNA for RANTES in 18. Grindlinger GA, Niehoff J, Hughes SL, Humphrey MA, Simpson G. antral mucosa when bacteria are found in the antrum (either Acute paranasal sinusitis related to nasotracheal intubation of head- as infection or colonization), thus demonstrating a prerequisite injured patients. Crit Care Med 1987;15:214–7. 19. Humphrey MA, Simpson GT, Grindlinger GA. Clinical characteristics for chemoattraction of several inflammatory cells. of nosocomial sinusitis. Ann Otol Rhinol Laryngol 1987;96:687–90. In nonallergic reactive inflammatory disease of the maxillary 20. Aebert H, Hu¨nefeld G, Regel G. Paranasal sinusitis and sepsis in ICU sinuses, prominent expression of P-selectin (an endothelial ad- patients with nasotracheal intubation. Intens Care Med 1988;15: hesive molecule) is found. In spite of what could be expected, 27–30. the frequency and character of inflammatory maxillary sinus 21. Bell R, Page G, Bynoe R, Dunham M, Brill A. Post-traumatic sinusitis. J Trauma 1988;28:923–30. disease is unchanged after tracheotomy in previously nasotra- 22. Linden B, Aguilar E, Allen S. Sinusitis in the nasotracheally intubated cheally intubated, critically ill long-term mechanically venti- patient. Arch Otolaryngol Head Neck Surg 1988;114:860–1. lated patients. Infectious sinusitis in critically ill mechanically 23. Bert F, Lambert-Zechovsky N. Microbiology of nosocomial sinusitis in ventilated patients is predominantly caused by anaerobic bacte- intensive care unit patients. J Infection 1995;31:5–8. ria that are forming empyemas. 24. Hansen M, Poulsen MR, Bendixen DK, Hartmann-Andersen F. Incidence of sinusitis in patients with nasotracheal intubation. Br J Anaesth 1988; Our results show that colonization of antra may occur in 61:231–2. critically ill mechanically ventilated patients, mainly by anaero- 25. Strange C, Wooten S, Schabel S, Heffner J, Sahn S. Radiographic sinusitis bic or facultative anaerobic bacteria and even by species nor- following endotracheal intubation. Am Rev Respir Dis 1988; mally not considered as pathogens. 137:63. 26. Fassoulaki A, Pamouktsoglou P. Prolonged nasotracheal intubation and its association with inflammation of paranasal sinuses. Anesth Analg 1989;69:50–2. References 27. Weymuller E, Bishop M. Problems associated with prolonged intubation 1. Johanson W, Pierce A, Sanford J. Changing pharyngeal bacterial flora in the geriatric patient. Otolaryngol Clin North Am 1990;23: of hospitalized patients. N Engl J Med 1969;281:1137–40. 1057–74. 2. Vallandigham JC, Johanson Jr WG. Infections associated with endotra- 28. Desmond P, Raman R, Idikula J. Effect of nasogastric tubes on the nose cheal intubation and tracheostomy. In: Bisno AL, Waldvogel FA, eds. and maxillary sinus. Crit Care Med 1991;19:509–11. Infections associated with indwelling devices. Washington, DC: Amer- 29. Pedersen J, Schurizek BA, Melsen NC, Juhl B. The effect of nasotracheal ican Society for Microbiology, 1989:179–97. intubation on the paranasal sinuses: a prospective study of 434 inten- 3. Emori T, Gaynes R. An overview of nosocomial infections, including sive care patients. Acta Anaesth Scand 1991;35:11–3. the role of microbiology laboratory. Clin Microbiol Rev 1993;6: 30. Lucchin F, Minicuci N, Ravasi M, et al. Comparison of A-mode ultra- 428–42. sound and computed tomography: detection of secretion in maxillary

/ 9c57$$oc04 09-28-98 06:32:23 cidal UC: CID CID 1998;27 (October) Ventilator-Associated Sinusitis 863

and frontal sinuses in ventilated patients. Intens Care Med 1996;22: 52. Evans Jr F, Sydnor J, Moore W, et al. Sinusitis of the maxillary antrum. 1265–8. N Engl J Med 1975;293:735–9. 31. Gue´rin JM, Meyer P, Le´vy C, et al. Sinusites aigue¨s spheńoidales conseću- 53. Druce H. Agents, diagnostic strategies and techniques, treatments, and tive a` l’intubation naso-tracheále. Sem Hoˆp Paris 1987;63(47–48): problems. Immunol Allergy Clin 1993;13:119–32. 3671–4. 54. Jorissen M. Recent trends in the diagnosis and treatment of sinusitis. Eur 32. Fougeront B, Bodin L, Lamas G, Bokowy C, Elbez M. Sinusites de Radiol 1996;6:170–6. reánimation. Ann Otolaryngol Chir Cervicofac 1990;107:329–32. 55. Revonta M. Ultrasound in the diagnosis of maxillary and frontal sinusitis. 33. Deutschman CS, Wilton PB, Sinow J, Dibbell D Jr, Konstantinides FN, Acta Otolaryngol Suppl 1980;370:1–54. Cerra FB. Paranasal sinusitis associated with nasotracheal intubation: 56. Jannert M, Andreásson L, Holmer N-G, Lo¨rinc P. A comparison between a frequently unrecognized and treatable source of sepsis. Crit Care different ultrasonic display techniques, radiography and invasive con- Med 1986;14:111–4. trol for different disorders of the paranasal sinuses. Acta Otolaryngol 34. Gue´rin JM, Meyer P, Segrestaa JM, Reizine D, Levy C. Sinusites noso- Suppl 1982;389:29–52. comiales et intubation nasotracheále. E´ tude prospective a´ partir de 53 57. DeWeese D, Saunders W. Acute and chronic sinusitis. In: Textbook of patients. Ann Med Interne 1989;140(2):106–7. otolaryngology. 6th ed. St. Louis: Mosby, 1982:223–37. Downloaded from https://academic.oup.com/cid/article/27/4/851/428661 by guest on 25 September 2021 35. Salord F, Gaussorgues P, Marti-Flich J, et al. Nosocomial maxillary 58. Seiden AM. Sinusitis in the critical care patient. New Horizons 1993;1: sinusitis during mechanical ventilation. a prospective comparison of 261–70. orotracheal versus the nasotracheal route for intubation. Intens Care 59. Sottile F, Marrie T, Prough D, et al. Nosocomial pulmonary infection: Med 1990;16:390–3. possible etiologic significance of bacterial adhesion to tracheal tubes. 36. Bowers B, Purdue G, Hunt J. Paranasal sinusitis in burn patients following Crit Care Med 1986;14:265–70. nasotracheal intubation. Arch Surg 1991;126:1411–2. 60. Dougherty S. Pathobiology of infection in prosthetic devices. Rev Infect 37. Bach A, Boehrer H, Schmidt H, Geiss K. Nosocomial sinusitis in venti- Dis 1988;10:1102–17. lated patients: nasotracheal versus orotracheal intubation. Anaesthesia 61. Peters G, Gray ED, Johnson GM. Immunomodulating properties of extra- 1992;47:335–9. cellular slime substance. In: Bisno AL, Waldvogel FA, eds. Infections 38. Borman K, Brown P, Mezera K, Jhaveri H. Occult fever in surgical associated with indwelling medical devices. Washington, DC: Ameri- intensive care unit patients is seldom caused by sinusitis. Am J Surg can Society for Microbiology, 1989:61–74. 1992;164:412–6. 62. Gristina AG, Giridhar G, Gabriel BL, Naylor PT, Myrvik QN. Cell 39. Michelson A, Schuster B, Kamp H-D. Paranasal sinusitis associated with biology and molecular mechanisms in artificial device infection. Int J nasotracheal and orotracheal long-term intubation. Arch Otolaryngol Artif Organs 1993;16:755–64. Head Neck Surg 1992;118:937–9. 63. Westergren V, Berg S, Lundgren J. Ultrasonography bedside evaluation of maxillary sinus disease in mechanically ventilated patients. Intens 40. Holzapfel L, Chevret S, Madinier G, et al. Influence of long-term oro- Care Med 1997;23:393–8. or nasotracheal intubation on nosocomial maxillary sinusitis and pneu- 64. Westergren V, Lundblad L, Timpka T. Interobserver variations in assess- monia: results of a prospective, randomized, clinic trial. Crit Care Med ment of antral disease from sinoscopy video recordings. Am J Rhinol 1993;21:1132–8. 1998;12:159–65. 41. Meduri G, Mauldin G, Wunderink R, Leeper K, Jones C, Tolley E, 65. Westergren V, Forsum U, Lundgren J. Possible errors in diagnosis of Mayhall G. Causes of fever and pulmonary densities in patients with bacterial sinusitis in tracheally intubated patients. Acta Anaesth Scand clinical manifestations of ventilator-associated pneumonia. Chest 1994;38:699–703. 1994;106:221–35. 66. Westergren V, Lundblad L, Forsum U. Ventilator-associated sinusitis: 42. Rouby J-J, Laurent P, Gosnach M, et al. Risk factors and clinical rele- antroscopy findings and bacteriology when excluding the contami- vance of nosocomial maxillary sinusitis in the critically ill. Am J nants. Acta Otolaryngol 1998;118:574–80. Respir Crit Care Med 1994;150:776–83. 67. Westergren V, Nilsson M, Forsum U. Penetration of antibiotics in dis- 43. Bert F, Lambert-Zechovsky N. Sinusitis in mechanically ventilated pa- eased antral mucosa. Arch Otolaryngol Head Neck Surg 1996;122: tients and its role in the pathogenesis of nosocomial pneumonia. Eur 1390–4. J Clin Microbiol Infect Dis 1996;15:533–44. 68. Baron E, Finegold S. Selection, collection, and transport of specimens 44. Levine S, Niederman M. The impact of tracheal intubation on host de- for microbiological examination. In: Baron E, Finegold S, eds. Bai- fences and risks for nosocomial pneumonia. Clin Chest Med 1991;12: ley & Scott’s diagnostic microbiology. 8th ed. St. Louis: CV Mosby, 523–43. 1990:49–63. 45. Fagon J-Y, Chastre J, Domart Y, Trouillet J-L, Pierre J, Darne C, Gibert 69. Baron E, Finegold S. Cultivation and isolation of viable pathogens. In: C. Nosocomial pneumonia in patients receiving continuous mechanical Baron E, Finegold S, eds. Bailey & Scott’s diagnostic microbiology. ventilation. Am Rev Respir Dis 1988;139:877–84. 8th ed. St. Louis: CV Mosby, 1990:81–99. 46. Rice DH. The microbiology of paranasal sinus infections: diagnosis and 70. Landis JR, Koch GG. The measurement of observer agreement for cate- management. Crit Rev Clin Lab Sci 1978;9:105–21. gorical data. Biometrics 1977;33:159–74. 47. Brook I. Aerobic and anaerobic bacterial flora of normal maxillary si- 71. Gjørup T. The kappa coefficient and the prevalance of a diagnosis. nuses. Laryngoscope 1981;91:372–6. Method Inform Med 1988;27:184–6. 48. Sobin J, Engquist S, Nord CE. Bacteriology of the maxillary sinus in 72. Haas M. Statistical methodology for reliability studies. J Manip Physiol healthy volunteers. Scand J Infect Dis 1992;24:633–5. Ther 1991;14(2):119–32. 49. Laranne J, Penttile` M, Paakala T, Pukander J. Diagnostic value of plain 73. Cars O. A microtechnique for the determination of antibiotics in muscle. radiographs in chronic maxillary sinusitis: a comparison between ra- J Antimicrob Chemother 1981;8:39–48. diological and endoscopic findings in 75 patients. Rhinology 1992;30: 74. Westergren V, Viale G, Dell’Orto P, Pellegrini C, Hellquist H. RANTES 205–15. is more prevalent in bacterial than non-bacterial maxillary sinusitis, 50. Finegold S, Jousimies-Somer H, Wexler H. Current perspectives on an- and P-selectin is preferentially upregulated in diseased mucosae. Arch aerobic infections: diagnostic approaches. Infect Dis Clin North Am Otolaryngol Head Neck Surg 1997;123:1103–10. 1993;7:257–75. 75. Cohen J. A coefficient of agreement for nominal scales. Educ Psychol 51. Axelsson A, Brorson J. The correlation between bacteriological findings Meas 1960;20:37–46. in the nose and maxillary sinus in acute maxillary sinusitis. Laryngo- 76. Gjørup T, Agner E, Jensen LB, Jensen AM, Møllman K-M. The endo- scope 1973;83:2003–11. scopic diagnosis of duodenal ulcer disease: a randomized clinical trial

/ 9c57$$oc04 09-28-98 06:32:23 cidal UC: CID 864 Westergren et al. CID 1998;27 (October)

of bias of interobserver variation. Scand J Gastroenterol 1986;21: 94. Noto G, Pravata` G, Malizia G, Aragona E, Arico` M. Expression of P- 261–7. and E-selectin on activated endothelial cells in skin inflammation. In: 77. Timpka T, Buur T. Inter-observer variations in decision-making regarding Schlossman SF, Boumsel L, Gilks W, et al., eds. Leucocyte typing. patients on chronic hemodialysis: a study using the kappa index. Meth- White cell differentiation antigens. Vol. 2. Oxford: Oxford University ods Inf Med 1990;29(2):23–9. Press, 1994:1820–1. 78. Davies M, Fleiss JL. Measuring agreement for multinominal data. Bio- 95. Heyland D, Mandell LA. Gastric colonization by gram-negative bacilli metrics 1982;38:1047–51. and nosocomial pneumonia in the intensive care unit patient. Chest 79. Petruson B. A form to show extension of tumors in the nose and sinus. 1992;101:187–93. Arch Otolaryngol Head Neck Surg 1982;108:460–1. 96. Bonten M, Gaillard C. Ventilator-associated pneumonia: do the bacteria 80. Westergren H, Westergren V, Forsum U. Propionebacterium acnes in come from the stomach? Neth J Med 1995;46:1–3. cultures from ventriculo-peritoneal shunts: infection or contamination? 97. Goldman DA, Weinstein RA, Wenzel RP, et al. Strategies to prevent and Acta Neurochir 1997;139:33–6. control the emergence and spread of antimicrobial-resistant microor- 81. Wimberley N, Faling L, Bartlett J. A fiberoptic bronchoscopy technique ganisms in hospitals. JAMA 1996;275:234–40. to obtain uncontaminated lower airway secretions for bacterial culture. 98. Hanberger H, Nilsson LE, the Swedish Study Group. High frequency of Downloaded from https://academic.oup.com/cid/article/27/4/851/428661 by guest on 25 September 2021 Am Rev Respir Dis 1979;119:337–43. antibiotic resistance among gram-negative isolates in intensive care 82. Wen-Yang S, Chen L, Shuian-Yeng H, Wei-Fu T. Bacteriological study units at 10 Swedish hospitals. Clin Microbiol Infect 1997;3:208–15. in chronic maxillary sinusitis. Laryngoscope 1983;93:931–4. 99. Creamer E, Smyth EG. Suction apparatus and the suctioning procedure: 83. Verbist L. What and where are the pathogens? Research and Clinical reducing the infection risks. J Hosp Infect 1996;34:1–9. Forums 1990;12(4):9–14. 100. Rello J, Sonõra R, Jubert P, Artigas A, Rue´ M, Valle´s J. Pneumonia in 84. Erkan M, Aslan T, O¨ zcan M, Koç N. Bacteriology of antrum in adults intubated patients: role of respiratory airway care. Am J Respir Crit with chronic maxillary sinusitis. Laryngoscope 1994;104:321–4. Care Med 1996;154:111–5. 85. Brown MRW, Collier PJ, Gilbert P. Influence of growth rate on suscepti- 101. Hanberger H, Hoffman M, Lindgren S, Nilsson LE. Increased antibiotic bility to antimicrobial agents: modification of the cell envelope and resistance among bacteria in four intensive care units at a university batch and continuous culture studies. Antimicrob Agents Chemother hospital in Sweden. Scand J Infect Dis 1997;29:607–14. 1990;34:1623–8. 102. Harty D, Oakey H, Patrikakis M, Hume E, Knox K. Pathogenic potential 86. Anwar H, Strap JL, Costerton JW. Establishment of aging biofilms: possi- of lactobacilli. Int J Food Microbiol 1994;24:179–89. ble mechanism of bacterial resistance to antimicrobial therapy. Antimi- 103. Pouwels P, Leer R, Boersma W. The potential of Lactobacillus as a crob Agents Chemother 1992;36(7):1347–51. carrier for oral immunization: development and preliminary character- 87. Chastre J, Fagon J-Y, Trouillet J. Diagnosis and treatment of nosocomial ization of vector systems for targeted delivery of antigens. J Biotechnol pneumonia in intensive care units. Clin Infect Dis 1995;21(suppl 3): 1996;44:183–92. S226–37. 104. Miettinen M, Vuopio-Varkila J, Varkila K. Production of human tumor 88. Schall TJ. The chemokines. In: Thomson AW, ed. The cytokine hand- necrosis factor alpha, interleukin-6, and interleukin-10 is induced by book. 2nd ed. London: Academic Press, 1994:419–60. lactic acid bacteria. Infect Immun 1996;64:5403–5. 89. Davidsson A˚ , Danielsen A, Viale G, et al. Positive identification in situ 105. Brook I, Frazier E. Significant recovery of nonsporulating anaerobic rods of mRNA expression of IL-6 and IL-12, and the chemotactic cytokine from clinical specimens. CID 1993;16:476–80. RANTES in patients with chronic sinusitis and polypoid disease. Acta 106. Lipchik R, Kuzo R. Nosocomial pneumonia. Radiol Clin North Am 1996; Otolaryngol 1996;116:604–10. 34:47–58. 90. Humbert M, Ying S, Corrigan C, et al. Bronchial mucosal expression of 107. Campbell G Jr, Niederman M, Broughton W, et al. Hospital-acquired the genes encoding chemokines RANTES and MCP-3 in symptomatic pneumonia in adults: diagnosis, assessment of severity, initial antimi- atopic and nonatopic asthmatics: relationship to the eosinophil-active crobial therapy, and preventative strategies. Am J Respir Crit Care cytokines interleukine (IL)-5, granulocyte-macrophage-colony-stimu- Med 1996;153:1711–25. lating factor, and IL-3. Am J Respir Cell Mol Biol 1997;16:1–8. 108. Sterling T, Ho E, Brehm W, Kirkpatrick M. Diagnosis and treatment 91. Kuna P, Alam R, Ruta U, Gorski P. RANTES induces nasal mucosal of ventilator-associated pneumonia—impact on survival. Chest 1996; inflammation rich in eosinophils, basophils, and lymphocytes in vivo. 110:1025–34. Am J Respir Crit Care Med 1998;157:873–9. 109. Neu H. The crisis in antibiotic resistance. Science 1992;257:1064–73. 92. Karlsson M, Hellquist H. Adhesion molecules for nasal-homing T cells 110. Kollef M. Antibiotic use and antibiotic resistance in the intensive care in allergy. Virchows Arch 1996;429:49–54. unit: are we curing or creating disease? Heart Lung 1994;23:363–7. 93. Symon F, Walsh G, Watson S, Wardlaw A. Eosinophil adhesion to nasal 111. Finegold SM, Strong CA, McTeague M, Marina M. The importance of polyp endothelium is P-selectin dependent. J Exp Med 1994;180: black-pigmented gram-negative anaerobes in human infections. FEMS 371–6. Immunol Med Microbiol 1993;6:77–82.

/ 9c57$$oc04 09-28-98 06:32:23 cidal UC: CID