Macroscopic Purulence, Leukocyte Counts, and Bacterial Morphotypes in Relation to Culture Findings for Sinus

Transcription

1 JOURNAL OF CLINICAL MICROBIOLOGY, OCt. 1988, p /88/ $02.00/0 Copyright 1988, American Society for Microbiology Vol. 26, No. 10 Macroscopic Purulence, Leukocyte Counts, and Bacterial Morphotypes in Relation to Culture Findings for Sinus Secretions in Acute Maxillary Sinusitis HANNELE R. JOUSIMIES SOMER,1* SEPPO SAVOLAINEN,2 AND JUKKA S. YLIKOSKI2 Anaerobe Reference Unit, National Public Health Institute,' and Department of Otolaryngology, Central Military Hospital,2 SF Helsinki, Finland Received 1 March 1988/Accepted 14 June 1988 Macroscopic purulence, leukocyte counts, and bacterial morphotypes in Gram-stained smears were investigated in 335 sinus secretions (240 aspirates and 95 injection aspirates) obtained by puncture in 234 young patients with acute maxillary sinusitis. Over 90% of the 147 aspirates macroscopically classified as purulent also contained high numbers of leukocytes (>20 per oil immersion field). A total of 82% of the 147 macroscopically purulent aspirates and 79% of the 156 aspirates containing high numbers of leukocytes yielded presumed sinus pathogens by culture in quantities of > 103 CFU/ml. Streptococcus pneumoniae or Streptococcus pyogenes was associated relatively more often (92 or 87%, respectively) with high numbers of leukocytes than Haemophilus influenzae, which was not infrequently (29%) recovered from the less purulent aspirates. When a bacterial morphotype was seen in the Gram-stained smear, a corresponding sinus pathogen was isolated in quantities of >103 CFU/ml in 92% of aspirates. Other bacterial species (most often staphylococci) were usually isolated in low numbers and were almost never seen in the smear, suggesting nasal contamination. The 95 injection aspirates behaved, to a large extent, like diluted aspirates, with the exception that there was a higher frequency of probable nasal contamination. Macroscopic purulence, high leukocyte counts, and bacterial morphotypes seen in Gram-stained smears each predicted well the isolation of a presumed sinus pathogen and in some cases supported the significance of an otherwise doubtful culture finding. However, the macroscopic appearance of the secretion should not be used to screen samples for culture, because in several cases H. influenzae grew from nonpurulent samples as well. The bacterial etiology of acute maxillary sinusitis (AMS) has been the subject of many studies during the past few years. Streptococcus pneumoniae and Haemophilus influenzae are the main etiological agents present in over half of the patients with AMS (1, 7, 10-12, 15, 19, 20, 23). Streptococcus pyogenes and Branhamella catarrhalis have usually been isolated in less than 5% of adult patients with AMS (1, 3, 10, 11). The reported frequencies of isolation of anaerobic bacteria have varied considerably, from 0 to 23% (2, 7, 18, 22, 25-27), with the lowest isolation rates being in children (26, 27). We recently conducted a study of the bacterial etiology of AMS in 238 young adults by paying special attention to the collection and processing of specimens (14). In this homogeneous patient population, a sinus pathogen was found in the secretions of 75% of the patients. Somewhat in contrast to previous findings, H. influenzae was the most commonly isolated species; it was recovered more than twice as often as Streptococcus pneumoniae, whereas anaerobes were a significant finding in only 2% of the patients. From clinical and laboratory standpoints, it is important to know how the macroscopic and microscopic qualities of the sinus secretions correlate with culture results. Specimens macroscopically classified as purulent have significantly more often yielded pathogenic bacteria in high quantities than the nonpurulent ones, which are often negative by culture (5-8, 20). Likewise, a good correlation has been demonstrated between high leukocyte counts (>1,000/mm3) of the secretions and infection with high titers of bacteria (6, 8, 10). A correlation of the bacterial morphotypes seen in Gram-stained smears with culture findings has only occa- * Corresponding author sionally been reported for isolates from patients with AMS (5, 21, 26), although examination of Gram-stained smears is a routine procedure in clinical microbiology laboratories. Our bacteriological survey of young adult patients with AMS (14) offered an opportunity to correlate both macroscopic and microscopic findings of sinus secretions with culture results. Specifically, we wanted to know whether recording of macroscopic purulence and examination of Gram-stained smears, both of which are easily performed in a routine microbiology laboratory, could help in predicting the bacterial etiology of AMS or in assessing the significance of culture results. MATERIALS AND METHODS Patients. The study group consisted of 234 consecutive patients diagnosed with AMS based on clinical and radiographic findings (14). Of the 238 patients described in the accompanying report (14), 4 were omitted from the present study because their specimens were not evaluated macroscopically or microscopically. Of the 234 patients, 8 were women and 226 were men (ages, 17 to 46 years; mean, 21 years) serving in the Finnish military service who were sent to the Ear, Nose, and Throat Department of the Central Military Hospital, Helsinki, between September 1983 and March 1986 for consultation and medical care. None of the patients had experienced symptoms for more than 3 weeks, and none of them had received antimicrobial agents during the 2 weeks before the examination or earlier for the present infection. Patients with a concomitant dental root canal infection suggesting dentogenic sinusitis were excluded. Informed consent was obtained from the patients before the invasive procedures were performed.

2 VOL. 26, 1988 PURULENCE AND MICROSCOPIC FINDINGS IN ACUTE SINUSITIS 1927 Specimen collection and culture. Altogether, 335 sinus secretions (240 aspirates and 95 injection aspirates) were collected and quantitatively cultured for bacteria, as described previously (14). Macroscopic examination of secretions. The specimens were received in the laboratory in 20-ml syringes within 5 min of collection. They were immediately inspected for purulence, homogenized with a Vortex mixer (Vortex- Genie; model K-550-GE; Scientific Industries Inc., Bohemia, N.Y.), passed on a clear petri dish, and carefully reinspected under a spotlight for flakes of purulent material. They were classified as purulent (often opaque, greenish, or yellowish secretions with lumps or flakes of nonviscous pus) or nonpurulent (often clear or pale yellow) secretions of low viscosity frequently forming a gel or as grey or brownish semiopalescent mucous secretions of high viscosity, as described by Carenfelt and Lundberg (5). An additional category of bloody specimens was established for secretions that were heavily contaminated with blood. Microscopic examination. (i) Leukocytes. After careful inspection of the secretion in the petri dish, a purulent part (when present) was selected, and approximately 10 iul/cm2 (13) was smeared gently and evenly onto a microscope slide. The slides were dried in air and cautiously passed through a flame for fixation. The smears were stained by the Hucker modified Gram stain procedure (16) and examined under an oil immersion objective at x 1,000 magnification. All smears were examined by one of us independently of the culture results. Polymorphonuclear leukocytes (PMNs) were counted in 20 oil immersion fields (OIFs), and the average number of PMNs per field was calculated. The secretions were classified into four groups according to the following number of PMNs seen per OIF: >20 (group 1; high number of PMNs), 6 to 20 (group 2; moderate number of PMNs), 1 to 5 (group 3; few PMNs), or less than 1 (group 4; very few PMNs). (ii) Bacterial morphotypes. We searched for bacteria in each smear (minimum of 20 fields) by paying special attention to the possible presence of small, faintly staining, gram-negative rods (concordant with H. influenzae. We grouped the morphotypes into the following categories: gram-negative rods, gram-negative rods in pairs, gram-negative diplococci, gram-positive cocci, gram-positive diplococci, gram-positive cocci in chains, and gram-positive rods (non-spore-forming). Statistical methods. The data were analyzed by the chisquare test. The predictive values were determined as described by Vecchio (24). RESULTS Macroscopic purulence and leukocyte counts. Of the 240 aspirates, 147 (61%) were macroscopically purulent, and 92% of them also contained high numbers of leukocytes, as determined by microscopy (Table 1). In fact, most of these aspirates (111 of 135 [82%]) contained >100 PMNs/OIF (data not shown). In addition, high numbers of leukocytes were also seen in 25% of the 69 macroscopically nonpurulent aspirates. The 95 injection aspirates mainly had the same characteristics as the aspirates. Conversely, most (86%) of the 156 aspirates with high numbers of PMNs were already identified as purulent by macroscopic examination (Table 2). In samples with moderate numbers of PMNs, the macroscopic appearance was a less reliable predictor, whereas samples with very few PMNs were, as a rule, nonpurulent. TABLE 1. Macroscopic purulence in relation to leukocyte counts of Gram-stained smears of sinus secretions Macroscopic purulence % Smears in each macroscopic category with the following no. of PMNs/OIF > <1 Aspirates (n = 240) Purulent (n = 147) (61") Nonpurulent (n = 69) (29) Bloody (n = 24) (10) Injection aspirates (n = 95) Purulent (n = 23) (24") Nonpurulent (n = 55) (58) Bloody (n = 17) (18) Percentage of all aspirates. "Percentage of ail injection aspirates. Macroscopic purulence versus culture findings. H. influenzae and Streptococcus pneumoniae were the most commonly found organisms in the purulent aspirates (57 and 29%, respectively) (Table 3). Streptococcus pyogenes, B. catarrhalis, or a mixture of anaerobes was found in 12% of the purulent aspirates. Other bacterial species were recovered in 13% of the samples, but most often they were recovered in small quantities (<103 CFU/ml) and were often recovered together with a presumed sinus pathogen. The culture of 12 (8%) purulent aspirates remained negative. On the other hand, in 82% of the purulent aspirates pathogens were present in quantities exceeding 103 CFU/ml. H. influenzae was also recovered by culture relatively often from the nonpurulent aspirates (42%), but Streptococcus pneumoniae and Streptococcus pyogenes were recovered by culture much less often (7 and 1%, respectively; a significant difference [P < 0.01] from recovery from purulent samples). No anaerobes were isolated from the nonpurulent specimens, and other bacterial species were present in only 4% of specimens. About one-half of the nonpurulent aspirates were negative by culture, and pathogens in quantities of >10' CFU/ml were present in just less than one-third of the specimens. About one-third of the 24 bloody aspirates were negative by culture, and pathogens in quantities exceeding >103 CFU/ml were present in one-half of them. Results for injection aspirates followed the same pattern, with the exception that the proportion (25%) of nonpurulent specimens yielding other bacterial species in culture was considerably (P < 0.01) higher than that (4%) in the aspirates. More than one-half of the specimens containing other bacterial species did not simultaneously contain presumed sinus pathogens. When the culture results were analyzed in relation to macroscopic purulence of the secretions (Fig. 1), it was found that a high proportion of the aspirates growing a sinus pathogen were purulent (for H. influenzae, 68%; for Streptococcus pneumoniae, 84%; for Streptococcus pyogenes, 73%; for B. catarrhalis, 75%; for anaerobes, 100%). In contrast, the majority (64%) of the culture-negative aspirates were nonpurulent. The same trend was seen among the injection aspirates, but a larger proportion of both H. influenzae and Streptococcus pneumoniae found in them originated from nonpurulent specimens. All of the injection aspirates that yielded other species of bacteria, together with the presumed sinus pathogens, were either nonpurulent or bloody. Leukocyte counts versus culture findings. The frequencies

3 1928 JOUSIMIES-SOMER ET AL. J. CLIN. MICROBIOL. TABLE 2. Leukocyte counts in relation to macroscopic purulence of sinus secretions % of secretions in each leukocyte count category PMN/OIF Aspirates (n = 240) Injection aspirates (n = 95) No. (%a) Purulent Nonpurulent Bloody No. (%") Purulent Nonpurulent Bloody Group 1 (>20) 156 (65) 86 il 3 31(33) Group 2 (6-20) 22 (9) (8) Group 3 (1-5) 20 (8) (15) Group 4 (<1) 42 (18) (44) a Percentage of all aspirates. b Percentage of all injection aspirates. of culture findings from the 156 aspirates containing high When the culture results were analyzed in relation to the numbers of leukocytes very closely resembled the frequen- leukocyte counts of the secretions (Fig. 2), it was found that cies of findings from the macroscopically purulent aspirates the majority of all aspirates in which a sinus pathogen grew (Table 4), with H. influenza and Streptococcus pneumoniae had >20 PMNs/OIF. The actual frequencies were very being found most often (Table 3). A high proportion (64%; 14 similar to those seen in Fig. 1 for macroscopically purulent of 22) of the aspirates containing 6 to 20 PMNs/OIF (group 2) aspirates: 71% for H. influenza, 92% for Streptococcus yielded H. influenza, whereas Streptococcus pneumoniae pneumoniae, 87% for Streptococcus pyogenes, 75% for B. (5%), Streptococcus pyogenes (9%), B. catarrhalis (5%), and catarrhalis, and 100% for the anaerobes. The majority 66% anaerobes (0%) were found considerably less often (P < (33 of 53) of the culture-negative specimens contained only a 0.01). Other bacterial species were cultured from the group 2 few (1 to 5) or very few (<1) leukocytes per OIF. specimens with relatively high frequencies (27%; 6 of 22). All Among the injection aspirates, a somewhat larger proporexcept two of these isolates were present in small quantities tion of both H. influenza and Streptococcus pneumoniae (<103 CFU/ml). The exceptions were two specimens with that were found originated from specimens containing.20 >103 CFU of Staphylococcus aureus per ml: one in pure PMNs/OIF. culture and the other together with Streptococcus faecalis. Culture results versus bacterial morphotypes. Of the 95 A sinus pathogen was isolated from three other specimens aspirates that were positive for H. influenza in quantities of along with other presumed nonpathogens. H. influenza was >103 CFU/ml, the corresponding morphotype (gram-negstill a common isolate (45%) in the aspirates containing 1 to ative rods) was seen in the smear of 85 aspirates (89%), while 5 PMNs/OIF (group 3). Of the other sinus pathogens, only 10 aspirates (11%) were negative (Table 5). More than 90% Streptococcus pneumoniae was isolated in 5% (1 of 20) of of the aspirates with positive smears contained >20 PMNs/ the group 3 specimens. One-half of the specimens remained OIF. In 77 aspirates, gram-negative rods were the only negative after culture, and altogether 35% of the specimens morphotype, whereas in 8 aspirates another morphotype were positive for sinus pathogens in quantities exceeding 103 was also present. This was a gram-positive diplococcus in CFU/ml. A similar distribution pattern of culture findings six smears and a gram-positive coccus in chains in one was also encountered in aspirates containing <1 PMN/OIF smear; this corresponded to the culture finding of Strepto- (group 4), but with still fewer specimens being positive for coccuts pneumoniae or Streptococcus pyogenes, respecsinus pathogens in quantities of >103 CFU/ml. tively, along with H. influenzae. In one specimen the smear The distribution of the culture findings in the injection contained gram-negative diplococci, but a corresponding isoaspirates resembled that in the aspirates. late was not recovered; the culture grew H. influenza only. TABLE 3. Macroscopic purulence in relation to culture findings of sinus secretions % of specimens yielding the indicated culture result in each macroscopic category" One or more pathogensd Macroscopic purulence Haemophilus Streptococcus Stireptococcis Brantharella Anaerobesb Other' Negative present at: influenzae pneumoniae pyogenes catarrhalis >îo3 CFU/ml Aspirates (n = 240) Purulent (n = 147) Nonpurulent (n = 69) Bloody (n = 24) Injection aspirates (n = 95) Purulent (n = 23) Nonpurulent (n = 55) 29 il Bloody (n = 17) a The sum of percentages is >100%, because some specimens contained multiple isolates, e.g., two pathogens or a pathogen together with other bacteria. <îo3 CFU/ml b Multiple species per specimen counted as one isolate; this includes only patients considered to have true anaerobic infections (see Table 5 in the accompanying report [14]). ' Other includes staphylococci, streptococci, gram-negative rods, Propionibacterium acnes, Bacteroides asaccharolyticus, and Peptostreptococcus magnus (see Tables 3 and 5 in the accompanying report [14]); multiple species per specimen counted as one isolate. d Haemophilus influenza, Streptococcus pneurmoniae, Streptococcufs pyogenes, Branhamella catarrhalis, and anaerobes.

4 VOL. 26, 1988 PURULENCE AND MICROSCOPIC FINDINGS IN ACUTE SINUSITIS r p cn (n 0 4- (0 O o (O z 50 z 50 p CD i N SD.e! Bs 0 c O I C cd<d O& O Z FIG. 1. Culture results in relation to macroscopic purulence of sinus secretions. In each column the bottom part indicates organisms derived from purulent secretions, the middle part indicates organisms from bloody secretions, and the top part indicates organisms from nonpurulent secretions. Likewise, a morphotype concordant with the culture results was present in 90% of the 48 aspirates yielding Streptococcus pneumoniae and in 93% of the 14 aspirates yielding Streptococcus pyogenes. The few isolates of B. catarrhalis (two aspirates) and anaerobes (three aspirates with mixed bacterial species) were all detected in the smears. Staphylococci were isolated in quantities of >10' CFU/ml from three aspirates only. In the smears of two of these, TABLE 4. n P3 Fln._. Ii I c> 0 O C D > C î-?c CdC 0Q FIG. 2. Culture results in relation to leukocyte counts of sinus secretions. In each column the bottom part indicates organisms derived from samples with >20 PMNs/OIF, the second part indicates organisms from samples with 6 to 20 PMNs/OIF, the third part indicates organisms from samples with 1 to 5 PMNs/OIF, and the top part indicates organisms from samples with <1 PMN/OIF. gram-positive cocci were seen; in one case Staphylococcus aureus was present in pure culture and in the other Staphylococcius aureus was present with Streptococcusfaecalis. Of the three specimens that yielded other bacterial species by culture, the smear of one revealed gram-negative rods in pairs only and the culture grew Moraxella sp. along with H. influenzae; the other showed gram-positive cocci in chains, and the culture was positive for Streptococcus anginosus Leukocyte counts in smears in relation to culture findings of sinus secretions % Culture findings in each leukocyte count category' One or more Microscopic purulence pathogens (PMN/OIF) Haernophilus Streptcoccus Steptococcus Branhainella Anaerobes'> Other' Negative present at: ijluenzae plneumoniae pyogenes catarrhalis >103 <103 CFU/ml CFU/ml Aspirates (n = 240) Group 1 (>20; n =156) Group 2 (6-20; n =22) Group 3 (1-5; n = 20) Group 4 (1; n = 42) Injection aspirates (n = 95) Group 1 (>20; n =31) Group 2 (6-20; n =8) Group 3 (1-5; n = 14) Group (1; n =42) " See footnote a to Table 3. b See footnote b to Table 3. See footnote c to Table 3. d See footnote d to Table 3.

5 1930 JOUSIMIES-SOMER ET AL. J. CLIN. MICROBIOL. TABLE 5. Culture results in relation to bacterial morphotypes in Gram-stained smears of sinus secretions No. in aspirates Smear No. in injection aspirates Count and organism Corresponding No corresponding Corresponding No corresponding Culture morphotype morphotype Culture morphotype morphotype Smear Alone combination Other only Negative Alone In Other only Negative obinain combination O >103 CFU/ml by culture Haemophilus ' b 8 influenza Streptococcus pneumoniae Streptococcus pyogenes Branhamella catarrhalis Anaerobes 3 3' 2 i1 Staphylococci 3 2` 1 0 Other <10i CFU/ml by culture Haemophilus 29 il influenza Streptococcus pneumoniae Streptococcus pyogenes Branhamella catarrhalis Anaerobes 0 0 Staphylococci 15 if Other Culture negative a Includes one smear positive for both gram-negative rods and gram-negative diplococci; the culture was positive only for Haemophilus influenza. b Includes one smear positive for both gram-negative rods and gram-positive diplococci; the culture was positive only for Haemophilus influenza. C Includes three specimens all containing several gram-positive and gram-negative anaerobic species and one with Streptococcus anginosus as well. d Morphologically indistinguishable gram-negative rods were seen; the culture was positive for Haernophilus influenza, Bacteroides oralis, and Bacteroides intermedius. e Gram-positive cocci seen in both smears; one was culture positive for Staphylococcus aureus alone; the other was positive for Staphylococcus aureus and Streptococcus faecalis. f Gram-negative rods seen in smear, but the culture remained positive just for coagulase-negative staphylococci after enrichment. 9 Gram-negative rods seen in smear; the culture was positive only for Staphylococcus aureus after enrichment. h Gram-negative rods seen in smear, but the culture remained negative. and several anaerobic species. The smear of one aspirate was negative, but the culture grew Propionibacterium acnes. In those 59 aspirates that yielded only scanty growth (<103 CFU/ml), the smears were most often (78%) negative. However, in 13 aspirates the smear was positive and in 12 of them the morphotype corresponded to the isolate: H. influenzae in 11 smears and B. catarrhalis in 1 smear. All these aspirates except one (yielding H. influenza) contained >20 PMNs/OIF. In one case the culture yielded coagulasenegative staphylococci, but the smear was positive for gram-negative rods. The smears of 52 of 53 (98%) culturenegative aspirates were also negative, and in only 1 case gram-negative rods were seen in the smear; this aspirate contained >20 PMNs/OIF. Results for injection aspirates mainly followed the pattern obtained with the aspirates, but the number of positive smears was smaller. Thus, of the 27 injection aspirates that were positive for H. influenza and 12 injection aspirates that were positive for Streptococcus pneumoniae in quantities of >10i CFU/ml, the corresponding morphotypes were seen in 70 and 83% of the smears, respectively. In one case the culture yielded only H. influenzae, but the morphotype, a gram-positive diplococcus, present in the smear was not cultured. In those 48 injection aspirates that yielded only scanty growth (<103 CFU/ml), the smears were most often (83%) negative. However, in eight injection aspirates the smear was positive, and in seven of them the morphotype corresponded to the isolate: H. influenza in four and Streptococcus pneumoniae, coagulase-negative staphylococci, and P. acnes in one each. In one case the culture yielded Staphylococcus aureus after enrichment, but the smear was positive for gram-negative rods only. All the smears of the 28 culture-negative injection aspirates were negative. In summary, when bacteria were cultured in quantities of >10i CFU/ml from the aspirates, the corresponding morphotype was detected in the smears in 89% of the aspirates (sensitivity of the gram-stained smear). The positive smears usually contained high numbers of bacteria in each OIF, corresponding to 107 to 108 organisms per ml. The corresponding sensitivity of the smear was only 20% when the culture was positive in quantities of <103 CFU/ml. When all the aspirates were combined, irrespective of the culture finding, a bacterial morphotype was identified in the smears

6 VOL. 26, 1988 PURULENCE AND MICROSCOPIC FINDINGS IN ACUTE SINUSITIS times. A corresponding culture finding was obtained 162 times, giving an estimated positive predictive value of 98%. Of these 162 culture findings, 150 (93%) represented cultures that exceeded >103 CFU/ml. There were some differences between the individual bacteria. For example, for H. influenzae the positive predictive`value of the smear was 98%, and the negative predictive value was 80%, while the corresponding values for Streptococcus pneumoniae were 100 and 96%. When the injection aspirates yielded isolates in quantities of >103 or <103 CFU/ml, the corresponding sensitivities of the Gram-stained smears in detecting concordant morphotypes were 76 and 15%. Both of these values were significantly (P < 0.01) smaller than the corresponding values for the aspirates. A bacterial morphotype was identified in these smears 43 times. A concordant culture finding was obtained 41 times, giving an estimated positive predictive value of 95%. Of these 41 culture findings, 34 (83%) occurred in quantities of >103 CFU/ml. For H. influenza the positive predictive value of the smear in the injection aspirates was 96% and the negative predictive value was 76%. The corresponding values for Streptococcus pneumoniae were 92 and 95%. DISCUSSION In this study macroscopic purulence of sinus secretions from patients with AMS predicted well (-80% in each case) a high leukocyte count (>20 PMNs/OIF) in the smear and the isolation of presumed sinus pathogens in quantities of >103 CFU/ml (positive culture). In addition, however, some macroscopically nonpurulent specimens had a high leukocyte count (25% of the aspirates and 18% of the injection aspirates) and a positive culture (29 and 27%, respectively). The frequency of positive cultures was the highest (79%) in the aspirates with the highest leukocyte counts and decreased gradually to 14% in the aspirates with very few leukocytes. H. influenza and Streptococcus pneumoniae were the most common pathogens isolated (63 and 26% of all pathogens in the aspirates, respectively). However, H. influenza was relatively more common in the nonpurulent aspirates (83% of the pathogens compared with 59% in the purulent aspirates; P < 0.01), and Streptococcus pneumoniae was more common in the aspirates with >20 PMNs/OIF (30% of the pathogens, compared with 9% in the aspirates with fewer leukocytes; P < 0.01). The presence of bacteria in the Gram-stained smear predicted with a high accuracy (.95%) the isolation of a corresponding species, usually in quantities of >103 CFU/ ml, whereas a pathogen was found by culture in only few specimens with a negative smear (11% of the positive cultures). From the beginning of this study it was evident that macroscopic classification of sinus secretions, as described by Carenfelt and Lundberg (5), as just purulent and nonpurulent causes problems in certain cases. Secretions that were grossly contaminated with blood were impossible to assign to either group and were hence classified as bloody. Furthermore, scanty specimens (<0.1 ml), several saline-diluted injection aspirates, and frothy specimens were also difficult to evaluate and were often classified as nonpurulent. The distribution of the aspirates to macroscopically purulent (61%), nonpurulent (29%), and bloody (10%) categories resembled the distribution of sinus secretions from children with AMS reported by Wald et al. (26). A higher percentage of the injection aspirates (58%) were classified as nonpurulent. The lack of visible purulence in them may have been partially caused by dilution of the secretion with the injected saline and in some cases to failure of the saline to mobilize the secretion. The injection aspirates mostly behaved as diluted aspirates. The following discussion therefore concentrates on the results of the aspirates, with the special features of the injection aspirates summarized at the end of this report. We used Gram-stained smears to determine the number of leukocytes in the secretions, since such smears are routinely prepared in laboratories in which pus specimens are examined. The secretions were divided into four groups based on leukocyte counts (Table 1). Macroscopic purulence of the aspirates predicted very well a high leukocyte count (>20 PMNs/OIF) in the smear (in 92% of the aspirates). In fact, 76% of the purulent aspirates contained >100 PMNs/OIF. It was notable, however, that some macroscopically nonpurulent (25%) and bloody (17%) aspirates also contained high numbers of leukocytes. Macroscopically purulent secretions have been almost exclusively associated with high numbers of leukocytes, varying from 104 to 106 cells per mm3, as determined in counting chambers, while nonpurulent secretions have contained <101 to 103 cells per mm3 (6, 8). As 104 cells per mm3 corresponds to an average of 14 PMNs/OIF (13), our results are in good agreement with those of previous studies (6, 8). A high proportion (approximately 80%) of the aspirates judged to be purulent either macroscopically or based on a high leukocyte count yielded presumed sinus pathogens in quantities exceeding 103 CFU/ml (positive culture). The majority of Streptococcus pneumoniae isolates came from purulent samples (84% from macroscopically purulent aspirates and 92% from samples with >20 PMNs/OIF). H. influenza was less strictly associated with purulence (68 and 71%, respectively). A total of 87% of Streptococcus pyogenes, 75% of B. catarrhalis, and 100% of the anaerobic isolates also came from samples with >20 PMNs/OIF. On the other hand, positive cultures were also obtained from 29% of the macroscopically nonpurulent and 50% of the bloody secretions. Likewise the frequencies of positive cultures decreased gradually with decreasing leukocyte counts (79, 64, 35, and 14% in leukocyte groups 1 to 4, respectively). The predominant pathogen in the less purulent aspirates was H. influenza (79% of positive cultures in leukocyte groups 2 to 4 compared with 55% of positive cultures in group 1). In previous studies an even higher proportion (>90%) of pathogenic bacteria was found in macroscopically or microscopically purulent secretions (5-8, 10). Evans et al. (10) have demonstrated that leukocyte counts of >5,000/mm3 in 13 aspirates from patients with AMS were, without exception, associated with the isolation of bacteria (at 105 CFU/ ml), a virus (2 aspirates), or a fungus (2 aspirates). On the other hand, the 19 aspirates from patients with AMS that contained <1,000 leukocytes per mm3 were either sterile or contained <103 bacteria per ml, which were considered as probable contaminants. In the present study, in which the number of samples studied was much larger than in those discussed above, we also found an overall correlation between culture findings and purulence. However, 10% of the aspirates with high or moderate numbers of leukocytes (groups 1 and 2, corresponding to >3,600 leukocytes per mm3) remained negative

7 1932 JOUSIMIES-SOMER ET AL. by culture, and 14% of the aspirates containing less than 700 leukocytes per mm3 (group 4) yielded pathogens in quantities exceeding 103 CFU/ml. The nonpurulent, pathogen-positive samples may reflect an earlier phase of sinusitis in our patient population (army recruits), which is likely because of their easy access to medical services. Purulent AMS is generally seen as a later phase of sinusitis; this is preceded by serous and mucopurulent phases (4, 9, 17; G. B. Healy, Editorial, N. Engl. J. Med. 304: , 1981). Interestingly, H. influenza, which was a more common finding in our material (14), was isolated relatively often from the nonpurulent aspirates. By contrast, Streptococcus pneumoniae, the major pathogen in most other studies (5-7), was more strictly confined to the purulent secretions. The culture-negative purulent aspirates might have contained viruses, mycoplasmas, chlamydia, or other microbes that were not cultivable by the methods we used in this study. The presence of a common sinus pathogen species was unlikely, since only once were bacteria (resembling H. influenza) seen in the Gram-stained smear of a culturenegative aspirate. Treatment with antimicrobial agents or prolonged transport of the specimens also could not account for the negative cultures, since none of our patients received therapy with antimicrobial agents within the 2 weeks before the study and the transport and processing of the specimens was exceptionally rapid (14). When bacteria were seen in the Gram-stained smear, a pathogen corresponding to the morphotype was isolated in quantities exceeding 103 CFU/ml in 150 of 163 (92%) aspirates. When the smear was negative, the correlation with the culture results was not as good; 18 positive cultures (10%) came from aspirates with negative smears, which was not surprising, considering the detection limit of the Gramstained smear (104 to 105 organisms per ml). On the other hand, the positive cultures classified here as those containing >103 CFU/ml, as described by Evans et al. (10), often (84%) had >i04 CFU/ml (14). The relatively high percentage (89%) of positive smears corresponding to the isolation of H. influenzae was probably achieved as the result of the fact that we paid special attention to faintly staining, small, gram-negative rods. In our earlier studies, these morphotypes were easily lost in the pink background, in particular; in the presence of numerous leukocytes. In fact, a positive smear was even found in 11 of the 29 aspirates (38%), yielding only scanty growth (<103 CFU/ml) of H. influenza. Of these 11 smears, 10 also had high numbers of leukocytes, supporting the pathogenic role of H. influenza. Published reports of studies correlating bacteria seen in Gram-stained smears with culture findings of sinus secretions are scarce or are based on a small number of samples (5, 26). Urdal and Berdal (21) have recorded bacterial morphotypes in smears of samples from 81 patients with sinusitis. They found concordant results for H. influenza in 88% of smears, for Streptococcus pneumoniae in 91%, for Streptococcus pyogenes in 100%, and for anaerobes in 100%; these results are closely similar to those of this study. We found staphylococci by culture in 7% of the aspirates, but usually in small numbers, suggesting that they were derived from contamination ofthe samples (14). The absence of the corresponding morphotype in the smear (Table 5) further supports this view. However, in one sample the culture yielded >103 CFU of Staphylococcus aureus per ml in pure culture, and the smear was positive, suggesting a possible pathogenic role for Staphylococcus aureus in this case. J. CLIN. MICROBIOL. As we pointed out above, the injection aspirates behaved to a large extent like diluted aspirates. However, a major difference was the more frequent isolation of bacteria other than the presumed sinus pathogens (Tables 3 and 4). They were usually isolated in small numbers (<103 CFU/ml), often together with a sinus pathogen that grew at >103 CFU/ml. We believe that they represent contamination associated with the sampling method. Even with this limitation, examination of injection aspirates should not be discouraged: 23% of all findings of a sinus pathogen came from them (14). ACKNOWLEDGMENTS This study was supported in part by a grant from the Paulo Foundation. We thank Inga Laukkonen, Marja Piekkola, Arja Kanervo, and Paula Summanen for excellent technical assistance and the staff of the Ear, Nose, and Throat Department, Central Military Hospital, for skillful assistance and Kaija Helisjoki for typing the manuscript. Our thanks are given to P. H. Makela, chief of the Bacteriological Department, National Public Health Institute, Helsinki, for constructive criticism and valuable help with preparing the manuscript. LITERATURE CITED 1. Axelsson, A., and J. E. Brorson The correlation between bacteriological findings in the nose and maxillary sinus in acute maxillary sinusitis. Laryngoscope 83: Bridger, R. C Sinusitis: an improved regimen of investigation for the clinical laboratory. J. Clin. Pathol. 33: Brorson, J.-E., and S. E. Holm The role of Branhamella catarrhalis, p In P. van Cauwenberge and C. Ekedahl (ed.), Advances in sinusitis. Scientific Society for Medical Information, Gent, Belgium. 4. Carenfelt, C Pathogenesis of sinus empyema. Ann. Otolaryngol. 88: Carenfelt, C., and C. Lundberg Purulent and non-purulent maxillary sinus secretions with respect to P02, pco2, and ph. Acta Otolaryngol. 84: Carenfelt, C., and C. Lundberg The role of local gas composition in pathogenesis of maxillary sinus empyema. Acta Otolaryngol. 85: Carenfelt, C., C. Lundberg, C.-E. Nord, and B. Wretlind Bacteriology of maxillary sinusitis in relation to quality of the retained secretion. Acta Otolaryngol. 86: Engquist, S., and C. Lundberg Bacteria and inflammatory cells in maxillary sinusitis. Arch. Otorhinolaryngol. 239: Engquist, S., and C. Lundberg Akut sinuit-nar, hur och av vem bor den behandlas? Lakartidningen 83: (In Swedish.) 10. Evans, F. O., Jr., J. B. Sydnor, W. E. C. Moore, G. R. Moore, J. L. Manwaring, A. H. Brill, R. T. Jackson, S. Hanna, J. S. Skaar, L. V. Holdeman, G. S. Fitz-Hugh, M. A. Sande, and J. M. Gwaltney, Jr Sinusitis of the maxillary antrum. N. Engl. J. Med. 293: Gwaltney, J. M., Jr., A. Sydnor, Jr., and M. A. Sande Etiology and antimicrobial treatment of acute sinusitis. Ann. Otol. Rhinol. Laryngol. 90(Suppl. 84): Hamory, B. H., M. A. Sande, A. Sydnor, Jr., D. L. Seale, and J. M. Gwaltney, Jr Etiology and antimicrobial therapy of acute maxillary sinusitis. J. Infect. Dis. 139: Holdeman, L. V., E. P. Cato, and W. E. C. Moore (ed.) Anaerobe laboratory manual, 4th ed., p Virginia Polytechnic Institute and State University, Blacksburg. 14. Jousimies-Somer, H. R., S. Savolainen, and J. S. Ylikoski Bacteriological findings of acute maxillary sinusitis in young adults. J. Clin. Microbiol. 26: Kortekangas, A. E Antibiotics in the treatment of maxillary sinusitis. Acta Otolaryngol. 188(Suppl.): Lennette, E. H., A. Balows, W. J. Hausler, Jr., and J. P. Truant (ed.) Manual of clinical microbiology, 3rd ed. American Society for Microbiology, Washington, D.C.

8 VOL. 26, 1988 PURULENCE AND MICROSCOPIC FINDINGS IN ACUTE SINUSITIS Ludman, H ,BC of ENT. Paranasal sinus diseases. Br. Med. J. 282: Lundberg, C., C. Carenfeit, S. Engquist, and C.-E. Nord Anaerobic bacteria in maxillary sinusitis. Scand. J. Infect. Dis. 19(Suppl.): Lystad, A., P. Berdal, and L. Lund-Iversen The bacterial flora of sinusitis with an in vitro study of the bacterial resistance to antibiotics. Acta Otolaryngol. 188(Suppi.): Mann, W., K. Peiz, W. Schlenter, and W. Niebling The role of streptococci and Haemophilus in sinusitis, p In P. van Cauwenberge and C. Ekedahl (ed.), Advances in sinusitis. Scientific Society for Medical Information, Gent, Belgium. 21. Urdal, K., and P. Berdal The microbial flora in 81 cases of maxillary sinusitis. Acta Otolaryngol. 37: van Cauwenberge, P., P. Kluyskens, and L. van Renterghem The importance of the anaerobic bacteria in paranasal sinusitis. Rhinology 13: van Cauwenberge, P., G. Verschraegen, and L. van Renterghem Bacteriological findings in sinusitis ( ). Scand. J. Infect. Dis. 9(Suppl.): Vecchio, T. J Predictive values of a single diagnostic test in unselected populations. N. Engl. J. Med. 274: Verschraegen, G., and P. van Cauwenberge The role of anaerobes in sinusitis, p In P. van Cauwenberge and C. Ekedahl (ed.), Advances in sinusitis. Scientific Society for Medical Information, Gent, Belgium. 26. Wald, E. R., G. J. Milmoe, A. Bowen, J. Ledesma-Medina, N. Salamon, and C. C. Bluestone Acute maxillary sinusitis in children. N. Engl. J. Med. 304: Wald, E. R., J. S. Reilly, M. Casseibrant, J. Ledesma-Medina, J. Milmoe, C. D. Bluestone, and D. Chiponis Treatment of acute maxillary sinusitis in childhood: a comparative study of amoxicillin and cefaclor. J. Pediatr. 104: