Utility of gram stain of endotracheal aspirates onempiric therapy in children with hospital-acquiredpneumonia*

Dear Editor,

We read “Gram stain/aolc screening for detection ofcatheter related bloodstream infection”1 with great inter-est which demonstrated that Gram/AOLC screening hasmodest sensitivity and high specificity for the diagnosis ofcatheter-related bloodstream infection. Early and promptdiagnosis of healthcare-associated infections (HAI) is im-portant because it leads to appropriate antimicrobial se-lections and decrease the morbidity and mortality ofpatients.

To evaluate the other common HAI, hospital-acquiredpneumonia (HAP), we investigated the usefulness of Gramstain of endotracheal aspirates (ETA) for the early di-agnosis of HAP in children to select the appropriate initialantibiotic therapy. Guidelines for the management of HAPin adults emphasize that broad-spectrum combinationantibiotic therapy (two different types of antipseudomo-nal agents, and vancomycin or linezolid) should beempirically initiated in patients with risk factors formultidrug-resistant organisms including Pseudomonasaeruginosa, Acinetobacter species, Extended-spectrumb-lactamase producing Klebsiella pneumoniae andmethicillin-resistant Staphylococcus aureus.2 Whereasthe overuse of broad-spectrum antibiotics can lead to in-creased hospital expenditures, while also potentially pro-moting antibiotic resistance.3 Gram stain of sputum hasbeen used for the early diagnosis of pneumonia and theselection of initial antibiotic therapy in adults4; however,the usefulness of Gram stain in children has not been re-ported except for one report in extremely pretermneonates.5

The study subjects included 53 children diagnosed withHAP at the largest pediatric hospital in Japan, betweenApril 2009 and February 2011. Pneumonia was defined bythe CDC criteria, including radiologic, clinical, andlaboratory findings.6 HAP is defined as pneumonia that oc-curs �48 h after admission. We selected patients whowere qualified to receive broad-spectrum combinationantibiotics based on the risk factors for multidrug-resistant organisms according to the adult guidelines2;the risk factors are antimicrobial therapy in preceding90 days, current hospitalization of �5 days, hospitaliza-tion for �2 days in the preceding 90 days, chronic dialysiswithin 30 days, and immunosuppressive disorders and/ortherapies.

All patients with suspected HAP on mechanicalventilation (MV) or tracheostomy were evaluated bycollecting ETA. All Gram stain specimens were examinedfor the number of polymorphonuclear neutrophils andsquamous epithelial cells with Geckler classifications.Nurses, who submitted the sputum specimens otherthan Geckler 5 classification, were suggested to resub-mit the other sputum specimens. Data was collectedfrom only children whose Gram stain findings wereevaluated as Geckler 5. There were six samples

classified as Geckler 4 or 6 which were excluded fromthe study. Bacterial counts were recorded as: þ1, þ2,þ3 and þ4 were <1 bacterium, 1, 2e10 and >10 notedper high-powered field, respectively. The criteria ofdominant organism were utilized when one or two typesof morphology are seen on Gram stain. Bacterial growthfrom sputum culture was semiquantified as few, 1þ, 2þand 3þ. By comparing with the quantitative culturemethod, bacteria semiquantified as 1þ, 2þ and 3þcorresponded to at least 104, 106 and 108 cfu/ml, re-spectively. The causative organisms were consideredto be “definitive” when one or more pathogenic organ-isms were isolated from ETA at concentrations of �þ2and those organisms were consistent with the clinicalcourses of the patients.

Empiric antibiotic therapies for HAP were initiatedbased on the results of Gram stain. The use ofvancomycin was considered when Gram-positive cocciwere dominantly observed (Group 1). One or twoantipseudomonal agents were initiated when Gram-negative bacilli (GNB) were observed with consider-ation of the patients’ background information (Group3). Gram-negative cocci were mainly observed in othercases, and a narrow-spectrum antibiotic was selected(Group 2).

The majority of HAP (77%) cases occurred while thepatients were on MV, and half of these required MV for�10 days (Table 1). Among the causative organisms(n Z 68), multidrug-resistant organisms were responsiblefor 21% of HAP cases (Fig. 1). Moraxella catarrhalis (37%)was the most common organism, followed by Haemophilusinfluenzae (19%), P. aeruginosa (16%). S. aureus was iso-lated (9%), and six percent was methicillin-susceptibleS. aureus and three percent was methicillin-resistantS. aureus.

Gram stain of ETA was highly sensitive (92%) and specific(81%) to identify causative organisms, and the specificity(97%) was enhanced when dominant organisms wereobserved.

The current study results showed that Gram stain ofETA was useful for predicting the causative organisms ofHAP in children if stringent criteria were fulfilled usingGeckler classifications. The present findings of Gram stainare superior to other studies in adults (sensitivity 62e91%and specificity 64e74%).7e9 The most probable reason ofthe difference was the use of stringent criteria for theGram stain in the current study. We only selected pa-tients with HAP, who were qualified to receive broad-spectrum combination antibiotics according to the riskfactors for multi-drug resistant bacteria based on theadult guideline.2 All antibiotic selections were deter-mined based on the results of Gram stain findings. As a re-sult, vancomycin or anti-pseudomonal agents wereselected initially in only 40% of patients and no patientreceived both antipseudomonal agents and vancomycinor linezolid; whereas, the selection of antibiotics was ap-propriate in 92% of the patients on the basis of drug sen-sitivity results. These favorable results are consistentwith previous reports in adults7 and in extremely pretermneonates.5

In conclusion, this study showed that Gram stain ofETA based on stringent criteria is useful for predicting

368 Letters to the Editor

the causative organisms of HAP and selecting appropri-ate narrow-spectrum antibiotics in children. Furtherprospective, randomized studies are necessary to eval-uate the usefulness of Gram stain of ETA in childrenwith HAP.

Financial disclosure

None.

Contributor’s statement

Chikara Ogimi: Dr. Ogimi conceptualized and designed thestudy, drafted the initial manuscript, and approved thefinal manuscript as submitted.

Kensuke Shoji: Dr. Shoji conceptualized and designedthe study, acquired data, revised the article critically for

important intellectual content, and approved the finalmanuscript as submitted.

Tomohiro Katsuta: Dr. Katsuta conceptualized and de-signed the study, acquired data, revised the article criti-cally for important intellectual content, and approved thefinal manuscript as submitted.

Yasushi Watanabe: Mr. Watanabe designed the study,coordinated and supervised data collection, revised thearticle critically for important intellectual content, andapproved the final manuscript as submitted.

Akihiko Saitoh: Dr. Saitoh conceptualized and designedthe study, coordinated and supervised analysis and in-terpretation of data, reviewed and revised the manuscripttotally, and approved the final manuscript as submitted.

References

1. Zach J, Zollner-Schwetz I, Krause R, Zebisch A, Rohn A, Sill H.Gram stain/aolc screening for detection of catheter relatedbloodstream infection. J Infect 2010;60(3):254e5.

2. Guidelines for the management of adults with hospital-acquired, ventilator-associated, and healthcare-associatedpneumonia. Am J Respir Crit Care Med 2005;171(4):388e416.

3. Trouillet JL, Chastre J, Vuagnat A, Joly-Guillou ML, Combaux D,Dombret MC, et al. Ventilator-associated pneumonia caused bypotentially drug-resistant bacteria. Am J Respir Crit Care Med1998;157(2):531e9.

4. Rea-Neto A, Youssef NC, Tuche F, Brunkhorst F, Ranieri VM,Reinhart K, et al. Diagnosis of ventilator-associated pneumonia:a systematic review of the literature. Crit Care 2008;12(2):R56.

5. Katayama Y, Minami H, Enomoto M, Takano T, Hayashi S, Lee YK.Usefulness of gram staining of tracheal aspirates in initialtherapy for ventilator-associated pneumonia in extremely pre-term neonates. J Perinatol 2010;30(4):270e4.

6. Horan TC, Andrus M, Dudeck MA. CDC/NHSN surveillance defini-tion of health care-associated infection and criteria for specific

Other Gram negative bacilli 7%

Staphylococcus aureus 9%

Klebsiella pneumoniae 3%

Other Gram-negative bacilli 7%

Moraxella catarrhalis 37%

Staphylococcus aureus 9%

iae 3%

%%

Streptococcus pneumoniae 9%

Pseudomonas aeruginosa 16%

Haemophilus influenza 19%

Figure 1 Causative organisms of Hospital-acquired pneumo-nia based on the semiquantitative culture results (n Z 68).Culture-identified organisms were recognized as causative or-ganisms when organisms were isolated from ETA at concentra-tions of �þ2 and those organisms were consistent with theclinical courses of patients.

Table 1 Baseline characteristics.

Characteristics All patients(n Z 53)

Group 1(n Z 7)

Group 2(n Z 11)

Group3(n Z 35)

Gender, male 26 (49%) 2 (29%) 4 (36%) 20 (57%)Age, year (median) 2.80 0.81 2.63 3.24

<1 14 (26%) 4 (57%) 4 (36%) 6 (17%)1e12 29 (55%) 3 (43%) 7 (64%) 19 (54%)�13 10 (19%) 0 0 10 (29%)

Length of hospital stay, �10 days 35 (66%) 6 (86%) 5 (45%) 24 (69%)Ventilator-associated pneumonia 41 (77%) 7 (100%) 9 (82%) 25 (71%)

Duration of mechanical ventilation, �10 days 22/41 (54%) 5/7 (71%) 2/9 (22%) 15/25 (60%)Tracheostomy 24 (45%) 0 5 (45%) 19 (54%)Admission to Intensive Care Unit 35 (66%) 7 (100%) 8 (73%) 20 (57%)Underlying diseases

Neuromuscular diseases 25 (47%) 3 (43%) 5 (45%) 17 (49%)Gastrointestinal diseases 17 (32%) 0 4 (36%) 13 (37%)Anatomical abnormalities in respiratory system 11 (21%) 1 (14%) 2 (18%) 8 (23%)Patients with immunosuppressive agents 11 (21%) 0 2 (18%) 9 (26%)Chromosomal abnormalities 10 (19%) 1 (14%) 3 (27%) 6 (17%)Heart diseases 8 (15%) 1 (14%) 3 (27%) 4 (11%)Metabolic diseases 7 (13%) 2 (29%) 0 5 (14%)Others 12 (23%) 1 (14%) 3 (27%) 7(20%)

Recent antibiotic changes within 72 h 13 (25%) 1 (14%) 5 (45%) 7 (20%)

Letters to the Editor 369

types of infections in the acute care setting. Am J Infect Control2008;36(5):309e32.

7. Veinstein A, Brun-Buisson C, Derrode N, Alvarez A, Pinsard M,Robert R, et al. Validation of an algorithm based on direct exam-ination of specimens in suspected ventilator-associated pneu-monia. Intensive Care Med 2006;32(5):676e83.

8. Papazian L, Autillo-Touati A, Thomas P, Bregeon F, Garbe L,Saux P, et al. Diagnosis of ventilator-associated pneumonia: anevaluation of direct examination and presence of intracellularorganisms. Anesthesiology 1997;87(2):268e76.

9. Blot F, RaynardB, ChachatyE, TancredeC, Antoun S,NitenbergG.Value of gram stain examination of lower respiratory tract secre-tions for early diagnosis of nosocomial pneumonia. Am J RespirCrit Care Med 2000;162(5):1731e7.

Chikara Ogimi*Kensuke Shoji

Tomohiro KatsutaDivision of Infectious Diseases,

Department of Medical Specialties,National Center for Child Health and Development,

Tokyo, JapanE-mail address: chikara.ohgimi.510311@gmail.com

(C. Ogimi)

Yasushi WatanabeDivision of Microbiology,

National Center for Child Health and Development,Tokyo, Japan

Akihiko SaitohDivision of Infectious Diseases,

Department of Medical Specialties,National Center for Child Health and Development,

Tokyo, Japan

*Corresponding author. Pediatric Resident, ElmhurstHospital Center, 79-01 Broadway, Elmhurst, NY 11373,

United States. Tel.: þ1 718 334 2156; fax: þ1 718 334 2862.

Accepted 9 June 2012

Available online 18 June 2012

*This was presented in part at the 49th Infectious Diseases Societyof America Annual Meeting in October, 2011 in Boston, MA, USA.

ª 2012 The British Infection Association. Published by Elsevier Ltd.All rights reserved.

http://dx.doi.org/10.1016/j.jinf.2012.06.002

CD4D T-lymphocytopenia in HIV-negative tuberculosispatients in Sudan

As Wang already stated population frequencies of CD4þT cells in tuberculosis (TB) patients vary significantly indifferent studies.1 CD4þ T lymphocytes are important inthe protective immunity to TB, since they prevent re-acti-vation of the infection through cell-mediated responses.In humans, the importance of CD4þ T lymphocytes in thecontrol of TB can be demonstrated by the high increased

susceptibility to both acute and reactivated TB in HIV-in-fected individuals.2 In the past it has been demonstratedthat CD4þ T-lymphocytopenia is also found in patientswith severe pulmonary tuberculosis without evidence ofHIV infection in patients originating from Argentina, SaudiArabia and Botswana,3e5 but no data is available from theSudan, another country with a high TB prevalence. Tuber-culosis patients with CD4þ T-lymphocytopenia usuallyhave a more severe disease with extra pulmonary involve-ment (pleural effusion, adenopathy or military disease)and co-infections associated with low CD4þ counts.2 In Su-dan because of the lack of normal reference ranges for he-matological parameters in healthy subjects, manyinvestigators interpret their data using normal values de-rived from populations in Europe and the United States.6

This range is 450e1350 cell/ml.6 Due to differences in thegenetic, environmental, nutritional, gender, age and atti-tude backgrounds, it has been shown that CD4þ countsare different for each population tested both in the abso-lute and the percent quantities.6 Therefore in order to de-termine if CD4þ T-lymphocytopenia is also common amongHIV-negative tuberculosis patients in Sudan, local CD4þreference counts need to be established first. We per-formed a prospective, cross sectional, observational, de-scriptive case study in 2008e2010 in Abu-Anjaa ChestHospital at Omdurman, Omdurman Teaching Hospital, El-shab Hospital, Academy Hospital, Baharri, Ebrahim Malik,Mayoo Hospital, Sudan. A population comprised of 55healthy unrelated people from the Sudan (Table 1) wasused to determine the normal CD4þ T-cells in the Sudanesepopulation by performing CD4 differential count using flow-cytometry method. Flow-cytometry was performed at Vol-untary Counseling and Testing Centre (VCT) (OmdurmanTeaching Hospital) using Partec Flow-Cytometry instrument(Cyflow�Counter) and the Partec CD4 easy count kit accord-ing to the instructions. As seen in the Tables 1 and 2, ourhealthy controls had lower CD4þ cell counts319e1075 cell/ml (student’s t test, p Z 0.0001) than thisreference range. Our finding of low CD4þ count among con-trols in Sudan adds to emerging data supporting the impor-tance of establishing CD4þ count reference ranges for localpopulations. Differences in CD4þ count among healthy per-sons has been shown in previous studies and has been relatedto a variety of factors including genetics factors, and to dif-ferences in the methodologies used for CD4þ-cell count.5

Published reference ranges for CD4þ count in HIV-negativepopulations from Africa and Asia vary widely.6e13 When wecompare the CD4þ counts found for the healthy controls inour study with a mean CD4þ cell counts of 458.8 cell/ml tothe values reported for healthy adult Dutch population(mean Z 993),10 Swiss (mean Z 691),7Tanzanians(mean Z 843),13 Kuwaitis (mean Z 1050),8 Indians (meanZ 865)12 andChinese (meanZ 785)9 a significant lowerCD4þ count is found in the Sudanese population. The CD4þcount of the Sudanese population was even lower than thatof populations from neighboring countries such as the Ethio-pian population (mean Z 1235)10 and the Ugandan popula-tion (mean Z 1256).11

In order to determine if Sudanese HIV-seronegativetuberculosis patients also suffered from CD4þ T-lymphocy-topenia, CD4þ counts were determined in a population of100 HIV-negative tuberculosis patients (based on the

370 Letters to the Editor