ww.sciencedirect.com

Journal of Hospital Infection 75 (2010) 220–224

Available online at w

Journal of Hospital Infection

journal homepage: www.elsevierheal th.com/journals / jh in

Establishing nurse-led ventilator-associated pneumonia surveillance inpaediatric intensive care

M. Richardson a, S. Hines a, G. Dixon b, L. Highe a, J. Brierley a,*

a Paediatric and Neonatal Intensive Care Unit, Great Ormond Street Hospital for Children NHS Trust, London, UKb Microbiology Department, Great Ormond Street Hospital for Children NHS Trust, London, UK

a r t i c l e i n f o

Article history:Received 26 June 2009Accepted 4 December 2009Available online 14 April 2010

Keywords:Healthcare-associated infectionPaediatric intensive careVentilator-associated pneumonia

* Corresponding author. Address: Consultant PaedCare Unit, Great Ormond Street Hospital for ChildrenLondon WC1N 3JH, UK. Tel.: þ44 2078298889; fax: þ

E-mail address: brierj@gosh.nhs.uk (J. Brierley).

0195-6701/$ – see front matter � 2009 The Hospitaldoi:10.1016/j.jhin.2009.12.015

s u m m a r y

Preventing ventilator-associated pneumonia (VAP) is one of the Department of Health ‘SavingLives’ initiatives. Whereas morbidity and mortality from VAP is well-documented in adults, it ispoorly studied in children. We describe the establishment of a nurse-led VAP surveillanceprogramme as part of an overall drive for patient safety and healthcare-associated infectionreduction. All children admitted to a tertiary referral paediatric intensive care unit over a four-month period were studied. VAP was defined as pneumonia occurring >48 h post intubation.Diagnostic criteria were: (i) radiological: new/progressive infiltrates, consolidation or cavita-tion on chest X-ray; (ii) clinical: �3 of new onset purulent bronchial secretions, leucopaenia orleucocytosis, core temperature �38.5�C or �36�C without other cause, significant positiverespiratory culture or culture from another relevant site of infection. A flow diagram andteaching programme was developed for bedside nurses to facilitate investigations of suspectedVAP. The nurse in charge collected data daily at midnight until 24 h post extubation, dischargeor death. Suspected cases of VAP were referred to infection control for secondary verification. Atotal of 158 intubated children were admitted over four months with 58 excluded (ventilated<24 h). Full data were obtained on all 100 children. VAP incidence was 5.6 per 1000 ventilator-days. We report successful introduction of a nurse-led VAP surveillance programme. Dataacquisition in this study was dependent on nursing workload, however, and placed a signifi-cant time burden on the study leads. Although a relatively low VAP rate was demonstrated,VAP bundles with automated surveillance are being introduced.

� 2009 The Hospital Infection Society. Published by Elsevier Ltd. All rights reserved.

Introduction

Ventilator-associated pneumonia (VAP), the second mostcommon healthcare-associated infection (HCAI) in paediatricintensive care (PIC), accounts for 20% of nosocomial infections.1,2

Defined as pneumonia developing later than 48 h after intubationand initiation of mechanical ventilation, VAP is associated withincreased morbidity and mortality.3 One prospective PIC studyshowed a 20% mortality rate for children with VAP compared with7% in those without.4 Significant morbidity is reported, between 3.7

iatric and Neonatal IntensiveNHS Trust, Great Ormond St,44 2078138206.

Infection Society. Published by Els

and 10.0 additional ventilation days in neonates and children,resulting in prolonged admission and hospital costs.2,4–6

The risk of developing VAP increases with duration of mechanicalventilation and is expressed as infectious episodes per 1000 venti-lator-days. Reported prevalence ranges from 2.9 to 11.6 per 1000ventilator-days although data in children remain limited.4,6,7

An established relationship exists between VAP and aspirationof colonised oropharyngeal secretions.8 Factors contributing tocolonisation include pH-altering drugs that permit overgrowth ofgastric bacteria and feeding tubes that encourage bacterial migra-tion.9 Micro-aspiration of colonised secretions occurs because ofinadequate glottic closure around endotracheal tubes (ETTs) andalso as neuromuscular blocking and sedative agents impair coughand mucociliary clearance, especially in those nursed supine.10,11

Suctioning has also been implicated in VAP through directcontamination due to inadequate hand washing, oro/

evier Ltd. All rights reserved.

M. Richardson et al. / Journal of Hospital Infection 75 (2010) 220–224 221

nasopharyngeal suctioning followed by ETT suction and mucosaltrauma from deep suctioning.12 ETT colonisation occurs early afterintubation and theoretically suctioning might dislodge bacteria intolower airways.13 Significantly, colonisation occurs first in theoropharynx and stomach, followed by lower airways and finally theETTs. This suggests that aspiration rather than contaminatedequipment may be implicated in VAP.8,14 Paediatric studies havespecifically implicated immunodeficiency, syndromes associatedwith neuromuscular weakness, transport off PIC and re-intubationas independent risk factors.4,9

Historically, VAP has been viewed as an inevitable consequence ofcritical illness, but increasingly it is accepted as an avoidable adversehealthcare incident. Although its prevention is a core element in theUS 5 Million Lives campaign, there have been concerns raised aboutthe suitability of such a generic programme to PIC, with similarconcerns relating to the Department of Health Saving Lives initia-tive.15,16 For this reason we describe our establishment of a purpose-designed nurse-led VAP surveillance programme within our PIC,which formed part of our overall HCAI reduction programme.

Methods

At the time of establishing the surveillance no UK nationalprotocol existed, and to our knowledge no individual unit wassystematically measuring VAP incidence. A committee of a leadnurse, consultant intensivist, consultant microbiologist and aninfection control nurse established a consensus definition of VAPwith emphasis on developing a simple, clinically viable tool.

Diagnosis was based on radiological, clinical and laboratorycriteria from the US National Nosocomial Infection SurveillanceSystem but adapted for local use.17

VAP was defined for all ages as pneumonia occurring>48 h postintubation, diagnosed by specific chest radiograph (CXR) changeswith at least three clinical or laboratory findings (Figure 1).

Radiological + at least three clinical/laboratory crite

Radiological Clinical/laboratory

New/worsening purulent br

Core temperature ≥38.5°C

cause)

Leucopenia or leucocytos

0 days–1 week

1 week–1 month

1 month–1 year

2–5 years

6–12 years

13 to <18 years

Significant positive culture

New or progressive

pulmonary

infiltrates,

consolidation or

cavitation,

pneumatoceles for

infants ≤1 year of

age on chest

radiograph (two or

more serial chest

X-rays)

Relevant culture from alter

Figure 1. Criteria for diagnosing ventilato

Radiological

Although radiographic evidence is an important diagnostic toolin VAP, it cannot be interpreted in isolation as infiltrates are only50–75% sensitive for pneumonia.18 Differentiating atelectasis andconsolidation is problematic as pulmonary contusions, interstitiallung disease or community-acquired pneumonia may be mistakenfor VAP without peri-admission CXR.9 VAP was suggested by newor progressive pulmonary infiltrates, consolidation or cavitation(Figure 1) on at least two serial CXRs with gradual resolution; rapidresolution suggesting a non-infective process such as pulmonaryoedema or atelectasis.

To reduce inter-observer variability, all CXRs were interpretedby a senior physiotherapist and the study lead and validated bya consultant radiologist.

Clinical

New onset or worsening of bronchopurulent secretionsTraditionally purulence of bronchial secretions is a quantita-

tive laboratory diagnosis, though many laboratories report itqualitatively (i.e. purulent þþþ, epithelials –) and bedside clin-ical descriptions are often highly variable.18 A more objectivemeasurement, the BronkoTest� Sputum Chart was thereforeused. Purulent secretions in pneumonia arise from neutrophilrecruitment to inflamed airways. The BronkoTest, a colour-graduated scale, correlates sputum colour directly with theneutrophil myeloperoxidase in secretions and is a reliable indi-cator of neutrophil counts.19 All ETT sputum samples wereroutinely compared with this bedside benchmark of purulence,which acted as a standardised trigger for samples to besubmitted for laboratory culture. As an indicator of activeinfection, the BronkoTest correlates well with C-reactive proteinand accurately predicts bacterial load, with positive bacterial

ria

onchial secretions

or <36°C (no other recognised

is (by age)

>34 × 109 L

>19.5 or <5 × 109 L

>17.5 or <5 × 109 L

>15.5 or <6 × 109 L

>13.5 or <4.5 × 109 L

>11 or <4.5 × 109 L

from respiratory secretions

native site of infection

r-associated pneumonia in children.

M. Richardson et al. / Journal of Hospital Infection 75 (2010) 220–224222

culture demonstrated in 84% of adult patients with green(purulent) sputum (94.4% sensitive, 77% specific).20 Nurses andclinicians were encouraged to send secretions if they were con-cerned about clinical chest infection or secretion colour, irre-spective of BronkoTest.

Irrespective of BronkoTest, any child with clinical concern of VAPfrom other criteria had secretions sent for culture.

Core temperature �38.5 �C or <36 �CHyper/hypothermia was defined as at least two consecutive

abnormal readings, using standard measurement techniques, ina 24 h period, which was not clearly attributable to extrapulmonaryinfection, the environment or blood/drug reactions.

Laboratory

Leucopenia or leucocytosisThis was defined by age according to the International

Consensus Conference on Paediatric Sepsis statement.21

Significant culture of respiratory secretionsObtaining minimally contaminated respiratory cultures from

children is difficult. Bronchoscopic techniques used in adultsurveillance are not routine in UK PIC. Whereas blind bron-choalveolar lavage and protected brush samples are reliable,reproducible, diagnostic tools, the former is associated withpneumothorax and raised intracranial pressure and the latter hashigh false-negative rates in PIC with relatively high antibioticuse.22–25

Tracheal aspirates are sensitive for VAP, and the lack ofspecificity due to upper respiratory tract pathogen contamina-tion is outweighed by the advantage of rapid safe bedsidesampling.18

Admission

Enroll all intubated,

ventilated patientsNon-routine investigat

Routine investigations

Continue until 24 h post extub

Routine CXR onadmission

CXR as clinicallyindicated

Figure 2. Investigation flow diagram. FBC, full blood count; CRP, C-reactiv

Relevant culture from alternative site of infectionPositive blood cultures of likely respiratory tract pathogens

unrelated to another source of infection were considered in thediagnosis of VAP, as were significant cultures from pleural fluid andlung parenchyma biopsy, or pathogens detected by validatedimmunofluorescence.

Immunocompromised children required specific CXR changes,but only two clinical or laboratory findings for the diagnosis, as thepresence of leucopaenia/leucocytois in this patient group wasdeemed an unreliable marker of potential VAP.17

Children were categorised as immunocompromised if they hadany of: neutropaenia (count <1.0�109/L), human immunodefi-ciency virus, chemotherapy, leukaemia/lymphoma/post bonemarrow transplant, post splenectomy, immunosuppressive drugtherapy (long-term steroids (any dose >7 days), azothiaprine,cyclosporin, mycophenolate mofetil, tacrolimus, methotrexate,anti-tumour necrosis factor).

A flow diagram (Figure 2) was developed for bedside nurses tofacilitate completion of investigations, which included daily fullblood counts, sputum samples if indicated by the colour sputumchart and blood cultures if the patient’s temperature became�38.5 �C.

The nurse in charge collected data daily from the bedside nurseson all children admitted to PIC over a four-month period until 24 hpost extubation, discharge or death. The study lead reviewed thedata to ensure completeness and accuracy and all suspected VAPcases were referred to microbiology for verification. Literaturereported VAP occurring in 3.0–10.3% of PIC admissions.6,26,27 Aminimum of 100 patient episodes were, therefore, included topower the surveillance programme sufficiently to identify at leastthree cases. Calculations were performed on Stata 9.0 (StataCorp,College Station, TX, USA) using the Wilcoxon rank sum test(significance set at P< 0.05) and bootstrapping for the differencebetween two medians.

Daily FBCCRP on alternate days

ions

Sputum(daily test: BronkoTestTM)If sputum 1–2 = no sampleIf sputum 3–5 = send MC+S sample

Representation only

Temperature

If ≥38.5°C or ≤36.0°C send blood

ation, discharge or death

4 5321

e protein; CXR, chest X-ray; MCþS, microscopy culture and sensitivity.

M. Richardson et al. / Journal of Hospital Infection 75 (2010) 220–224 223

Results

Surveillance data

In total, 158 intubated admissions occurred (March 2007 to July2007) with 58 excluded (ventilated <24 h). One hundred admis-sions (89 patients) were analysed, with 20 (17 patients) defined asimmunocompromised.

Fifty percent were female with a median age and admissionpaediatric index of mortality (PIM) score of 2.06 years (0–15.4years) and 0.05 (0.0–0.61) (�100 (%)), respectively.

Complete data were obtained for all patients. Data entry wasclearly related to unit workload and staffing. When a deficit instaff:patient ratio was observed (calculated using daily augmentedcare periods), the number of mandatory data fields completed bystaff decreased from 11 to 8 out of 14 (P� 0.004).

Sixteen cases were referred to microbiology for analysis: fourdetermined as nosocomial pneumonias acquired prior to PICadmission, six as community-acquired pneumonia and three asdeteriorating underlying disease (one empyema and two varicellapneumonitis). There were three cases of VAP in three patients(median age: 0.96 years), giving a mean pooled rate of 5.6 per 1000ventilator-days.

Two cases were ‘early-onset’ VAP occurring on days 2 (Strepto-coccus pneumoniae) and 4 (Staphylococcus aureus) respectively, theremaining ‘late-onset’ episode occurring on day 12 (Escherichiacoli). Children with VAP had a longer duration of ventilation [9.27 vs3.76 days (95% CI: –9.82 to –0.44; P¼ 0.085)] and a longer PICadmission [11.33 vs 5.05 days (95% CI: –11.33 to –0.02; P ¼ 0.10)].There was no difference in the PIM score, the standard mortalityprediction model reported in UK PIC, between patients with orwithout VAP [0.04 vs 0.05 (95% CI: –0.06 to 0.03; P¼ 0.94)].

No VAP occurred in immunocompromised children, thoughthey had a significantly longer duration of ventilation compared toimmunocompetent patients without VAP (6.8 vs 3.3 days [95% CI:–5.75 to –1.31; P¼ 0.005)]. No deaths occurred within the VAPgroup.

All VAP cases were associated with growth from respiratorysecretions, no patient had positive blood cultures or negativecultures overall.

Workload analysis for surveillance

Surveillance was achieved by the creation of a multidisciplinarycommittee of motivated individuals from relevant specialties,which allowed engagement with the entire PIC staff and manage-ment. Once surveillance was established, data collection proveda significant burden for the nurse in charge of each shift and studylead who completed missing data daily. Data were collectedmanually from non-interfaced information technology systems,including pathology, radiology and electronic care charts, needing2 h of data collection per night. Initial data completeness andquality depended on nursing workload and the number of agencystaff unfamiliar with the programme per shift. Nursing comple-ment was measured on a shift-by-shift basis using augmented careperiod data, which calculates ideal staff:patient ratios, based onpatient complexity and severity. On shifts where staffing was belowthe required establishment for activity, the nurse-in-charge onlymanaged to complete 8 out of the possible 14 data-fields. Thiscompared with 11 out of 14 when the unit had the correct numberor a surplus of staff. Junior and agency staff were noted to be morelikely to omit BronkoTest scoring.

Towards the end of the four-month pilot period, quality controlchecks highlighted a noticeable decline in the amount of completedata returned. However, cross-reference of ward admission records

with study data completed by the lead showed capture of everysingle eligible patient.

Discussion

We implemented, to our knowledge, the first comprehensiveVAP surveillance programme in a UK PIC identifying three cases infive months. Each patient had unrelated underlying pathology andmortality scores were not predictive. The two cases occurringwithin 5 days of intubation were due to fully susceptible organisms(S. aureus and S. pneumoniae) often implicated in early-onsetpaediatric VAP and associated with aspiration of gastriccontents.26,28 The other, late-onset, case was caused by a multiplyresistant E. coli. This type of organism is associated with late-onsetVAP (�5 days after intubation).9

Despite the absence of specific interventions to reduce VAP wefound a relatively modest VAP rate of 5.6 per 1000 ventilator-days incomparison with previously published data.4,6 Increased morbiditywas demonstrated in those with VAP who were ventilated fora median of 5.97 days longer. Although this was not statisticallysignificant due to the small numbers, the 95% CI values suggest thatthere may be grounds to reject a null hypothesis of no difference inventilation duration between the two groups. Due to the smallsample size and low VAP incidence, no risk factor analysis waspossible although neuromuscular blocking agents (52%) and gastriculcer prophylaxis (73%) use, both associated with VAP, was frequent.

The 20% of children defined as immunocompromised had twicethe predicted mortality of immunocompetent patients [0.10 versus0.05 (�100 (%)), P¼ 0.07] and were ventilated longer. Althoughimmunodeficiency and immunosuppression are suggested riskfactors for VAP, changes to contemporary ventilation strategies maybe important.27 Recognition of the complications and mortalityassociated with invasive ventilation in immunocompromisedchildren has resulted in greater use of non-invasive ventilation, orearly aggressive extubation.29,30 During the study period, 11immunocompromised children were intubated <24 h, and soexcluded; several others were managed entirely on non-invasiveventilation. Ultimately this may have reduced the immunocom-promised children’s risk of VAP.

Overall, our diagnostic criteria were sensitive, insofar as theyidentified all patients whom physicians considered to displaysymptoms of pneumonia, and furthermore they were feasible toimplement in a clinical environment. It is evident, however, thatnone was specific for VAP, as, even when combined, children withnosocomial bacterial tracheitis might fulfil them. For the purpose ofsurveillance, aiming to assemble a comprehensive picture bycapturing all adverse events, we concluded any potential lack ofspecificity unimportant providing that future surveillance adoptedthe same criteria.

Adopting specific CXR changes in immunocompromised chil-dren as an essential criterion is debatable. Such children frequentlypresent with diffuse pulmonary infiltrates related to non-infectiouscauses such as interstitial oedema, fibrosis, radiation pneumonitis,graft-versus-host disease or chemotherapeutic/other drug reac-tions, and distinguishing new VAP infiltrates can be unfeasible.31,32

Although this might account for lack of VAP in the immunocom-promised, on review no patient in this group was clinically thoughtto have VAP. Further specific studies regarding VAP diagnosticcriteria in this group are necessary.

Establishing a VAP surveillance programme was a huge chal-lenge. An important aspect of effecting change is clear leadership,which ‘sells’ a vision of the desired future and provides direction forchange. Clear explanation to all groups enabled widespread under-standing of the project, its desired objectives and benefits. Delega-tion of responsibility to staff for investigations and data collection

M. Richardson et al. / Journal of Hospital Infection 75 (2010) 220–224224

gave individuals the opportunity to own the change themselves. Akey issue for future surveillance is maintaining staff engagementand interest. Ideally a rolling programme of education and real-timefeedback regarding unit VAP rates will be established; whereas theworkload, largely from data collection, prohibited this, such feed-back now occurs in our ongoing surveillance programme.

One further concern regards a potential conflict for clinicianswith clinical and financial oversight for PIC and responsibility forHCAI identification. For this reason final confirmation of VAPremained the responsibility of the consultant microbiologist withbedside clinical staff limited to identification of possible cases,although ideally VAP surveillance should be undertaken by stafffrom outside the PIC team.

Full engagement with the PIC team occurred, although newways of sustaining interest and motivation in surveillance arenecessary.

Purpose-designed VAP bundles and automated VAP surveil-lance, using upgraded bedside information technology systems, arebeing introduced and it is hoped this will impact on VAP rates andease the collection of data.

Acknowledgements

We would like to thank all the medical, nursing and alliedprofessions whose enthusiasm enabled this surveillance to occur.

Conflict of interest statementNone declared.

Funding sourcesNone.

References

1. Elward AM. Pediatric ventilator-associated pneumonia. Pediat Infect Dis J2003;22:445–446.

2. Foglia E, Meier MD, Elward A. Ventilator-associated pneumonia in neonatal andpediatric intensive care unit patients. Clin Microbiol Rev 1997;20:409–425.

3. Langley JM, Bradley JS. Defining pneumonia in critically ill infants and children.Pediat Crit Care Med 2005;6(Suppl):S9–S13.

4. Elward AM, Warren DK, Fraser VF. Ventilator-associated pneumonia in pedi-atric intensive care unit patients: risk factors and outcomes. Pediatrics2002;109:758–764.

5. Fischer JE, Allen P, Franconi S. Delay in extubation of neonates and childrenafter cardiac surgery: impact of ventilator-associated pneumonia. IntensiveCare Med 2000;26:942–949.

6. Almuneef M, Memish ZA, Balkhy HH, Alalem H, Abutaleb A. Ventilator-asso-ciated pneumonia in a pediatric intensive acre unit in Saudi Arabia: a 30 monthprospective surveillance. Infect Control Hosp Epidemiol 2004;25:753–758.

7. National Nosocomial Infections Surveillance (NNIS) system report. Datasummary from January 1992–June 2004. J Am Infect Control 2004;32:470–485.

8. Berdal JE, Bjørnholt J, Blomfeldt A, Smith-Erichsen N, Bukholm G. Patterns anddynamics of airway colonisation in mechanically-ventilated patients. ClinMicrobiol Infect 2007;13:476–480.

9. Pieracci FM, Barie PS. Strategies in the prevention and management of venti-lator-associated pneumonia. Amer Surg 2007;73:419–433.

10. Drakulovic MB, Torres A, Bauer TT, Nicolas JM, Nogue S, Ferrer M. Supine bodyposition as a risk factor for nosocomial pneumonia in mechanically ventilatedpatients: a randomised trial. Lancet 1999;354:1851–1858.

11. Hampton DC, Griffith D, Howard A. Evidence-based clinical improvement formechanically ventilated patients. Rehab Nurs 2005;30:160–166.

12. Kunis K, Puntillo K. Ventilator-associated pneumonia in the ICU: its patho-physiology, risk factors and prevention. Am J Nurs 2003;103(8): 64AA–64GG.

13. Feldman C, Kassel M, Cantrell J, et al. The presence and sequence of endotra-cheal tube colonization in patients undergoing mechanical ventilation. EurRespir J 1999;13:546–551.

14. George DL, Falk PS, Wunderink RG, et al. Epidemiology of ventilator-acquiredpneumonia based on protected bronchoscopic sampling. Am J Respir Crit CareMed 1998;158:1839–1847.

15. Curley MA, Schwalenstocker E, Deshpande JK, et al. Tailoring the Institute forHealth Care Improvement 100,000 Lives Campaign to pediatric settings: theexample of ventilator-associated pneumonia. Pediatr Clin North Am2006;53:1231–1251.

16. Department of Health. The Health Act 2006 – code of practice for the prevention andcontrol of healthcare associated infections. London: Department of Health; 2006.

17. Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDC definitions fornosocomial infections. In: Olmsted RN, editor. APIC infection control and appliedepidemiology: principles and practice. St Louis, MO: Mosby; 1996A1–A19.

18. Wright ML, Romano MJ. Ventilator-associated pneumonia in children. SeminPediatr Infect Dis 2006;17:58–64.

19. Stockley RA, Bayley D, Hill SL, Hill AT, Crooks S, Campbell EJ. Assessment ofairway neutrophils by sputum colour: correlation with airways inflammation.Thorax 2001;56:366–372.

20. Stockley RA, O’Brien C, Pye A, Hill S. Relationship of sputum colour to natureand outpatient management of acute exacerbations of COPD. Chest2000;117:1638–1645.

21. Goldstein B, Giroir B, Randolph A, the Members of the International ConsensusConference on Pediatric Sepsis. International pediatric sepsis consensusconference: definitions for sepsis and organ dysfunction in pediatrics. PediatCrit Med 2005;6:12–18.

22. Labenne M, Poyart C, Rambaud C, et al. Blind protected specimen brush andbronchoalveolar lavage in ventilated children. Crit Care Med 1999;27:2537–2543.

23. Gauvin F, Dassa C, Chaıbou M, Proulx F, Farrell CA, Lacroix J. Ventilator-asso-ciated pneumonia in intubated children: comparison of different diagnosticmethods. Pediatric Crit Care Med 2003;4:437–443.

24. Gauvin F, Lecroix J, Guertin MC, et al. Reproducibility of blind protectedbronchoalveolar lavage in mechanically ventilated children. Am J Respir CritCare Med 2002;165:1618–1623.

25. De Jaeger A, Litalien C, Lacroix J, Geurtin MC, Infante-Rivard C. Protectedspecimen brush or bronchoalveolar lavage to diagnose bacterial nosocomialpneumonia in ventilated adults: a meta-analysis. Crit Care Med 1999;27:2548–2560.

26. Richards MJ, Edwards JR, Culver DH, Gaynes RP. Nosocomial infections inpediatric intensive care units in the United States. Pediatrics 1999;103:e39.

27. Fayon MJ, Tucci M, Lacroix J, et al. Nosocomial pneumonia and tracheitis ina pediatric intensive care unit: a prospective study. Am J Resp Crit Care Med1997;155:162–169.

28. Kollef MH, Shorr A, Tabak YP, Gupta V, Liu LZ, Johannes RS. Epidemiology andoutcomes of health-care-associated pneumonia: results from a large US data-base of culture-positive pneumonia. Chest 2005;128:3854–3862.

29. Crawford SW, Schwartz DA, Petersen FB. Mechanical ventilation after bonemarrow transplantation: risk factors and clinical outcome. Am Rev Respir Dis1988;137:682–687.

30. Piastra M, Antonelli M, Chiaretti A, Polidori G, Polodori L, Conti G. Treatment ofacute respiratory failure by helmet-delivered non-invasive pressure supportventilation in children with acute leukemia: a pilot study. Intensive Care Med2004;30:472–476.

31. Maimon N, Barski L, Sion-Vardy N, Almog Y. A man with interstitial pneumoniaand pancytopenia during radiotherapy. Chest 2004;126:1368–1371.

32. Crawford S. Noninfectious lung disease in the immunocompromised host.Respiration 1999;65:385–395.