Prevention of Gram-Negative Bacillary Pneumonia

Using Polymyxin Aerosol as Prophylaxis

II. EFFECT ONTHE INCIDENCE OF PNEUMONIA

IN SERIOUSLY ILL PATIENTS

JAMESM. KLICK, GARYC. DUMOULIN, JOHNHEDLEY-WHYTE,DANIEL TERES, LEONARDS. BUSHNELL, and DAVID S. FEINGOLD

From the Department of Anaesthesia and the Thorndike Laboratory of theDepartment of Medicine of the Harvard Medical School at the Beth IsraelHospital, Boston, Massachusetts 02215

A B S T R A CT All 744 patients admitted to a Respira-tory-Surgical Intensive Care Unit (RSICU) were in-cluded in a prospective study of the effects of a poly-myxin (2.5 mg/kg body wt/day in six divided doses) ora placebo aerosol sprayed into the posterior pharynxand tracheal tube (if present), during 11 alternating2-mo treatment cycles. The incidence of upper airwaycolonization in the RSICUwith Pseudomonas aeruginosawas 1.6% during the polymyxin treatment cycles (total374 patients) and 9.7% during the placebo cycles (370patients) (xi = 23.2, P < 0.01). 3 patients in the RSICUacquired Pseudomonas pneumonia, as defined by inde-pendent "blinded" assessors, during the polymyxin cy-cles while 17 acquired a Pseudomonas pneumonia duringthe placebo cycles (x' = 10.2, P < 0.01). The overallmortality was similar in both placebo and polymyxin-treated groups (12.2 vs. 12.0%). Systemic antibiotic us-age was similar in the different cycles; 49% of patientsin the placebo and 53% in the polymyxin-treated groupsreceived systemic antibiotics while in the RSICU.

This work was presented in part at the Annual Meetingof the Society of Critical Care Medicine, Anaheim, Calif.,February 1975.

Lieutenant commander Klick, Medical Corps, U. S. Navy,was seconded from the Boston Naval Hospital, Departmentof Anesthesiology. Dr. Klick's present address is the De-partment of Anesthesia, National Naval Medical Center,Bethesda, Md. 20014; Dr. Feingold's address is the De-partment of Medicine (Dermatology), Boston's VeteransAdministration Hospital, Boston, Mass. 02130; and Dr.Teres' address is Medical Center of Western Massachusetts,Springfield, Mass. 01107.

Received for publication 13 March 1974 and in revisedform 7 October 1974.

INTRODUCTIONlAn ever increasing problem in modern hospital care hasbeen the development of nosocomial respiratory infec-tions preceded by colonization of the upper respiratorytract (1). Sanford and Pierce have cited occurrencesof nosocomial infections of the respiratory tract asranging from 0.5 to 5.0% of all hospitalized patients(2). Much progress has been made in elucidating reser-voirs of pathogens existing in the hospital environment,including anesthesia equipment, room humidifiers, dis-infectants, water sources, and jellies (3-8). However,dangerous pathogens, such as Pseudomonas aeruginosaand Klebsiella pneumoniae, still account for increasingmorbidity and mortality in certain patient populations,particularly critically ill patients (9-11).

To assess the extent of nosocomial infection we previ-ously carried out a 30-mo retrospective analysis of in-fection in our Respiratory-Surgical Intensive Care Unit(RSICU)' (12). During that time P. aeruginosa wasthe most common pulmonary pathogen (50% of 158episodes of bacterial pneumonia). Pseudomonas pneu-monia carried a 75% mortality rate as compared to 25%in other bacterial pneumonias. The mortality of pneu-monia-free patients was only 3.8%. Consequently, analy-sis of the potential reservoirs from which Pseudomonascould be recovered became a major objective. We then

'The opinions or assertions contained herein are thoseof the authors, and are not to be construed as official ornecessarily reflecting the views of the Navy Department,or the naval service at large.

'Abbreviation used in this paper: RSICU, Respiratory-Surgical Intensive Care Unit.

The Journal of Clinical Investigation Volume 55 March 1975 514-519514

undertook the monitoring of 640 patients and their en-vironment in the RSICU during a 19-mo period. 71%of the sputum or throat isolates of Pseudomonas wereclearly hospital acquired, and at least 10 of 16 patientswho had P. aeruginosa on first culture were suspected ofhaving hospital-acquired organisms. Pyocine type 10was the "resident" strain, outbreaks of which tended tooccur in clusters. Pyocine type 10 was regularly cul-tured from only one environmental source (the sinks),including 1-2-wk intervals during which no patient inthe RSICU was colonized with Pseudomonas.

We have presented data which show that treatmentof Pseudomonas pneumonia with systemic antibiotics isat best minimally effective (12). Since the portal ofentry is usually the upper respiratory tract, the feasi-bility of specific prophylaxis with localized antibiotictherapy was tested in a pilot study. Polymyxin B wasgiven by aerosol to reduce colonization of the upperrespiratory tract with nosocomial gram-negative bacilli.58 high-risk patients from the RSICU entered the pilottrial. 33 were randomly selected to receive 2.5 mg/kg/day of polymyxin B by a hand atomizer into the pharynxand, if present, tracheal tube: 17 of 25 control patientsbecame colonized with gram-negative bacilli as com-pared with 7 of 33 polymyxin-treated patients (P <0.01). Polymyxin B was selected for reasons whichhave been previously detailed (13).

Since the upper respiratory tract is rapidly colonizedduring acute illnes and hospitalization (1, 14) and sincepneumonia often occurs before prolonged controlledventilation is instituted (12), prevention must be at-tempted early in the course of a given critical illness(15). Thus, after completion of the pilot study, we re-ceived permission from the appropriate hospital com-mittees to put all patients in our RSICU on 2-mo alter-nating cycles of placebo and polymyxin aerosols to de-termine the effect on the development of pneumonia.

METHODSThe present study was conducted in the RSICU at theBeth Israel Hospital where patients in respiratory distressare admitted from all services. The majority of the pa-tients were those in whom respiratory impairment was an-ticipated postoperatively.

The interval from 23 October 1972 to 14 June 1974 wasdivided into 10 8-wk cycles; the last cycle, cycle 11, was4 wk in duration. During this time, 744 patients wereadmitted to the RSICU, 370 during cycles when a normalsaline placebo aerosol was administered (mean age 60 yr,average number patient days 5.3) and 374 during cycleswhen polymyxin was administered (mean age 57 yr, averagenumber of patient days 5.1). The systemic antibiotics givento the patients were recorded. No attempt was made byanyone connected with the study to influence systemic anti-biotic administration.

Within the first 24 h of admission to the RSICU allpatients had their posterior pharynx and tracheal tube (ifpresent) sprayed with an aerosol which consisted either of

normal saline (placebo) or a 0.5% polymyxin in normalsaline aerosol every 4 h to a total daily dose of 2.5 mg/kgin the manner described by Greenfield, Teres, Bushnell,Hedley-Whyte, and Feingold (13). The placebo and poly-myxin aerosols were alternated so that during each 8-wkperiod all patients admitted to or present in the RSICUreceived the same aerosol. The content of the aerosol wasunknown to the RSICU staff, and the code was brokenby one of us (J. H-W.) at the end of the study.

Tracheal aspirate and stool cultures were obtained within24 h of admission. Daily tracheal aspirates for culturewere usually collected in the morning from each patientfollowing chest physiotherapy and within 1 h before the 10a.m. aerosol treatment. Urine was cultured at the time ofbladder catheterization and when clinically indicated.

The drains of the sinks used for handwashing in theRSICU were cultured at least biweekly, and the cultureswere examined for the presence of gram-negative organ-isms, particularly Pseudomonas. During the duration of thisstudy a daily practice of sink sterilization was in effectwithin the RSICU. This consisted of filling the sinks, whichhad previously been fitted with drain cutoff valves, witha 5% phenol solution and heating the disinfectant with a500-W immersion heater to 70'C for 90 min. The sinkoverflow drains, reservoir nebulizers, and ventilator trapbottles were all cultured intermittently. During the twiceyearly closing of the RSICU for cleaning, floors, beds, vents,etc. were cultured using cotton swabs moistened with dex-trose-phosphate broth. Settling plates were positioned aroundthe unit and exposed for 2 h during all stages of cleaningto monitor airborne contamination.

All bacteriology was performed by the same microbiolo-gist (G. C. D.). The isolation media for P. aeruginosa wascetrimide agar, and identification was based on colonymorphology, pigment production, and biochemical reactionsin oxidative-fermentative media. Pyocine typing was doneby the technique of Gillies and Govan (16). Fermentativegram-negative bacilli were identical by the criteria of Ed-wards and Ewing (17). The nonfermentative gram-nega-tive bacilli were identified by the criteria of Weaver,Tatum, and Hollis.' Polymyxin sensitivities were performedon all gram-negative bacilli using a modified method ofdisk diffusion antibiotic susceptibility testing as describedby Bauer, Kirby, Sherris, and Turck (18).

Colonization was defined as the isolation of a neworganism from throat or sputum on more than one con-secutive culture. If only a few colonies (nine or less onthe primary plating) of bacteria were present in the throatculture on admission to the unit, any increase on subsequentculture was considered to represent colonization. For gram-negative bacilli this is an identical definition to that used inour pilot study (13). The diagnosis of pneumonia wasbased upon three separate opinions: that of the house officersin the RSICU having direct care for the patient; an inde-pendent physician reviewing the records of the patients atthe end of every two cycles; and, finally, a third opinionfrom a group consisting of a different physician and para-medical personnel reviewing the records of the patients atthe end of every two cycles. The treatment given duringcycles was not known to those passing judgment on thepresence or absence of pneumonia. Each group was asked to

'Weaver, R. E., H. W. Tatum, and D. G. Hollis. 1972.The Identification of Unusual Pathogenic Gram NegativeBacteria. Atlanta, Ga.: U. S. Department of Health, Edu-cation and Welfare, Public Health Service, Center for Dis-ease Control. Preliminary revision.

Polymyxin Aerosol in the Prevention of Pneumonia 515

20r

Kr;k

15

10

5 n

* Polymyxin Aerosol

H. *0

:....

:.:....,.........

H..1Hfl-.

Cycles I 11 III IV V VI VIl VilI Ix x xiIOtieats 66 6 72 56 78 55 76 82 85 41

FIGURE 1 Upper airway colonization rate with P. aeru-ginosa in patients in the RSICU in each of the 11 cycles.Patients in the even-numbered cycles received polymyxinaerosol. Total number of patients in each cycle is indicated.36 patients were colonized with Pseudomonas during theplacebo cycles and 6 during the polymyxin cycles (x2 =23.2, P < 0.01).

decide whether the pneumonia was "probably" present or only"possibly" present and whether the pneumonia was presenton admission to the RSICU or acquired while the patientwas in the RSICU. Only cases of "probable" pneumoniawere considered.

The criteria for the diagnosis of pneumonia were basedon the presence of persistent X-ray evidence of an alveolarinfiltrate on 2 or more consecutive days observed at a timewhen sputum cultures or tracheal aspirates were positivefor potentially pathogenic organisms; an elevated alveolar-

arterial oxygen tension difference while breathing 100%oxygen (AaDO2s1); and an evaluation of the temperaturecourse, daily weights, central venous pressure, and Swan-Ganz catheter data (19).

The Beth Israel Hospital's Committee on Clinical In-vestigation and New Procedures and Forms of Therapyaccepted this methodology and was kept abreast of resultsat intervals. The present study was terminated by theChairman of the Committee on Clinical Investigation in themiddle of the 11th cycle, a placebo cycle, because of thesudden increase in colonization and infection with P. aeru-ginosa (Figs. 1 and 2). Because of the number of patientsat risk in the RSICU, the treatment code was broken, andall patients were placed on polymyxin aerosol, thus endingthe blinded study.

At the end of our analysis to account for those patientswho, during their stay in the RSICU, overlapped the 2-mocycles, the data were reexamined with these patients ex-cluded. This was done to rule out the possibility of a patientbeing colonized in one cycle and developing a pneumoniain the following cycle. The criteria for overlapping werestrict in that, for example, a patient whose stay in theRSICU was 2 wk and who overlapped the following cycleby a day was considered as an "overlap".

RESULTS

Systemic antibiotic usage. Full information on sys-temic antimicrobial therapy prescribed by the patient'sphysician was available for 331 of 371 (89.5%) of thepatients in the placebo aerosol cycles and for 324 of 374(86.6%) of the patients in the polymyxin aerosol cycles.49% of those patients given the placebo saline aerosolwere being treated with systemic antibiotics vs. 53%during the polymyxin periods. During the placebo aero-sol cycles systemic antibiotics were given as follows:oxacillin 19% of patients, ampicillin 13%, cephalothin

TABLE I

Number of Acquired Microorganisms Other than P. aeruginosa by Cycle

PolymyxinPlacebo Polymyxin

Cycle ................... I II III IV V VI VII VIII IX X XI total total

Patients with acquired GNB 22 19 24 13 23 25 18 12 22 17 15 124 86

I. Polymyxin-sensitive organismsEnterobacter species 6 0 14 4 15 8 12 7 10 3 5 62 22Acinetobacter species 8 1 3 0 4 1 6 0 8 1 2 31 3Other GNB* 2 1 22 3 12 5 6 6 6 2 3 51 17

II. Polymyxin-resistant organismsSerratia species 0 6 4 3 4 2 2 1 2 1 0 12 13Proteus species 3 10 4 2 3 10 3 4 8 6 2 23 32Flavobacterium species 11 15 9 7 5 10 1 3 8 9 4 38 44Pseudomonas mallophilia 0 0 0 0 0 0 0 0 0 0 2 2 0

III. Gram-positive organismsStaphylococcus aureus 1 0 2 3 5 9 6 12 5 8 1 20 32Enterococcus 5 4 0 0 0 0 2 1 7 1 3 17 6

IV. YeastCandida albicans 0 0 0 0 1 4 4 12 5 23 2 12 39Candida species 0 0 0 0 0 0 0 1 2 6 0 2 7

* Escherichia coli, Citrobacter, Klebsiella.GNB, gram-negative bacilli.

516 Klick, du Moulin, Hedley-Whyte, Teres, Bushnell, and Feingold

TABLE IISummary of Pneumonias Acquired within the RSICU

Placebo Polymyxincycles cycles

Pseudomonas aeruginosa 17 3Enterobacter species, Klebsiella pneumoniae

and Escherichia coli 3 1Slaphylococcus aureus 2 5Serratia species or Proteus species 6 8Flavobacterium meningosepticum 0 1Pseudomonas maltophilia I 0Candida albicans I 0

Total 30 18

13%, and kanamycin 6%. During the polymyxin aerosolcycles corresponding percentages were 20, 12, 9, and 6.Systemic gentamicin, systemic carbenicillin, and/or sys-temic polymyxin were used in 11.8% of the placebo aero-sol patients and 6.5% of the polymyxin aerosol patients:this difference may be a response of the patient's primaryphysicians to the increased incidence of P. aeruginosainfection in the placebo aerosol periods.

Colonization rate. Fig. 1 shows the colonization rateof P. aeruginosa in RSICU patients during each of the11 cycles. The overall rate of upper airway coloniza-tion with Pseudomonas was 5.6%. During the placebocycles, the upper airways of 36 patients (9.7%) werecolonized with Pseudomonas while during the polymyxinaerosol cycles the upper airways of 6 patients (1.6%)were colonized (x = 23.2, P<0.01). This decrease incolonization is consistent with the previous investigationby Greenfield et al. (13). As expected, the incidence ofairway colonization with other species of Enterobac-teriaceae sensitive to polymyxin was also diminishedduring those cycles in which polymyxin was employed(Table I). Colonization with Escherichia coli, Entero-bacter species, and Klebsiella species occurred on 22 oc-casions during the polymyxin cycles and on 62 occasionsduring the placebo cycles (x' = 22, P<0.01).

Perhaps of equal importance is the colonization ratewith those organisms resistant to polymyxin. Eight pa-tients were colonized with Staphylococcus aureus dur-ing the placebo cycles, and nine were colonized duringpolymyxin cycles (Tables I). 38 patients (9.5%) in theplacebo cycle were colonized with Serratia species orProteus species while 44 patients (12.0%) were colo-nized during the use of the polymyxin aerosol (NS).The most common nonfermentative gram-negative bacil-lus in the RSICU was Flavobacterium species. 10.3%of the patients were colonized with this organism dur-ing the placebo cycles, and 11.8% were colonized duringthe polymyxin cycles (Table I).

Since the start of sink decontamination, sink drain andoverflow drain cultures have been taken twice weekly.Only occasional cultures positive for Pseudomonas have

been obtained. Routine cultures of heated nebulizershave been consistently negative, and only rare positivecultures have been obtained from the expiratory trap ofthe ventilator.

Pneumonia (Table II). During the placebo periods,30 patients acquired pneumonia while in the RSICU; 17of these were due to Pseudomonas (4.6% of the 370 pa-tients), and 4 of these latter patients died (Fig. 2). Inthe polymyxin cycles, 18 acquired pneumonia, 3 due toPseudomonas (0.8% of the 374 patients; x2 = 10.2, P <0.01) 2 of whom died. Five patients acquired staphylo-coccal pneumonia while in the RSICU during the poly-myxin cycles, two of whom died, while two acquired astaphylococcal pneumonia during the placebo cycles(NS), one of whom died. There was a single case ofpneumonia attributed to polymyxin-resistant Flavobac-terium meningosepticum, subtype D during a polymyxincycle (20). The number of patients with pneumonia ac-quired in the RSICU due to Serratia species and Pro-teus species during the placebo and polymyxin cycles wassix and eight, respectively. During polymyxin cyclesthere was one case of pneumonia attributed to K. pneu-moniae. During the placebo cycles, one case of acquiredpneumonia was attributed to Enterobacter species, andtwo cases of pneumonia were attributed to K. pneu-moniae (Table II).

Cycle-overlapping patients. A total of 52 patientswere found to overlap two cycles: none overlappedthree. 33 patients began in placebo periods and completedtheir stay in the RSICU in polymyxin periods. 19 pa-

K~K

12

10

8

6 - ':

4 "

2

Cycles IPatiens 68

* Polymyxin Aerosol

1-.:

..

.'.,..........

. '....II III IV V VI VIl Vil IX X xl63 66 72 58 78 55 76 82 85 41

FIGURE 2 Incidence of Pseudomonas pneumonia acquiredby patients in the RSICU during each of the 11 cycles.Total number of patients in each cycle is indicated. 17 pa-tients acquired Pseudomonas pneumonia during the placebocycles and 3 acquired Pseudomonas during the polymyxincycle (X2= 10.2, P < 0.01).

Polymyxin Aerosol in the Prevention of Pneumonia 517

-

.-:

.. .

tients began in the polymyxin-cycles and overlappedinto a placebo period. Six patients in the placebo cyclesacquired pneumonia and then overlapped into polymyxincycles. Three of these pneumonias were due to P. aerugi-nosa, two were due to Proteus species, and one was dueto Serratia species. Only two patients began in theRSICU in polymyxin cycles, developed pneumonia, andthen completed their RSICU stays in placebo cycles.One of these two patients had a P. aeruginosa pneu-monia, and the other pneumonia was due to Proteusspecies. No patient was free of pneumonia in the firstcycle and then developed pneumonia in the next cycle.Thus with all cycle-overlapping patients excluded, 14patients acquired P. aeruginosa pneumonias during theplacebo periods, and two acquired P. aeruginosa pneu-monias during the polymyxin periods (x2 = 9.3, P <0.01).

DISCUSSION

Patients in the RSICU were placed on alternating poly-myxin-placebo aerosol treatments to determine if the in-cidence of gram-negative bacillary pneumonia is re-duced by polymyxin. Such aerosol treatment with poly-myxin B has previously been shown not to contaminatethe culture media so as to inhibit the growth of viableorganisms deposited in the plate (13). There is there-fore no diminution in the ability to culture existing poly-myxin B sensitive gram-negative bacilli after suchtreatment. As shown by other studies (21), inhibition ofgram-negative bacteria by commensal flora in the re-spiratory tract has been presumed to be a factor in thedetermination of colonization. The majority of patientsentering the RSICU, however, had recent prior anti-biotic administration and consequently showed little"normal flora" on their original plates. We thus con-cluded that interbacterial inhibition would not play asignificant role in this study.

The sinks used for handwashing continued to undergoa daily sterilization program throughout this study. Theywere previously identified as the only known reser-voir of Pseudomonas within the RSICU (8), theirdrains being a source of pyocine type 10 P. aeruginosa.This pyocine type previously accounted for 57.5% of theupper airway isolates acquired within the RSICU, whilefollowing sink sterilization, pyocine type 10 was re-sponsible for only 10% of the acquired isolates. Thediminished colonization by gram-negative bacilli of theupper respiratory tract has again been demonstrated inthose patients receiving polymyxin aerosol, the incidenceof Pseudomonas colonization being 9.7% in the placebocycles vs. 1.6% in the polymyxin cycles (x' = 23.2,P < 0.01). Our initial concern that we would see alarge increase in the number of patients colonized withpolymyxin-resistant organisms was unfounded during

the use of alternating polymyxin and normal salineplacebo cycles. Of particular concern to us was the pos-sibility that patients might become colonized withstaphylococci, Proteus species, Serratia species, or otherresistant organisms during a polymyxin cycle only todevelop pneumonia during a subsequent placebo cycle.Analysis of the overlapping patients showed that thisinfection pattern did not occur to any significant extent.Cycle-overlapping patients were not at greater risk ofdeveloping pneumonia than comparably sick patients inthe placebo cycles. Overall conclusions about our studywere not altered by excluding cycle-overlapping patients.

In a recent study using endotracheal gentamicin pro-phylaxis, some selection of resistant organisms oc-curred. In particular, gentamicin-resistant Providencestrains emerged (15). During the use of polymyxin,there was an insignificant increase in colonization ofSerratia species and Proteus species of 2.5% and an in-crease in colonization with Flavobacterium species of1.5%, while only one additional patient was colonizedwith S. aureus. It is possible that having used such anti-biotics prophylactically in an alternating manner witha placebo, we were able to minimize colonization withresistant organisms.

Despite the significant decrease in Pseudomonas pneu-monia acquired within the RSICU, we were unable todemonstrate any effect on the overall mortality. 12.2%of the patients (45) died during placebo cycles and12.0% (45) during the polymyxin cycles. That theequality in overall mortality rates is to some extent dueto the difference in mortality of Pseudomonas pneumoniaacquired in the RSICU during placebo vs. polymyxincycles should be considered. Two of the three patientsacquiring Pseudomonas pneumonia died during the useof polymyxin prophylaxis, while 4 of the 17 patients(24%) acquiring Pseudomonas pneumonia died in theplacebo group (NS). That three patients acquiredPseudomonas pneumonia during polymyxin prophylaxisand that two of them died may be a reflection of theirdegree of illness. One of these patients was hospitalizedwith a large intracerebral hemorrhage and renal failure;he succumbed to a cardiac arrest following an acutetension pneumothorax. The other death occurred in anelderly man following massive peritonitis and shock.Both patients required mechanical controlled ventila-tion; the neurosurgical patient, in addition, requiredpositive end-expiratory pressure.

In addition to the Pseudomonas pneumonia acquiredby patients while in the RSICU, there were 13 cases ofacquired pneumonia during the placebo cycles and 15cases during the polymyxin cycles of etiology otherthan P. aeruginosa (Table II). The most striking dif-ference in acquired pneumonias other than those due toPseudomonas were the five cases of S. aureus pneumonia

518 Klick, du Moulin, Hedley-Whyte, Teres, Bushnell, and Feingold

during the polymyxin cycles and two cases during theplacebo cycles (NS). The mortality rate for patientswith pneumonia was 30.4% vs. the 12.1%. Pneumoniawas a diagnosis in 18. 1% of the 744 patients in thestudy, and 35.5% of the pneumonias were acquired whilethe patients were in the RSICU.

At present we are encouraged that the incidence ofPseudomonas pneumonia acquired within the intensivecare setting can be reduced by the techniques we haveemployed. We are continuing our efforts to reduce theincidence of nosocomial pneumonias and the evaluationof uninterrupted polymyxin prophylaxis.

ACKNOWLEDGMENTSWe thank Ms. Gail Gomez, R.N., the Nursing Staff in theRSICU, Ms. J. Deborah Welsh, and the Beth Israel Phar-macy Staff. Dr. John P. Gilbert, Staff Statistician in theHarvard Computing Center and Lecturer on Statistics gaveinvaluable advice. Members of The Radiology Departmentwere of enormous help.

Dr. Klick was supported by the U. S. Navy; Mr. duMoulin was supported by Center Grant GM15904; Dr.Hedley-Whyte was supported in part by Grant HL15219;Dr. Teres was supported by Training Grants HE05422 andGM01273; and Dr. Feingold was supported in part byCenter Grant GM15904 and a grant-in-aid from the Bur-roughs Wellcome Co.

REFERENCES1. Johanson, W. G., Jr., A. K. Pierce, J. P. Sanford, and

G. D. Thomas. 1972. Nosocomial respiratory infectionswith gram-negative bacilli. The significance of coloniza-tion of the respiratory tract. Ann. Intern. Med. 77: 701-706.

2. Sanford, J. P., and A. K. Pierce. 1971. Current infec-tion problems-respiratory. Proceedings of the Interna-tional Conference on Nosocomial Infections. Chicago,Ill. American Hospital Association. 77-81.

3. Babington, P. C., A. B. B. Baker, and H. H. Johnston.1971. Retrograde spread of organisms from ventilatorto patient via expiratory limb. Lancet. 1: 61-62.

4. Moody, M. R., V. M. Young, D. M. Kenton, and G.D. Vermeulen. 1972. Pseudomonas aeruginosa in a centerfor cancer research. I. Distribution of intraspecies typesfrom human and environmental sources. J. Infect. Dis.125: 95-101.

5. Lowbury, E. J., B. T. Thom, H. A. Lilly, J. Babb, andK. Whitfall. 1970. Sources of infection with Pseudo-monas aeruginosa in patients with tracheostomy. J.Med.Microbiol. 3: 39-56.

6. Thomas, M. E., E. Piper, and I. M. Maurer. 1972.Contamination of an operating theatre by gram-nega-tive bacteria. Examination of water supplies, cleaningmethods, and wound infections. J. Hyg. 70: 63-73.

7. Mushin, R., and G. Ziv. 1973. Epidemiological aspects of

Pseudomonas aerugintosa in man, animals, and the en-vironment. Application of pyocin typing. Isr. J. Med.Sci. 9:155-161.

8. Teres, D., P. Schweers, L. S. Bushnell, J. Hedley-Whyte, and D. S. Feingold. 1973. Sources of Pseudo-monas aeruginosa infection in a respiratory/surgical in-tensive-therapy unit. Lancet. 1: 415-417.

9. Johanson, W. G., A. K. Pierce, and J. P. Sanford. 1969.Changing pharyngeal bacterial flora of hospitalized pa-tients. Emergence of gram-negative bacilli. N. Engl. J.Med. 281: 1137-1140.

10. DuPont, H. L., and W. W. Spink. 1969. Infections dueto gram-negative organisms. An analysis of 860 patientswith bacteremia at the University of Minnesota MedicalCenter, 1958-1966. Medicine (Baltimore). 48: 307-332.

11. Lerner, A. M., and M. J. Federman. 1971. Editorial.Gram-negative bacillary pneumonia. J. Infect. Dis. 124:425-427.

12. Stevens, R. M., D. Teres, J. J. Skillman, and D. S. Fein-gold. 1974. Pneumonia in an intensive care unit: athirty month experience. Arch. Intern. Mcd. 134: 106-111.

13. Greenfield, S., D. Teres, L. S. Bushnell, J. Hedley-Whyte, and D. S. Feingold. 1973. Prevention of gramnegative bacillary pneumonia using aerosol polymyxinas prophylaxis. I. Effect on the colonization pattern ofthe upper respiratory tract of seriously ill patients. J.Clin. Invest. 52: 2935-2940.

14. Tillotson, J. R., and M. Finland. 1969. Bacterial col-onization and clinical superinfection of the respiratorytract complicating antibiotic treatment of pneumonia.J. Infect. Dis. 119: 597-624.

15. Klastersky, J., E. Huysmans, D. Weerts, C. Hensgens,and D. Daneau. 1974. Endotracheally administered gen-tamicin for the prevention of infections of the respira-tory tract in patients with tracheostomy: a double-blindstudy. Chest. 65: 650-654.

16. Gillies, P. R., and J. R. W. Govan. 1966. Typing ofPseudomonas pyocanea by pyocine production. J. Pathol.Bacteriol. 91: 339-345.

17. Edwards, P., and W. H. Ewing. 1972. Identification ofEnterobacteriaceae. Burgess Publishing Company, Min-neapolis, Minn. 3rd edition. 747.

18. Bauer, A. W., W. M. M. Kirby, S. Sherris, and M.Turck. 1966. Antibiotic susceptibility testing by a stan-dardized single disk method. Am. J. Clin. Pathol. 45:493-496.

19. Hedley-Whyte, J., H. Pontoppidan, and M. J. Morris.1966. The response of patients with respiratory failureand cardiopulmonary disease to different levels of con-stant volume ventilation. J. Clin. Invest. 45: 1543-1554.

20. Teres, D. 1974. ICU-acquired pneumonia due to Flavo-bacterium meningosepticum. J. Am. Med. Assoc. 228:782.

21. Sprunt, K., and W. Redman. 1968. Evidence suggestingimportance of rate of interbacterial inhibition in main-taining balance of normal flora. Ann. Intern. Med. 68:579-590.

Polymyxin Aerosol in the Prevention of Pneumonia 519