In the neonatal intensive care unit (NICU) of one hospital, 16 infants became colonized or infected with multiply-resistant Escherichia coli (MR-E.coli) over an 8-month period. Isolates were obtained from blood, urine, and sputum of three patients and from rectal surveillance cultures of 13 patients. The one patient with the blood isolate died. A matched case-control study identified continuous feeding (nine of 16 cases vs. one of 16 controls, p less than or equal to 0.001) and receipt of aminoglycosides (p less than or equal to 0.03) as risk factors. For case-babies not exposed to continuous feeding, duration of bolus feeding was significantly greater than for their controls (cases, 22 days; controls, 7 days; p less than or equal to 0.02). All 16 isolates were the same serotype and were resistant to amikacin, tobramycin, kanamycin, and gentamicin. The epidemiologic investigation suggested that MR-E. coli may have spread from person-to-person on the hands of personnel and that MR-E. coli persisted in the NICU for 8 months until effective control measures were instituted.

Clinical isolates of Klebsiella pneumoniae and Serratia marcescens at a hospital that had used amikacin as its principal aminoglycoside for the preceding 42 months demonstrated high-level resistance to amikacin (greater than or equal to 256 micrograms/ml), kanamycin (greater than or equal to 256 micrograms/ml), gentamicin (greater than or equal to 64 micrograms/ml), netilmicin (64 micrograms/ml), and tobramycin (greater than or equal to 16 micrograms/ml). The resistant strains contained an identical 6.8-kilobase plasmid, pRPG101. Transformation of pRPG101 into Escherichia coli produced high-level resistance to amikacin (greater than or equal to 256 micrograms/ml) and kanamycin (greater than or equal to 256 micrograms/ml) but unchanged susceptibilities to gentamicin, netilmicin, and tobramycin. The clinical isolates and transformants produced a novel 3'-phosphotransferase, APH(3'), that modified amikacin and kanamycin in vitro. The location and orientation of the amk gene encoding this APH(3') were determined by analysis of insertions in pRPG101 of the defective gene fusion phage Mu dII1681 (mini-Mulac). Cells containing plasmids with insertions into amk that had the lac operon fused to the amk promoter were selected as Lac+ and amikacin susceptible. A collection of these mini-Mulac insertions was mapped by restriction enzyme analysis. This characterization of amk facilitated its cloning as a 1.8-kilobase EcoRI-Bg/I fragment of pRPG101 into the pUC19 vector. E. coli strains containing this recombinant plasmid had APH(3') activity and demonstrated high-level resistance to amikacin and kanamycin (greater than or equal to 256 micrograms/ml) but were as susceptible to gentamicin, tobramycin, and netilmicin (less than or equal to 1.0 microgram/ml) as the strains harboring the original pRPG101 plasmid.

We investigated an increase in the number of patient specimens yielding Mycobacterium terrae in 1986, isolation of M. terrae was associated with specimens obtained from inpatients at our new hospital, but not with specimens referred from other hospitals [37(+)/144 inpatient specimens versus 2(+)/26 referred specimens, p less than 0.05]. By October 31, 1987, we had identified 163 positive specimens from 131 patients. All M. terrae were isolated from specimens obtained from non-sterile sites, i.e., respiratory, gastrointestinal, or urine. No clinical disease related to M. terrae occurred. Review of procedures and cultures of solutions used in the Microbiology Laboratory suggested the source of M. terrae was not in the Microbiology Laboratory. An analysis of case location showed an association with hospital tier (p less than 0.05), a pattern matching the design of the potable water system of the hospital. M. terrae was cultured from multiple outlets of this system. There appeared to be multiple modes of transmission of M. terrae from this reservoir. Control measures included avoidance of water sources during specimen collection and hyperchlorination of the potable water system. These measures appeared to result in the disappearance of M. terrae from subsequent clinical specimens. We believe this to be the first report defining the epidemiologic aspects of M. terrae contaminating clinical specimens.

Surveillance for bacteremic or pyrogenic episodes associated with hemodialysis was undertaken before and after the reconstruction of the water treatment system at our University medical center. The new water system included a holding tank with iodination treatment. The water delivered to individual dialysis stations had only occasional positive bacterial cultures (3 of 21 samples before completion of construction, 2 of 16 samples afterwards) and intermittent detection of endotoxin (6 of 21 samples before completion of construction, 9 of 16 samples afterwards) at monthly sampling. Among 51 individual dialysis treatments (25 patients) before reconstruction and 56 treatments (29 patients), after, only 2 and 3 febrile events were identified, respectively. All of these were associated with underlying infectious illness and not with the hemodialysis procedure itself. Overall, we conclude that pyrogenic episodes associated directly with hemodialysis treatment are infrequent, and that the addition of a water storage tank with iodination treatment does not appear to increase the risk of bacteremia or pyrogenic episodes.