A model of catheter-associated urinary tract infection initiated by bacterial contamination of the catheter tip

Authors

  • Jessica M.T. Barford,

    1. Medical Microbiology, Centre for Infection, Division of Cellular and Molecular Medicine, St George’s, University of London, and *Department of Urology, St George’s Hospital, London, UK
    Search for more papers by this author
  • Ken Anson,

    1. Medical Microbiology, Centre for Infection, Division of Cellular and Molecular Medicine, St George’s, University of London, and *Department of Urology, St George’s Hospital, London, UK
    Search for more papers by this author
  • Yanmin Hu,

    1. Medical Microbiology, Centre for Infection, Division of Cellular and Molecular Medicine, St George’s, University of London, and *Department of Urology, St George’s Hospital, London, UK
    Search for more papers by this author
  • Anthony R.M. Coates

    1. Medical Microbiology, Centre for Infection, Division of Cellular and Molecular Medicine, St George’s, University of London, and *Department of Urology, St George’s Hospital, London, UK
    Search for more papers by this author

Anthony R.M. Coates, St George’s, University of London – Medical Microbiology, Cranmer Terrace, London SW17 0RE, UK.
e-mail: acoates@sgul.ac.uk

Abstract

OBJECTIVE

To examine the microbial colonization of urinary catheters that have been used by patients, to model catheter colonization in vitro and thus provide information about the way bacteria gain access to the bladder during catheterization.

MATERIALS AND METHODS

Microbial growth patterns from patients’ indwelling catheters and from catheters used in an in vitro model of the catheterized urinary tract were compared. Catheters were cut into short segments, microorganisms from the inside and outside of each segment of the catheters were removed by sonication, and viable bacteria counted. DNA was extracted from selected patient catheter isolates and the DNA fragment of 16S ribosomal RNA was amplified by polymerase chain reaction and confirmed by DNA sequencing. The DNA sequences from the isolates obtained from different catheter sections and from urine in the same patient were compared.

RESULTS

After 1 day of catheterization there was significant bacterial growth on the outside of all the segments of patient catheters; there was significant growth on the inside of all segments by 4 days. Higher viable counts and a wider spectrum of genera were found on the outside than on the inside of these catheters. The same strains of bacteria, as determined by ≥98% similarity of the 16S ribosomal DNA sequence, were found on the outside and inside of catheters and in the urine. In the in vitro model, when the distal urethra was inoculated before inserting the catheter, the viable counts after incubation were variable along the outside of the catheter and the inside counts were uniformly high. By contrast, there was a different pattern after inoculating the inside of the distal end of the catheter, leading to an ascending biofilm on the inside. A smaller inoculum delayed but did not prevent infection in the model.

CONCLUSION

The present results are consistent with the hypothesis that contamination of the tip of the catheter while it is being inserted is a possible means by which bacteria gain access to the bladder.

Abbreviations
CAUTI

catheter-associated UTI

CFU

colony-forming units

AU

artificial urine

BA

blood agar

NA

nutrient agar

TSB

tryptic soy broth.

INTRODUCTION

Catheter-associated UTI (CAUTI) is a major clinical problem leading to considerable morbidity and mortality [1–3]. An ascending biofilm within the lumen of the catheter has been shown to be a major cause of CAUTI but the introduction of closed drainage systems has reduced the incidence of this route of infection [4,5]. However, in recent years, little further progress has been made in understanding the mechanisms of CAUTI and reducing the morbidity associated with them [6,7].

The precise route of entry of bacteria into the bladder is unclear. There are very few data about the proportion of CAUTI caused by bacteria accessing the bladder along the inside or outside of the catheter, or through contamination of the tip during insertion. Only indirect measures have been used, e.g. assuming that bacteriuria detected within 24 h of catheterization was due to the introduction of organisms during insertion [8]. Since closed drainage systems were introduced, it is thought that most bacteria gain access to the bladder along the outside of the catheter rather than intraluminally from the drainage bag or due to contamination of the catheter/bag junction [5,8,9].

Clinical trials have generally focused on the number of bacteria in the urine [6,10] although a few have used scanning electron microscopy [11] or limited culture to examine the catheters themselves [12,13]. In vitro models of CAUTI have examined bacterial growth on the inside of the catheter and in the urine or in artificial urine [14–17]. No studies have assessed whole-catheter cultures nor studied microbiological differences between the inside and outside of catheters. Thus the aim of the present study was to generate new hypotheses about how CAUTI might be initiated, by the detailed examination of patient catheters and by the development of an in vitro model of the catheterized lower urinary tract.

MATERIALS AND METHODS

The study group of urethral catheters were retrieved from a mixture of urology patients in a Teaching Hospital urology ward between January 2004 and January 2005. Urinary catheters that had been routinely removed after various indwelling times were placed into sterile containers and stored in a refrigerator (for ≤2 days in most cases) until collection. The Wandsworth Research Ethics Committee approved the study protocol (REC reference number: 05/Q0803/139).

Catheters were removed from their containers and placed onto sterile paper with 5 cm sections marked on it. The catheters were then cut into 5 cm sections, designated A to G (Fig. 1). The end of the tip above the drainage holes was removed from A to allow both ends of this section to be sealed, and in some cases this was also cultured. Section G was normally outside the body but above the balloon valve.

Figure 1.

A diagram showing the first step in processing patient catheters. The tip and drainage end were removed and the remainder cut into 5-cm sections. The outside and the inside of each section were then cultured separately.

To examine the bacterial cells adherent to the outside of the catheters, each section was tapped on the paper to remove any excess liquid and each end was dipped into melted wax to seal the ends. The catheter sections were placed into universal containers containing 10 mL PBS (Sigma-Aldrich, Dorset, UK) supplemented with 0.05% Tween 80 (polyoxyethylene sorbitan monooleate; Sigma-Aldrich), shaken briefly and sonicated for 1.5 min in a water bath sonicator. The sections were then removed and placed in 70% ethanol for ≈1 min to kill any remaining organisms on the outside, then removed and allowed to dry. To quantify the bacteria inside the catheters, the ends of each section were removed at the point where the wax covered them, to ensure that the same length of catheter was exposed during sonication on the inside and the outside. The sections were then placed into fresh universal containers with 10 mL PBS with Tween, shaken and sonicated for 1.5 min to remove any organisms from the inside (lumen) of the catheter. Colony forming units (CFU) were estimated for each universal container on blood agar (BA, Oxoid, Basingstoke, Hampshire, UK) plates incubated at 37 °C for 42–48 h. Control catheters (taken straight from their sterile packaging and cultured according to the patient catheter method) were also included to determine contamination levels from the testing procedure.

Bacterial viability was estimated by CFU counts; serial 10-fold dilutions of cultures were made in sterile water and 100 µL dilutions were added to one-third segments of nutrient agar (NA; Oxoid) or BA plates. These were then incubated overnight at 37 °C. Viability was expressed as logCFU/mL or logCFU/cm.

After incubation, the BA plates were examined; for each outside and inside section the three or four most dominant organisms, distinguished by colony morphology, were counted at an appropriate dilution (when individual colonies could be counted). If it was impossible to count individual colonies, further dilutions were carried out and plated.

Any colony type that was recorded as dominant on one section of the catheter was also recorded on other sections if it was found, whether or not it was one of the three or four most dominant in these other sections. If there were more than four colony types per catheter section and they were not found to be dominant in any other section on the same catheter, these were counted and recorded as ‘other’. The total number of CFU per section were also counted at an appropriate dilution.

Each dominant colony type recovered from a particular catheter was subcultured onto BA and incubated at 37 °C overnight or longer if necessary. Bacteria were identified to the genus level using Gram stain, colony morphology, catalase reaction, oxidase reaction and bile aesculin hydrolysis. Organisms that were from the same genus but with different colony morphology were considered to be different strains.

To examine if the same strains of bacteria were isolated from different locations on the catheter or urine, catheter isolates from four patients were subjected to DNA analysis. A single colony of the selected isolates from different locations on the same catheter was inoculated into 10 mL of Bacto tryptic soy broth (TSB; Becton Dickinson Co., Oxford, UK) and incubated at 37 °C on a shaker with shaking at ≈110 rpm overnight. Then 109−1010 bacterial cells were harvested by centrifugation at 3116 g for 10 min. The cell pellet was resuspended with 200 µL of water and then added to a 2-mL tube containing 500 µL of detergent solution (9.6 mL Divolab no. 1, 90.4 mL 10 mm Trizma and 1 mm EDTA buffer pH 7.5), 500 µL of phenol (pH 8.0), 100 µL of chloroform/isoamyl alcohol (24:1) and glass beads (75–150 µm, acid washed, Sigma-Aldrich) and then lysed in a reciprocal shaker for 45 s at 6.5 speed (Hybaid, Hants, UK). Cell debris was removed by centrifugation at 13 000 rpm in a microcentrifuge for 10 min. The aqueous phase was carefully removed and re- extracted with an equal volume of chloroform/isoamyl alcohol by vortexing the tube for 15 s. DNA was precipitated by adding 0.6 mL of isopropanol at −20 °C overnight.

DNA fragments of 16S ribosomal RNA were amplified by PCR. The primers used were EUB f933 (GCACAAGCGGTGGAGCATGTGG) and EUB r1387 (GCCCGGGAACGTATTCACCG) [18]. PCR was performed in a total volume of 50 µL containing 250 µm each of dATP, dCTP, dGTP and dTTP, 1 µm of each primer, ≈10 ng of DNA, 1 unit of Hotstar Taq polymerase (Qiagen, Crawley, West Sussex, UK) and the buffers supplied with the enzyme. The PCR amplification was carried out for 30 cycles (94 °C 1 min, 55 °C 2 min and 72 °C 3 min), followed by a final extension of 10 min at 72 °C. The sequences of the purified PCR products were determined with a DNA sequencer by using the AmpliTaq FS enzyme for cycle sequencing with d-Rhodamine dye terminators (Qiagen, Hilden, Germany). A homology search was performed in the database through the National Center for Biotechnology Information by using the BLAST algorithm [19]. The DNA sequences were compared using the ClustalW program [20].

The in vitro model of a catheterized lower urinary tract was set up to create realistic flow conditions for investigating factors such as the site of inoculation and inoculum size, and their effect on bacteriuria and catheter colonization. The model was set up according to (Fig. 2), on a trolley. Standard length, 16 F uncoated latex Foley catheters (Metacot AB, Stockholm, Sweden) were coated with a sterile water-based lubricating jelly (Aplicare Inc., West Virginia, USA or Aquagel, Adams Healthcare, Leeds, West Yorkshire, UK), then inserted using sterile forceps into the inoculated urethra tube, balloon-end first (Fig. 2). The catheter was pushed through and when the balloon protruded from the other end, a rubber ring cut though one side was placed on the catheter below the balloon (to prevent the leakage that occurred during preliminary tests when the balloon was inflated, and mimicking the internal urethral sphincter). The attached screw-cap was then screwed onto the bladder chamber, adjusting the catheter so the balloon was just inside the chamber.

Figure 2.

Photograph of the catheterized urinary tract model, and a diagram showing the detail of one of the bladders with catheter in situ.

Sterile catheter bags (Simpla Plus, 2 L, with lever action tap, 1000 mm tube) were then attached to the free end of the catheters and tied to the trolley so that they were hanging below the level of the catheters. The taps (off position) on the bottom of the bags were attached to tubing, which allowed drainage into the waste container.

After adding artificial urine (AU, 5 L) to the reservoir chamber, the entire apparatus on the trolley was wheeled into a hot-room which was set at 37 °C. The waste container was placed on the floor to obtain optimum drainage. The pump was operated at 40–50 mL/h and the model left running for 24 h. The catheter bags were emptied into the waste container by turning the taps after 9 h.

A clinical isolate of Escherichia coli (SGH031) from a UTI (isolated at St George’s Hospital, London), was used in all in vitro experiments. For overnight broth cultures, one colony was inoculated into 10 mL of TSB, then incubated at 37 °C overnight with shaking.

The AU was based on Difco urea broth and contained 0.1 g of selected yeast extract (Gibco BRL Life Technologies, Paisley, UK), 9.1 g potassium phosphate, monobasic (Sigma-Aldrich), 9.5 g potassium phosphate, dibasic (Sigma-Aldrich) and 10 g urea (BDH, VWR International Ltd, Poole, Dorset, UK) in 1 L of distilled water, sterilized by filtration.

For the inoculum, an overnight broth culture of E. coli was diluted 10 000-fold in AU. For some experiments, this was diluted 10- and 100-fold further; 100 µL of the inoculum was placed just inside the distal end of the urethra tube on the latex cuff.

Samples from the bladder and catheter bags were taken at regular intervals, e.g. 2, 4, 6, 8, 10 and 24 h, by unscrewing the spare inlet and removing 2–3 mL from the bladder with a pipette. The waste tubing was detached from the catheter-bag tap, then the tap was turned on to fill a universal container with 5–10 mL AU. The tubing was replaced after this procedure.

After the model had been running for 24 h, the pump tubing was diverted so that AU dripped into a 50-mL tube. The amount collected in 20 min was measured and the flow-rate calculated. The trolley with the rest of the apparatus on it was wheeled out of the hot-room into the laboratory.

The urethra tube was unscrewed from the bladder chamber, the catheter pushed slightly from the bottom to enable the rubber ring to be removed, then the catheter was drawn out of the urethra by the drainage end and placed on sterile paper. The catheters were cultured according to the patient-catheter method, except for plating the dilutions on NA instead of BA, incubating for 16–24 h, and only counting E. coli. Model experiments were done with different initial concentrations of bacteria and periods of incubation.

The viability of bacteria was expressed as CFU/mL for bladder and catheter bag samples or CFU/cm for the catheter counts. CFU counts were transformed by taking the logCFU count plus 1 and then used for statistical analysis, including the mean and sd, plotted on bar charts. The data were analysed using t-tests or Wilcoxon signed-ranks tests for ordinal data or one-way anova, to compare different conditions.

RESULTS

Preliminary tests of the method involving incubating catheters with Enterococcus on the inside and E. coli on the outside before processing, showed that there was very little cross-contamination between the inside and outside. The results show that there was almost 20 times more E. coli than Enterococcus on the outside after processing and, no E. coli but almost 105 CFU/cm of Enterococcus on the inside (data not shown). These tests also showed that the ethanol treatment of the outside of catheter sections reduced viable bacterial numbers by >2000 times. To control for contamination during the long and detailed method used for culturing the catheters, 10 control catheters were studied, which showed very few contaminating bacteria.

Microbiological data from 34 catheters from 33 patients were assessed; the catheters had been in place for up to 5 days (mean 2.2 days). The mean (range) age of the patients was 65 (23–88) years and the group included 29 men and four women. Figure 3 shows the mean total CFU/cm of each catheter segment for catheters in situ for 1, 2, 3, 4 and 5 days. After 1 day there was a significant difference in the number of bacteria cultured from the outside of the patients’ catheters compared to the controls (P < 0.005) but there was no significant difference between the inside CFU and the control. CFU counts on both the inside and outside of the patient catheters were higher after 4 or 5 days. More bacteria were cultured from the outside than the inside for all periods except 4 days, and a paired t-test showed that the mean difference was statistically significant (P < 0.001).

Figure 3.

Summary of bacterial growth from 34 catheters from patients that had been catheterized for up to 5 days, plus 10 control catheters. Error bars, sd; blue rhombus, outside; red square, inside; turquoise triangle, outside control; yellow cross, inside control.

There were significantly more different strains (subtypes) and genera (groups of closely related species) of bacteria on the outside of the catheters than on the inside (Table 1) (both P < 0.001).

Table 1. 
A comparison of the number of different strains and genera of microorganisms cultured from the outside and inside of patient catheters
Number ofNMedian (range)P, Wilcoxon signed-ranks test
Strains
 Outside347.0 (3–16)<0.001
 Inside344.5 (0–12) 
Genera
 Outside343.0 (1–6)<0.001
 Inside342.5 (0–6) 

Figure 4 shows the same data in more detail, with mean counts for each section of the catheter for each period that the catheters were in situ. Bacterial counts were fairly even along the outside of the catheters, but with a slight peak around section F, roughly where the catheter exits the body (Fig. 4a). On the inside the counts were also fairly even along the catheters, but there was a wider range of bacterial densities, and this appears to depend on the length of time the catheters were in place (Fig. 4b).

Figure 4.

Bacterial growth from the outside (a) and inside (b) of different sections of patient catheters. Data from 34 catheters from patients who had been catheterized for up to 5 days were used. Also shown are the counts from the in vitro model when incubated for 1 and 7 days, for comparison. ‘A’ is the tip inside the bladder and ‘G’ is adjacent to the connection with the drainage bag; blue rhombus, 1 day (15 samples), pink square 2 days (eight), yellow triangle 3 days (two), turquoise cross 4 days (six), purple asterisk 5 days (three), brown circle control (10), green plus sign, 1 day (model), blue dash, 7 days (model).

DNA was extracted from seven Staphylococcus isolates from catheter 1 (an additional catheter in situ for 17 days) and a further isolate from the urine of the same patient. The 16S ribosomal DNA sequences for the eight Staphylococcus isolates were compared and four of these, including the isolate from the urine, were ≥99% identical (Table 2). The others all had a similarity of ≤98%.

Table 2.  Genotypic matches of selected isolates from different catheter sections with those found in the urine of patients. DNA was extracted and a section of the 16S ribosomal DNA was amplified using PCR. The products were sequenced and compared using ClustalW. Staphylococcus strains were considered to be identical if there was ≥99% similarity, and for the other organisms a similarity of ≥98% was assumed to indicate identical strains. ‘A’ is the tip inside the bladder and ‘G’ is adjacent to the connection with the drainage bag
Catheter (days of catheterization)SpeciesCatheter sectionUrine
ABCDEFG
  1. I, inside; o, outside.

1 (17)S. epidermidis+ (o + i)    + (o) +
1 (17)P. putida+ (i)      +
2 (22)K. oxytoca + (i)     +
3 (4)Ent. faecalis+ (o)      +
4 (4)Ent. faecalis+ (i)+ (o)     +

Table 2 also shows that a Pseudomonas putida isolate from the inside of section A of the same catheter (1) had a 98% similar sequence to the P. putida isolate from the urine. Similarly, a Klebsiella oxytoca isolate from the inside of section B of a different catheter (2, an additional catheter in situ for 22 days) had a 98% similar sequence to the Klebsiella isolated from the urine of the same patient.

The sequence from an Enterococcus isolate from the outside of section A of catheter 3 (in situ for 4 days) was 98% identical to one from the urine isolate from the same patient (Table 2). Enterococcus isolates from another catheter (4, in situ for 4 days) were compared in a similar way. The sequence from an isolate from the outside of section B was 99% identical to one from the urine and these were 98% identical to an isolate from the inside of section A.

To determine the shape of the growth curve of bacteria in the model, it was run for a week with samples taken every 2 h for the first 10 h and twice daily thereafter. Figure 5a shows that the bacterial numbers rapidly increased during the first 24 h, then levelled off and remained constant for ≥180 h after inoculation. Bacterial growth in the bladder and catheter bag followed the same pattern.

Figure 5.

(a) CFU counts in the bladder and catheter bag when the distal urethra of the model was inoculated with E. coli before catheter insertion at time 0 and incubated for 7 days. Mean of three catheters; error bars, sd; green rhombus, bladder; pink square catheter bag. (b) CFU counts from the outside and inside of the catheter when the distal urethra of the model was inoculated with E. coli before catheter insertion and incubated for 7 days. Mean of three catheters; error bars, sd; blue, outside; red, inside.

The study catheters were removed from the model after 180 h and cultured in separate sections (as described for the patient catheters) to show the patterns of growth that E. coli produces (Fig. 5b). A similar density of bacteria was reached in the inside of all sections of the catheter, whereas the outside was more variable, shown by the larger sd s. The highest levels on the outside were at the tip end (sections A and B) because these sections were within the bladder, but they were lower than on the inside. There were also relatively high numbers around where the urethra ends (section F).

The catheter CFU counts after 1 and 7 days of incubation in the model were also plotted with the patient catheter CFU counts for comparison (Fig. 4). As seen in Fig. 4a the outside catheter counts from the model were more variable along the catheter than were the patient counts, but spanned a similar although wider range. The inside catheter counts in Fig. 4b from the model showed a similar even distribution along the catheter as the patient counts. However, the counts were much higher; even the 24 h incubation of the model catheters gave higher CFU counts than catheters that were in situ in patients for 4 or 5 days.

In separate experiments, the catheter bag tubing was inoculated on the inside, just below the catheter. After 24 h E. coli was found only on the inside of the catheter, with more lower down (data not shown). This clearly results in a different pattern of growth than when the distal urethra was inoculated. In addition, the effect of inoculum size on the growth of bacteria in the model was studied. Three different concentrations of E. coli were pipetted into the end of the urethra before the catheter was inserted. Figure 6 shows that the smaller the inoculum the more delayed the growth of E. coli was in the bladder. However, by 24 h after inoculation the numbers of bacteria reached the same level (P = 0.9) regardless of the initial inoculum size. One-way anova showed that this was significant at 2, 4, 6 and 8 h (P < 0.01) and at 10 h (P < 0.05). The pattern was similar with CFU counts from the catheter bags. Finally, there were no differences in any individual section or the total CFU count from each catheter, inside or outside, when catheters with different inocula were compared.

Figure 6.

E. coli counts in the bladder of the model when different numbers were inoculated into the end of the urethra before catheter insertion. Mean of three experiments; error bars, sd; blue, 34 000 CFU; turquoise, 3600 CFU; yellow 330 CFU inoculum.

DISCUSSION

In this study we describe a novel method of culturing bacteria adherent to urethral catheters, and provide detailed information about the colonization of the catheter on the outside and inside separately. The culture of patient catheters suggests that a mixed biofilm forms on the outside of the catheters (Table 1 and Fig. 3). The biofilms contained 3–16 different strains of bacteria. These organisms might originate from those colonizing the external urethral meatus, and which are smeared up the urethra as the catheter is being inserted. These grow into a biofilm over the whole external surface of the catheter. The mixed nature of the genera in the biofilm suggests their origin is periurethral skin [21]. The early appearance of bacteria on the outside of the tip after only one day (Fig. 4a) suggests that contamination on insertion is the mechanism of infection, rather than a biofilm originating from around the external urethral meatus ascending up the outside of the catheter. The relatively even distribution of CFU along the whole catheter on days 1–5 supports the hypothesis that the contaminated tip smears bacteria evenly up the urethra during insertion. Could this effect be due to bacteria around the external urethral meatus, being smeared up the outside of the catheter when it is withdrawn? This is unlikely to be the case, as bacteria were also found on the inside of the catheter before withdrawal. The data in Table 2 suggest that the origin of the bacteria inside the catheter is from contaminating bacteria on the outside of the tip. The flow of urine carries them through the hole in the tip into the lumen of the catheter, where they adhere to the inside surface. It is improbable that the bacteria on the inside of the tip come from an ascending biofilm, because this would form a gradient of CFU, the highest being towards the distal end and the lowest at the tip. Figure 4b clearly shows an even internal biofilm from tip to distal end, which strongly argues against an ascending infection. Furthermore, significant bacterial growth (Fig. 3) appears at 1 day on the outside of patient catheters, but not until 4 days on the inside, strongly suggesting that infection of the outside occurs before growth on the inside.

It is also possible that bacteria might already be present higher up the urethra and colonize the catheter from this position. However, studies have shown that only the distal few centimetres are colonized in healthy people [21], although, in the patient group being studied, previous catheterization or instrumentation might cause the urethra to be colonized higher up.

What is the relationship between the bacteria in the bladder and those on the catheter? Our hypothesis is that the bacteria on the outside of the tip colonize the bladder epithelium. Data from other laboratories partly support this hypothesis. For example, perineal bacteria have been shown to access the inside of the catheter, presumably by either colonizing the tip on catheterization or ascending the outside of the catheter [9]. With the flow of urine, bladder bacteria or those from the biofilm in the drainage hole of the catheter (which come from the outside of the tip), are carried by the flow into the lumen of the catheter. Here, some of them attach to the inside walls of the catheter, whilst others are flushed out in the urine. So the bacteria in the urine are likely to originate, at least in part, from the bladder. Curiously, there are fewer genera of bacteria inside the catheter than outside. This suggests that there might be some kind of selection of genera for bladder growth and multiplication in urine. After long periods, more bacteria can grow inside than outside and this might be due to the absence of immune cells, or simply more nutrients in the urine. Only 34 catheters were examined in this study, so it is possible that different patterns of bacterial growth might emerge with a larger sample.

The DNA analysis of patient catheter isolates showed that the same strains can be found on the inside and outside of catheters and in the urine, with the assumption that a ≥98% sequence similarity indicates the same strain (≥99% for Staphylococcus isolates). For example, catheter 1 (Table 2) has the same strain of S. epidermidis on the outside and inside of section A, which is the tip inside the bladder, and on the outside of section F, which is likely to be at the external urethral meatus, as well as in the urine. This suggests that the bacteria that colonize the catheter and cause bacteriuria have a common source, most likely the flora of the distal urethra. In three further catheters (2, 3 and 4 in Table 2), Klebsiella and Enterococcus isolates provide additional support, as the same strains were found both on the catheter and in the urine. It is unlikely that the bacteria causing bacteriuria have come from an external source, e.g. from contamination of the catheter bag or the junction between the catheter and catheter bag tubing, as in this case it would not be expected to find the same organism on the outside of the catheter, as was found with the Staphylococcus and Enterococcus isolates.

These examples are consistent with the hypothesis that bacteria might originate from the distal urethra and be pushed into the bladder when the catheter is inserted. They can then multiply in the urine and colonize the inside of the catheter as the urine drips through it. The origin of the bacteria on the catheters could be determined in future studies by culturing swabs of the external urinary meatus and distal urethra before catheter insertion.

A further explanation for finding the same strains on different areas of the same catheter is that contamination from one section to another has occurred during the processing of the catheter. Contamination between the catheter samples and urine samples is very unlikely, as they were processed separately. Enterococcus was cultured from catheters 3 and 4, but it is not known whether there was cross-infection between these patients.

We also describe a novel model of the catheterized lower urinary tract which, unlike previous models [14–17], includes the urethra. Catheters from this model were cultured using the same method as was used for the patient catheters. There was a similar pattern of bacterial growth between this model and patient catheters when the distal urethra was inoculated before the catheter was inserted (Fig. 4). This similarity was not found, e.g. when contamination of the catheter/bag junction was simulated (data not shown). We propose that contamination of the tip of the catheter during insertion cannot be excluded as a route by which bacteria are introduced into the bladder. This conclusion is in agreement with other authors who have found that intraluminal ascending infection has decreased since the widespread adoption of closed drainage [5,8,9]. However, most authors have also concluded that bacteria gain entry to the bladder by ascending the outside of the catheter/urethra after the catheter has been inserted [5,8,22]. The present study provides evidence that contamination of the tip upon insertion also needs to be considered when designing new catheters or other interventions to prevent infection.

The main difference found between the patient catheters and the model was the rate of growth of the bacteria. As shown in Fig. 5, most growth in the model occurs in the first 24 h (up to ≈108 CFU/mL) and only very gradually after that. However, growth of bacteria on the patient catheters appears to have a much slower time course, with only 105 CFU/cm being recovered from the inside of patient catheters after 4 days, and 107 CFU/cm from the inside of the catheters from the model after 1 day of incubation (Fig. 4b). Possible reasons for this discrepancy include the composition of urine (AU might contain more nutrients and have a pH more suitable for the growth of bacteria than real urine) and the lack of immune response (macrophages or secretory IgA) which might slow the rate of bacterial growth in vivo. Another unrealistic feature of this model is the rigid glass urethra. Presumably in vivo, the urethra is more moist and flexible, although there is little published evidence to show what the catheterized urethra is really like. A preliminary experiment was carried out to determine whether bacteria inoculated in the distal urethra after the catheter was inserted would ascend into the bladder (data not shown). This test showed that no bacteria ascended the outside of the catheter. These data need to be interpreted with caution, as the conditions in the model might have been too dry. However, even after applying lubricating gel to the outside of the catheter, bacteria still did not ascend the urethra.

Experiments with the in vitro model showed that when the distal urethra is inoculated before catheter insertion, a smaller inoculum delays but does not prevent infection (Fig. 6). Even when only 300 bacteria were placed into the urethra and fewer than one viable bacterial cell/mL was present in the bladder urine after 4 h of incubation, there were >107 CFU/mL after 24 h. This final count was the same as that when an inoculum 100 times larger was used. However, a delay of a few hours in the model might translate into a much longer period in a patient, due to the slower time course in vivo. This delay in bacterial growth might be beneficial in a patient, as it might prevent infection during a short-term catheterization. Also, other factors in vivo might be more effective in preventing infection if the bacterial inoculum is small, such as the immune system (macrophages, IgA, etc.), mucus and shedding of cells [23]. In the future this in vitro model will be used to examine the influence of differing catheter materials or other interventions on infection rates.

In conclusion, a novel method was devised to culture bacteria from different sections of urinary catheters from patients. In addition, a new model of the catheterized lower urinary tract was developed. These two novel techniques have proved valuable in testing various hypotheses of CAUTI. It appears from this original work that contamination of the tip of catheters during insertion is one of the main mechanisms of bacterial bladder inoculation. We propose a scientific explanation for the previous clinical observations of early bacteriuria within 24 h of catheterization, and go some way to dispel the myth of ascending bacterial growth along an indwelling catheter as the main source of CAUTI. These research methods will hopefully prove useful in future studies assessing new catheter materials and designs.

ACKNOWLEDGEMENTS

We are grateful to the Pathology Research Fund (St George’s) for supporting J.B., to Metacot AB for the supply of catheters, to Gunnar Walstam for helpful discussion, to Peter Fraser for helping to build the model, to Richard Latham for help with laboratory work, to Joanna Philips for collecting patient consents and clinical data, to Jan Poloniecki for statistical advice and to all the nurses on Vernon ward, St George’s Hospital, London, for collecting the catheters.

CONFLICT OF INTEREST

Source of funding: see Acknowledgements.

Ancillary