A prospective, randomized, controlled trial comparing transparent polyurethane and hydrocolloid dressings for central venous catheters

A prospective, randomized, controlled trial comparing transparent polyurethane and hydrocolloid dressings for central venous catheters Suzanne Nikoletti, RN, BScHons, PhDa Gavin Leslie, BApSc, PGDipClinN, FRCNAb Sylvia Gandossi, RN, BNc Geoffrey Coombs, BApSc(MedTech), PGDip(BiomedSc)b Roger Wilson, MBChB, FRCAP, DipMicrobiold Churchlands and Myaree, Western Australia, and Penrith, New South Wales, Australia

Background: This study was undertaken to determine the frequency of skin colonization, hub colonization, and central venous catheter colonization in transparent hydrocolloid versus standard polyurethane dressings. Methods: Adult patients requiring the insertion of a multilumen central venous catheter in an intensive care unit were randomized to receive either a standard polyurethane dressing or a transparent hydrocolloid dressing. Cultures were obtained from 125 skin insertion sites, 141 catheter hubs, 128 catheter tips, and blood samples from 132 patients. Extensive data on patient and catheter characteristics were collected. Results: Skin and hub cultures revealed no significant difference in degree of colonization. However, the hydrocolloid group had a significantly higher level of catheter colonization than the polyurethane group (P = .048). Conversely, there was a significantly higher frequency of positive blood cultures in the polyurethane group (P = .03), although the majority were considered to be potential contaminants. There were only 6 cases in which the same species was simultaneously isolated from a positive blood culture and a colonized catheter, 5 from the hydrocolloid group and 1 from the polyurethane group. Conclusions: The results of this study suggest that an increased risk of catheter colonization is associated with the use of hydrocolloid dressings, despite previous research suggesting that they significantly reduce microbial growth compared with standard polyurethane. The clinical significance of increased numbers of positive blood cultures in the polyurethane group requires further examination, although distinguishing between contamination and true infection in intensive care settings continues to be methodologically challenging. Further studies are required to determine whether these findings are generalizable across different study settings and whether similar outcomes are obtained when different brands of hydrocolloid dressing are used. (AJIC Am J Infect Control 1999;27:488-96)

The type of insertion site dressing is one of the many risk factors associated with central venous catheter infections, yet there is no consensus on the optimal dressing type despite more than a decade of research and debate. Hoffman et al1 attempted to resolve this debate by meta-analysis of studies fulfilling specific methodologic criteria. Their analysis concluded that

From Edith Cowan University, Churchlands, Western Australiaa; Royal Perth Hospital, Western Australiab; formerly Royal Perth Hospital, currently at Western Diagnostic Pathology, Myaree, Western Australiac; and formerly Royal Perth Hospital, currently at Nepean Hospital, Penrith, New South Wales.d Funded by Edith Cowan University and the Olive Anstey Nursing Fund. Hydrocolloid dressings were supplied free of charge by Coloplast Pty Ltd (Australia). Reprint requests: Mr Gavin Leslie, Nurse Coordinator, Critical Care Division, Royal Perth Hospital, GPO Box X 2213, Perth, Western Australia 6001. Copyright © 1999 by the Association for Professionals in Infection Control and Epidemiology, Inc. 0196-6553/99/$8.00 + 0

488

17/46/95513

the widely used transparent polyurethane dressings (Tegaderm and Opsite) were associated with a significantly increased risk of catheter colonization compared with gauze and tape dressings. Maki2 claims that this meta-analysis was flawed by the inclusion of studies with noncomparable groups and by the omission of more recent comparative trials. The risk associated with polyurethane dressings is thought to be a result of pooled fluid such as blood, exudate, and sweat accumulating under the semipermeable dressing. Microorganisms proliferate in such environments and may then migrate along the catheter surface to the catheter tip. This problem has been addressed by the development of Opsite IV3000 (Smith and Nephew, United Kingdom), which has a moisture vapor transmission rate 3 to 8 times higher than conventional polyurethane dressings.3 However, one of the largest prospective, randomized, controlled trials that compared gauze with standard polyurethane and Opsite IV3000 did not show significant differences in risk.4 Smaller studies have concurred with this finding; they have not conclusively shown significant advan-

AJIC Volume 27, Number 6

tages of Opsite IV3000 over conventional polyurethane or gauze dressings for a variety of central venous catheters.5-7 Little research has been conducted on alternatives such as transparent hydrocolloid dressings. These are normally used on open wound sites to promote moist healing, but at the same time, the hydrocolloid matrix absorbs excess moisture away from the skin surface. A study by Haffejee et al8 examined the effect of transparent hydrocolloid dressings on central venous catheter sites of 8 critical care patients in South Africa. The study demonstrated a statistically significant decrease in the number of organisms detected under the dressings during a period of 2 months but was limited by the small sample size and lack of data on catheter colonization. The present study evaluates the risk of infection associated with a thin, transparent hydrocolloid dressing (Comfeel, Coloplast Pty Ltd, Australia) compared with the conventional transparent polyurethane dressing (Tegaderm, 3M, St Paul, Minn). METHODS Sample, setting, and design The study was conducted as a prospective, randomized, controlled trial during 16 months from September 1994 in the 24-bed intensive care unit of a 900-bed metropolitan teaching hospital in Perth, Western Australia. Institutional ethics committee approval was obtained before commencing the study. The sample comprised 204 patients older than 18 years who required insertion of a multilumen central venous catheter in the intensive care unit. Catheters were ineligible for inclusion if they were inserted for fewer than 24 hours, inserted outside the intensive care unit, exchanged over a guidewire, or if they were tunneled or implantable. Arterial catheters and pacing wires were also excluded. Insertion of catheters followed a standard aseptic technique that involved the use of sterile gloves, drapes, masks, gown, and disinfection of the insertion site with 0.5% chlorhexidine in 70% alcohol. After insertion, catheters were randomized to receive either a conventional polyurethane dressing or a transparent hydrocolloid dressing. All catheters were cared for by swabbing with 0.5% chlorhexidine in 70% alcohol and allowing the skin to dry before application of a fresh dressing. Dressings were changed every 5 days or earlier if soiled or nonadherent. Antimicrobial ointments were not used. Data collection Data relating to patient and catheter characteristics were entered onto data collection sheets by nursing staff or a part-time research nurse. Risk of infection was determined by measuring skin colonization levels at the catheter insertion site and by culture of the

Nikoletti et al 489 catheter tip after removal. In addition, culture of the hub was undertaken to determine whether the 2 dressing groups had an equal risk of catheter colonization via the intralumenal route. Microbial growth on the skin was cultured by collecting agar impressions before skin cleansing, before insertion of the catheter, and before its removal. The device used for obtaining agar impressions was specifically prepared for the study, using 20 mL freshly poured horse blood agar in sterile 50 mL syringes from which the tops had been removed. These agar syringes were stored at 4° centigrade with the plunger retracted to protect the agar surface. The tops of the syringes were sealed with aluminum foil. As many as 50 colonies could be counted accurately on the 30 mm surface. Colony counts exceeding 50 colonies were categorized as either >50 colonies but less than confluent, or confluent. In a minority of cases (7 in the polyurethane group and 5 in the hydrocolloid group), samples were excluded because bacterial swarming obscured the agar surface. Skin colonization as an outcome measure was expressed semiquantitatively by comparing growth at catheter removal with the baseline growth before insertion (increase, decrease, or no difference). All hubs of the multilumen catheters were swabbed immediately after removal with a single swab moistened with sterile saline. Swabs were cultured directly onto horse blood agar in the laboratory and incubated in 5% carbon dioxide. Growth from hubs was scored as no growth, <20 colonies, ≥20 colonies but less than confluent, or confluent. Catheter tips were removed by cutting off the distal segment (approximately 5 cm) and cultured by the semiquantitative method of Maki et al.9 The presence of 15 or more colonies was defined as catheter colonization. Blood cultures were obtained when signs of possible sepsis such as fever, unexplained tachycardia, hypotension, or tachypnea were observed and when ordered by the attending physician. In such cases, 20 mL of blood was drawn from a peripheral site after cleansing with 0.5% chlorhexidine in 70% alcohol. A culture set was prepared by dispensing 10 mL aliquots into standard aerobic and anaerobic bottles. Cultures were incubated at 37° centigrade in a continuous monitoring system (BacTAlert Blood Culture System, Organon Teknika, Durham, NC) according to the manufacturer’s instructions. Species were identified by using conventional and automated identification systems. Laboratory records were retrospectively examined to determine whether there was any temporal relationship among positive blood cultures, colonized catheters, and colonized hubs in the 2 dressing groups. Temporal relationship was defined as a positive blood culture taken at any time during the period of catheterization. If positive blood cultures were recorded for any time as long

AJIC

490 Nikoletti et al

December 1999

Table 1. Frequency of reasons for exclusion, according to dressing group

Protocol violation Missing data Insertion for <24 hours Catheter not inserted Coroner’s case Allergy

Polyurethane (%)

Hydrocolloid (%)

n = 78

n = 77

3 61 8 2 3 1

(4) (78) (10) (3) (4) (1)

7 58 1 3 6 2

(9) (75) (1) (4) (8) (3)

as 2 days before insertion of the study catheter, any subsequent positive culture of the same species was not counted as being temporally related. Statistical analysis The randomization of patient and catheter characteristics between the 2 dressing groups was assessed by testing for significant differences in each patient and catheter characteristic between the 2 groups, by using χ2 analysis or the 2-tailed Fisher exact test for categorical variables and the Mann-Whitney U test adjusted for ties for continuous variables. The significance of differences in levels of colonization at each site was determined by χ2 analysis or the 2-tailed Fisher exact test. RESULTS Characteristics of the study population Of 310 catheters entered into the trial, data were available from 155 catheters. Of the 310 catheters, 119 (38.3%) were excluded because of missing data, 10 (3.2%) were excluded because of protocol violation, 9 (2.9%) were excluded because of insertion for fewer than 24 hours, 9 (2.9%) were unavailable because of coroner’s investigation, 5 catheters (1.6%) were not inserted, and 3 (1%) were excluded because of allergy to either dressing. Exclusions were evenly distributed between the 2 dressing groups (Table 1). For statistical analysis, the various categories of reasons for exclusion were collapsed into a single category “exclusions,” and the characteristics of patients and catheters were compared with those of the study participants. There were no significant differences for all characteristics listed in Table 2, with 1 exception; the median number of days in hospital was higher among those included (26 days) than in those excluded (24 days). The magnitude of this difference was small but significant (P = .007). There was no significant difference in the proportion of each dressing type in those excluded versus those included (Fisher exact test = .734). Of the 155 available catheters, cultures were obtained from 125 (81%) skin insertion sites, 141 (91%)

catheter hubs, and 128 (82%) catheter tips. Blood samples from 132 patients were available for analysis. Summary statistics of the patient and catheter characteristics are shown in Table 2. The results indicate that there were no significant differences between the 2 groups across the range of variables, confirming randomization of the allocation. The susceptibility of this group of patients to nosocomial infection is reflected by the median age of 55 to 56 years, the APACHE II10 score of 19 to 20, the median length of stay of 24 to 27 days, and in all cases, the presence of other invasive devices. In each group, there were slightly more men than women, less than half the patients underwent surgery, and the majority were on antimicrobial therapy while the study catheter was in place. The majority of catheters were inserted through the subclavian route to deliver fluids and drugs. Within each group, the period of hospitalization before catheter insertion and the duration of catheterization were the same. Skin colonization in the 2 dressing groups To determine the effect of the dressings on growth of microorganisms, an agar impression of organisms at the insertion site was made just before insertion (baseline) and again before removal of the catheter. The difference was expressed either as an increase, decrease, or no difference relative to the baseline for each site. Baseline levels of skin colonization at the catheter insertion site were similar for the 2 groups (Table 3). Of 125 sites, 61 (49%) showed decreased growth under the dressing, 33 sites (26%) showed increased growth, and for 31 sites (25%) there was no difference. When the 2 dressing groups were compared (Table 3), no significant difference was detected in the level of skin colonization (χ2 = 0.72, df = 2, P = .697). Hub colonization Cultures from 141 catheter hubs were available for analysis; of these, 38 (27%) were colonized and 103 (73%) showed no growth at the time of removal (Table 3). Of the 38 colonized catheters, 26 (18%) had <20 colonies, 9 (6%) had ≥20 colonies, and 3 (2%) had confluent growth. When the proportion of colonized hubs was compared between the 2 dressing groups, there was no significant difference detected by the 2-tailed Fisher exact test (P = .348) (Table 3). Catheter tip colonization Cultures from 128 catheter tips were obtained by the method of Maki et al;9 of these, 36 (28%) were classified as catheter colonization (Table 3). Of these, 24 (67%) were from the hydrocolloid group and only 12 (33%) were from the polyurethane group. The 2-tailed Fisher

AJIC

Nikoletti et al 491

Volume 27, Number 6

Table 2. Comparison of patient and catheter characteristics between the 2 dressing groups P value

Characteristics

Polyurethane

Hydrocolloid

Patients Median age (y) (mean, SD, range) Sex, men Diagnostic category Medical Surgical Median APACHE II score (mean, SD, range) Median hospital day of catheter insertion (mean, SD, range) Median days in hospital (mean, SD, range) Steroid therapy Other invasive devices Catheters Purpose of catheter General access Hyperalimentation Hemodialysis

n = 74 - 79 56 (53, 19, 18-81) 48 (64%)

n = 71 - 79 55 (52, 18, 16-84) 44 (56%)

41 (53%) 37 (47%) 20 (20, 6, 0-33) 6 (9, 10, 1-46) 27 (37, 28, 1-163) 8 (11%) 75 (100%)

44 (56%) 35 (44%) 19 (20, 7, 4-33) 6 (9, 10, 1-50) 24 (43, 37, 4-163) 11 (16%) 78 (100%)

.693

n = 73 63 (86%) 2 (3%) 8 (11%) n = 75 5 (5, 2,1-10) n = 76 58 (76%) 11 (15%) 1 (1%) 6 (8%) n = 73 6 (8%) 43 (59%) 24 (33%) n = 68 15 (22%) 17 (25%) 23 (34%) 4 (6%) 4 (6%) 4 (6%) 1 (1%)

n = 69 60 (87%) 4 (6%) 5 (7%) n = 73 5 (6, 3, 1-13) n = 75 60 (80%) 6 (8%) 5 (7%) 4 (5%) n = 68 5 (7%) 43 (63%) 20 (29%) n = 66 22 (33%) 24 (36%) 12 (18%) 4 (6%) 4 (6%) 0 0

.517

Median number of days in situ (mean, SD, range) Catheter location Subclavian Jugular Femoral Supraclavicular Number of lumens 2 3 4 Reason for catheter removal Not required Suspected infection Routine Dislodged Patient deceased Malfunction Allergy

.627 .409

.847 .681 .999 .466

.303 .694*

.870

.185†

*Two-tailed Fisher exact test for subclavian versus other sites. †Two-tailed Fisher exact test for suspected infection versus all other reasons.

exact test revealed that the difference was statistically significant (P = .048). There was a 17% higher rate of colonization in the hydrocolloid group (36% colonized) compared with the polyurethane group (19% colonized) (Table 3). Blood cultures A total of 397 blood culture sets were collected from 132 patients. Of 397 culture sets, 35 (9%) were positive, 23 (6%) from the polyurethane group and 12 (3%) from the hydrocolloid group. This difference is statistically significant (2-tailed Fisher exact test = .03). For the purposes of this study, cultures were scored as positive when growth was detected in at least one bottle of a set. Results for growth in single bottles (Table 3) suggest that a large proportion of positive cultures are likely to be contaminants. When expressed as a proportion of patients tested, the percentage of positive blood cul-

tures was 26%, with 17% from the polyurethane group and 9% from the hydrocolloid group. Of the positive cultures, only 6 (17%) were temporally related to catheter tips colonized with the same species (Table 4). Five were from the hydrocolloid group and one was from the polyurethane group. In the hydrocolloid group, the same species was isolated from the skin, hub, catheter tip, and blood in 2 of the 5 cases, suggesting that either the skin or the hub could be potential sources of infection (Table 4). In another 2 cases, the same species was isolated from the skin, tip and blood, but not the hub, suggesting that the skin, rather than the hub, may be a source of infection. In the fifth case, the same species was isolated from the hub, tip and blood, but not the skin, suggesting that the hub, rather than the tip may be a source of infection. In the polyurethane group, the skin was not implicated as a source of infection (Table 4).

AJIC

492 Nikoletti et al

December 1999

Table 3. Frequency of colonization at skin insertion site, catheter hub, and catheter tip

Skin colonization Preinsertion CFU (baseline) 0 1-50 >50 but 50 but
Polyurethane

Hydrocolloid

Total

n (%)

n (%)

n (%)

n = 74 6 (8) 19 (26) 33 (45) 11 (15) 5 (7) n = 71 18 (25) 21 (30) 26 (37) 4 (6) 2 (3) n = 63 3 (52) 16 (25) 14 (22) n = 69 53 (77) 16 (23) n = 62 50 (81) 12 (19) n = 65 42 (65) 19 (29) 4 (6)

n = 77 2 (3) 22 (29) 38 (49) 12 (16) 3 (4) n = 70 20 (29) 13 (19) 23 (33) 12 (17) 2 (3) n = 62 28 (45) 17 (27) 17 (27) n = 72 50 (69) 22 (31) n = 66 42 (64) 24 (36) n = 67 55 (82) 5 (8) 7 (10)

n = 151 8 (5) 41 (27) 71 (47) 23 (15) 8 (5) n = 141 38 (27) 34 (24) 49 (35) 16 (11) 4 (3) n = 125 61 (49) 33 (26) 31 (25) n = 141 103 (73) 38 (27) n = 128 92 (72) 36 (28) n = 132 97 (74) 24 (18) 11 (8)

P value

.781*

.104

.697

.348

.048

.030‡

CFU, Colony-forming units. *For χ2 analysis, categories were collapsed to ≤50, >50, and confluent. †For χ2 analysis, this category was merged with ≤50 as organisms were identified as low numbers of swarming Proteus. ‡Fisher exact test for categories collapsed into positive and negative.

Types of organisms The most common organisms isolated from the insertion site, catheter hub, catheter tip, and blood were coagulase-negative Staphylococcus species. Table 5 shows the range of microorganisms isolated from each site and the frequency with which each species was isolated. Bacillus species were frequently isolated from the skin before cleansing at the time of insertion but were rarely isolated at catheter removal or from the hub or tip. For blood cultures, the number of culture sets positive for only one bottle is shown in the legend. Risk factors for local catheter infection Univariate analysis revealed statistically significant relationships between catheter colonization and skin colonization (χ2 = 27.27, df = 2, P < .001), as well as catheter colonization and hub colonization (2-tailed Fisher exact test, P = .001). No statistically significant relationships were demonstrated for any of the following patient and catheter variables by using χ2 analysis

or the Mann-Whitney U test: age, sex, APACHE II score, diagnostic category, purpose of catheter, location of catheter, number of lumens, duration of hospitalization before catheter insertion, duration of catheterization, use of antimicrobial or steroid therapy, presence of drains, other intravenous devices, ventricular drains, mechanical ventilation, and duration of hospitalization. DISCUSSION The present study was undertaken to extend the work of Haffejee et al,8 whose results indicated that transparent hydrocolloid dressings could be an improvement over conventional polyurethane. Specifically, their study demonstrated that there was a significant reduction of microbial growth under hydrocolloid dressings compared with polyurethane. However, the present study has revealed that hydrocolloid dressings increase the risk of catheter colonization. The two studies differ in several aspects. The results of the former study were based on skin cultures taken repeatedly during a period of 2 months, when dressings were changed and

AJIC

Nikoletti et al 493

Volume 27, Number 6

Table 4. Bacterial species and sites associated with simultaneous positive blood culture and catheter colonization Dressing group

Hydrocolloid Case 1 Case 2 Case 3 Case 4 Case 5 Polyurethane Case 1

Skin baseline

Skin removal

Hub

Catheter tip

Blood culture

CNS CNS Bacillus CNS Bacillus CNS

S aureus CNS CNS CNS Xanthomonas

S aureus CNS No growth No growth

S aureus CNS CNS CNS

CNS

Pseudomonas

Proteus

S aureus CNS CNS Bacillus CNS Enterococcus Citrobacter Proteus CNS

No growth

Enterobacter

–

CNS

CNS

Proteus

CNS, Coagulase-negative Staphylococcus aureus; –, missing data.

catheters removed. However, these samples were obtained from only 8 patients receiving total parenteral nutrition and catheter tips were not cultured. Skin was cleansed with povidone-iodine followed by 0.5% chlorhexidine in 70% alcohol. In contrast, the present study sampled skin sites before insertion and removal of catheters but not during dressing changes. The majority of catheters were inserted for general access, and only 4% were for parenteral nutrition. Skin was cleansed with 0.5% chlorhexidine in 70% alcohol without prior use of povidone-iodine. Our study used a much larger sample size, with randomization to dressing groups and an additional outcome measure, catheter colonization. The 2 studies also used different brands of hydrocolloid dressings, because Visiband (Convatec Squibb, Skillman, NJ) was not available for the present study. It is possible that different brands of hydrocolloid dressing may not be comparable in their effect on microbial growth. In the present study, skin cultures from 125 catheter sites did not reveal any significant difference in growth under the 2 dressings. Unexpectedly, however, hydrocolloid dressings were associated with a significantly increased risk of catheter colonization, with two thirds of the colonized catheters occurring in this group. This finding cannot be explained by differences in skin colonization or by hub colonization because the 2 dressings groups had similar outcomes for these variables. These negative findings may be explained by inadequate sample size because catheter colonization was significantly associated with skin and hub colonization when data from both dressings groups were combined. The latter findings are consistent with those of other studies that have identified the skin insertion site11,12 and the hub12,13 as the 2 major routes of central catheter infection. Although the type of dressing was not expected to affect the frequency of hub colonization, hub cultures were routinely performed to ensure that both dressing groups had an equal risk of catheter colonization via this route. In addition, data on the overall frequency of colonization of this site was sought. Systematic sam-

pling of hub sites is rarely undertaken in studies comparing catheter dressings, therefore the relative importance of hub colonization is unclear.14 Overall, 27% of catheters showed some level of hub colonization, with 6% having more than 20 colonies and only 2% showing confluent growth. This frequency is higher than the rates of 1.9% to 5.7% reported in central venous and arterial catheters by Maki et al.15 However, their figures included only those hubs colonized with more than 10 CFU. In a more recent study,16 23% of hub cultures were positive, with the criteria for positive culture being any number of organisms isolated. Our study demonstrated a statistically significant relationship between catheter colonization and hub colonization, which concurs with previous findings.12,16 In addition, Moro et al12 demonstrated that hub colonization, when compared with skin colonization, was more frequently responsible for severe infections. Therefore, health care professionals should seek every opportunity for reducing the risk of infection via the hub. A novel method of assessing skin colonization was used for this study because there is no agreement in the literature regarding the best method.17 Agar impressions were chosen instead of swabs because the former are not affected by delays in laboratory analysis and thus were considered more suitable for the study setting. In addition, skin cultures were to be taken by a large number of bedside nurses and the part-time research nurse, and it was believed that variations in technique would be less likely to occur with an agar impression. The agar syringe device was used as an alternative to the Rodac plate because the projecting surface of the former provided more effective contact with the skin surface. It is not possible to draw any conclusions about causal relationships between the dressing type and catheter-related bacteremia. Although there were several instances of simultaneous bacteremia and colonized catheters, these could have arisen from several alternative sources to the hub or the skin, particularly given that all patients had at least one other invasive

AJIC

494 Nikoletti et al

December 1999

Table 5. Frequency of organisms cultured from each site for polyurethane (PU) and hydrocolloid (HC) dressing group

Organism

Skin (baseline)

Skin (removal)

PU

HC

PU

HC

PU

HC

PU

HC

PU

HC

n = 75

n = 77

n = 71

n = 70

n = 70

n = 72

n = 63

n = 66

n = 23

n = 12

65 6

41 2

40 2

13 0

18 2

17 1

30 1

13* 1

7† 1

1 2 0 74

0 2 0 45

1 2 0 45

0 0 0 13

0 0 0 20

0 1 0 19

2 4 0 37

0 2‡ 1‡ 17

1 0 0 9

33 0 0 0 33

2 0 0 0 2

1 0 0 0 1

0 0 0 0 0

2 0 0 0 2

0 0 0 0 0

1 1 0 0 2

0 0 1‡ 2‡ 3

0 0 1‡ 1‡ 2

0

0

0

0

0

0

1

0

0

2 0 2 2 1 1 1 0 9

2 0 2 2 4 1 3 0 14

1 0 0 3 5 2 1 1 13

0 0 1 1 2 0 0 0 4

0 0 0 0 0 0 1 0 1

0 0 1 2 0 0 0 0 3

2 0 0 2 1 1 1 1 8

3‡ 2‡ 0 0 0 0 0 0 5

0 0 0 1 0 1‡ 1 0 3

0 116§

2 63§

0 59§

1 18

1 24

2 24

1 49

0 25§

0 14§

Gram-positive cocci Coagulase-negative Staphylococcus sp 58 S aureus 5 Alpha haemolytic Streptococcus sp 2 Enterococcus sp 2 Micrococcus sp 0 Subtotal 67 Gram-positive bacilli Bacillus sp 31 Corynebacterium jeikeium 0 Corynebacterium sp 0 Propionibacterium acnes 0 Subtotal 31 Gram-negative cocci Neisseria sp 0 Gram-negative bacilli Acinetobacter sp 6 Bacteroides sp 0 E coli 1 Enterobacter sp 2 Klebsiella sp 1 Pseudomonas sp 0 Proteus sp 1 Xanthomonas sp 0 Subtotal 11 Fungi Candida sp 0 Total frequency 109§

Hub

Catheter tip

Blood culture

*Ten of thirteen isolates were positive in only one bottle of a culture set. †Three of seven isolates were positive in only one bottle of a culture set. ‡Isolates were positive in only one bottle of a culture set. §Isolates were sometimes recovered as mixed cultures, or in the case of blood cultures, repeated sets yielded different isolates, therefore column totals do not reflect the number of positive cultures.

line in situ. Furthermore, the finding that the majority of positive blood cultures in this study were not temporally related to colonized catheter tips suggests that sources of contamination or infection other than the catheter are common in this group of vulnerable patients. This conclusion is supported by the finding that a large proportion of positive blood cultures were obtained from single bottles, suggesting contamination or pseudobacteremia. However, the reason for a significantly increased frequency of pseudobacteremia in the polyurethane group is not clear and requires further investigation. Methodologic problems associated with distinguishing pseudobacteremia from true infection are well documented.18 For these reasons, catheter-related bacteremia was not used as a primary outcome measure. A further limitation on the use of bacteremia or sepsis as an outcome measure is that the relatively low prevalence

of catheter-related sepsis reported for Australia (2.3%)19 creates the need for a sample size of many hundreds to achieve adequate statistical power.20 The use of catheter colonization as an outcome measure allows for smaller samples as the frequency is generally much higher, ranging from 7.5%12 to 42%.20 In the present study, the overall frequency of catheter colonization was 28%. The most common organisms isolated from all sites (skin, hub, catheter tip, and blood) were coagulase-negative staphylococci. No attempt was made to speciate or subtype these because the study was not designed to analyze causal relationships. Isolation of anaerobes was attempted from all blood cultures, but not from the skin, the catheter hub, or tip. A similar range of isolates was recovered for both dressing groups but frequencies of each species were too low to permit further analysis. Bacillus species were recovered with relatively high fre-

AJIC

Nikoletti et al 495

Volume 27, Number 6

quency from baseline skin cultures but were rarely found in cultures taken from the skin at catheter removal, which suggests that conditions under the dressing do not support the growth of these organisms or that the skin-cleansing procedure effectively removes or suppresses their growth. In addition, Bacillus species were rarely isolated from catheter hubs or tips. The univariate analysis of risk factors associated with catheter colonization showed significant association with skin and hub colonization, concurring with other studies.12 However, no other statistically significant associations were detected for patient-related and catheter-related variables. One limitation of this study is that the number of dressing changes for each catheter and the timing of the last dressing change before catheter removal were not documented. Dressing changes could have been more frequent among febrile patients in the hydrocolloid group because of problems with dressing adherence and transparency when sweating occurred. Recently published guidelines for central venous catheters make no specific recommendation on the frequency of dressing changes, reflecting the lack of conclusive data from published studies.22 However, some studies on polyurethane dressings for central venous catheters suggest that the risk of infection is reduced with more frequent dressing changes.8,23 Therefore it is unlikely that an increase in frequency of dressing changes in the hydrocolloid group would have contributed to the increased risk of catheter colonization. Another limitation of the study is the high rate of attrition of catheters subsequent to enrollment in the study. This rate was largely a result of the lack of 24hour research assistance, which meant that some specimens were not taken and many patients were not followed up when they were transferred to other wards with the catheter in place. However, a comparison of patient and catheter characteristics for those included versus those excluded indicates that the outcomes of this study were not likely to have been influenced by selection bias. Craven et al24 report even higher levels of attrition (55%-62%) arising from similar circumstances. The hydrocolloid dressing was generally not favored by nursing staff because of problems with adherence and a tendency for the hydrocolloid matrix to break down with prolonged use in febrile patients. The breakdown of the matrix gives the appearance of pus, leading to the false impression that the site is infected. The opacity of the dressing under such conditions not only negates the potential benefit of transparent hydrocolloid over gauze but also creates concerns about localized infection, which may lead to unnecessary resiting of the catheter.

In conclusion, the results of this study suggest that there is an increased risk of catheter colonization associated with the use of hydrocolloid dressings. The clinical significance of increased numbers of positive blood cultures in the polyurethane group requires further examination, although distinguishing between contamination and true infection in intensive care settings continues to be methodologically challenging. Further studies are needed to determine whether this finding is generalizable across different study settings and whether similar outcomes are obtained when different brands of hydrocolloid dressing are used. Recent guidelines for intravascular devices recommend the use of either polyurethane or gauze.22 While the debate regarding the relative advantages and disadvantages of gauze, standard polyurethane, and Opsite IV3000 continues, further research is required to identify suitable alternative dressings. The question of dispensing with dressings altogether has been examined, with favorable outcomes, but only for catheters that are subcutaneously tunneled and used in the outpatient setting.25 Uncuffed catheters in ventilated patients in critical care settings may require different guidelines for care. Irrespective of the type of dressing used, there appears to be unanimous agreement in the literature about the importance of adherence to aseptic technique and thorough handwashing in the prevention of central venous catheter infection. We gratefully acknowledge Dr Brian Mee for the design of the agar syringe, Ms Christina Cook, Ms Michelle Dunlop, and the intensive care unit nursing staff for data collection, Department of Microbiology staff for microbiological analysis, Megan Phillips and Margaret Downie for laboratory assistance, and Dr Ian Jacobs and Dr Pender Pedler for statistical advice.

References 1. Hoffmann KH, Weber DJ, Samsa GP, Rutala WA. Transparent polyurethane film as an intravenous catheter dressing: a metaanalysis of the infection risks. JAMA 1992;267:2072-6. 2. Maki DG. Yes, Virginia, aseptic technique is very important: maximal barrier precautions during insertion reduce the risk of central venous catheter-related bacteremia. Infect Control Hosp Epidemiol 1994;15:227-30. 3. Thomas S, Loveless P, Hay NP. Comparative review of the properties of six semipermeable film dressings. Pharm J 1988;241:785-9. 4. Maki DG, Stolz SM, Wheeler SJ, Mermel LA. A prospective, randomized, three-way clinical comparison of a novel highly permeable polyurethane dressing with 442 Swan-Ganz catheters [abstract 825]. In: Programs and abstracts of the 32nd interscience conference on antimicrobial agents and chemotherapy, Washington (DC): American Society for Microbiology; 1992. 5. Brandt B, DePalma J, Irwin M, Shogan J, Lucke JF. Comparison of central venous catheter dressings in bone marrow transplant recipients. Oncol Nurs Forum 1996;23:829-36. 6. Reynolds MG, Tebbs SE, Elliott TSJ. Do dressings with increased permeability reduce the incidence of central venous catheter related sepsis? Intensive Crit Care Nurs 1997;13:26-9. 7. Wille JC, Blussé van Oud Alblas A, Thewessen EAPM. A compari-

AJIC

496 Nikoletti et al

8.

9.

10.

11.

12.

13.

14. 15.

16.

son of two transparent film-type dressings in central venous therapy. J Hosp Infect 1993;23:113-21. Haffejee AA, Moodley J, Pillay K, Singh B, Thomson S, Bhamjee A. Evaluation of a new hydrocolloid occlusive dressing for central catheters used in total parenteral nutrition. Suid-Afrikaanse Joernaal Vir Chirurgie 1991;29:142-46. Maki DG, Weise CE, Sarafin HW. A semiquantitative culture method for identifying intravenous-catheter-related infection. N Engl J Med 1977;296:1305-9. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med 1985;13:818-29. Maki DG. Pathogenesis, prevention and management of infections due to intravascular devices used for infusion therapy. In: Bisno AL, Waldvogel FA. Infections associated with indwelling medical devices. Washington (DC): American Society for Microbiology; 1989. Moro ML, Viganò EF, Lepri AC. The central venous catheter-related infections study group. Risk factors for central venous catheterrelated infections in surgical and intensive care units. Infect Cont Hosp Epidemiol 1994;15:253-64. Linares J, Sitges-Serra A, Garau J, Perez JL, Martin R. Pathogenesis of catheter sepsis: a prospective study with quantitative and semiquantitative cultures of catheter hub and segments. J Clin Microbiol 1985;21:357-60. Goldmann DA, Pier GB. Pathogenesis of infections related to intravascular catheterization. Clin Microbiol Rev 1993;6:176-92. Maki DG, Ringer M, Alvarado CJ. Prospective randomized trial of povidone-iodine, alcohol, and chlorhexidine for prevention of infection associated with central venous and arterial catheters. Lancet 1991;338:339-43. Segura M, Alvarez-Lerma F, Tellado JM, Jimenez-Ferreres J, Oms

December 1999

17.

18.

19. 20.

21.

22.

23.

24.

25.

L, Rello J, et al. A clinical trial on the prevention of catheter-related sepsis using a new hub model. Ann Surg 1996;223:363-9. Hambraeus A, Hoborn J, Whyte W. Skin sampling: validation of pad method and comparison with commonly used methods. J Hosp Infect 1990;16:19-27. Souvenir D, Anderson DE, Palpant S, Mroch H, Askin S, Anderson J, et al. Blood cultures positive for coagulase-negative staphylococci: antisepsis, pseudobacteremia, and therapy of patients. J Clin Microbiol 1998;1923-6. Collignon PJ, Munro R, Sorrell TC. Systemic sepsis and intravenous devices; a prospective survey. Med J Aust 1984;141:345-8. Collignon PJ, Munro R. Laboratory diagnosis of intravascular catheter associated sepsis. Eur J Clin Microbiol Infect Dis 1989;8:807-14. Conly JM, Grieves K, Peters B. A prospective, randomized study comparing transparent and dry gauze dressings for central venous catheters. J Infect Diseases 1989;159:310-9. Pearson ML, Hospital Infection Control Practices Advisory Committee. Guideline for prevention of intravascular device-related infections Part I and Part II. AJIC Am J Infect Control 1996;24:262-93. Engervall P, Ringertz S, Hagman E, Skogman K, Björkholm M. Change of central venous catheter dressings twice a week is superior to once a week in patients with haematological malignancies. J Hosp Infect 1995;29:275-86. Craven DE, Lichtenberg DA, Kunches LM, McDonough AT, Gonzalez MI, Heeren TC, et al. A randomized study comparing a transparent polyurethane dressing to a dry gauze dressing for peripheral intravenous catheter sites. Infect Control 1985;6:361-6. Petrosino B, Becker H, Christian B. Infection rates in central venous catheter dressings. Oncol Nurs Forum 1988;15:709-17.