Securing Jugular Central Venous Catheters With Dressings Fixed to a Liquid Adhesive to Prevent Dressing Failure in Intensive Care Patients (the STICKY Trial): A Randomized Controlled Trial

Abstract

OBJECTIVES:

Central venous catheters (CVCs) are vital for treating ICU patients. However, up to a quarter of CVCs fail from mechanical or infective complications. Poor securement of CVCs to the skin contributes to catheter failure, particularly CVCs placed in the jugular vein, which are highly vulnerable to pullout forces. This study evaluated the effectiveness of medical liquid adhesive (MLA) for improving jugular CVC dressing adhesion.

DESIGN:

Multisite parallel group, superiority, randomized controlled trial.

SETTING:

Four metropolitan Australian ICUs.

PATIENTS:

Eligible patients were 18 years old or older, within 12 hours of jugular CVC insertion, expected to need the CVC for greater than or equal to 72 hours, and remain in ICU for greater than or equal to 24 hours.

INTERVENTIONS:

Patients were randomly allocated (stratified by hospital and gender) to standard CVC dressings with the application of MLA to skin under the dressing border (intervention) or standard care CVC dressings (control).

MEASUREMENTS AND MAIN RESULTS:

The primary endpoint was dressing failure within 7 days due to lifting edges. Secondary outcomes included the total number of dressing changes, skin injury, and CVC failure. In total, 160 participants (82 control; 78 intervention) were enrolled. There were 22 (28%) and 41 (50%) cases of premature dressing failure in the intervention and control groups respectively (odds ratio, 0.39; 95% CI, 0.20–0.76; p = 0.005). The intervention group had fewer dressing changes (incidence rate ratio [IRR], 0.74; 95% CI, 0.55–0.99). Time to dressing failure (log-rank test; p = 0.12) and all-cause CVC failure (IRR, 1.44; 95% CI, 0.36–5.79) did not differ between groups. Three skin injuries occurred: one in the intervention (blister) and two in the control (maceration and skin tear) groups.

CONCLUSIONS:

MLA is associated with significantly decreased jugular CVC dressing failure and longer dressing dwell, with an acceptable safety profile. MLA should be considered to preserve jugular CVC dressings in ICU.

KEY POINTS.

Question: What is the effectiveness of medical liquid adhesive, compared with standard dressings, for improving jugular central venous catheter dressing adhesion?

Findings: Medical liquid adhesive reduced jugular central venous catheter dressing failure and resulted in longer dressing dwell.

Meaning: Medical liquid adhesive should be considered as an adjunct to routine jugular central venous catheter dressings in the ICU.

Central venous catheters (CVCs) in the ICU deliver supportive and interventional therapies, monitor central venous pressures and sample blood from critically ill patients. Millions of CVCs are inserted globally each year, with up to a quarter failing due to noninfectious (e.g., thrombosis, dislodgement) or infectious (local or systemic) complications (1–4). CVC failure leads to longer ICU stays, increased morbidity, and mortality (5, 6).

CVCs are often inserted in the jugular vein, which have higher risk of central line-associated bloodstream infection (CLABSI) and failure compared with subclavian vein insertion (7–9). Optimizing dressing efficacy can limit CVC failure and complications risk (10). Yet, finding a dressing that provides security and stability at this challenging anatomical location is difficult (2, 11). Traditionally, polyurethane dressings, with or without sutures or stabilization devices, have been used (12). However, poor adherence to the skin can cause these dressings to fail from lifting edges or complete detachment, increasing the opportunity for extrinsic contamination of the insertion wound and causing local infection or CLABSI. Indeed, a three-fold increase in CLABSI has been reported following two or more dressing disruptions (13). Treating CLABSI places economic burden on hospitals with attributable costs per episode ranging from $33,696 to $71,443 U.S. dollars (14–16). Furthermore, poorly secured CVCs allow catheter micromotion, irritating the vessel wall and causing complications, including thrombosis or catheter malposition (3, 12).

Current international guidelines recommend 7-day routine dressing changes (12, 17). However, jugular CVC dressings fail as early as 25–46 hours after placement (2, 13, 18). A randomized controlled trial (RCT) examining dressing failure found lifting edges in nearly 40% of jugular CVC dressings (18). Jugular CVC dressings are exposed to multivector pull forces from patient mobilization, lighter sedation, “catching” infusion sets (e.g., on bedrails), and the “drag” of multiple infusions (2, 3, 19). Repeated dressing changes increase medical adhesive-related skin injury (MARSI) risk (20, 21), to which critically ill patients are vulnerable due to factors including organ dysfunction and hemodynamic instability (22, 23).

A latex-free, medical liquid adhesive (MLA), applied to the skin following antisepsis but before dressing application, may improve dressing integrity. This nonwater-soluble gum mastic, manufactured from Pistacia lentiscus tree resin, is purported to reduce unnecessary dressing changes and CLABSI and MARSI occurrence (11, 20). However, due to the lack of high-quality evidence examining its use (24), global guidelines cannot make firm practice recommendations (12). In this RCT, we tested the effectiveness and safety of MLA, compared with usual care, for jugular CVC securement in ICU patients. Our aim was to provide clinicians and policy makers the best evidence to support decisions for dressing securement.

MATERIALS AND METHODS

Study Design and Participants

This multicenter, parallel-group, superiority RCT was conducted in four secondary, tertiary, or quaternary Australian ICUs: three in Brisbane (The Prince Charles, Royal Brisbane and Women’s, and Logan Hospitals) and one in Sydney (Liverpool Hospital). Recruitment occurred between September 2021 and February 2023. Research nurses (ReNs) screened daily for participants meeting eligibility criteria: 18 years old or older, expected to have a jugular CVC greater than or equal to 72 hours, expected ICU admission greater than or equal to 24 hours, and within 12 hours of CVC insertion. Patients were excluded if they had: a bloodstream infection (BSI) in the previous 24 hours (25), emergency CVC insertion (with the potential for a breach in aseptic technique), a concurrent CVC expected to dwell greater than 24 hours, end-of-life care orders, or previous enrollment. The Metro North Health ethics committee approved the trial (HREC/2021/QRBW/73896; title: Securing jugular central venous access devices with dressings fixed to a liquid adhesive (Mastisol) in an Intensive Care Unit population: a randomized controlled) on May 21, 2021, and the protocol was approved by all participating institutions. The trial was registered with the Australian New Zealand Clinical Trial Registry (ACTRN12621001012864) and the protocol published (26). Study procedures adhered to the Helsinki Declaration 1975. Written informed consent or consent-to-continue was obtained. The trial is reported according to the Consolidated Standards of Reporting Trials guidelines (27).

Interventions

We compared standard CVC dressing plus the application of MLA (intervention; Mastisol, Eloquest Healthcare, Ferndale, MI) with standard CVC dressings (control). Standard dressings had minor variations between study sites according to local policy but were predominately a bordered transparent polyurethane dressing ± chlorhexidine disc or gel pad (Supplementary Table 1, http://links.lww.com/CCM/H627). For intervention group participants, MLA was provided to clinical staff before dressing application. MLA was applied after skin antisepsis from the single-use 0.66 mL vial in a half-inch wide line where the outside border of the dressing was to be placed (manufacturer’s instructions), at the time of CVC insertion or within 12 hours of insertion. It was allowed to dry for 30 seconds before dressing application. MLA was applied by hospital staff who had received training on MLA application from research staff at each site.

Outcomes

The primary outcome was CVC dressing failure, defined as the requirement for initial dressing replacement due to lifting at the edges before routine dressing change at 7 days (12). The need for dressing change was determined by clinical staff (not research staff).

Secondary outcomes were: premature dressing removal during CVC dwell less than 7 days; total dressing changes during CVC dwell; loss of dressing integrity not requiring dressing change (i.e., edges lifting with/without reinforcement); staff and patient satisfaction on dressing application and removal (11-point ordinal scale); all-cause CVC failure from infectious and/or noninfectious complications; subtypes of CVC failure (CLABSI [28], primary BSI [28], local infection defined by “arterial or venous infection” criteria [25, 28], pain, infiltration/extravasation, blockage/occlusion, fracture, thrombosis, or dislodgement); CVC dwell time; serious adverse skin events relating to MARSI (pain, itch, erythema, skin stripping, blister, skin tear, irritant contact dermatitis, maceration, folliculitis, and infection) (29); mortality; skin colonization and measured descriptively (i.e., organism) and quantitatively (i.e., colony-forming units [CFUs]); and cost analysis (cost and number of products used, cost of treating complications, and staff time).

Randomization and Masking

ReNs randomized patients via a central, web-based service in a 1:1 ratio, with randomly varied blocks (four or six). Randomization was generated by a statistician separate to the study team and was stratified by hospital and gender to account for facial/neck hair differences. Allocation was concealed until after randomization.

Participants, clinical staff, or ReNs could not be masked to dressing allocation. The infectious diseases physician and microbiology laboratory staff were masked to group allocation when assessing infectious outcomes, as was the statistician for data analysis.

Data Collection

Study data were collected by ReNs and stored in Research Electronic Data Capture (REDCap; Vanderbilt University, Nashville, TN) (30). A project manager undertook quality checks for allocation integrity and source data verification for the first patient at each site and a random 5% of data for all patients. Data were collected at recruitment, removal, and 48 hours post-removal (see protocol) (26). ReNs assessed insertion sites daily to identify whether dressings were still in place, if they were lifting at one or more edges, and for adverse events potentially associated with the dressings (e.g., blister, redness). Masked inter-rater site assessments (adverse skin events related to MARSI) were completed by two ReNs for a subset of 20 patients.

Sample Size

We hypothesized a 25% reduction in dressing failure with MLA use (from 50% to 25%) based on prior research (2, 18). Consequently, with 90% power (alpha = 0.05), outcome data was required from 77 participants per group. To account for potential attrition, three additional participants per group were recruited (total 160 participants).

Statistical Analysis

Summary statistics are presented as mean and sd or median and interquartile range (IQR) for continuous data, depending on distribution, and as frequency and percentage for categorical data. Statistical analyses were performed as prespecified (26). Missing data was not imputed, and a complete-case analysis was performed. Analysis followed a modified intention-to-treat principle, which included all randomized participants where the primary endpoint was available. The association between intervention group and primary outcome was investigated using a logistic regression model with study group included as the fixed effect. The effect estimate was presented as odds ratio (OR) and 95% CI. Findings are also presented as absolute risk difference (ARD), calculated using a generalized linear model (GLM) with a binomial family and identity link. We investigated possible interactions between study group and stratification variables (site and gender) using the likelihood-ratio test. The robustness of the primary outcome was explored in a per-protocol population, which excluded participants not receiving the allocated study intervention within 12 hours of CVC insertion. The sensitivity of the primary outcome was investigated by rerunning analyses adjusting for stratification factors, then stratifying by site and gender (Supplementary Methods, http://links.lww.com/CCM/H627). The key secondary outcome between-group time to dressing failure was assessed using the log-rank test. Secondary time-to-event outcomes were assessed using Cox proportional hazards models and are presented as hazard ratio (HR) and 95% CI. Secondary count outcomes were assessed using GLMs with Poisson family and log link, offset by the natural logarithm of time-at-risk, with effect estimates presented as incidence rate ratio (IRR) and 95% CI.

For cost analyses, healthcare resources cost was estimated by assessing the sum product of unit costs and respective utilization. Resources included CVCs, affiliated dressings and securements, technology used for insertion, number of subsequent dressings used, and costs of treating complications. Convenience samples of staff time associated with CVC insertion and dressing changes were costed separately from the full sample. A simple economic model estimated total per participant cost (2022 Australian dollars; Supplementary Methods, http://links.lww.com/CCM/H627). For secondary outcomes and subgroup analyses, formal adjustment of CIs for multiplicity was not performed, and no definitive inferences should be drawn. All statistical and economic analyses were performed in Stata v13.1 (StataCorp, College Station, TX).

Microbiology Substudy

A convenience sample of 20 patients (ten per arm) was included in a microbiological substudy to investigate skin colonization underneath the dressing. Upon dressing removal/change, skin surrounding the insertion site was swabbed using a moistened (sterile saline) swab, applied (twisting/back-forth) for 5 seconds. Swabs were placed in a sterile tube containing transport medium. Swabs were inoculated onto horse blood agar and incubated in ambient air (35°C) for 5 days, with plates examined on days 1 and 5. Bacterial growth was semi-quantified by determining CFUs/swab, and single-pick colony morphotypes identified to species level using Matrix Assisted Laser Desorption Ionization-Time of Flight (MALDI-TOF) (Vitek MS; v3.2 BioMerieux, Marcy-l’Étoile, France).

RESULTS

Two hundred ninety patients were screened for eligibility, with 163 randomized to MLA (n = 80) or control (n = 83) groups (Fig. 1). Three participants were excluded from the modified intention to treat analysis due to prespecified reasons: two (MLA) did not provide consent-to-continue and consequently their data could not be collected, and one (control) had a failed CVC insertion and was deemed ineligible post-randomization. In the MLA and control groups, 66 (84.6%) and 82 (100.0%) participants were treated per-protocol, respectively. One participant in the MLA group incorrectly received standard care and 11 protocol deviations occurred where MLA was added outside the 12-hour inclusion period. Overall, 23,180 catheter hours were studied (10,138 hr MLA; 13,040 hr control).

Figure 1.

Consolidated Standards of Reporting Trials diagram.

Participant and CVC characteristics were distributed similarly between study groups (Table 1). Most participants were male (n = 102, 64%), with a median age of 65 years (IQR 51–70 yr), emergently admitted (n = 82, 51%) with a median Acute Physiology and Chronic Health Evaluation II score of 17 (IQR, 13–22). CVCs were mostly four lumens (n = 142, 89%), inserted in ICU (n = 96, 60%) by ICU registrars (n = 90, 56%) on first attempt (n = 135, 84%) using ultrasound (n = 133, 83%) in the mid-neck (n = 115, 79%).

TABLE 1.

Baseline Characteristics of Participants, Central Venous Catheters, and Dressings

Characteristic	Medical Liquid Adhesive (n = 78)	Control (n = 82)
	n (%) or Median (IQR)	n (%) or Median (IQR)
Catheter hours studied	10,139	13,041
Age, yr	62 (50–70)	65 (52–70)
Gender, female	28 (36)	30 (37)
Body mass index—overweight/obese	58 (74)	59 (72)
Acute Physiology and Chronic Health Evaluation II score	16.5 (13–21)	17.0 (13–23)
Location of enrollment
The Prince Charles Hospital	29 (37)	31 (38)
Liverpool Hospital	37 (47)	38 (46)
Logan Hospital	4 (5)	6 (7)
The Royal Brisbane and Women’s Hospital	8 (10)	7 (9)
Planned ICU admission	33 (42)	37 (45)
Emergent ICU admission	43 (55)	39 (48)
≥ 3 comorbidities	38 (49)	36 (44)
Infection at recruitment	21(27)	26 (32)
CVC inserted by ICU registrar	46 (59)	45 (55)
CVC inserted by anesthetic registrar	18 (23)	18 (22)
CVC inserted by ICU/anesthetic consultant	10 (13)	14 (17)
CVC inserted in ICU	49 (63)	47 (57)
CVC inserted in operating theater	25 (32)	30 (37)
Insertion side, right	59 (76)	73 (89)
One insertion attempt	69 (88)	66 (80)
≥ 2 insertion attempts	1 (1)	2 (2)
Technology assisted
Ultrasound	66 (85)	67 (82)
X-ray after	53 (68)	56 (68)
Three lumen CVC	6 (8)	6 (7)
Four lumen CVC	68 (87)	74 (90)
Five lumen CVC	4 (5)	2 (2)
Mid neck jugular placement	56 (72)	59 (72)
Antimicrobial catheter, yes	55 (71)	53 (65)
Chlorhexidine disc or transparent dressing with chlorhexidine gel pad	42 (54)	44 (54)
Additional dressings/securements (at insertion site)a	11 (14)	19 (23)
Use of sutures/tissue adhesive (at insertion site)	65 (83)	68 (83)
Diaphoretic (ever)	16 (21)	23 (28)

CVC = central venous catheter, IQR = interquartile range, n = number.

For the primary outcome of initial CVC dressing failure, there were 22 (28%) and 41 (50%) failures in the MLA and control groups, respectively (OR, 0.39; 95% CI, 0.20–0.76; ARD, –21.8%; 95% CI, –36.5% to –7.2%; p = 0.005) (Table 2). There were similar findings in the per-protocol analysis (OR, 0.40; 95% CI, 0.20–0.80). Supplementary Table 2 (http://links.lww.com/CCM/H627) contain sensitivity analyses. Results were similar after adjusting for stratification factors. There was no significant interaction between study group and site (p = 0.11); however, the interaction between study group and sex was significant (p = 0.04). When considered by sex, MLA dressings appear to have a stronger effect for female (OR, 0.14; 95% CI, 0.04–0.48) than males (OR, 0.65; 95% CI, 0.29–1.44).

TABLE 2.

Study Outcomes by Treatment Group

Primary Outcome (n = 160)	MLA (n = 78)	Control (n = 82)	MLA vs. Control Group
	n (%) or Mean (sd) or Median (IQR)	n (%) or Mean (sd) or Median (IQR)	Effect Estimate (95% CI)	p
First dressing failure, n (%)	22 (28)	41 (50)	Absolute risk difference = –21.8% (–36.5% to –7.2%)	0.005
Time to first dressing failure due to lifting, hr, median (IQR)	58.5 (29.6–81.2)	23.8 (12.3–61.1)	MedD = 43.3 (16.4–70.2)	0.002
Secondary outcomes
Dressing dwell time (all dressings), hr, median (IQR)	56.7 (33.8–94.8)	49.0 (21.3–91.6)	MedD = 8.74 (–6.5 to 23.9)	0.38
Premature dressing change (all dressings), n (%)	56 (88)	117 (94)	OR = 0.42 (0.14–1.21)	0.11
Total dressing changes, n	64	124	Incidence rate ratio = 0.74 (0.55–0.99)	0.04
Loss of dressing integrity, first dressing lifting edges, median (IQR)	0 (0–1)	1 (0–2)	MedD = –1 (–1.44 to –0.56)	< 0.001
Staff satisfaction on application, median (IQR)	9.5 (8–10), n = 38	8 (6–10), n = 22	MedD = 1.00 (–0.48 to 2.48)	0.18
Staff satisfaction on removal, median (IQR)	10 (8–10), n = 32	9 (8–10), n = 35	MedD = 1.00 (–0.02 to 2.02)	0.055
Patient satisfaction on application, median (IQR)	8 (4–8.5), n = 8	7 (5–10), n = 11	MedD = 1.00 (–2.65 to 4.65)	0.57
Patient satisfaction on removal, median (IQR)	8 (7–10), n = 35	8 (7.5–9), n = 32	MedD = 0.00 (–0.78 to 0.78)	> 0.999
All-cause central venous catheter failure, n (%)	4 (5)	4 (5)	HR = 1.44 (0.36–5.79)	0.61
Per 1000 catheter days, mean (95% CI)	9.5 (3.6–25.2)	7.4 (2.8–19.6)
Infectious complications
Local infection at insertion site, n (%)	2 (3)	0 (0)	n/c	n/c
Per 1000 catheter days (95% CI)	4.7 (1.2–18.9)	0 (0.0–6.8)
Mechanical complications
Blockage/occlusion, n (%)	1 (1)	2 (2)	HR = 0.74 (0.07–8.32)	0.81
Per 1000 catheter days (95% CI)	2.4 (0.3–16.8)	3.7 (0.9–14.7)
Partial dislodgment, n (%)	2 (3)	0 (0)	n/c	n/c
Per 1000 catheter days (95% CI)	4.7 (1.2–18.9)	0 (0.0–6.8)
Complete dislodgement, n (%)	1 (1)	2 (2)	HR = 0.74 (0.07–8.32)	0.81
Per 1000 catheter days (95% CI)	2.4 (0.3–16.8)	3.7 (0.9–14.7)
Central venous catheter dwell time, median (IQR)	103.2 (62.9–172.0)	138.5 (85.7–196.8)	MedD = –34.48 (–66.89 to –2.06)	0.037
Adverse events, n (%)
Medical adhesive-related skin injury	1 (1)	2 (2)	OR = 0.52 (0.05–5.85)	0.60
Mortality	5 (6)	9 (11)	OR = 0.55 (0.18–1.74)	0.31

HR = hazard ratio, IQR = interquartile range, MedD = median ratio, MLA = medical liquid adhesive, n = number, n/c = not computable, OR = odds ratio.

Infectious complications = central line-associated bloodstream infection and primary bloodstream infection not included as they had zero events.

Mechanical complications = infiltration/extravasation, too painful to tolerate, catheter dislodgement by patient and thrombosis not included as they had zero events.

MLA CVC dressings dwelled longer (median 58.5 vs. 23.8 hr; median ratio [MedD], 43.3; 95% CI, 16.4–70.2) with fewer dressing changes (64 vs. 124; IRR, 0.74; 95% CI, 0.55–0.99). There were significantly fewer lifting edges on MLA secured dressings compared with control (MedD, –1.0; 95% CI, –1.44 to –0.56). The Kaplan-Meier plot of time to first dressing failure (HR, 0.66; 95% CI, 0.39–1.11) shows no significant difference between groups (log-rank test; p = 0.12) (Fig. 2).

Figure 2.

Kaplan-Meier survival curve for time to dressing failure. MLA = medical liquid adhesive.

The mean occurrence of all-cause CVC failure in the MLA and control groups was 9.5 and 7.4 per 1000 catheter days (HR, 1.44; 95% CI, 0.36–5.79). No CLABSIs or primary BSIs were identified in either group; however, two CVC insertion site infections were reported in the MLA group. Mechanical complications were occlusion (1/78; 1%), partial dislodgement (2/78; 3%), and complete dislodgement (1/78; 1%) in the MLA group and occlusion (2/82; 2%) and dislodgement (2/82; 2%) in the control group.

Adverse skin events were infrequent; one MARSI in the MLA group (blister, likely related to the intervention) and two in the control group (skin maceration and skin tear). All resolved within 7 days. Inter-rater reproducibility testing between ReNs for adverse skin events related to MARSI showed 100% agreement for all signs/symptoms except redness (90% agreement) (Supplementary Table 3, http://links.lww.com/CCM/H627).

Estimated expected cost per participant was $310.43 (95% credible interval [CrI], $308.42–312.44) for the MLA group vs. $321.45 (95% CrI, $319.29–323.61) for the control group, resulting in expected savings of $11.02 (95% CrI, $13.96–8.08) per participant (Supplementary Tables 4 and 5, http://links.lww.com/CCM/H627).

In the microbiological analysis, growth was observed in four of ten (40%) swabs in the MLA group (1/10 with growth ≥ 100 CFU) and eight of ten (80%) swabs in the control group (4/10 with growth ≥ 100 CFU) (Supplementary Table 6, http://links.lww.com/CCM/H627). Swabs had a range of 1–9 colony types per patient. Species identified were consistent with commensal skin flora. Skin swabs were collected from one hospital site and chlorhexidine dressings were not part of the dressing combination in either the MLA or the control groups at this site.

DISCUSSION

Among ICU patients with jugular CVCs who had MLA applied, compared with standard dressings alone, there was an absolute 22% reduction in dressing failure due to lifting. Time to first dressing change was a median 43 hours longer for dressings with MLA, extending median time to dressing change due to lifting from 24 hours in the control group to 57 hours in the MLA group. A cost analysis of labor and materials associated with dressing changes and failure demonstrated cost savings associated with MLA use. Importantly, MLA was acceptable to both staff and participants at both application and removal. Finding effective durable dressings for jugular CVCs has been challenging, and these results present a simple effective intervention targeting jugular CVC dressing failure.

The finding of improved dressing performance with MLA application is congruent with a pre-post study of jugular CVCs that reported half (n = 15/36) of dressings had lifting edges without MLA, and 100% (n = 30/30) of dressings were intact after MLA was introduced (11). Similarly, improved dressing adhesion with MLA was described in an early healthy volunteer study that found significantly higher pull-out force required to dislodge angiocatheters with a dressing and MLA, compared with a dressing alone (31). However, our subgroup analysis comparing males and females shows significantly improved dressing adherence for females in the MLA group compared with males. This is likely driven by the presence of facial hair in males, which has also been described in a previous RCT of dressing regimes (2). Future research should examine improving dressing adherence for males with jugular CVCs.

Bacterial skin colonization at the insertion site in our study was higher in the control (80%) than the MLA group (40%); however, for this outcome, the study was underpowered to detect between-group differences. Nevertheless, these findings could indicate that the properties of gum mastic (Pistacia lentiscus) improve dressing adherence to the skin (32), and thereby reduce colonization around the insertion site. Seminal work by Timsit et al (13) established the relationship between dressing disruption and CVC tip colonization (increasing from 8.6 to 13.1 per 1000 catheter days for the first disruption alone); Soulaidopoulos et al (32) described the broad-spectrum antimicrobial activity of mastic, which inhibited Gram-positive and Gram-negative bacteria growth, including Staphylococcus aureus, in broth cultures. In our study, skin swabs yielded growth of greater than or equal to 100 CFU, four times higher in the control (40%) vs. MLA group (10%). In both groups, only common commensals were grown. However, these microorganisms, such as coagulase-negative Staphylococci, present a significant risk for vulnerable populations, accounting for approximately 40% of BSIs in ICUs with a crude mortality rate of 21% (33, 34). Our finding regarding less colonization at the MLA insertion site requires further investigation to establish if MLA is a potentially useful infection prevention strategy for CVC post-insertion care. There were no CLABSI events in this study. A previous multisite RCT, which collected data from ICUs included in our study, reported a CLABSI rate of 3.6–5.5% (35). Therefore, this confirms the need for future adequately powered studies prioritizing infection outcomes to determine if MLA reduces local and systemic CVC-associated-infections.

Applying skin adhesive has the potential to cause skin injury, particularly in ICU populations who are at high risk due to increased exposure to medical adhesives; mechanical ventilation; renal insufficiency; use of vasopressors, anticoagulants, and corticosteroids; and edema (23, 36, 37). MARSI is a skin abnormality lasting greater than or equal to 30 minutes after adhesive removal, with occurrence ranging from 11% to 42% in ICU patients (22, 23) Individual case studies using MLA for wound or epidural catheter dressings have reported cases of allergic dermatitis (38, 39). In our study, MLA did not increase occurrence of MARSI compared with control. These results are consistent with an audit observing 30,049 vascular access dressings secured with MLA, which reported that 96.7% of dressings were intact with no evidence of MARSI (24). Therefore, MLA appears safe and a useful strategy to decrease patients’ exposure to additional dressing changes and likely the risk of MARSI.

All-cause CVC failure was similar between groups (MLA: 5%, 9.5/1000 catheter days; control: 5%, 7.4/1000 catheter days), with occurrence rates consistent with published estimates (2, 4, 40). Despite significantly reduced dressing disruption with MLA use, we did not reduce CVC failure, as the study was not designed to detect between-groups differences for this outcome. Further research, specifically adequately powered RCTs with CVC failure as the primary outcome, are needed to establish if MLA application reduces mechanical and infectious complications leading to CVC failure.

Unplanned dressing replacement places increased burden on healthcare budgets (staff time, dressing products), which is particularly challenging for low- and middle-income countries with heavy workloads and low staffing levels (41). Our study showed reduced costs associated with CVCs dressed with MLA. Based on this trial, implementing MLA plus standard care resulted in an expected saving of $11.02 per participant. Internationally, millions of CVCs are inserted annually in ICUs, so the impact of reducing dressing failure could be substantial.

Strengths of this study are its randomized design, allocation concealment, masking of infectious diseases physician and microbiology laboratory staff for infection outcomes and lack of missing outcome data. The inclusion of multiple sites, with varying standard care dressings, maximizes generalizability of study results and, although only jugular CVC were included, previous research has identified that this line type has the greatest need for innovative catheter securement (2, 11). Limitations included a lack of statistical power for some outcomes (e.g., microbiology, CVC failure). It was also not possible to mask patients and clinicians to the intervention; however, the risk of bias was minimized by high inter-rater reproducibility between ReNs for skin assessment, blinding staff allocating infection outcomes, and the statistician.

CONCLUSIONS

This multisite RCT demonstrated that MLA reduces dressing disruption and frequency of dressing changes in jugular CVCs. Therefore, the addition of MLA to the primary CVC dressing provides an opportunity for meaningful patient safety benefits and cost savings. A large RCT is urgently required to determine the impact of improved dressing integrity from the application of MLA on CLABSI and CVC failure rates.

ACKNOWLEDGMENTS

We thank participating patients and intensive care staff for their support of this trial; Jill Campbell and John Gowardman for assistance with study planning; and Nicholas Mifflin, Lynette Morrison, Joanne Sutton, Annabel Levido, Ryan Leese, Andrew Thomas, and Jiville Latu for patient recruitment and data collection.

Supplementary Material

ccm-53-e282-s001.docx ^{(48.6KB, docx)}