The aim of this systematic review was to assess the efficacy of heparin flushing in the lock of central venous catheters.
We searched MEDLINE and CINAHL databases. Eligible studies were randomized controlled trials evaluating the use of heparin versus normal saline or other solution in the flushing of central catheter among adult patients. No language restrictions were applied. Two reviewers independently screened titles and abstracts in order to identify relevant publications. The same two reviewers retrieved and evaluated full texts. Parameter estimates regarding catheter occlusion were pooled using network meta-analysis with Bayesian hierarchical modeling.
We identified 462 references. Eight studies were included. There was no evidence that heparin was more effective than normal saline in reducing occlusions. It was unclear whether urokinase and lepirudin were more effective than heparin in reducing occlusions. Vitamin C solution does not appear to prolong catheter patency.
There is no evidence of a different effectiveness between heparin flushing and normal saline or other solutions in reducing catheter occlusions. Due to the little and inconclusive evidence available in this field, further studies might be necessary.
J Vasc Access 2014; 15(4): 241 - 248
Article Type: REVIEW
Article Subject: Nursing
AuthorsAlberto Dal Molin, Elias Allara, Doriana Montani, Simona Milani, Cristina Frassati, Simonetta Cossu, Simone Tonella, Dania Brioschi, Laura Rasero
- • Accepted on 12/02/2014
- • Available online on 25/04/2014
- • Published in print on 31/07/2014
This article is available as full text PDF.
In order to prevent some complications such as obstruction, adequate flushing of central venous catheter is necessary (1, 2). The device must be flushed before and after drug administration or transfusion of blood components, after obtaining blood specimens and for device maintenance when not in use (3). Most manufacturers indicate that flushing must be performed every 4 weeks when the device is not in use (4). However, some studies suggest that less frequent flushing could be safe and feasible (5-6-7-8).
The aim of flushing is to clear the infused medication and to prevent contact between other drugs; the final flush solution is thought to be important to decrease the risk of occlusion (2, 9). The nurse must flush the catheter using a pulsated push-pause technique that creates turbulence within the device lumen allowing removal of debris. In order to prevent reflux of blood, the positive pressure for lock of catheter must be used (9). Flushing protocols vary by facility and type of vascular catheter, but in most cases it is performed with 10 to 20 mL of normal saline, followed by 5 mL of heparin solution (3). However, the effectiveness of heparin is unproven (10) and diversity of practice was documented by some surveys (11, 12). In particular, Sona et al (12) reported that in most cases the nurses use only physiological saline to flush central catheters and maintain patency. The authors concluded that a randomized controlled trial (RCT) was necessary to determine the adequate flushing solution.
Some adverse effects are associated with heparin use, such as autoimmune-mediated heparin-induced thrombocytopenia, allergic reactions and the potential for bleeding complications following multiple, unmonitored heparin flushes (10, 13, 14).
The use of normal saline instead of heparin in the lock of catheter is reported in some studies (15-16-17), and systematic reviews agree that in the peripheral venous catheter, heparin intermittent flushing is no more effective than flushing with normal saline alone (18-19-20).
The aim of this study was to compare the effectiveness of heparin over other solutions in catheter flushing among adult patients with central venous catheter.
We identified relevant primary studies by searching the MEDLINE and CINAHL databases.
We defined the clinical question by using the PICO framework (21) (
|Population||Patients using central venous catheter|
|Intervention||Use of heparin solution in the catheter flushing|
|Comparison||Other substances/solutions (such as normal saline)|
|Outcome||Obstructions, infections, venous thrombosis, heparin-induced thrombocytopenia, other complications related to the management of central venous catheter|
|PubMed||“Heparin” [Mesh] AND “Catheterization, central venous” [Mesh]||No||326|
|CINAHL||“Heparin” [Mesh] AND “Catheterization, central venous” [Mesh]||No||53|
|CINAHL||“Catheter Irrigation, vascular” [Mesh] AND “Catheterization, central venous” [Mesh]||No||10|
|CINAHL||“Catheter Irrigation, vascular” [Mesh] AND “Heparin” [Mesh]||No||72|
|Additional study found in review bibliography||1|
Eligible studies were RCTs evaluating the use of heparin versus normal saline or other solution in the flushing of central catheter in adult patients. No language restrictions were applied.
Observational studies, reviews or studies conducted in pediatrics patients or in patients with hemodialysis catheters or with peripheral venous catheter were excluded.
Additional primary studies were identified by the first author of reference list in published reviews.
Two reviewers (SM and CF) independently screened titles and abstracts in order to identify relevant publications. Full texts were retrieved and evaluated by the same two reviewers. Discrepancies were resolved by discussion with first author (ADM) and final decision was made by him.
The concordance of their revision was assessed using Cohen Kappa: if
For each included study, the following data were extracted:
First author, name of journal and year of publication
Number of patients included in the study
Characteristics of patients included in the study (mean age, sex)
Data were extracted by three independent reviewers (ST, SC and DM). Discrepancies were resolved by discussion.
Quality assessment of primary studies
The quality of the included studies was assessed with Critical Appraisal Skills Programme (CASP) for RCTs (23). This appraisal tool was structured by ten questions.
Outcomes of interest
Primary outcome of interest was occlusion. Secondary outcomes were venous thromboembolism, catheter-related bloodstream infection and heparin-induced thrombocytopenia (HIT). Studies without at least one of these outcomes were excluded from the analysis.
Parameter estimates of included studies were pooled for the primary endpoint, that is, occlusion of catheter, which was measured in all of the included studies. Estimates for secondary endpoints were summarized narratively due to the small number of studies available for each comparison.
Due to the presence of multiple interventions (heparin, sodium chloride, urokinase, lepirudin, and vitamin C), a network meta-analysis was performed to allow head-to-head comparisons for all the treatments considered in this review—even when a direct comparison was not available.
A Bayesian hierarchical modeling approach was chosen for its flexibility in accounting for the inherent correlation in multiarm trials and for using the exact likelihood of the data (e.g., Binomial or Poisson) rather than approximation to Normal (24). Parsimoniously, a random effects model with logit link was fitted. The posterior distribution was generated using the JAGS Monte Carlo Markov Chain sampler. We specified 500,000 iterations, 5,000 adaptation iterations and a thinning of 10 for each chain. Convergence to the target distribution was assessed via visual inspection of the sample plots, which revealed consistent trajectory of the chains over time. Due to a minimal asymmetry of the posterior distributions, median log odds ratios (ORs) were preferred over the empirical mean log ORs. The lower bound of each credible interval was obtained from the 2.5th percentile and the upper bound from the 97.5th percentile. Both point estimates and bounds of credible intervals were exponentiated to allow interpretation as ORs and their 95% credible intervals. The gemtc R package was used for such analyses and for generating diagnostic plots.
Main issues of network meta-analyses are loop inconsistency and design inconsistency (25, 26). Loop inconsistency, that is, the lack of agreement between direct and indirect comparisons, was not a problem in this review because all comparisons were either purely direct or purely indirect (
Structure of the network of included studies. Clockwise: L = Lepirudin; H = Heparin; VitC = Vitamin C; UK = Urokinase; NaCl = Sodium chloride.
Since network meta-analysis is a relatively recent and nonfully consolidated methodology, direct comparisons were also analyzed with traditional meta-analysis. Heterogeneity of intervention effects was assessed with Cochran’s Q tests. A random effects model was fitted and forest plots were generated for each direct comparison (heparin vs. sodium chloride, heparin vs. urokinase, heparin vs. lepirudin and heparin vs. vitamin C). The metaphor R package was used. Since the point estimates and inferences obtained with traditional and network meta-analysis were very similar, only the latter were shown.
We identified 462 references by the search strategy, conducted on January 20, 2014; 400 were excluded after title/abstract review. A very good correlation was observed between the two reviewers (Cohen’s
|Author||Study design||Patients||Interventions||Main Results|
|CI = confidence interval; CVC = central venous catheter; RCT = randomized controlled trial; RR = risk ratio.|
|Ray et al (1999)||RCT||n=105||Twice-daily heparin flushing with heparin (10 IU/mL)||
|Male = 65||versus||twice-daily heparin flushing with once-weekly urokinase instillation = 2 (4%)|
|Female = 40||Twice-daily heparin flushing with once-weekly urokinase instillation (9000 U/1.8 mL)||P<0.05|
|Heparin = 52 (49.5%)||
|Heparin and urokinase = 53 (50.5%)||twice-daily heparin flushing with once-weekly urokinase instillation = 3 (6%)|
|≥18 years of age Cancer patients with Hickwmann catheter||P=0.02|
|Twice-daily heparin flushing with heparin = 5 (10%)|
|twice-daily heparin flushing with once-weekly urokinase instillation = 6 (11%)|
|Solomon et al (2001)||RCT open label Conducted in two centers||n=100||Twice-weekly flushes of heparin (50 IU/5 mL) versus||
|Male = 33||Twice-weekly flushes of urokinase (5000 IU/2 mL)||
|Female = 67||
|>16 years Oncology patients with double-lumen Hickmann catheter||
|Heparin flushing = 48 (48%) Urokinase flushing = 52 (52%)|
|Rabe et al (2002)||RCT blinding||n=99||Sodium chloride 0.9% solution||
|Male = 42||versus||heparin solutions 5000 IU/mL|
|Female = 57||versus|
|≥18 years of age||Vitamin C 200 mg/mL|
|Patients, with three-lumen CVC, admitted on a bed medical Intensive care unit|
|Sodium chloride = 33 (33%) Heparin = 33 (33%) Vitamin C = 33 (33%)|
|Horne et al (2006)||RCT, double blinded||n=49||3 mL heparin 100 IU/mL||
|Male = 33||versus||RR 1.6; 95% CI 0.40-13.86; p=0.70|
|Female = 16||3 mL lepirudin 100 μg/mL|
|≥ 18 years of age|
|Cancer patients with an open-ended, double or triple lumen, subcutaneously tunneled central venous access device|
|Heparin group = 24 (49%) Lepirudin group = 25 (51%)|
|Bowers et al (2008)||RCT open label||n=102||0.9% sodium chloride injections versus||
|Male = 51||Heparin lock flush 100 U/mL||
|Female = 51|
|≥18 years of age|
|Peripherally inserted central catheters (PICC)|
|Normal saline group = 50 (49%) Heparin group = 52 (51%)|
|Fuentes i Pumarola et al (2007)||RCT, double blinded||Phase 1 n = 128 Heparin 500 IU/5 mL=49 (38.3%) Heparin 100 IU/5 mL=79 (61.7%)||Phase 1 Heparin 500 IU/5 mL versus Heparin 100 IU/5 mL||
|Phase 2 n=95 Normal saline = 57 (60%) Heparin = 38 (40%)||Phase 2 Normal saline versus Heparin 100 IU/5 mL||Phase 2 Normal saline = 0 Heparin 100 IU/5 mL = 0 p=ns|
|≥18 years of age Patients admitted to critical care unit three-lumen CVC|
|Schallom et al (2012)||RCT open label||n=341 patients randomized n=295 patients included in the analysis||0.9% sodium chloride versus Heparin 10 U/mL||
|Male = 151 Female =144||
|0.9% sodium chloride = 150 Heparin = 145||
|≥18 years of age Patients admitted to critical care unit Multilumen CVC|
|Goossens et al (2013)||RCT open label||n=802 patients randomized n=765 patients included in the analysis||Normal saline versus Heparin (300 U/3 mL)||
|Normal saline = 382 Heparin = 383||
|≥18 years of age Patients with port access|
Occlusion of catheter was measured in seven two-arm studies (17, 28-29-30-31-32-33) and in one three-arm study (34). Direct comparisons were feasible for heparin vs. sodium chloride, heparin vs. urokinase and heparin vs. vitamin C.
Results of pooled analyses are shown in
RANDOM EFFECTS NETWORK META-ANALYSIS FOR OCCLUSION OF CATHETER
|Heparin vs. sodium chloride||5 (1)||110/800 vs. 136/885||0.55 (0.12 to 1.37)|
|Heparin vs. urokinase||2 (0)||38/100 vs. 32/105||1.99 (0.44 to 12.48)|
|Heparin vs. lepirudin||1 (0)||3/24 vs. 5/25||0.54 (0.04 to 7.09)|
|Heparin vs. vitamin C||1 (1)||25/33 vs. 31/33||0.22 (0.01 to 2.16)|
|Urokinase vs. sodium chloride||None||–||0.27 (0.02 to 1.39)|
|Urokinase vs. lepirudin||None||–||0.27 (0.01 to 5.05)|
|Urokinase vs. vitamin C||None||–||0.10 (0.00 to 1.60)|
|Lepirudin vs. sodium chloride||None||–||0.98 (0.04 to 14.59)|
|Lepirudin vs. vitamin C||None||–||0.39 (0.01 to 12.75)|
|Vitamin C vs. sodium chloride||None||–||2.44 (0.19 to 33.35)|
Heparin vs. normal saline
Rabe et al randomized 99 patients with three-lumen central venous catheters, in three treatment groups: sodium chloride 0.9%, vitamin C (200 mg/mL) and heparin (5,000 IU/mL). They found significant differences in catheter patency among the groups (p<0.03, long-rank test). In particular, catheter survival was higher in the catheter group flushing with heparin than the group flushing with sodium chloride 0.9% (p<0.04, long-rank test). No statistical difference in catheter patency was found comparing sodium chloride and vitamin C flushing (p<0.56, long-rank test) (34).
Fuentes i Pumarola et al have structured a blind RCT in two phases. In the first phase they compared flushing with sodium heparin 100 IU and sodium heparin 500 IU, while in the second phase they randomized catheters in two groups: heparin and saline flushing. They found no statistically significant differences in catheter patency between groups. This study was characterized by high attrition rate: in the first phase only 128 of 291 catheters were analyzed (49 in 500 IU heparin group and 79 in 100 IU heparin group), while in the second phase only 95 out of 250 were analyzed (38 heparin flushing and 57 saline flushing) (28).
Similar results were highlighted in the randomized open label trial by Schallom et al where they randomized patients, with multilumen central venous catheters, in heparin flushing versus saline 0.9% sodium chloride group. The occlusion rate was higher in the NaCl group, but this difference was not statistically significant (6.3 vs. 3.8; risk ratio [RR] 1.66; 95% CI 0.86 to 3.22, p=0.136). Four catheter-related bloodstream infections developed in the saline group (3.1 per 1,000 catheter day vs. 0 per 1,000 catheter days, p=0.125). The authors suggested that the 0.9% sodium chloride might be used in catheter flushing for short term (17).
Bowers et al structured a nonblinded RCT in which they randomized 102 patients with peripherally inserted central catheters (PICCs) in two groups: 0.9% sodium chloride injections and heparin 100 U/mL lock flush. Significant differences were present in patient characteristics between the groups. The no-patency rate was higher in the saline group (6% vs. 0%), but this difference was not statistically significant. The average duration of PICC was 2.1 in the normal saline group and 2.9 in the heparin group (p=ns) (29).
In a recent noninferiority open trial 802 cancer patients were randomized to heparin lock or to normal saline lock. The incidence rate of easy injection, impossible aspiration was 3.70% in the sodium chloride group and 3.92% in heparin group. The relative risk was 0.94% (95% CI 0.67-1.32%) (30).
Heparin vs. other solutions
Ray et al designed a prospective, controlled, randomized study in which 105 patients with Hickmann catheter were allocated to two groups: heparin flushing or heparin flushing and urokinase flushing. The results of their study indicate that the use of urokinase reduces catheter-related complications. In particular, the infections and fibrin sheath formed was higher in the heparin group than in the group where catheters were flushed with heparin and urokinase (31).
Solomon et al also compared heparin flush to a urokinase flush in another RCT open-label study, but they concluded that urokinase was not reducing the frequency of Hickmann complications. An elevated drop-out rate was present (32).
In another study (33) 49 adults undergoing bone marrow transplantation for hematologic malignancies or metastatic solid tumors were randomized to lepirudin flushes or heparin flushes. The authors concluded that lepirudin was not more effective than heparin to reduce withdrawal occlusion (RR 1.6; 95% CI 0.40 to 13.86; p=0.70).
Rabe et al have structured a RCT in which patients were randomized to three intervention groups and conclude that vitamin C solution does not prolong catheter patency (34).
One study compared different dosages of heparin for flushing the central venous catheter. Fuentes i Pumarola et al indicate that the occlusion rate is not different if flushing is performed with heparin 500 IU/5 mL or with heparin 100 IU/5 mL (4.9% vs. 4.5%; p=0.937) (28).
Quality assessment of RCT studies
In all trials the focus of the study was clearly defined and the RCT was appropriately carried out. The methods of randomization and allocation have not always been clearly described and one study (Bowers et al) showed a statistically significant difference (gender) between the groups. Many studies were conducted in a open-label or single-blind fashion, thus determining the possibility of the presence of some bias, while in some other studies there has been an important dropout of patients (such as Fuentes i Pumarola et al). In some cases the sample of study was small. Calculations for sample power calculations were not always being shown.
The central venous catheter is widely used in clinical practice, but not without complications (6, 35). Nursing is important in order to reduce complications. Flushing is required for assuring the function of the catheter and it must be performed using turbulent flush technique and positive-pressure locking techniques (9, 36). Heparin flushes are normally used to prevent thrombus formation and to reduce occlusion of catheter. However, there is still no consensus about this practice, and the use of heparin can be associated with complications such as autoimmune-mediated HIT, allergic reactions and the potential for bleeding complications following multiple, unmonitored heparin flushes (10).
The aim of this systematic review of RCTs was to determine the efficacy of heparin flushing in the central venous catheter. In our review we did not include studies conducted in pediatric patients and in patients with hemodialysis catheters.
One retrospective observational cohort study conducted in 610 patients with totally implantable long-term central vascular access shows no statistically significant differences for occlusive events between the group where the catheter was flushed with heparin solution and that of normal saline (39).
The use of heparin is not risk free. Garajová et al reported one case of heparin-induced delayed hypersensitivity after Port-a-Cath heparinization in a 79-year-old female patient, for whom heparin flushing (50 IU/5 mL) was performed every 30-40 days to prevent clotting. This reaction was developed after 52 months of Port-a-Cath maintenance (14).
The heparin concentration documented in the trials was wide. In accordance with the International Infusion Nursing Society heparin lock solution 10 units/mL is the preferred lock solution after each intermittent use. In order to reduce the risk of HIT, all patients must be monitored and heparin should be discontinued immediately if signs or symptoms of HIT appear (2).
Another not completely resolved issue is whether the use of urokinase is effective for patency. One study (31) suggests that twice-daily heparin flushing with once-weekly urokinase instillation is more effective than twice-daily heparin flushing with heparin. However, another study (32) in which patients were randomized to twice-weekly flushes of heparin or to twice-weekly flushes of urokinase indicates that there is no evidence of a difference in rates of occlusions, infections and venous thrombosis. These results suggest the possibility that the catheter complications may be reduced with the associated use of heparin and urokinase. However, further randomized controlled studies should be conducted to confirm this possibility.
Occlusion rate is higher in patients for whom flushing is performed with vitamin C rather than heparin. No different rate of occlusion was identified between lepirudin and heparin.
This review has some limitations. We searched only MEDLINE and CINAHL, without searching for gray literature. However, those are the major medical/nursing databases and we did not set any language constraints. Thus, we feel that our search provides an acceptable overview of the studies which are currently in the public domain.
In conclusion, our data suggest that heparin can be used in the clinical practice for flushing the catheter when indicated by the manufacturer. Nevertheless, further studies may be necessary in this field to clarify whether saline solution may be a viable and cheaper alternative to heparin. However, to this day there is not enough evidence supporting the use of saline solution in catheter flushing.
The authors wish to thank Mauro Pittiruti for his advice in reviewing the manuscript.
- Dal Molin, Alberto [PubMed] [Google Scholar] 1, * Corresponding Author (email@example.com)
- Allara, Elias [PubMed] [Google Scholar] 2
- Montani, Doriana [PubMed] [Google Scholar] 3
- Milani, Simona [PubMed] [Google Scholar] 4
- Frassati, Cristina [PubMed] [Google Scholar] 4
- Cossu, Simonetta [PubMed] [Google Scholar] 4
- Tonella, Simone [PubMed] [Google Scholar] 4
- Brioschi, Dania [PubMed] [Google Scholar] 4
- Rasero, Laura [PubMed] [Google Scholar] 5
School of Nursing, Biella Hospital, Avogadro University, Biella - Italy
Department of Translational Medicine, Avogadro University, Novara - Italy
School of Nursing, Novara Hospital, Avogadro University, Novara - Italy
Biella Hospital, Biella - Italy
Florence University, Florence - Italy