Introduction

Almost all patients admitted to hospital require a peripheral intravenous catheter to provide access for administration of drugs and fluids and parenteral nutrition. In addition, many critically ill patients require arterial catheterisation for haemodynamic monitoring and blood sampling. Maintenance of the patency of these indwelling catheters is important for minimising patients' discomfort and the expense associated with replacement. Vascular thrombosis,¹ visible scarring, and infection related to the catheter² are complications associated with use of these indwelling vascular devices.

The anticoagulant properties of heparin led clinicians to use heparin flushes or heparinised infusion in an attempt to prevent thrombus formation and to prolong the duration of catheter patency. The effective dose of heparin, however, has not been clearly established for venous and arterial catheters. Two meta-analyses evaluating use of heparin flush solutions for peripheral intermittent infusion devices concluded that the effect of heparin flushes was equivalent to that of 0.9% sodium chloride flushes. Both meta-analyses combined the results of controlled and uncontrolled trials. ^{3 4} Goode et al included 17 studies (seven randomised controlled trials)³ and Peterson et al included 20 studies (three randomised controlled trials).⁴ Peterson et al combined trials that evaluated continuous infusion of heparinised solution with trials that assessed intermittent flushing in catheters locked between use.⁴ None the less, these results led some organisations to state that sodium chloride injection should be the standard of care for maintaining intravenous catheters used for peripheral intermittent infusion.⁵

Despite its beneficial antithrombotic effects, decreasing unnecessary exposure to heparin is important to minimise the complications resulting from sensitisation. Autoimmune mediated thrombocytopenia induced by heparin occurs in about 3% of patients exposed to unfractionated heparin, which greatly increases the risk of thrombotic events. ^{6 7} Heparin induced thrombocytopenia is a risk even in newborns.⁸ Other risks of heparin use include allergic reactions and the potential for bleeding complications after multiple, unmonitored heparin flushes.⁵

A large percentage of heparin exposure in patients in hospital is through heparin use in peripheral venous catheters. If the risks associated with heparin use are considered what is the benefit of using heparin in peripheral venous or arterial access devices? Individual trials of heparin in peripheral venous catheters are contradictory, and there are no systematic reviews assessing various heparin dosing strategies for arterial catheters. We therefore conducted this systematic review to resolve and synthesise the conflicting literature. We have critically appraised the clinical trials evaluating use of heparin in continuous and intermittent infusion solutions on the duration of patency of peripheral venous and arterial catheters and on phlebitic complications.

	Methods

Study identification
Trials included in this review were identified by cross referencing the following MeSH terms from Medline from 1966 to April 1997: "catheterisation" and "catheters, indwelling" and "heparin" with "randomisation," "random allocation," "randomised controlled trial(s), randomised response technique," and "(controlled) clinical trials, randomised." Embase was searched from 1974 through 1996 by using the search terms "catheter" and "catheterisation, intravascular, random," and "heparin." After examining the full manuscripts of all abstracts deemed potentially relevant we reviewed the reference lists of each retrieved article and obtained the manuscript of any reference considered to be a randomised controlled trial. The trials included in two meta-analyses ^{3 4} were retrieved. Package inserts from catheter kits were searched for references regarding published and unpublished data. We also contacted companies manufacturing heparin bonded catheters regarding other unpublished and published randomised controlled trials. In addition, we hand searched the National Intravenous Therapy Association Journal from 1985 to 1992. 

Study selection
The following selection criteria were used to identify studies for inclusion in this analysis: study designrandomised controlled clinical trial; populationadult or paediatric patients; interventionheparin infused through the catheter via intermittent or continuous flush versus a control group with no heparin; outcomescatheter patency, catheter related phlebitis, catheter thrombus, infusion failure.

Data abstraction
Data abstraction was conducted by two investigators; disagreement was resolved by consensus. To evaluate agreement we calculated a quadratic weighted  for each item. Data on the number of catheters or the numbers of patients, or both, were abstracted in the form in which they were reported. Catheters were the unit of analysis when data were pooled because this was the way that most results were reported. We tried to contact authors to provide further information when the data necessary for critical appraisal or analysis, or both, were missing or unclear.

Definitions
The following definitions of terms were used. Duration of catheter patency was the number of hours the catheters were in place. Loss of patency was removal of the catheter because of inability to flush it. Catheter thrombus referred to a clot adherent to or occluding the catheter. Catheter related phlebitis indicated the presence of any one or more of the following: pain, erythema, induration, or a palpable venous cord at the catheter site. Infusion failure was loss of patency, phlebitis, or infiltration resulting in premature removal of the catheter.

Data analysis
We combined data to estimate the relative risks and associated 95% confidence intervals across studies by using the DerSimonian and Laird random effects model.⁹ We tested for heterogeneity (major differences in the apparent effect of the interventions across studies) by using the method proposed by Fleiss.⁹ We have reported tests of heterogeneity of variance in the results only when they were significant (P<0.05).

	Results

Study identification and selection
Twenty six trials of heparin use in peripheral venous catheters were identified in which random assignment was used, and 13 were included12 published trials^10-21 and one unpublished (FD Craig and SR Anderson, Harrison Methodist Fort Worth Hospital, personal communication). Two trials of heparin use in peripheral arterial catheters were identified in which random assignment was used and both were included. ^{22 23} Three trials claiming random allocation that actually used alternate assignment or assignment by odd-even hospital number were excluded.^24-26 Five trials were excluded that randomised by hospital unit or wards instead of individual patient because only two units were randomised and a before-after design was applied within each unit.^27-31 One randomised study was excluded because all patients received 5000 U heparin subcutaneously for prophylaxis of deep venous thrombosis.³² Three randomised studies were excluded because more than 40% of observations were not reported after randomisation.^33-35 One randomised study was excluded because only half the patients were randomised and the rest were allocated to treatment arms at the discretion of the physician.³⁶ The authors of one unpublished trial were unable to provide the necessary primary data (N Bell, D Brown, L Poon, Eden Hospital Medical Centre, California, personal communication). One randomised trial of peripheral venous catheters was performed in patients treated with cephalothin³⁷ and another was done in patients receiving a lignocaine infusion,³⁸ both of which are associated with higher rates of phlebitis leading us to exclude these trials from our analysis. We were unable to include the results of one unpublished randomised study because they reported the number of events per patient, many of whom had more than one catheter, and the primary author could not re-extract the data (A Kasparek, J Wenger, R Feldt, Mercy Medical Centre, Iowa, personal communication).

Trial characteristics and assessment of quality
The populations, interventions, number of patients, number of catheters, catheter gauges used, and methodological characteristics of the studies included in the final analysis are described in table 1. For peripheral venous catheters, intermittent heparin flushes varied from 10 U/ml to 100 U/ml and continuous heparin infusion was 1 U/ml. In the trial by Moclair et al all patients received a glyceryl trinitrate transdermal patch and twice daily application of hydrocortisone cream to the infusion site in an attempt to prolong vein survival and decrease phlebitis.¹⁰ The doses of heparin used in trials evaluating continuous flush in arterial catheters varied, and the actual dose was not reported in the largest trial as it was an effectiveness study of any amount of heparin versus no heparin.²³

Duration of catheter patency
Table 2 shows the effect of heparin on duration of catheter patency in six trials. We were unable to pool the results because of differences in reporting. The two trials of intermittent heparin flushes at concentrations of 10 U/ml¹¹ and 100 U/ml¹⁵ showed no effect on duration of catheter patency. Two trials showed that heparin added to the infusion to make a concentration of 1 U/ml effectively prolonged peripheral venous catheter patency ^{18 20} and two trials showed non-significant trends in the direction of benefit. ^{10 17} The results could not be pooled because not all investigators reported the standard deviation around the mean and some reported the median. In peripheral arterial catheters, heparin significantly prolonged the duration of catheter patency in two trials, ^{22 23} although investigators reported the percentage of catheters patent at 72 hours (4 U/ml normal saline; 90% of heparin catheters v 79% of control catheters; difference 11%; P<0.05²³) and 96 hours (variable dose; 86% v 52%; difference 34%; P<0.01²²) and not the average number of hours catheters were patent.

Catheter clotting and loss of patency
Figure 1 shows that use of 10 U intermittent heparin flushes had no effect on catheter clotting compared with normal saline when the results of two trials were pooled (FD Craig and SR Anderson, Harrison Methodist Fort Worth Hospital, personal communication, and Shoaf and Oliver¹²). At doses of 100 U/ml flushed every 6 or 8 hours heparin was associated with a significantly lower loss of catheter patency when the results of two trials were pooled. ^{15 16} Heparinised infusion significantly decreased loss of patency in arterial catheters when the results of two trials were pooled. ^{22 23}

View larger version (19K): [in this window] [in a new window]   Fig 1.    Relative risks (95% confidence intervals) for catheter clot formation in and loss of patency of peripheral venous and arterial catheters according to treatment with or without heparin. Last two columns are numbers of catheters with clots/total number of catheters

Infusion failure
Figure 2 shows that use of intermittent 10 U heparin flushes had no effect on infusion failure rates for peripheral intravenous catheters when the results of two trials were pooled (FD Craig and SR Anderson, Harrison Methodist Fort Worth Hospital, personal communication, and Shoaf and Oliver¹²). Addition of heparin at a concentration of 1 U/ml to the infusion was associated with a reduced risk of infusion failure when the results of three trials were pooled. ^{10 17 20}

View larger version (13K): [in this window] [in a new window]   Fig 2.    Relative risks (95% confidence intervals) for failure of infusion in peripheral venous catheters according to treatment with or without heparin. Last two columns are numbers of catheters with failure of infusion/total number of catheters

Catheter related phlebitis
Figure 3 shows that there was a significant difference in the risk of phlebitis when the results of three trials of 10 U/ml intermittent heparin flushes versus normal saline were pooled (FD Craig and SR Anderson, Harrison Methodist Fort Worth Hospital, personal communication). ^{11 14} When the results of two trials of 100 U/ml of intermittent heparin flush were pooled phlebitis was significantly decreased. ^{15 16} The test of homogeneity, however, was significant (P=0.0006) for the decreased risk of phlebitis, with one trial that used 100 U/ml every 6 hours¹⁶ showing a much larger but non-significant trend in the direction of heparin being beneficial than the trial that used 100 U/ml every 8 hours.¹⁵ When the data from seven trials of heparin at concentrations of 1 U/ml infusion flushed continuously through the catheter were pooled (see figure 2) there was a significant decrease in phlebitis with use of heparin.

View larger version (26K): [in this window] [in a new window]   Fig 3.    Relative risks (95% confidence intervals) of phlebitis from peripheral venous catheters according to treatment with or without heparin. Last two columns are numbers of catheters associated with phlebitis/total number of catheters

	Discussion

Use of heparin as an antithrombotic agent in catheters has been widespread for over 20 years. Despite almost universal use, the benefit of heparin has not been firmly established. Half of the available trials claiming to be randomised had to be excluded because of quality considerations or the presence of potentially confounding cointerventions. The evidence supporting use of heparin in peripheral arterial catheters comes mainly from one large (5139 patients) randomised trial including 198 sites in which various heparin dosing strategies were used.²³ The limited evidence available suggests that use of heparin as an intermittent flush solution at a concentration of 10 U/ml in catheters locked between episodes of use is not beneficial. Use of heparin in peripheral arterial catheters will prolong their life and utility. Current evidence does not allow us to make firm conclusions regarding the benefit of adding heparin to the solutions infused continuously through peripheral venous catheters, but this intervention warrants further study.

Use of heparin in peripheral venous catheters
Our meta-analyses included three randomised controlled trials of intermittent heparin flushes and seven randomised controlled trials of continuous infusion of heparinised solution that were not included in the two previously published meta-analyses. ^{3 4} Our finding that heparin at doses if 10 U/ml for intermittent flushing is no more beneficial than flushing with normal saline alone is in agreement with the results of these meta-analyses, which combined controlled and uncontrolled studies. ^{3 4} This intervention has been evaluated in only four truly randomised controlled, double blind trials including a total of 652 catheters. These trials involved different populations and evaluated different outcomes. Added to the larger number of uncontrolled studies, however, the weight of the evidence supports discontinuation of use of 10 U/ml heparin flush in intermittent intravenous infusion devices.

Use of heparin in peripheral arterial catheters
Mostly on the basis of the results of one multicentre study heparin has been shown effectively to prolong the life of peripherally placed arterial pressure monitoring devices. The minimal effective dose of heparin, however, has not been established. Bolgiano et al reported no significant difference in duration of arterial catheter patency when heparin was used at 0.25 U/ml versus 1 U/ml in adults.⁴² Butt et al reported that increasing the heparin concentration from 1 U/ml to 5 U/ml in 22 gauge catheters in children significantly prolonged arterial catheter patency.⁴³ The type of solution, however, may be important as Rais-Bahrami et al reported that neonatal peripheral arterial lines infused continuously with heparinised normal saline functioned significantly longer (107 (SD 71) hours) than those with heparinised 5% dextrose (39 (32) hours).⁴⁴ Other agents besides heparin have also been shown to be effective in prolonging the duration of patency of radial arterial catheters. Arterial catheter solutions containing papaverine⁴⁵ and 1.4% sodium citrate⁴⁶ effectively prolong the duration of catheter patency and their risk profile should be compared with that of heparin.

Conclusions
In this systematic review we have clarified that low dose heparin is beneficial for maintaining peripherally placed arterial catheters when added to the continuously infused solutions. Heparin at a concentration of 1 U/ml infused continuously through peripheral venous catheters is a promising intervention to prolong catheter life but requires further study. While the use of 100 U/ml of intermittent heparin flushes for peripheral intravenous catheters needs further evaluation, evidence currently available suggests that the current use of 10 U/ml as an intermittent flush is no more effective than normal saline flush.