Heparin versus 0.9% sodium chloride locking for prevention of occlusion in central venous catheters in adults

Eduardo López-Briz; Vicente Ruiz Garcia; Juan B Cabello; Sylvia Bort-Martí; Rafael Carbonell Sanchis

doi:10.1002/14651858.CD008462.pub4

Sellado con heparina versus con cloruro de sodio al 0,9% para la prevención de la oclusión del catéter venoso central en adultos

Declaraciones de intereses de los autores

Versión publicada: 18 julio 2022 Historial de versiones

https://doi.org/10.1002/14651858.CD008462.pub4

Contraer todo Desplegar todo

Resumen

disponible en

Antecedentes

El sellado intermitente de los catéteres venosos centrales (CVC) se realiza para ayudar a mantenerlos permeables y funcionales. Existen variaciones sistemáticas en la asistencia: algunos profesionales utilizan heparina (a diferentes concentraciones), mientras que otros utilizan cloruro de sodio al 0,9% (suero fisiológico normal). Esta revisión analiza la efectividad y la seguridad del sellado intermitente con heparina en comparación con el suero fisiológico normal para ver si la evidencia establece que una es mejor que la otra. Esta es una actualización de una revisión Cochrane anterior.

Objetivos

Evaluar los efectos beneficiosos y perjudiciales del sellado intermitente de los CVC con heparina versus suero fisiológico normal en adultos para prevenir la oclusión.

Métodos de búsqueda

Se utilizaron los métodos exhaustivos estándares de búsqueda de Cochrane. La última fecha de búsqueda fue el 20 de octubre de 2021.

Criterios de selección

Se incluyeron los ensayos controlados aleatorizados en adultos ≥ 18 años de edad con un CVC que compararon el sellado intermitente con heparina a cualquier concentración versus suero fisiológico normal. En esta revisión se excluyeron los estudios en lactantes y niños.

Obtención y análisis de los datos

Se utilizaron los métodos Cochrane estándares. Los desenlaces principales de esta revisión fueron la oclusión de los CVC y la duración de la permeabilidad del catéter. Los desenlaces secundarios de esta revisión fueron las infecciones de la sangre relacionadas con el CVC y la colonización relacionada con el CVC, la mortalidad, la hemorragia, la trombocitopenia inducida por la heparina, la trombosis relacionada con el CVC, el número de inserciones adicionales de CVC, la anomalía del perfil de coagulación y las reacciones alérgicas a la heparina. Se utilizó GRADE para evaluar la calidad de la evidencia de cada desenlace.

Resultados principales

En esta actualización se identificó un nuevo ECA con 30 participantes. Se incluyeron 12 ECA con 2422 participantes. Todos los ECA aportaron datos para el metanálisis. Se observaron diferencias en los métodos utilizados por los estudios incluidos y hubo variación en las concentraciones de heparina (10 a 5000 UI/ml), el período de seguimiento (uno a 251,8 días) y la unidad de análisis utilizada (participante, catéter, línea de acceso). Cinco estudios incluyeron a pacientes de la UCI (unidad de cuidados intensivos), dos estudios incluyeron a pacientes oncológicos y los restantes incluyeron a pacientes diversos (nefropatía crónica, hemodiálisis, pacientes con atención domiciliaria, etc.).

Desenlaces principales

En general, los desenlaces combinados podrían mostrar menos oclusiones con la heparina en comparación con el suero fisiológico normal, pero no está claro (razón de riesgos [RR] 0,70; intervalo de confianza [IC] del 95%: 0,51 a 0,95; diez estudios; 1672 participantes; evidencia de certeza baja). Se agruparon los estudios que utilizaron el participante o el catéter como unidad de análisis.

Se realizó un análisis de subgrupos por unidad de análisis. No se detectaron diferencias claras tras analizar las diferencias entre subgrupos (p = 0,23).

No se encontró evidencia clara de una diferencia en la duración de la permeabilidad del catéter con heparina en comparación con la solución salina normal (diferencia de medias [DM] 0,44 días; IC del 95%: ‐0,10 a 0,99; seis estudios; 1788 participantes; evidencia de certeza baja).

Desenlaces secundarios

No se encontró evidencia clara de una diferencia en los siguientes desenlaces: las infecciones de la sangre relacionadas con el CVC (RR 0,66; IC del 95%: 0,08 a 5,80; tres estudios; 1127 participantes; evidencia de certeza muy baja); la mortalidad (RR 0,76; IC del 95%: 0,44 a 1,31; tres estudios; 1100 participantes; evidencia de certeza muy baja); la hemorragia (RR 1.54; IC del 95%: 0,41 a 5,74; tres estudios; 1197 participantes; evidencia de certeza muy baja); o la trombocitopenia inducida por la heparina (RR 0,21; IC del 95%: 0,01 a 4,27; tres estudios; 443 participantes; evidencia de certeza muy baja).

Los motivos principales para disminuir la certeza de la evidencia de los desenlaces principales y secundarios fueron la ocultación poco clara de la asignación, la sospecha de sesgo de publicación, la imprecisión y la inconsistencia.

Conclusiones de los autores

Debido a la evidencia de certeza baja, no está claro si el sellado intermitente con heparina da lugar a menos oclusiones de los catéteres venosos centrales que el sellado intermitente con suero fisiológico normal en adultos. Evidencia de certeza baja indica que la heparina podría tener poco o ningún efecto en la duración de la permeabilidad del catéter. Aunque no se encontró evidencia de diferencias en la seguridad (infecciones de la sangre relacionadas con el CVC, mortalidad o hemorragia), los estudios combinados no tenían suficiente poder estadístico para detectar eventos adversos poco frecuentes como la trombocitopenia inducida por la heparina. La realización de nuevos estudios de investigación durante períodos más prolongados reduciría las incertidumbres actuales.

PICO

Population

Intervention

Comparison

Outcome

El uso y la enseñanza del modelo PICO están muy extendidos en el ámbito de la atención sanitaria basada en la evidencia para formular preguntas y estrategias de búsqueda y para caracterizar estudios o metanálisis clínicos. PICO son las siglas en inglés de cuatro posibles componentes de una pregunta de investigación: paciente, población o problema; intervención; comparación; desenlace (outcome).

Para saber más sobre el uso del modelo PICO, puede consultar el Manual Cochrane.

Resumen en términos sencillos

disponible en

¿El sello con heparina previene la obstrucción de los catéteres venosos centrales en adultos, en comparación con el sellado con suero fisiológico normal?

Mensaje clave

No se encontró evidencia clara de una diferencia entre la heparina y el suero fisiológico normal (solución estéril de sal en agua) en la prevención de obstrucciones (oclusión) de los catéteres venosos centrales, ni en el tiempo que los catéteres permanecieron desobstruidos, ni en el número de efectos secundarios como infecciones, muerte, hemorragias, etc. Se necesitan más estudios bien diseñados y a gran escala para reducir las incertidumbres.

¿Por qué es importante esta pregunta?

Los catéteres venosos centrales son tubos (también llamados "vías") que se deben colocar temporalmente en las venas de los pacientes, a las que hay que acceder regularmente por razones médicas. Se insertan en vasos grandes que conducen al corazón. Mientras no se utiliza, se inyecta un líquido en el catéter hasta su próximo uso para evitar que se formen coágulos de sangre que puedan obstruir el catéter. A este procedimiento se le llama sellar el catéter. La sustitución de los catéteres aumenta el coste de la asistencia, podría retrasar el tratamiento y supone un riesgo adicional de episodios adversos para el paciente relacionados con el catéter. El catéter también se podría infectar, dando lugar a infecciones de la sangre. Los líquidos utilizados para el sellado son la heparina o el suero fisiológico normal. La heparina, que es un anticoagulante, se utiliza para evitar la coagulación de la sangre. También podría ayudar a evitar que los catéteres se obstruyan; sin embargo, también puede provocar hemorragias, reacciones alérgicas y un descenso del número de plaquetas en la sangre. Esto ha planteado la pregunta de si la heparina es mejor que el suero fisiológico para evitar obstrucciones, y la seguridad de cada método.

¿Qué se hizo?

Se buscaron ensayos controlados aleatorizados que evaluaran si el sellado de los catéteres con heparina fue más eficaz para reducir el riesgo de obstrucción y de infecciones en comparación con el suero fisiológico normal. En los ensayos controlados aleatorizados, los tratamientos que reciben las personas se deciden de forma aleatoria y estos ofrecen la evidencia más fiable sobre los efectos del tratamiento.

¿Qué se encontró?

En esta actualización se identificó un nuevo estudio. Se incluyeron 12 estudios con 2422 personas. Cinco estudios incluyeron pacientes de la UCI, dos estudios incluyeron pacientes con cáncer y los restantes incluyeron pacientes diversos (hemodiálisis, pacientes de atención domiciliaria, etc.). No se puede concluir que el sellado de los catéteres con heparina evite la obstrucción mejor que el lavado con suero fisiológico normal. Se observó poca o ninguna diferencia en el período de tiempo que el catéter permaneció desobstruido o en el número de efectos secundarios entre el uso de heparina o de solución salina.

¿Qué certeza se tiene en la evidencia?

Al comparar la heparina con la el suero fisiológico, la certeza de la evidencia de los resultados varió de muy baja a baja debido al diseño de los estudios y a que el resultado general incluía la probabilidad de efectos beneficiosos y perjudiciales.

¿Cuál es el grado de actualización de esta evidencia?

Esta revisión Cochrane actualiza la evidencia anterior. La evidencia está actualizada hasta el 20 de octubre de 2021.

Authors' conclusions

Implications for practice

Given the low‐certainty evidence, we are uncertain whether intermittent locking with heparin results in fewer central venous catheter occlusions than intermittent locking with normal saline in adults. Low‐certainty evidence suggests that heparin may have little or no effect on catheter patency duration. Although we found no evidence of differences in safety (CVC‐related bloodstream infections, mortality, or haemorrhage), the combined studies were not powered to detect rare adverse events such as heparin‐induced thrombocytopaenia. We are uncertain about the effects of heparin compared to normal saline, and review findings should be interpreted with caution.

Implications for research

Better designed large‐scale randomised controlled trials are needed to definitively establish or rule out a net benefit of locking with heparin versus normal saline; these trials should also explore effectiveness in different patient groups, such as patients under haemodialysis or those with onco‐haematological malignancies. Trials should report the outcome using both the participant and the catheter as units of analysis to allow evidence to be combined more consistently. Occlusions and adverse events must be the focus of future studies, and we suggest at least one month of follow‐up. In addition, assessment by type of line (i.e. dialysis/apheresis versus peripherally inserted central catheter (PICC) or versus other) is important. Addressing the question of harm from rare events requires high‐quality prospective cohort studies with sufficient follow‐up. Decision analytical modelling incorporating the costs of heparin and normal saline and the probabilities and costs of alteplase use and catheter replacement may also help to establish the thresholds required to conclude which method is the most appropriate and efficient choice.

Summary of findings

Open in table viewer

Summary of findings 1. Heparin versus normal saline solution locking for prevention of occlusion in central venous catheters in adults

Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	Number of participants (studies)	Certainty of the evidence (GRADE)	Comments
Heparin versus normal saline solution locking for prevention of occlusion in central venous catheters in adults
Patient or population: adults with CVCs Settings: hospital Intervention: heparin Comparison: normal saline solution (0.9% NaCl)
Outcomes	Risk with normal saline	Risk with heparin	Relative effect (95% CI)	Number of participants (studies)	Certainty of the evidence (GRADE)	Comments
Occlusion of CVC (combining participant, and catheter as unit of analysis) Blood withdrawing Follow‐up: 1 to 231 days	Study population		RR 0.70 (0.51 to 0.95)	1672 participants, 1025 catheters (10 RCTs)	⊕⊕⊝⊝ Low¹	Heparin may reduce the rate of CVC occlusion when compared to normal saline, NNTB: 42 (95% CI 32 to 252). Considering only participant as unit of analysis: RR 0.79 (95% CI 0.58 to 1.08; 7 studies, 1672 participants) Considering only catheter as unit of analysis: RR 0.53 (95% CI 0.29 to 0.95; 3 studies, 1025 catheters), NNTB: 35 (95% CI 23 to 326) Considering only line access as unit of analysis: RR 1.06 (95% CI 0.84 to 1.33; 2 studies, 6835 lines accessed)
	103 per 1000	77 per 1000 (62 to 86)	RR 0.70 (0.51 to 0.95)	1672 participants, 1025 catheters (10 RCTs)	⊕⊕⊝⊝ Low¹
Duration of catheter patency (days; combining participant and catheter as unit of analysis) Blood withdrawing Follow‐up: 3 to 180 days	Study population		‐	1788 (6 RCTs)	⊕⊕⊝⊝ Low²	No clear evidence of a difference in duration of catheter patency was shown ‐ less than 1 day longer with heparin locking Considering only participant as unit of analysis: MD 0.66 (95% CI ‐0.66 to 1.97; 4 studies, 1036 participants) Considering only catheter as unit of analysis: MD 0.40 (95% CI ‐0.20 to 0.99; 2 studies, 752 catheters)
	Mean catheter patency in the normal saline group was 9 days (8.36 to 9.7 days)	Mean catheter patency in the heparin group was 0.44 days more (‐0.1 less to 0.99 more) than in the normal saline group	‐	1788 (6 RCTs)	⊕⊕⊝⊝ Low²
CVC‐related bloodstream infections (positive microbiological culture, participant as unit of analysis) Follow‐up: 22 to 180 days	Study population		RR 0.66 (0.08 to 5.80)	1127 (3 RCTs)	⊕⊝⊝⊝ Very low³	No clear evidence of a difference in CVC‐related bloodstream infections between locking methods was shown.
	11 per 1000	8 per 1000 (0 to 212)	RR 0.66 (0.08 to 5.80)	1127 (3 RCTs)	⊕⊝⊝⊝ Very low³
Mortality Follow‐up: 17 to 180 days	Study population		RR 0.76 (0.44 to 1.31)	1100 (3 RCTs)	⊕⊝⊝⊝ Very low⁴	No clear evidence of a difference in mortality between locking methods was shown.
Mortality Follow‐up: 17 to 180 days	52 per 1000	40 per 1000 (24 to 57)	RR 0.76 (0.44 to 1.31)	1100 (3 RCTs)	⊕⊝⊝⊝ Very low⁴
Haemorrhage from any site Follow‐up: 1 to 180 days	Study population		RR 1.54 95% CI 0.41 to 5.74	1197 (3 RCTs)	⊕⊝⊝⊝ Very low⁴	No clear evidence of a difference in haemorrhage between locking methods was shown.
Haemorrhage from any site Follow‐up: 1 to 180 days	5 per 1000	5 per 1000 (‐10 to 11)	RR 1.54 95% CI 0.41 to 5.74	1197 (3 RCTs)	⊕⊝⊝⊝ Very low⁴
Heparin‐induced thrombocytopaenia Follow‐up: 7 to 22 days	Study population		RR 0.21 (0.01 to 4.27)	443 (3 RCTs)	⊕⊝⊝⊝ Very low⁴	No clear evidence of a difference in HIT between locking methods was shown. Studies are likely to be underpowered to detect low adverse events.
	9 per 1000	2 per 1000 (0 to 38)	RR 0.21 (0.01 to 4.27)	443 (3 RCTs)	⊕⊝⊝⊝ Very low⁴
**The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; CVC: central venous catheter; HIT: heparin‐induced thrombocytopaenia; MD: mean difference; NNTB: number needed to treat for an additional beneficial outcome; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate. The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited. The true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate. The true effect is likely to be substantially different from the estimate of effect.
¹ We downgraded the certainty of evidence by two levels, i.e. one level for risk of bias due to suspicion of publication bias (see Figure 1) and one level due to imprecision. ² We downgraded the certainty of evidence by two levels, i.e. one level for risk of bias due to unclear allocation concealment and one level due to imprecision. ³ We downgraded the certainty of evidence by three levels, i.e. two levels due to imprecision (low number of events and CIs were very wide) and one level due to inconsistency. ⁴ We downgraded the certainty of evidence by three levels, i.e. one level due to risk of bias and two levels due to imprecision (low number of events and CIs were very wide).

Figure 1

Funnel plot of comparison: 1 Occlusion of CVCs, outcome: 1.1 All studies

Background

Description of the condition

Vascular access devices (VADs) are commonly used in healthcare. They encompass a wide range of devices that include, among others, central venous catheters (CVCs). A CVC is a catheter with a tip that lies within the proximal third of the superior vena cava, the right atrium, or the inferior vena cava. Catheters can be inserted through a peripheral vein or a proximal central vein, most commonly the internal jugular, subclavian, or femoral vein. Four types of CVCs are available: non‐tunnelled catheters, tunnelled catheters (e.g. Hickman catheters, tunnelled dialysis catheters), peripherally inserted central catheters (PICCs), and totally implantable ports (Port‐a‐cath) (Smith 2013).

In the United States, more than five million CVCs are inserted every year (Merrer 2001), leading to approximately 15 million central line days per year in intensive care units (ICUs) (Mermel 2000). CVCs allow measurement of haemodynamic variables that cannot be measured accurately by non‐invasive methods (although some minimally invasive methods are now available), and they allow delivery of blood, medication, and nutritional support that cannot be given safely through peripheral venous catheters. Unfortunately, use of CVCs is associated with adverse events. Among them, mechanical complications during insertion (arterial puncture, haematoma, and pneumothorax) in 5% to 29% (Eisen 2006; McGee 2003), infectious complications in 5% to 26% (Merrer 2001; Raad 1997; Veenstra 1999), and thrombosis in 2% to 26% (Lee 2007) are the most common. Almost all of catheter occlusions are thrombotic. Non‐thrombotic occlusions represent only a small percentage and are caused mainly by medication, lipids deposits, mineral precipitates or mechanical obstructions (Jacobs 2003).

To some extent, thrombi are formed on CVCs during the first few hours of use in the form of fibrin tail, fibrin sheath, intraluminal occlusion, or mural thrombus (Jonker 2010), and thrombosis of large vessels occurs after long‐term catheterisation (Valerio 1981). The incidence of CVC‐related thrombosis varies depending on the patient's condition, catheter tip position and diameter, side and technique of insertion, and the chemical structure and nature of the infusate, among other factors (Verso 2003). CVC‐related thrombosis represents an important source of morbidity and mortality among affected patients, not only for its inherent risks but also because thrombus creates a medium for bacterial proliferation that promotes infection (Mermel 2000). Pulmonary embolism, a severe medical condition, occurs in approximately 15% of patients with CVC‐related upper extremity deep venous thrombosis (Burns 2008).

To avoid thrombus formation in CVCs, clinicians are currently applying several measures with different levels of success. Among others, heparin‐locking catheters (Bishop 2009), heparin‐bonded catheters (Shah 2008), systemic heparinisation with unfractionated heparin or with low molecular weight heparin (Randolph 1998b), anticoagulation with warfarin (Bern 1990), or administration of alteplase or urokinase, as in Hemmelgarn 2011 and Ray 1999, respectively, may be used. Heparin locking is the most commonly used procedure. According to some trial authors, the use of heparin may be justified with some types of VADs when they are not used frequently (Bishop 2009), but the efficacy of this practice remains unproven (López‐Briz 2005).

Description of the intervention

Heparin locking essentially consists of filling the lumens of CVCs with solutions of unfractionated heparin of varying strength. To rinse out the catheter after every use the catheter needs to be flushed. Flushing helps keep the catheter clean. It also prevents blood clots from blocking the catheter.

How the intervention might work

People that have CVCs are at risk of vascular thrombosis via vessel wall injury (during catheter placement), hypercoagulability, and alterations in normal blood flow. The balance between haemostatic systems producing thrombi and fibrinolytic systems dissolving them regulates blood vessel lumen patency, but placement of a CVC can alter this fine‐tuned process, leading to a persistent thrombotic state. Catheter composition also plays a role in this thrombotic situation, allowing adsorption of fibrin and fibrinogen on its surface, thereby worsening the problem (Jacobs 2003). The anticoagulant properties of heparin have led clinicians to use heparin flushes in an attempt to prevent thrombus formation and to prolong the duration of catheter patency between uses. However, this physiopathological rationale may be wrong when applied to peripheral venous catheters, for which no benefit in using heparin locking versus normal saline solution (a crystalloid solution that contains 9.0 g of sodium chloride (NaCl) per litre of water) locking has been demonstrated, as two published systematic reviews have independently shown (Goode 1991; Randolph 1998a).

Why it is important to do this review

Bishop and colleagues reported in 2009 that heparin locking of catheters is a standard practice in the maintenance of CVCs (Bishop 2009), but the effectiveness of this practice so far has not been established in a systematic review. Moreover, variation in nursing practice is considerable because current guidelines provide conflicting recommendations about locking frequency and heparin concentration and volume (Mitchell 2009). A survey conducted in ICUs in the United States shows that 64.6% of respondents used normal saline and 31% used heparin (Sona 2012). The concentrations of heparin most commonly used were 100 IU/mL (37.5%) and 10 IU/mL (29.7%), and the most common intervals for locking catheters were every eight hours and after each use (74.4%). No information is available on CVC maintenance practices in other countries, so could clinical expertise be the guiding principle on this topic?

There are reasons to think that heparin locking catheters might be helpful. This makes pathophysiological sense. One systematic review studied the benefits of heparin in central venous and pulmonary artery catheters (Randolph 1998b). This paper showed that prophylactic systemic heparin decreases catheter‐related venous thrombosis (risk ratio (RR) 0.43, 95% confidence interval (CI) 0.23 to 0.78) and bacterial colonisation of CVCs (RR 0.18, 95% CI 0.06 to 0.60) and may decrease catheter‐related bacteraemia (RR 0.26, 95% CI 0.07 to 1.03). Randolph 1998b included combined data from trials using several doses of systemic prophylactic heparin, including unfractionated heparin (treatment regimens of 1 IU/kg, 3 IU/kg, 50 IU q12h, and 5000 IU intermittently), low molecular weight heparin (2500 IU given subcutaneously daily), or heparin‐bonded catheters and did not include trials that provided periodic flushing of CVCs with heparin.

However, there are also potential harms associated with heparin use. Heparin‐induced thrombocytopaenia (HIT), a severe immunological drug reaction known to cause arterial and venous thromboembolism, raises serious concerns regarding the use of heparin (Warkentin 2007). Exposure of surgical patients to unfractionated heparin for longer than four days implies an overall risk of HIT of 2.6% (Martel 2005). A recent paper highlights an incidence of HIT in the USA of 0.76% in patients treated with therapeutic doses of non‐fractionated heparin and less than 0.1% with prophylactic doses, leading to an amputation rate of 3 to 4% (Gruel 2020).

This adverse effect of heparin treatment is a common late‐onset complication that can develop five or more days after initiation of the drug. Another potential harm that may be associated with heparin use is the incidental administration of a heparin bolus through a catheter line intended for heparin locking.

From an economic point of view, avoiding heparin locking would represent a very important cost savings (Sona 2012). Another systematic review estimated yearly savings of USD109 million to USD218 million when peripheral venous lines were flushed with normal saline instead of heparin (Goode 1991).

In summary, the effectiveness of heparin locking of CVCs has not yet been demonstrated, and wide systematic variations in both guideline recommendations and practice have surrounded its use. Moreover, use of heparin is not free of risk and has a considerable economic impact. We developed a protocol and performed a systematic review about this topic (López‐Briz 2010; López‐Briz 2014). This is the second update of our review first published in 2014.

Objectives

To evaluate the benefits and harms of intermittent locking of central venous catheters with heparin versus normal saline in adults to prevent occlusion.

Methods

Criteria for considering studies for this review

Types of studies

We included only randomised controlled trials (RCTs) of heparin locking versus normal saline locking of central venous catheters (CVCs) in adults. We excluded studies when researchers used alternative methods of randomisation (quasi‐randomised), such as alternate days of the week, odd and even numbers, dates of birth, hospital numbers, or historical controls.

Types of participants

We included studies of adults 18 years of age or older with a CVC. We excluded from this review studies on infants and children, as they are the topic of another Cochrane review (Bradford 2020).

Types of interventions

Interventions included intermittent locking with heparin (any dose with or without systemic drugs ‐ except systemic heparin) compared with normal saline solution. All locking protocols were accepted for inclusion.

Types of outcome measures

Primary outcomes

Occlusion of CVCs (defined as inability to infuse fluids through the catheter because of blockage)
Duration (in days) of catheter patency

Secondary outcomes

Episodes of CVC‐related bloodstream infections; CVC‐related bloodstream infections are defined as the presence of positive blood cultures from both the catheter and peripheral veins and fever or chills in absence of other infection sources (Goosens 2013).
Episodes of CVC‐related colonisation; CVC‐related colonisation is defined as the presence of micro‐organisms in the CVC only and not at another sterile site.
Mortality
Haemorrhage from any site in the body
Heparin‐induced thrombocytopaenia (HIT) (development of thrombocytopaenia after heparin locking of a CVC in an adult with a previously normal platelet count after exclusion of all other causes of thrombocytopaenia, along with a positive antibody test)
CVC‐related thrombosis (determined by colour‐coded Doppler ultrasonography, venography, computerised tomography, or magnetic resonance venography)
Number of additional CVC insertions
Abnormality of coagulation profile
Allergic reactions to heparin

Outcomes were assessed using the description and definitions used by the included studies and reported using the time points reported by the studies, generally at the end of the study period.

Search methods for identification of studies

We applied no restriction on language of publication.

Electronic searches

The Cochrane Vascular Information Specialist conducted systematic searches of the following databases for randomised controlled trials and controlled clinical trials without language, publication year or publication status restrictions:

The Cochrane Vascular Specialised Register via the Cochrane Register of Studies (CRS‐Web);
The Cochrane Central Register of Controlled Trials (CENTRAL; via the Cochrane Register of Studies Online (CRSO);
MEDLINE (Ovid MEDLINE Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations, Ovid MEDLINE Daily and Ovid MEDLINE);
Embase Ovid;
CINAHL EBSCO.

We developed search strategies for other databases from the search strategy designed for MEDLINE. Where appropriate, they were combined with adaptations of the highly sensitive search strategy designed by Cochrane for identifying randomised controlled trials and controlled clinical trials (as described in the Cochrane Handbook for Systematic Reviews of Interventions (Lefebvre 2022)). Search strategies for major databases are provided in Appendix 1.

We searched the following trials registries:

The World Health Organization International Clinical Trials Registry Platform (trialsearch.who.int/);
ClinicalTrials.gov (clinicaltrials.gov).

The most recent searches were carried out on 20 October 2021.

Searching other resources

We searched the reference lists of relevant studies identified through the electronic searches in order to find missing studies.

Data collection and analysis

Selection of studies

Two review authors (ELB and VRG) independently read the abstract and, if necessary, the full text of potentially relevant references, to identify studies that needed to be further examined. We excluded letters, editorials, commentaries, reviews, and lectures that did not contain original research data. We contacted authors of unpublished and ongoing trials to obtain further information. When differences in opinion arose, we consulted a third review author (RCS).

Data extraction and management

Three review authors (ELB, VRG, and RCS) independently extracted data regarding populations, interventions, relevant outcomes, funding source and declarations of interest from the study authors, using the standard Cochrane Vascular forms for dichotomous data and continuous data. We contacted study authors to obtain additional data, if necessary (Goosens 2013; Schallom 2012).

Assessment of risk of bias in included studies

We assessed the risk of bias in included studies by using standardised criteria from Cochrane for the following (Higgins 2011).

Adequacy of random sequence generation.
Allocation concealment.
Blinding of participants and personnel.
Blinding of outcome assessment.
Incomplete outcome data.
Selective reporting.
Other bias.

Measures of treatment effect

We used risk ratio (RR) with 95% confidence interval (CI) and number needed to treat for an additional beneficial outcome (NNTB) to measure any effect on dichotomous variables (i.e. occlusion of CVCs, mortality, adverse events, etc.). We calculated NNTB values from the RR according to the formula NNTB (or number needed to treat for an additional harmful outcome (NNTH)) = 1/ACR*(1‐RR), for which ACR is the assumed control risk (McQuay 1997).

Unit of analysis issues

The unit of analysis differed between studies and was either the participant or catheter or line access (i.e. each time a line of a CVC is used to administer drugs, blood, etc.). We performed analysis separately for each different unit of analysis for outcomes that could have been influenced by the unit of analysis (occlusions and patency), if sufficient data were available. The main analyses stratified studies by the unit of analysis type, but we also reported the main results irrespective of the unit of analysis. For secondary outcomes, when considering adverse effects, we used the participant as the denominator for analysis.

Dealing with missing data

We contacted the principal investigators of two studies to request additional data (Goosens 2013; Schallom 2012). These study authors provided relevant data that were later published.

Assessment of heterogeneity

We attempted to explain relevant clinical, methodological, or statistical heterogeneity using forest plots, and we quantified heterogeneity using the I² statistic (Higgins 2021). Thresholds for interpretation of I² can be misleading in that the importance of inconsistency depends on several factors. Higgins 2021 prepared the following rough guide to interpretation.

0% to 40%: might not be important.
30% to 60%: may represent moderate heterogeneity.
50% to 90%: may represent substantial heterogeneity.
75% to 100%: shows considerable heterogeneity.

We considered I² values > 50% to indicate significant heterogeneity.

Assessment of reporting biases

We assessed reporting bias using funnel plots, since we found a sufficient numbers of studies.

Data synthesis

We summarised data statistically, if possible. We performed statistical analysis according to the statistical guidelines referenced in the current version of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2021). We used Review Manager for review production and data analysis (Review Manager 2020). We used a random‐effects model, even though I² values were low because, although the same drug was used across trials (heparin), we noted clear clinical heterogeneity in the study methods applied (i.e. different doses with systemic heparin or not, different follow‐up times, different kinds of patients, etc.).

Subgroup analysis and investigation of heterogeneity

We performed subgroup analyses for each different unit of analysis for the primary outcomes (participant, catheter or line access). The incidence of CVC‐related thrombosis varies depending on clinical background of the participants (suffering malignancies or other onco‐haematological diseases, admitted to intensive care units, on dialysis, etc.), CVC implantation site, CVC type, and perfusion‐related factors. We planned to perform subgroup analyses to take these factors into account, if sufficient data were available.

Sensitivity analysis

We carried out sensitivity analyses to explore the robustness of results by investigating the influence of the following factors on effect size for occlusions.

Published or unpublished studies.
Methodological quality of studies. We explored quality of studies according to the risk of bias of the allocation concealment.
Weight of different studies. We categorised most weighted studies as those with more than 30% of total weight.
Different measures of effect size (odds ratio (OR) and RR).

Summary of findings and assessment of the certainty of the evidence

We created summary of findings Table 1 to present the results for the comparison of heparin versus normal saline intermittent locking for prevention of occlusion in central venous catheters in adults. We used GRADEpro GDT software to present the main findings of the review (gradepro.org) (GRADEproGDT 2015), and assessed the certainty of the evidence as high, moderate, low, or very low, based on within‐study risk of bias, directness of evidence, heterogeneity, precision of effect estimates, and risk of publication bias (Guyatt 2008). We judged the outcomes of CVC occlusion, duration of catheter patency, CVC‐related bloodstream infections, mortality, haemorrhage, and heparin‐induced thrombocytopaenia to be the most clinically relevant to healthcare professionals and patients. For each outcome, we calculated assumed control intervention risks from the mean number of events reported in the control groups of selected studies.

Results

Description of studies

Results of the search

See Figure 2.

Figure 2

Study flow diagram 2021

Included studies

One new study met the inclusion criteria for this update (Klein 2018), bringing the number of included studies to 12, involving a total of 2422 patients (Babu 2014; Beigi 2014; Bowers 2008; Dal Molin 2015; Goosens 2013; Heidari 2015; Kaneko 2004; Klein 2018; Lyons 2014; Pumarola 2007; Rabe 2002; Schallom 2012). See Characteristics of included studies.

Babu 2014 performed an RCT in 100 participants from the Respiratory Intensive Care Unit with triple lumen CVC. This study compared heparin (3 mL, 10 IU/ mL) versus normal saline (10 mL) flushes every eight hours. The primary outcome of the study was lumen non‐patency, defined as inability to both withdraw blood and flush through a lumen, and the unit of analysis was the participant. Researchers reached the conclusion of lumen non‐patency after the following interventions: (1) if the lumen could not be flushed, the participant was repositioned and the flush re‐attempted; and (2) if the lumen still could not be flushed, the syringe was changed and the flush was re‐attempted. Investigators assessed the secondary outcome, HIT, using daily platelet count starting on day 4 from the time of giving heparin flushes to all participants in the heparin group.

Beigi 2014 was a single‐blinded randomised controlled trial with 100 participants with chronic kidney disease. Researchers randomly assigned participants to locking with heparin (1000 IU) versus normal saline. The unit of analysis was the participant. Only three in the heparin group and one in the normal saline group withdrew. We sent a letter to study authors to request more information, but they did not respond to our request. Length of follow‐up was 24 hours.

Bowers 2008 conducted a single‐centre randomised study in 102 adult participants with single‐lumen peripherally inserted central catheters (PICCs) with luer‐activated devices. Trial authors used a random block design with allocation concealment to randomly assign participants to receive normal saline or heparin lock flush (100 USP U/mL). The main outcome studied was occlusion rate, and the secondary outcome was duration of PICCs (in days). The unit of analysis was the participant for occlusion rate as well as for patency. All participants completed the study (50 in the normal saline group and 52 in the heparin group).

Dal Molin 2015 was a multicentre, open‐label randomised trial with 430 oncology participants. Investigators randomly assigned participants to locking with heparin 5 mL (50 IU) versus normal saline 5 mL. Trial authors used the participant as the unit of analysis for occlusion. Study authors reported 5% withdrawals from the normal saline group and 2.5% from the heparin group without providing details.

Goosens 2013 conducted a randomised controlled open‐label non‐inferiority trial in 802 participants older than one year scheduled for first insertion of a totally implantable venous access device (TIVAD) through the superior vena cava (SVC) system, with an onco‐haematological malignancy and with sufficient life expectancy to complete the planned follow‐up of 180 days at the study centre. After randomisation via computerised random number generation, researchers assigned 398 participants to receive a normal saline lock and 404 to receive a heparin lock in a non‐blinded manner. Although participants were randomly assigned, the main unit of analysis was the number of catheters accessed. However, the study authors provided us with additional information about occlusions per participant. Participants who had difficulties with aspiration through the catheter were registered. Investigators considered outcomes of withdrawal occlusion, catheter‐related bloodstream infections, and catheter duration within 180 days (unit of analysis ‐ participant), as well as adverse events. The study authors also provided data on thrombosis, and mortality. As this study included adults and children, we also requested data for the adult participants only. The study author responded as follows: "Only 3.5% of patients were <18 years old; given that small number we didn't perform any sub analysis. Moreover we don't presume any difference in results between adults and peds" [sic].

Heidari 2015 conducted a double‐blinded RCT in 84 participants from the intensive care unit. This study compared a flush of 3 mL of heparin (10 IU/mL) versus normal saline locking. The main outcome was CVC patency, and the unit of analysis chosen was the participant. We requested additional information from study authors, but they did not respond to our request. Follow‐up period was 21 days.

Kaneko 2004 performed a single‐centre, open‐label, randomised controlled clinical trial in adult participants with an inserted double‐lumen CVC. This study compared a flush of 20 mL of normal saline versus a flush of 20 mL of normal saline followed by locking with 2 mL heparin (1000 IU/mL). Researchers used low molecular weight heparin at 8 IU/kg/h during each haemodialysis session. They randomly allocated 48 participants to the normal saline (26) or heparin group (22). They studied the outcomes: days of catheter survival and thrombotic occlusion (both considered the participant as the unit of analysis), as well as coagulation analytical parameters such as activated coagulation time, activated partial thromboplastin time, and prothrombin time.

Klein 2018 performed a RCT in 30 patients from the blood and bone transplantation unit. The study was not blinded. Fifteen were flushed with normal saline and 15 with different concentrations of heparin according to the lines (triple or double lumen). In addition, every line was flushed depending on the type of line. Outcomes of interest were patency and safety. The unit of analysis was line access.

Lyons 2014 performed a single RCT on 90 participants from home care and tried to find the most effective locking solution for maintenance of PICCs. This study compared three arms: 10 mL of normal saline, 5 mL of low‐dose heparin (10 IU/mL), and 3 mL of high‐dose heparin (100 IU/mL). The main outcome was the development of patency‐related complications (sluggishness, occlusions, etc.), and researchers used the participant as the unit of analysis. One participant in the normal saline group and one in the high‐dose heparin group withdrew. We sent a letter to study authors to request more information and they kindly provided us with the protocol of study.

Pumarola 2007 carried out a two‐phase clinical trial in a polyvalent ICU. Participants were adults with multiple pathological processes in whom a three‐lumen CVC had been inserted. Researchers used a registered software program for randomisation. However, the study was not blinded. In the first phase, trialists compared two concentrations of heparin (20 IU/mL and 100 IU/mL), establishing patency at 24 hours after catheter implantation and at discharge. In the second phase, 125 participants were randomised to each group and heparin at a concentration of 100 IU/mL was compared to normal saline. Patency was assessed at 24 hours, at 72 hours, and at discharge. Only this second phase fulfilled our inclusion criteria. Although study authors randomised 125 to each group, 95 CVCs were assessed (38 in the heparin group and 57 in the normal saline group) for occlusion rates and mean days of catheter duration, using the catheter as the unit of analysis for both.

Rabe 2002 studied 99 three‐lumen CVCs inserted into 91 adult participants locked with one of the following solutions: normal saline, heparin (5000 IU/mL), or vitamin C (200 mg/mL). Researchers assigned catheters randomly (using a list of random numbers prepared by the study authors) to one of three groups. They assessed patency every two days to a maximum of 20 days. Study outcomes included thrombotic obstruction and catheter survival, with the catheter used as the unit of analysis.

Schallom 2012 conducted a single‐centre study wherein researchers randomly assigned patients in the ICU with a newly placed three‐ or four‐lumen CVC (simple randomisation, sequence concealed) to be flushed with normal saline or with heparin (10 IU/mL every 8 hours). Among the randomly assigned participants, 295 had at least one lumen with a minimum of two flushes, resulting in 326 catheters (170 allocated to the normal saline group and 156 to the heparin group) with 709 lumens (395 in the normal saline group and 314 in the heparin group). The primary outcome was lack of lumen patency (unit of analysis was the catheter). Secondary outcomes included rates of loss of blood return, flush failure, HIT, and catheter‐related bloodstream infection.

Excluded studies

We excluded 12 additional studies for this update (IRCT20151228025732N56; IRCT20190325043107N4; IRCT20191218045773N2; Kaewsangsai 2021; Liu 2018; NCT02923830; Roberts 2020; Saini 2018; Silva 2021; TCTR20200630005; Wathanavasin 2021; Wouters 2020).

The total number of excluded studies in the current review is 189. We excluded these studies for the following reasons:

Studies did not meet the criteria established for intervention (heparin) or comparison (normal saline).
Studies focussed on peripheral catheters.
Studies focussed on arterial catheters.
Studies did not provide data stratified by arterial or venous catheters.
Studies were in fact protocols of data unpublished/published.

We excluded some studies for more than one reason.

See the Characteristics of excluded studies section for further details.

Ongoing studies

We identified six new studies as ongoing (ChiCTR1800018391; CTRI/2021/04/033007; IRCT20190905044704N1; JPRN‐UMIN000033713; NCT02354118; NCT05029596). See Characteristics of ongoing studies for further details.

Risk of bias in included studies

Figure 3 and Figure 4 show the risk of bias according to the methodological quality of included trials.

Figure 3

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies

Figure 4

Risk of bias summary: review authors' judgements about each risk of bias item for each included study

Allocation

Eight studies provided sufficient information on random sequence generation, so we assessed the risk of bias for these studies as low (Beigi 2014; Bowers 2008; Dal Molin 2015; Goosens 2013; Heidari 2015; Pumarola 2007; Rabe 2002; Schallom 2012). In four studies, the information provided about the sequence generation process was either insufficient or undisclosed, so we judged them to be at unclear risk of bias (Babu 2014; Kaneko 2004; Klein 2018; Lyons 2014).

Eight studies provided insufficient information about allocation concealment, so we assessed the risk of selection bias for these studies as unclear (Babu 2014; Beigi 2014; Bowers 2008; Heidari 2015; Kaneko 2004; Klein 2018; Pumarola 2007; Rabe 2002). Pumarola 2007 reported a method of closed envelopes, but it remains unclear whether the envelopes were opaque or sealed to conceal information. Goosens 2013 concealed the allocation sequence from researchers who enrolled participants by using sequentially numbered participant cards stored in a separate room; Schallom 2012 stated that the allocation sequence was concealed from the researcher enrolling participants; Dal Molin 2015 used a web‐based method to conceal allocation; and Lyons 2014 used a sequentially numbered, opaque sealed envelope method, so we assessed these studies as having low risk of selection bias.

Blinding

Nine studies were open‐label or did not blind participants or research staff to the intervention received. We rated these studies as having a unclear risk of performance and detection bias (Babu 2014; Bowers 2008; Dal Molin 2015; Goosens 2013; Kaneko 2004; Klein 2018; Pumarola 2007; Rabe 2002; Schallom 2012).

Beigi 2014 and Lyons 2014 used single‐blinding design, and we classified the risk of performance and detection bias as unclear for both.

Heidari 2015 was at low risk of bias as both participants and researchers were unaware of which locking fluid was used (the solution was made up by nurses). However, neither occlusion nor patency was likely to be influenced by lack of blinding. We judged that the secondary outcomes, namely, CVC‐related thrombosis, episodes of CVC‐related bloodstream infections and colonisation, numbers of additional CVC insertions, mortality, coagulation profile, HIT, and allergic reactions to heparin and haemorrhage, were also unlikely to be influenced by lack of blinding.

Incomplete outcome data

We considered Beigi 2014 (two in heparin groups and one in saline group withdrew), Bowers 2008 (no withdrawals), Dal Molin 2015 (five participants in heparin group and 10 in saline group withdrew), Heidari 2015 (no withdrawals), Lyons 2014 (no withdrawals), Babu 2014 (no withdrawals), and Schallom 2012 (no withdrawals), to have a low risk of attrition bias because missing outcome data were either none or few and were balanced in numbers across intervention groups, and reasons for missing data were similar across groups.

Researchers from Rabe 2002 and Goosens 2013 reported attrition or exclusions insufficiently to permit judgement, and information about the number of catheters losing patency in each treatment group was lacking in Rabe 2002. In Klein 2018, one patient was excluded from the normal saline group and two from the heparin group, because of incomplete flushing records; furthermore, in Klein 2018, the presence of occlusion was evaluated several times per day resulting in a huge number of observations; hence, the impact of losing a participant could not be properly estimated. We judged these three studies as having an unclear risk of attrition bias.

We rated both Kaneko 2004 and Pumarola 2007 as having a high risk of bias. Kaneko 2004 reported 40% withdrawals in the heparin group (9/22) and 30% in the normal saline group (8/26) and provided unclear reasons for withdrawal. Pumarola 2007 reported a withdrawal rate of 69.6% (87/125) in the heparin group and 54.4% (68/125) in the normal saline group; the main reason for withdrawal was cancellation of the procedure (74/125 and 52/125, respectively).

Selective reporting

Dal Molin 2015, Goosens 2013, and Lyons 2014 reported all expected outcomes, so we rated these studies as having a low risk of selective reporting bias. The remaining studies were at unclear risk owing to lack of available protocols or insufficient information retrieved from researchers (Beigi 2014; Bowers 2008; Heidari 2015; Kaneko 2004; Klein 2018; Babu 2014; Pumarola 2007; Rabe 2002; Schallom 2012).

Other potential sources of bias

Bowers 2008, Klein 2018 and Lyons 2014 were rated at low risk of bias. Pumarola 2007 might be underpowered as researchers analysed only 38 and 57 catheters per group, but the predetermined sample size was 185 catheters per group; trialists stopped the study early for 74 and 52 catheters in the heparin and normal saline groups, respectively, but did not provide the reason for this. Therefore, we rated the risk of other bias as high. In Goosens 2013, 3.5% of participants were children and study authors did not perform separate analyses; therefore we rated the risk of other bias as unclear. In the remaining studies, the risk of other bias was also rated as unclear because there was not enough information to permit a low‐risk judgement of bias.