Reducing Peripheral Intravenous Catheter Use in Adult Acute Care: A Systematic Review

ABSTRACT

Peripheral intravenous catheters (PIVCs) are the most frequently inserted invasive device in acute care, yet up to 52% remain unused, exposing patients to avoidable risks and increasing healthcare costs. Despite growing awareness of inappropriate PIVC use, unnecessary insertion persists, driven by cultural and organisational norms and risk‐averse practices. Interventions to reduce PIVC use have been implemented, but their effectiveness and sustainability have not been systematically synthesised. A systematic review was conducted in accordance with PRISMA guidelines and registered in PROSPERO (CRD420251033201). Searches of CINAHL, MEDLINE, Scopus, and Web of Science identified studies reporting interventions to reduce PIVC insertion, unused PIVCs, and common complications. Data were extracted for intervention characteristics and PIVC outcomes. Implementation strategies were mapped to Consolidated Framework for Implementation Research domains for summative analysis. Random‐effects meta‐analysis was performed when appropriate. Twelve studies were included from emergency departments and inpatient settings across high‐income countries. Most reported reductions in PIVC insertion and unused catheters, with improvements in secondary outcomes including phlebitis and bloodstream infections. Sustainability of practice change was demonstrated in two multi‐year follow‐up studies. Interventions showed a significant trend for increasing complexity over time, with recent strategies incorporating multi‐modal, theory‐informed approaches. Meta‐analysis showed substantial heterogeneity, limiting pooled estimate interpretation, while methodological limitations restricted causal inference. Multi‐modal interventions addressing behavioural, organisational, and contextual determinants appear most effective for reducing PIVC insertion rates and unnecessary PIVC use. Future research should evaluate sustainability, cost‐effectiveness, and scalability using implementation science frameworks.

1. Introduction

Peripheral intravenous catheters (PIVCs) are the most frequently inserted vascular access devices used in hospitalised patients worldwide [1], with up to half of all inpatients undergoing PIVC placement for anticipated parenteral treatment [2, 3]. Between 34% and 60% of adults attending emergency departments (EDs) will have a PIVC inserted, yet 34%–52% of the devices placed in the ED remain unused; often inserted as a precautionary measure rather than when clinically indicated [4, 5, 6, 7].

Although PIVC insertion is a routine clinical procedure, it can be associated with significant complications including pain, phlebitis, dislodgement, and catheter‐related bloodstream infections [8]. Among the most serious complications is hospital‐acquired Staphylococcus aureus bacteraemia (HA‐SAB), a bloodstream infection with significantly increased morbidity and mortality [9]. Reports estimate between 14% and 42% of PIVCs remain unused for any therapeutic purpose during patients' hospital stays [1, 4, 5]. These ‘just‐in‐case’ catheter insertions are unnecessary invasive procedures that expose patients to avoidable risks, consume limited resources and contribute to healthcare waste and carbon emissions [10]. A ‘snowball effect’, may also occur whereby PIVC insertion can trigger additional investigations, initiation of intravenous therapies despite suitable oral alternatives, and potentially unnecessary hospital admissions [11].

National and international guidelines emphasise assessing clinical need before PIVC insertion [12, 13, 14, 15], yet the gap between evidence‐based recommendations and clinical practice persists. PIVCs continue to be inserted when not clinically indicated, with decisions to insert PIVCs described as a learned reflex with little cognitive deliberation [16, 17]. This pattern is likely sustained by organisational norms, clinician habits, and risk‐averse cultures that favour insertion over restraint. Persistent overuse of PIVCs constitutes low‐value care; conferring little or no clinical benefit while exposing patients to avoidable risk [18] and adversely affecting patient experience through procedural pain, anxiety, and reduced satisfaction [19].

Various interventions have been proposed to reduce unnecessary insertions, including clinician education, policy changes, and clinical decision‐support tools. However, their effectiveness and factors shaping implementation and sustainability have not been systematically synthesised. Few interventions have been evaluated using implementation science frameworks, limiting understanding of contextual and behavioural factors that drive PIVC use. This systematic review aims to identify and synthesise studies evaluating interventions to reduce PIVC insertion rates in adult acute care, assess their effectiveness, and identify factors influencing implementation and sustainability.

2. Methods

2.1. Study Design and Registration

This systematic review was conducted in accordance with PRISMA guidelines [20] and prospectively registered with PROSPERO (CRD420251033201).

2.2. Eligibility Criteria

We included randomised and non‐randomised studies evaluating interventions in adult acute care settings aimed at reducing PIVC insertions. Studies exclusively involving paediatric patients, pre‐hospital care, case reports, and conference abstracts were excluded. As randomised designs are rarely feasible for service‐level interventions, the review focused on high‐quality observational studies.

2.3. Participants and Settings

The population of interest was adult patients (≥ 18 years) treated in acute hospital settings, including EDs and inpatient settings where PIVC placement is routinely considered. Patients and the public were not involved in the design, conduct, or reporting of this review, as it is a synthesis of previously published clinical and health‐services data. No primary data involving patients were collected.

2.4. Interventions and Comparators

Studies reporting interventions designed to reduce the frequency of PIVC insertion or the proportion of PIVCs assessed to be unnecessary or unused were eligible. Secondary outcomes included insertion attempts, insertion site location, PIVC‐associated infections, complications at sites of insertion, and adverse events associated with not having a PIVC. Comparators were usual care or baseline practice.

2.5. Information Sources and Search Strategy

Searches were conducted in CINAHL, MEDLINE, Scopus, and Web of Science. Only English‐language publications were included, with no date restrictions. All studies published by January 2026 were eligible. Additional studies were identified through reference list screening. The full search strategy is publicly available via PROSPERO (Table S1).

2.6. Study Selection and Data Extraction

Search results were imported into Covidence (Veritas Health Innovation, Melbourne, Australia), [21] and duplicate records removed. Titles, abstracts and full text articles were screened independently by two reviewers, with disagreements resolved by a third person. Data were extracted using a standardised extraction template capturing study characteristics, population, setting, intervention details, and outcomes. Discrepancies were resolved by discussion or third‐party adjudication.

2.7. Quality Assessment and Risk of Bias

Reporting quality was assessed for all included studies using the STROBE statement for observational studies checklist [22] (Table S2). As a reporting guideline rather than a methodological quality tool, scores reflect completeness of reporting and should be interpreted accordingly. Risk of bias was assessed independently by two reviewers using three tools according to study design: ROBINS‐I V2 for non‐randomised pre‐post intervention studies, [23] RoB 2 CRT [24] for cluster randomised trials and EPOC criteria for interrupted time‐series analyses (ITS) [25]. All six pre–post intervention studies were rated ‘Critical’ on the ROBINS‐I V2 Domain 1 for confounding, reflecting a structural inability of uncontrolled before‐and‐after designs to account for temporal confounding and concurrent co‐interventions, rather than failures in study conduct (Table S3).

2.8. Data Synthesis

Unadjusted relative risks (RR) with 95% confidence intervals were calculated from raw counts extracted from included studies. Data analysis was performed using Stata 18.1 (StataCorp LLC, Texas, USA). Exposure and outcome definitions were standardised across studies to ensure comparability, and when definitions varied, efforts were made to harmonise them conceptually. Random effects meta‐analysis was attempted to pool estimates across studies. A summative content analysis of intervention components was undertaken and mapped to the Consolidated Framework for Implementation Research (CFIR) [26] domains by one reviewer and validated by a second (Table S4). CFIR was selected for its comprehensive taxonomy of implementation determinants across individual, intervention, and organisational levels [26]. An annotated subset of CFIR domains was applied. The Outer Setting, encompassing policy environments, incentives and external networks, was omitted from the analysis [27]. This domain is influential in the choice and motivation for undertaking a behavioural change intervention but was not directly addressed or reported in the included studies. An unadjusted linear regression was conducted to examine the association between the number of CFIR domains identified through intervention mapping and publication year.

The authors used an AI tool for verification of statistical calculations, risk of bias assessments, and proofreading (Claude‐Sonnet‐4.6, Anthropic, 2026).

3. Results

A total of 2019 records were identified through database and citation searches. After removal of 647 duplicates, 1372 publications were screened for relevance. Of these, 1316 were ruled ineligible and excluded based on title and abstract review. Fifty‐six full text articles were assessed for eligibility, with 44 excluded and 12 articles included for review (Figure 1).

FIGURE 1.

PRISMA flow diagram of study selection.

Studies were from diverse geographic regions, including Australia (n = 7), Europe (France, Spain, Netherlands; n = 4), and the United States of America (n = 1), conducted in EDs and inpatient settings. Sample sizes ranged from 267 to 4173 participants and the studies were published between 1994 and 2024. Research designs included prospective cohort studies (n = 6) [5, 28, 29, 30, 31, 32], retrospective cohort studies (n = 2) [6, 33], controlled interrupted time series (n = 3) [34, 35, 36] and one mixed methods follow‐up study [37] One prospective cohort study was a cluster randomised trial [28] Risk of bias assessments are summarised in Table S3, and are referenced when they materially affect result interpretation. Study and participant characteristics are presented in Table 1.

TABLE 1.

Study characteristics.

Study author, year	Country	Design	Setting	Sample size	Population	Inclusion criteria
Champion et al., 2013	France	Multicentre, prospective 3‐step interventional study	56 public hospitals	N = 3653, n = 1910 Baseline, n = 832 intervention phase control, n = 911 intervention	Internal medicine patients	All hospitalised patients were screened during each study visit
Decker et al., 2016	Australia	Prospective cohort study	Level 4 adult emergency department	N = 300, n = 150 pre‐intervention, n = 150 post‐intervention	Emergency department	Persons aged ≥ 18 years with an Australasian Triage Scale (ATS) Categories 2, 3, 4 or 5 were eligible
Egerton‐Warburton et al., 2019	Australia	Retrospective cohort study	Adult tertiary emergency department	N = 760 patients, n = 380 pre‐MMI, n = 380 post‐MMI	Emergency department	Adult patients aged ≥ 18 years with a record of PIVC insertion in the ED EMR
Hawkins et al., 2018	Australia	Prospective cohort study	Adult tertiary emergency department	N = 4173, n = 2063 pre intervention, n = 2110 post intervention	Emergency department	Persons aged ≥ 18 years presenting to the ED
Laan et al., 2020	The Netherlands	Controlled interrupted time series	3 University and 4 general hospitals	N = 3843, n = 1781 baseline, n = 2062 intervention	Hospital inpatients—internal medicine and non‐surgical subspeciality	Adult patients admitted internal medicine/medical sub‐specialty with PIVC
Lim et al., 2020	Australia	Retrospective cohort study	Adult tertiary emergency department	N = 1127 n = 376 pre‐intervention, n = 378 post‐intervention, n = 373 long term follow‐up	Emergency department	Patients aged ≥ 18 years with a recorded PIVC insertion in ED
Parenti et al., 1994	United States of America	Prospective cohort study	University affiliated veterans' hospital	N = 746, n = 378 pre‐intervention, n = 368 post	Medical ward patients	Presence of PIVC
Perez‐Granda et al., 2015	Spain	Prospective cohort study	General hospital	N = 1435, n = 753 pre‐intervention, n = 682 post‐intervention	Hospital inpatients (maternity and psychiatry excluded)	Presence of PIVC
Ray‐Barruel et al., 2023	Australia	Controlled interrupted time series	3 Metropolitan hospitals	N = 867, n = 410 pre‐intervention, n = 457 post‐intervention	Seven adult inpatient wards, three medical, three surgical, and one infectious disease ward	Persons aged ≥ 18 years with PIVC
Rhodes et al., 2016	Australia	Controlled interrupted time series	Tertiary hospital	N = 552, n = 273 pre‐intervention, n = 279 post‐intervention	Whole of hospital patient population	Presence of PIVC
Ruegg et al., 2022	Australia	Pre‐post intervention study	Tertiary teaching hospital	N = 646 during intervention period	Emergency department, two medical and two surgical wards	Presence of PIVC
van Horrik et al., 2024	The Netherlands	Mixed methods (follow‐up study)	5 hospitals	N = 1113	Hospital inpatients: Internal medicine and non‐surgical subspeciality	Adult patients admitted internal medicine/medical sub‐specialty with PIVC

Abbreviations: ATS, Australasian Triage Scale; ED, emergency department; EMR, electronic medical record; PIVC, peripheral intravenous catheter.

Across the 12 included studies, interventions demonstrated consistent reductions in PIVC insertion rates and the proportion of unused or inappropriate PIVCs, although the magnitude and statistical significance varied. This consistency of direction across studies of varying design and methodological quality strengthens confidence in the overall finding.

Eight studies reported PIVC insertion rates, with six studies finding a statistically significant reduction post intervention [5, 6, 28, 29, 33, 35]. Among the before‐and‐after studies, Hawkins et al. [5] and Egerton‐Warburton et al. [33] were the most methodologically rigorous within their design tier. Egerton‐Warburton et al. [33] observed a reduction in PIVC use from 33.9% to 21.0% (RR = 0.71, 95% confidence interval [CI] 0.67–0.75), while Lim et al. [6] reported sustained reductions at the same sites at 5 years' follow‐up (RR = 0.60, 95% CI 0.56–0.65).

Nine studies assessed appropriate use of or rates of unused PIVCs, with most showing a reduction in idle PIVCs. Definitions varied but generally included catheters not used for medication, fluids, or blood products. The strongest causal evidence came from Champion et al. [28], the only cluster‐randomised trial, which demonstrated a 39% relative reduction in inappropriate intravenous line use (p = 0.05). Laan et al. [34] demonstrated a significant reduction in inappropriate PIVC use across seven hospitals (incidence rate ratio (IRR) = 0.65, 95% CI 0.56–0.77, p = 0.0001), which was sustained in the follow‐up study by van Horrik et al. [37] (odds ratio (OR) = 0.76, 95% CI 0.68–0.84, p < 0.001) (Table 2).

TABLE 2.

Summary of intervention outcomes across included studies.

Study (author, year)		Primary outcome change in PIVC use		Test for significance	Primary outcome appropriate use “unused cannulas”		Test for significance	Criteria for determining an inappropriate cannula	Secondary outcome(s)		Test for significance	Study findings/outcomes
	Metric	Pre	Post		Pre	Post			Pre	Post
Champion et al., 2013 f	Mean (SD)	24.9 (10.8)	20.1 (10.5)	—	16.8 (17.4)	10.2 (32.7)	p = 0.05 e	“Any intravenous line that was no longer being used”	—	—	—	The 38% reduction in the number of appropriate intravenous lines was mainly due to junior doctors who were the most sensitive to this teaching program.
Decker et al., 2016	n (%)	74 (49.3%)	50 (33.3%)	p = 0.05 a	50 (67.6%) 24 (32.4%) f	28 (60.0%) 22 (44.0%) f	p = 0.69 a (p = 0.191) p = 0.191 a , f	PIVCs Used PIVC unused f	PIVC in ACF 54 (36%)	PIVC in ACF 26 (17.3%)	p = < 0.01 a	Simple interventions were successful in significantly reducing the number of PIVCs
Egerton‐Warburton et al., 2019	n (%)	1413 (33.9%)	928 (21.0%)	−12.9% RR b 0.71 CI 0.67–0.75	139 (37.0%)	73 (19.3%)	−17.7% RR b 0.52 CI 0.41–0.67	Unused PIVC—A cannula was considered ‘used’ if medications, fluids, blood products or contrast were administered intravenously to the patient	Appropriately unused PIVC 42 (30.2%) CI 22.9–38	Appropriately unused PIVC 20 (27.4%) CI 18.0–39.3	RR b  = 0.91 CI 0.58–1.42	MMI was effective in reducing the rate of unused PIVC insertions. MMI reduced the proportion of adults having a PIVC inserted. The appropriateness of unused PIVC insertions did not change.
Hawkins et al., 2018	n (%)	869 (42.1%)	682 (32.4%)	−9.8% CI −12.7% to −6.8%	585 (67.4%) 283 (32.6%) f	541 (79.4%) 140 (20.6%) f	12.0% CI 8.7–17.0 −12.0% CI −8.7 to −17.0 f	PIVC Used in ED Unused PIVC in ED f	PIVC Used within 24 h 612 (70.4%)	PIVC Used within 24 h 568 (83.4%)	12.86% CI 8.7%–17.0%	The intervention reduced PIVC placement in the ED and increased the percentage of PIVCs placed that were used
Laan et al., 2020	n (%)	—	—	—	366 (22%)	275 (14%)	IRR 0.65 CI 0.56–0.77 p = 0.0001	Inappropriate use	PIVC associates BSI 3 (0.2%)	PIVC associates BSI 2 (0.1%)	IRR 0.59 CI 0.10–3.50 p = 0.670	De‐implementation strategy reduced inappropriate use PIVCs in patients who were not in the intensive care unit.
—		—	—	—	34 (42%)	23 (26%)	IRR 0.61 CI 0.36–1.05 p = 0.027	Inappropriate use of 2nd PIVC	—	—	—
Lim et al., 2020	n (%)	1413 (33.9%)—Initial study	925 (20.2%)—long term follow‐up	−13.7% RR b 0.60 CI0.56–0.65	139 (37.0%)	101 (27.1%)	−9.9% RR b 0.73 CI 0.59–0.90	PIVC was considered ‘unused’ if it was unused after insertion or used solely for pathology collection.	Appropriately unused—pre intervention 42 (30.2%)	Appropriately unused—long term follow‐up 23 (22.8%)	—	MMI aimed at reducing unused PIVC insertions in ED has been effective in eliciting sustained change.
Parenti et al., 1994	n (%)	234 (61%)	251 (68%)	NS	107 (42%)	77 (29%)	p < 0.01 a	An idle IV catheter is defined as an IV catheter in place 2 or more consecutive days without therapeutic use	—	—	—	This quality improvement effort successfully reduced unnecessary PIVC use
Perez‐Granda et al., 2015	n (%)	653 (81.9)	556 (82%)	NS	183 (22.9%)	48 (7.1%)	p < 0.001 a	The VL to be unnecessary on the study day when the patient was hemodynamically stable, had no indication for IV fluids via that line, and did not require IV medication.	Clinical evidence of local infection 18 (2.2%)	Clinical evidence of local infection 12 (1.8%)	p = 0.52	A multidisciplinary teaching program to improve VL care and compliance with recommendations is effective.
Ray‐Barruel et al., 2023	n (%)	1052 Utilisation ratio (0.42)	1216 Utilisation ratio (0.49)	Difference 7.5% RR b 1.18 CI 1.11–1.25 p < 0.001	52 (12.7%)	38 (8.3%)	Difference −4.4% RR b 0.66 CI 0.44–0.97 p = 0.035	PIVC Idle—PIVC in situ without a clear purpose	Primary bloodstream infection 0	Primary bloodstream infection 1	n/c	Implementation of a comprehensive device assessment and decision tool (I‐DECIDED) reduced idle catheters and catheter complications, despite higher device utilisation
—	—	—	—	—	—	—	—	—	Complications 66 (16.1%)	Complications 50 (10.9%)	Difference −5.2% RR b 0.68 CI 0.48–0.96 p = 0.026
Rhodes et al., 2016	n (%)	—	—	—	273 Utilisation ratio (0.37)	279 Utilisation ratio (0.42)	p = 0.16 a	PIVC Used	HA‐SAB 24 (35% PIVC related), 0.39 per 10,000 OBD	HA‐SAB 12 (14.4% PIVC associated), 0.14 per 10,000 OBD	Difference −63% RR b 0.36 CI 0.17–0.76 p = 0.018	HA‐SAB rates were 63% lower in the post‐intervention period compared to baseline with a change point observed following full bundle implementation
—	—	—	—	—	—	—	—	—	Phlebitis score = 0, 211 (77.6%)	Phlebitis score = 0, 257 (92.1%)	p < 0.05 a
—	—	—	—	—	—	—	—	—	PIVC in ACF 102 (37.4%)	PIVC in ACF 88 (31.5%)	p = 0.15 a
Ruegg et al., 2022	%	—	—	—	—	—	—	—	Rate ACF PIVC (ED) 78% on week 1	Rate ACF PIVC (ED) 33% at week 9	p = 0.4 c	Demonstrated a multi‐faceted intervention reduced the proportion of ACF PIVC in the ED.
van Horrik et al., 2024	n (%)	—	—	—	282 (22%)	154 (13.8%)	OR d 0.76 CI 0.68–0.84 p < 0.001	Inappropriate use				A small temporary investment in a de‐implementation strategy was sufficient to achieve sustained effects
		—	—	—	Second PIVC, 30 (40.5%)	Second PIVC, 27 (40.9%)	OR d 1.01 CI 0.72–1.41 p = 0.97	Inappropriate use

Note: Reported changes in PIVC insertion rates, appropriateness of use, and secondary outcomes. Study findings, statistical significance and definitions of inappropriate use are provided when available.

Abbreviations: ACF, anterior cubital fossa; HA‐SAB, hospital‐acquired Staphylococcus aureus bacteraemia; MMI, multi‐modal interventions; n/c, cannot be statistically calculated; OBD, occupied bed days; RR, relative risk; VL, venous line.

^a Chi squared.

^b Relative risk.

^c Linear regression.

^d Logistic regression.

^e Unpaired students t‐test.

^f Used PIVC value recalculated to unused PIVC, BSI, Blood stream infection.

Nine studies reported secondary outcomes including anterior cubital fossa (ACF) insertion rates [29, 32, 36], bloodstream infections [34, 35, 36], and complications such as localised infection and phlebitis [31, 35, 36]. ‘Appropriately’ unused PIVCs did not change significantly in either study reporting this outcome, [6, 33] implying that interventions selectively reduced genuinely unnecessary cannulation rather than deterring clinically justified insertions. Rhodes et al. [36] showed a 63% reduction in HA‐SAB post‐intervention (RR = 0.36, 95% CI 0.17–0.76), however, a pre‐existing downward trend in HA‐SAB rates during the baseline period limits attribution solely to the intervention. Ray‐Barruel et al. [35] observed significant reduction in catheter‐related complications (RR = 0.68, 95% CI 0.48–0.96, p = 0.026) despite a paradoxical increase in overall device utilisation, consistent with the study's primary aim of improving assessment and documentation practices rather than reducing insertion rates (Table 2).

Interventions were mapped to the CFIR domains to characterise design and implementation strategies (Tables 3 and S4).

TABLE 3.

Intervention components mapped to CFIR domains.

CFIR domain	Intervention characteristics		Inner settings			Characteristics of individuals		Process
	Equipment	Educational material/guidelines	Audit and feed back	IT	Removal of PIVC (actions)	Education	Clinical champions	Events	Theoretical framework
Parenti et al., 1994		Criteria for appropriate use of peripheral IV catheters were established (guidelines intervention). (administrative intervention). IV catheters discontinued the day following coronary arteriography. on patient transfer from a monitored bed, any IV catheter be discontinued unless the physician specifically wrote an order to continue catheter use. Prior to the post‐intervention study period, both recommendations had become official hospital policy, and guidelines for catheter use were published and distributed.
Champion et al., 2013		An educational leaflet was sent to all prescribing physicians, and posters summarising appropriate indications for intravenous lines were displayed in all clinical areas				An educational slide show was presented twice to these physicians by members of the Investigator Committee.
Perez‐Granda et al., 2015		Distribution of pocket leaflets with recommendations on catheter care, posters in all nursing units				An interactive on‐line training program and talks to small groups of nurses during the different shifts. “We preferred bedside visits because they provide a more direct way of assessing clinical practice and the training program”
Rhodes et al., 2016	Standardised equipment: introduction of standardised PIVC insertion trolleys across the organisation.	Poster campaign, newly developed observation chart and revised phlebitis scoring system. Alerts: A flagging alert sticker was introduced to tag PIVCs requiring removal within 24 h				Medicine and nursing education
Decker et al., 2016	New venepuncture devices were piloted with feedback encouraged, and the layout of the ‘IV trolleys’ were changed with blood collection equipment in the top drawer and PIVC insertion equipment in the second draw.	A specimen collection guide and order of draw laminated poster was attached to the trolley to aid in the process of venepuncture.			During the interventional phase red coloured ‘remove within 24 h’ stickers for ambulance inserted PIVCs and blue coloured ‘PIVC insertion’ stickers to record PIVC insertions in the ED were introduced.	“PINK Week” also involved two education sessions about the appropriateness of PIVC insertion versus venepuncture, including educational quizzes and baking competitions with prizes		“a ‘PINK week for the PIVC Project’ was launched in [sic] ED”
Hawkins et al., 2018		Advertising posters with the 80% logo and “PIVC are you sure?” were displayed in all clinical areas of the ED.	Surveillance and feedback			Education and training, comprised clinical nurse and ED consultant physician led training to educate staff on PIVC risks, placement, and care	Champions wore shirts bearing the cannulation rates in ED intervention trial (CREDIT) insignia
Egerton‐Warburton et al., 2019	Equipment changes include renaming of PIVC trollies to ‘Intravenous and venepuncture’ trollies and making venepuncture equipment easier to access and use	Guideline implementation, promotional material (such as screen savers)	Regular published snapshots of unused PIVC rates			Education sessions	Executive and clinician champions	Logo competition	Human factor analysis and human factor engineering [38]
Laan et al., 2020		Disseminated a list of appropriate indications for catheter use through posters, pocket cards	Baseline data from all hospitals were presented as a competitive feedback report	Smart phrase for the daily patient report in electronic health records	“Empowerment of nurses depending on the local situation of the participating hospital”	Educational meetings	Local champion
Lim et al., 2020	Introduction of separate trolleys for PIVC and venepuncture procedures	Guideline implementation, slogans and promotional materials				Education sessions		Logo competitions	human factor‐designed [38]
Ruegg et al., 2022		Advertising and promotion of the study, using posters and screen savers.	A baseline digital survey to identify root causes for clinical decision making related to PIVCs. Weekly audits and feedback			Widespread education	Recruiting clinical change champions among senior medical and nursing
Ray‐Barruel et al., 2023		Dedicated vascular access device form for nurses to document PIVC, Pocket‐sized lanyard cards displaying the tool, posters, tool were placed in each patient's bedside folder				Education program was conducted by the lead author at all sites	Local champions		Promoting Action on Research Implementation in Health Services (PARiHS) framework [39]
van Horrik et al., 2024		Pocket cards, posters, digital protocols with the appropriateness criteria	Audit and feedback on the number of inappropriate catheters	Smart phrase in electronic health record (originally in hospital 2 only).	Empowerment for nurses to independently remove catheters (originally in hospital 3 only).	Educational meetings or e‐learnings for physicians and nurses.	Local champion

Note: Summary of intervention strategies used to reduce unnecessary PIVC use across included studies, categorised by CFIR domains. Components include changes to equipment, educational materials, audit and feedback, ICT infrastructure, clinical champions, and implementation processes. Studies are ordered chronologically to illustrate the increasing complexity and theoretical grounding of interventions over time.

Abbreviations: ED, emergency department; IV, intravenous; PIVC, peripheral intravenous catheter.

All studies included components under the Intervention Characteristics domain, most commonly changes to equipment and educational materials. Eleven studies addressed the Characteristics of Individuals domain through clinician education and clinical champions. Six studies addressed the Inner Setting domain via audit and feedback mechanisms [5, 32, 33, 34, 37], information and communication technology (ICT) infrastructure [34, 37], and procedural changes to support timely PIVC removal [34, 37]. Three studies incorporated Process domain structured engagement activities including logo competitions, themed awareness weeks and promotional social events [6, 29, 33]. Three studies explicitly reported behaviour change theories [6, 33, 35]. Theoretical frameworks referenced included human factors engineering [6, 33] and the Promoting Action on Research Implementation in Health Services tool [35].

Summative analysis showed intervention complexity increased significantly with publication year (Figure 2, Table S5), from simple policy changes in early studies to multi‐modal strategies integrating educational programmes, audit and feedback, equipment redesign, clinical champions and digital tools in later reports. This trajectory reflects a broader shift in healthcare behavioural change interventions towards implementation science‐informed designs that address individual and system‐level determinants. Studies addressing more CFIR domains were generally associated with larger reductions in PIVC insertion rates. This association was not observed for the unused PIVCs, where effect size appeared more strongly influenced by outcome definition, setting, and study design rather than by intervention complexity.

FIGURE 2.

Summative analysis of reported interventions and CFIR domains addressed within the included studies.

Unadjusted risk ratios for PIVC insertion rates are presented in Figure 3 (Table S6–S7). A temporal trend of increasing effect size for PIVC insertion rates was observed with publication year. The four studies reporting the largest reductions, Egerton‐Warburton et al. [33] (RR 0.618), Lim et al. [6] (RR 0.596), Hawkins et al. [5] (RR 0.767), and Decker et al. [29] (RR 0.675), are all Australian ED studies published between 2016 and 2020. In subgroup analysis excluding Ray‐Barruel et al. [35], whose documentation‐focused intervention produced an increase in recorded device utilisation (RR 1.179), linear regression showed a significant association between publication year and effect size (p = 0.031). This trend is substantially confounded by the clustering of Australian ED studies within the 2016–2020 period, which share not only greater intervention complexity but also setting, population, and a contemporaneous national focus on PIVC‐associated infection prevention. The independent contribution of intervention complexity to effect size therefore cannot be isolated from these contextual factors and should be interpreted accordingly.

FIGURE 3.

Risk ratios (RR) and 95% confidence intervals for primary outcome PIVC use across pre‐ and post‐intervention periods. A vertical reference line at RR = 1 indicates no effect or difference. Studies are ordered chronologically to illustrate temporal trends.

Unadjusted risk ratios for unused PIVCs are presented in Figure 4 (Table S9–S10). Most studies returned an RR less than 1.0, indicating a significant reduction in unused PIVCs post intervention. Non‐significant findings were observed in two studies [29, 36] and for second/additional PIVCs in van Horrik et al. [37], no temporal trend was observed.

FIGURE 4.

Risk ratios (RR) and 95% confidence intervals for the primary outcome ‘unused PIVC across pre‐ and post‐intervention periods. A vertical reference line at RR = 1 indicates no effect. Studies are ordered chronologically to illustrate temporal trends. Van Horrik reports use of both first cannula and second cannulas.

Random effects meta‐analysis was attempted for both primary and secondary outcomes; however, heterogeneity was substantial (I ² = 97.3% for PIVC insertion and I ² = 86.6% for unused PIVCs), indicating that nearly all variability was attributable to between‐study differences (Figures S1 and S2, Tables S8 and S11). Studies varied considerably in design, intervention complexity, setting, outcome definitions, data collection methodology, and risk of bias profiles. Pooling would not yield a meaningful summary estimate [40], findings are therefore presented as a narrative synthesis with unadjusted risk ratios to illustrate the direction and approximate magnitude of effect.

4. Discussion

This review identified 12 studies evaluating the effectiveness of interventions to reduce PIVC insertion rates and unnecessary cannulations in adult acute care. Most studies reported reductions in PIVC insertions and unused cannulas, with associated improvements in secondary outcomes including reduced site complications and HA‐SAB infections. The consistency of direction across studies of varying design, setting and methodological quality is the most robust finding of this review.

Interventions were mapped to the CFIR domains which provides a comprehensive, theory‐informed taxonomy of implementation determinants spanning individual, intervention, and organisational levels and enables characterisation of contextual factors and explanation of implementation outcomes across settings [26]. Studies addressing more CFIR domains were generally linked to larger reductions in PIVC insertion rates, supporting the greater effectiveness of multi‐modal approaches over single‐component interventions. This association was not observed for the unused PIVC outcomes, where effect size appeared more strongly influenced by outcome definition, setting, and study design than by intervention complexity. This trajectory potentially reflects an evolving understanding of the implementation determinants required for sustained practice change in PIVC management. The Increasing complexity of interventions required to effect and maintain reductions in PIVC use will need to be considered when designing programmes for reduction of healthcare emissions and waste [10].

Four studies reported interventions across multiple hospitals, demonstrating the applicability of multi‐modal approaches across varied settings and strengthening the external validity of findings [28, 34, 35, 37] Laan et al. [34] observed a positive association between ward level interventional compliance and reductions in inappropriate PIVC use, suggesting that intervention fidelity is an important determinant of effect within multi‐site programs.

Sustainability of effect was demonstrated in two pairs of studies. Lim et al. [6] confirmed sustained reductions in PIVC use (−13.7%) and unused PIVCs (−9.9%) in a five‐year follow up of Egerton‐Warburton et al.'s [33] study. In the Netherlands, Van Horrik et al. [37] demonstrated sustained effects 5 years after the Laan et al. [34] multi‐site de‐implementation study for both inappropriate PIVCs and urinary catheters. While no individual or set of maintained de‐implementation components explained the sustained reduction, clinicians' intrinsic motivation to reduce harm facilitated ongoing positive improvement [37]. These findings suggest a front loaded effort to change clinician behaviour can become embedded in routine practice when clinicians believe the new practice will benefit patients. Compelling change requires addressing the objectivity, consistency and plausibility of new evidence as well as overcoming the social contexts of beliefs held among professionals [41, 42].

A temporal trend of increasing intervention complexity was accompanied by an increasing effect size for PIVC insertion rate reduction, although this cannot be isolated from contemporaneous clustering of Australian ED studies, baseline trends and concurrent advances in study design methodology. It is beyond the scope of this review to attribute the magnitude of effect for PIVC reduction or unused PIVCs to either individual or grouped interventional components, and the independent contribution of intervention complexity cannot be isolated from secular changes in practice. Studies addressing multiple aspects of clinical and behavioural practice with broad interventional approaches appear to be more effective.

Three studies explicitly referenced theoretical frameworks, including Promoting Action on Research Implementation in Health Services (PARiHS) and Human Factors Engineering, implementation science methodologies which offer structured approaches to address context‐specific barriers and enablers to enhance programme uptake [38, 43]. Findings support theory‐informed, structured assessment of barriers and enablers of behaviour change, to develop context‐specific integration of decision‐support tools, targeted education, structural and system‐level changes to minimise low‐value care.

Notably, none of the included studies incorporated consumer feedback or patient‐partnered elements, representing an important gap in current approaches to improving PIVC practice. Patient involvement in intervention design is increasingly recognised as essential to relevance, acceptability, and sustained uptake [44].

The findings of this review should be interpreted in the context of several limitations. Most included studies were uncontrolled before‐and‐after designs, all rated Critical on ROBINS‐I V2 for confounding, reflecting the structural inability of this design to account for temporal confounding and concurrent co‐interventions in the absence of a comparator group. The single cluster‐randomised trial by Champion et al. [28] was rated as having some concerns due to absent detail on allocation concealment and blinding. The ITS studies provided the least confounded evidence but were also limited in number and quality. Collectively, the design limitations of the included studies mean that while the consistent direction of effect across the evidence base is informative, causal attribution of observed reductions in PIVC use to interventions cannot be established. Outcome definitions and measurement approaches varied considerably across studies, affecting comparability and resulting in the substantial heterogeneity observed. The systematic review yielded studies from high‐income countries, and findings may not be generalisable to resource‐limited settings. Publication bias cannot be excluded. Long‐term sustainability and implementation fidelity were not consistently reported across included studies.

5. Conclusion

This review demonstrates a range of interventional strategies that were effective in reducing PIVC insertions, unused PIVCs and associated complications within the context of individual studies. Interventions varied widely in scope and complexity, with most associated with reductions in PIVC use and inappropriate cannulation. The results highlight the importance of multi‐modal, theory‐informed strategies that address behavioural, organisational, and contextual determinants of practice. Future research should prioritise rigorous evaluation of intervention effectiveness and sustainability through careful selection of study designs to minimise biases inherent in unblinded behavioural interventions. Additionally, it is important to address scalability across diverse healthcare settings, incorporating consumer or patient‐partnered elements in their design.

Author Contributions

Study concept and design: Giles Barrington, Lauren E. Thurlow, Diana Egerton‐Warburton and Lisa Kuhn. Acquisition of the data: Giles Barrington, Lauren E. Thurlow, Sarah Wiggs, Sundy Ni‐Yen Yang, Suzanne Bumpstead, Bibesh Pokhrel and Lisa Kuhn. Analysis and interpretation of the data: Giles Barrington and Lauren E. Thurlow. Drafting of the manuscript: Giles Barrington and Lauren E. Thurlow. Critical revision of the manuscript for important intellectual content: Sarah Wiggs, Sundy Ni‐Yen Yang, Suzanne Bumpstead, Bibesh Pokhrel, Viet Tran, Diana Egerton‐Warburton and Lisa Kuhn. Acquisition of funding: Viet Tran, Diana Egerton‐Warburton and Lisa Kuhn.

Funding

This research was funded by the Medical Research Future Fund (MRFF)—Clinical Trials Activity Initiative—2021: Clinical Trials Activity Grant Opportunity—Stream 5 Grant MRF2023389.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Literature review search terms by data base and results.

Table S2: Reporting quality assessment of studies using the Equator Networks Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement for observational studies checklist.

Table S3: Combined Risk of Bias assessment of included studies.

Table S4: Reported Intervention classification, attributed CFIR domain and rationale.

Table S5: Summative analysis of reported interventions and CFIR domains addressed within the included studies.

Table S6: Raw data extracted from included studies showing cannula presence (case) before and after intervention (pre/post), primary outcome “PIVC use.”

Table S7: Calculated risk ratios (RR) and 95% confidence intervals (CI) comparing cannula use pre‐ and post‐intervention, primary outcome “PIVC use.”

Figure S1: Meta‐analysis forest plot with pooled estimate for primary outcome “PIVC use,” weights are for a random effects meta analysis. Lim et al. 2020 removed as not independent study.

Table S8: Test for Heterogeneity for Meta‐analysis with pooled estimate for primary outcome “PIVC use.”

Table S9: Raw data extracted from included studies showing cannula presence (case) before and after intervention (pre/post), primary outcome “unused PIVC.”

Table S10: Calculated risk ratios (RR) and 95% confidence intervals (CI) comparing cannula use pre‐ and post‐intervention, primary outcome “unused PIVC.”

Figure S2: Meta‐analysis forest plot with pooled estimate for primary outcome “unused PIVC”, weights are for a random effects meta‐analysis.

Table S11: Test for Heterogeneity for Meta‐analysis with pooled estimate for primary outcome “unused PIVC.”

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.