Central Venous Access Device Complications and Premature Removal in Patients With Haematological Malignancies: A Multi‐Site Cohort Study

Abstract

Background

Patients with haematological malignancies require urgent and reliable venous access for the administration of systemic anticancer therapies (SACTs) commonly via central venous access devices (CVADs). Disease pathophysiology and side effects of SACTs increase the risk of complications during the dwell time and premature removal. CVAD complications are associated with treatment disruption, increased morbidity and mortality. This study aimed to comprehensively describe CVAD performance over a 12‐month period in patients with haematological malignancies.

Methods

A multi‐site cohort study at four tertiary hospitals in Melbourne, Australia was undertaken using multidisciplinary data from patient health records and administrative datasets including patient, device, insertion, maintenance, complication and removal data. Cases of interest were CVADs, ascertained using lists provided by the insertion services.

Findings

A total of 1078 CVADs were inserted in 673 patients between 1 September 2020 and 31 August 2021. Of the 1078 CVADs, 197 (18%) remained in situ, and 881 (82%) were removed, of which 369 (42%) were removed prematurely due to infection (n = 208, 57%) and non‐infection related reasons (n = 201, 54%). Most CVADs (n = 919, 85%) had documented complications during their dwell time and the proportion of premature removals in these CVADs was over two‐fold higher than CVADs with no documented complications. Multivariable Cox regression results indicated that CVAD type, urgency of the procedure, concurrent CVADs and insertion technology were associated with an increased risk of premature removal. Clinical variations in insertion and management care throughout the life of a CVAD and current evidence were identified.

Conclusion

An unacceptably high proportion of CVADs had complications documented during the dwell time and were prematurely removed. Inconsistencies in current evidence and clinical practice highlight opportunities to positively impact CVAD outcomes in this cohort.

Trial Registration

The authors have confirmed clinical trial registration is not needed for this submission.

1. Introduction

Central venous access devices (CVADs) are ubiquitous in the management of patients with haematological malignancies for the administration of systemic anticancer therapies (SACTs) (chemotherapy, immunotherapy, targeted therapies and stem cell transplantation), supportive therapies (blood products, antimicrobials and parenteral nutrition) and frequent blood sampling [1]. CVADs can be in situ for days, weeks, months or years if required for prolonged treatment and do not develop unresolvable complications. CVADs are managed by clinicians from multiple disciplines as patients move between a variety of inpatient settings (potentially intensive care) and ambulatory, community, and home settings throughout the courses of SACT.

Coagulopathic and inflammatory processes associated with cancer pathophysiology and side effects of SACTs, for example, immunosuppression, bleeding and thrombosis, increase the risk of CVAD‐related complications and premature removal [2]. CVAD‐related complications and premature removal, such as CVAD associated blood stream infection (CABSI) and thrombosis, are associated with delayed treatment, increased morbidity and mortality, and considerable financial costs which represents low value care [3, 4]. Value‐based health care is a two‐fold construct, value to the patient by optimising outcomes, and value for the health care system by reducing health care expenditure [5].

Standardised, evidence‐based management decreases the risk of CVAD‐related complications [6]. However, considerable variation in CVAD management exists in clinical practice representing low value care and presenting diverse opportunities to align patient care with current evidence and optimise patient management and safety [7, 8]. CVAD‐related complications and premature removal rates in patients with haematological malignancies are not publicly reported. The primary aim of this study was to provide a comprehensive description of CVAD performance in patients being treated for haematological malignancies during a 12‐month period. The secondary aim was to identify low value health care practices and opportunities to inform initiatives to mitigate preventable, premature CVAD removal, to be tested in future research.

1.1. Objectives

The study objectives were to identify and describe reasons for CVAD removal; identify and describe all documented CVAD‐related complications during the dwell time, not at the time of removal; identify baseline patient, CVAD, and insertion factors associated with premature removal of the first CVAD in the study period; and describe multidisciplinary insertion and management practices of CVADs for patients with haematological malignancies attending four metropolitan hospitals.

2. Methods

2.1. Study Design

A multi‐site cohort study was undertaken on CVADs inserted between 1 September 2020 and 31 August 2021. An a priori protocol defined all study variables which facilitated the capture of data related to all CVADs inserted in patients receiving care for haematological malignancies over a 12‐month period. Multi‐site ethics approval was granted before commencement. This study is reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement and The REporting of studies Conducted using Observational Routinely‐collected health Data (RECORD) Statement [9, 10].

2.2. Setting

The study was conducted in four metropolitan hospitals that treat and care for patients with haematology cancers, of the seven hospitals comprising the Victorian Comprehensive Cancer Centre (VCCC) Alliance. The hospitals are tertiary, teaching referral hospitals providing medical, surgical and specialist services to adult patients in metropolitan, Melbourne, Victoria, Australia.

2.3. Participants and CVADs

Adult patients (≥ 18 years) diagnosed with haematological malignancies who had a CVAD inserted during the study period were included. Malignancies were identified using ICD‐10‐AM codes identified from the patient health records in Hospitals 1 and 2, the procedural list from insertion services in Hospital 3 and from administrative services in Hospital 4. CVADs were the unit of interest, as some patients had multiple CVADs over the study period. CVADs commonly included peripherally inserted central catheters (PICC) and tunnelled cuffed‐centrally inserted central catheters (tc‐CICC), or shorter term devices (CICCs), femorally inserted central catheters (FICC), apheresis‐CICCs (A‐CICCs) or haemodialysis CICCs (H‐CICC) and longer term devices (totally implantable venous access devices (TIVAD), apheresis TIVADs (A‐TIVAD), tunnelled cuffed apheresis catheters (tc‐A‐CICCs) and tunnelled cuffed haemodialysis CICCs (tc‐H‐CICCs). CVADs inserted before the study period, inserted in external hospitals or in patients with confidentially sealed health records were not included.

2.4. Variables

Variables collected aligned with the recommendations for a minimum CVAD dataset, published literature and clinical experience with patients with haematological malignancies [11]. Variables included patient demographic data, disease and treatment characteristics; insertion and device information including dwell time (i.e., the duration of time between insertion and removal); documented complications during the dwell time (see Table 1); and removal reasons. At the end of the study period, CVADs that were not removed either remained in situ or the patient was transferred to an external health care service with the CVAD in situ to continue treatment. CVADs removed in the study period were either due to the end of treatment or expected removal reason/cessation of use, prematurely removed, or the removal reason was not documented and could not be established from available patient health record documentation. Premature removals were categorised into clinically meaningful groups: dressing and securement related, clinician related, occlusion related, thrombosis or phlebitis were included (Table 1). All variables were defined, and definitional sources included the standards of infusion therapy from the United States of America, Canada and the United Kingdom [12, 13, 14]; and, when absent, were author devised (Supporting Information S1). Relevant variables were re‐coded for the regression analysis (Supporting Information S2).

TABLE 1.

Documented CVAD‐related complications during dwell time and premature removal reasons.

Complications during dwell time

Premature removal reasons

Bleeding/bruising at insertion site External catheter migration Occlusion CVAD associated skin impairment Local infection Dressing issues (wet, not intact, soiled) Secondary catheter tip malposition Thrombosis Insertion site—inflamed, ooze, pain Needleless connector (disconnected, soiled) Paraesthesia related to CVAD TIVAD flipped Phlebitis Thrombophlebitis Upper extremity generalised subcutaneous swelling

Infection‐related removal reasons ‘Suspected CLABSI’ Local infection ‘CLABSI’ Infection related other a Non infection‐related removal reasons Dressing and securement related Secondary catheter tip malposition Accident removal Catheter migration CVAD associated skin impairment Clinician‐related Removal to insert different type of CVAD ‘Routine’ 7‐day replacement in ICU Sufficient veins to complete treatment peripherally Patient request related to physical issue with device ‘Routine’ 7‐day replacement in ICU & local infection Clinician related other b Occlusion related Occlusion CVAD device rupture/damage c Thrombosis related Phlebitis

^a ‘Two‐day CVAD holiday’ before CVAD reinserted in bacteraemia episode, ‘to avoid infection’, part of ongoing management of bacteraemia.

^b Based on ethically approved protocol to prevent identifiable data.

^c Rupture with documented preceding episode/s of occlusion and forceful flushing.

2.5. Data Sources

Cases were ascertained from procedural lists obtained from the hospitals' insertion services. All other data were obtained from hospital administrative services and multidisciplinary documentation in patients' health records including nursing, medical (haematology, infectious diseases, intensive care, renal, radiology), pathology and pharmacy data.

2.6. Bias

Selection bias was minimised by inclusion of all patients with haematological malignancies with a CVAD inserted in the participating hospitals during the study period. Information bias was reduced by the ethics approved a priori study protocol including a priori definitions of variables and outcomes, an experienced clinical researcher collecting data from patient health records, access of multidisciplinary records and extensive cross referencing (minimise missing data) and exporting the extracted data into a standardised collection form.

2.7. Study Size

No definitive information on numbers of patients or CVADs in each hospital was available before data collection. Consequently, the sample size was pragmatic based on patients meeting eligibility criteria and time available for data collection. Using the ‘10 events per variable rule’ for Cox regression [15, 16], there were a sufficient number of events (premature removals, n = 195) amongst the subset of first CVADs inserted in the study period (n = 655) to examine 16 predictors/variables (taking into consideration the number of categories in categorical variables).

2.8. Statistical Analysis

Data from the purpose‐built, secure electronic case report form in Research Electronic Data Capture (REDCap) [17, 18] was imported into SPSS Version 25 (Chicago, IL, USA) for initial exploratory and descriptive analysis (recodes described in Supporting Information S2). Data were checked for missing and out of range values in REDCap and SPSS. Missing and conflicting data were addressed by cross‐referencing multiple sources in the patient health records. Absent data post‐examination of multiple sources were noted as an important clinical finding. Missing data were not imputed. The unit of analysis was the CVAD, to capture the proportion of patients who had multiple devices inserted during the study period. Descriptive statistics were used to summarise patient, device and insertion characteristics for the full sample and by device type. The remaining analyses were performed in R [19]. Complications at any stage during the dwell time and reasons for removal were summarised using counts and percentages with 95% confidence intervals (CIs), and incidence rates (IRs) per 1000 catheter days with 95% CIs using the ‘binom’ [20] and ‘epiR’ [21] packages, respectively. Reasons for removal calculations only included CVADs removed during the 12‐month study period (i.e., CVADs that remained in situ at the end of the study period and CVADs in patients transferred to an external facility were excluded).

Survival analyses were performed using the ‘survival’ package [19]. Only the first device inserted in the study period was analysed. Devices not removed during the study period and devices in patients transferred to another facility were censored at the date of last follow‐up. Premature removal probabilities per device type were calculated using the Kaplan–Meier method and plotted using the ‘survminer’ package [22]. Univariable and multivariable Cox regression models were used to assess the association between device type and time to premature removal. All models included a shared frailty term to account for the clustering of patients within hospitals [23] and alpha was set at 0.05 (two‐tailed). Baseline covariates—patient, device and insertion factors—were selected using clinical knowledge and evidence published in the professional literature, however, collinearity and missing data were also considered [24]. ‘Inserter’, which exhibited a strong association with ‘insertion technology’ (p < 0.001, Cramer's V = 0.68), was excluded from regression modelling, as were ‘catheter to vein ratio’ and ‘insertion attempts’ because these were not documented for all/most patients (n = 655 and n = 506, respectively). The multivariable model included all pre‐specified covariates apart from stated exclusions; results of univariable analyses were not considered [25, 26]. Descriptive statistics were used to explore clinical variation [27] (e.g., documentation of catheter to vein ratio, technology used to guide catheter tip placement).

3. Results

3.1. Patients

Patient characteristics for the 673 patients with one or more CVADs included in the study period are summarised in Table 2. The mean age was 58 years (range 19–90 years) and the majority were male (n = 375, 56%). The most common diagnoses were non‐follicular lymphoma (n = 167, 25%), multiple myeloma (n = 164, 24%) and myeloid leukaemia (n = 148, 22%). One or more comorbidities were documented for 434 patients (64%). ‘Diabetes’ was documented for 77 (12%) patients and renal disease for 61 (9%).

TABLE 2.

Patient and CVAD characteristics at insertion.

n = (%)	PICC	tc‐CICC	A‐CICC	CICC	tc‐A‐CICC	H‐CICC	FICC	chest‐TIVAD	tc‐H‐CICC	A‐TIVAD	Total
CVAD	644 (60)	188 (17)	69 (6)	49 (5)	41 (4)	40 (4)	18 (2)	14 (1)	9 (1)	6 (1)	1078
Catheter days	32,905	13,673	246	248	2158	241	97	2204	702	906	53,380
Patient characteristics (n = 673)
Age
Mean	61	51	58	59	57	73	34	59	71	43	58
Standard deviation	14	14	13	20	13	2	9	17		17	15
Median	64	54	61	63	61	73	34	61	71	43	61
Interquartile range	54–71	43–64	52–67	45–76	49–66			55–72		27–60	50–69
Range	19–90	19–79	22–76	21–78	23–74	71–74	26–43	23–78		24–63	19–90
BMI
Mean	28	29	28	29	28	26	24	26	34	29	28
Standard deviation	6	6	6	4	6	8	5	8		9	6
Median	28	28	28	29	28	26	24	25	34	26	28
Interquartile range	24–32	25–32	24–32	27–33	24–34			21–32		23–37	24–32
Range	15–61	17–55	19–48	21–33	17–44	20–32	20–29	15–39	34	20–46	15–61
Missing	2		1
Sex
Male	232 (57)	77 (54)	33 (61)	4 (44)	19 (53)	0	3 (100)	4 (29)	1 (100)	2 (33)	375 (56)
Female	174 (43)	65 (46)	21 (39)	5 (56)	17 (47)	2 (100)	0	10 (71)	0	4 (67)	298 (44)
Diagnosis a
C83 Non‐follicular lymphoma	125 (31)	7 (5)	20 (37)	1 (11)	10 (28)	0	0	4 (29)	0	0	167 (25)
C90 Multiple myeloma, malignant plasma cell neoplasms	120 (30)	3 (2)	13 (24)	2 (22)	20 (56)	1 (50)	0	4 (29)	1 (100)	0	164 (24)
C92 Myeloid leukaemia	69 (17)	73 (52)	0	4 (44)	0	0	1 (33)	0	0	1 (17)	148 (22)
C91 Lymphoid leukaemia	20 (5)	26 (18)	8 (15)	1 (11)	0	0	1 (33)	0	0	3 (50)	59 (9)
C81 Hodgkin lymphoma	15 (4)	4 (3)	3 (6)	0	3 (8)	0	0	1 (7)	0	0	26 (4)
C82 Follicular lymphoma	8 (2)	6 (4)	2 (4)	0	1 (3)	0	0	1 (7)	0	0	18 (3)
C84 Mature T/NK‐cell lymphomas	10 (2)	4 (3)	2 (4)	0	3 (8)	0	0	0	0	2 (33)	21 (3)
C85 Other and unspecified types of NHL	19 (5)	1 (1)	7 (13)	0	0	1 (50)	1 (33)	0	0	0	29 (4)
C86 Other specified types of T/NK‐cell lymphoma	6 (1)	3 (2)	1 (2)	0	0	0	0	2 (14)	0	0	12 (2)
C88 Malignant immunoproliferative diseases	2 (0)	0	0	0	0	0	0	1 (7)	0	0	3 (0)
C93 Monocytic leukaemia	5 (1)	2 (1)	0	0	0	0	0	0	0	0	7 (1)
C94 Other leukaemias of specified cell type	0	1 (1)	0	0	0	0	0	0	0	0	1 (0)
C95 Leukaemia of unspecified cell type	1 (0)	1 (1)	1 (2)	0	0	0	0	0	0	0	3 (0)
C96 Other and unspecified malignant neoplasms of lymphoid, haematopoietic and related tissue	3 (1)	0	0	0	0	0	0	0	0	0	3 (0)
D45 Polycythaemia vera	1 (0)	0	0	0	0	0	0	0	0	0	1 (0)
D46 Myelodysplastic syndromes	13 (3)	20 (14)	0	2 (22)	0	0	0	0	0	0	35 (5)
D47 Other neoplasms of uncertain or unknown behaviour of lymphoid, haematopoietic and related tissue	9 (2)	8 (6)	1 (2)	0	0	0	0	2 (14)	0	0	20 (3)
Comorbidities
None documented	140 (34)	62 (44)	26 (48)	1 (11)	9 (25)	0	1 (33)	6 (43)	0	2 (33)	247 (37)
Diabetes b	59 (15)	8 (6)	5 (9)	0	3 (8)	1 (50)	0	1 (7)	0	0	77 (11)
Renal disease b	45 (11)	8 (6)	1 (2)	2 (22)	1 (3)	1 (50)	1 (33)	1 (7)	1 (100)	0	61 (9)
Other c	245 (60)	76 (54)	26 (48)	9 (100)	25 (69)	2 (100)	2 (67)	7 (50)	1 (100)	4 (67)	397 (59)
History of venous thrombosis	38 (9)	13 (9)	2 (4)	1 (11)	5 (14)	0	1 (33)	0	0	0	60 (9)
Solid malignancy—current or history	39 (10)	6 (4)	2 (4)	1 (11)	2 (6)	0	0	0	0	2 (33)	52 (8)
Device characteristic (n = 1078)
Priority for insertion
Planned	517 (80)	174 (93)	64 (93)	8 (16)	39 (95)	4 (10)	2 (11)	14 (100)	4 (44)	5 (83)	831 (77)
Urgent	127 (20)	14 (7)	5 (7)	41 (84)	2 (5)	36 (90)	16 (89)	0	5 (56)	1 (17)	247 (23)
Side of insertion
Right	268 (42)	173 (92)	64 (93)	30 (63)	40 (98)	22 (56)	12 (67)	14 (100)	8 (89)	5 (83)	636 (59)
Left	376 (58)	15 (8)	5 (7)	18 (38)	1 (2)	17 (44)	6 (33)	0	1 (11)	1 (17)	440 (41)
Missing				1		1
Number of lumens
1 lumen	14 (2)	1 (1)	0	0	0	1 (3)	0	13 (93)	0	6 (100)	35 (3)
2 lumens	620 (96)	177 (94)	68 (99)	0	41 (100)	4 (11)	0	1 (7)		0	919 (86)
3 lumens	10 (2)	10 (5)	1 (1)	2 (4)	0	32 (86)	0	0	0	0	55 (5)
4 or 5 lumens	0	0	0	46 (96)	0	0	18 (100)	0	0	0	64 (6)
Missing				1		3
Vessel of insertion
Basilic	471 (73)	0	0	0	0	0	0	0	0	0	471 (44)
Internal jugular	0	185 (98)	66 (96)	45 (92)	38 (93)	21 (53)	0	11 (79)	9 (100)	6 (100)	381 (35)
Brachial	146 (23)	0	0	0	0	0	0	0	0	0	146 (14)
Femoral	0	0	3 (4)	0	0	18 (45)	18 (100)	0	0	0	39 (4)
Cephalic	11 (2)	0	0	0	0	0	0	0	0	0	11 (1)
External jugular	0	0	0	2 (4)	0	0	0	0	0	0	2 (0)
Subclavian	0	0	0	1 (2)	0	0	0	0	0	0	1 (0)
Not documented	16 (2)	3 (2)	0	1 (2)	3 (7)	1 (3)	0	3 (21)	0	0	27 (3)
Indication d
Chemotherapy	336 (52)	92 (49)	0	4 (8)	7 (17)	0	0	10 (71)	0	0	449 (42)
Autologous stem cell transplant	120 (19)	15 (8)	1 (1)	3 (6)	30 (73)	5 (13)	2 (11)	0	0	0	176 (16)
Allogenic stem cell transplant	36 (5)	82 (43)	0	16 (32)	0	7 (18)	7 (39)	0	1 (11)	0	149 (14)
Supportive therapies	110 (17)	2 (1)	1 (1)	4 (8)	0	0	6 (33)	1 (7)	0	0	124 (12)
Apheresis procedures	0	1 (1)	69 (100)	0	29 (71)	0	0	0	0	6 (100)	105 (10)
Irritant/vesicant administration e	15 (2)	0	0	41 (84)	0	0	15 (83)	0	0	0	71 (7)
Poor venous access	55 (9)	0	2 (3)	4 (8)	0	0	0	9 (64)	0	1 (17)	71 (7)
Haemodialysis procedures	0	0	0	0	0	40 (100)	0	0	9 (100)	0	49 (5)
Car‐T	40 (6)	1 (1)	0	0	0	0	0	0	0	0	41 (4)
Haemodynamic monitoring	0	0	0	20 (41)	0	0	12 (67)	0	0	0	32 (3)
Insertion of different type of CVAD	10 (2)	8 (4)	0	2 (4)	1 (2)	0	0	0	0	0	21 (2)
Patient preference	0	1 (1)	0	0	0	0	0	2 (14)	0	0	3 (0)
Other	1 (0)	1 (1)	0	1 (2)	0	0	0	1 (7)	0	0	4 (0)
Unknown	3 (0)	0	0	0	0	0	0	0	0	0	3 (0)

^a Multiple diagnoses possible, sum may exceed total number, percentages exceed 100%. Less aggressive diagnoses: C82 Follicular lymphoma, C90 Multiple myeloma and malignant plasma cell neoplasms, D45 Polycythaemia vera, D46 Myelodysplastic syndromes, C96 Other and unspecified malignant neoplasms of lymphoid, haematopoietic and related tissue, D47 Other neoplasms of uncertain or unknown behaviour of lymphoid, haematopoietic and related tissue, Other. More aggressive: C83 Non‐follicular lymphoma C84 Mature T/NK‐cell lymphomas C81 Hodgkin lymphoma C86 Other specified types of T/NK‐cell lymphoma, C85 Other and unspecified types of NHL C88 Malignant immunoproliferative diseases C91 Lymphoid leukaemia, C92 Myeloid leukaemia, C93 Monocytic leukaemia, C94 Other leukaemias of specified cell type, C95 Leukaemia of unspecified cell type.

^b Comorbidities documented in at least 5% of participants.

^c Noted in patient notes.

^d Multiple responses possible.

^e Vasopressors, parenteral nutrition, antibiotics.

3.2. CVADs

In total, 1078 of the 1112 (97%) CVADs were successfully inserted in the study period (n = 34 insertion failures), with a total dwell time of 53,380 days (Table 2). The most common CVADs inserted were PICCs (n = 644/1078, 60%, 32,905 CVAD days) and tc‐CICCs (n = 188/1078, 17%, 13,673 CVAD days). Most CVADs were inserted on the right side (n = 636, 59%) and the majority had 2 lumens (n = 919, 86%). The most common indication for insertion was chemotherapy (n = 449, 42%), autologous stem cell transplant (n = 176, 16%), allogeneic stem cell transplant (n = 149, 14%), supportive therapies (n = 124, 11%) and apheresis procedures (n = 105, 10%). Most patients (n = 455, 68%) had only one CVAD inserted in the 12‐month study period, 125 (19%) patients had two CVADs, 93 (14%) had three or more CVADs inserted (range 3–11).

3.3. CVAD Status at End of Study Period

Of the 1078 CVADs successfully inserted, 881 (82%) were removed and 197 (18%) were not removed during the study period. CVADs that were not removed either remained in situ for use at one of the participating hospitals (n = 145, 74%) or the patients were transferred to another hospital with the CVAD in situ (n = 52, 26%). Of those removed, most (n = 500, 57%) were removed due to expected reasons or cessation of use. Expected removal reasons or cessation of use included end of therapy (n = 405, 81%), CVADs in situ at the time of a non‐CVAD related death (n = 86, 17%), and other reasons in 9 (2%) CVADs (de‐escalation of treatment, n = 8; post disconnection of needleless connector from CVAD, n = 1). The remaining CVADs were removed prematurely (n = 369, 42%) or the removal reason was not documented (n = 12, 1%) and could not be established from other documentation in the patient health record.

3.4. CVAD Performance

3.4.1. Premature Removals

Of the 881 CVADs removed during the study period, 369 were removed prematurely (42%; 95% CI: 39–45), at an IR of 9.61 per 1000 catheter days (95% CI: 8.65–10.64). CVADs were removed due to one or more premature reasons including infection related reasons (n = 208; 23%; 95% CI: 20–26; IR 5.41; 95% CI: 4.70–6.20) and non‐infection‐related reasons (n = 201; 23%; 95% CI: 20–26; IR 5.32; 95% CI: 4.53–6.01) (Table 3).

TABLE 3.

CVAD performance (CVADs removed during study period) (n = 881, 38,382 catheter days).

	Episodes (N)	Proportion of CVADs (%)	95% CI	Incidence rate per 1000 catheter days	95% CI
Premature removal a	369	42	39–45	9.61	8.65–10.64
Infection‐related premature removal reasons b	208	24	20–26	5.41	4.70–6.20
‘Suspected CLABSI’	146	16	14–19	3.80	3.21–4.47
Local infection	31	3	2–5	0.80	0.54–1.14
‘CLABSI’	26	3	2–4	0.67	0.44–0.99
Other c	8	1	0–2	0.20	0.08–0.41
Non‐infection‐related removal reasons	201	23	20–26	5.32	4.53–6.01
Dressing and securement related	66	7	6–9	1.71	1.32–2.18
Clinician related	60	7	5–9	1.56	1.19–2.01
Occlusion related	41	5	3–6	1.06	0.76–1.44
Thrombosis related	33	4	3–5	0.85	0.59–1.20

^a Counts may exceed total number and percentages may exceed 100 due to multiple reasons for removal.

^b Three CVADs had two different infection‐related premature removal reasons (total infection related reasons n = 211 for 208 CVADs).

^c Includes ‘2‐day CVAD holiday’ before CVAD reinserted in bacteraemia episode n = 2, ‘to avoid infection’ n = 3, part of ongoing management of bacteraemia n = 2.

3.4.2. Documented Complications

Documented complications were reported for all CVADs (n = 1078) as these events occurred during the dwell time, not at the time of removal so the complete sample was analysed. Of the 1078 CVADs inserted, 919 had documented complications during the dwell time (85%, 95%CI: 83.2–87.2), with an IR of 17.1 per 1000 catheter days (95% CI: 16.01–18.2). The most common complications (alone or in combination) were bleeding from or bruising at the exit site n = 539 (50%), catheter migration in or out at the exit site n = 363 (34%), occlusion n = 344 (32%), CASI n = 284 (26%) or local infection (n = 271, 25%) (Table 4).

TABLE 4.

CVAD performance (CVADs n = 1078, 53,345 catheter days).

	Episodes (i)	Proportion of CVADs (%)	95% CI	Incidence rate per 1000 catheter days	95% CI
CVADS with complications documented during the dwell time a	919	85	83–87	17.21	16.12–18.36
Bleeding/bruising at catheter exit site b	539	50	47–53	10.09	9.26–10.98
External catheter migration	363	34	31–36	6.80	6.12–7.54
Occlusion	344	32	29–35	6.44	5.78–7.16
CVAD associated skin impairment	284	26	24–29	5.32	4.71–5.97
Local infection	271	25	23–28	5.07	4.49–5.71
Dressing issues—wet, not intact, soiled	73	7	4–8	1.36	1.07–1.71
Secondary catheter tip malposition	67	6	5–8	1.25	0.97–1.59
Thrombosis	54	5	3–6	1.01	0.75–1.31

^a Counts may exceed total number and percentages may exceed 100 due to multiple documented complications during dwell time. Complications per CVAD, not number of events per CVAD during dwell time, potentially an underestimate.

^b Calculated from total of removed CVADs, n = 881 and 38,382 catheter days.

3.4.3. Premature Removal of CVADs With and Without Documented Complications

The proportion of premature removals of CVADs with complications documented during the dwell time (n = 919) was over two‐fold (n = 342, 37%) compared to CVADs with no complications documented during the dwell time (n = 27 of 159, 17%).

3.5. Predictors of Premature Removal

Univariable and multivariable Cox regression results are detailed in Supporting Information S3. These were quite similar overall, so multivariate results only are presented here for brevity. Compared with the reference group (PICC, the most common CVAD inserted), there was an increased risk of premature removal at any point in time for tc‐CICC (HR [95% CI]: 1.75 [1.04, 2.95]) and other short‐term devices (HR [95% CI]: 4.02 [1.41, 11.50]) and a decreased risk for long‐term devices (HR [95% CI]: 0.43 [0.20, 0.94]). There was also an increased risk of premature removal at any point in time in patients whose insertion procedures were urgent (HR [95% CI]: 1.93 [1.07, 3.48]), in those with concurrent CVADs (HR [95% CI]: 2.22 [1.35, 3.65]) and for devices inserted using x‐ray only (HR [95% CI]: 1.67 [1.09, 2.55]; reference group: fluoroscopy only).

For the first CVAD, compared with PICCs, premature removal of short‐term devices (CICC, FICC, A‐CICC, H‐CICC) and tc‐CICC occurred earlier in the dwell time, whereas premature removal of long‐term devices (chest‐TIVAD, A‐TIVAD, tc‐A‐CICC, tc‐H‐CICC) occurred later (Figure 1).

FIGURE 1.

Kaplan–Meier survival estimates of premature removal per device type.

3.6. Health Care Variation

Clinical variation, the difference in clinical practice and the standards of therapy (evidence‐based guidelines) [27] were identified throughout various stages in the life of a CVAD (device selection, insertion, maintenance and removal). For example, no insertion procedural documentation included catheter to vein ratio or the use of intracavity electrocardiograph (ECG) for catheter tip placement. Over one third of primary catheter tip positions were suboptimal (n = 351, 33%). Nearly 60% of CVAD dressings were soiled from a bleeding catheter exit site, 37% had occlusions and 39% had documented migration in or out of the catheter at some stage during the dwell time. Approximately 10% of PICCs (n = 67) were in situ for less than 7 days.

4. Discussion

This multi‐site study comprehensively described removal reasons and documented complications during the dwell time of a wide variety of CVADs in patients with haematological malignancies over a 12‐month period of their SACTs and supportive therapies. There are few published studies that have investigated the full range of complications and removal reasons across the range CVAD types in this patient cohort. Studies commonly focus on one type of CVAD, one complication or premature removal reason, for example, infection or thrombosis. This study identified approximately two in five (42%) CVADs were removed prematurely for various infection and non‐infection related reasons, and in some cases both. The majority of CVADs (n = 919, 85%) had complications documented at some stage during the dwell time, and the proportion of premature removals of these devices was two‐fold compared to those with no documented complications. Multivariable Cox regression results indicated that device type, urgency of the procedure, the presence of a concurrent CVADs and insertion technology for catheter tip placement (x‐ray only vs. fluoroscopy only) were associated with an increased risk of premature removal at any point in time. Clinical variation or deviation from evidence‐informed practice recommendations across multidisciplinary clinical practices were identified throughout the life of the CVADs. These findings identify important opportunities to inform clinically relevant initiatives to mitigate preventable, premature CVAD removal and complications including supporting clinicians with contemporary education, technology, materials and products within the health care services.

Study findings suggested that, of the CVADs removed, nearly two in every five (42%) were removed prematurely, with relatively equal proportions removed due to infection (23%) and non‐infection related reasons (23%). This was considerably higher than a study of PICCs and tc‐CICCs in patients with AML (7.1% and 7.9% respectfully) [28], and 34% in PICCs and patients with haematological malignancies [29]. The patient cohort in this study, for example, those undergoing allogeneic and matched unrelated donor stem cell transplants, had higher acuity clinical needs, heterogeneity of infection related outcomes in the literature (i.e., catheter related blood stream infection [CRBSI], catheter associated bloodstream infection [CLABSI], ‘suspected CLABSI’ and CVAD associated bloodstream infection [CABSI]), hampers comparison between studies. The diverse range of infection and non‐infection related removals, categorised into clinically relevant groups, demonstrate opportunities to implement and standardise an evidence‐informed bundle of care approach to CVAD insertion, care and management to mitigate low value, unsafe practices by aligning current multidisciplinary CVAD management with current evidence and contemporary materials.

Reliable venous access is critical for uninterrupted delivery and timely completion of SACTs for patients with haematological malignancies. CVAD dysfunction was a predictor of poor overall survival in paediatric patients with cancer [30]. Our study identified more than 93 (12%) patients required three or more CVADs, with a maximum of 11, to continue SACTs over the 12‐month study period and 32% of CVADs (n = 344) had occlusion documented during the dwell time. This potentially represents disruption or delays in SACTs, in some cases at times in acute periods post bone marrow transplant or in emergent situations in ICUs as patency or new venous access was established. Treatment was mostly delivered through PICCs and tc‐CICC; however, a diverse range of other types of CVADs for apheresis or haemodialysis procedures, intensive and short‐term support in the ICU, and longer term CVADs for prolonged or repeated vascular access was required. This illustrates the complexity of supportive care and CVAD management of patients with haematological malignancies during their SACTs. There were no studies identified from a literature search with multiple types of CVADs in this patient cohort for comparison.

The vast majority of CVADs (85%) had complications documented at some stage during the dwell time, commonly dressing related complications (i.e., insertion site bleeding or haematoma, catheter migration, CASI) and occlusion. While we acknowledge this patient cohort commonly encounter episodes of impaired immunity, inflammation or coagulopathies, bundled practice approaches have been demonstrated to positively impact CVAD‐related outcomes such as infection [31, 32]. A dressing management bundle presents a potential multifaceted, clinically relevant approach to addressing exit site bleeding, catheter migration and skin impairment incorporating contemporary evidence, products and material. This study also identified considerable increased proportions of premature CVAD removal (two‐fold) when CVADs had documented complications during the dwell time. So, reducing documented complications during the dwell time and the associated proportion of premature removals aligns with value‐based health care by optimising patient outcomes.

Cox regression analysis was undertaken to identify predictive factors for premature CVAD removal. Multivariate regression results indicated that CVAD type, urgency of the procedure, concurrent CVADs and insertion technology increased the risk of premature removal. For the most part, these are clinically reasonable findings for this patient population. Insertion of multiple concurrent CVADs for urgent commencement of therapies and insertion of shorter term CVADs (e.g., CICCs and FICCs) may be associated with increased patient acuity and concurrent comorbidities, and an increased risk of premature removal. Technological confirmation of the primary catheter tip position using x‐ray post procedure may result in delayed detection of tip malposition requiring adjustment post procedure, possibly replacement or removal compared to intraprocedural fluoroscopy.

CVAD management in this vulnerable patient population is complex and dynamic. Standardised, evidence‐based clinical practice is required to reduce variation and enhance patient safety [33]. This is pertinent as this cohort travels through intensive, consolidation or recovery phases of SACTs in inpatient, ambulatory and home settings. Variation in clinical management in response to differences in patient needs is acceptable; however, unwarranted variation is not [27]. In this study, clinical variations from standard guidelines were evident; for example, shorter term CVADs (CICC or FICC) replaced solely due to time, longer‐term CVADs being removed to be immediately replaced by another type of longer‐term CVAD, and catheter to vein ratio or the use of intracavity electrocardiogram technology were not documented in any insertion procedural records. Catheter tip positions were documented acceptable for use in locations other than at the cavoatrial junction, for example, brachiocephalic or subclavian vein. Tissue adhesive or haemostatic dressing materials were not documented for management of exit site bleeding. External catheter migration in or out of up to 11 cm with no use of subcutaneous securement devices. Management was varied, according to the individual clinician. This aligns with the findings of a survey by Yuen et al. of cancer centres in Australia and New Zealand and the considerable variation from device selection to removal and a call for future research to inform clinical practice [8]. These findings highlight opportunities for collaborative, informed and multidisciplinary revision and alignment of current practices to current evidence and contemporary products, and technology to positively impact CVAD care and outcomes [34].

Suboptimal CVAD practices and poor CVAD outcomes represent low value care. Standardised, evidence‐based management decreases the risk of CVAD‐related complications [6] and aligns with a value‐based approach to address and inform standardised prevention, and early identification and management of complications during the dwell time to minimise premature removal. This presents challenges and opportunities at hospital and health system levels. Clinicians require ready access to appropriate evidence‐based contemporary materials, equipment, technology, education and information technology, as well as organisational policies and procedures. For example, the use of tissue adhesive or haemostatic agents to manage bleeding at the catheter exit sites [35], subcutaneous engineered securement devices to prevent catheter migration, micromotion or accidental dislodgement [36, 37], and flushing and locking practices, patency assessment and proactive management to minimise occlusions [38]. Hospitals require adequate levels of funding and resources to provide efficient, effective health care services. However, barriers to implementation of evidence‐based CVAD practice include a lack of awareness of the prevalence of the problem, inadequate or infrequent education, poor communication and standardised procedures or policies [39]. This study highlights the prevalence of premature removal and documented complications in this patient cohort, as well as potential areas of clinical practice to align with current evidence and contemporary products.

Our study has some limitations including a reliance on pre‐existing data in patient health care records collected for purposes other than research or surveillance, potentially impacting data quality. Nonetheless, this was optimally managed by using a research protocol with a priori definitions and multidisciplinary cross‐referencing to maximise the robustness of the data collected. Data extraction was completed by one researcher potentially introducing bias. Data are reported by sex, but not possible to report gender as these data were not recorded in the patient health records. Furthermore, the timing of the multiple complication events over the prolonged dwell time of the CVAD was not recorded, so could not be included regression modelling. However, this was not a study objective, with the focus being on baseline factors associated with premature CVAD removal. Infection related complications and premature removal, that is ‘suspected CLABSI’, ‘CLABSI’, local infection and other infection related reasons were not confirmed through application of standardised definitions and were reported as recorded by the treating medical team in the patient health records. This reflects the variation in clinical practice where a multitude of terms are used.

Findings from this multisite study indicate that the proportions of CVADs with documented complications and premature removals are considerable. It underscores the need to better understand and rectify suboptimal CVAD outcomes in the target population. Inconsistencies in multidisciplinary insertion and management practices, along with current evidence, highlights opportunities to mitigate potentially preventable complications and low value care. We advocate strategic interventions at both the local and system levels, prioritising the enhancement of patient‐centred outcomes.

Author Contributions

Kerrie Curtis: conceptualisation, data curation, formal analysis, investigation, methodology, project administration, resources, software, supervision, validation, visualisation, writing–original draft, writing–review and editing. Samantha Keogh: conceptualisation, investigation, methodology, project administration, supervision, validation, visualisation, writing–review and editing. Meinir Krishnasamy: conceptualisation, investigation, methodology, project administration, resources, supervision, validation, visualisation, writing–review and editing. Karla Gough: conceptualisation, formal analysis, investigation, methodology, project administration, software, supervision, validation, visualisation, writing–review and editing.

Ethics Statement

Multisite ethics approval granted by Peter MacCallum Cancer Centre Human Research Ethics Committee [EC00235], HREC Project ID: HREC/69997/PMCC before commencement of the study.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting Information

Supporting Information

Supporting Information

Data Availability Statement

All data generated and analysed during this study are included in this published article [and its supplementary information files].