Study design: prospective cohort observational study

An unselected cohort was constituted by including every consecutive PICC inserted during a four month period (July through October 2010), regardless of demographic or medical status of the recipient patient or indication for PICC use. PICC placement was exclusively performed by the radiology department of the Montpellier University Hospital. Every PICC was prospectively and weekly followed until removal or until the end of the study in February 2011. Some patients were enrolled more than once, if they had more than one PICC during the inclusion period. All patients gave informed consent.

Data collection

At the time of PICC insertion, we collected data concerning patients’ demographic characteristics, comorbidities, immunosuppressive therapy, hospitalization ward and location of the patient 72 h after insertion. Data concerning the PICC were also collected: date of insertion, operator (junior or senior radiologist), treatment indication, rank of the PICC, device characteristics, compliance with pre-operative antisepsis protocol (Fig. 1), site of insertion, length and success of the procedure and type of PICC fixation.

Data were collected by performing patient chart review and/or phone calls to healthcare professionals involved in the patient’s care or directly to the patient. Information was obtained on patient outcome and occurrence of a catheter dysfunction or signs of infection. Data on PICC utilization concerned number of daily PICC accesses, frequency of dressings, type of antiseptic used for manipulating PICC lines, catheter flushing procedure and frequency of intravenous administration set change.

An information note about PICC care rules was given to the patient at time of placement and standardized protocols of antisepsis and PICC care were available in healthcare units of our hospital (Fig. 1). In instances of premature PICC removal, data were collected regarding the circumstances warranting the removal. If PICC-related infection was suspected, insertion site swabbing and catheter tip culture (according to Brun-Buisson method) were required, along with blood cultures if clinically relevant (general infectious symptoms). For each PICC, data were collected in a standardized questionnaire which was used throughout PICC follow-up.

The “inpatients” subgroup included all PICCs monitored from insertion to removal in a health-care setting and the “outpatients” subgroup included all PICCs monitored in the outpatient setting. PICCs used alternatively in hospital and outside were classified in a “mixed setting” subgroup.

Definitions

Diagnoses of catheter related infections were established according to the French definitions:

- Confirmed catheter-related BSI was defined as the association of a positive blood culture in a patient having had a central line within 48 h prior to the onset of symptoms, AND one of the following criteria: 1) a positive culture of either catheter tip or exit site swabbing (≥10³ CFU/ml) involving the same organism as blood culture, 2) blood cultures from peripheral venous puncture and central lines positive with the same organism with a quantitative ratio (central sample/peripheral sample) > 5, or 3) a differential time to positivity > 2 h in favor of central line sample.

- Confirmed catheter-related local infection (LI) was defined as a positive culture of the PICC segment (≥10³ CFU/ml) with pus emerging from the exit site or a tunnel infection, with local manifestations of infection but no general signs of sepsis and negative blood cultures.

When all these criteria were not present or bacteriological culture not realized, or realized when the patient was under antibiotic therapy, we classified suspected infections as “possible infection”. When cultures remained negative (in the absence of antibiotics) or another cause of infection was diagnosed, the case was classified as “infection not confirmed”.

Local inflammation was defined by redness and/or soreness at the catheter exit site. General inflammatory signs were defined as isolated fever/chills without focal signs of infection.

Catheter-related venous thrombosis was defined by thrombus presence by ultrasonography.

Among catheter dysfunctions, pre occlusive events were defined as either a significant reduction of infusion flow or an impairment of blood back-flow. Lumen occlusion was defined by the permanent inability to flush the catheter or obtain blood back-flow.

Statistical analysis

Quantitative variables were described as mean (+/-SD) or median (Q25-Q75) according to normality of distribution. For each variable, Odds Ratios (ORs) were obtained using logistic regression with the type of complication as the dependent variable. Each PICC insertion was considered as a new event. For each type of complication, a multivariate logistic model was then performed with all variables that were close to significance in the first model (p < 0.20). Otherwise, a p-value <0.05 was used for statistical significance. All statistical analyses were performed with SAS software (SAS Institute Inc, Cary, NC).

Patients’ characteristics

Two PICCs were lost to follow-up, hence 192 PICCs were monitored from insertion to removal, for a total of 5218 PICC-days: 3337 PICC-days for inpatients (2700 in our hospital, 637 in other hospitals) and 1881 PICC-days for outpatients (Fig. 2). Overall mean PICC dwell time was 27.2 days (median 17 days; range 2–174), with mean dwell times of 23 days in the inpatient setting, 27.5 days in the outpatient setting and 46.9 days in patients managed in a mixed setting. Longest dwell times (>100 days) were observed mainly in oncology patients.

Placement conditions and type of PICCs

All PICCs were inserted in the department of radiology, mostly by trained senior interventional radiologists (91.2%). Surgical aseptic conditions were reportedly applied in 100% of cases, in compliance with local recommendations. The PICC was inserted mostly for multiple indications (67%). The mean duration of the procedure was 15.20 min (3–120). No immediate complication was observed during or after insertion. A majority of PICCs were single lumen catheters (90.2%), medium size (96.9% were high flow 5 French devices), with a distal valve (79.4%) and mainly introduced under echographic control in the basilic vein (66.5%). Most PICCs were held in place with sutures (95.9%); 8 PICCs were attached with StatLock adhesive dressings (StatLock, Bard, Murray Hill, NJ, USA), which was privileged for young patients (4 had cystic fibrosis), with a mean PICC dwell time of 16 days, versus 27.7 days with suture.

Modalities of PICC utilization

Frequency of PICC utilization (i.e: number of accesses per day) decreased over the course of care, with a higher proportion of seldom or never used PICCs in the outpatient settings (Fig. 3a and b).

Whatever the care setting, 55% of intravenous administration sets were replaced every 3–5 days; dressing frequency, type of antiseptic used and protocol of catheter flushing complied with local recommendations in 39.5% of the cases. Alcoholic chlorhexidine was mostly used in our hospital (78%), while povidone-iodine was mostly used in outpatient settings (94%).

PICC outcomes

The global complication rate was 30.2% (11.1 per 1000 PICC-days) with a mean time to onset of 16.1 days. This rate was higher in the inpatient setting (36.1%; 14.38 per 1000 PICC-days) than in the outpatient setting (19.4%; 3.19 per 1000 PICC-days).

Occlusions and accidental withdrawals occurred on average 16 days and 8 days respectively after PICC insertion, and the 3 episodes of deep vein thrombosis occurred 4, 5 and 39 days after PICC insertion.

PICC-related infectious complications

Overall, 12 confirmed PICC-related infections (3 BSI and 9 LI) occurred, amounting to an infection rate of 2.3 per 1000 PICC-days. The overall PICC-related BSI and LI rates were 0.57 and 1.72 per 1000 PICC-days respectively.

All infections but one LI occurred in the inpatient setting: thus the global infection rates among in- and outpatients were 3.3 and 0.53 per 1000 PICC-days respectively.

The BSI occurred 6, 9 and 39 days after insertion; the mean time to onset for the 9 LI was 17 days. With respect to microbiology, 2 BSI were caused by coagulase-negative staphylococci and 1 by Candida albicans. Seven additional cases were possible infections according to definition: 4 inpatients presented a possible BSI, and 3 outpatients a possible LI. Nineteen PICCs were removed as a matter of principle in febrile patients, but infection was not ultimately confirmed.

Other incidents

Forty-two pre-occlusive events occurred, with 55% of these occurring within the first week after insertion. Pre-occlusive events were managed by pulsed normal saline flush and/or heparin flush. Seventeen catheters (8.9%) were ultimately occluded, 16 of which had had a pre-occlusive event.

Twenty-five patients (15.3%) presented local or general inflammatory signs, with onsets occurring mostly within the first 4 weeks of PICC use. Interestingly, in 14 cases (56%), these precursor inflammatory signs were not followed by a positive diagnosis of catheter-related infection.

Univariate and multivariate analysis

The univariate comparison of baseline patient characteristics between inpatient and outpatient subgroups showed significant differences concerning the number of cases of solid tumor (12 vs 29% respectively) and cystic fibrosis (0 vs 12.9%). Concerning infusion therapy, hydration was more frequent among inpatients (60.9 vs 16.1%) and so was total parenteral nutrition (27.1 vs 6.5%). The small number of outpatients did not allow to statistically compare subgroups in terms of outcome and risk factors.

Multivariate logistic regression analysis performed on the whole study population showed that the occurrence of PICC-related occlusion was significantly associated with 2 risk factors: age > 65 years (OR = 4.19; 95% CI [1.1–15.81]), and presence of a pre-occlusive event the week before PICC removal (OR = 76.35; 95% CI [9.36–622.97]). Interestingly, catheter dwell time was not associated with any of the complications.