Design and participants

A single-center prospective cohort observational study was carried out between January 1, 2013, and November 30, 2017, in an acute-care 600-bed teaching hospital (reference population, 250,000) located in Seville, Spain. The primary objective was to determine the incidence, time to appearance, and risk factors of major PICC-related infectious and thrombotic complications. Secondary objectives were the rate and incidence density of complications, causative pathogens, and crude and attributable mortality. This study was approved by the institutional review board, and written informed consent was obtained from all participants.

Consecutive adult inpatients and outpatients aged ≥18 years who had a PICC inserted between January 1, 2013, and December 31, 2016, and for whom follow-up was conducted by November 30, 2017, were eligible. Only the first PICC inserted was considered for each patient. Exclusion criteria for PICC insertion were life expectancy <7 days, anatomical abnormalities of the venous system of the upper extremities, bilateral upper-extremity deep vein thrombosis (UEDVT), lymphedema, or skin infection at the site of insertion. The following exclusion criteria were applied: patients aged <18 years, PICCs not placed in the peripheral veins of the upper extremity, and refusal to provide written informed consent to participate.

Study procedures

All PICCs were inserted as an elective procedure using a PowerPICC Catheters (Bard Access Systems, Salt Lake City, UT). The PICC catheterization inserted by an experienced intensivist was performed by ultrasound-guided puncture of the deep veins in the upper mid-arm using a standard 10 MHz linear ultrasound probe, maximal barrier, and strict sterile precautions.^{Reference O’Grady, Alexander and Burns13,Reference Tian, Zhu, Qi, Guo and Xu18,Reference Pittiruti, Hamilton, Biffi, MacFie and Pertkiewicz19} The position of the catheter tip was assessed during implantation using fluoroscopy or the Sherlock 3CG Tip Confirmation System (Becton Dickinson, Franklin Lakes, NJ). PICCs were secured in suture-free manner with the StatLock PICC Plus Stabilization Device (Becton Dickinson). Routine prophylaxis of UEDVT according to clinical practice guidelines was not performed.^{Reference Lyman, Bohlke and Falanga20} A standardized protocol of PICC care and maintenance, with aseptic techniques used at all times, was applied.^{Reference O’Grady, Alexander and Burns13} After catheter insertion, patients were followed during their hospital stay and thereafter by telephone contact at 1, 3, and 6 months after PICC placement using an ad hoc questionnaire. After 6 months, the patient’s family physician contacted us to report any complication, and we also made frequent nonprotocolized telephone calls until the catheter was removed or until November 30, 2017, if the catheter was still in place at the end of the study. Catheters with confirmed or suspected CABSI were considered for antibiotic lock. Antibiotic lock was carried out with vancomycin 1–2 mg/dL for 14 days, and alcohol lock was performed with ethanol (70%) for 24 hours.^{Reference Bestul and Vandenbussche21}

Definitions

CABSI was defined as either CRBSI or PB. CRBSI was diagnosed in a patient with PICC and clinical manifestations of infection (fever >38°C or <35°C, or hypotension) with at least 1 positive blood culture (bacteremia or fungemia) and isolation of the same microorganism in the peripheral blood and the catheter tip (semiquantitative culture, ≥15 colony-forming units [CFU]/mL), or a differential time to positivity of ≥120 min. PB was diagnosed in a patient with PICC and clinical manifestations of infection (fever >38°C or <35°C, or hypotension) with at least 1 positive blood culture (bacteremia or fungemia) and negative catheter culture (or not performed)^{Reference Mermel, Allon and Bouza12,Reference Eggimann, Pagani and Dupuis-Lozeron14} without signs or symptoms of another documented source of infection. For common skin contaminant 2 or more positive blood cultures with the same microorganism on separate occasions were required.^{Reference Mermel, Allon and Bouza22} Symptomatic UEDVT (pain, swelling, redness, or alteration of the venous flow) with suggestive signs of partial or complete obstruction of the vein confirmed by Doppler ultrasound studies were only considered in the group of thrombotic complications. Catheter dwell time was the number of days between catheter insertion and removal. Incidence density was defined as the number of episodes (or events) per 1,000 catheter days. Hospitalized patients were inpatients referred for PICC placement. Outpatients were those referred to the hospital for PICC placement who returned home after the insertion procedure. Mortality attributable to CABSI was defined in the presence of documented bacteremia before the death of the patient, assuming that the progression of infection was the cause of death.^{Reference Ziegler, Pellegrini and Safdar23}

Data collection

The following demographic data were recorded: type of patient (hospital ward or outpatient setting referral for PICC insertion); underlying disease; indication of PICC; laterality of PICC placement; vein accessed; catheter-related characteristics; tip position; reasons for PICC removal related to the catheter (eg, local infection, obstruction, phlebitis, CRBSI, and UEDTV); reasons for PICC removal unrelated to the catheter (eg, end of intravenous therapy, accidental withdrawal, intolerance, not confirmed suspicion of CRBSI, and death); complications during insertion; complications during use (eg, CRBSI, PB, UEDTV, local infection, phlebitis, catheter colonization); and mortality.

Statistical analysis

In patients with 2 simultaneous major complications, infection (CRBSI) and thrombosis (UEDVT) were considered independent events and were analyzed separately. When PB was followed by a CRBSI, only PB was considered. Categorical variables are expressed as frequencies and percentages, and continuous variables are expressed as mean and standard deviation (SD) or as median and interquartile range (IQR, ie, 25th–75th percentile). Categorical variables were compared using the χ^{Reference Johansson, Hammarskjöld, Lundberg and Arnlind2} test or the Fisher exact test, and continuous variables were compared using the Student t test or the Wilcoxon rank-sum test. Logistic regression analysis was used to assess risk factors for outcome. Those variables with P < .10 in univariate testing were entered into the model. Catheter lifetime was defined as the number of days from insertion to catheter failure. Failure included local infection, obstruction, deep venous thrombosis, suspicion of CRBSI, phlebitis and intolerance. Otherwise, the observation was considered right censored. Survival curves according to the levels of several factors were estimated using the Kaplan-Meier method and were compared using the log-rank test. To obtain the proportional hazard model by the catheter lifetime, a selection of variables based on the Akaike information criterion (AIC) was carried out. Probability density functions for the catheter lifetime in the groups CRBSI and PB (considering catheter removal due to suspected CRBSI as failure) were estimated using the local likelihood approach to density estimation with censored data.^{Reference Loader24} The study period spanned January 2013 to December 2016. For each of the 16 quarters of the study period, the number of patients in whom the PICC was inserted, the total number of days of exposure (D _t) and finally, the number of patients who presented the study event (ie, infectious or thrombotic complications) during the follow-up period (N _t) were calculated. To evaluate the evolution of the incidence, it was assumed that the random variable N _t followed a Poisson distribution of mean μ _t:

$${\rm{log}}{\mu _t} = {\rm{log}}{D_t} + \alpha + s\left( t \right)$$

Here, s(t) is a smooth function of time, which was nonparametrically estimated by means of a cubic spline. The goodness of fit was evaluated using the overdispersion coefficient, defined as the ratio of the deviance and its degree of freedom. Notably, for each quarter (t = 1, …,16), 1,000 × μ_t/D _t corresponded to the expected number of events per 1,000 catheter days. Also, a subanalysis of the distribution of variables of interest in hospitalized versus outpatients was performed. Statistical significance was set at P < .05. Data were analyzed using the R package software, version 3.6.1.²⁵