Design and setting
This was an observational cohort study in southeast Queensland, Australia, studying adult patients from both a large tertiary level emergency department (ED) and the local statutory ambulance service who had a PIVC inserted. Human research ethics committee approval, including waiver of consent, was obtained. The ED is a mixed adult and paediatric Level 1 trauma centre with over 110,000 ED presentations annually.
Participants
During 9 February − 18 March 2017 and 5 January − 4 February 2018, data on 1507 eligible ED presentations were prospectively collected by a research nurse(22). Inclusion criteria were patients who were adults and either had a PIVC inserted within the ED, or by a paramedic in the pre-hospital setting. Patients who had an ATS category of 1,(23) had a PIVC inserted in another hospital and were transferred between hospitals were excluded (Fig. 1).
Primary Outcome Measure
The primary outcome measure was the idle PIVC rate, defined as the proportion of PIVCs inserted but not used within 24 hours of placement. PIVC use was defined as administration of intravenous medication, fluid or contrast. PIVC use for a "flush dose" of normal saline to ensure cannula patency and/or drawing blood from a PIVC for pathology was not defined as PIVC use. Other outcomes of interest included comparison of PIVC insertion practices, including PIVC size, anatomical insertion site and inserter staff level, as well as identifying predictors of idle PIVC.
Data Sources
For participants who were prospectively identified as having a PIVC inserted by the ambulance service, the electronic ambulance report form (eARF), ED and inpatient electronic medical records (EMRs) were retrospectively interrogated for relevant variables. From these sources, data on PIVC use within 24 hours were extracted and included: patient demographics (e.g. gender, age), PIVC characteristics (e.g. gauge, insertion site) and inserting clinician characteristics (e.g. designation, years of service) were collected (Table 1). For participants who had a PIVC inserted within the ED the same features were extracted from an existing prospectively captured data set(22)(Table 2).
Table 1
Data captured from electronic ambulance report forms
Patient features
|
PIVC insertion features
|
Clinician features
|
Demographics, age, sex
Presenting complaint
Triage category
Distance from scene to hospital (kms)
Time from scene to hospital (mins)
Time on scene (mins)
Time of day
|
Anatomical insertion site
Cannula size/gauge
Number of insertion attempts
Insertion success (yes/no)
Complications#
PIVC use (yes/no and if yes, what was it used for?)
|
Years of service
Clinicians scope of practice*
Clinician medal number
|
*Clinicians scope of practice refers to the paramedic’s clinical level, paramedics included in the study where student paramedics, advanced care paramedics, critical care paramedics and High Acuity Response (HARU) paramedics. #Arterial puncture/fluid extravasation/haematoma or haemorrhage/ venous air embolus
Table 2.
Data captured from ED medical records
PIVC insertion features
|
Patient features
|
Clinician features
|
Anatomical insertion site
|
Demographics, age, sex
|
Doctor
|
Cannula size/gauge
|
Presenting complaint
|
Nurse
|
Number of insertion attempts
|
Triage category
|
|
Insertion success (yes/no),
|
Time of day
|
|
Complications
|
|
|
PIVC use (yes/no and if yes, what used for)
|
|
|
ED Emergency Department; PIVC Peripheral intravenous catheter
Sample Size And Data Analysis
We calculated the sample size necessary to identify a clinically significant difference in idle PIVC rates between pre-hospital and ED settings. Based on existing literature showing the idle PIVC rate to be around 30%, we considered a difference of +/- 10% to be clinically significant(20). In order to be powered at 80% with an alpha of 0.05, we required data on 390 patients with PIVC insertions in both pre-hospital and ED groups.
Data analysis involved both univariate and multivariate methods, using SPSS v26.0. Simple descriptive statistics (proportions, frequencies, and means of central tendency) were calculated. A 95% confidence interval around the proportion of "idle" PIVC was calculated using statistical software (Open Epi), and the Wilson score method. Comparisons of proportion of idle PIVCs, and insertion practices between pre-hospital and ED practice employed chi-square tests. Univariate associations of potential predictors of idle PIVC were examined in the pre-hospital setting, and across the emergency setting. Crude odds ratios (OR) and 95% confidence intervals (CI) were calculated.
Two logistic regression models using a backwards conditional approach were developed with the dependent variable of idle PIVC. The first assessed predictors of idle PIVC for pre-hospital PIVCs only, with independent variables of age, sex, gauge, insertion site, cannulator characteristics, time (minutes) at site, time of day, distance from site to hospital, and time from scene to hospital (Table 1). The second assessed predictors of idle PIVC for all PIVCs placed in the emergency setting (pre-hospital and ED combined), with a more limited set of potential predictors that were applicable to both settings (age, sex, cannulator experience, presentation time of day, gauge, anatomical insertion site, and insertion setting). The pre-hospital specific variables, such as time on scene were not included. The ATS for the pre-hospital group was determined from the ED ATS. The number of records with missing data was presented for each variable. In all inferential univariate comparisons, the missing data category was excluded from the chi-square tests. In the multivariable modelling, records missing potential predictors were automatically excluded case-wise from the initial backwards conditional analysis. Variables that remained significant in the backwards model were then forced into a new model with only those variables. Adjusted odds ration (OR) and 95% confidence interval (CI) were calculated for this final model.