Effect of multimodal interventions on peripheral intravenous catheter–associated <i>Staphylococcus aureus</i> bacteremia and insertion rates: An interrupted time‐series analysis

Abstract

Hospital-acquired Staphylococcus aureus bacteremia (SAB) is a significant cause of morbidity and mortality worldwide.1 Serious clinical outcomes (e.g., septic shock, localized infection, and infective endocarditis) contribute to the health care burden at the individual, health service, and national level.2, 3 Peripheral intravenous catheter (PIVC)-associated SABs account for 23% to 36% of all hospital-acquired SAB events.4, 5 About half of the PIVCs in emergency departments (EDs) are never used for any clinical purpose.6 PIVC-associated SAB is more likely when initial insertion occurs in the ED.4 Human factor–designed multimodal interventions (MMIs) have reduced PIVC insertions among ED patients by 10% to 13%.7, 8 However, ED-based studies to date have included relatively small samples with short follow-up times. Therefore, the long-term effectiveness of reported ED-based intervention programs is unknown. Two MMIs (one ED-based and the other health service–wide) were implemented with the objective of reducing PIVC-associated SAB through insertion of “fewer” PIVCs initially in ED and subsequently “cleaner” PIVCs health service–wide. To our knowledge, no previous studies have examined the effect of an intervention that combines measures to reduce frequency of insertion and an emphasis on aseptic PIVC insertion techniques with subsequent incidence of PIVC-associated SABs. The aim of this study was to assess the effect of the two MMIs on the rates of PIVC-associated SAB rates and PIVC insertions in ED over a 9-year period. This observational study included data on adult patients (aged ≥ 18 years) who presented to Monash Health EDs from January 1, 2010, to December 31, 2018. Monash Health has three EDs and a combined annual census in excess of 230,000 patients. It is the largest public health service in Victoria, Australia. The three EDs have the same workforce, guidelines, and procedures. MMI 1 “Just Say No To The Just In Case Cannula” was an ED-based intervention developed using a human factor design aimed at changing the heuristics, environment, and equipment to create sustained change.7, 8 It focused on “fewer” PIVCs, although education on aseptic technique was included. MMI 2, a health service–based intervention, included the introduction of a new PIVC starter kit, procedure, and video designed to standardize insertion and improve aseptic technique. It focused on “cleaner” PIVCs with continuation of some of the preceding elements from MMI 1. The study had three phases: a 40-month preintervention period (January 1, 2010, to April 30, 2013), a 20-month postintervention MMI phase 1 (May 1, 2013, to December 31, 2014), and a 48-month postintervention MMI phase 2 (January 1, 2015, to December 31, 2018). The Monash Health Human Research Ethics Committee approved the study (RES-18-0000-527Q). Study outcomes were PIVC-associated SAB cases per 10,000 ED presentations (primary), PIVC-associated SAB per 10,000 PIVCs, and PIVC insertions per 10,000 ED presentations (secondary). Due to the constraints of the low numbers of SAB infections, we chose to examine monthly event rates. This along with hospital determined intervention dates meant sample size calculation was not feasible. Data on the number of ED presentations, number of ED-inserted PIVCs, and patients’ characteristics were collected from the ED electronic database (Symphony Version 4). Data on the number of PIVC-associated SAB infections and patient characteristics were collected from Monash Infection Prevention and Infectious Diseases database. The overall hospital-acquired SAB rate has been a nationally required key performance indicator for acute care hospitals in Australia since 2010.9 Specific PIVC-related SAB data are not required to be reported. Descriptive analyses were conducted to estimate the prevalence of PIVC and PIVC-associated SAB. An interrupted time-series analysis, using segmented regression to take into account correlations across time, was performed to assess the impact of the MMIs on rates of PIVC insertion and PIVC-associated SAB.10 The resultant time series were modeled using generalized linear regression. The count of PIVC-associated SABs was offset by the number of adult patients monthly to enable presentation of the PIVC-associated SAB rate per 10,000 adult presentations monthly, the number of PIVCs inserted per 10,000 patients monthly, and the number of PIVC-associated SAB per 10,000 PIVC monthly. Outliers were accounted for by including indicator variables in the model. The number of PIVCs, PIVC-associated SABs, and ED presentations were collapsed by month to estimate monthly rates of PIVC insertion and PIVC-associated SAB. To determine whether there was an immediate reduction in PIVC use with introduction of the MMIs, the estimated endpoint of the preintervention period was compared to the estimated rate at the beginning of the post–MMI 1 period. The analysis excluded data of May 2013 (the first month of MMI 1) to compensate for 1-month lag period and potential immediate effect of this period. Further, comparison of PIVC-associated SAB rates between the different periods (pre-MMI vs. both post-MMIs) was determined by estimating and comparing the midpoint rate of each period. Statistical significance was defined as a two-sided p-value of ≤0.05. All analyses were carried out using Stata version 15.1 (StataCorp, College Station, TX). Of 1,365,720 ED presentations during 108 months, 337,706 PIVC insertions and 46 PIVC-associated SABs (25 [preintervention], two [MMI 1], and 19 [MMI 2]) were recorded. Median (interquartile range) age of SAB patients was 73 (60–82) years, and 17 (37%) patients were female. In total, 30% of all ED presentations had a PIVC inserted. Seasonality was not apparent, and outliers were not excluded from the model. Table 1 presents the rates of PIVC-associated SABs and PIVC insertions and summarizes the midpoint rate ratios during pre- and post-MMI phases, including number of PIVC insertions and PIVC-associated SABs. The midpoint rate data indicate that there was a reduction in the PIVC-associated SAB per 10,000 ED presentations from pre-intervention to MMI phases 1 and 2. Similarly, a reduction in the PIVC-associated SAB per 10,000 PIVC insertions was also noted from preintervention to MMI phases 1 and 2. Also, the reduction in the PIVC insertions per 10,000 ED presentations was also significant from preintervention to MMI phases 1 and 2. The study demonstrated a reduction in the rate of PIVC-associated SAB with the implementation of an ED-based human factor–designed intervention. A second health service–wide intervention resulted in a concerning rise in PIVC-associated SAB rates. The rates of PIVC insertion were also observed to decrease dramatically with the introduction of the intervention. This reduction was maintained over the study period with a small upward trend toward the end of the study. The reduction in PIVC insertions in this study was similar to those reported in previous studies.6-8 Reduced SAB rates related to PIVC insertions in the ED reported in our study may be partly due to other factors such as the initiatives related to improved hand hygiene practices.9 Separating out the “fewer” and “cleaner” elements of our MMI is not straightforward. However, the dramatic reduction in SAB rates observed in early MMI phase 1 was likely due to the relatively large decrease in the number of PIVCs inserted. This is not surprising since although education on aseptic technique was included, the main thrust of the MMI 1 was on reducing PIVC insertions in the ED. The concerning rise in SAB rates during the MMI 2 could have been an unintended consequence of the introduction of a new PIVC insertion kit and PIVC insertion guidelines. Anecdotal evidence and local audits suggest that the kit itself may have been a factor—many staff reported use of the kits to be confusing making sterility difficult to maintain. There was a concern over the fenestrated dressing with local audits demonstrating failure to cover PIVC insertion site in more than half of cases (personal communication DEW). The PIVC kit and procedure were subsequently withdrawn from the health service in 2019. The inclusion of a large number of ED patients and PIVC insertions during 9 years is a key strength of the study. We are not aware of specific changes to relevant policies/definitions that would have introduced systematic bias during the study period. Australian national data suggest overall hospital-acquired SAB rates have been stable during the MMI phase 2.9 Our study limitations include: i) baseline incidence of a relatively small number of SABs; ii) any contribution of improved hand hygiene could not be controlled for; iii) denominator of the PIVC-associated SAB measure did not include PIVC insertion duration; and iv) compliance with the measures to prevent PIVC-associated SAB, site of PIVC insertion, and other time-varying confounders (e.g., staff turnover, changes in the population), including patient medical history, were not controlled for. The study did not include other bacteremia or change in antibiotic use or have a control group. The observational nature of the study only allows for hypothesis generation rather than confirmation. In conclusion, the ED- and health service–based interventions resulted in a significant sustained reduction in PIVC insertions among ED attendees. A reduction of PIVC-associated SAB rate was noted during MMI phase 1, but it increased by the end of the MMI phase 2. The authors have no potential conflicts to disclose. Chhavi R. Bhatt undertook literature search, managed the data, partly conducted the statistical analyses, and prepared the manuscript. Robert Meek, Rhonda L. Stuart, and Diana Egerton-Warburton designed the study. Catherine Martin conducted the statistical analyses. Zhengjie Lim and Suzanne Bumpstead contributed to data management and contributed to manuscript preparation. All authors contributed to manuscript preparation and reviewed and approved the final version of the manuscript.

Topics & Concepts