Central line-associated bacteraemia (CLAB) in intensive care units (ICUs) is associated with a mortality rate of between 10% and 50% and a significant burden of morbidity.1-3 The cost is estimated at around NZ$20,000 per CLAB infection.4
CLAB rates can be reduced effectively by aseptic insertion, use of an insertion checklist, monitoring of line days, standardised management and early catheter removal.5-7 Quality improvement collaboratives to implement these interventions in the US and New South Wales, Australia, achieved reductions of CLAB incidence between 60% and 74%.6,8,9
In 2008, Counties Manukau Health used Institute for Healthcare Improvement (IHI) insertion and maintenance care bundles in a local improvement programme in the Critical Care Complex at Middlemore Hospital, Auckland. The incidence of CLAB dropped from 6.6 cases per 1,000 line days to 0.9 cases.5 This compares to a drop in the mean incidence of CLAB from 3.73 to 0.97 per 1,000 line days in Rhode Island, from 1.5 to 0.6 in Hawaii, and from 3.0 to 1.2 in New South Wales.6,8,9
This successful local programme, combined with the weight of research evidence and an encouraging policy environment, created a powerful case for action across New Zealand ICUs.
The Target CLAB Zero campaign commenced in October 2011. The campaign was a national collaborative of New Zealand ICUs. Ko Awatea, the health system innovation and improvement centre at Counties Manukau Health, led the campaign and the Health Quality & Safety Commission (HQSC) funded it. The campaign’s objective was to reduce the rate of CLAB in New Zealand ICUs to less than one incident per 1,000 line days by the end of March 2013, and to establish a national measurement system for CLAB.
We report on the results of the collaborative and the lessons learnt from its development and implementation.
At the time of the study, healthcare in New Zealand was configured into 20 district health boards (DHB) covering four regions. There were 24 adult and one paediatric ICUs, which collectively admitted around 19,000 patients per year.5
Intensive care services are configured as ICUs, high dependency units (HDUs) or mixed ICU/HDU units. Considerable variation exists in size, access to specialist services, and staff and patient mix.
Patients in smaller ICUs requiring specialist care are often transferred to hospitals with larger units. The patient mix is therefore broader, and acuity higher, in the larger units. As a result, CLAB rates at the beginning of the collaborative ranged from zero to six incidents per month, and were higher in the larger units.
We identified key challenges. Firstly, because of New Zealand’s geography and varying population density, some ICUs are geographically isolated and there was considerable variation in the experience of CLAB. Furthermore, there was no mechanism for shared learning and improvement. Additionally, New Zealand lacked a national surveillance system for CLAB to establish an accurate baseline incidence rate. Finally, definitions of CLAB and data collected were inconsistent across ICUs.
We used two methods to obtain a baseline:
- We extrapolated from data collected from a local improvement programme in the Auckland Region. The programme covered four of the largest ICUs in New Zealand, which collectively serve a third of the country’s population. The region had three to six CLAB infections per month, giving a regional annual extrapolated rate of 48 to 72 CLAB infections.
- We also conducted a 12-month retrospective audit of clinical notes from all patients admitted to all New Zealand ICUs who had a central line inserted from November 2010 to December 2011 against the Centers for Disease Control and Prevention (CDC) definition of central line-associated bloodstream infection (Table 1). The audit was completed at each site by a person who understood the criteria and had training in the audit process. The person nominated was the project leader, a registered nurse or an infection prevention control nurse.
Faculty and stakeholders
Ko Awatea assembled a national project team and a steering group to lead the collaborative. The project team comprised a national project manager, a clinical leader and an improvement advisor. The steering group included clinicians, experts in improvement science from Ko Awatea and IHI, and representation from the HQSC. An expert group and a measurement group, comprising intensivists, microbiologists and infectious disease specialists, provided overall clinical leadership and guidance to support the project team and steering group. Each participating site had a clinical leader (ICU physician, anaesthetist or clinical head) and a project leader. In addition, there were four regional clinical leaders (ICU clinical director or consultant).
We secured the support of key DHB stakeholders (chief executive, operating and medical officers; directors of nursing and ICU management) with a targeted invitation letter and background document. A follow-up letter acknowledged those who had committed to take part and encouraged those yet to respond to do so. All ICUs agreed to participate.
IHI bundles of care for the insertion and maintenance of central lines used in the 2008 CLAB reduction programme at Middlemore Hospital, Auckland, were adopted for Target CLAB Zero (Table 1).10 Due to the wide variation among ICUs, teams at each site adapted implementation of the bundles to suit their local context.
Teams were trained in IHI Breakthrough Series Collaborative Model (BTS) methodology to support adoption of the bundles.11 The clinical leader and project leader from each site were sponsored to attend two 3-day training programmes in improvement science and BTS methodology.
We structured the BTS model as three national learning sessions interspersed with action periods. The first learning session, in November 2011, covered how to implement the insertion and maintenance bundles, collect data and interpret operational definitions. A guide developed from the 2008 programme at Middlemore Hospital was provided. The second learning session, in November 2012, focused on testing the validity and reliability of data and overcoming barriers to collection of bundle compliance data. The third, in March 2013, addressed how to sustain the improvements achieved. Clinical leaders were sponsored to attend sessions.
Regional meetings and WebEx online conferencing supplemented learning sessions. These provided additional coaching in improvement methods and measurement, and support from peers and the project team.
Teams applied learning during the action periods, using the Model for Improvement.12 This model required teams to set specific aims and measures, then develop and test change ideas using plan, do, study, act cycles.
A communication plan supported the intervention. A website enabled participants to access resources, report progress and interact through a discussion forum. We added a dedicated blog to the Ko Awatea website and produced a quarterly newsletter. Each ICU received promotional resources and displayed a progress board.
We collected two outcome measures—incidence of CLAB (O1) and a count of central venous line days (O2). We used these to calculate our primary outcome measure, the national rate of CLAB per 1,000 line days (O3) (Table 1).
We also collected two composite process measures—compliance with the insertion bundle (P1) and compliance with the maintenance bundle (P2). These comprised three and four process measures respectively. The rate of compliance with the insertion and maintenance bundles was calculated by dividing the number of patients who had all elements of a bundle correctly executed by the number of checklists completed. Patients admitted to ICUs with a central line from operating theatres were included in the data, as staff were trained in the use of the insertion and maintenance bundles and were able to check insertion compliance. Operational definitions were agreed for all measures (Table 1).
To judge the degree of engagement with the collaborative, we measured attendance at learning sessions, regional meetings and WebEx sessions. We also monitored monthly data submissions.
Data collection and analysis
A national web-based database was established for collecting data. Each ICU entered data monthly from January 2012 to March 2013. Following March 2013, the HQSC took over the data collection and reporting. Data were checked for completeness 10 working days after the end of the month. Follow-up contact was made with teams who had incomplete data and/or data anomalies. Data were then extracted from the database into an Excel spreadsheet.
Monthly progress reports were fed back to teams. Project managers and clinical leads reported monthly data to DHB stakeholders from January 2012.
Estimates of effect were determined using random and non-random variation flags on run charts.
The collaborative ran from October 2011 to March 2013. Data are reported on 23 units—one unit had no line days for the duration of the campaign, and one unit implemented the insertion and maintenance bundles, but had not determined a method for data collection on the three key measures.
The estimated baseline CLAB rate for the 12 months prior to the project was 3.32 per 1,000 line days. The period January to March 2012 provided data gathered against a defined, consistent standard (Table 1). During this period, CLAB rates were between 1.6 and 2.7 per 1,000 line days. By April 2012, all ICUs had implemented both the insertion and maintenance bundles. The CLAB rate fell immediately to 0.28 per 1,000 line days and was sustained throughout the collaborative period (Figure 1). This represented a relative reduction of 90% from the baseline of 3.32.
Table 1: Operational definitions for outcome and process measures
O1 (CLAB): An infection associated with central venous catheters which meets the CDC criteria and is independently confirmed by a qualified third party (microbiologist, intensive care consultant).
CDC technical description:
i. Presence of a recognised pathogen cultured from one or more blood cultures and organism cultured from blood not related to infection at another site.
ii. Fever (>38°C), chills, or hypotension and organism cultured from blood is not related to infection at another site and presence of at least one of the following:
O2 (Count of central venous line days): A count of the number of central venous catheters inserted in a great vessel (eg, superior vena cava).
O3 (Rate of CLAB/1,000 line days): The rate of CLAB per 1,000 line days, calculated by dividing the number of confirmed cases of CLAB (numerator) by the number of central line days per month (denominator) and multiplying the result by 1,000.
Process measures for insertion and management of the line
P1 (Compliance with insertion bundle):
P2 (Compliance with maintenance bundle):
Figure 1: CLAB rate per 1,000 line days (all ICUs)
- End of the implementation period
- Learning session 2
- Regional workshops
- Christmas period
- Learning session 3
Between January 2012 and June 2012, New Zealand experienced an average of 4.6 days between CLAB infections. This rose to 13 days after June 2012 (Figure 2).
Figure 2: Days between CLAB infections
The first 3 months of the collaborative (January–March 2012) were spent establishing consensus on how to measure compliance with the insertion and maintenance bundles. By April 2012, all but one unit had set up measurement systems and were reporting compliance rates. By the final 3-month period, January to March 2013, average compliance with the insertion bundle was 80%, and with the maintenance bundle 75%.
Participation in the collaborative was high. Over 90% of those invited attended all three national learning sessions and the meetings held in each of the four regions (Table 2). All sites submitted data and posted reports each month.
Table 2: Characteristics of New Zealand ICUs, bed numbers, participation, and compliance with reporting and quarterly line days
CLAB rates were significantly reduced, and participation in the collaborative and process compliance was high. This improvement in the CLAB rate and engagement with the collaborative occurred despite differences in ICU size, staff and patient mix, population and geographic location among participating DHBs.
Our results reinforce existing studies that demonstrate the effectiveness of hand hygiene, aseptic skin preparation, barrier precautions and removing unnecessary catheters for reducing CLAB.6,7,8 Burrell et al. report a drop in the rate of CLAB from 3.0 to 1.2 in a quality improvement collaborative across 37 ICUs in New South Wales (NSW) using comparable bundles of care.6 An initiative covering 45 states in the US, On the Cusp: Stop BSI, reported a baseline rate of 1.87 CLAB infections per 1,000 central line days for units that began participating in the project in 2009 and 2010. After 10–12 months of participation, this rate decreased to 1.25.13 Target CLAB Zero covered all New Zealand ICUs and achieved a decrease in the average rate of CLAB from 3.32 to 0.28. However, the NSW and US projects were considerably larger in scale than the New Zealand project. New Zealand’s small size, and centralised public health service, made a national approach feasible and contributed to the high participation throughout the collaborative.
Collaborative methodology was used in New South Wales, Rhode Island and Hawaii to successfully implement evidence-based best practices for CLAB reduction.6,8,9 To our knowledge, Target CLAB Zero is the first nationwide improvement collaborative in the New Zealand healthcare context.
Carter et al. identify accounting for context, setting realistic goals, providing sufficient time and resources, and careful management of collaboratives as factors that mitigate over-competitiveness, inertia, conflicting organisational pressures and free-riding in collective action.14 We found that the flexibility of the BTS methodology gave frontline staff the tools, knowledge and direction to adapt the change package according to local context. This enabled Target CLAB Zero to work effectively across ICUs of varied size, patient mix and degree of specialism. The national project team avoided being prescriptive about how the bundles were implemented. Instead, the focus was on agreed definitions and guiding principles, and on building improvement capability and capacity through a learning network. The emphasis was on measurement for improvement rather than performance imperatives. Competitiveness was apparent without resultant hostility or inertia.
Our results are likely to be accompanied by a reduced mortality rate, shorter lengths of stay in hospital, and a reduction in the economic burden of CLAB. Although our data finish in March 2013, when HQSC took over data collection and reporting, data available from HQSC show that the improvements have been sustained.15
The cost per CLAB infection is approximately NZ$20,000.4 The implied monthly incidence of CLAB at the baseline level was up to 8.9 per month (3.3/1,000 * 2,700). Extrapolating to the post set-up period (April 2012 to March 2013), we would have expected up to 105 incidents of CLAB had no corrective action been taken. There were 15 incidents of CLAB during this period. This represents potential savings of $1.8 million.
Other benefits of the campaign were:
- development and agreement on process flow for obtaining blood cultures
- an agreed national approach to determining a CLAB from a positive blood stream infection
- establishment and development of four regional reusable networks
- increase in staff awareness and commitment to reducing CLAB
- increased health sector capacity and capability in using IHI improvement methodology
- reduction in patient harm.
Target CLAB Zero achieved high national participation. We identified six key facilitators of participation:
- The existence of strong international evidence for the effectiveness of the intervention.6-9,16-18
- The existence of a proven example of international evidence being successfully applied in the New Zealand context.5
- The flexibility of BTS methodology.
- The peer-led nature of the campaign. The ability to attract influential clinicians to leading roles was important. Clinical respect was leveraged within and across DHBs.
- The involvement of HQSC. HQSC sponsored attendance at learning sessions and lent weight to the campaign as a national initiative.
- An effective communication plan. Targeted messages secured engagement from key stakeholders. The communication plan, and the use of regional and virtual meetings, also enabled participants to connect despite the geographical isolation of some units.
The lack of an existing CLAB surveillance system using standard definitions of CLAB presented challenges in establishing an accurate baseline rate. The retrospective audit conducted to establish the baseline depended upon information in clinical notes that were not compiled according to a consistent standard. Counting mechanisms for line days were still under development at the time the baseline data were collected. Furthermore, the definition of a CLAB infection used during the collaborative required two positive blood cultures, but this standard was not in place during the baseline period. The baseline may therefore under- or over-represent the national CLAB rate at the beginning of the collaborative.
The period from January to March 2012 is probably more indicative of the true national baseline, even though some ICUs began implementing the intervention from January 2012. It is also possible that the Northern regional ICUs “drove” the bulk of the CLAB incidence. Therefore the early focus by clinical staff in this region was already resulting in service improvements that were quickly realised and sustained by the collaborative process and the national monthly surveillance system.
Target CLAB Zero exceeded its aim of reducing the rate of CLAB in New Zealand ICUs to less than one CLAB incident per 1,000 line days, and established a national measurement system for CLAB. The project demonstrates that national improvement collaboratives are feasible in New Zealand. Barriers of geographical isolation and variation in the size and characteristics of participants can be overcome by effective communication and the use of an improvement methodology that accommodates differences in local context. The lessons learned from the Target CLAB Zero experience would inform successful development and implementation of improvement collaboratives aimed at addressing other deficiencies in quality of health care, where a strong evidence base exists to support definable best practice. The Enhanced Recovery after Surgery (ERAS) collaborative, which was launched in 18 DHBs in November 2013 to improve care for orthopaedics patients, provides one such example.19,20