Comparison of studies with strongest study design
| Bion 2013 | Marsteller 2012 | |
|---|---|---|
| Study design | Non-randomised stepped wedge | Cluster randomised clinical trial |
| Number of ICUs | 215 ICUs, 196 adult and 9 paediatric | 45 ICUs |
| Period of intervention and measurements | September 2009–July 2010 depending on cluster | Baseline: 2006 Intervention: March 2007–September 2008 Control: October 2007–September 2008 |
| Bundle elements | Hand hygiene Full barrier precautions Chlorhexidine skin antisepsis Avoid femoral vein Daily check unnecessary catheters | Hand hygiene Full barrier precautions Chlorhexidine antisepsis Avoiding femoral vein Removal unnecessary lines |
| Co-interventions and heterogeneity | Dressing maintenance Disinfection connector Aseptic maintenance Heterogeneity reported, wide variation in generic infection control practices and in blood culture sampling rates, which may affect reported infection rates | Dressing change checklist CUSP: engage staff, educate staff, execute activities and practices, evaluate process and outcomes regularly, expand program to other units, plan intervention to endure as regular practice Previous exposure to CUSP and safety culture did not alter intervention effect. However, West ICUs, larger ICUs and ICUs with lower baseline rates were more likely to have produced exact zero infections |
| Compliance with bundle | Reported as limitation of study that adoption of the intervention and compliance with best practices was not measured | Assessed by an observer using a checklist to ensure that line was inserted following evidence-based practices |
| Outcome measure | CLABSI or central line-related blood stream infection, following definitions of CDC NNISS and Hospital in Europe Link for Infection Control through Surveillance programme | CLABSI following CDC definition: laboratory confirmed bloodstream infection in patient who had central line in place within the 48 h period before the infection developed |
| Exclusions in outcome measure | Distinction between ICU-acquired and pre-ICU acquired. Ratio between ICU-acquired and all central line infections remained stable, suggesting a similar declining trend in both pre-ICU and ICU-acquired central line infections | Infections present or incubating within 72 h of ICU admission were excluded. Not reported whether these have decreased as well |
| Difference in outcome at baseline | Median CLABSI rate 2.56 (intervention) vs 1.78 (control) p=0.52 Mean: 4.48 vs 2.71 | |
| Effect of intervention | Decline from 4.4/1000 to 1.7/1000 Each successive cluster joining the intervention, had an entry-level central line infection rate similar to post-intervention level of preceding cluster. Thus the reduction in central line infection rate is likely to be due to a general secular trend | Decline in intervention group from mean 4.5 to 1.3 per 1000; decline in control group from 2.7 to 2.2 per 1000 (70% vs 21% reduction). After crossover, at 19-month follow-up: 0.85/1000 in intervention group versus 0.83/1000 in control group (81% vs 69 reduction compared with baseline) |
CDC, Centers for Disease Control and Prevention; CLABSI, central line-associated blood stream infection; CUSP, Comprehensive Unit-based Safety Program; ICU, intensive care unit; NNISS, National Nosocomial Infections Surveillance.