Nasal Iodophor Antiseptic vs Nasal Mupirocin Antibiotic in the Setting of Chlorhexidine Bathing to Prevent Infections in Adult ICUs: A Randomized Clinical Trial

Abstract

Importance: Universal nasal mupirocin plus chlorhexidine gluconate (CHG) bathing in intensive care units (ICUs) prevents methicillin-resistant Staphylococcus aureus (MRSA) infections and all-cause bloodstream infections. Antibiotic resistance to mupirocin has raised questions about whether an antiseptic could be advantageous for ICU decolonization.

Objective: To compare the effectiveness of iodophor vs mupirocin for universal ICU nasal decolonization in combination with CHG bathing.

Design, setting, and participants: Two-group noninferiority, pragmatic, cluster-randomized trial conducted in US community hospitals, all of which used mupirocin-CHG for universal decolonization in ICUs at baseline. Adult ICU patients in 137 randomized hospitals during baseline (May 1, 2015-April 30, 2017) and intervention (November 1, 2017-April 30, 2019) were included.

Intervention: Universal decolonization involving switching to iodophor-CHG (intervention) or continuing mupirocin-CHG (baseline).

Main outcomes and measures: ICU-attributable S aureus clinical cultures (primary outcome), MRSA clinical cultures, and all-cause bloodstream infections were evaluated using proportional hazard models to assess differences from baseline to intervention periods between the strategies. Results were also compared with a 2009-2011 trial of mupirocin-CHG vs no decolonization in the same hospital network. The prespecified noninferiority margin for the primary outcome was 10%.

Results: Among the 801 668 admissions in 233 ICUs, the participants' mean (SD) age was 63.4 (17.2) years, 46.3% were female, and the mean (SD) ICU length of stay was 4.8 (4.7) days. Hazard ratios (HRs) for S aureus clinical isolates in the intervention vs baseline periods were 1.17 for iodophor-CHG (raw rate: 5.0 vs 4.3/1000 ICU-attributable days) and 0.99 for mupirocin-CHG (raw rate: 4.1 vs 4.0/1000 ICU-attributable days) (HR difference in differences significantly lower by 18.4% [95% CI, 10.7%-26.6%] for mupirocin-CHG, P < .001). For MRSA clinical cultures, HRs were 1.13 for iodophor-CHG (raw rate: 2.3 vs 2.1/1000 ICU-attributable days) and 0.99 for mupirocin-CHG (raw rate: 2.0 vs 2.0/1000 ICU-attributable days) (HR difference in differences significantly lower by 14.1% [95% CI, 3.7%-25.5%] for mupirocin-CHG, P = .007). For all-pathogen bloodstream infections, HRs were 1.00 (2.7 vs 2.7/1000) for iodophor-CHG and 1.01 (2.6 vs 2.6/1000) for mupirocin-CHG (nonsignificant HR difference in differences, -0.9% [95% CI, -9.0% to 8.0%]; P = .84). Compared with the 2009-2011 trial, the 30-day relative reduction in hazards in the mupirocin-CHG group relative to no decolonization (2009-2011 trial) were as follows: S aureus clinical cultures (current trial: 48.1% [95% CI, 35.6%-60.1%]; 2009-2011 trial: 58.8% [95% CI, 47.5%-70.7%]) and bloodstream infection rates (current trial: 70.4% [95% CI, 62.9%-77.8%]; 2009-2011 trial: 60.1% [95% CI, 49.1%-70.7%]).

Conclusions and relevance: Nasal iodophor antiseptic did not meet criteria to be considered noninferior to nasal mupirocin antibiotic for the outcome of S aureus clinical cultures in adult ICU patients in the context of daily CHG bathing. In addition, the results were consistent with nasal iodophor being inferior to nasal mupirocin.

Trial registration: ClinicalTrials.gov Identifier: NCT03140423.

Figure 1.. Mupirocin-Iodophor Swap Out Trial Flow Diagram

Aggregated baseline hospital data were used to establish similar hospital pairs based on key variables using the Goldilocks approach. Members of each pair were randomly assigned, 1 to each group. ICU indicates intensive care unit. ^aAs-randomized analyses used health system data from hospitals regardless of withdrawal. ^bPatients were included in the analysis until the time of withdrawal for all 3 hospitals that withdrew during the intervention period. The 1 hospital (2 ICUs) that withdrew from the iodophor-chlorhexidine group midintervention contributed 3737 admissions and 18 062 ICU-attributable days during baseline and 786 admissions and 3679 ICU-attributable days during intervention. The 2 hospitals (2 ICUs) that withdrew from the mupirocin-chlorhexidine group shortly after the intervention period began contributed 3010 admissions and 11 976 ICU-attributable days during baseline and 141 admissions and 480 ICU-attributable days during intervention. ^cAs-treated analyses used data up until the time of withdrawal. The as-treated analysis also removed admissions where the patient did not receive at least 2 doses of the assigned nasal product.

Figure 2.. As-Randomized Unadjusted Clustered Comparison of Iodophor-Chlorhexidine Gluconate (CHG) vs Mupirocin-CHG Comparing Differences in Hazard Ratios of Trial Outcomes in the Intervention to Baseline Period

Graphic shows impact of trial interventions on trial outcomes attributable to participating intensive care units (ICUs). Group-specific hazard ratios (and their 95% confidence limits) comparing outcomes in the intervention and baseline periods are derived from proportional hazard models (as-randomized, unadjusted) accounting for clustering by hospital and person using parameter estimates and their variance matrix. These are shown for Staphylococcus aureus clinical cultures (A), methicillin-resistant S aureus (MRSA) clinical cultures (B), and all-cause bloodstream infections (C) based on 7696, 3718, and 4624 respective events in the iodophor-CHG group; 6277, 3206, and 4096 respective events in the mupirocin-CHG group; 1 678 331, 1 714 574, and 1 710 232 respective attributable ICU days in the iodophor-CHG group; and 1 549 099, 1 574 114, and 1 567 355 respective attributable ICU days in the mupirocin-CHG group. P values for the difference in within-group hazard ratios are derived from these same models. Bubble plots of the hospital-specific hazard ratios (accounting for clustering by patient) are shown adjacent to group-specific hazard ratios and confidence intervals. The size of the bubble shows the relative number of admissions contributing data to the trial. For readability, hazard ratios greater than 4 or less than 0.25 are not depicted: 2 mupirocin-CHG (5.7 and 10.6) and 1 iodophor-CHG (5.7) for MRSA and 2 mupirocin-CHG (0.24 and 11.4) for all-cause bloodstream infections. For each outcome, the panel displays a horizontal dotted line indicating the 10% absolute noninferiority margin from the overall hazard ratio of the mupirocin-CHG group.

Figure 3.. Cumulative Hazard of Trial Outcomes by Day of Patient’s Intensive Care Unit (ICU) Stay

Graphic showing the durability of clinical effects for trial outcomes due to universal ICU decolonization over a span of more than 7 years in the same hospital system (post hoc analysis). Group-specific cumulative hazards for the first 30 days of ICU-attributable events are shown for ICU-attributable Staphylococcus aureus clinical cultures (A), methicillin-resistant S aureus (MRSA) clinical cultures (B), and all-cause bloodstream infections (C). Day 3 is noted on the x-axis because cases attributed to the ICU are defined by the Centers for Diseases Control and Prevention as occurring from ICU day 3 through 2 days after ICU discharge. Thus, the first 30 days of ICU-attributable events occur between ICU day 3 through 33. At that point, the x-axis is truncated because data after that period are sparse. Full figure is found in eFigure 2 in Supplement 2 with associated at-risk days in eTable 5 in Supplement 2. Note the different y-axis scales across the panels. Comparisons are made between the mupirocin-CHG group of the current trial (2017-2019) to data from the mupirocin-CHG group of a prior trial in the same hospital system (2010-2011). Effect of mupirocin-CHG (2017-2019) vs mupirocin-CHG (2010-2011) was unchanged for S aureus clinical cultures (P = .50), MRSA clinical cultures (P = .09), and all-cause bloodstream infections (P = .41). The cumulative hazard of all 3 outcomes with iodophor-CHG (2017-2019) in the current trial was superior to the hazard in the prior trial during the baseline period with no decolonization (2009): S aureus clinical cultures (P = .001), MRSA clinical cultures (P < .001), and all-cause bloodstream infections (P < .001).