ADAPT-Sepsis: PCT or CRP Guided Antibiotic Duration
ADAPT-Sepsis: PCT or CRP-guided Antibiotic Duration for Patients with suspected Sepsis
Dark et al. JAMA Dec 9 2024. doi:10.1001/jama.2024.26458
Clinical Question
- For critically ill adults with suspected sepsis, does a procalcitonin (PCT) guided or C-reactive protein (CRP) guided antibiotic discontinuation prompt, compared with usual care, reduce the total duration of antibiotics without increasing all-cause 28-day mortality?
Background
- Shorter antibiotic durations are likely better for most conditions;
- Recent summaries for non-bacteraemic patients include this 2024 review and this website
- The BALANCE RCT also recently demonstrated non-inferiority between 7 and 14 days in bloodstream infections
- PCT and CRP are inflammatory biomarkers that show promise in guiding decisions to stop antibiotic treatment in critically unwell patients with suspected sepsis. The 2021 Surviving Sepsis guidelines suggest using PCT along with clinical evaluation to decide when to discontinue antibiotics
- The most important previous individual RCTs on PCT-guided discontinuation are:
- PRORATA (2010, French, n=621, 23% reduction in antibiotic exposure with non-inferior mortality)
- ProGUARD (2014, Australian, n=394, did not achieve 25% reduction in duration)
- SAPS (2016, Netherlands, n=1546, PCT guidance reduced duration of treatment by an average of 1.2 days and was associated with a 5.4% reduction in 28 day mortality)
- PROGRESS (2021, Greek, n=266, effective in reducing their composite outcome of infection-associated adverse events)
- Recent published meta-analyses (2018, 2023, 2024) have shown that PCT-guided discontinuation strategies shortened antibiotic treatment duration by an average of 1.8 – 1.9 days, and were associated with a mortality benefit
- The most important previous individual RCTs on PCT-guided discontinuation are:
Design
- Multicentre randomised controlled trial
- Enrolment paused during the first UK COVID lockdown (March-August 2020)
- Randomised in a 1:1:1 ratio of PCT-guided : CRP guided : Standard care, via a central web-based system
- Stratification factors of
- Severity (sepsis v septic shock)
- Recruitment centre
- Surgery within last 72h
- Clinicians were blinded to the intervention; they only received the protocol advice
- Biomarker blood values were not reported back to the treating team
- Clinicians were allowed to measure CRP outside of the protocol; non-trial use of CRP was recorded and monitored locally
- Clinicians were not permitted to request procalcitonin for the trial period (28 days after randomisation); non-trial use was counted as a protocol deviation
- Sample size calculation: 2760 patients (with a 5% drop-out rate) would give 90% power to detect a reduction of 1 day of antibiotic administration from a baseline of 7 days with a 5% false positive rate
- Intention to treat analysis
Setting
- 41 UK NHS intensive care units
- Recruitment from January 2018 to June 2024
Population
- Inclusion: Patients ≥18 years old admitted to participating intensive care units with “suspected sepsis” whose intravenous antibiotics had been initiated within 24 hours and were likely to be continued for at least 72 hours
- “Suspected sepsis” was defined as “acute organ dysfunction associated with suspected infection”. No individual definition of “acute organ dysfunction” was mandated
- Exclusions: Patients who:
- required mandated prolonged (>21 days) antibiotic therapy
- were severely immunocompromised from a cause other than sepsis
- were expected to receive IL-6 receptor inhibitors
- were expected to have antibiotics stopped within next 24h due to futility
- were unlikely to receive IV antibiotics for 72h
- had previously been enrolled in this trial
- declined consent (patient, family, consultee)
- 16109 screened; 2761 randomised; 2695 in effectiveness analysis and 2631 in safety analysis
- 8046 had already had >24h IV antibiotic treatment
- 1239 declined consent
- 908 severely immunocompromised
- 671 needed prolonged antibiotic therapy,
- 396 no reason given
- Baseline characteristics were well matched and representative of a general adult ICU – PCT v CRP v Standard
- Age 61 v 60 v 60
- Female 39% v 40% v 41%
- Medical admissions 62% v 61% v 61%
- Community-acquired sepsis 68% v 68% v 68%
- Respiratory infections 48% v 50% v 50%, intra-abdominal 26% v 23% v 22%
- Causative microorganism identified in 47% v 46% v 47%
- Median SOFA score 7.0. Mean APACHE II score 17.5 v 17.3 v 17.2
- 19 trial patients (0.7%) had COVID
- Within 3 hours of presentation 93% of patients had a lactate measured, 94% were started on broad-spectrum antibiotics, and 53% had blood cultures prior to starting antibiotics. 31% received systemic steroids within 12 hours
Intervention
- PCT-guided and CRP-guided
- Groups had daily serum values taken, with written advice delivered daily by local research team to treating clinician until antibiotic discontinuation
- PCT-guided group:
- PCT < 0.25 μg/l: “Protocol STRONGLY SUPPORTS stopping antibiotics”
- PCT fall by >80% from baseline or PCT > 0.25 & < 0.50 μg/l: “Protocol SUPPORTS stopping antibiotics”
- Otherwise: “Protocol supports usual care”
- CRP-guided group:
- CRP < 25 mg/l: “Protocol STRONGLY SUPPORTS stopping antibiotics”
- CRP fall by 50% from baseline: “Protocol SUPPORTS stopping antibiotics”
- Otherwise: “Protocol supports usual care”
Control
- Blood samples not tested for serum PCT or CRP
- Written advice delivered daily was always “Protocol supports usual care”
Management common to both groups
- Standard NHS care for sepsis and antibiotic stewardship, including 72-hourly MDT antibiotic stewardship reviews
Outcome
- Primary outcomes:
- Effectiveness: Total antibiotic treatment duration to 28d after randomisation
- Significant reduction for PCT-guided protocol compared with standard care (9.8d for PCT v 10.7d standard care, 95% CI 0.19 to 1.58 days, p=0.01)
- Pre-planned Bayesian analysis of primary outcome with non-informative priors showed a 85% posterior probability of a reduction in antibiotic days in the PCT-guided arm of greater than 0.5 days compared with standard care, 62% probability of >0.75 days
- No difference in duration for CRP-guided protocol (10.6 v 10.7d)
- Significant reduction for PCT-guided protocol compared with standard care (9.8d for PCT v 10.7d standard care, 95% CI 0.19 to 1.58 days, p=0.01)
- Safety: 28d all-cause mortality (non-inferiority, margin 5.4%)
- Non-inferior increase in PCT-guided protocol compared with standard care (20.9% v 19.4%; absolute difference 1.57%, 95% CI -2.18% to 5.32%)
- Inconclusive (i.e. top end of 95% CI crosses non-inferiority margin) for all-cause mortality in CRP-guided protocol (21.1% v 19.4%, absolute difference 1.69%, 95% CI -2.07 to 5.45)
- Effectiveness: Total antibiotic treatment duration to 28d after randomisation
- Secondary outcomes:
- No differences in: (PCT v CRP v Standard)
- 90d all-cause mortality (25.6 v 26.4 v 25.5)
- Length of ICU stay (6.2 v 6.0 v 5.8)
- Total antibiotic daily defined dose, unscheduled care escalation or re-admission, time to first deemed fit for hospital discharge, time to hospital discharge, infection relapse or recurrence requiring further antibiotic treatment, new infection or superinfection at different anatomical site, suspected clinically relevant antibiotic-related events
- PCT-guided protocol and CRP-guided protocol better than standard care for antibiotic treatment duration for initial sepsis period (PCT 7.0d v CRP 7.4d v Standard 8.1d)
- No differences in: (PCT v CRP v Standard)
- Pocock’s win ratios: PCT-guided v Standard care odds of win 1.12 (1.0 to 1.25); CRP v Standard care 0.98 (0.88 to 1.10)
- None of these effects were significantly modified in the pre-specified subgroups
- Process measures and notes:
-
- “Strong stop” advice was more common and produced earlier in PCT group compared with CRP group
- The first “Strong Stop” recommendation occurred at 3.5 days in the PCT group and 5.1 days in the CRP group. Antibiotics were on average continued for 7.0 days and 7.4 days respectively
- Non-trial PCTs were requested infrequently (60 v 95 v 75 times in the PCT, CRP, and Control arms)
Authors’ Conclusions
- “In critically ill hospitalized adults with sepsis, there is a significant safe reduction in the total antibiotic days when a daily PCT-guided protocol is administered compared with standard care. A daily CRP-guided protocol does not reduce the total duration of antibiotics”
Strengths
- Large, excellently run, pragmatic, multicentre, randomised, intervention-concealed, trial, with transparent documentation and analyses
- Cohort of patients is representative of UK ICUs
- Protocol advice intervention thresholds were set very reasonably
Weaknesses
- Alert-based advice prevented clinicians from setting personalised antibiotic discontinuation threshold based on clinical status and pretest probabilities, or on trends over time, or clinician preference
- This is a reasonable trade-off for maintaining blinding, but makes real-world deployment more difficult
- The degree to which clinicians followed the protocol in the trial is “baked in” to the result in this intention-to-treat analysis. Replicating this positive result in the real world may require a similar degree of adherence as clinicians demonstrated during the trial; i.e. continuing antibiotics for an average of 3.5 days further than the protocol recommends for the PCT group. We do not know if better adherence would be safe, or if poorer adherence would be effective
- CRP would have been routinely monitored for many patients, which could conceivably have led to unblinding if clinicians were aware of the intervention thresholds and then compared routinely collected results with the protocol advice
- I think this is unlikely to have happened often, and given the range of opinions on PCT and CRP it will likely have biased toward standard care
- Wide pre-specified non-inferiority margin of 5.4% for absolute increase in mortality. However, I am reassured given the 90d mortality data and previous trials on PCT
The Bottom Line
- This trial provides further evidence that a PCT-based antibiotic discontinuation protocol may be a safe and effective way of reducing the duration of antibiotics in critically unwell patients with suspected sepsis
- Given the way the intervention was blinded, further process evaluation work will be needed before a real-world implementation plan can be recommended
External Link
Metadata
Summary author: Hrishee Vaidya – bsky / twitter
Summary date: 16/12/24
Peer-review editor: Dave Slessor
Picture by: Pexels – Alexas Fotos