Skip to main content

Foundational Trials

Kress 2000

Image: Shutterstock / ame435

Publication

  • Title: Daily interruption of sedative infusions in critically ill patients undergoing mechanical ventilation
  • Year: 2000
  • Journal published in: The New England Journal of Medicine
  • Citation: Kress JP, Pohlman AS, O’Connor MF, Hall JB. Daily interruption of sedative infusions in critically ill patients undergoing mechanical ventilation. N Engl J Med. 2000;342(20):1471-7.

Context & Rationale

  • Background
    • Continuous IV sedation (often benzodiazepines, frequently with opioids) was routine for ventilated ICU patients to manage anxiety/agitation and facilitate synchrony.
    • Prolonged/accumulated sedation was believed to prolong mechanical ventilation, extend ICU/hospital length of stay, and impede daily neurological assessment.
    • Altered mental status during deep/continuous sedation often triggered diagnostic testing (e.g., neuroimaging) with cost and iatrogenic risks from transport.
  • Research Question/Hypothesis
    • Whether a strategy of daily interruption of sedative (midazolam or propofol) and morphine infusions (until patients awakened) would shorten mechanical ventilation duration and ICU length of stay versus usual care.
    • Whether this strategy would reduce diagnostic testing for changes in mental status without increasing clinically important adverse events (e.g., unplanned device removal).
  • Why This Matters
    • Mechanical ventilation duration and ICU length of stay are high-impact outcomes linked to morbidity (e.g., VAP, barotrauma), resource use, and costs.
    • Sedation delivery is a modifiable iatrogenic exposure, making a low-cost, bedside-deliverable strategy potentially practice-changing.

Design & Methods

  • Research Question: In mechanically ventilated adults receiving continuous sedative-drug infusions, does daily interruption of sedative and morphine infusions (until awake) reduce ventilation duration and ICU length of stay compared with clinician-discretion interruption only?
  • Study Type:
    • Randomised, controlled, single-centre trial in a medical ICU (University of Chicago).
    • Open-label drug administration; investigators knew allocation; specific study end points were not disclosed to caregivers.
    • Within each strategy arm, patients were also randomised to midazolam vs propofol (both with concurrent morphine infusion).
    • Institutional review board approval with waiver of patient consent (intervention considered within established local standard of care).
  • Population:
    • Adults receiving mechanical ventilation and continuous infusions of sedative drugs in a medical ICU.
    • Analgesia: all patients received a morphine infusion alongside the non-analgesic sedative (protocolised ranges).
    • Target sedation: nurse titration to Ramsay sedation scale score 3–4 (responsive to commands only; or asleep with brisk response).
    • Exclusions: pregnancy; transfer from another institution with sedatives already administered; admission after resuscitation from cardiac arrest.
  • Intervention:
    • Daily interruption of midazolam or propofol infusion and morphine infusion starting 48 hours after enrolment.
    • Infusions stopped until patient was awake and could follow instructions, or until uncomfortable/agitated and deemed to require resumed sedation.
    • If receiving cisatracurium, sedative infusion not interrupted; resumed daily interruption after paralytic stopped.
    • After interruption, infusions restarted after awake (or if agitation prevented successful waking) at half prior rates, then titrated as needed.
    • Wakefulness definition (objective): able to perform ≥3/4 actions (open eyes to voice; track examiner; squeeze hand; protrude tongue) at any point that day.
  • Comparison:
    • Continuous infusion of sedatives (midazolam or propofol) plus morphine with interruption only at ICU clinician discretion.
    • Research staff recorded total daily sedative doses; management decisions (other than mandated interruption/resumption in intervention arm) remained with ICU team.
  • Blinding: Unblinded (open-label sedation strategy); caregivers were not told study end points, but daily interruption is operationally apparent.
  • Statistics:
    • Power calculation: Not reported.
    • Analysis: stated as intention-to-treat, but patients dying or extubated on ICU day 1–2 (before first possible interruption) were excluded from analysis (modified ITT).
    • Time-to-event: Kaplan–Meier and Cox proportional hazards (adjusted for age, sex, weight, APACHE II, and type of respiratory failure).
    • Continuous outcomes: Mann–Whitney U; categorical outcomes: chi-square/Fisher’s exact; two-sided testing.
  • Follow-Up Period: Until hospital discharge (in-hospital outcomes).

Key Results

This trial was not stopped early. No interim stopping rationale was reported.

Outcome Daily interruption (n=68 analysed) Usual care (n=60 analysed) Effect p value / 95% CI Notes
Duration of mechanical ventilation (days), median (IQR) 4.9 (2.5–8.6) 7.3 (3.4–16.1) RR (extubation) 1.9 95% CI 1.3 to 2.7; P<0.001 Median difference 2.4 days shorter; P=0.004
ICU length of stay (days), median (IQR) 6.4 (3.9–12.0) 9.9 (4.7–17.9) RR (ICU discharge) 1.6 95% CI 1.1 to 2.3; P=0.02 Median difference 3.5 days shorter
Hospital length of stay (days), median (IQR) 13.3 (7.3–20.0) 16.9 (8.5–26.6) Not reported P=0.19 No significant between-group difference reported
Days classified as “awake” while receiving infusions (% days), median (IQR) 85.5 (50.0–100.0) 9.0 (0.0–50.0) Not reported P<0.001 Awake if ≥3/4 objective actions at any time that day
Total midazolam dose (mg), midazolam subgroup, median (IQR) 229.8 (59–491) 425.5 (208–824) Not reported P=0.05 Midazolam subgroup: n=37 vs n=29
Total morphine dose (mg), midazolam subgroup, median (IQR) 205 (68–393) 481 (239–748) Not reported P=0.009 Reduced opioid exposure alongside reduced benzodiazepine exposure
Diagnostic testing for mental status change, n/N (%) 6/68 (9%) 16/60 (27%) Not reported P=0.02 Most tests reportedly non-diagnostic
Complications (patient removed ETT or central venous catheter), n/N (%) 3/68 (4%) 4/60 (7%) Not reported P=0.88 Intervention: 2 self-extubations + 1 line removal; Control: 4 self-extubations
In-hospital mortality, n/N (%) 24/68 (36.0%) 28/60 (46.7%) Not reported P=0.25 No statistically significant between-group difference
Discharged home, n/N (%) 39/68 (59%) 24/60 (40%) Not reported P=0.06 Trend favouring daily interruption; not statistically significant
  • Daily interruption accelerated extubation and ICU discharge (RR 1.9 for extubation; RR 1.6 for ICU discharge), with median reductions of 2.4 ventilator days and 3.5 ICU days.
  • There was marked protocol separation in patient wakefulness (85.5% vs 9.0% of infusion-days classified as awake) and lower cumulative midazolam/morphine exposure in the midazolam subgroup.
  • Major safety outcomes were uncommon and similar between arms (4% vs 7% for device removal complications), but the trial acknowledged limited power for rare harms.

Internal Validity

  • Randomisation and Allocation
    • Computer-generated random assignments, concealed in sealed envelopes.
    • Two-stage allocation: sedation strategy (daily interruption vs usual care) and sedative agent (midazolam vs propofol) within each strategy arm.
  • Drop out or exclusions
    • 150 patients randomised (75 per group); 22 excluded post-randomisation (7 intervention; 15 control) due to extubation or death on ICU day 1–2 before the infusion could be interrupted.
    • Analysed population: 68 intervention vs 60 control (modified intention-to-treat), which may introduce attrition bias.
  • Performance/Detection Bias
    • Unblinded intervention; bedside staff could observe interruption events, and extubation decisions can be clinician-influenced.
    • Primary outcomes (ventilation duration, ICU length of stay) are objective but susceptible to practice patterns that sedation strategy can directly influence (intended causal pathway).
    • End points were not disclosed to caregivers, a partial mitigation against behavioural co-intervention.
  • Protocol Adherence and Separation of the Variable of Interest
    • Awake-days separation: 85.5% (IQR 50.0–100.0) vs 9.0% (IQR 0.0–50.0).
    • Ventilation duration: 4.9 (2.5–8.6) vs 7.3 (3.4–16.1) days.
    • ICU length of stay: 6.4 (3.9–12.0) vs 9.9 (4.7–17.9) days.
    • Control-group contamination: 18/60 control patients had infusions stopped temporarily on non-final days; the proportion of non-final infusion-days stopped ranged 0–54%.
    • Midazolam subgroup cumulative exposure: total midazolam 229.8 vs 425.5 mg; total morphine 205 vs 481 mg.
  • Baseline Characteristics
    • Groups were similar at baseline (e.g., APACHE II medians 20 vs 22; comparable admission diagnoses distribution and permissive hypercapnia use reported).
    • Population severity appears typical for a medical ICU cohort requiring continuous sedation and invasive ventilation.
  • Timing
    • Daily interruption began 48 hours after enrolment, potentially excluding early extubations and limiting influence on very early ventilator liberation.
    • Neuromuscular blockade deferred interruption until paralytic was stopped.
  • Dose
    • Sedation target Ramsay 3–4 and concurrent opioid infusion reflect a “comfort-first but moderate-to-deep sedation” paradigm; by contemporary practice, this may be deeper than many light-sedation targets.
    • Restart strategy: after awakening (or if agitation prevented successful waking), resumed at half prior infusion rates and titrated.
  • Outcome Assessment
    • Primary outcomes were clearly defined and measured until hospital discharge.
    • Wakefulness was assessed using an objective command-based definition (≥3/4 actions) rather than solely sedation score documentation.
  • Statistical Rigor
    • Cox models adjusted for prespecified baseline variables; effects presented with 95% CIs for time-to-event outcomes.
    • No prespecified power calculation reported; acknowledged limited power for rare adverse events.

Conclusion on Internal Validity: Overall, internal validity is moderate: randomisation and allocation concealment were appropriate and the intervention achieved large separation, but unblinded delivery and post-randomisation exclusions (modified ITT) introduce risks of bias and imprecision for safety outcomes.

External Validity

  • Population Representativeness
    • Single-centre medical ICU cohort of ventilated adults on continuous sedative and morphine infusions.
    • Excluded pregnancy, transfer with prior sedation exposure, and post-cardiac arrest admissions, which may limit applicability to those subgroups.
  • Applicability
    • Intervention is operationally simple and nurse-deliverable, but requires staffing to monitor agitation/distress and coordinate restarting infusions.
    • Modern ICU practice increasingly targets lighter sedation and uses sedation protocols; effect sizes may be smaller where baseline oversedation is already minimised.
    • Generalises best to settings using continuous infusions of sedatives/opioids with opportunities to standardise daily neurological assessment and weaning readiness.

Conclusion on External Validity: External validity is moderate: the mechanism is broadly relevant, but single-centre conduct and historical sedation practices limit confidence that identical magnitude of benefit persists in contemporary, protocolised light-sedation ICUs.

Strengths & Limitations

  • Strengths:
    • Clear, pragmatic sedation strategy with objective operational definition of “awake”.
    • Clinically meaningful primary end points (ventilator duration, ICU length of stay) with adjusted time-to-event analyses and effect estimates with 95% CIs.
    • Demonstrated strong protocol separation (wakefulness and sedative/opioid exposure) consistent with the intended mechanism.
  • Limitations:
    • Unblinded intervention, with potential for practice-pattern influence on extubation and discharge decisions.
    • Post-randomisation exclusions of early deaths/extubations (modified ITT) and small sample size for safety and mortality outcomes.
    • Single-centre design and a sedation target (Ramsay 3–4) that may not reflect current light-sedation paradigms.
    • No assessment of patient-reported distress, recall, cardiovascular stress, withdrawal syndromes, or longer-term neuropsychological outcomes.

Interpretation & Why It Matters

  • Clinical signal
    • Daily interruption produced faster ventilator liberation and ICU discharge without an observed increase in major immediate safety events, suggesting oversedation is a clinically important, reversible iatrogenic exposure.
  • Mechanistic plausibility
    • The intervention increased wakefulness (85.5% vs 9.0% of infusion-days), facilitating daily neurological assessment and likely earlier readiness recognition for weaning/extubation.
    • Reduced cumulative benzodiazepine and opioid exposure in the midazolam subgroup supports an “accumulation avoidance” mechanism.
  • Practice consequence
    • The trial helped establish routine “spontaneous awakening” as a testable ICU process of care and informed later bundled approaches to sedation minimisation and ventilator weaning.

Controversies & Subsequent Evidence

  • Potential harms were incompletely measured
    • Patient distress and recall during daily awakenings, cardiovascular stress responses, and withdrawal syndromes were not evaluated, limiting harm-benefit assessment for certain subgroups (e.g., coronary disease, prolonged benzodiazepine exposure). 1
  • Attribution of benefit: “interruption” versus “avoidance of oversedation”
    • The observed benefit plausibly reflects prevention of oversedation and reduced sedative accumulation; the correspondence argued that comparable outcomes might be achievable through lower-dose continuous infusions targeting lighter sedation, rather than mandated daily interruption. 2
  • Attention and co-intervention concerns
    • The intervention was operationalised by research staff with structured monitoring during interruption periods, raising the possibility that increased attention (rather than interruption alone) contributed to earlier weaning/extubation decisions. 2
  • Later randomised evidence refined “when” daily interruption adds value
    • Pairing spontaneous awakening with spontaneous breathing trials (ABC trial) improved outcomes, supporting a bundled approach linking sedation minimisation with ventilator liberation. 3
    • When high-quality protocolised sedation is already in place, adding routine daily sedation interruption did not improve outcomes and increased sedative use and nursing workload in at least one large RCT (SLEAP), suggesting redundancy in well-run light-sedation systems. 4
    • More aggressive minimisation strategies (e.g., “no-sedation” approaches) demonstrated feasibility in selected settings but raise different trade-offs around comfort and workload. 5
  • Syntheses and guidelines
    • Systematic reviews conclude that evidence for daily sedation interruption versus alternatives is mixed and context-dependent, particularly where protocolised light sedation is practised. 6
    • International guidelines endorse strategies that achieve and maintain light sedation (including daily sedation interruption and/or nurse-protocolised sedation), emphasising that either approach can be effective depending on local implementation. 7
    • A recent focused update continues to refine guidance on pain/sedation/delirium bundles and implementation in adult ICU care. 8

Summary

  • In a single-centre medical ICU, daily interruption of sedative (midazolam/propofol) and morphine infusions shortened median ventilation duration (4.9 vs 7.3 days) and ICU length of stay (6.4 vs 9.9 days).
  • Time-to-event analyses showed earlier extubation (RR 1.9; 95% CI 1.3 to 2.7) and earlier ICU discharge (RR 1.6; 95% CI 1.1 to 2.3).
  • Wakefulness separation was large (85.5% vs 9.0% of infusion-days classified as awake), with lower cumulative midazolam and morphine exposure in the midazolam subgroup.
  • Diagnostic testing for mental-status change occurred less often (9% vs 27%), consistent with improved ability to perform daily neurological assessment.
  • Immediate safety events were uncommon and similar between groups (4% vs 7% device removal), but the trial was underpowered for rare harms and did not measure distress/withdrawal/cardiovascular effects.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • Where protocolised light sedation is consistently achieved, later evidence suggests a mandated daily interruption may add little and can increase workload; the key target is avoidance of oversedation rather than the “holiday” itself.

Overall Takeaway

KRESS reframed sedation as a modifiable determinant of ventilator dependence and ICU length of stay by demonstrating that routine daily interruption of sedative and opioid infusions can accelerate extubation and ICU discharge in a medical ICU cohort. Its landmark status lies in shifting sedation from “background support” to a measurable ICU process of care, while catalysing later trials and guideline frameworks that prioritise reliably achieving light, goal-directed sedation with structured daily awakening/weaning assessments.

Overall Summary

  • Daily interruption of sedative + morphine infusions shortened ventilation (4.9 vs 7.3 days) and ICU stay (6.4 vs 9.9 days) in a single-centre medical ICU cohort.
  • Intervention achieved substantial separation in wakefulness (85.5% vs 9.0% awake-days) and reduced cumulative midazolam/morphine exposure in the midazolam subgroup.
  • Unblinded delivery and modified ITT (post-randomisation exclusions) limit certainty; subsequent evidence supports the broader principle of avoiding oversedation, with daily interruption most useful when light sedation is otherwise not reliably achieved.

Bibliography