MIDEX-PRODEX: Dexmedetomidine vs Midazolam or Propofol for Sedation During Prolonged Mechanical Ventilation: Two Randomized Controlled Trials
Jakob. JAMA 2012; 307:1151-1160. doi:10.1001/jama.2012.304.
- In adult intensive care patients requiring mechanical ventilation, is dexmedetomidine inferior to propofol or midazolam in achieving target sedation level?
- Parallel trial design, depending on the study centre’s usual practice
- MIDEX: midazolam
- PRODEX: propofol
- Non-inferiority statistical design
- ‘Double-dummy’ design with opaque syringe and giving set to mask propofol administration
- Double blind design – study personnel and patients were unaware of the group allocation
- Deemed non-inferior if the 95% confidence intervals for the dexmedetomidine to standard sedation ratio for the time within target RASS score did not cross 0.85 (i.e. 15% inferiority margin)
- For 90% power a sample size of 450 in each trial was required
- (RASS score is Richmond Agitation-Sedation Scale – see external links at bottom)
- MIDEX: 44 centres in 9 European countries
- PRODEX: 31 centres in 6 European countries and 2 in Russia
- 2007 through 2010
- Inclusion: over 18 years; invasive mechanical ventilation; target RASS score 0 to -3 (light to moderate sedation); expected to require sedation for at least 24 hours; within 72 hours of ICU admission and within 48 hours of starting either midazolam or propofol
- Exclusion: acute severe neurological disorder; MAP < 55 mmHg despite optimisation; heart rate < 50/min; AV conduction defects; recent use of alpha agonists or antagonists
- MIDEX: 8301 ventilated patients screened of which 501 randomised
- PRODEX: 11610 ventilated patients screened of which 500 randomised
- Dexmedetomidine 0.2-1.4 mcg/kg/hr
- MIDEX: midazolam 0.03-0.2 mg/kg/hr
- PRODEX: propofol 0.3-4.0 mg/kg/hr
- Primary outcome: non-inferiority of dexmedetomidine confirmed against both midazolam and propofol
- Time at target sedation (RASS 0 to -3) without rescue medication was similar for midazolam vs dexmedetomidine and for propofol vs dexmedetomidine. Small differences were not significant and the 95% confidence intervals for the ratio did not cross 0.85.
- Duration of mechanical ventilation was significantly shorter for dexmedetomidine compared to midazolam, but the slightly shorter duration for dexmedetomidine against propofol was not statistically significant.
- Secondary outcome: superiority of dexmedetomidine was suggested but not confirmed statistically
- Median length of stay in ICU was shorter for dexmedetomidine against both midazolam and propofol, but differences were not statistically significant.
- The nurses’ assessment of patient arousal, cooperation and communication by visual analogue scale were higher (better) for dexmedetomidine against both midazolam and propofol.
- Other data outcomes:
- Mean sedation scores were higher for dexmedetomidine against both midazolam and propofol with statistical significance.
|Study||Dexmedetomidine||Usual Drug||Estimated ratio||95% CI||p|
|Time at target sedation without rescue medicine – proportion of total time|
|MIDEX||60.7%||56.6%||1.07||0.97 – 1.18||0.15|
|PRODEX||64.6%||64.7%||1.00||0.92 – 1.08||0.97|
|Median duration of MV (including NIV) – hours [IQR]|
|MIDEX||123 [67–337]||164 [92–380]||0.03|
|PRODEX||97 [45–257]||118 [48–327]||0.24|
|CI = confidence interval; p = p-value; MV = mechanical ventilation; NIV = non-invasive ventilation|
|Study||Dexmedetomidine||Usual Drug||MD||95% CI||p|
|Median length of ICU stay – hours [IQR]|
|MIDEX||211 [115–831]||243 [140–630]||n/a||0.27|
|PRODEX||164 [90–480]||185 [93–520]||n/a||0.54|
|Combined VAS (as mm out of 100) for nurses’ assessment of arousal, cooperation and communication|
|15.2 – 24.2||<0.001|
|6.4 – 15.9||<0.001|
|MD = mean difference; CI = confidence intervals; p = p-value; VAS = visual analogue scale|
- Dexmedetomidine is non-inferior to midazolam and propofol for long-term mild to moderate sedation on intensive care.
- It may reduce time to extubation and it does improve communication between staff and patients.
- Efficient use of parallel trial design
- Clinically useful outcome measure
- Independent statistical analysis helps toward preventing any bias from sponsoring manufacturer
- Per-protocol analysis for non-inferiority primary outcome will prevent bias toward non-inferiority conclusion, whereas the intention-to-treat analysis for efficacy secondary outcomes prevent bias toward positive difference. This approach will limit the chance of false positive conclusions.
- Extensive online publication of raw data
- Sponsored by manufacturer of dexmedetomidine with significant trial design, oversight and manuscript input
- Recruitment rate averages 3 to 4 patients per year per centre! Was there selection bias?
- Only short term measures of delirium and neurocognitive problems used
- The finding that mean sedation scores were higher in patients managed with dexmedetomidine may be a confounding factor (by chance or by accidental unblinding) leading to a bias that favours dexmedetomidine. Alternatively, it may be a positive finding that sedation was easier to manage and titrate with dexmedetomidine.
The Bottom Line
- This non-inferiority trial opens the door to using dexmedetomidine in intensive care, but further trials with longer-term neurocognitive follow-up and cost-analysis are required before it can be recommended as first-line.
- [article abstract] Dexmedetomidine vs Midazolam or Propofol for Sedation During Prolonged Mechanical Ventilation Two Randomized Controlled Trials
- [further reading] Dexmedetomidine by Life in the Fast Lane
- [further reading] CONSORT guidance for non-inferiority trials
- [further reading] The pros and cons of noninferiority trials by Pocock
- [further reading] The Richmond-Agitation Sedation Scale by ICUDelirium.org