Correlation of Serum and Urine Midazolam Levels with Consciousness Tests after Discontinuation of Sedation in the Intensive Care Unit


  • Marielena Papadaki 1st Department of Intensive Care, Evangelismos Hospital, University of Athens Medical School, 45-47 Ipsilantou Str., 106 76 Athens, Greece
  • Maria Pratikaki
  • Achilleas Giannopoulos
  • Theodora Ntaidou
  • Eleftheria Mizi
  • Marios Kougias
  • Georgios Bouboulis
  • Aikaterini Sarri
  • Charikleia S. Vrettou



Title: Correlation of serum and urine midazolam levels with observed level of consciousness after discontinuation of midazolam sedation in the intensive care unit


Marilena Papadaki1, Maria Pratikaki2, Achilleas Giannopoulos1, Theodora Ntaidou1, Eleftheria Mizi1, Marios Kougias1, Georgios Bouboulis1, Aikaterini Sarri1 and Charikleia S Vrettou1


1  1st Department of Intensive Care, Evangelismos Hospital, University of Athens Medical School, 45-47 Ipsilantou Str., 106 76 Athens, Greece

2 Department of Clinical Biochemistry, Evangelismos Hospital, Athens, Greece


Introduction: Continuous infusion of midazolam is related to prolonged activity and delayed awakening in the critically ill. Serum benzodiazepine levels can be helpful in differentiating residual benzodiazepine activity from other causes of impaired level of consciousness (LOC) [1]. Although benzodiazepine levels can also be measured in the urine, the relationship between serum and urine levels with the observed LOC has not been studied in clinical practice.

Objectives: To investigate the correlation between serum and urine benzodiazepine levels in the critically ill and their correlation with the observed level of consciousness estimated with the Glasgow Coma Scale (GCS) and the Full Outline of UnResponsiveness Score (FOUR score).

Patients and Methods: This prospective observational study involved patients admitted to a 30 bed General Intensive Care Unit, who were intubated and mechanically ventilated, with GCS prior to intubation > 8. Midazolam infusion was discontinued for at least 12 hours before sampling. Serum and urine sampling and clinical evaluation to calculate the GCS and FOUR score were done simultaneously. Gathered data included age, sex, weight and height, reason for admission to intensive care, renal function, daily fluid balance, daily and hourly urine output, liver function, serum proteins, hemoglobin and the application of renal replacement therapy. Serum benzodiazepine measurements were performed on the Integra system (Roche), which is suitable for semiquantitative detection of benzodiazepines in the serum. Urine benzodiazepine levels were measured with the Cobas C501 system, which is suitable for semiquantitative detection of benzodiazepines in human urine. The Scientific and Ethics committee of Evangelismos hospital approved the study protocol.

Results: Twenty patients were included in the study, 10 male and 10 female. Reasons for ICU admission were septic shock (n=7), respiratory failure and ARDS (n=7), and acute surgery and trauma (n=6). Patients’ age ranged from 20 to 90 years old (median 66 years) and their weight from 45 to 160 Kg  (median 77.5 Kg). The SOFA score ranged from 4 to 15 (median 8). The GCS score from 3 to 14 (median 7) and the FOUR score from 3 to 15 (median 10). Six patients were on continuous veno-venous haemodiafiltration (CVVHD) at sampling time. Serum benzodiazepine levels correlate moderately with the GCS (R =-0.496, p=0.026) and better with the FOUR score (R =-0.685, p=0.001), but did not correlate with measured levels in the urine (R =-0.029 p=0.904), even when patients without AKI were analysed separately (n= 12, R = 0.173, p=0.572). Figure 1 presents the scatter plot of measured urine and serum benzodiazepine levels in our sample.

Conclusions: In patients treated in the intensive care unit, after discontinuation of midazolam sedation, the LOC (GCS and FOUR score) correlate significantly with the benzodiazepine levels measured in the serum. Urine benzodiazepine levels do not correlate with serum levels or with the observed LOC and therefore cannot be helpful in the differential diagnosis of drowsiness or coma in this population.

References: (1) Rosich Andreu et al. Intensive Care Medicine Experimental 2015, 3(Suppl 1):A330