Criteria abstracted from The Users' Guide to Medical Literature, from the Health Information Research Unit and Clinical Epidemiology and Biostatistics, McMaster University

Highlighted lines and questions below provide links to the pertinent description of criteria in The EBM User's Guide, now available at the Canadian Centres for Health Evidence

Article Reviewed:

Please visit the new Evidence Based Journal Club Reviews

Sedation in the Intensive Care Unit. A Systematic Review

Ostermann ME, Keenan SP, Seiferling RA, Sibbald WJ.

JAMA. 2000;283:1451-1459. [abstract] [full-text for subcribers]

Reviewed by Patricia Jines, MD, University of California, Los Angeles

Review posted September 21, 2001

---

I. Are the results of the study valid?

A. Primary questions:

1. Did the overview address a focused clinical question?

This overview is focused on identifying which sedatives provide the optimal sedation, shortest time to extubation and length of ICU stay in mechanically ventilated adult patients of intensive care units. The clinical question the authors were trying to answer was very broad. They included a wide variety of studies, patient populations and drug combinations which resulted in an inability to formulate any meaningful conclusions. The article created interest because it exposed the lack of a systematic approach to an area of vital importance to intensivists.

2. Were the criteria used to select articles for inclusion appropriate?

The selection of the articles for review was appropriate. Criteria included comparison of at least two sedative drugs, outcomes measuring quality of sedation, time to extubation, or length of stay in the ICU; and study designs were randomized trials. The study population did not include patients undergoing withdrawal of life support. Studies available only in abstract were excluded. Sedation was analyzed separately as short-term sedation if it was 24 hours duration and long-term sedation if it was > 24 hours duration. Each group compared short and long-term sedation in cardiac patients and other ICU patients separately. Scoring systems were used to assess quality of sedation in 30 of the 32 trials. The most frequently used was the Ramsay score (60%). Other scores were the modified Ramsay (6%), Cook and Palma modified Glasgow Coma Scales (10%) and in 30% scoring system were created for each specific study.

B. Secondary questions:

3. Is it unlikely that important, relevant studies were missed?

Thirty-two randomized control trials enrolling adult patients receiving mechanical ventilation and requiring short-term or long-term sedation were selected for the review. The selected studies were obtained through a key word search of MEDLINE, EMBASE and Cochrane collaboration from 1980 to June 1998. Also a hand search of six anesthesiology journals: Anesthesiology, Anesthesia and Analgesia, Canadian Journal of Anesthesia, British Journal of Anesthesia, Anesthesia and Intensive Care and Anaesthesia in the same time period was done. Experts and first authors of selected articles as well as 18 pharmaceutical companies were contacted and reference lists of retrieved articles and personal files were reviewed.

Considering this degree of attention, it is unlikely that important, relevant studies were missed.

4. Was the validity of the included studies appraised?

The authors examined eight validity criteria in all the trials. Four of them were applicable to most systematic reviews of randomized controlled trials and the other four criteria were of specific importance in studies of sedative agents. The only criteria that can be assessed as presented in the article are: 1) Where health care workers blinded? Blinding was accomplished in only 16% of the studies. 2) Were cointerventions that could have affected the outcome standardized between groups in the trial? Cointerventions such as weaning strategy was mentioned (3/32, 9%), anesthesia for postoperative patients (7/9, 78%), use of analgesia (7/32, 22%) and use of neuromuscular blockers (13/32, 41%). These interventions were used in a varied fashion in the trials, and most of them only considered the use of anesthesia for postoperative patients as a cointervention. 3) Was intention to treat considered? Intention to treat was present in most of the studies (81%). 4) Was follow up considered? Follow up was assessed in most of the studies if it was applicable. 5) Was a baseline data considered? The baseline data included: age, weight, hepatic, renal and lung function and 6) Was the use of sedation scored? As it was mentioned before, sedation was scored using Ramsay, Modified Ramsay Scores, Cook and Palma Glasgow Modified Scores or specific scores created for the studies. 7) Was masking of allocation reported? Masking of allocation was not reported in 11/32 (34 %) of the studies. We were able to determine only 7 of the 8 criteria they used. The authors state they used the validity criteria but did not clarify whether they interpreted the results of the studies differently, i.e., different weights based on better validity.

5. Were assessments of studies reproducible?

The assessment of validity was done through duplicate data abstraction that was conducted by 2 of 3 investigators. In that way, all the decisions were subject to less mistakes (random errors) or bias (systematic errors) than if there was just one investigator doing all the decisions. Differences between the reviewers were resolved by consensus, in consultation with a third investigator.

6. Were the results similar from study to study?

No. The heterogeneity of the studies allowed only subgroup analysis. Most of the studies included in this overview were grouped according to drug used, duration of sedation and also the type of patient receiving sedation in order to compare similar results. Two big groups were outlined: short-term sedation studies that included 18 of 32 (56 %) trials that evaluate sedation in less than 24 hours and long-term sedation studies that included 14 of 32 (44 %) trials that evaluate sedation for more than 24 hours.

In the short-term sedation group the 18 studies included were classified according to two patient populations: cardiac patients and surgical or mixed ICU patients. There were 9 studies (50%) in cardiac patients group that compared propofol and midazolam. Of this group only 7 studies compared quality of sedation between propofol and midazolam; in 5 there was no difference reported and in only 2 studies (Ground et al, 1987 and Murray et al., 1990) propofol was demonstrated to be more effective than midazolam. There was a very wide range for the propofol doses (mean from 8.3 to 45 mcg/kg/min) and for midazolam doses (mean from 0.3 to 3.3 mcg/kg/min) as well if we really try to consider these studies similar.

In the short-term sedation group, there were also 9 studies (50%) done in surgical or mixed ICU patients. Of this group, 6 studies compared effectiveness of propofol versus midazolam. In 3 of these 6 studies (Boyd et al, 1993, Carrasco et al, 1993 and Costa el al, 1994) propofol was found more effective than midazolam in quality of sedation achieved, and although in these studies time of extubation was reported there was no mention of duration on ventilation or ICU stay. We should also remark that in these 6 studies there were seven different score system used to compare sedation level. The other 3 studies in the surgical or mixed ICU patients showed different conclusions. One study (Cernaiau et al, 1996) compared midazolam to lorazepam and found no difference in quality of sedation; in this case a study-specific scale of sedation was used and no important outcomes as time of sedation or time of extubation or length of ventilation or ICU stay were given. Another study (Kong et al, 1989) compared midazolam to Isoflurane and found superior quality of sedation and no difference in hemodynamic effect in the isoflurane use. The last study in this group (Heinrichs et al, 1992) compared propofol to a lytic solution (pethidine, promethazine and dihydroergotamine) and showed similar quality of sedation and hemodynamics in both groups but a shorter time of extubation in the patients receiving propofol. For these last 3 studies due to the different drugs compared and different doses used, we cannot consider these as a similar subgroup either. Also, as in the above mentioned studies few clinical outcomes were considered and different sedation scales used.

In the long-term sedation group that included 14 studies, 7 studies (50%) compared propofol versus midazolam and only one of them found midazolam favored over propofol (Weinbroum et al 1997). In 3 other trials (Kress et al 1996, Barrientos-Vega et al 1997 and Sanchez-Izquierdo et el, 1998) sedation was similar. Another 2 trials (Carrasco et al, 1993 and Costa et al, 1994) favored propofol rather than midazolam but only in the last study was a statistical difference (p < 0.05) was reported. Time of extubation was reported in 4 trials and was statistically shorter for the propofol use in only two of them (Carrasco et al, 1993 and Barrientos-Vega et al 1997). It should also be noted that in any these studies duration of ventilation was not reported. There is 1 study in this group (Pohlman et al, 1994) that compared midazolam and lorazepam found no difference in the quality of sedation or hemodynamic effects. The study that compared midazolam with isoflurane (Spencer and Williatts, 1992) found no difference in quality of sedation, hemodynamics and duration of ICU stay but did find a shorter time to extubation in the use of isoflurane compared to midazolam. Also 1 study (Kolenda et al, 1996) compared ketamine with fentanyl in head injury patients and found in the patients receiving ketamine higher blood pressure and heart rate compared to the those receiving fentanyl.

Finally 4 trials compared different combination of sedatives (Adams et al, 1998, Ledingham et al, 1988, Harris et al, 1990 and Manley et al, 1997) and only one trial showed a difference in outcome. The combination of alfentanil and propofol provided better quality of sedation than morphine and midazolam. As it was noted in the previous studies no significant clinical outcomes were assessed in most of these studies, sedation scales were mostly study-specific with only 1 study using Ramsay level (Harris et al, 1990), and drug doses were reported only in two trials (Harris et al, 1990 and Manley et al, 1997). We can agree that because of the marked clinical differences amongst all the studies, no statistical pooling of data was performed. Therefore, formal tests of heterogeneity were not appropriate.

II. What are the results?

1. What are the overall results of the review?

The overall results of the review are descriptional only (1). No additional statistics were done and the reader is asked to refer to the original articles cited in the review for specifics of their conclusions. It is clear that only a small proportion of sedative agents used in surveys in North America and Europe have been evaluated rigorously by more than 1 or 2 RCT's. Propofol appears as effective as midazolam to achieve similar level of sedation. Use of propofol resulted in more problems with hypotension compared to midazolam. More studies are required to assess the cost effectiveness of propofol (in settings where rapid awakening of patients is needed) and the risk of hypotension. Isoflurane use may result in a shorter time to extubation than midazolam and better quality of sedation for short-term sedation. Ketamine had the same sedative effect as fentanyl in head injury patients and results in less hemodynamic instability, so perhaps it can have a role in patients receiving inotropes. As it was mentioned by the authors, no cost evaluations were included and no general conclusions were drawn mostly due to the heterogeneity of patients and dosing regimens. A meta-analysis of study results was not possible.

2. How precise were the results?

The results obtained were limited to partial comparisons in between selected group of trials. No estimates of precision can be offered, as no statistical analysis was performed.

III. Will the results help me in caring for my patients?

1. Can the results be applied to my patient care?

The series of studies were done in adult population and the results are of questionable value in their applicability to children. For example, the use of propofol is limited in pediatric ICU's due to the initial lactic academia and bradyarrhythmia reported in 1992 (2) and the "propofol infusion syndrome" reported in subsequent pediatric studies (3). Although there may not be a causal relationship between these fatal events and propofol administration, they have prompted the manufacturer to forward a letter warning against its use in children. In addition, a recent unpublished trial showing excess mortality in pediatric patients receiving propofol has resulted in a strong letter from the FDA and the manufacturer (pdf of letter). None of the results obtained in this review bring a new concept or a change in the current way sedation in our PICU is performed.

2. Were all clinically important outcomes considered?

Clinically important outcomes were considered: hemodynamic stability, time of sedation, time for extubation, patient diagnosis, weaning strategy, use of neuromuscular blockade. Unfortunately, not all the trials included these outcomes equally, in order to generate a general conclusion.

3. Are the benefits worth the harms and costs?

No obvious benefits of any specific drug compared to harms and costs were identified in this review. The wide variety of patients, drugs, and assessment techniques used in these studies defy a simple cost-benefit summary. We need to keep in mind also that there is not a single appropriate level of sedation for all ICU patients and the levels are suppose to change according to the patient condition.

Finally, we should mention new devices that attempt to assess the level of sedation in different patient populations: the Bispectral Index and the middle latency auditory evoked response(4). The Bispectral Index (BIS) monitor is a methodology that uses a processed EEG signal from a forehead sensor that purports to quantify the level of sedation. It is considered useful during neuromuscular blockade, mechanical ventilation, barbiturate coma and bedside procedures. In cases of propofol-induced hypnosis, it is highly predictive of depth of sedation, as judged by responsiveness of the patient to command and tactile stimulation. (5) A significant limitation, however, is that BIS thresholds do not appear to be independent of the combinations of anesthetic agents administered. That is, comparable BIS values achieved with different combinations of agents do not represent the same depth of anesthesia (4). The middle latency auditory evoked response (MLAER) is a technology that is not yet commercially available (as is the BIS monitor). This is another EEG signal recorded from a mastoid-vertex electrode that produce wave forms that can discriminate between conscious and anesthetized states (6). Although it is still too early to draw conclusions, and as with every new technology there is need of more assays in different patient types and conditions, comparison of sedation regimens will continue to be problematic until such quantification techniques are validated and available.

References

1. Oxman, AD, Cook DJ, Guyatt GH: Evidence Based Medicine Working Group: How to Use an Overview . JAMA 1994;272 (17): 1367-71 [full-text]
2. Parke TJ, Stevens JE, Rice ASC et al. Metabolic acidosis and fatal myocardial failure after propofol infusion in children: five case reports. Br Med J 1992; 305:613-616. [abstract]
3. Bray RJ. Propofol infusion syndrome in children. Paediatric Anesthesia 1998; 8: 491-499. [abstract]
4. Drummond JC. Monitoring depth of anesthesia: with emphasis on the application of the bispectral index and the middle latency auditory evoked response to the prevention of recall. Anesthesiology 2000; 93(3):876-82 . [abstract]
5. Liu J, Singh H, White PF. Electroencephalographic bispectral index correlates with intraoperative recall and depth of propofol-induced sedation. Anesth Analg 1997; 84 (1): 185-9. [abstract]
6. Schwender D, Kaiser A, Klasing S, Peter K, Poppel E. Midlatency auditory evoked potentials and explicit and implicit memory in patients undergoing cardiac surgery. Anesthesiology 1994; 80 (3): 493-50. [abstract]

Comments

Back to the EB Journal Club Index