Criteria abstracted from The Users' Guides to the Medical Literature series in JAMA
Effect of Mechanical Ventilator Weaning Protocols on Respiratory Outcomes in Infants and Children: A Randomized Controlled Trial.
Randolph AG, Wypij D, Venkataraman ST, Hanson JH, Gedeit RG, Meert KL, Luckett PM, Forbes P, Lilley M, Thompson J, Cheifetz IM, Hibberd P, Wetzel R, Cox PN, Arnold JH, for the Pediatric Acute Lung Injury and Sepsis Investigators (PALISI) Network.
Reviewed by Mark Twite MA MB BChir MRCPCH, Pediatric Critical Care Fellow, The Children's Hospital, Denver, Colorado
Review posted April 25, 2003
I. What is being studied?:
- The study objective:
There are four study objectives - two primary and two secondary objectives:
- To evaluate whether ventilator weaning protocols are better than no protocol (standard care)
- To compare two ventilator weaning protocols, PSV (Pressure Support Ventilation) and VSV (Volume Support Ventilation) to each other. The difference between the two weaning protocols is the method by which pressure support is adjusted. In the PSV protocol, the pressure support is manually weaned according to defined criteria whereas in the VSV protocol the pressure support is adjusted automatically by the ventilator according to expiratory tidal volumes.
- To evaluate a set of extubation criteria on predicting extubation success and failure
- To evaluate sedative use during the first 24hrs of weaning on respiratory outcomes
- The study design:
Randomized, controlled, multicenter prospective trial
- The patients included:
182 children who had been ventilated for > 24 hours and failed a test for extubation readiness, i.e., actually required weaning in the first place, were included in this study.
2349 patients were screened and only 752 were eligible for the study. Of the eligible patients 313 were tested for entry into the study using the extubation readiness test. 439 were not tested for entry into the study and the reasons are given for this in the paper, but include: consent window missed, parents or guardian declined or was unavailable, physician declined etc. The 182 patients that failed the extubation readiness test were entered into the study and randomized to one of three groups: PSV group, VSV group or no protocol group.
- The patients excluded:
A large number of patients were excluded based on the following criteria:
Although the exclusion criteria are clear cut and extensive they do exclude a large proportion of 'typical' pediatric intensive care unit patients - 68% in this study.
- Age: 18years or older or corrected gestational age less than 38 weeks
- Diaphragmatic hernia or paralysis
- Ventilator use prior to admission when the patient was at baseline health status (including use of noninvasive ventilator support)
- Cyanotic congenital heart disease with unrepaired or palliated right-to-left intracardiac shunt or history of single ventricle defect at any stage of repair
- Significantly diminished lung capacity (estimated resting tidal volume < 6ml/kg), decreased lung vascularity, anatomical obstruction of lower airways, tracheal or upper airway obstructive conditions, status asthmaticus in children 2 years or older, primary pulmonary hypertension or anticipated need for nitric oxide after extubation
- Bone marrow or lung transplant
- Spinal cord injury above the lumbar region or progressive neuromuscular weakness
- Children currently enrolled in another trial in which the intervention may influence the patient's respiratory outcome or if a patient already has a decision in place to withdraw or limit life support.
- The interventions compared:
Two different weaning protocols, VSV and PSV, were compared to each other and to no protocol. The ventilator management prior to entry into the study was at the discretion of the physician. The study used an extubation readiness test (ERT) to determine if the subject was ready for extubation. The test uses objective criteria. The test involves placing the patient on a pressure support trial where the amount of pressure support is adjusted for endotracheal tube size. The trial lasts for two hours and the patient fails the test if at any time the pulse-oximetry saturations fall below 95%, the patient's respiratory rate increases to a rate that is outside of the age appropriate range or if the tidal volume of the breath falls below 5ml/kg (ideal body weight). Children who failed the ERT were eligible for study enrollment and randomization. All children in the study were placed on a Siemens Servo 300 ventilator. There were two study hypotheses:
The time to successful extubation for children receiving protocol-directed weaning (PSV and VSV combined) would be equivalent to or less than those receiving physician directed weaning (no protocol). Decreasing the time to extubation from 4 days to 2.65 days was considered clinically important. A power of 80% was used for sample size estimates.
The time to successful extubation for children randomized to the VSV protocol would be less than those randomized to the PSV protocol, because in VSV the pressure support level was continually adjusted as the patient improved. Decreasing the median time to extubation from 3 days to 2 days was considered clinically important. A power of 80% was used for sample size estimates.
- The outcomes evaluated:
Two outcomes are evaluated:
- Duration of weaning time - the time from randomization into the study until successful extubation
- The number of extubation failures defined as the patient requiring invasive or non-invasive ventilator support within 48 hours of extubation. In this study children who did not wean within 28 days of randomization, or who died or were transferred out of the study center, were treated as weaning failures.
- Primary questions:
- 1. Was the assignment of patients to treatments randomized?
Yes. Patients in the study were randomized using a standard randomization technique (permuted block design, stratified by ICU with allocation concealment).
- 2. Were all patients who entered the trial properly accounted for and attributed at its conclusion?
- Was followup complete?
Yes. Of the 182 patients enrolled 179 were included in the analysis. Of the three patients excluded, two were excluded as they were ineligible for the study and one patient was considered to have invalid consent.
Follow up was complete for all subjects except for 1 child in the VSV protocol who was transferred to another hospital. This child was treated as an extubation failure.
- Were patients analyzed in the groups to which they were randomized?
Subjects were analyzed in their randomized groups i.e., PSV protocol, VSV protocol or no protocol by intention-to-treat.
The reference groups for both of the analyses are:
- No protocol
- Neonatal age group
- Fourth quartile for both sedation score and time intubated prior to randomization.
- Secondary questions:
- 3. Were patients, health workers, and study personnel "blind" to treatment?
There was no 'blinding' in this study. Practically it would be very difficult to perform 'blinding' for this type of study.
- 4. Were the groups similar at the start of the trial?
The baseline characteristics of the subjects randomly assigned to the three groups are similar e.g., age, sex, race, diagnosis on admission and the number of days intubated prior to randomization. However, with the extensive exclusion criteria for this study, subjects that are not very sick have been enrolled.
This is demonstrated by the:
- low mean PRISM scores around 10
- wide standard deviation of the number of days intubated prior to randomization
- average time to extubation in the enrolled children is only 2 days.
- 5. Aside from the experimental intervention, were the groups treated equally?
Although the baseline characteristics of the three groups were similar this is a heterogeneous collection of subjects according to diagnoses i.e., subjects were admitted for a range of pulmonary, cardiac, neurological and other conditions. Therefore, the management of the underlying disease process was disease specific but the protocol for weaning should have been implemented uniformly between the centers. Protocol compliance was monitored prospectively and 66% of subjects completed the protocol in each PSV and VSV protocol groups.
- 1. How large was the treatment effect?
The Data Safety Monitoring Board stopped this trial after 182 patient enrollments due to lack of efficacy. The investigators determined that a clinically important difference between trial groups was a minimum of 1 day and that the study should be stopped if the differences in weaning times between study groups were unlikely to exceed 1 day.
Analysis of weaning failure (n=179), in addition to raw analysis, was only done with logistic regression, reporting odds ratios with a 95% confidence interval. Analysis of weaning time (n=146 successful extubations), in addition to raw analysis by group was supplemented by proportional hazards regression, reporting the hazard ratio with a 95% confidence interval.
The results may be considered with respect to the four objectives stated at the start of the review:
With respect to weaning failure and weaning times among successes, there are no significant differences between the three groups (see table below). In other words, this study did not show weaning protocols were any better than standard care and neither protocol is better than the other. This is in contrast to adult studies where protocols have been found to shorten the duration of ventilator weaning (4, 5).
* By Fisher exact test
| ||Pressure Support N=61 ||Volume Support N=59 ||No Protocol N=59 ||P Value
|Weaning failure No. (%) ||9 (15) ||14 (24) ||10 (17) ||0.44*
|Weaning times among successes, median (interquartile range), d
||1.6 (0.9 - 4.1) ||1.8 (1.0 - 3.2) ||2.0 (0.9 - 2.9) ||0.75**
** By log-rank test
The average extubation failure rate was 19% using an ERT and 17% using physician judgment. In other words, this study did not show that objective extubation criteria, as defined by the ERT, were any better than physician judgment. The only problem with this conclusion is that many physicians will use many of the ERT objective criteria in their judgments already!
This study showed that sedation used during the first 24 hours of weaning significantly predicted extubation failure and weaning time (see table below).
| ||Weaning failure N = 179 ||Weaning time N = 146 successes
|Sedation score (points) ||Odds Ratio (95% confidence interval) ||P value ||Hazard Ratio (95% confidence interval) ||P value
|Quartile 1 |
|0.41 (0.44 - 5.25) ||.52 ||4.11 (2.29 - 7.41) ||<.001
|Quartile 2 |
(6 - 21.9)
|0.24 (0.07 - 0.83) ||.03 ||2.01 (1.16 - 3.49) ||.01
|Quartile 3 |
(22 - 54.9)
|0.16 (0.04 - 0.71) ||.02 ||1.84 (1.06 - 3.20) ||.03
|Quartile 4 |
|Reference || ||Reference ||
(Hazard Ratios > 1 imply shorter weaning times compared to the reference group)
The investigators developed what they call a 'sedation score' where points are given for any sedative drug administered in the first 24 hours of weaning. A potency scale was modified to convert opiates and benzodiazepines to morphine and midazolam equivalents. Although this initially seems like a good way to manage a diverse range of sedative drugs and equate them between patients the concept is flawed for a couple of reasons:
i. The amount of sedation a child receives does not necessarily correlate with how sedated they are. Some children may be sedated with very small amounts of drugs whereas others require large amounts. It might have been better to assess the level of sedation using a validated scoring system in ventilated children such as the COMFORT score (1) and then show that extubation failure was related to the COMFORT score during the weaning time. The use of the terminology 'sedation score' is slightly misleading in this study. Perhaps a more accurate term would be 'sedation use score.'
ii. It is unlikely that critically ill children respond in the same way to sedative and analgesic drugs as healthy children or adults. Studies that look at the pharmacokinetics of drugs are mainly performed in healthy adult volunteers (2). Therefore to extrapolate a potency scale for use in children from adult data and then not to validate its use does not seem appropriate.
Despite the potentially flawed design concept the study conclusion with respect to sedation is nonetheless very interesting. It certainly appears to be consistent with the Kress study (3) in adults where simply stopping sedative drips on a daily basis resulted in less time on the ventilator. The challenge is translating this to a pediatric population where cognitive immaturity impedes the ability of children to be awake and tolerate having an endotracheal tube in place.
- 2. Did the treatment effect differ in subgroups of patients?
One of the major findings of this study showed that being male is a significant predictor of extubation failure (p < .001). Overall there were 33 extubation failures, of which 29 (87%) occurred in boys. The authors speculate whether anatomical, hormonal or other influences underlie the differences. To further speculate, perhaps boys, as well as having worse outcomes for many disease processes, actually respond in a different way to many drugs including sedatives and analgesics.
- 1. Can the results be applied to my patient care?
Yes and no. The patients that most pediatric intensivists would consider applying a weaning protocol to are the 'difficult to wean' patients. The study does not really address this subgroup of patients, who were purposefully excluded by the extensive study exclusion criteria. The study was designed to test ventilator weaning protocols in a carefully selected group of patients before trying to apply it to the more 'difficult to wean' patients. Adopting a weaning protocol for such a subgroup of patients may prove beneficial but has not yet been adequately studied. The conclusion that gradual weaning may not be necessary in children with acute respiratory failure may lead some intensivists to be more rigorous and frequent in their assessment of whether a child is ready to extubate.
The fact that sedative use appears to strongly influence both length of time on the ventilator and extubation failure in children is a major finding and should prompt intensivists to question how they are using sedative drugs. The question is not so much about the drugs themselves but are we adequately monitoring how they are utilized? Do we want all children to receive drugs to maintain a deep level of sedation? How do we accurately measure and then tailor sedative use in different children? Simply changing the way we use sedative drugs may have a profound effect on time to wean on the ventilator.
- 2. Were all clinically important outcomes considered?
Lengths of time weaning and extubation failure were the two outcomes measured. Other outcome measures that could also have been included are length of ICU and hospital stay. Including these outcome measures would not have changed the conclusion of the primary objectives but may have been interesting with respect to the secondary sedation question.
- Ambuel B, Hamlett K, Marx C, Blumer J. Assessing distress in pediatric intensive care environments: The COMFORT Scale. Journal of Pediatric Psychology 1992; 17(1): 95-109 [abstract]
- Kress J, Pohlman A, Hall J. Sedation and Analgesia in the Intensive Care Unit. Am J Respir Crit Care Med 2002; 166: 1024-1028 [citation]
- Kress J, Pohlman A, O'Connor M, Hall J. Daily Interruption of sedative infusions in critically ill patients undergoing mechanical ventilation. N Eng J Med 2000; 342:1471-1477 [abstract]
- Kollef M, Shapiro S, Silver P, St. John R, Prentice D, Sauer S, Ahrens T, Shannon W, Baker-Clinkscale D. A randomized, controlled trail of protocol-directed versus physician-directed weaning from mechanical ventilation. Crit Care Med 1997; 25(4):567-574 [abstract]
- Ely EW, Meade M, Haponik E, et al. Mechanical ventilator weaning protocols driven by non-physician health care professionals: evidence-based clinical practice guidelines. Chest 2001; 120(suppl 6): 454S-463S [abstract]
Last Updated: April 25, 2003