Desconexión protocolizada versus no protocolizada para la reducción de la duración de la ventilación mecánica invasiva en pacientes pediátricos graves
Resumen
Antecedentes
La ventilación mecánica es un componente crítico del tratamiento de cuidados intensivos pediátricos. Se indica cuando la respiración espontánea del paciente es inadecuada para mantener la vida. La desconexión es la reducción gradual del apoyo ventilatorio y la transferencia del control respiratorio nuevamente al paciente. La desconexión puede representar gran parte del período ventilatorio. La asistencia respiratoria prolongada se asocia con morbilidad, costos hospitalarios, riesgos psicosociales y físicos significativos para el niño e incluso la muerte. La desconexión oportuna y efectiva puede reducir la duración de la ventilación mecánica y puede reducir la morbilidad y la mortalidad asociadas con la asistencia respiratoria prolongada. Sin embargo, no se ha logrado consenso sobre los criterios que se pueden utilizar para identificar cuándo los pacientes están preparados para la desconexión o la mejor manera de lograrla.
Objetivos
Evaluar los efectos de la desconexión protocolizada en los pacientes pediátricos graves con ventilación mecánica invasiva. Comparar la duración total de la ventilación mecánica invasiva en pacientes pediátricos graves que se desconectaron mediante los protocolos versus los que se desconectaron mediante la práctica habitual (no protocolizada). Evaluar cualquier diferencia entre la desconexión protocolizada y la atención habitual en cuanto a mortalidad, eventos adversos, duración de la hospitalización en la unidad de cuidados intensivos y calidad de vida.
Métodos de búsqueda
Se hicieron búsquedas en el Registro Cochrane Central de Ensayos Controlados (Cochrane Central Register of Controlled Trials, CENTRAL; The Cochrane Library, número 10, 2012), MEDLINE (1966 hasta octubre de 2012), EMBASE (1988 hasta octubre de 2012), CINAHL (1982 hasta octubre de 2012), ISI Web of Science y LILACS. Se identificaron datos no publicados en la Web of Science (1990 hasta octubre de 2012), ISI Conference Proceedings (1990 hasta octubre de 2012) y en Cambridge Scientific Abstracts (desde el principio hasta octubre de 2012). Se estableció contacto con los autores principales de los estudios incluidos en la revisión para obtener más información sobre estudios no publicados o en curso. Se realizaron búsquedas en las listas de referencias de todos los estudios identificados y artículos de revisión para obtener estudios pertinentes adicionales. No hubo restricciones de idioma ni de publicación.
Criterios de selección
Se incluyeron los ensayos controlados aleatorios que compararon la desconexión protocolizada (realizada por profesionales o por computadora) versus la práctica de desconexión no protocolizada realizada en niños mayores de 28 días y menores de 18 años de edad.
Obtención y análisis de los datos
Dos revisores examinaron de forma independiente los títulos y los resúmenes identificados mediante la búsqueda electrónica. Tres revisores recuperaron y evaluaron las versiones de texto completo de los estudios potencialmente pertinentes, extrajeron los datos de forma independiente y evaluaron el riesgo de sesgo.
Resultados principales
En el análisis se incluyeron tres ensayos con bajo riesgo de sesgo con 321 niños. La desconexión protocolizada redujo significativamente el tiempo total de asistencia respiratoria en el ensayo más grande (260 niños) en una media de 32 horas (intervalo de confianza [IC] del 95%: 8 a 56; p = 0,01). Otros dos ensayos (30 y 31 niños, respectivamente) informaron reducciones no significativas con una diferencia de medias de ‐88 horas (IC del 95%: ‐228 a 52; p = 0,2) y ‐24 horas (IC del 95%: ‐10 a 58; p = 0,06). La desconexión protocolizada redujo significativamente el tiempo de desconexión en estos dos ensayos más pequeños para una reducción media de 106 horas (IC del 95%: 28 a 184; p = 0,007) y 21 horas (IC del 95%: 9 a 32; p < 0,001). Estos estudios no informaron efectos significativos para la duración de la ventilación mecánica antes de la desconexión, la duración de la estancia en la unidad de cuidados intensivos pediátrica (UCIP) ni hospitalaria, la mortalidad en la UCIP ni los eventos adversos.
Conclusiones de los autores
Pruebas limitadas indican que los protocolos de desconexión reducen la duración de la ventilación mecánica, pero las pruebas no son adecuadas para mostrar si lograr una asistencia respiratoria más corta mediante la desconexión protocolizada tiene efectos beneficiosos o perjudiciales en los niños.
PICO
Resumen en términos sencillos
Utilidad de los protocolos para la reducción del tiempo que los niños están con ventilación mecánica en la unidad de cuidados intensivos
En una unidad de cuidados intensivos pediátrica, la ventilación mecánica se utiliza para ayudar a los niños a respirar cuando se encuentran muy enfermos y la respiración espontánea es inadecuada para mantenerlos con vida. No obstante, si se utiliza durante períodos largos, la ventilación mecánica puede causar problemas. La asistencia respiratoria se asocia con complicaciones como la lesión pulmonar inducida por la asistencia respiratoria, neumonía, complicaciones de la sedación y recuerdos negativos de la experiencia. Por este motivo, es importante reconocer el momento en que el niño se recuperó suficientemente para comenzar a respirar por sí mismo y reducir (o desconectar) la asistencia respiratoria. Lamentablemente, no se ha logrado un acuerdo sobre la mejor manera de desconectar a los niños del respirador artificial.
En los adultos, los investigadores han estudiado la utilidad de protocolos estandarizados que ayuden a guiar a los médicos y enfermeras en los cuidados intensivos a desconectar a los pacientes del respirador artificial de una manera segura y oportuna. El objetivo de esta revisión Cochrane fue analizar los estudios de protocolos de desconexión en niños para ver si se pueden establecer conclusiones con respecto a su utilidad en niños.
Se encontraron tres estudios controlados aleatorios que analizaron a 321 niños mayores de 28 días y menores de 18 años. Los estudios fueron de buena calidad y se realizaron en Brasil, Canadá y Estados Unidos. El estudio más grande mostró que la desconexión protocolizada redujo la duración del tiempo con ventilación mecánica un promedio de 32 horas; los otros dos estudios no mostraron un efecto significativo. Dos estudios informaron reducciones significativas del tiempo desde el comienzo hasta el final de la desconexión del respirador artificial. Los protocolos de desconexión no afectaron el período de tiempo que el niño permaneció en la unidad de cuidados intensivos ni en el hospital, ni afectaron el número de complicaciones asociadas con la ventilación mecánica.
En dos estudios, los participantes representaron a una población amplia de niños en cuidados intensivos, aunque estos estudios no incluyeron a niños a los que se les realizó cirugía del corazón ni con enfermedad neuromuscular crónica, cardíaca o pulmonar. El tercer estudio solamente incluyó a niños con neumonía, bronquiolitis y síndrome de dificultad respiratoria aguda. Los estudios incluidos utilizaron diversos criterios para establecer la disposición para la desconexión y los protocolos siguieron diferentes abordajes para el proceso de desconexión. Estos estudios tuvieron un riesgo de sesgo bajo o incierto.
Pruebas limitadas indican que los protocolos de desconexión reducen la duración de la ventilación mecánica, pero las pruebas no son suficientes para demostrar si lograr una asistencia respiratoria más breve mediante la desconexión protocolizada tiene efectos beneficiosos o perjudiciales en los niños.
Authors' conclusions
Background
The most frequent cause of acute respiratory failure in infants and children leading to admission to a paediatric intensive care unit (PICU) is bronchiolitis with pneumonia (Randolph 2002). Mechanical ventilation is an important component of critical care (Byrd 2010). A pressurized volume of air is delivered via a tracheal tube or by tracheostomy, mask or nasopharyngeal tube. Evidence is insufficient to show the best ventilation modes in critically ill children (Duyndam 2011). Prolonged mechanical ventilation is associated with morbidity and mortality; as a result, clinical and research efforts have focused on early identification of weaning readiness to reduce unnecessary delays.
Description of the condition
Weaning consists of the gradual reduction of ventilatory support and the transfer of respiratory control and the work of breathing back to the patient, resulting in discontinuation of mechanical ventilation (Byrd 2010; Hess 2001; Intensive Care Society 2007). The most common ventilator weaning modes used in weaning children are pressure support ventilation, volume support ventilation, synchronized intermittent mandatory ventilation (Randolph 2002; Wolfler 2011) and a spontaneous breathing trial (Farias 2001). In pressure support mode, the level of pressure is adjusted to achieve acceptable respiratory parameters followed by gradual weaning to minimal pressure support. Volume support is an automated mode whereby the amount of pressure support required to maintain a pre‐set tidal volume is reduced automatically as respiratory mechanics improve. Synchronized intermittent mandatory ventilation is a combination mode by which patients receive mandatory (set) breaths synchronized with their breathing efforts and according to a pressure‐ or volume‐selected mode. Patients breathe spontaneously with pressure support between ventilator breaths; this results in patient‐ventilator synchrony. In this mode, weaning often involves combined reduction of all of the above. A spontaneous breathing trial involves allowing the child to breathe spontaneously on minimal pressure support or through a T‐piece attached to the ventilatory circuit. Each approach may be managed with or without written protocols, or with partial or fully automated ventilator loop algorithms. Investigators have tried to determine how to most effectively wean and extubate patients (Clement 1996; Ely 2001; Fortenberry 2009; Hubble 2000; Kahn 1996). Unfortunately no current, standard method is used to wean children (Newth 2009), and weaning practices vary even within the adult population (Blackwood 2010).
More than 50% of ventilated children are extubated within 48 hours of admission (Newth 2009), often without weaning; up to 50% of unplanned extubations are successful (Little 1990). Weaning refers to a gradual withdrawal of ventilatory support through a stepwise process, rather than extubation from full ventilatory support (Blackwood 2010; Cook 2000). For some children weaning may take weeks or months, and a few remain ventilator‐dependant.
Prolonged intubation and ventilation in children can compromise the child's comfort, feeding and mobility (Hoskote 2005). Furthermore, the requirement for continued sedation and risks of accidental extubation, vocal cord dysfunction, subglottic stenosis, ventilator‐induced lung injury and nosocomial pneumonia can present important morbidity (Hoskote 2005; Newth 2009). No less important are the well‐documented psychological sequelae, which include children's memories of pain and anxiety associated with inability to communicate and with the endotracheal tube (Noyes 2000; Playfor 2000); parental experiences of stress and emotional intensity (Latour 2011; Noyes 1999; Pooni 2013); and post‐traumatic stress disorder for both child and parents (Colville 2012). Ventilation is a life‐saving intervention; however, if unnecessarily prolonged, the child is needlessly exposed to these risks. Therefore, safely minimizing the duration of invasive mechanical support is an important goal of critical care medicine (MacIntyre 2001).
Extubation, which is defined as removal of the endotracheal tube, is a separate but closely related aspect of care (Alia 2000; Byrd 2010; Newth 2009). Concerns that must be addressed before extubation include level of consciousness, respiratory muscle strength, haemodynamic stability and airway oedema or trauma (Walters 2008). Once a patient has achieved a low level of ventilatory support and is capable of sustaining independent spontaneous breathing, his or her ability to safely maintain the airway should be assessed (Byrd 2010). Patients may require additional respiratory assistance after extubation, often in the form of non‐invasive positive‐pressure ventilation. Although non‐invasive ventilation is recognized as a form of mechanical ventilation, its place in weaning protocols has yet to be fully determined (Leclerc 2010).
Description of the intervention
Weaning protocols aim to safely and efficiently liberate patients from mechanical ventilation, reducing unnecessary or harmful variations in approach (Ely 2001). A protocol is defined by the United Kingdom National Health Service Institute as, "descriptions of the steps taken to care for and treat a patient..." enabling "...staff to put evidence into practice by addressing the key questions of what should be done, when, where and by whom at a local level" (NHS Institute 2010). A weaning protocol generally consists of an assessment of the patient's readiness to wean that is based on objective measurement of his or her clinical stability (cardiovascular and metabolic status) and adequate oxygenation, pulmonary function and mental status. This is followed by a method of removing or reducing support. One method involves undertaking a spontaneous breathing trial to identify patients ready for extubation (Farias 1998) that has been shown in children to be equally effective when performed with a T‐piece or with pressure support (Farias 2001). Another method involves following an algorithm outlining a step‐wise reduction in ventilatory support using pressure support ventilation or synchronized intermittent mandatory ventilation (Kollef 1997). Several automated, closed loop weaning systems have become commercially available that propose to wean in real time in accordance with patient ventilatory status (Rose 2008).
A written protocol requires the vigilance and compliance of the clinician in the process, whereas an automated protocol changes the level of support provided by the ventilator in accordance with a computerised algorithm to decrease support and in some cases perform an automated spontaneous breathing trial (Lellouche 2006). An advantage of automated systems is that they may reduce adherence difficulties associated with paper‐based protocols (Rose 2008). Notwithstanding their potential benefits, weaning protocols have attracted criticism. In a review of the evidence for protocols for weaning and sedation management, Girard 2008 provided a sound rationale for the use of weaning protocols, but equally these investigators found that protocol applicability and efficacy remained a source of controversy. The clinical decision to wean or discontinue mechanical ventilation has traditionally been based on clinician judgement and experience (Sahn 1973); protocols may be perceived as removing clinical judgement and hindering consideration of all facets of the care of the participants involved, thereby creating resentment and frustration among healthcare professionals (Ely 2001). It may be argued that most protocols are not replicable because of their dependence on bedside clinician judgement for many decisions, which then become tacitly incorporated into the protocol. Because these judgements occur in a variable manner, it may not be possible to fully describe the protocol rules (Morris 2007).
How the intervention might work
Protocolized weaning, an intervention used by clinicians, may affect the duration of mechanical ventilation in a number of ways. First, it may reduce unwanted variability in weaning practice. In part because of their different experiences, skills and philosophies, clinicians may wean patients differently. Protocols are generally developed consensually by expert groups within the intensive care unit (ICU). They are intended to exemplify best practice, to provide guidelines and thus to reduce needless variation, thereby improving effectiveness and efficiency (Murtagh 2007). Second, weaning often is not considered early enough in the course of ventilation, and a protocol that incorporates assessment for readiness to wean will direct attention to patient readiness. Third, weaning protocols also have the potential to enable non‐medical healthcare professionals to lead or have responsibility in weaning from ventilation: this may reduce unnecessary delays in the weaning process due to limited availability of physicians (Blackwood 2010). Thus, using protocols to guide weaning may encourage best practice through timely recognition of readiness to wean, and adoption of effective weaning processes, so reducing risks and costs associated with unnecessary time on the ventilator.
Why it is important to do this review
A systematic review and meta‐analysis of 11 trials in adults indicated that protocolized weaning significantly reduced the duration of mechanical ventilation, weaning duration and ICU length of stay without adverse effects (Blackwood 2010). However, research evidence from studies of adult participants may not apply for children as children have a dynamic respiratory physiology, affected by growth demands and vulnerable to damage; they are not little adults (WHO 2008). With the growing interest amongst clinicians in developing weaning protocols, it is important to ensure that practice is evidence based and safe. Consequently, our review will rigorously and systematically examine the evidence concerning benefits and harms of protocols to wean children from mechanical ventilation. This is essential in guiding decisions on whether or not to adopt weaning protocols as a quality improvement measure.
Objectives
To assess the effects of weaning by protocol on invasively ventilated critically ill children.
To compare the total duration of invasive mechanical ventilation of critically ill children who are weaned using protocols versus those weaned through usual (non‐protocolized) practice.
To ascertain any differences between protocolized weaning and usual care in terms of mortality, adverse events, ICU length of stay and quality of life.
Methods
Criteria for considering studies for this review
Types of studies
We included in the review randomized controlled trials (RCTs) that compared protocolized with non‐protocolized weaning.
Types of participants
We included studies of children (younger than 18 years old) who were cared for in a PICU and were mechanically ventilated via a nasal or oral tracheal tube.
We excluded studies of neonates (from birth to 28 completed days after birth) (WHO 2010)) because of differences in their ventilation and weaning strategies (Alander 2013) and because a Cochrane systematic review of protocolized weaning in neonates is under way (Wielenga 2013).
We excluded studies in which children were ventilated exclusively via non‐invasive techniques or tracheostomy.
Types of interventions
We included studies that evaluated protocolized weaning compared with non‐protocolized weaning. For the purpose of this review, protocolized weaning was defined as the use of an algorithm (paper based or automated) intended to result in removal of children from invasive mechanical ventilation. Non‐protocolized weaning was defined as usual care, standard practice or clinician‐led care that incorporated any non‐protocolized practice.
Types of outcome measures
Primary outcomes
Duration of mechanical ventilation (MV), measured in hours.
Secondary outcomes
-
Weaning duration (from identification of weaning readiness to invasive MV discontinuation).
-
MV before weaning.
-
PICU and hospital length of stay.
-
PICU and hospital mortality.
-
Quality of life as defined by the authors.
-
Adverse events (such as re‐initiation of MV within 48 hours of removal, tracheostomy, self‐extubation or re‐admission within 48 hours).
Search methods for identification of studies
Electronic searches
We searched the literature using the standard strategy of the Cochrane Anaesthesia Review Group of the Cochrane Collaboration. We searched the Cochrane Central Register of Controlled trials (CENTRAL) (The Cochrane Library, Issue 10, 2012); MEDLINE In‐Process and other Non‐Indexed Citations and OVID MEDLINE (1946 to 22 October 2012); CINAHL Plus via EBSCO host (1982 to 22 October 2012); EMBASE via OVID (1980 to 22 October 2012); LILACS (1982 to 22 October 2012); unpublished data－Web of Science (1990 to 22 October 2012) and ISI Conference Proceedings (1990 to 22 October 2012). We did not restrict language of publication.
We used a specific search strategy for each database with descriptors that included synonyms for ventilator weaning, clinical protocols and randomized controlled trials; reflecting the clinical condition, intervention and research design, respectively. Search strategies for each database can be found in the appendices (Appendix 1: MEDLINE; Appendix 2: CINAHL; Appendix 3: EMBASE; Appendix 4: LILACS; Appendix 5: CENTRAL; Appendix 6: Web of Science).
Searching other resources
In our efforts to obtain grey literature, we searched reference lists of included studies; contacted authors of included studies by electronic mail for information; searched the major clinical trials registries (ProQuest; www.ClinicalTrials.gov) and searched for theses (www.theses.com).
Data collection and analysis
Selection of studies
Three review authors (MM, POH, BB) independently scanned identified titles and abstracts and excluded records that did not meet eligibility requirements. We obtained full‐text copies of potentially relevant studies.
Data extraction and management
Three review authors (AC, BB, MM) independently extracted data from the included studies using a piloted paper form (Appendix 7). We extracted information about study design, study setting and participants, inclusion and exclusion criteria, interventions and outcomes. We also collected information on sources of funding for the study and on ethical approval. Furthermore, we collected information, where available, regarding physician and nurse staffing numbers and sedation strategies as these can influence ventilator weaning (Hansen 2009; Playfor 2006). After independent data extraction, we met to resolve any disagreement through discussion and consultation. We did not require additional arbitration by a fourth review author (POH).
Assessment of risk of bias in included studies
Risk of bias in included studies was assessed independently by the same three review authors (AC, BB, MM) using the domain‐based evaluation described in the Cochrane Handbook for Systematic Reviews of Interventions, Chapter 8, version 5.1.0 (Higgins 2011). The 'Risk of bias' form (Appendix 7) extracted from Chapter 8.5.1 was used to evaluate each included study, and the review authors' judgements were directed by the criteria set out in Chapter 8.5.3 and Table 8.5c. Each study was judged as 'Yes' (low risk of bias), 'Unclear' (uncertain risk of bias) or 'No' (high risk of bias) for the following domains.
-
Random sequence generation.
-
Allocation concealment.
-
Blinding (of participants, personnel and outcome assessors).
-
Incomplete outcome data.
-
Selective reporting.
-
Free of other bias.
We categorized the risk of bias in all included studies according to the following:
-
Low risk of bias (plausible bias unlikely to seriously alter the results) if all criteria were met;
-
Unclear risk of bias (plausible bias that raises some doubt about the results) if one or more criteria were assessed as unclear; or
-
High risk of bias (plausible bias that seriously weakens confidence in the results) if one or more criteria were not met.
These assessments are reported in the 'Risk of bias' tables in the review (Appendix 7). We also discuss in the review result section the impact of methodological quality on the results.
Measures of treatment effect
We planned to combine data using RevMan 5.2, when appropriate, by intervention, outcome and population.
Unit of analysis issues
The child was the unit of analysis in each trial. Children were randomly allocated to one of two parallel intervention groups, and a single measurement for each outcome from each participant was collected and analysed.
Dealing with missing data
Where necessary, we contacted the first author of included studies to obtain data.
Assessment of heterogeneity
Clinical heterogeneity was judged by the review authors (MM, AC), and these results are noted in the review. We planned to investigate heterogeneity by conducting subgroup analyses defined by type of PICU, protocol and approach to delivery.
Assessment of reporting biases
Studies were insufficient to allow the review authors to explore small study effects.
Data synthesis
Data were entered into RevMan (RevMan 5.2) by BB and were checked independently by MM. For the primary outcome (duration of mechanical ventilation), data were reported differently: median with 95% confidence interval (CI) using Kaplan‐Meier survival curves (Foronda 2011); mean and standard deviation (SD) (Jouvet 2013) or median and interquartile range (IQR) (Maloney 2007). Foronda 2011 supplied raw data to enable us to calculate the mean and SD. For the Maloney 2007 study, we approximated the mean using the median, and approximate SD estimates were calculated from the IQR, as suggested in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). Ventilation data from all three studies had skewed distributions. Whilst one study (Foronda 2011) provided raw data, the other two did not and would require approximations to calculate the mean and SD on the log scale before the meta‐analysis was performed. It was unclear how well these approximations would perform, particularly as two studies had small numbers (Jouvet 2013; Maloney 2007); therefore we did not conduct a meta‐analysis. Results from each study are presented in tables, along with mean differences and 95% CIs. If further trials are identified in the future, we will calculate pooled estimates of the difference in means and risk ratios (RRs) using the fixed‐effect model (FEM) or the random‐effects model (REM), depending on the degree of heterogeneity.
Subgroup analysis and investigation of heterogeneity
Studies were insufficient for review authors to conduct subgroup analyses.
Sensitivity analysis
Studies were insufficient for review authors to conduct sensitivity analyses.
Results
Description of studies
The studies were RCTs conducted with mechanically ventilated children older than 28 days.
Results of the search
The electronic searches identified a total of 10,983 citations: 9891 from electronic databases and 1092 from additional records. Three review authors (MM, POH, BB) reviewed these citations and listed eight studies for possible inclusion. Full papers for these citations were retrieved. Where necessary, the authors were contacted to clarify whether their study met inclusion criteria for our review. A flow diagram detailing the selection of studies is presented in Figure 1.
Review flow diagram.
Included studies
We included three RCTs conducted on mechanically ventilated children in PICUs. The intervention groups were weaned from mechanical ventilation in accordance with written or automated weaning protocols. The control groups were weaned by healthcare professionals without the use of written, formal guidelines.
Participants and settings
These studies analysed 321 children in two published papers (Foronda 2011; Jouvet 2013) and one thesis (Maloney 2007). Details are provided in the Characteristics of included studies table. The trials were conducted in Sao Paulo, Brazil (Foronda 2011), Montreal, Canada (Jouvet 2013), and Salt Lake City, Utah, United States (Maloney 2007). Trial sample sizes ranged from 30 to 260, and participants were recruited from mixed patient population paediatric (Foronda 2011; Jouvet 2013) or cardiothoracic (Maloney 2007) PICUs. The age range of participants was 28 days to 18 years. None of the studies provided information on physician and nurse staffing in the units, and none provided information on their sedation strategies or sedation protocols.
Interventions
Children were allocated to different interventions at different times in each study (see Appendix 8 for details). Weaning interventions included daily evaluation for readiness to wean and a spontaneous breathing trial (Foronda 2011); a computerized protocol using a commercially available closed‐loop system, SmartCare/PSTM (Jouvet 2013) and a non－commercially available computerized decision support tool and weaning protocol (Maloney 2007). Randomization to groups was conducted at different time points in the three trials: before meeting readiness to wean criteria (Foronda 2011); after passing a 30‐minute pressure support test (Jouvet 2013) and after two consecutive reductions in ventilator support (Maloney 2007). Only Foronda 2011 described usual care in which the most frequently used modes were pressure support, synchronized intermittent mandatory ventilation and pressure‐controlled ventilation; and weaning consisted of reductions in respiratory frequency, peak inspiratory pressure, fraction of inspired oxygen and positive end‐expiratory pressure as determined by the presence of ventilatory parameters.
Studies pending classification
Two studies (Randolph 2002; Schultz 2001) met the inclusion criteria but their study samples included neonates. All authors were contacted to ascertain whether the data for children and neonates could be separated for analysis. We are awaiting this information.
Excluded studies
Three studies were excluded. Two studies (Restrepo 2004, Oliveria 2002) did not meet the eligibility criteria, and one study (Rushforth 2004) included only three children. Details are presented in the Characteristics of excluded studies table.
Risk of bias in included studies
We assessed the risk of bias using the domain‐based evaluation of risk of bias tool of the Cochrane Collaboration (Higgins 2011). Low or unclear risk was identified across all six domains. Our judgement on the classification of bias for individual studies is presented in the Characteristics of included studies table and is summarized in Figure 2.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Allocation
All studies used adequate methods for random sequence generation and allocation concealment: Foronda 2011 and Jouvet 2013 used random selection of opaque sealed envelopes and Maloney 2007 used computer‐generated randomization.
Blinding
In the Foronda 2011 study, medical personnel were blinded to allocation up until the point at which the participant passed the spontaneous breathing trial, indicating low risk of performance bias. In the other two studies (Jouvet 2013; Maloney 2007), blinding was not possible, and it is unclear whether this produced performance bias. In all three studies, blinding of outcome assessors was not reported, and therefore risk of detection bias was unclear.
Incomplete outcome data
Data on recruitment and attrition were reported, and no evidence of attrition bias was found in the three studies.
Selective reporting
A trial protocol was registered by Foronda 2011 and Jouvet 2013, and no evidence of selective reporting was found. Maloney 2007 did not register a protocol but reported usual outcomes for trials in this area, so we assessed the risk of reporting bias as low.
Other potential sources of bias
We found no other potential sources of bias in included studies.
Effects of interventions
All study authors were contacted to confirm and supplement information related to methods and data, when needed. Results are reported for each outcome.
Primary analysis: Comparison of protocolized versus non‐protocolized weaning
Primary outcome: Total duration of mechanical ventilation (hours)
All three studies reported the review’s primary outcome, which was the total duration of mechanical ventilation (see Table 1). In all studies, this outcome was defined as initiation of mechanical ventilation to extubation. Jouvet 2013 further defined the endpoint of this outcome as including subsequent ventilation episodes if reintubation occurred within 48 hours of extubation; the endpoint in the other two studies was time to first extubation. All three studies reported results favouring protocolized weaning, but only the largest of the three trials (Foronda 2011) (with 260 participants) showed a statistically significant mean (95% CI) reduction of 32 (8 to 56) hours (P = 0.01). Jouvet 2013 reported a mean difference of ‐88 (‐228 to 52) hours (P = 0.2); and Maloney 2007 reported a difference of ‐24 (‐10 to 58) hours (P = 0.06).
| Study | Protocolized weaning N mean (SD) | Non‐protocolized weaning N mean (SD) | Difference in mean | 95% CI | P‐value | ||
| 134 | 111 (85) | 126 | 143 (107) | ‐32 | ‐55.58 to ‐8.42 | 0.01± | |
| 15 | 200 (186) | 15 | 288 (206) | ‐88 | ‐228.46 to 52.46 | 0.20§ | |
| 15 | 93.6 (27)* | 16 | 117.8 (64)* | ‐24.2† | ‐10.0 to 58.4 | 0.055|| | |
* Standard deviation approximated from the interquartile range; † difference in median; ± from t‐test; § from Mann Whitney U‐test; || from Mann Whitney t‐test; NR not reported
Secondary outcomes
Two studies (Jouvet 2013; Maloney 2007) reported secondary outcomes relevant to the review. These are presented in Table 2. A statistically significant mean reduction in weaning duration was reported in the protocolized weaning group for Jouvet 2013 (106 hours, 95% CI 28 to 184, P = 0.007) and Maloney 2007 (21 hours, 95% CI 9 to 32, P < 0.001). Both studies defined weaning duration as initiation of weaning to extubation, but each study used different criteria for determining the start and endpoint of this outcome (see Appendix 8 for details). No significant differences in outcomes between protocolized and non‐protocolized weaning groups were reported for duration of MV before weaning or for PICU and hospital length of stay. No study reported quality of life.
| Study | Protocolized weaning | Non‐protocolized weaning | |||||
| N | Mean (SD) | N | Mean (SD) | Mean difference | 95% CI | P‐value | |
| Duration of mechanical ventilation before weaning (hours) | |||||||
| 15 | 157 (189) | 15 | 141 (104) | ‐16 | ‐125.17 to 93.17 | 0.89 | |
| 15 | 74.5 (39.3)* | 16 | 84 (53.3)* | 9.5 | ‐23.33 to 42.33 | 0.50 | |
| Weaning duration (hours) | |||||||
| 15 | 36 (36) | 15 | 142 (150) | 106 | 27.94 to 184.06 | 0.007 | |
| 15 | 8 (9.3)* | 16 | 28.5 (22.2)* | 20.5 | 8.65 to 32.35 | <0.001 | |
| PICU length of stay (hours) | |||||||
| 15 | 216 (120) | 15 | 696 (504) | 480 | 217.82 to 742.18 | 0.11 | |
| 15 | 176 (64) | 16 | 217 (114) | 41 | ‐23.57 to 105.57 | 0.23 | |
| Hospital length of stay (hours) | |||||||
| 15 | 648 (432) | 15 | 696 (504) | 48 | ‐287.93 to 383.93 | 0.68 | |
| 15 | 312 (88) | 16 | 436 (338) | 124 | ‐47.50 to 295.50 | 0.18 | |
* Standard deviation approximated from the interquartile range.
Adverse events
Adverse events are presented in Table 3. Foronda 2011; Jouvet 2013 and Maloney 2007 reported no significant differences in reintubation and self‐extubation rates. Jouvet 2013 reported one death in PICU in the automated group and none in the control group. Foronda 2011 reported no significant differences in PICU mortality between groups, with 23 (14.8%) and 15 (10.8%) deaths, respectively, reported in the protocol and control groups. Most deaths occurred before weaning; only two deaths per group occurred after weaning (personal communication). Foronda 2011 and Jouvet 2013 reported no significant differences in the use of non‐invasive ventilation post extubation. Foronda 2011 reported no significant differences in ventilator‐associated pneumonia, and Jouvet 2013 reported no significant differences in prolonged mechanical ventilation. No study reported hospital mortality.
| Study | Protocolized weaning n/N (%) | Non‐protocolized weaning n/N (%) | Risk Ratio | 95% CI | P‐value |
| PICU Mortality | |||||
| 23/155 (14.8) | 15/139 (10.8) | 1.44 | 0.72 to 2.89 | 0.30 | |
| 1/15 (6.7) | 0/15 (0.0) | 3.0 | 0.13 to 68.26 | NR | |
| Reintubation | |||||
| 15/134 (11.2) | 18/126 (14.3) | 0.78 | 0.41 to 1.49 | 0.45 | |
| 2/15 (13.3) | 1/15 (6.7) | 2.0 | 0.2 to 19.78 | NR | |
| 2/15 (13.3) | 3/16 (12.5) | 0.71 | 0.14 to 3.68 | 1.0 | |
| Self‐extubation | |||||
| 3/134 (2.2) | 8/126 (6.3) | 0.35 | 0.1 to 1.3 | 0.10 | |
| 1/15 (6.7) | 0/15 (0.0) | 3.0 | 0.13 to 68.26 | NR | |
| 0 | 0 | NE | NE | NE | |
| Non‐invasive ventilation post extubation | |||||
| 29/134 (21.6) | 39/126 (31.0) | 0.7 | 0.46 to 1.06 | 0.09 | |
| 1/15 (6.7) | 2/15 (13.3) | 0.5 | 0.05 to 4.94 | NR | |
| Ventilator‐associated pneumonia | |||||
| 9/123 (6.7) | 12/126 (9.5) | 0.77 | 0.34 to 1.76 | 0.41 | |
| Prolonged MV | |||||
| 0/15 (0.0) | 2/15 (13.3) | 0.20 | 0.01 to 3.85 | NR | |
NE, not estimable; NR, not reported.
Discussion
A thorough search of the literature identified five studies that could potentially be included in our review. Two studies (Randolph 2002; Schultz 2001) included a proportion of neonates (17% and unknown proportion, respectively); the authors were unable to provide us with disaggregated data. Furthermore, Randolph 2002 included two weaning protocol groups (using pressure support and volume support ventilation) and one control group, and we were unable to obtain and combine intervention group data. Consequently, only three studies were included in the review. Ventilation outcome data in these studies were skewed and consequently require conversion to the log scale for meta‐analysis; these approximations are complicated by small numbers in two studies (Jouvet 2013; Maloney 2007), and this made meta‐analysis inadvisable. As a result, the review findings cannot provide sufficient strength of evidence to demonstrate benefit or harm.
Summary of main results
Only the largest study (Foronda 2011) showed a significant effect on the total duration of mechanical ventilation. In this study, the protocolized weaning group received a daily evaluation of readiness to wean and a two‐hour spontaneous breathing trial; duration of ventilation was reduced by a mean (95% CI) of 32 (8 to 56) hours. Using a SmartCare/PSTM automated system and a computerized weaning protocol, respectively, Jouvet 2013 and Maloney 2007 showed a statistically significant reduction in weaning duration in the protocolized groups by 106 and 21 hours, respectively, which is promising; however, this reduction did not significantly reduce total mechanical ventilation time or PICU or hospital length of stay. Foronda 2011 reported no significant differences in PICU mortality, reintubation, self‐extubation or use of non‐invasive ventilation after extubation. Adverse events and deaths were too few in the two smaller studies for the review authors to draw significant conclusions. Because of the small number of studies included in the review and inability to pool the data, we are not able to provide a meaningful summary of findings table.
Overall completeness and applicability of evidence
All three included studies included a population of children with respiratory conditions; therefore these studies are applicable to the general PICU, where respiratory disorders are the main cause of respiratory failure necessitating mechanical ventilation (Newth 2009). Jouvet 2013 was the only study that included postoperative surgical and trauma participants; additionally, this study group restricted participant age, as the SmartCare/PSTM automated system currently is not licensed for children younger than two years of age. The average age of children admitted to the PICU is seven months (Farias 2012), which explains the current lack of trials using this weaning method. All studies excluded children with complex conditions such as primary pulmonary hypertension, cyanotic heart disease and neuromuscular disease, which are associated with prolonged mechanical ventilation (Polito 2011); therefore the impact of protocolized weaning on prolonged mechanical ventilation in these groups is unknown.
Conclusions cannot be drawn on the effectiveness of specific weaning methods, as each study used a different approach to protocolized weaning. Methods included a professional‐led approach with a daily evaluation of readiness to wean, and with those meeting set criteria undergoing a two‐hour spontaneous breathing trial; an automated closed loop system with automatic adjustment of pressure support and physician‐led adjustment of positive end‐expiratory pressure and a computer‐driven protocol that automatically analysed data relevant to the participant's respiratory performance, formulated a recommended change in ventilator support and transmitted a paged message to a respiratory therapist to manually adjust settings in accordance with a protocol. Resources available to individual PICUs, including availability of computerized systems, may place restrictions on the choice of weaning method.
Context‐related information such as physician and nurse staffing, sedation strategies and sedation protocols are known to cause delays in the weaning process (Brattebø 2002; Marcin 2005). None of the studies provided contextual information; consequently, the influence of these factors on study outcomes cannot be assessed.
Quality of the evidence
The three studies included sample sizes ranging from 30 to 260 and involved 321 randomized children. Methodological quality of the studies was high. We assessed the largest study (Foronda 2011) as having low risk of bias in all domains of the domain‐based evaluation of risk of bias tool of the Cochrane Collaboration (Higgins 2011) and the two smaller studies as having low or unclear risk across all six domains. Blinding of the intervention is not feasible in studies comparing a weaning protocol with usual care; however, Foronda 2011 was able to conceal participant assignment up until the child passed the daily evaluation of weaning readiness and a spontaneous breathing trial was indicated, thus removing potential performance bias.
Potential biases in the review process
We adhered closely to our protocol, which outlined our procedures for minimising bias in the review: these included independent screening for trial inclusion, data extraction and assessment of risk of bias by three review authors. With the assistance of the Cochrane Anaesthesia Group's Search Trials Co‐ordinator and an experienced librarian, we conducted a thorough search strategy and believe we have identified all relevant studies.
Agreements and disagreements with other studies or reviews
This is the first published systematic review of trials comparing protocolized weaning with usual care in critically ill children in intensive care.
Review flow diagram.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
| Study | Protocolized weaning N mean (SD) | Non‐protocolized weaning N mean (SD) | Difference in mean | 95% CI | P‐value | ||
| 134 | 111 (85) | 126 | 143 (107) | ‐32 | ‐55.58 to ‐8.42 | 0.01± | |
| 15 | 200 (186) | 15 | 288 (206) | ‐88 | ‐228.46 to 52.46 | 0.20§ | |
| 15 | 93.6 (27)* | 16 | 117.8 (64)* | ‐24.2† | ‐10.0 to 58.4 | 0.055|| | |
| * Standard deviation approximated from the interquartile range; † difference in median; ± from t‐test; § from Mann Whitney U‐test; || from Mann Whitney t‐test; NR not reported | |||||||
| Study | Protocolized weaning | Non‐protocolized weaning | |||||
| N | Mean (SD) | N | Mean (SD) | Mean difference | 95% CI | P‐value | |
| Duration of mechanical ventilation before weaning (hours) | |||||||
| 15 | 157 (189) | 15 | 141 (104) | ‐16 | ‐125.17 to 93.17 | 0.89 | |
| 15 | 74.5 (39.3)* | 16 | 84 (53.3)* | 9.5 | ‐23.33 to 42.33 | 0.50 | |
| Weaning duration (hours) | |||||||
| 15 | 36 (36) | 15 | 142 (150) | 106 | 27.94 to 184.06 | 0.007 | |
| 15 | 8 (9.3)* | 16 | 28.5 (22.2)* | 20.5 | 8.65 to 32.35 | <0.001 | |
| PICU length of stay (hours) | |||||||
| 15 | 216 (120) | 15 | 696 (504) | 480 | 217.82 to 742.18 | 0.11 | |
| 15 | 176 (64) | 16 | 217 (114) | 41 | ‐23.57 to 105.57 | 0.23 | |
| Hospital length of stay (hours) | |||||||
| 15 | 648 (432) | 15 | 696 (504) | 48 | ‐287.93 to 383.93 | 0.68 | |
| 15 | 312 (88) | 16 | 436 (338) | 124 | ‐47.50 to 295.50 | 0.18 | |
| * Standard deviation approximated from the interquartile range. | |||||||
| Study | Protocolized weaning n/N (%) | Non‐protocolized weaning n/N (%) | Risk Ratio | 95% CI | P‐value |
| PICU Mortality | |||||
| 23/155 (14.8) | 15/139 (10.8) | 1.44 | 0.72 to 2.89 | 0.30 | |
| 1/15 (6.7) | 0/15 (0.0) | 3.0 | 0.13 to 68.26 | NR | |
| Reintubation | |||||
| 15/134 (11.2) | 18/126 (14.3) | 0.78 | 0.41 to 1.49 | 0.45 | |
| 2/15 (13.3) | 1/15 (6.7) | 2.0 | 0.2 to 19.78 | NR | |
| 2/15 (13.3) | 3/16 (12.5) | 0.71 | 0.14 to 3.68 | 1.0 | |
| Self‐extubation | |||||
| 3/134 (2.2) | 8/126 (6.3) | 0.35 | 0.1 to 1.3 | 0.10 | |
| 1/15 (6.7) | 0/15 (0.0) | 3.0 | 0.13 to 68.26 | NR | |
| 0 | 0 | NE | NE | NE | |
| Non‐invasive ventilation post extubation | |||||
| 29/134 (21.6) | 39/126 (31.0) | 0.7 | 0.46 to 1.06 | 0.09 | |
| 1/15 (6.7) | 2/15 (13.3) | 0.5 | 0.05 to 4.94 | NR | |
| Ventilator‐associated pneumonia | |||||
| 9/123 (6.7) | 12/126 (9.5) | 0.77 | 0.34 to 1.76 | 0.41 | |
| Prolonged MV | |||||
| 0/15 (0.0) | 2/15 (13.3) | 0.20 | 0.01 to 3.85 | NR | |
| NE, not estimable; NR, not reported. | |||||
