Monitorización con BIS versus evaluación clínica para la sedación en adultos con ventilación mecánica en la unidad de cuidados intensivos y la repercusión sobre los resultados clínicos y la utilización de recursos
Resumen
Antecedentes
Los pacientes que ingresan a la unidad de cuidados intensivos y que son sometidos a ventilación mecánica, reciben fármacos sedativos y analgésicos para mejorar tanto su comodidad como la interacción con el respirador. La optimización de la práctica de sedación puede reducir la mortalidad, mejorar la comodidad del paciente y reducir el costo. La práctica actual incluye el uso de escalas o puntuaciones para evaluar la profundidad de la sedación basada en criterios clínicos como la conciencia, la comprensión y la respuesta a los comandos. Sin embargo, las mismas son percibidas como herramientas de evaluación subjetivas. Los monitores del índice biespectral (BIS), que se basan en el procesamiento de las señales electroencefalográficas, pueden superar las restricciones de las escalas de sedación y proporcionar una orientación más fiable y consistente para el aumento gradual de la profundidad de la sedación.
Los beneficios de la monitorización con BIS de los pacientes bajo anestesia general para los procedimientos quirúrgicos ya han sido confirmados por otra revisión Cochrane. Al llevar a cabo una revisión sistemática bien realizada, el objetivo fue determinar si la monitorización con BIS mejora los resultados en los pacientes adultos de la unidad de cuidados intensivos (UCI) sometidos a ventilación mecánica.
Objetivos
Evaluar los efectos de la monitorización con BIS en comparación con la evaluación clínica de la sedación sobre la estancia en la UCI, la duración de la ventilación mecánica, la mortalidad por cualquier causa, el riesgo de neumonía asociada al respirador (NAR), el riesgo de eventos adversos (p.ej. autoextubación, desconexión no planificada de los catéteres permanentes), la duración de la estancia hospitalaria, la cantidad de agentes sedativos utilizados, el costo, los resultados funcionales a más largo plazo y la calidad de vida según lo informado por los autores para los adultos con ventilación mecánica en la UCI.
Métodos de búsqueda
Se realizaron búsquedas en CENTRAL, MEDLINE, Embase, CINAHL, ProQuest, OpenGrey y SciSearch hasta mayo de 2017, se verificaron las referencias bibliográficas y se estableció contacto con los autores de los estudios para identificar estudios adicionales. Se hicieron búsquedas en registros de ensayos, que incluyeron clinicaltrials.gov y controlled‐trials.com.
Criterios de selección
Se incluyeron todos los ensayos controlados aleatorios que comparaban el BIS versus evaluación clínica (EC) para el control de la sedación en adultos con enfermedades graves sometidos a ventilación mecánica.
Obtención y análisis de los datos
Se utilizaron los procedimientos metodológicos estándar Cochrane. Se realizaron análisis utilizando el programa Revman 5.3.
Resultados principales
Se identificaron 4245 posibles estudios a partir de la búsqueda inicial. De esos estudios, cuatro (256 participantes), cumplieron los criterios de inclusión. Otro estudio está a la espera de clasificación. Los estudios fueron realizados en UCI quirúrgicos y médicos‐quirúrgicos combinados de centro único. El monitor BIS se usó para evaluar el nivel de sedación en el brazo de intervención en todos los estudios. En el brazo de control, las herramientas de evaluación de la sedación para la EC incluyeron la Sedation‐Agitation Scale (SAS), la Ramsay Sedation Scale (RSS) o la EC subjetiva que utiliza los signos clínicos tradicionales (frecuencia cardíaca, presión arterial, nivel de consciencia y tamaño de las pupilas). Sólo un estudio se clasificó como en riesgo bajo de sesgo, los otros tres estudios se clasificaron como en alto riesgo.
No hubo evidencia de una diferencia en un estudio (N = 50) que midió la duración de la estancia en la UCI (mediana (Rango Intercuartil; IQR) 8 (4 a 14) en el grupo de EC; 12 [6 a 18] en el grupo de BIS; evidencia de baja calidad). Hubo poco o ningún efecto sobre la duración de la ventilación mecánica (DM ‐0,02 días [IC del 95%: ‐0,13 a 0,09; dos estudios; N = 155; I2 = 0%; evidencia de baja calidad)). Se informaron eventos adversos en un estudio (N = 105) y los efectos sobre la agitación después de la succión, la resistencia al tubo endotraqueal, la tolerancia al dolor durante la sedación o el delirio después de la extubación fueron inciertos debido a la evidencia de muy baja calidad. Los eventos adversos clínicamente relevantes como la autoextubación no se informaron en ningún estudio. Tres estudios informaron la cantidad de agentes sedativos usados. No fue posible medir la diferencia combinada en la cantidad de agentes sedativos usados debido a los diferentes protocolos de sedación y los agentes sedativos usados en los estudios. La calidad de la evidencia GRADE fue muy baja. Ningún estudio informó otros resultados secundarios de interés para la revisión.
Conclusiones de los autores
Se encontró evidencia insuficiente acerca de los efectos de la monitorización con BIS para la sedación en adultos con enfermedades graves sometidas a ventilación mecánica sobre los resultados clínicos o la utilización de recursos. Los resultados son inciertos debido a la evidencia de calidad baja y muy baja derivada de un número limitado de estudios.
PICO
Resumen en términos sencillos
Comparación de la monitorización con BIS versus evaluación clínica para determinar el nivel de sedación en adultos con ventilación mecánica en las unidades de cuidados intensivos
Pregunta de la revisión
Se examinó la evidencia sobre los beneficios de la monitorización del índice biespectral (BIS, por sus siglas en inglés) en comparación con los métodos de evaluación clínica (EC) en adultos conectados a un respirador artificial en la unidad de cuidados intensivos (UCI).
Antecedentes
La monitorización con BIS sigue la actividad eléctrica del cerebro para producir puntuaciones. Dichas puntuaciones pueden ayudar al personal del hospital a decidir si un paciente de la UCI sometido a ventilación mecánica está recibiendo suficientes agentes sedativos como para estar cómodo y aceptar el respirador. Los sedantes son fármacos administrados por sus efectos calmantes e inductores del sueño. La administración de demasiados agentes sedativos, o muy pocos, podría dar lugar a efectos perjudiciales. En el método de EC, la observación de los factores clínicos como la conciencia, la comprensión y la respuesta a los comandos ayuda a evaluar la profundidad de la sedación o el sueño. La puntuación proporcionada por el monitor del BIS no depende de una persona. La monitorización con la EC podría variar entre los cuidadores.
El objetivo fue determinar si la monitorización con BIS es beneficiosa en comparación con la EC en adultos con enfermedades graves sometidos a ventilación mecánica.
Características de los estudios
La evidencia identificada en la búsqueda bibliográfica está actualizada hasta mayo de 2017. Cuatro estudios controlados aleatorios cumplieron los criterios de inclusión para esta revisión (con la participación de 256 adultos). Otro estudio está a la espera de clasificación. Estos estudios se realizaron en UCI quirúrgicos y médicos‐quirúrgicos combinados para adultos y compararon la monitorización con BIS con diversas medidas para la EC.
Fuentes de financiación de los estudios
Para un estudio, el fabricante de dispositivos de monitorización con BIS proporcionó el equipo. La empresa no tuvo ninguna función en la realización del estudio. Otro estudio fue financiado como parte de un proyecto científico y tecnológico. No había información disponible sobre financiación para los otros dos estudios.
Resultados clave
Con la monitorización con BIS, no se encontró ninguna diferencia significativa en la estancia en la UCI (un estudio, 50 adultos), la duración de la ventilación (dos estudios, 155 adultos) y el riesgo de eventos adversos (un estudio, 105 adultos) en comparación con la EC. No se informaron eventos adversos clínicamente relevantes, p.ej. la autoextracción accidental del tubo respiratorio. No fue posible medir la diferencia combinada en la cantidad de uso de sedativos debido a los diferentes protocolos de sedación y los sedantes utilizados. No se informó ninguno de los otros resultados de interés para la revisión, p.ej. la muerte, la neumonía asociada al respirador, la calidad de vida, etc., en ninguno de los estudios.
Calidad de la evidencia
Los hallazgos de la revisión provienen de un número limitado de estudios que proporcionaron evidencia de calidad “baja a muy baja” según GRADE.
Conclusión
Los autores de esta revisión concluyen que se encontró evidencia insuficiente acerca de los efectos de la monitorización con BIS en comparación con la EC de la sedación en adultos con enfermedades graves sometidos a ventilación mecánica.
Authors' conclusions
Summary of findings
| BIS monitoring compared to clinical assessment for sedation in mechanically ventilated adults in the intensive care unit and its impact on clinical outcomes and resource utilization | ||||||
| Patient or population: Mechanically ventilated adults in the intensive care unit | ||||||
| Outcomes | Anticipated absolute effects* | Relative effect | № of participants | Quality of the evidence | Comments | |
| Risk with Clinical assessment | Risk with BIS monitoring | |||||
| Intensive care unit length of stay (ICU LOS) | Median ICU LOS was 8 Days | Median ICU LOS was 4 Days higher | Mdn D 4 [Range 4 to 18] | 50 | ⊕⊕⊕⊝ | |
| Duration of mechanical ventilation (measured in days) | Mean duration of mechanical ventilation was 2.49 days | Mean duration of mechanical ventilation was 0.02 days lower | MD ‐0.02 (‐0.13, 0.09) | 155 | ⊕⊕⊝⊝ | |
| Adverse events: Measured as number of patients with adverse events | 105 (1 RCT) | ⊕⊝⊝⊝ | Clinically relevant adverse events such as self‐extubation or unplanned disconnection of indwelling catheters were not reported in any study. | |||
| 809 patients with restlessness after suction per 1000 patients | 16 less patients with restlessness after suction | RR 1.11 (0.90,1.37) | ||||
| 714 patients with endotracheal tube resistance per 1000 patients | 32 more patients with endotracheal tube resistance | RR 0.96 (0.75, 1.22) | ||||
| 928 patients with pain tolerance during sedation per 1000 patients | 8 more patients with pain tolerance during sedation | RR 0.99 (0.89, 1.10) | ||||
| 47 patients with delirium after extubation per 1000 patients | 32 less patients with delirium after extubation | RR 3 (0.28, 32.04) | ||||
| Other important secondary outcomes like Any‐cause mortality, ventilator‐associated pneumonia, hospital LOS, amount of sedative agents used, long term functional outcomes and quality of life were not reported in any studies | ||||||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 Downgraded two levels due to very serious concerns about imprecision (very small sample size of the study and large confidence interval). 2 Downgraded two levels due to serious concerns about risk of bias (Zhao 2011 which carries 98.3% weight for this outcome, Random sequence generation, Allocation concealment and selective reporting were graded as unclear risk of bias) and imprecision (Difference in duration of mechanical ventilation was less than one day which is clinically insignificant). 3 Downgraded three levels due to serious concerns about risk of bias (Random sequence generation, Allocation concealment and Selective reporting were assessed as unclear risk of bias), indirectness (Clinically relevant adverse events were not reported) and imprecision (Small number of patients in the study Zhao 2011). | ||||||
Background
Description of the condition
A significant proportion of the patients admitted to an intensive care unit (ICU) undergo mechanical ventilation (Esteban 2002; Metnitz 2009). It is common practice to administer sedative and analgesic drugs to these patients, to improve their comfort and their interaction with the ventilator. Different sedative and analgesic drugs are used for this purpose (Gommers 2008; Patel 2012). Careful titration of analgesia and sedation is important to prevent pain and discomfort in this population of patients, but oversedation has been associated with increased mortality and morbidity (Kollef 1998; Kress 2000). Optimizing sedation practice may reduce mortality, and may reduce the duration of mechanical ventilation and ICU length of stay, resulting in reduced costs and improved resource utilization (Jackson 2010). The recommended strategy to titrate sedation is to use scales or scores based on clinical criteria (Jacobi 2002). Many sedation tools have been developed, but not all have been validated and tested in clinical practice (Barr 2013). There is variability in the specific domains (e.g. consciousness, cognition, and comprehension) they assess (Sessler 2008), and in their implementation (about 88% of units use a sedation scale, with variability in the sedation scale used) (Martin 2007; Reschreiter 2008; Soliman 2001). Furthermore, these scales are perceived to provide a subjective assessment of patient sedation, also their usefulness in patients receiving neuromuscular blocking medications or requiring deep sedation may be limited.
Description of the intervention
With the aim to overcome the restraints of the subjective sedation scales, many techniques and devices (e.g. Bispectral Index (BIS) monitoring, State Entropy (SE), Auditory evoked potentials (AEPs), Narcotrend Index (NI), Patient State Index (PSI)) have been developed with the purpose of providing an objective measurement of patient's sedation (Carrasco 2000). The BIS monitoring is possibly the most studied and adapted.
BIS monitoring is based on the processing of electroencephalographic signals from the brain. The device uses three or four electrodes applied to the patient's forehead. The electrodes record the raw electroencephalogram (EEG) signal and process it through a proprietary algorithm, producing a dimensionless number, ranging from zero to 100, where 90 to100 indicates a state of wakefulness and zero represents absence of brain electrical activity. BIS monitoring is available in different hardware and software versions (LeBlanc 2006). The set up and maintenance cost of BIS monitoring is quite high. The monitor cost is around USD 6500.00 and a sensor, which includes four electrodes costs around USD 25.00 per set (Sedation Equipment & Supplies 2017), but this cost may be offset by a reduction in the usage of sedative drugs. In one study, titration of sedation with BIS monitoring in ICU patients resulted in an 18% reduction in cost over two months period (about USD 150.00 per patient) mainly as a result of reduction in lorazepam, midazolam and propofol usage (Kaplan 2000).
BIS monitoring is quite well established for monitoring anaesthesia depth (Punjasawadwong 2014), but there are differences in patient characteristics in critical care compared to anaesthesia. A critical care patient's brain may be abnormal. Delirium and neurological impairment are extremely common in the intensive care setting (Singhal 2014). Sepsis is often characterized by an acute brain dysfunction (Sonneville 2013). There are several other conditions that can also cause encephalopathy in critical care patients (Fugate 2013; Hu 2014; Ma 2013; Stevens 2008; Ziaja 2013). The effect of hypoglycaemia (low blood sugar level), temperature, nerve‐muscle electrical activity and drugs such as catecholamines on BIS monitoring scores might vary (Barr 2013; LeBlanc 2006). Also, there are already well‐established validated clinical sedation scores, such as the Richmond Agitation Sedation scale (RASS) and Sedation Agitation Scale (SAS) available in critical care, hence it is not clear if BIS monitoring in critically ill patients is equally as effective as in anaesthesia.
How the intervention might work
Significant under‐sedation occurs using subjective analysis of sedation in the ICU (Kaplan 2000). BIS monitoring has been reported to be better than clinical assessment (CA) methods for ICU patients undergoing short‐term mechanical ventilation in terms of reduction in the amount of sedative use and time to wakefulness (Zhao 2011). It has also been reported that BIS monitoring can reliably differentiate between inadequate and adequate sedation (Karamchandani 2010); helps in faster emergence and improved recovery from sedation; and reduces recall phenomenon thereby, reducing the posttraumatic stress disorder (PTSD) (Kaplan 2000). When compared with four commonly used subjective clinical scales (Ramsay Sedation Scale (RSS), RASS, SAS and Adaptation to Intensive Care Environment scale), BIS monitoring showed significant correlation with all the scales (Yaman 2012). In another study comparing BIS monitoring with RASS in mechanically ventilated critically ill patients, BIS monitoring correlated well with RASS (Karamchandani 2010). With the production of an objective measurement in the form of a dimensionless number, BIS monitoring might be able to overcome some of the limitations of the subjective clinical sedation scales and provide a more reliable and consistent guidance for the titration of sedation in ICU.
Why it is important to do this review
The benefits of BIS monitoring in patients undergoing general anaesthesia for surgical procedures have been confirmed by a Cochrane review (Punjasawadwong 2014). The use of BIS monitoring in intensive care has many advantages. Using BIS monitoring to guide sedative administration would allow optimizations of drug delivery to the needs of the individual patients in order to avoid unnecessary deep or light sedation. Compared to CA, BIS monitoring can distinguish between lightly and deeply sedated patients (Dewhurst 2000). It has a special role in critically ill brain injured patients with or without sedation (Deogaonkar 2004). It has also been reported to reduce consumption of sedative drugs (Kaplan 2000). All this may lead to reduced duration of mechanical ventilation, ICU length of stay, hospital length of stay and ultimately result in cost saving. Although several studies have evaluated the use of BIS monitoring in the ICU, there are only two systematic reviews that have been undertaken to establish its benefit for ICU patients (Finger 2016; Bilgili 2017). However both of these reviews included studies where sedation monitoring based on CA was used in both the intervention and control arm (i.e. BIS monitoring and CA versus CA alone). By undertaking a well‐conducted systematic review we aim to answer the question, does the use of BIS monitoring alone compared to clinical sedation assessment lead to improvement in clinical outcomes and resource utilisation.
Objectives
To assess the effects of BIS monitoring compared with clinical sedation assessment on intensive care unit (ICU) length of stay (LOS), duration of mechanical ventilation, any cause mortality, risk of ventilator‐associated pneumonia (VAP), risk of adverse events (e.g. self‐extubation, unplanned disconnection of indwelling catheters), hospital LOS, amount of sedative agents used, cost, longer‐term functional outcomes as reported by authors and quality of life as reported by authors for mechanically ventilated adult study participants in the ICU.
Methods
Criteria for considering studies for this review
Types of studies
We included all randomized controlled trials (RCTs) comparing BIS monitoring versus clinical assessment (CA) for the management of sedation in mechanically ventilated critically ill adults, regardless of language and publication status.
We planned to include cluster‐randomized trials in our review but none were identified .
Non‐randomized and quasi‐randomized trials were not eligible for inclusion because of the significant risk of bias.
Cross‐over trials were also not eligible for inclusion because this methodology is not suitable for investigating the intervention topic of our study.
Types of participants
We included trials involving adults undergoing mechanical ventilation in ICUs, irrespective of the admission diagnosis.
Types of interventions
The intervention group comprised all participants whose sedation was managed by a strategy based on BIS monitoring with, or without, the use of a protocol to titrate the sedation level. The control group included all participants whose sedation was managed by monitoring with any clinical method (using clinical judgement or a specific clinical sedation scoring tool), with or without the use of a titration protocol.
Types of outcome measures
Primary outcomes
-
Intensive care unit (ICU) length of stay (LOS), measured in days.
Secondary outcomes
-
Duration of mechanical ventilation, measured in days.
-
Any‐cause mortality.
-
Risk of ventilator‐associated pneumonia (VAP).
-
Risk of adverse events (e.g. self‐extubation, unplanned disconnection of indwelling catheters).
-
Hospital LOS in days.
-
Amount of sedative agents used. (See Differences between protocol and review).
-
Cost.
-
Longer‐term functional outcomes as reported by study authors.
-
Quality of life as reported by study authors using SF36 or similar tools.
Search methods for identification of studies
Electronic searches
We searched the latest issue of the Cochrane Central Register of Controlled Trials (CENTRAL, Issue 6 of 12, June 2017; Appendix 1), MEDLINE (Ovid SP, from 1994 to May 2017 Appendix 2), Embase (Ovid SP, from 1994 to May 2017; Appendix 3) and the Cumulative Index to Nursing and Allied Health Literature (CINAHL) (EBSCOhost, from 1994 to May 2017; Appendix 4).
We searched the databases from 1994 onwards, because BIS monitor was introduced by Aspect Medical Systems, Inc. (Norwood, Massachusetts, USA) for the first time in 1994.
In the relevant databases (MEDLINE and Embase) the sensitivity‐maximizing strategy was applied as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).
We adopted our ProQuest search strategy in searching all other databases (Appendix 5).
We also searched clinicaltrials.gov, controlled‐trials.com and other national and regional registries for ongoing trials.
We did not impose any language restrictions.
Searching other resources
In addition to searches of electronic databases;
-
we searched OpenGrey for Information on grey literature (up to June 2017);
-
screened the reference lists of all eligible trials and relevant reviews;
-
undertook cited reference searching using SciSearch (up to June 2017);
-
identified relevant studies published in dissertations or theses by searching ProQuest Dissertations and Theses database (up to June 2017);
-
we tried to contact experts in the field and the manufacturer of the device, however we did not receive any response from them.
Data collection and analysis
Selection of studies
We merged the results of the searches (described above) using reference management software, and removed all duplicates.
Two review authors (RS, AB) independently examined the titles and abstracts of identified studies and removed obviously irrelevant reports. We (RS, AB) were not blinded to any details of the published study. After this first screening process, we (RS, AB) compared our results and were able to resolve disagreements by discussion. In cases of inability to reach a consensus, we consulted a third review author (RJ).
We produced a list of potentially relevant studies. The same two review authors independently assessed studies for potential inclusion in the review by using the Cochrane Anaesthesia, Critical and Emergency Review Group's (ACE's) study selection and data extraction form (Appendix 6). We independently noted the reasons for exclusion.
We resolved disagreements in study selection by discussion. In cases of inability to reach a consensus, we consulted a third review author (AK). We contacted the journal/ corresponding author of the relevant studies for additional data or clarifications.
We compiled a list of all eligible studies, along with a list of excluded studies.
Data extraction and management
Two review authors (RS, AB) extracted data independently according to the predetermined criteria provided on the ACE study selection and data extraction form (Appendix 6). If any relevant data were missing, we contacted the first author or corresponding author of the study to obtain this information. Data extraction or translation from studies of languages other than English were undertaken by Cochrane experts arranged by the Cochrane Anaesthesia, Critical and Emergency Review Group. One Japanese article (Inaba 2007), was translated and data extracted by two Japanese speaking healthcare professionals in addition to the Cochrane organized expert.
We (RS, AB) resolved disagreements by discussion. If we were unable to reach an agreement, we consulted the third review author (AK).
We collected the following information about study context where available.
-
Country where the study was conducted.
-
Number of beds in the hospital.
-
Number of beds in the Intensive care unit (ICU).
-
Number of admissions to the ICU per year.
-
Nurse‐to‐patient ratio.
-
Type of ICU (medical, surgical, cardiac, neurological, trauma, burn).
-
Type of sedation used in both groups, as well as dose and total amount given.
-
Whether paralytics were used in both groups.
-
Confounders: drugs (e.g. catecholamines, aminophylline), electromyography (EMG), sleep, temperature, hypoglycaemia, excessive muscle movement, etc.
-
Diagnosis.
-
Severity of illness scoring.
Assessment of risk of bias in included studies
Two review authors (RS, AK) independently assessed risk of bias using the Cochrane 'Risk of bias' tool (Higgins 2011). We were not blinded to the names of the study authors, institutions, journal and results. We judged the quality of studies on the basis of risk of bias in the following domains.
-
Selection bias.
-
Random sequence generation.
-
Allocation concealment.
-
-
Detection bias.
-
Blinding of outcome assessors.
-
Blinding of personnel.
-
-
Attrition bias.
-
Incomplete outcome data.
-
-
Reporting bias.
-
Selective reporting.
-
We classified studies as low risk, high risk or unclear risk of bias for the above domains using information available from the studies. We considered a study as having low risk of bias if all domains (except blinding of personnel, as blinding is not possible because of the nature of the study), were assessed as adequate (low risk). We considered a study as having high risk of bias if one or more domains (except blinding of personnel) were assessed as inadequate (high or unclear risk), and as having an unclear risk if insufficient detail of what happened in the study was reported. Primary analysis was planned to be restricted to studies at low risk of bias. We planned to perform a sensitivity analysis excluding studies assessed as having high risk of bias. We (RS, AK) resolved any cases of disagreement about classification of risks by discussion. If we were unable to reach an agreement, we planned to consult a third review author (MH), however this was not required.
We constructed a 'Risk of bias' table as part of the 'Characteristics of included studies,' a 'Risk of bias' summary figure (Figure 1) and a 'Risk of bias' graph (Figure 2), with details of all judgements made for all studies included in the review. For the 'Risk of bias' table, we have provided a text box that includes a description of the design, conduct or observations that underline the judgement.
'Risk of bias' summary: review authors' judgements about each risk of bias item for each included study.
'Risk of bias' graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Measures of treatment effect
We undertook analysis using RevMan 5.3 software.
For continuous outcomes (duration of mechanical ventilation), we presented the treatment effect as a mean difference (MD). ICU LOS is presented as median with range as only one study reported this outcome (Weatherburn 2007) and it was reported as median. For dichotomous outcomes (risk of adverse events), we presented treatment effect as a risk ratio (RR). We presented effect estimates along with 95% confidence intervals (CI).
Unit of analysis issues
We included in our review only randomized controlled trials with a parallel‐group design. The issue of repeated measures is not relevant for the outcomes under investigation.
We planned, if the review included cluster‐randomized studies, to perform a sensitivity analysis that excludes cluster‐randomized studies to determine the impact of including them in the analysis. Our search did not find any cluster‐randomized trials.
Dealing with missing data
We performed quantitative analysis on an intention‐to‐treat (ITT) basis and planned to contact the study authors for missing data. Data for Zhao 2011, was converted from hours to days and the standard deviations (SD) calculated from the reported 95% CI.
Assessment of heterogeneity
We had planned not to perform meta‐analysis if we suspected important clinical heterogeneity on examination of the included studies. We used the Chi2statistic to test statistical heterogeneity between studies and considered a P value ≤ 0.10 as indicating significant heterogeneity; we used the I2 statistic to assess the magnitude of heterogeneity (Higgins 2002). We considered an I2 > 50% would indicate problematic heterogeneity between studies and in such case we would carefully consider the value of any pooled analysis. We planned to use a random‐effects model analysis if an I2 was greater than 30%. We planned to use a fixed‐effect model of analysis to determine the best estimate of the intervention effect. If the two did not coincide, we would not consider the random‐effects estimate as the actual intervention effect in the population under study. We constructed forest plots to summarize findings from the included studies.
Assessment of reporting biases
We undertook a comprehensive electronic search and a search of other sources such as trial registries, as described above, to minimize the effects of publication bias. We planned to construct a contour‐enhanced funnel plot to differentiate asymmetry due to publication bias. As we had less than 10 studies, funnel plots of effect estimates against their standard errors (on a reversed scale) were not created as per the guideline.
Data synthesis
We quantitatively reviewed the included data and combined the data by intervention, outcome and population using the Cochrane's statistical software (Revman 5.3). We synthesized the data only in the absence of important clinical or statistical heterogeneity, and we expressed pooled estimates of the mean difference for continuous variables and risk ratios for proportions.
We planned to use the inverse‐variance fixed‐effect method of meta‐analysis for continuous variables. For studies reporting median and range, we took estimation of the mean and standard deviation using the method described by Hozo and colleagues (Hozo 2005).
Had we identified cluster‐randomized studies, we planned to determine whether the results had been correctly analysed by using an appropriate method such as a multi‐level mode, variance component analysis or generalized estimating equations (GEEs). Had this been done, we would have included in the meta‐analysis the effect estimates from these studies and their standard errors.
If substantial heterogeneity was present, and if sufficient studies were available, we planned to perform a random‐effects meta‐analysis.
We have presented the results in the form of a forest plot.
Subgroup analysis and investigation of heterogeneity
When appropriate, with obvious clinical or statistical (I2 > 50%) heterogeneity, we planned to consider subgroup analysis based on participants with neurological injury, including:
-
head injury;
-
cardiopulmonary bypass; and
-
use of neuromuscular blocking agents.
if the data had indicated heterogeneity on that basis, patients with neurological injury were excluded from our selected studies. Not enough data were available to undertake subgroup analysis based on patients on cardiopulmonary bypass or the use of neuromuscular blocking agents.
Sensitivity analysis
We planned to perform sensitivity analyses to explore the consistency of effect size measures in studies with low risk of bias versus those with high risk of bias. We did not perform a sensitivity analysis, as there were not enough studies included in the review.
'Summary of findings' table and GRADE
We present study findings in a standard 'Summary of findings' table (summary of findings Table for the main comparison), which includes a list of all important outcomes; a measure of the typical burden of these outcomes; the absolute and relative magnitude of effect; the numbers of participants and studies addressing each outcome and a grade for the overall quality of the body of evidence for each outcome.
We used the principles of the GRADE system (Guyatt 2008) to assess the quality of the body of evidence associated with specific outcomes (intensive care unit length of stay, duration of mechanical ventilation and risk of adverse events (e.g. self‐extubation, unplanned disconnection of indwelling catheters)) and constructed summary of findings Table for the main comparison using GRADE software. The GRADE approach appraises the quality of a body of evidence according to the extent to which one can be confident that an estimate of effect or association reflects the item being assessed. The quality of the body of evidence considers within‐study risk of bias (methodological quality), directness of the evidence, heterogeneity of the data, precision of effect estimates and risk of publication bias.
Results
Description of studies
Results of the search
We identified 4245 possible studies from the initial search. From these studies we identified seven potentially relevant studies and retrieved them for further assessment (Figure 3).
Study flow diagram.
Included studies
Of the seven identified studies, we included four trials with 256 participants (Inaba 2007; Li 2009; Weatherburn 2007; Zhao 2011) that fulfilled the inclusion criteria and compared Bispectral Index (BIS) versus clinical assessment (CA) method in monitoring sedation in adult mechanically ventilated Intensive care unit (ICU) participants. We excluded two studies because sedation monitoring was based on CA in addition to BIS monitoring in the intervention group and hence did not fit with the aim of our review (Binnekade 2009; Olson 2009). One study is awaiting classification (Ou 2016). In all the included studies, sedation was assessed with BIS monitoring in the intervention group. BIS monitoring was assessed hourly in all studies but one (Li 2009), where it was assessed four times in a 48‐hour period. In the control group, sedation was assessed using a variety of methods. In Inaba 2007, the Ramsay score was used, in Zhao 2011 , the Sedation Agitation Scale (SAS) was used, and in Li 2009, both the SAS and the Ramsay score were used. In Weatherburn 2007, sedation assessment was conducted clinically, based on heart rate, blood pressure, conscious level and pupillary size. In the control group, frequency of sedation assessment was conducted hourly in Inaba 2007 and Zhao 2011, four times in an 48‐hour period in Li 2009, and not reported in Weatherburn 2007.
Participants and settings
We reported full participant details in the Characteristics of included studies. All were single‐centre studies. Inclusion and exclusion criteria were fairly similar across studies. Main differences included study sample size (ranging from 18 (Inaba 2007), to 105 (Zhao 2011)), age (39.3 years in Zhao 2011, and 53 years in Weatherburn 2007), and duration of mechanical ventilation (immediate postoperative period in Inaba 2007 and longer than 12 hours in Weatherburn 2007 and Zhao 2011). Trials were conducted in different parts of the world; China (Li 2009; Zhao 2011), Japan (Inaba 2007), and Australia (Weatherburn 2007). Three of the four studies were published in languages other than English: two in Chinese (Zhao 2011; Li 2009), and one in Japanese (Inaba 2007).
Interventions
Intervention was sedation titration based on BIS monitoring. Target BIS score varied between studies; it was 40 to 70 in Inaba 2007, greater than 70 in Weatherburn 2007, 50 to 70 in Zhao 2011. Target BIS score was not mentioned in the Li 2009 study. There were large differences in the sedation protocol used in different studies. Both sedative drugs and administration methods varied. In Inaba 2007, fentanyl and propofol were administered as an infusion, in Li 2009, midazolam was given both as boluses and infusion, propofol and midazolam infusion were given in Zhao 2011. In Weatherburn 2007, morphine and midazolam were given, however the exact protocol was not described.
Control group
The same sedatives were given in the control group compared to intervention group in all the studies with similar bolus and infusion protocols. In Inaba 2007, the target Ramsay score was four to five, in Li 2009, the target SAS was three to four, but the target for Ramsay score was not described. In Zhao 2011, the target SAS was three to four. In Weatherburn 2007, the target for sedation with CA was not described. Muscle relaxants were used in both groups in Li 2009; no information was available about use of paralytics in other studies.
Funding sources
Funding sources for Weatherburn 2007 included Abbott Australasia and manufacturers of the device. Authors reported that funders of the study had no role in the study concept, design, data collection, data analysis, data interpretation or writing of the reports. Funding for Li 2009 was from Scientific and technological project Chengdu Sichuan. No information was given about the role of the funders. No information about funding was given for Inaba 2007 and Zhao 2011. Author conflict of interest was not reported in the studies.
Excluded studies
We excluded two studies as sedation monitoring was based on CA in addition to BIS monitoring in the study group and hence did not fit with the aim of our review (Binnekade 2009; Olson 2009) (Characteristics of excluded studies).
Studies awaiting classification
Ou 2016 is only published as an abstract, not enough data are provided for analysis. No contact details were provided for authors. Publishers when contacted did not provide authors' contact details.
Ongoing studies
We found no ongoing studies
Risk of bias in included studies
All studies were randomized controlled trials. Risk of bias has been described in the 'Risk of bias' table for each study (Characteristics of included studies). Figure 1 and Figure 2 summarize the risk of bias within and across studies, respectively.
Allocation
Allocation concealment was classified as 'low risk' in one study (Weatherburn 2007). Allocation concealment was classified as high risk in Inaba 2007 and unclear risk in Li 2009 and Zhao 2011.
Blinding
Because of the nature of the intervention, it was not possible to blind participants and personnel (performance bias). No information was reported about blinding of outcome assessment in any of the studies, but review authors judge that the outcome measurements of interest are unlikely to be influenced by lack of blinding of outcome assessment.
Incomplete outcome data
All four studies were classified as 'low risk' as all the participants completed the study and there was no loss to follow‐up.
Selective reporting
One study was classified as 'low risk' because they had published the protocol (Weatherburn 2007), and the study's pre‐specified (primary and secondary) outcomes were reported. The remaining three studies were classified as 'unclear risk' as we could not find a record in the trials registry.
Effects of interventions
See Summary of findings table 1 (summary of findings Table for the main comparison)
Primary outcomes
1. Intensive care unit (ICU) length of stay (LOS), measured in days
One study reported this outcome (N = 50) (Weatherburn 2007). There was no significant difference in ICU length of stay in days between the two arms of the study (Median (Interquartile Range IQR) 8 (4, 14) in the clinical assessment (CA) group; 12 (6, 18) in the BIS group; P = 0.20). ). The GRADE quality of evidence was downgraded by two levels to low due to concerns about imprecision (because of small size of the study and large confidence interval (CI)).
Secondary outcomes
1. Duration of mechanical ventilation, measured in days
This outcome was reported in two studies (N = 155) (Weatherburn 2007; Zhao 2011) (Analysis 1.1). The pooled analysis showed no effect in the duration of mechanical ventilation between the BIS monitoring group and the CA group (mean difference (MD) ‐0.02 days (95% CI ‐0.13 to 0.09; Chi2 = 0.01; I2= 0%). The GRADE quality of evidence was judged as low due to serious concerns about risk of bias (Zhao 2011, which carries 98.3% weight for this outcome, random sequence generation, allocation concealment and selective reporting were graded as unclear risk of bias) and imprecision (the difference in duration of mechanical ventilation is less than one day which is not clinically significant).
2. Any cause mortality
This outcome was not reported in included studies.
3. Risk of ventilator‐associated pneumonia
This outcome was not reported in included studies.
4. Risk of adverse events
This outcome was reported by only one study (N = 105) (Zhao 2011). The number of patients with adverse events analysed included restlessness after suction, endotracheal tube resistance, pain tolerance during sedation and delirium after extubation. There was no significant difference between the two groups. Restlessness after extubation: risk ratio (RR) 1.11 (95% CI 0.90 to 1.37), endotracheal tube resistance: RR 0.96 (95% CI 0.75 to 1.22), pain tolerance during sedation: RR 0.99 (95% CI 0.89 to 1.10), delirium after extubation: RR 3 (95% CI 0.28 to 32.04), all P > 0.05. The GRADE quality of evidence was downgraded to very low due to serious concerns about risk of bias (random sequence generation, allocation concealment and selective reporting were assessed as unclear risk of bias), indirectness (clinically relevant adverse events were not reported) and imprecision (small number of patients in the study).
Other clinically important adverse events such as self‐extubation and unplanned disconnection of indwelling catheters were not reported.
5. Hospital LOS in days
This outcome was not reported in included studies.
6. Amount of sedative agents used
This outcome was reported in three studies (Inaba 2007; Weatherburn 2007; Zhao 2011, ). We could not pool results because the studies used different sedation protocols and sedative agents. Results are presented in Additional Table 1. The GRADE quality of evidence was judged as very low due to serious concerns about risk of bias (allocation concealment and selective reporting in Zhao 2011, and Inaba 2007 was assessed as either high risk or unclear risk), inconsistency (because of heterogeneity of data) and imprecision (effect estimate of amount of sedative agents used was imprecise).
| Study | BIS group | Clinical assessment group | |||||
| N | Mean (SD) | N | Mean (SD) | Mean difference | 95% CI | P value | |
| Average propofol dose (mg/kg/hour) | 9 | 5.3 (1) | 9 | 5.1 (0.9) | 0.2 | ‐0.68, 1.08 | 0.670 |
| Time to eye opening (minutes) | 9 | 5.7 (5.7) | 9 | 4.1 (2.8) | 1.6 | ‐2.55, 5.75 | 0.771 |
| Time to consciousness (minutes) | 9 | 7.6 (5.3) | 9 | 7.6 (3.6) | 0 | ‐4.19, 4.19 | NA |
| Number of flow rate changes | 9 | 4.4 (2.5) | 9 | 3.6 (1.7) | 0.8 | ‐1.18, 2.78 | 0.779 |
| Number of boluses | 9 | 1.4 (2.3) | 9 | 0.89 (1.4) | 0.51 | ‐1.25,2.27 | 0.719 |
| Mean morphine total daily dosage (mg) | 25 | 22.6* | 25 | 26.6* | 0.67 | ||
| Mean midazolam total daily dosage (mg) | 25 | 18.4* | 25 | 14.6* | 0.85 | ||
| Mean midazolam dose (mg/kg/hour) | 42 | 0.10 (0.02) | 63 | 0.09 (0.02) | 0.01 | 0.00, 0.02 | 0.993 |
| Mean propofol dose (mg/kg/hour) | 42 | 0.95 (0.23) | 63 | 0.86 (0.20) | 0.09 | 0.00, 0.18 | 0.979 |
| Mean time to wake up (minutes) | 42 | 0* | 63 | 15* | <0.05 | ||
* Standard deviation not reported
7. Cost
This outcome was not reported in the included studies.
8. Longer‐term functional outcomes as reported by study authors
This outcome was not reported in included studies.
9. Quality of life as reported by study authors
This outcome was not reported in the included studies.
Discussion
This review includes randomized controlled trials (RCTs) comparing bispectral index (BIS) monitoring versus clinical assessment (CA) for sedation in mechanically ventilated adult intensive care unit (ICU) patients. We collected data on clinically relevant outcomes such as ICU length of stay (LOS), which was the primary outcome and the secondary outcomes such as duration of mechanical ventilation, any‐cause mortality, risk of ventilator‐associated pneumonia (VAP), risk of adverse events, hospital LOS, amount of sedative agents used, cost, longer‐term functional outcomes and quality of life. Data on the primary and secondary end points were available for only ICU LOS, duration of mechanical ventilation, risk of adverse events and amount of sedative agents used.
Summary of main results
Our primary objective was to assess the effect of mode of sedation assessment on ICU LOS. Evidence from one study (Weatherburn 2007), with 50 participants showed no statistically and clinically significant difference between the BIS monitoring and CA group. The GRADE quality of evidence was low for this outcome.
Of our secondary objectives, only duration of mechanical ventilation, risk of adverse events and amount of sedative agents used were reported. Two studies (155 participants) reported the duration of mechanical ventilation (Weatherburn 2007; Zhao 2011), with no significant difference between the groups (GRADE Low quality of evidence). The number of patients with adverse events (restlessness after suction, endotracheal tube resistance, pain tolerance during sedation and delirium after extubation) was reported in only one study (105 participants) (Zhao 2011). There was no statistically significant difference between the two groups (GRADE very low quality of evidence). Adverse events of interest for the review, such as self‐extubation and unplanned disconnection of indwelling catheters, were not reported. Three studies (173 participants) reported the amount of sedative agents used (Inaba 2007; Weatherburn 2007;Zhao 2011). The studies used different sedation protocol and sedative agents; therefore it was not possible to pool results (GRADE very low‐quality of evidence)(Table 1).
Overall completeness and applicability of evidence
Our protocol proposed the following outcomes: ICU LOS, duration of mechanical ventilation, any cause mortality, risk of VAP, risk of adverse events, hospital LOS, amount of sedative agents used, cost, long‐term functional outcomes and quality of life. The outcomes we sought are consistent with the recommended four core areas of outcomes: death, life impact, pathological manifestations, and resource used by other specialties such as rheumatology (The OMERACT Handbook 2014). Most of the studies included in our review did not report many of these outcomes. However some of the outcomes even though reported were not defined (duration of mechanical ventilation), or they used different methods of measurements (sedation) leading to the possibility of inconsistency in outcomes between trials. Development and utilization of core outcome sets (COS) may help to prevent these issues in the future. Several COS for critical care research are still in various stages of development (Blackwood 2015).
There are some outcomes, which were not mentioned in the protocol, but may be of importance for patients on sedation in ICU. Posttraumatic stress disorder (PTSD) is one such example. Systematic review of studies has shown that one‐fifth of general ICU survivors have either substantial PTSD symptoms or clinician‐diagnosed PTSD (Davydow 2008). Another systematic review showed that early post‐ICU memories of in‐ICU frightening or psychotic experiences were associated with increased risk of post‐ICU PTSD in over 80% of the studies that examined this factor (Parker 2015). Therefore PTSD may be a useful outcome to look for in studies assessing depth of sedation monitoring. Delirium and mild cognitive impairment in ICU survivors may be other useful outcome measures.
Quality of the evidence
Our review included four studies with 256 patients. Only one study (Weatherburn 2007) was judged to be at low risk of bias. Other studies were judged to be at high risk of bias. The GRADE quality of evidence ranked from low to very low across the different outcomes. Methodological limitations of the studies included small numbers (256 patients), risk of bias (random sequence generation, allocation concealment and selective reporting), inconsistency (duration of mechanical ventilation not defined) and imprecision (large confidence interval).
External validity of this review may be limited because there was a large heterogeneity in the patient population. Zhao 2011 and Inaba 2007 enrolled patients who were admitted postoperatively and required ventilation for less than 24 hours, whereas Weatherburn 2007 included patients from a mixed medical‐surgical ICU who required ventilation for longer duration of time.
Potential biases in the review process
We followed the guidelines provided in the Cochrane Handbook for Systematic Reviews of Interventions (Cochrane 2008). The eligibility for inclusion and exclusion and assessment for risk of bias was carried out independently by two review authors (RS, AB). In our protocol (Shetty 2014), we stated that we would include all adults (18 years of age or older) undergoing mechanical ventilation in ICU for longer than 24 hours, irrespective of the admission diagnosis. We made two changes to this section. We removed the criterion: "longer than 24 hours" because three of the four included studies otherwise could not fulfil the criteria. We changed "18 years of age or older" to only 'adults' because all of the included studies mentioned adults, but did not provide the exact range and we were unable to obtain additional data from the study authors. Hence the criteria for types of participants now reads "We included all adults undergoing mechanical ventilation in an ICU, irrespective of the admission diagnosis" (Differences between protocol and review).There were no other major departures from the protocol (Shetty 2014), that could have affected our findings or introduced any risk of bias. However difference in duration of mechanical ventilation less than one day is clinically insignificant. Hence inclusion of three more studies with less than 24 hours of mechanical ventilation may not result in clinically significant difference in duration of mechanical ventilation.
Agreements and disagreements with other studies or reviews
Our Cochrane review compared BIS monitoring versus clinical assessment for sedation in mechanically ventilated adult ICU patients. BIS monitoring and clinical assessment versus clinical assessment alone was investigated in two recently published meta‐analysis/systematic reviews (Bilgili 2017; Finger 2016). In these reviews there was no benefit of adding BIS monitoring to clinical assessment. Also ICU LOS was actually better in the control group (mean difference (MD) 1.4; 95% confidence interval (CI) 0.29, 0.5; P = 0.01) indicating addition of BIS monitoring to usual clinical monitoring could be harmful (Finger 2016). In our review median ICU LOS was four days higher in the BIS monitoring group even though this was not statistically significant. We are not aware of any other systematic review or meta‐analysis comparing BIS monitoring versus clinical assessment in this patient group. The American College of Chest Physicians, American College of Critical Care Medicine, Society of Critical Care Medicine, and the American Society of Health System Pharmacists clinical practice guidelines for the sustained use of sedatives and analgesics in the critically ill patient (Barr 2013), recommend that the routine use of BIS is not recommended (moderate quality of evidence rated as strongly against the intervention).
The benefits of BIS monitoring in patients undergoing general anaesthesia for surgical procedures have been confirmed by a Cochrane review (Punjasawadwong 2014). This benefit is not shown in our review. The reason for this may be the difference in level of target sedation (anaesthesia needs deeper level of sedation). Also endpoints are different; the aim in anaesthesia is avoiding awareness, whereas target of ICU sedation is keeping patient alert and calm to lightly sedated and hence the patient is always aware.
There is evidence to show that muscular activity may affect BIS values (Dahaba 2005). The magnitude of BIS overestimation significantly correlates to both BIS and electromyographic activity before neuromuscular blockade (Vivien 2003). BIS monitoring may be a reasonable approach in assessing depth of sedation in ICU patients receiving neuromuscular paralysis. However, no studies so far have looked at outcome benefits in this group of patients.
'Risk of bias' summary: review authors' judgements about each risk of bias item for each included study.
'Risk of bias' graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Study flow diagram.
Comparison 1 Bispectral Index versus Clinical assessment, Outcome 1 Duration of mechanical ventilation.
| BIS monitoring compared to clinical assessment for sedation in mechanically ventilated adults in the intensive care unit and its impact on clinical outcomes and resource utilization | ||||||
| Patient or population: Mechanically ventilated adults in the intensive care unit | ||||||
| Outcomes | Anticipated absolute effects* | Relative effect | № of participants | Quality of the evidence | Comments | |
| Risk with Clinical assessment | Risk with BIS monitoring | |||||
| Intensive care unit length of stay (ICU LOS) | Median ICU LOS was 8 Days | Median ICU LOS was 4 Days higher | Mdn D 4 [Range 4 to 18] | 50 | ⊕⊕⊕⊝ | |
| Duration of mechanical ventilation (measured in days) | Mean duration of mechanical ventilation was 2.49 days | Mean duration of mechanical ventilation was 0.02 days lower | MD ‐0.02 (‐0.13, 0.09) | 155 | ⊕⊕⊝⊝ | |
| Adverse events: Measured as number of patients with adverse events | 105 (1 RCT) | ⊕⊝⊝⊝ | Clinically relevant adverse events such as self‐extubation or unplanned disconnection of indwelling catheters were not reported in any study. | |||
| 809 patients with restlessness after suction per 1000 patients | 16 less patients with restlessness after suction | RR 1.11 (0.90,1.37) | ||||
| 714 patients with endotracheal tube resistance per 1000 patients | 32 more patients with endotracheal tube resistance | RR 0.96 (0.75, 1.22) | ||||
| 928 patients with pain tolerance during sedation per 1000 patients | 8 more patients with pain tolerance during sedation | RR 0.99 (0.89, 1.10) | ||||
| 47 patients with delirium after extubation per 1000 patients | 32 less patients with delirium after extubation | RR 3 (0.28, 32.04) | ||||
| Other important secondary outcomes like Any‐cause mortality, ventilator‐associated pneumonia, hospital LOS, amount of sedative agents used, long term functional outcomes and quality of life were not reported in any studies | ||||||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 Downgraded two levels due to very serious concerns about imprecision (very small sample size of the study and large confidence interval). 2 Downgraded two levels due to serious concerns about risk of bias (Zhao 2011 which carries 98.3% weight for this outcome, Random sequence generation, Allocation concealment and selective reporting were graded as unclear risk of bias) and imprecision (Difference in duration of mechanical ventilation was less than one day which is clinically insignificant). 3 Downgraded three levels due to serious concerns about risk of bias (Random sequence generation, Allocation concealment and Selective reporting were assessed as unclear risk of bias), indirectness (Clinically relevant adverse events were not reported) and imprecision (Small number of patients in the study Zhao 2011). | ||||||
| Study | BIS group | Clinical assessment group | |||||
| N | Mean (SD) | N | Mean (SD) | Mean difference | 95% CI | P value | |
| Average propofol dose (mg/kg/hour) | 9 | 5.3 (1) | 9 | 5.1 (0.9) | 0.2 | ‐0.68, 1.08 | 0.670 |
| Time to eye opening (minutes) | 9 | 5.7 (5.7) | 9 | 4.1 (2.8) | 1.6 | ‐2.55, 5.75 | 0.771 |
| Time to consciousness (minutes) | 9 | 7.6 (5.3) | 9 | 7.6 (3.6) | 0 | ‐4.19, 4.19 | NA |
| Number of flow rate changes | 9 | 4.4 (2.5) | 9 | 3.6 (1.7) | 0.8 | ‐1.18, 2.78 | 0.779 |
| Number of boluses | 9 | 1.4 (2.3) | 9 | 0.89 (1.4) | 0.51 | ‐1.25,2.27 | 0.719 |
| Mean morphine total daily dosage (mg) | 25 | 22.6* | 25 | 26.6* | 0.67 | ||
| Mean midazolam total daily dosage (mg) | 25 | 18.4* | 25 | 14.6* | 0.85 | ||
| Mean midazolam dose (mg/kg/hour) | 42 | 0.10 (0.02) | 63 | 0.09 (0.02) | 0.01 | 0.00, 0.02 | 0.993 |
| Mean propofol dose (mg/kg/hour) | 42 | 0.95 (0.23) | 63 | 0.86 (0.20) | 0.09 | 0.00, 0.18 | 0.979 |
| Mean time to wake up (minutes) | 42 | 0* | 63 | 15* | <0.05 | ||
| * Standard deviation not reported | |||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Duration of mechanical ventilation Show forest plot | 2 | 155 | Mean Difference (IV, Random, 95% CI) | ‐0.02 [‐0.13, 0.09] |
