Secondary Logo

Journal Logo

Patient-Controlled Versus Clinician-Controlled Sedation With Propofol: Systematic Review and Meta-analysis With Trial Sequential Analyses

Kreienbühl, Lukas MD*; Elia, Nadia MD, MSc*,†‡; Pfeil-Beun, Elvire MD*; Walder, Bernhard MD*,†; Tramèr, Martin R. MD, DPhil*,†

doi: 10.1213/ANE.0000000000003361
Anesthetic Clinical Pharmacology

BACKGROUND: Sedation with propofol is frequently used to facilitate diagnostic and therapeutic procedures. Propofol can be administrated by the patient (patient-controlled sedation [PCS]) or by a clinician (clinician-controlled sedation [CCS]). We aimed to compare these 2 techniques.

METHODS: PubMed, Embase, CENTRAL, and trial registries were searched up to October 2017 for randomized controlled trials comparing PCS with CCS with propofol. The primary end points were the risks of presenting at least 1 episode of oxygen desaturation, arterial hypotension, and bradycardia, and the risk of requiring a rescue intervention (pharmacologic therapies or physical maneuvers) for sedation-related adverse events. Secondary end points were the dose of propofol administrated, operator and patient satisfaction, and the risk of oversedation. A random-effects model and an α level of .02 to adjust for multiple analyses were used throughout. Trial sequential analyses were performed for primary outcomes. Quality of evidence was assessed according to the Grades of Recommendation, Assessment, Development, and Evaluation system.

RESULTS: Thirteen trials (1103 patients; median age, 47 years; American Society of Anesthesiologists physical status I–III) describing various diagnostic and therapeutic procedures with propofol sedation were included. PCS had no impact on the risk of oxygen desaturation (11 trials, 31/448 patients [6.9%] with PCS versus 46/481 [9.6%] with CCS; risk ratio, 0.74 [98% confidence interval, 0.35–1.56]) but decreased the risk of requiring a rescue intervention for adverse events (11 trials, 29/449 patients [6.5%] with PCS versus 74/482 [15.4%] with CCS; risk ratio, 0.45 [98% confidence interval, 0.25–0.81]). For both outcomes, Trial sequential analyses suggested that further trials were unlikely to change the results, although the quality of evidence was graded very low for all primary outcomes. For the risk of arterial hypotension and bradycardia, the required sample size for a definitive conclusion had not been reached. Analysis of secondary outcomes suggested that PCS decreased the risk of oversedation and had no impact on propofol dose administrated, or on operator or patient satisfaction.

CONCLUSIONS: PCS with propofol, compared with CCS with propofol, had no impact on the risk of oxygen desaturation, but significantly decreased the risk of rescue interventions for sedation-related adverse events. Further high-quality trials are required to assess the risks and benefits of PCS.

From the *Division of Anesthesiology, Geneva University Hospitals, Geneva, Switzerland

Faculty of Medicine, University of Geneva, Geneva, Switzerland; and Geneva Cancer Registry, Institute of Global Health, Faculty of Medicine, University of Geneva, Geneva, Switzerland.

Published ahead of print May 9, 2018.

Accepted for publication February 16, 2018.

Funding: Departmental.

The authors declare no conflicts of interest.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website.

Reprints will not be available from the authors.

Address correspondence to Lukas Kreienbühl, MD, Department of Anesthesiology, Intensive Care and Pain Medicine, Helios Klinikum Bad Saarow, Pieskower Straße 33, 15526 Bad Saarow, Germany. Address e-mail to

Sedation for diagnostic and therapeutic procedures improves patient comfort and compliance. Propofol is widely used in this context1,2 because of its rapid onset and offset. It is usually administered by anesthesiologists or other clinicians who are trained in the administration of this drug. Patient-controlled sedation (PCS) is an alternative technique to clinician-controlled sedation (CCS). In analogy to patient-controlled analgesia (PCA) with opioids, which has become a widely used and safe method of postoperative analgesia in the surgical setting,3,4 PCS allows patients to self-administer a sedative drug via a hand-held device connected to an infusion pump, using a preprogrammed bolus dose, and an optional lock-out time and background infusion.

This systematic review was designed to compare PCS with propofol (experimental intervention) with CCS with propofol (control intervention), in patients undergoing a diagnostic or therapeutic procedure requiring sedation. We hypothesized that PCS decreases the risks of oxygen desaturation, arterial hypotension, bradycardia, and rescue interventions due to adverse events.

Back to Top | Article Outline


For the reporting of this systematic review, we followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses recommendations.5

Back to Top | Article Outline

Protocol and Registration

The protocol of this systematic review has not been registered.

Back to Top | Article Outline

Eligibility Criteria

We included full reports of randomized controlled trials (RCTs) with a parallel group design, reporting on the risk of sedation-related adverse events with PCS with propofol compared with CCS with propofol, in adults undergoing a diagnostic or therapeutic procedure. CCS was defined as sedation provided by a medical doctor, dentist, or nurse. Propofol had to be used as the main sedative drug in both groups, and any additional medication given before or during the procedure (eg, premedication, opioids, local or spinal anesthetics) had to be administered equally in both groups. Patient-controlled target-controlled infusion (TCI), which is also known as patient-maintained sedation, was considered as PCS. Trials were not excluded when clinicians administered loading doses of propofol in the PCS group before passing the push button to the patient. A minimal group size of 10 patients was set.

Back to Top | Article Outline

Information Sources and Search

We searched Medline (through PubMed), Embase, and the Cochrane Library (CENTRAL) combining the following search terms with Boolean operators: (“patient-controlled” or “patient-maintained”) and “sedation” and “propofol” (Supplemental Digital Content, Table A, No filters were used. Electronic searches were performed up to October 2017. There was no restriction to language or year of publication. In addition, 5 trial registries were searched for interrupted (after the start of enrollment) or completed unpublished trials, fulfilling the inclusion criteria of this systematic review.

Back to Top | Article Outline

Study Selection

Study selection was performed by 1 author (L.K.) and, in case of queries, discussed with another author (M.R.T.). Titles of retrieved reports were screened. Abstracts of potentially relevant articles were assessed for eligibility.

Back to Top | Article Outline

Data Collection Process

Two authors (L.K., E.P.-B.) independently extracted data from original reports, and a third author (M.R.T.) resolved discrepancies. Extracted data were transferred into an Excel spreadsheet that was designed for the purpose of this analysis. Authors of original reports were contacted when additional unpublished data were required for analysis.

Back to Top | Article Outline

Data Items and Definitions

We collected information on year of publication, first author’s name, study sample size, patient characteristics (American Society of Anesthesiologists [ASA] physical status, mean age, body mass index and weight, and sex ratio), inclusion and exclusion criteria, type of procedure, drug regimen, targeted depth of sedation in the CCS group, health profession or medical specialty responsible for CCS, type of PCS pump device used, proportion of and reasons for PCS failures, proportion of patients who declined trial participation, and sponsorship.

The primary end points were the risks of oxygen desaturation, arterial hypotension, and bradycardia, and the risk of requiring rescue interventions for any adverse events related to sedation. Rescue interventions considered were pharmacological therapies (eg, atropine) or physical maneuvers (eg, chin lift) to counteract an adverse event. Prodding was not considered a rescue intervention, nor was supplementary oxygen administration. Reported airway obstructions were considered as adverse events requiring a chin lift.

Secondary end points were total propofol dose administrated, risk of oversedation, and operator and patient satisfaction. Oversedation was defined as an absent response to loud and repeated spoken commands. Data on patient and operator satisfaction were reported on a 0–10 cm visual analog scale (VAS; 0 = not satisfied at all, 10 = maximally satisfied) and were converted to this scale, if reported on a 0–100 mm VAS or 0–10 points numerical scale.

Back to Top | Article Outline

Risk of Bias and Quality of Evidence Assessment

Limitations related to trial design and implementation were assessed with the Cochrane risk of bias assessment tool. Quality of evidence was assessed according to the principles of the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) Working Group.6 The GRADE approach categorizes evidence into high, moderate, low or very low quality, taking into account the limitations of trial design and implementation, indirectness of evidence, heterogeneity of results, imprecision of results, and risk of publication bias. Risk of publication bias was assessed using Funnel plots, if at least 5 trials with at least 1 event each could be included.

Two authors (L.K., N.E.) independently performed risk of bias and quality of evidence assessments. A third author (M.R.T.) resolved discrepancies.

To test for the robustness and validity of the results and to take into account the trial heterogeneity, we performed subgroup analyses for primary end points of trials that used local or locoregional anesthesia, and trials that used a propofol bolus regimen in both groups. These analyses were only performed if the subgroup size reached at least 100 patients or if data from at least 3 trials were available.

Back to Top | Article Outline

Statistical Analyses

For primary end points, risk ratios (RRs) with confidence interval (CI) were computed comparing experimental with control groups at the study level, based on group sizes and the number of events reported in each group. The estimates of effect of individual studies were combined into a pooled weighted estimate using the random-effects model by DerSimonian and Laird.7 In case of zero events, a constant continuity correction was performed by adding 0.5 to each cell. To decrease the risk of type I error, the threshold for statistical significance was adapted to the 4 primary end points according to the recommendation of Jakobsen et al.8 This resulted in an α-level of .02 (0.05/[{4 + 1}/2]), instead of the conventional .05, and a 98% CI, instead of the conventional 95% CI around the point estimates.

For primary end points, trial sequential analyses (TSAs) were performed to identify the information sizes required to verify the study hypotheses, that is, the number of patients needed to reach a definitive conclusion of either harm or futility. If the information size was not reached, O’Brien–Fleming α-spending boundaries were computed (80% power, α level of 2% and 2-sided test, including trials with zero events). We considered that a definitive conclusion of effect or futility was reached when the cumulative Z-curve crossed the O’Brien–Fleming α-spending boundaries. There was a pre hoc consensus to set the threshold of a minimal clinically relevant effect at 50%.

Outcomes in subgroup analyses and secondary outcomes were computed as RRs or weighted mean differences (WMDs). Ninety-eight percentage CIs were used to take into account the 4 secondary outcomes. For continuous secondary outcomes, we computed mean differences in effects between PCS and CCS at study level, before combining them into a pooled WMD, using an inverse variance random-effects model.

Study heterogeneity and inconsistency were evaluated using the Q statistics and I2, respectively. Significant heterogeneity was defined as P < .1 and inconsistency as I2 > 75%.

Statistical analyses were performed using RevMan version 5.2.7 (The Cochrane Collaboration, Copenhagen, Denmark), STATA 15 (StataCorp, College Station, TX), Microsoft Excel version 12.0, and TSA viewer, version beta (Copenhagen Trial Unit, Copenhagen, Denmark).9

Back to Top | Article Outline


Study Selection

Figure 1.

Figure 1.

Of 967 retrieved references, 130 were assessed for eligibility (Figure 1). Of those, 117 were subsequently excluded: 54 were not RCTs, 32 did not have CCS as a control group, 18 used different drug combinations in PCS and CCS groups, 7 were published as abstracts only, 1 was not performed during a procedure, 4 had a crossover design, and 1 had <10 patients per group. We finally included 13 RCTs, with data from 1103 patients (CCS 567 patients; PCS 536 patients).10–21

Back to Top | Article Outline

Trial Characteristics

Trials were published between 1993 and 2015 (Table; Supplemental Digital Content, Table B, The median trial size was 74 patients (range, 3216 to 18121). Procedures were colonoscopies in 4 trials,13–15,22 extracorporeal shock wave lithotripsies10,16 and endoscopic retrograde cholangiopancreatographies17,21 in 2 each, and cataract surgery in 1.20 In 4 trials, a variety of procedures were performed.11,12,18,19 In the CCS groups, propofol was administered by anesthesiologists in 9 trials,10,12,13,15–19,22 anesthesiology nurses in 2,20,21 and gastroenterologists14 and emergency physicians11 in 1 trial each.



In 4 trials, patients and clinicians used the same propofol administration device and administered the same bolus dose of propofol. In 3 of these 4 trials, lock-out times were identical in both groups,12,15,20 and differed in 1 trial.14 In 5 trials, propofol boluses in the PCS group were compared with continuous propofol infusions in the CCS group.10,18,19,21,23 In 2 trials, propofol boluses were used in the PCS group and TCI in the CCS group.13,17 In 1 trial, patient-controlled propofol boluses were compared with propofol boluses as required.11 In 1 trial, patient-maintained sedation (ie, patient-controlled TCI) was compared with propofol boluses as needed.22

In 8 trials, clinicians were encouraged to titrate propofol according to a defined sedation depth (eg, mild slowing of speech).12,14,15,17–19,21,22 In these trials, the suggested sedation depth corresponded to a level of moderate sedation, that is, a relaxed patient, responding to loud spoken or mild tactile stimulation. In the remaining 5 trials, the targeted sedation depth was less precisely defined (eg, patient comfort) or was left at the discretion of the clinicians.10,11,13,16,20 In 9 trials, the clinician administered or encouraged patients to administer a propofol loading dose in the PCS group.11,12,14–16,18–21 In 4 trials, patients of both groups had a concomitant local or regional anesthesia.12,18–20 In 8 trials, patients of both groups received concomitant opiates intravenously; in 3 trials as a preoperative bolus,12,14,22 in 2 as a rescue drug during the procedure,11,17 in 2 as an adjuvant to the propofol solution,15,16 and in 1 trial as a baseline infusion.10

The number of PCS failures was reported in 13 trials10–22 and occurred in 26 of 539 (4.6%) patients. Twenty-two patients required additional propofol,11,13,17,21 2 patients received midazolam,11 and 2 patients had a failing PCS device as a result of an interrupted power supply.22 For all events except in 1 trial (3 events),17 an intention-to-treat analysis was performed.

Back to Top | Article Outline

Patient Characteristics

In 4 trials, patients were classified ASA physical status I or II,10,12,16,18 in 7 trials I–III,11,13,14,17,19–21 and in 1 trial II or III (Supplemental Digital Content, Table B, One trial did not provide information on the ASA physical status.15 The median of mean ages was 47 years (range, 3618 to 6820,21). None of the trials included children. In 4 trials, the median of the mean body mass index was 25.5 kg·m−2 (range, 23.417 to 29.619).13,17,19,21 In 10 trials, median of mean weights was 75 kg (range, 5118 to 8115,22).10–12,14–16,18,20–22 The median percentage of males was 59% (range, 37%12 to 90%16). In 7 trials,10,11,13,14,17,21,22 median rate of refusal of participation in the trial was 25.1% (range, 3.3%21 to 63%22). One trial reported the reasons for refusal, which were fear to take responsibility for sedation (73%) and preference to be asleep during the whole procedure (20%).14

We contacted the corresponding authors of 8 trials for additional patient data; 6 replied (75%),10,11,15,21,22,24 and unpublished data of 4 trials could eventually be included into our analyses (Supplemental Digital Content, Table C,,15,21,22

Back to Top | Article Outline

Risk of Bias and Quality of Evidence Assessment

In 2 trials, the method of randomization was judged unclear because of insufficient description or unclear validity of the method used (Supplemental Digital Content, Table D,,17 The risk of bias regarding concealment of treatment allocation was judged high for 1,15 low for 4,11,13,15,22 and unclear for all other trials. Blinding of participants and personnel (performance bias) was insufficient for all trials except 2.10,22 Blinding of outcome assessors (detection bias) was insufficient for all trials except 3.10,19,22 Incomplete data reporting (attrition bias) was judged as likely in 1 trial17 and possible in 4 trials.11,12,14,16 The risk of selective data reporting (reporting bias) could not be determined and was considered unclear for all trials except 1, where the results of randomized and nonrandomized patients were merged in the PCS group.21 No other biases were detected.

Funnel plots for the risk of oxygen desaturation, arterial hypotension, and rescue intervention for adverse events did not suggest publication bias (Supplemental Digital Content, Figures A–C, The literature search in trial registries did not reveal interrupted or terminated, unpublished trials (Supplemental Digital Content, Table E,

Quality of evidence according to the GRADE system was judged very low for all primary outcomes and for total propofol dose. It was judged low for the risk of oversedation and operator and patient satisfaction (Supplemental Digital Content, Table F,

Back to Top | Article Outline

Primary End Points

Oxygen Desaturation.

The number of patients experiencing peripheral oxygen desaturation was reported in 11 trials (929 patients).10–12,14–19,21,22 Desaturation was defined as peripheral oxygen saturation (Spo2) <90%,11,12,18 Spo2 <90% with O2 2,14 3,19,21 or 4 L·minute-1,17 Spo2 <95%16 and <94%22 with O2 4 L·minute-1, Spo2 <92% with O2 2 L·minute-1,10 or as Spo2 <90% for >30 seconds with an inspired oxygen fraction of 100%.15 Oxygen desaturation was found in 31 of 448 (6.9%) patients with PCS compared with 46 of 481 (9.6%) patients with CCS; RR, 0.74 (98% CI, 0.35–1.56) (Figure 2).

Figure 2.

Figure 2.

The required sample size needed to reach a definitive conclusion regarding the capacity of PCS to reduce oxygen desaturation by 50% compared with CCS was 1152 patients and was not reached (Supplemental Digital Content, Figure D, However, the cumulative Z-curve had crossed the line for futility since the trial by Mandel et al15 published in 2010, suggesting that future trials are unlikely to change our conclusion.

Back to Top | Article Outline

Arterial Hypotension.

The number of patients with at least 1 episode of arterial hypotension was reported in 6 trials (668 patients).11,17,18,20–22 Arterial hypotension was defined as a systolic blood pressure of <90 mm Hg,17,21,22 or <80 mm Hg,11,18 or a drop of >30% from baseline.20 Hypotension occurred in 21 of 320 (6.6%) patients with PCS compared with 40 of 348 (11.5%) patients with CCS; RR, 0.56 (98% CI, 0.34–0.93) (Figure 3).

Figure 3.

Figure 3.

The sample size required to reach a definitive conclusion on a decrease of 50% in the risk of arterial hypotension using PCS was 957 and was not reached (Supplemental Digital Content, Figure E, Although the pooled effect from the random-effects meta-analysis was statistically significant, the Z-curve did not cross the α-spending boundaries, suggesting that these results may be due to random chance, and that no definitive conclusion can yet be reached.

Back to Top | Article Outline


The number of patients with at least 1 episode of bradycardia was reported in 4 trials (451 patients).11,13,16,21 Bradycardia was defined as a heart rate <60 beats per minute,11 <40 beats per minute,21 or a decrease of >10% from baseline.13 In 1 trial, no numerical definition was provided.16 Bradycardia occurred in 11 of 214 (5.1%) patients with PCS compared with 12 of 237 (5.0%) with CCS; RR, 0.86 (98% CI, 0.35–2.09) (Figure 4).

Figure 4.

Figure 4.

To reach a definitive conclusion on a decrease of 50% in the risk of bradycardia using PCS, the required sample size was 2319 (Supplemental Digital Content, Figure F, The Z-curve did not cross the α-spending boundaries, and therefore, no definitive conclusion could be reached on this end point.

Back to Top | Article Outline

Intervention for Adverse Events.

The number of rescue interventions for adverse events was extracted from 11 trials (931 patients).10–12,14–18,20–22 Rescue interventions consisted of chin lift maneuvers (PCS 12 patients; CCS 45 patients),11,17,21,22 bag-mask ventilation (0 vs 6 patients),11,15 sympathomimetic drugs for hypotension (15 vs 23 patients),17,18,21 and atropine for bradycardia (2 vs 0 patients).16,21 At least 1 rescue intervention was needed in 29 of 449 (6.5%) patients with PCS compared with 74 of 482 (15.4%) patients with CCS; RR, 0.45 (98% CI, 0.25–0.81) (Figure 5).

Figure 5.

Figure 5.

A sample size of 713 was needed to reach a definitive conclusion on a 50% decrease in the risk of requiring a rescue intervention for an adverse event when using PCS compared with CCS (Supplemental Digital Content, Figure G, The present analysis has reached the required sample size suggesting that further trials are unlikely to change our conclusion.

Back to Top | Article Outline

Subgroup Analyses for Primary End Points

Subgroup analysis of trials, which used a propofol bolus regimen in both groups, confirmed the main results (ie, no difference between PCS and CCS) for the risk of oxygen desaturation, but failed to confirm a difference in the risk of arterial hypotension and interventions for adverse events (Supplemental Digital Content, Figures H–J, Analysis of trials with concomitant use of local or locoregional anesthesia confirmed the main results (Supplemental Digital Content, Figures K and L,

Back to Top | Article Outline

Secondary End Points

Total Propofol Dose Administrated.

Total propofol dose was reported in 8 trials (724 patients).10,13–16,19–21 The median of mean propofol doses was 122 mg with PCS (range, 3020 to 23221) and 126 mg with CCS (range, 3520 to 33721). The amount of propofol administrated when using PCS was not significantly decreased; WMD −21.8 mg (98% CI, −44.3 to 0.73) (Supplemental Digital Content, Figure M,

Back to Top | Article Outline

Risk of Oversedation.

Data on oversedation were extracted from 8 trials (840 patients). Oversedation occurred in 68 of 407 (16.7%) patients with PCS compared with 193 of 433 (44.6%) patients with CCS; RR, 0.37 (98% CI, 0.21–0.63) (Supplemental Digital Content, Figure N,

Back to Top | Article Outline

Operator Satisfaction.

Operator satisfaction was reported in 5 trials (410 patients).11,14,16,20,22 Median of mean ratings on the 10 cm VAS was 7.8 cm for both PCS (range, 6.722 to 9.120) and CCS (range, 7.722 to 9.220); WMD −0.18 cm (98% CI, −0.46 to 0.11) (Supplemental Digital Content, Figure O,

Back to Top | Article Outline

Patient Satisfaction.

Patient satisfaction was reported in 7 trials (687 patients).11,14,16,19–22 Median of mean ratings on the 10 cm VAS was 9.0 cm for both PCS (range, 7.916 to 9.619) and CCS (range, 7.216 to 9.619); WMD −0.05 cm (98% CI, −0.49 to 0.39) (Supplemental Digital Content, Figure P,

Back to Top | Article Outline


This systematic review and meta-analysis of RCTs showed that in patients undergoing a diagnostic or therapeutic procedure under sedation, PCS with propofol decreased the risk for rescue interventions to treat sedation-related adverse events, but had no impact on the risk of oxygen desaturation, compared with CCS. A definitive conclusion (risk reduction ≥50% as by our definition) could not be ascertained for arterial hypotension or bradycardia due to a lack of sufficient data. Analysis of secondary end points suggested a lower risk of oversedation with PCS, but no impact on the total dose of propofol administrated, or on operator or patient satisfaction. These conclusions are based on low (oversedation, operator and patient satisfaction) to very low (primary outcomes, total propofol dose) quality of evidence according to the GRADE system.

The first clinical study on PCS with propofol was performed in patients undergoing third molar extraction and was published in 1991.25 Subsequently, PCS with propofol has been studied in a variety of procedures. A recent meta-analysis comparing PCS with CCS during colonoscopy showed a decreased risk of oxygen desaturation and arterial hypotension with PCS.26 An observational single-center study including 1196 patients undergoing endoscopic retrograde cholangiopancreatography showed that PCS with propofol was less often associated with respiratory and cardiovascular depression than CCS.27

The present analysis adds further knowledge as it included exclusively RCTs comparing PCS with CCS, with propofol as the main sedative drug in both treatment groups, and it was not restricted to 1 type of procedure. Our analyses were conservative; we were using a random-effects model throughout to allow for treatment effects to differ across settings, a 98% CI to take into account multiple primary outcomes testing, and a risk reduction of ≥50% as a cutoff for TSA.

Several limitations of this systematic review need to be discussed. First, the majority of included trials had a high risk of bias, reported on different primary end points than those used for this meta-analysis and were small. Inclusion of small trials into meta-analyses may overestimate the beneficial effect of an experimental intervention,28 but neither our search for unpublished trials nor the funnel plots suggested publication bias. However, the value of funnel plots to determine publication bias is questioned, especially when <10 trials are included.29 Second, several trials compared a propofol bolus regimen in the PCS group with a continuous propofol infusion in the CCS group. Continuous propofol infusions may result in higher cumulative propofol doses compared with intermittent propofol boluses.30–33 Subgroup analyses of trials that were using bolus regimens in both experimental and control groups did not confirm the overall results for the risk of arterial hypotension and interventions for adverse events, although the former subgroup consisted of only 3 trials, and the 98% CI for the latter crossed 1 by a close margin (RR, 0.37 [98% CI, 0.14–1.01]). Third, outcome results may have varied according to the type of procedure. Unfortunately, the number of trials with identical procedures was too small for valid subgroup analyses. Finally, most trials included predominantly low to medium risk, young to middle-aged, nonobese adults. Although none of the trials explicitly excluded obese or elderly patients, these patient groups were not adequately represented in this systematic review and the results may not apply to them.

We retrieved 13 valid and relevant RCTs with data of 1103 patients, published over the last 23 years. This relatively low number of trials suggests that, contrary to PCA with intravenous opioids, the dissemination of, and interest in, PCS has remained limited. Several reasons may explain why PCS is not used more widely. First, the propofol titration window for a moderate level of sedation is narrow, and patient response to propofol is variable and cannot solely be predicted by sex, age, and weight. Until today, PCS is usually based on a one-size-fits-all dosing scheme (bolus and lock-out time), which may not be suitable for every patient. Second, only patients who are able and willing to use a pump device and to take responsibility for sedation are eligible for PCS. This is likely to exclude many patients. Third, it remains unclear whether PCS is cost-effective. One trial included a cost-effectiveness analysis, considering costs of nursing time and medication, and reported higher costs with PCS compared with CCS.14 Additional costs related to the time needed for the preparation of the PCS device and for patient instructions should be included in future analyses. Finally, PCS may not be considered safe and effective enough to be performed in the absence of a clinician dedicated to monitoring the patient and intervening when necessary. It is likely that PCS will only become cost-effective without the uninterrupted presence of a clinician or nurse in addition to the operator.

Despite these limitations, PCS with propofol may have the potential to become as popular for daily clinical practice as PCA with opioids. Several developments of the PCS technique deserve further investigation. TCI systems attenuate fluctuations of plasma concentrations compared with an on-demand bolus regimen,34 and may therefore be better adapted to an individual’s sedative drug requirements. Anesthesia responsiveness monitoring is a safety feature, which feeds a patient’s reaction time to a verbal command back into the PCS device and adapts drug delivery accordingly.35,36 Fuzzy logic PCS adjusts the size of boluses and basal infusion of the drug based on previous button press activity.37 These and other technological developments may further improve safety and acceptability of PCS.

In conclusion, in low- to medium-risk middle-aged nonobese patients undergoing a diagnostic or therapeutic procedure under sedation, PCS with propofol had no impact on the risk of oxygen desaturation, but reduced the risk for rescue interventions for adverse events, compared with CCS with propofol. The impact of PCS on the risk of arterial hypotension or bradycardia remained uncertain. Low-grade quality of available data demands further high-quality trials to validate these findings.

Back to Top | Article Outline


The authors thank Prof. A. Alhashemi, Prof. J. Greenslade, Prof. K. Leslie, Prof. J. E. Mandel, Dr A. Nilsson, and Prof. M. Zacharias for providing unpublished data.

Back to Top | Article Outline


Name: Lukas Kreienbühl, MD.

Contribution: This author helped conceive the project, design the study, collect and analyze the data, and write the manuscript.

Name: Nadia Elia, MD, MSc

Contribution: This author helped analyze the data and write the manuscript.

Name: Elvire Pfeil-Beun, MD.

Contribution: This author helped collect and analyze the data, and attests to the integrity of the original data and the analysis reported in this manuscript.

Name: Bernhard Walder, MD.

Contribution: This author helped write the manuscript.

Name: Martin R. Tramèr, MD, DPhil.

Contribution: This author helped design the study, analyze the data, and write the manuscript.

This manuscript was handled by: Ken B. Johnson, MD.

Back to Top | Article Outline


1. McQuaid KR, Laine L. A systematic review and meta-analysis of randomized, controlled trials of moderate sedation for routine endoscopic procedures. Gastrointest Endosc. 2008;67:910–923.
2. Newstead B, Bradburn S, Appelboam A, et al. Propofol for adult procedural sedation in a UK emergency department: safety profile in 1008 cases. Br J Anaesth. 2013;111:651–655.
3. Walder B, Schafer M, Henzi I, Tramèr MR. Efficacy and safety of patient-controlled opioid analgesia for acute postoperative pain. A quantitative systematic review. Acta Anaesthesiol Scand. 2001;45:795–804.
4. Ballantyne JC, Carr DB, Chalmers TC, Dear KB, Angelillo IF, Mosteller F. Postoperative patient-controlled analgesia: meta-analyses of initial randomized control trials. J Clin Anesth. 1993;5:182–193.
5. Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA Group. Preferred Reporting Items for Systematic Reviews and Meta-Analyses: the PRISMA statement. Ann Intern Med. 2009;151:264–269, W64.
6. Atkins D, Best D, Briss PA, et al.; GRADE Working Group. Grading quality of evidence and strength of recommendations. BMJ. 2004;328:1490.
7. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7:177–188.
8. Jakobsen JC, Wetterslev J, Lange T, Gluud C. Viewpoint: taking into account risks of random errors when analysing multiple outcomes in systematic reviews. Cochrane Database Syst Rev. 2016;3:ED000111.
9. Thorlund K, Jakobsen J, Wetterslev J, Jesper B, Imberger G, Gluud C. User Manual for Trial Sequential Analysis (TSA). 2017. Available at: Accessed January 15, 2018.
10. Alhashemi JA, Kaki AM. Anesthesiologist-controlled versus patient-controlled propofol sedation for shockwave lithotripsy. Can J Anaesth. 2006;53:449–455.
11. Bell A, Lipp T, Greenslade J, Chu K, Rothwell S, Duncan A. A randomized controlled trial comparing patient-controlled and physician-controlled sedation in the emergency department. Ann Emerg Med. 2010;56:502–508.
12. Cork RC, Guillory EA, Viswanathan S, Eyrich JE, Heaton JF. Effect of patient-controlled sedation on recovery from ambulatory monitored anesthesia care. Am J Anesthesiol. 1995;22:94–100.
13. Crepeau T, Poincloux L, Bonny C, et al. Significance of patient-controlled sedation during colonoscopy. Results from a prospective randomized controlled study. Gastroenterol Clin Biol. 2005;29:1090–1096.
14. Heuss LT, Drewe J, Schnieper P, Tapparelli CB, Pflimlin E, Beglinger C. Patient-controlled versus nurse-administered sedation with propofol during colonoscopy. A prospective randomized trial. Am J Gastroenterol. 2004;99:511–518.
15. Mandel JE, Lichtenstein GR, Metz DC, Ginsberg GG, Kochman ML. A prospective, randomized, comparative trial evaluating respiratory depression during patient-controlled versus anesthesiologist-administered propofol-remifentanil sedation for elective colonoscopy. Gastrointest Endosc. 2010;72:112–117.
16. Maroof M, Khan RM, Bhatti TH, Hamalawy H, Siddique MK. Evaluation of patient controlled sedation and analgesia for ESWL. J Stone Dis. 1993;5:240–243.
17. Mazanikov M, Udd M, Kylänpää L, et al. A randomized comparison of target-controlled propofol infusion and patient-controlled sedation during ERCP. Endoscopy. 2013;45:915–919.
18. Singh T, Ravishankar M. Conscious sedation with propofol for surgeries under spinal anaesthesia: anaesthesiologist versus patient controlled. J Anaesthesiol Clin Pharmacol. 2005;21:169–173.
19. Wahlen BM, Kilian M, Schuster F, Muellenbach R, Roewer N, Kranke P. Patient-controlled versus continuous anesthesiologist-controlled sedation using propofol during regional anesthesia in orthopedic procedures–a pilot study. Expert Opin Pharmacother. 2008;9:2733–2739.
20. Yun MJ, Oh AY, Kim KO, Kim YH. Patient-controlled sedation vs anaesthetic nurse-controlled sedation for cataract surgery in elderly patients. Int J Clin Pract. 2008;62:776–780.
21. Nilsson A, Grossmann B, Kullman E, Uustal E, Sjöberg F, Nilsson L. Sedation during endoscopic retrograde cholangiopancreatography: a randomized controlled study of patient-controlled propofol sedation and that given by a nurse anesthetist. Scand J Gastroenterol. 2015;50:1285–1292.
22. Stonell CA, Leslie K, Absalom AR. Effect-site targeted patient-controlled sedation with propofol: comparison with anaesthetist administration for colonoscopy. Anaesthesia. 2006;61:240–247.
23. Girdler NM, Rynn D, Lyne JP, Wilson KE. A prospective randomised controlled study of patient-controlled propofol sedation in phobic dental patients. Anaesthesia. 2000;55:327–333.
24. Zacharias M, Bridgman J, Parkinson R. Two methods of administration of propofol for dental sedation. Br J Oral Maxillofac Surg. 1998;36:19–23.
25. Rudkin GE, Osborne GA, Curtis NJ. Intra-operative patient-controlled sedation. Anaesthesia. 1991;46:90–92.
26. Lu Y, Hao LX, Chen L, Jin Z, Gong B. Systematic review and meta-analysis of patient-controlled sedation versus intravenous sedation for colonoscopy. Int J Clin Exp Med. 2015;8:19793–19803.
27. Jokelainen J, Udd M, Kylänpää L, et al. How patient-controlled sedation is adopted in clinical practice of sedation for endoscopic retrograde cholangiopancreatography? A prospective study of 1196 cases. Scand J Gastroenterol. 2017;52:166–172.
28. Finckh A, Tramèr MR. Osteoarthritis: small studies overestimate the benefit of therapies for OA. Nat Rev Rheumatol. 2010;6:617–618.
29. Lau J, Ioannidis JP, Terrin N, Schmid CH, Olkin I. The case of the misleading funnel plot. BMJ. 2006;333:597–600.
30. Brownlie GS, Baker JA, Ogg TW. Propofol: bolus or continuous infusion. A day case technique for the vaginal termination of pregnancy. Anaesthesia. 1991;46:775–777.
31. González-Santiago JM, Martín-Noguerol E, Vinagre-Rodríguez G, et al. Intermittent boluses versus pump continuous infusion for endoscopist-directed propofol administration in colonoscopy. Rev Esp Enferm Dig. 2013;105:378–384.
32. Grendelmeier P, Tamm M, Pflimlin E, Stolz D. Propofol sedation for flexible bronchoscopy: a randomised, noninferiority trial. Eur Respir J. 2014;43:591–601.
33. Bennett J, Shafer DM, Efaw D, Goupil M. Incremental bolus versus a continuous infusion of propofol for deep sedation/general anesthesia during dentoalveolar surgery. J Oral Maxillofac Surg. 1998;56:1049–1053.
34. Atkins JH, Mandel JE. Recent advances in patient-controlled sedation. Curr Opin Anaesthesiol. 2008;21:759–765.
35. O’Brien C, Urquhart CS, Allam S, et al. Reaction time-monitored patient-maintained propofol sedation: a pilot study in oral surgery patients. Anaesthesia. 2013;68:760–764.
36. Allam S, Anderson KJ, O’Brien C, et al. Patient-maintained propofol sedation using reaction time monitoring: a volunteer safety study. Anaesthesia. 2013;68:154–158.
37. Shieh JS, Chang LW, Wang MS, Sun WZ, Wang YP, Yang YP. Pain model and fuzzy logic patient-controlled analgesia in shock-wave lithotripsy. Med Biol Eng Comput. 2002;40:128–136.

Supplemental Digital Content

Back to Top | Article Outline
Copyright © 2018 International Anesthesia Research Society