Propofol vs. inhalational agents to maintain general anaesthesia in ambulatory and in-patient surgery: a systematic review and meta-analysis

It is estimated that of the more than 300 million surgical procedures undertaken worldwide annually, the majority requires anaesthesia services [1]. Propofol and volatile anaesthetic agents are the key components of modern general anaesthesia and account for many beneficial outcomes in clinical anaesthetic practice and in perioperative medicine [2]. Previous attempts have been made by systematic reviews and meta-analyses to establish particular benefits for one technique over the other with mixed or inconclusive results, but concerns were also raised which methodology is able to best capture these [3]. It is still not known in many domains whether or not propofol-based total intravenous anaesthesia (TIVA) techniques are more advantageous to patients than volatile-based anaesthesia. Often, surrogate and composite endpoints have precluded clinical application of results [4]. In addition, patient-reported outcome measures (PROM) are increasingly relevant to health care providers and will influence clinical choices [5]. We have therefore undertaken a new exhaustive and comprehensive literature review representing a wide spectrum of clinical applications on key parameters relevant to patient outcome, in particular the influence of choice of anaesthetic technique on post-operative nausea and vomiting (PONV) [6]. Potential confounders and bias were minimised by strict adherence to Cochrane guidelines [7].

We conducted a systematic review and meta-analyses of randomized controlled trials (RCTs) on the efficacy and safety of propofol (TIVA) vs. inhalational agents in maintaining general anaesthesia in ambulatory and in-patient surgical procedures. We also studied how those results vary by type of comparator, patient characteristics, type and setting of surgery. A systematic literature review of RCTs published up to date of the search was performed and predefined outcomes from selected studies were extracted, pooled and meta-analysed. A secondary meta-analysis also included observational studies. Reporting was done according to the PRISMA reporting guideline [8].

This systematic review was registered at PROSPERO, 22 July 2016, registration number CRD42016039539.

The systematic literature review and meta-analyses were performed and reported according to the Cochrane Handbook [7] and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA-P, PRISMA) guidelines [8, 9]. The quality of evidence for all outcomes was judged using the Grading of Recommendations Assessment, Development and Evaluation working group methodology (GRADE) [10]. A full and detailed description of the methodology and the search strings can be found in an additional file (Additional file 1).

The most important finding of this meta-analysis is a consistent and clinically relevant reduction on PONV when using TIVA over volatile maintenance of general anaesthesia. The relative risk reduction of 39% in our study is more than twice as high as the 18.9% reported by Apfel and colleagues in 2004 [13]. Although the anti-emetic effect of propofol is widely known [14] and documented elsewhere, the adoption of this as a first line technique to minimise PONV has not been routinely translated into clinical practice. In spite of grade A1 evidence to reduce baseline risk of PONV in recent consensus guidelines [15] the uptake of TIVA as a default anaesthetic technique is still slow but should be recommended based on our results. For instance, it is estimated that less than 20% of all general anaesthetics in the UK use propofol as maintenance [16]. The present results re-emphasize the range of advantages of propofol over inhalational anaesthetics, including the lower requirements of postoperative analgesia.

In contrast, most previous systematic reviews comparing TIVA with volatile anaesthetics were primarily focused on, and therefore limited to specific clinical domains, like ambulatory surgery [17], cardiac surgery [18, 19], neurosurgery [20], post-operative pain [21, 22], post-operative complications [23], robotic surgery [24] or were even looking at differences in survival [25]. In addition, some of the reviews significantly date back in time more than 20 years [26].

We found, as a secondary outcome, a moderate but consistent effect on lower pain scores by intravenous anaesthesia compared to volatile anaesthetics. Although associated with higher intraoperative use of analgesia, patients receiving TIVA experienced better pain control immediately after surgery (p = 0.001) and after 24 h (p = 0.1). Recently, some evidence has emerged linking intraoperative propofol exposure with an intrinsic analgesic effect [27] demonstrated as reduced post-operative analgesic requirements [28] and the absence of hyperalgesia [29]. However, in an attempt to address this in a systematic review and meta-analysis, Peng and colleagues recently found substantial heterogeneity among studies, limiting their conclusions [22].

Both PONV and post-operative pain are significant determinants of patient dissatisfaction as Myles and colleagues found in validating the quality of recovery score 40 (QoR40) [30]. Of all recovery domains analysed in a subsequent cohort study only pain and PONV contributed to incomplete patient satisfaction [31] whereas TIVA improves early patient well-being by reducing PONV and pain [32] dependent on the setting [33]. This hypothesis is supported by increased patient satisfaction with propofol in the present study. Another factor, influencing patient outcome is the incidence of post-operative agitation and delirium. In this context, it is noteworthy that after adding three observational studies (4387 patients) in the secondary analyses, there was a reduced risk for agitation with propofol. There are indicative studies associating volatile agents with the development neuro-cognitive dysfunction [34, 35], although an earlier meta-analysis could not demonstrate an association of general anaesthesia per se and risk of Alzheimer’s disease [36].

We found no difference between anaesthetic techniques for other outcomes such as, post-anaesthetic shivering, hospital length of stay or haemodynamic instability requiring unplanned interventions. Although time to tracheal extubation and respiratory recovery was significantly faster for volatile anaesthesia (on average one minute) and time in PACU significantly shorter for TIVA (on average three minutes), we would consider neither as clinically relevant. The lack of difference in these secondary outcomes might be explained in the varying standards and definitions of reporting the criteria in the individual studies as well as other procedure related factors that, independent of the anaesthetic technique, might have influenced outcomes. Some of the main and secondary outcomes were consistent in the domains of the subgroup analyses and in favour of propofol maintenance. Others, like differences in immediate recovery milestones, type of surgery, haemodynamic instability were not. There has been a debate in recent years whether or not volatile anaesthesia offers advantages in survival due to myocardial events in cardiac surgical [37–39] and non-cardiac patients [23, 40]. This systematic review did not demonstrate any contribution of the anaesthetic technique to perioperative survival, although we did not specifically include perioperative myocardial events or its impact on 1-year mortality. The malignancy status of patients undergoing oncological surgical procedures as a contributing factor for differences in outcome of observed parameters could also not be established by our data material although in a recent retrospective database trial an association of survival and type of anaesthetic could be demonstrated [41].

We have to consider some limitations to our analysis and report. Firstly, although large in number, the publications included in our systematic review and meta-analysis represent the scientific output across three decades. Clinical practice is continuously changing and adapting. Older publications may therefore no longer be representative of current scientific consensus and clinical practice. However, over 80% of the studies included data from 2001 onwards and 40% were published between 2011 and 2016 (Additional file 2). Nevertheless, the combination of varying conditions and settings and their heterogeneity into one estimate may limit the clinical interpretation when using a random effects model. Secondly, although the Cochrane tool for assessing risk of bias was systematically applied in the analysis, we have identified some areas of moderate risk of bias (Table 1) [42, 43] as well as domains with a low number of studies in the subgroup analysis. This needs to be considered when interpreting the results [44]. Based on the studies available, there is a lack of RCTs specifically comparing TCI with non-TCI techniques for propofol maintenance and its impact on meaningful outcome parameters. We could not establish enough study power for this comparison in the available collection of RCTs. A previous Cochrane review on TCI has recently been withdrawn owing to the age, retraction of studies, and possible errors in the conversion of dose for the purpose of analysis [http://cochranelibrary-wiley.com/doi/10.1002/14651858.CD006059.pub3/full].

Patient-reported outcomes after surgery is becoming increasingly relevant for health care providers to improve quality of care and to define key performance indicators in perioperative medicine. Reviewing 229 RCTs with 20,991 patients from peer-reviewed publications including a wide variety of clinical specialties and patient populations, we found reduced PONV and better analgesia with TIVA as compared to general anaesthesia with volatile anaesthetics. Although patient discomfort after surgery is common, it is less well understood how it affects patient satisfaction [45]. This large systematic review and meta-analysis shows that in a relevant clinical domain of perioperative care TIVA provides a better patient experience compared to volatile anaesthesia with a high degree of confidence in the effect estimate [46]. These promising findings will hopefully trigger the conduct of prospective validation studies.