Does a Combined Intravenous-volatile Anesthesia Offer Advantages Compared to an Intravenous or Volatile Anesthesia Alone

A Systematic Review and Meta-analysis


BMC Anesthesiol. 2021;21(52) 

In This Article


The study protocol of this meta-analysis was registered at PROSPERO (International prospective register of systematic reviews;; registration number CRD42019126627).

We searched for trials without any restriction in the databases PubMed, Scopus, Web of Science, and CENTRAL. We used the search terms "sevoflurane AND propofol", "desflurane AND propofol", "isoflurane AND propofol", "volatile AND propofol", "inhal AND propofol", "combined intravenous volatile", and "CIVA". Additionally, references of relevant studies were screened as well as current literature.

Only controlled studies, investigating the effect of combined intravenous volatile anesthesia (CIVA) versus balanced (BAL) or total intravenous anesthesia (TIVA) in English or German language and providing data on postoperative nausea and vomiting (PONV), time to extubation, or pain perception were included.

If a study had more than one active drug arm, data were extracted for each treatment arm and included separately in the analysis. Duplicate use of the same placebo group was then automatically factored in by the meta-analysis software used.

Furthermore, randomized and non-randomized studies were analyzed and compared separately.

All complete papers reporting trials were rated independently by two investigators (M.U. and A.W.). Data were extracted onto standard simple forms. Any disagreement was discussed with additional reviewers (HS, HH), and decisions were documented. If necessary, authors of studies were contacted for clarification. The risk of bias was assessed on a sectoral basis: generation of random sequences, concealment of assignments, blinding, incomplete result data, selective reporting.

The primary outcome was PONV in the post anesthesia care unit (PACU) or recovery room (RR). The secondary outcome was PONV within 24 h, time to extubation, movement during surgery, pain intensity in the PACU/RR and pain intensity within 24 h.

Statistical Analysis

We analyzed pooled studies using BAL and pooled studies using TIVA, as well as the overall effect. In a sensitivity analysis we excluded non-randomized studies and considered only randomized controlled trials.

The outcome data were combined in a meta-analysis. We calculated the risk ratio (RR) for dichotomous data such as the occurrence of PONV and movement during surgery. For continuous data like time to extubation and pain intensity we calculated the standardized mean difference (SMD) and their 95% confidence interval (CI) as effect size measure.

We used the model of random effects due to the inhomogeneity of the studies themselves, such as different types of surgery (thoracic, laparoscopic, ear/nose/throat) and study populations (gynecological vs. non-gynecological) and due to different heterogeneous results in the studies. Study heterogeneity was assessed by a Chi-square test and the I-square statistic.[1] The Chi-square test compares the effect sizes of the individual trials with the pooled effect size. Significance levels of p < 0.1 were determined a priori in order to assume the presence of heterogeneity. The I-square statistic provides an estimate of the percentage of variability due to heterogeneity rather than chance alone. We interpreted values ≥50% as considerable heterogeneity.[1] If the results were statistically significantly heterogeneous, reasons for the heterogeneity were searched for by re-reading the publications, verifying the extracted data and looking for deviations in the study methodology that explain the heterogeneity. Small studies with negative results are less likely to be published than studies with significant results. The possibility of such a publication bias was examined using the funnel plot method described by Egger et al..[2]

The meta-analytical calculations were performed using the Comprehensive Meta-analysis version 3. The exact formulas are reported there.[3] A p-value < 0.05 was considered statistically significant.