Annual Scientific Meeting - 2001

A Meta-analysis of PONV comparing propofol and sevoflurane anesthesia

Alison S Carr MB, Dave Wright PhD & J Robert Sneyd MD, Dept of Anaesthesia, Plymouth, UK.

Introduction: Post-operative nausea and vomiting (PONV) are common causes of morbidity after general anaesthesia. A number of prospective randomised comparator studies have suggested that PONV is reduced after propofol anaesthesia compared with sevoflurane. A previous meta-analysis (1) showed maintenance of anaesthesia with propofol caused less PONV than using inhalational agents, however few references studied sevoflurane. We performed a meta-analysis of published studies comparing propofol and sevoflurane that included data on the incidence of nausea (N) or vomiting (V) or gave a combined incidence of N/V.

Methods: The Zeneca pharmaceuticals MEDLEY database is compiled by regular searches of databases and hand searching of journals. By December 2000, 13068 publications mentioned propofol and 1122 (including duplicates) sevoflurane. A search of MEDLEY highlighted 388 publications comparing propofol and sevoflurane. These publications were examined and 29 prospective randomised comparator studies identified providing patient numbers for N, V or N/V after propofol and sevoflurane. These studies were examined in detail individually and independently by two authors, noting also the induction agent, agent for maintenance of anaesthesia, the presence/absence of nitrous oxide, age of patient, use of opioid and type of surgery undertaken. The significance of the difference in incidence of PONV between groups was demonstrated by applying a sign test to the number of studies. A Bayesian random-effects meta-analysis was used to estimate the odds ratios and expected number needed to treat comparing the different methods of anaesthesia on the incidence of PONV.

Results: Preliminary data from 2763 adult subjects in 29 randomised comparator studies on the incidence of PONV are summarised in Table 1. These studies compared 3 methods of anaesthesia: propofol induction and maintenance (PP), propofol induction followed by sevoflurane maintenance (PS) and sevoflurane induction and maintenance (SS). Note that some studies contain comparisons of all three methods of anaesthesia. Propofol is associated with a lower incidence of PONV than sevoflurane.

Ten out of 11 studies comparing PP with PS show less PONV with PP and there is one tie (P = 0.002). Fifteen out of 18 studies comparing PP and SS show less PONV with PP while 3 show a higher incidence (P = 0.008). Eight out of 9 studies comparing PS and SS show less PONV in the PS group and one is indeterminate (P = 0.008).

The odds ratios for PONV are summarised in table 1. Due to the large general variation in incidence of PONV between different studies, numbers needed to treat vary greatly, ranging upwards from around 5 for PS/PP and around 3 for SS/PP for studies where the incidence of PONV is high.

Conclusions: Patients anaesthetised with propofol-propofol had a lower incidence of PONV in comparison to propofol-sevoflurane and to sevoflurane-sevoflurane. Further analysis may indicate the effect of the presence/absence of nitrous oxide, age of patient, use of opioid and type of surgery.

Summary: A meta-analysis of 29 randomised studies (2763 subjects) comparing propofol (P) and sevoflurane (S) and the incidences of nausea (N) or vomiting (V) suggests PP produces less PONV than PS and SS. Odds ratios (posterior median, 95% Credibility interval) for N were 2.1 (1.2-3.2) for PS/PP and 5.2(2.6-7.7) for SS/PP and for V were 2.6(1.4-4.6) for PS/PP and 5.4(2.5-8.7) for SS/PP.

Reference: 

(1) Eur J Anaesthesiol 15: 433-445, 1998. 

Details of Financial Relationship:

I have applied for a small educational grant from Zeneca, UK to cover the cost of statistical analysis and secretarial support for publication of this study.