Epidural volume extension - a reply

We thank Drs. Tyagi and Ramanujam for their interest in and comments on our systematic review and meta-analysis 1 on epidural volume extension (EVE). We agree that baricity of the injected local anaesthetic, position of the patient and other factors can affect the outcomes of neuraxial anaesthesia. Consequently, we performed sub-group analysis of those studies in which different anaesthetic mixtures were used in the intervention and comparator groups, and a meta-regression to study the effects volume for EVE volume. Other factors like differences in study design, assessment of the outcomes etc, were discussed.

We included trials with combined spinal-epidural (CSE) and spinal anaesthesia as comparators. We would like to emphasise that in the CSE studies no epidural medication was given, thereby excluding an effect on our outcomes such as the height of the sensory block. There has been a discussion whether the spread of a spinal anaesthesia is higher after CSE, that is, when the epidural space has been punctured compared with spinal-only anaesthesia. Studies showing such a difference in the height of block 2 are contrasted by those not finding a difference 3. However, we feel that this question could have been addressed in a sub-group analysis.

The issues raised in the letter relate to heterogeneity, a common phenomenon in meta-analyses. A way to deal with heterogeneity is the performance of sub-group analyses. We registered our meta-analysis in the PROSPERO registry and we had to pre-define a priori our sub-group analyses, limiting the possibility of exploring other factors. The reason, however, why pre-specification is required is that with multiple analyses, the risk of spurious findings (type-1 error of statistical analysis) increases. In our meta-analysis, we chose to adhere to our published protocol and not to perform additional analyses. Therefore, some sources of heterogeneity remain to be discussed, and we are grateful to Drs. Tyagi and Ramanujam for pointing these out.

In a forthcoming editorial in Anaesthesia, we discuss multiple testing and pre-specifications of analyses 4, offering answers to the problem of heterogeneity.