• anesthesia
• local
• pain
• postoperative
• acute pain

In Regional Anesthesia & Pain Medicine, Hussain and colleagues1 publish the results of a meta-analysis where they ask the question: is there an analgesic benefit in adding erector spinae plane block (ESPB) to parenteral analgesia following surgery for breast cancer? Their results were pooled from 699 patients mostly having modified radical mastectomy. The authors should be congratulated for a priori defining the minimal clinically important difference (MCID) for postoperative pain intensity and opioid consumption. Opioid consumption and pain scores were reduced and although these results were statistically significant (generated low p values), the difference was small, and, therefore, declared clinically unimportant. For example, the mean differences in postoperative pain scores were −0.85 at 12 hours and –0.72 at 24 hours, which did not cross the MCID threshold set by the authors of 2/10 for an individual time point. Hussain and colleagues provide us a refreshing reminder that results that cross the threshold for statistical significance with low p values may be clinically unimportant. Regardless of the MCID used, clinicians and patients may question the therapeutic benefit of ESPB using their threshold of ‘common sense’ where opioid consumption (as reported in this review) is reduced by the equivalent of a few oxycodone tablets in the first 24 hours. As is so often the case in acute pain research, the trials in the Hussain review did not report improvements in health status (eg, patient-reported outcomes, long-term opioid use), further complicating the clinical interpretation of the reduced 24-hour opioid consumption.

A p value <0.05 is arbitrarily used as the threshold to indicate statistical significance. This statistic is immediately recognizable, it is ubiquitous in research and it helps us decide whether a research finding is real or whether it occurred by chance. P values are the first line of defense against being fooled by randomness.2 The words ‘statistically significant’ and p values appear quite frequently in the results and other sections of most manuscripts. However, p values do not contain information about the magnitude of the effect estimate or the precision of that estimate (ie, 95% CI). P values generated from statistical tests simply present the probability that the difference between treatment groups is due to chance. When there are two groups in a study such as ESPB versus paravertebral block, the statistical test tries to tell you if all the data belong to one distribution (no treatment effect) versus two different distributions—one for ESPB and one for paravertebral block (a treatment effect). Thus, a p value of 0.05 means that there is a 5% chance that all the sample data belong to only one distribution. Rejection of the null hypothesis in this situation simply means that we trust with a 95% probability that there really are two distributions (one for ESPB and one for paravertebral block). This seems like a reasonable bet.

However, a low p value alone does not tell us the chance that we correctly rejected the null hypothesis or indicate that a treatment is more plausible.3 In the case of ESPB, its proposed anatomical mechanism is important. My direct observation is that the dense layers of the intrinsic back muscles provide a significant physical obstacle for spread of the injectate to its proposed target, the thoracic spinal nerves.4 Anatomical studies support ESPB as being a dorsal rami block, with limited involvement of the ventral rami. The modest treatment effects reported in this review and in other studies5 support my belief of limited or absent ventral rami involvement. A proportion of the reported ESPB effect may not be a pharmacodynamic, but instead be a positive treatment effect borne out of enthusiastic practitioner endorsement of ESPB.6 In sum, the lack of defined and reproducible anatomical mechanism of action for ESPB, clinical trial results and experience to date cause me to question the anatomical mechanism for ESPB. In this scenario, a low p value alone does not necessarily make the ESPB more plausible as an intervention.3

Hussain and colleagues give us a masterclass on performing a meta-analysis and provide us with an estimate of the analgesic treatment effect of ESPB, which although statistically significant is not clinically important. Despite the modest analgesic benefits, we have not explored all potential improvements in health status that may result from use of ESPB in other breast surgery and patient subpopulations. We need to foster innovative research strategies that seek fundamental knowledge about how our interventions impact postoperative recovery. We should not yet reject ESPB for all patients scheduled for breast cancer surgery but consider the benefits and risks for an individual patient and apply a vigilant filter when results cross the threshold of statistical significance and consider asking ourselves, despite the low p value how strong is the evidence against the null hypothesis and is the effect size clinically important?