Prediction of postoperative pain: are we missing the target?

Despite much progress in our understanding of acute pain physiology and its treatment, several large clinical surveys continue to demonstrate suboptimal pain management [1, 2]. It may not be rational to apply all available, effective, but sometimes invasive and costly, analgesic techniques to all patients. Consequently, the potential to predict the so-called ‘high-pain responders’ has received much attention, and the problem is divided into the role of pre-, intra- and postoperative predictive factors. With regard to pre-operative factors, characterisation of the nociceptive nervous system function has been largely disappointing regarding the use of pre-operative quantitative sensory testing [3], while detailed characterisation of individual immune signatures may be promising [4], but have still not been tested on a large scale. Recently, Armstrong et al. [2] published a large database study aiming to predict patients at risk for severe pain after major surgery. Although the study included relevant statistical analyses and compared patient characteristics such as sex; diabetes; smoking; use of opioids; and other patient-reported data, the pre-operative factors had little clinical relevance in predicting significant postoperative pain. The use of intra-operative factors, such as type of surgery, improved predictability [2], but were still relatively disappointing. Nevertheless, the type of surgery (which part of the body is operated on, minimally invasive surgery, etc.) is important for the pain response and is well argued by the PROSPECT methodology [5]. However, intra-operative assessment of the nociceptive response has so far been somewhat disappointing in predicting postoperative pain [6]. As to the role of postoperative factors, the study by Armstrong et al. showed earlier achievement of the DrEaMing (delivery of drinking, eating and mobilising) recovery criteria was associated with less postoperative pain, reflecting the importance of enhanced recovery programmes [2]. Finally, the use of machine learning techniques to combine the many predictors may be promising but are, thus far, not clinically significant [7]. So, where do we go from here? There is a general agreement that a personalised medicine approach seems rational, especially regarding the recommendation of more labour-intensive, costly and potentially risky, invasive analgesic procedures [8]. In this context, it is surprising that one single obvious and important factor to determine the risk of early significant pain has not been analysed or planned in detail in all the mentioned studies, including the study by Armstrong et al. Exact data on the provided analgesic management for individual procedures were not available. Consequently, we are left in a vacuum that hinders sufficient interpretation of these ‘predictive’ pain studies, because the specific role of pre- and intra-operative factors cannot be analysed without sufficient standardisation and reporting of postoperative pain management according to the procedure-specific evidence [5]. Indeed, the problem cannot be solved by the hypothesis that ‘larger studies may allow better interpretation’, since they may just add too many confounding factors. Hopefully, the future possibilities to improve prediction of individual ‘high-pain responders’ may come from detailed, prospective studies, including the established relevant pre- and intra-operative factors, combined with implemented and detailed reporting of standardised procedure-specific pain management programmes. Otherwise, we may continue to ‘miss the target’ and the urgent need to improve overall peri-operative pain management.