Strike a pose

This Editorial accompanies the following original articles: POSE-Study group. Peri-interventional outcome study in the elderly in Europe: a 30-day prospective cohort study. Eur J Anaesthesiol 2022; 39:198–209. Kowark A, Berger M, Rossaint R, Schmid M, Coburn M., the POSE-Study group. Association between benzodiazepine premedication and 30-day mortality rate: A propensity-score weighted analysis of the Peri-interventional Outcome Study in the Elderly (POSE). Eur J Anaesthesiol 2022; 39:210–218. In this issue of the Journal, we publish side by side, primary and secondary analyses of the Peri-interventional Outcome Study in the Elderly (POSE).1,2 The POSE study is a prospective observational multicentre cohort study of 30-day peri-operative mortality and complications in approximately 9500 elderly patients exceeding 80 years of age in 177 hospitals from 20 European countries. The project was funded and sponsored by the Academic Clinical Research Organization of the European Society of Anaesthesiology, the legacy society of what has since become the European Society of Anaesthesiology and Intensive Care. The observed 30-day mortality was 4.2% with the majority of deaths occurring in hospital. Complications occurred in 17.4% in-hospital and a further 3.9% postdischarge.1 Albeit the overall complication rate of 17.4% might appear ‘high’ in this population, significant complications such as unplanned admissions to ICU and prolonged hospital stay were low, as were the incidence of serious complications (pulmonary, cardiac, renal and neurological) indicating that anaesthesia and the peri-operative period are generally well tolerated in fragile elderly citizens. Apart from the estimated 30-day mortality rate, the authors identified several independent risk factors associated with an increased risk of mortality.1 The majority of the independent risk factors are nonmodifiable such as age, frailty, limited mobility, degree of urgency of the procedure and multimorbidities. Some of the very few modifiable factors that were independently associated with an increased risk of mortality are already documented for all patients, such as transfusions of blood products. Interestingly, the authors found that admission to a unit with geriatric care was associated with improved survival. Otherwise, few modifiable factors associated with improved survival were identified in this study and are hardly surprising. As part of a planned secondary analysis, perhaps the most unexpected, and potentially contentious finding, was the identification of a possible signal to a beneficial effect of midazolam premedication on survival in this cohort.2 A significant unadjusted hazard ratio of 0.58 (95% CI 0.40 to 0.85) for all-cause mortality was found as part of the primary analysis. The secondary analysis reported an adjusted hazard ratio of 0.71 (95% CI 0.49 to 1.04). As this is an observational study, devoid of randomisation, the authors used inverse propensity score weighted regression to attempt to control for numerous confounding variables and baselines in estimating the adjusted hazard ratio. The authors suggest that the anxiolytic effect of midazolam may improve mortality in this vulnerable group by mitigating peri-procedure stress. For readers of this article,2 the unexpected beneficial effect on survival of premedication with midazolam contradicts current practice. Several prospective randomised studies were not able to demonstrate a benefit of benzodiazepines premedication on the quality of postoperative recovery.3–7 Considering the risks associated with benzodiazepine in the elderly, avoidance of premedication would seem reasonable. This editorial is provided to try to address the issues. During the review process, Editors and Reviewers of the manuscript did address, as much as was practical, the robustness of the statistical associations between premedication with midazolam and improved 30-day survival. Given the considerable imbalances in baseline measures and likely measurable and unmeasurable clinical heterogeneities in practice, we cannot expect statistical modelling alone to provide the answer. Whatever the robustness, we have to accept that the use of midazolam for a particular patient must be influenced by clinician, institutional and procedural factors and, as such, the interpretation of any potential benefit must be subject to confounding by indication in such an observational study.8 Even if such considerations were fully resolved, the potentially beneficial effect of premedication with benzodiazepines, specifically midazolam, and its adoption in routine clinical practice would require the demonstration of a causal link. Causality, or causal inference, is one of the most difficult problems in medicine and biology, even in the twenty-first century.9 If we attempt to apply the nine criteria, actually viewpoints, initially developed by Hill in 1965 to the article of Korwak et al.,2 very few would stand for a causal link between benzodiazepine premedication and improved survival.10Strength of association is weak and not overly supportive of causality. Consistency, where multiple studies confirm the association, is difficult to confirm because there are a few studies on this topic, despite the risks associated with premedication with benzodiazepines. Specificity implies that a credible mechanism linking benzodiazepine premedication with improved survival can be described. The authors suggest attenuated stress as a casual mechanism but this is purely speculative and hardly credible that a single dose of benzodiazepine could have such a specific effect on all-cause 30-day mortality. Temporality is there in that premedication preceded the occurrence of death, but again it is difficult to accept that a single dose would precede such improved survival. Biological gradient, such as a dose-response effect, cannot be assessed as no information is presented on different doses of benzodiazepines.11Plausibility, where current knowledge and understanding is consistent with the observation, is again difficult to evaluate due to a paucity of relevant data related to stress or other mechanisms. Coherence with other classes of drugs used for premedication is not supported in this report due to limited data on other drugs. Experiment, such as a randomised controlled trial (RCT) for a beneficial effect of benzodiazepines on survival is of course lacking, but this report does provide (some) support for conducting such an RCT. Analogy again is difficult to support due to the lack of evidence for other strong effects on mortality by other similar interventions. Taken together, although conceptually difficult, the causality viewpoints applied to the Korwak et al.2 article are currently not in favour of a causal link between the use of benzodiazepine premedication in the elderly and improved survival. Benzodiazepines are associated with side effects, especially in elderly patients. Respiratory depression and respiratory arrest have been reported up to 2 h after small doses of midazolam.12 Cognitive disorders and postoperative delirium are also well known effects of benzodiazepines that have been associated with poor outcomes. So, what would explain a beneficial effect of these drugs? It is important to note that only 16% of patients received a premedication in the POSE study. Hence, it would not make sense to suggest a benefit for all patients but only for selected ones. One effect could be related to attenuation of the stress response to anaesthesia and surgery in the most anxious patients.13 Stress-free anaesthesia associated with reduced stress hormone secretion was a popular concept 25 years ago. Another benefit would be related to improved patient satisfaction. Retrospective analyses of large databases have shown an association between satisfaction and a decreased risk of myocardial infarction, pneumonia, readmission rate and mortality.14,15 However, benzodiazepine premedication did not demonstrate improved patient satisfaction in a prospective randomised study.5 One should keep in mind that the most anxious patients, who asked for pre-operative anxiolysis, were not included in the study for obvious ethical reasons. Finally, the target population for premedication could be the most anxious patients, in good clinical condition, without risk factors for respiratory depression or cognitive alterations. Prospective randomised studies will have to test this hypothesis. The answer to some of these questions may come from a prospective RCT that is in progress, the PReOperative Midazolam on OuTcome of Elderly patients (I-PROMOTE trial). The primary outcome is global patient satisfaction, not mortality, however, which is a secondary outcome. With a planned enrolment of 614 patients, it is highly unlikely that the study will have sufficient power (<20%) to find a significant reduction in the mortality from that (4.45%) reported here,2 albeit that in the RCT16 the patients will all have been operated upon under general anaesthesia. So, although the POSE papers published here are of interest and raise some interesting questions, it is not appropriate at present to recommend benzodiazepines based on a putative, likely confounder with possible overestimation of a survival benefit in this patient population.17 Also, given the limited size of the I-PROMOTE trial,16 it is not likely to be answered in the near future. The strength of the POSE project1,2 is that it provides some justifications for such RCTs and the authors are to be congratulated on taking this forward.