Pre‐operative hypertension and intra‐operative hypotension: a reply

We thank Drs. Rook and Yeung for their comments on some of the important limitations of our study. The premise for our study was to determine if the proposed joint guidelines of the Association of Anaesthetists and the British Hypertension Society for the management of hypertensive patients undergoing surgery were reasonable, particularly considering the limited evidence to support them 1. In order to address this question, we adopted a pragmatic study design, with the ability to include the known independent risk factors for adverse peri-operative outcomes in non-cardiac surgery, in the logistic regression model 2. To be pragmatic, we did not demand invasive blood pressure measurement, but rather accepted standard-care blood pressure measurement techniques, which included non-invasive blood pressure measurement. The use of non-invasive blood pressure measurement may certainly increase the proportion of unidentified hypotensive episodes in the peri-operative period 3. However, the standard definitions (on which we base our daily practice) of clinically important hypotension have been derived in a similar manner, from patients who have had predominantly non-invasive blood pressure readings 4. It is, therefore, likely that the definitions of clinically important hypotension would change if we only studied patients with invasive blood pressures, as this would increase both the incidence and severity of reported hypotension 3. Therefore, if we had adopted an approach of recruiting only patients with an invasive blood pressure measurement, the definitions of clinically important blood pressure used in the study would have been questionable, and the generalisability of the findings to clinical practice would have been compromised. Furthermore, in order to ensure a pragmatic study design, we selected a simple binary outcome of a mean arterial pressure (MAP) of 55 mmHg. The systematic review by Wesselink et al. provides support for this decision 5. Indeed, the data presented in Table 4 of their review, suggests that the risk of end-organ damage is high (OR > 2) in 24 out of 27 (89%) associations with a MAP < 55 mmHg, and conversely, there is not a single high-risk association with end-organ damage between a MAP of 55 and 65 mmHg, provided the duration does not exceed 20 min at these pressures. We believe that our surrogate outcome definition was, therefore, surprisingly well selected. Unfortunately, as we did not collect other intra-operative blood pressure data, we cannot comment on the association between hypertension and other MAP hypotensive thresholds in this dataset. In our study, the attending anaesthetists were required to document the duration of hypotension in minutes. We expect that there may well have been a Hawthorne effect as a result. However, this approach to recording patient blood pressure data in studies has been used in important global studies that have and continue to define our understanding of adverse outcomes associated with peri-operative hypotension 6. Interestingly, although the introduction of electronic record keeping has shown to increase the number of identified hypotensive episodes (as would be expected 3), it did not increase the rate of hypotensive interventions 7. It is also important to note, that in our normal clinical practice, anaesthetists do not readily accept a MAP as low as 55 mmHg. For example, in the standard-care arm of Futier et al.'s trial, the MAP was 75 mmHg 8. Our attending anaesthetists did intervene more often in the hypertensive group based on a significantly higher use of vasopressors. However, vasopressor usage was not risk-adjusted for other factors that may have required more vasopressor usage, such as type of surgery, blood loss or other co-existing comorbidities, and therefore it is difficult to ascribe this purely to the presence of hypertension. Based on these findings, we do not believe that the Hawthorne effect had a substantial impact on our findings. Finally, as hypertension is not a consistently independent predictor of major adverse cardiac events 9, it is likely that other co-existing patient comorbidities and intra-operative factors may provide a larger proportional contribution to clinically important intra-operative hypotension, than hypertension itself 10. Although the review by Wesselink et al. highlights a number of studies of intra-operative hypotension 5, the risk adjustment for other risk factors contributing towards hypotension in most of these studies is limited. A strength of our study was the a priori decision to specifically include known independent predictors of adverse peri-operative outcomes in the logistic regression, in order to determine their contribution to clinically important hypotension in hypertensive surgical patients. We accept that this methodology has some inherent weaknesses, which have previously been eloquently discussed in an accompanying paper 11. In conclusion, although our study has numerous limitations and weaknesses, we believe that its strength lies in providing some support for the Association of Anaesthetists peri-operative hypertension guidelines. Furthermore, it did not find any compelling reasons not to follow the proposed guidelines 1. Finally, we agree that this is an area which requires further work, especially as perioperative hypotension has been associated with significant morbidity 4. Clinician researchers with an interest in this field could make an important contribution to the evidence by participating in the peri-operative ischemic evaluation-3 (POISE-3) trial, which includes a blood pressure management arm as one of the trial interventions (ClinicalTrials.gov NCT03505723) .