Reply from the authors

Abstract

Editor—In response to the letters of Neto and of Xue and colleagues, we would first like to thank both authors for their thoughtful comments. We agree that retrospective studies can only show association, not causation, and that even though the propensity score matching analysis supported the results found in the Cox regression model,1Levin MA McCormick PJ Lin HM Hosseinian L Fischer GW Low intraoperative tidal volume ventilation with minimal PEEP is associated with increased mortality.Br J Anaesth. 2014; 113: 97-108Abstract Full Text Full Text PDF PubMed Scopus (130) Google Scholar the magnitude of the biases in distributions between the tidal volume (TV) groups for certain covariates were considerable and unlikely to be fully offset by the propensity score. We stated this clearly in our discussion. In our final model however, many of the systematically distributed variables had no statistically significant effect on the observed outcome despite disproportionate representation. Among those that did, short length of intubation, low BMI, and female gender were all associated with lower TV IBW (ideal body weight) and increased mortality. Therefore, it is not surprising that after adjusting for these variables and others, the hazard ratios for low TV decreased as compared with the unadjusted hazard ratios shown in Table 1 below:Table 1Unadjusted Hazard Ratio for 30-day mortalityTV IBW groupUnadjusted Hazard RatioP-value3–65.109<0.00016–82.559<0.000110–120.8490.430212–201.2600.5521 Open table in a new tab Neto asked what the 30-day mortality rates were per tidal volume (TV) per IBW group. The rates are shown in Table 2. Unfortunately due to the retrospective design of this study we are unable to comment as to the causes leading to mortality. In regards to the overall mortality reported in our paper, we believe that 1.2% is consistent with another large population based study (1.85%).2Noordzij PG Poldermans D Schouten O Bax JJ Schreiner FA Boersma E Postoperative mortality in The Netherlands: a population-based analysis of surgery-specific risk in adults.Anesthesiology. 2010; 112: 1105-1115Crossref PubMed Scopus (162) Google ScholarTable 230-day mortality by TV IBW groupTidal Volume IBW group30-day mortality3–63.9%6–82%8–10 (reference)0.8%10–120.7%12–201% Open table in a new tab Neto also asked whether it would be possible to perform an analysis using PEEP as the primary variable of interest rather than TV. The range of PEEP in our entire cohort was 2.2–5 cm H2O. Consequently, we were unable to investigate a low TV, high PEEP combination. Additionally, we were unable to find a strong correlation between PEEP and tidal volume (r = 0.033) (Fig 2c of original article). Xue commented that we did not control for preoperative factors such as anaemia and for details of intraoperative management such as RBC transfusions, blood loss, and episodes of haemodynamic instability. Preoperative haemoglobin level was available for only 9177/29 343 patients (31.3%). Of these, only 124 patients (1.4%) had a haemoglobin level of less than or equal to 7 g/dl. Therefore we believe it is unlikely that preoperative anaemia was a significant contributor to post-operative mortality in this cohort. As for other preoperative risk factors, we believe that our use of the all patient refined-diagnosis related group (APR-DRG) Severity of Illness and Risk of Mortality scores provided sufficient risk adjustment to account for additional comorbid conditions. In regards to intraoperative factors, it is beyond the scope of this reply, and was well beyond the scope of our original analysis, to fully analyse the detailed anaesthetic course. We did however re-analyse our data set to examine the use of intraoperative blood transfusions. 2672/29 343 patients (9.1%) received at least one unit of packed red blood cells. The 30-day mortality among patients who were transfused was 5.3% vs 0.76% among patients who were not transfused. Among those patients who were transfused, the median number of units was 2 (1–3) regardless of tidal volume group. Median TV per kg IBW was 8.6 ml kg−1 IBW in patients who were transfused vs 8.62 ml kg−1 IBW in patients who were not. The percentage of cases that were transfused per TV IBW group is shown in Table 3.Table 3Percentage of patients transfused per TV IBW groupTidal Volume IBW group% receiving PRBCs3–613.5%6–89.4%8–10 (reference)8.4%10–129.8%12–2011.7%P<0.0001 for comparison between groups. Open table in a new tab P<0.0001 for comparison between groups. It is interesting to note that both the low and high TV groups had a higher percentage of transfusions than the reference group, yet mortality in the high TV group was significantly less than both the reference 8–10 ml kg−1 IBW group and the low TV groups. We re-ran the Cox regression analysis including an additional variable for transfusion (yes/no) and found that the hazard ratio for TV 6–8 ml kg−1 IBW vs 8–10 ml kg−1 IBW was essentially unchanged, 1.62 (1.26–2.1), P=0.0002. Similarly, when we repeated the propensity score analysis with the inclusion of transfusion as a matching variable, the adjusted hazard ratio for TV 6–8 ml kg−1 IBW vs 8–10 ml kg−1 IBW was 1.68 (1.25–2.27), P<0.001, again essentially unchanged (in fact slightly higher) than in the original analysis. Thus we feel comfortable concluding that use of RBC transfusion was not a significant confounder in this cohort. Finally, we disagree with Neto's comment implying that PEEP should not be used with low TV, since that combination in conjunction with recurrent recruitment manoeuvres is precisely what has been promulgated as a protective ventilation strategy. In conclusion, we once again thank the authors for their observations and hope that our reply adequately addresses their concerns. None declared.