BACKGROUND: There is a need for a clinically relevant and feasible outcome measure to facilitate clinical studies in perioperative care medicine. This large-scale retrospective cohort study proposed a novel composite outcome measure comprising invasive respiratory or vasopressor support (IRVS) and death. We described the prevalence of IRVS in patients undergoing major abdominal surgery and assessed the validity of combining IRVS and death to form a composite outcome measure. METHODS: We retrospectively collected perioperative data for 2776 patients undergoing major abdominal surgery (liver, colorectal, gastric, pancreatic, or esophageal resection) at Kyoto University Hospital. We defined IRVS as requirement for mechanical ventilation for ≥24 hours postoperatively, postoperative reintubation, or postoperative vasopressor administration. We evaluated the prevalence of IRVS within 30 postoperative days and examined the association between IRVS and subsequent clinical outcomes. The primary outcome of interest was long-term survival. Multivariable Cox proportional regression analysis was performed to adjust for the baseline patient and operative characteristics. The secondary outcomes were length of hospital stay and hospital mortality. RESULTS: In total, 85 patients (3.1%) received IRVS within 30 postoperative days, 15 of whom died by day 30. Patients with IRVS had a lower long-term survival rate (1- and 3-year survival probabilities, 66.1% and 48.5% vs 95.2% and 84.0%, respectively; P < .001, log-rank test) compared to those without IRVS. IRVS was significantly associated with lower long-term survival after adjustment for the baseline patient and operative characteristics (adjusted hazard ratio, 2.72; 95% confidence interval, 1.97-3.77; P < .001). IRVS was associated with a longer hospital stay (median [interquartile range], 65 [39-326] vs 15 [12-24] days; adjusted P < .001) and a higher hospital mortality (24.7% vs 0.5%; adjusted P < .001). Moreover, IRVS was adversely associated with subsequent clinical outcomes including lower long-term survival (adjusted hazard ratio, 1.78; 95% confidence interval, 1.21-2.63; P = .004) when the analyses were restricted to 30-day survivors. CONCLUSIONS: Patients with IRVS can experience ongoing risk of serious morbidity and less long-term survival even if alive at postoperative day 30. Our findings support the validity of using IRVS and/or death as a composite outcome measure for clinical studies in perioperative care medicine.
BACKGROUND: There is a need for a clinically relevant and feasible outcome measure to facilitate clinical studies in perioperative care medicine. This large-scale retrospective cohort study proposed a novel composite outcome measure comprising invasive respiratory or vasopressor support (IRVS) and death. We described the prevalence of IRVS in patients undergoing major abdominal surgery and assessed the validity of combining IRVS and death to form a composite outcome measure. METHODS: We retrospectively collected perioperative data for 2776 patients undergoing major abdominal surgery (liver, colorectal, gastric, pancreatic, or esophageal resection) at Kyoto University Hospital. We defined IRVS as requirement for mechanical ventilation for ≥24 hours postoperatively, postoperative reintubation, or postoperative vasopressor administration. We evaluated the prevalence of IRVS within 30 postoperative days and examined the association between IRVS and subsequent clinical outcomes. The primary outcome of interest was long-term survival. Multivariable Cox proportional regression analysis was performed to adjust for the baseline patient and operative characteristics. The secondary outcomes were length of hospital stay and hospital mortality. RESULTS: In total, 85 patients (3.1%) received IRVS within 30 postoperative days, 15 of whom died by day 30. Patients with IRVS had a lower long-term survival rate (1- and 3-year survival probabilities, 66.1% and 48.5% vs 95.2% and 84.0%, respectively; P < .001, log-rank test) compared to those without IRVS. IRVS was significantly associated with lower long-term survival after adjustment for the baseline patient and operative characteristics (adjusted hazard ratio, 2.72; 95% confidence interval, 1.97-3.77; P < .001). IRVS was associated with a longer hospital stay (median [interquartile range], 65 [39-326] vs 15 [12-24] days; adjusted P < .001) and a higher hospital mortality (24.7% vs 0.5%; adjusted P < .001). Moreover, IRVS was adversely associated with subsequent clinical outcomes including lower long-term survival (adjusted hazard ratio, 1.78; 95% confidence interval, 1.21-2.63; P = .004) when the analyses were restricted to 30-day survivors. CONCLUSIONS:Patients with IRVS can experience ongoing risk of serious morbidity and less long-term survival even if alive at postoperative day 30. Our findings support the validity of using IRVS and/or death as a composite outcome measure for clinical studies in perioperative care medicine.