Abstract

Editor’s Perspective
What We Already Know about This Topic
  • Arterial pressure is a complex signal that is characterized by three primary components — systolic, diastolic, and mean pressure, along with a derived component, pulse pressure (systolic minus diastolic pressure)

  • Each blood pressure component reflects distinct hemodynamic variables, and therefore presumably differently influences perfusion of various organs

  • Previous work identifies associations between intraoperative systolic and mean hypotension with myocardial and kidney injury

What This Article Tells Us That Is New
  • For each blood pressure component, the authors report significant and clinically meaningful associations between the lowest pressure sustained for 5 min and myocardial and kidney injury

  • Absolute population risk thresholds were similar for myocardial and kidney injury, being roughly 90 mmHg for systolic, 65 mmHg for mean, 50 mmHg for diastolic, and 35 mmHg for pulse pressures

  • The odds for myocardial and kidney injury progressively increased with duration and severity of hypotension below each threshold, even after adjusting for potential baseline confounding factors

Background

Arterial pressure is a complex signal that can be characterized by systolic, mean, and diastolic components, along with pulse pressure (difference between systolic and diastolic pressures). The authors separately evaluated the strength of associations among intraoperative pressure components with myocardial and kidney injury after noncardiac surgery.

Methods

The authors included 23,140 noncardiac surgery patients at Cleveland Clinic who had blood pressure recorded at 1-min intervals from radial arterial catheters. The authors used univariable smoothing and multivariable logistic regression to estimate probabilities of each outcome as function of patients’ lowest pressure for a cumulative 5 min for each component, comparing discriminative ability using C-statistics. The authors further assessed the association between outcomes and both area and minutes under derived thresholds corresponding to the beginning of increased risk for the average patient.

Results

Out of 23,140 patients analyzed, myocardial injury occurred in 6.1% and acute kidney injury in 8.2%. Based on the lowest patient blood pressure experienced for greater than or equal to 5 min, estimated thresholds below which the odds of myocardial or kidney injury progressively increased (slope P < 0.001) were 90 mmHg for systolic, 65 mmHg for mean, 50 mmHg for diastolic, and 35 mmHg for pulse pressure. Weak discriminative ability was noted between the pressure components, with univariable C-statistics ranging from 0.55 to 0.59. Area under the curve in the highest (deepest) quartile of exposure below the respective thresholds had significantly higher odds of myocardial injury after noncardiac surgery and acute kidney injury compared to no exposure for systolic, mean, and pulse pressure (all P < 0.001), but not diastolic, after adjusting for confounding.

Conclusions

Systolic, mean, and pulse pressure hypotension were comparable in their strength of association with myocardial and renal injury. In contrast, the relationship with diastolic pressure was poor. Baseline factors were much more strongly associated with myocardial and renal injury than intraoperative blood pressure, but pressure differs in being modifiable.

Arterial pressure is a complex signal that is characterized by three primary components—systolic, diastolic, and mean pressure—along with a derived component, pulse pressure (systolic minus diastolic pressure). Each blood pressure component reflects distinct hemodynamic variables, and therefore presumably differently influences perfusion of various organs. For instance, ventricular contractility and vascular resistance directly influences steady components—systolic, diastolic and mean pressures—whereas ventricular ejection and vascular compliance affects pulse pressure curve variations. Narrow pulse pressure is also thought to indicate reduced cardiac output and increased peripheral resistance.1 

Previous work identifies associations between intraoperative systolic and mean hypotension with myocardial and kidney injury.2–7  But diastolic pressure is an important determinant of myocardial perfusion, at least in noncardiac settings, and may also be important in surgical patients.8,9  Research on perioperative pulse pressure remains scarce, although elevated pulse pressure is reported to be associated with poor cardiovascular outcomes independent of hypertension.10,11  Which perioperative blood pressure component(s) are most related to critical organ perfusion remains unknown and there is no consensus about which blood pressure component should be targeted in surgical patients. Perhaps consequently, most studies of perioperative blood pressure have been based on essentially arbitrarily selected components.

Complex, time-varying biologic signals are often best characterized by their central tendency, explaining why we2,3,12–14  and others4,6,15–23  used the mean to evaluate relationships between blood pressure and outcomes such as myocardial and kidney injury. While using the mean is a reasonable mathematical approach, blood pressure presents a special challenge in being characterized by systolic, mean, and diastolic components, along with pulse pressure which is a measure of signal excursion. There was thus no a priori reason to believe that any one component would be especially predictive for myocardial or kidney injury. But if one were clearly superior, it would presumably become the basis of future pressure-directed trials.

Which blood pressure component is chosen matters because some may better predict organ injury than others, and of course harm thresholds will differ for each. We therefore evaluated associations of intraoperative systolic, mean, diastolic, and pulse pressure with myocardial and kidney injury. Specifically, we sought to determine which components were most strongly associated with each of myocardial and renal injury after noncardiac surgery.

Materials and Methods

We conducted a post hoc re-analysis of a large single-center retrospective cohort.2  Exposures and outcomes were defined a priori. The statistical plan for the current analysis was approved December 4, 2017 by our institutional review board (No. 17-1627; Cleveland, Ohio) before data were accessed for this analysis. We restricted analysis to patients who had blood pressure recorded from a radial artery catheter.

With institutional review board approval (No. 14-418) and waived consent, the original study2  included 57,315 adult patients who had noncardiac surgery between January 6, 2005 to March 1, 2014 at the Cleveland Clinic’s main campus, with pre- and postoperative creatinine value within 7 days of surgery, and baseline blood pressure recorded in preanesthetic clinic or appointment visits within 6 months before surgery.

Patients were excluded if they had urologic procedures (including relief of urinary obstruction, nephrectomy, or renal transplantation), chronic kidney disease (defined as 30-day preoperative estimated glomerular filtration rate of less than 60 ml · min−1 · 1.73 m−2, calculated using the Cockcroft–Gault equation; preoperative serum creatinine was extracted from the most recent available preoperative visit), or required preoperative dialysis. Finally, patients who had anesthesia lasting less than 60 min, missing baseline variables, or invalid or unavailable intraoperative blood pressure data for more than 10 consecutive min were excluded.

The original study2  demonstrated comparable associations between intraoperative absolute and relative mean pressure hypotension with postoperative myocardial and kidney injury. The main outcome was that patients who had postoperative myocardial and kidney injuries had higher area under various thresholds and more minutes below absolute thresholds.

Blood Pressure Components and Artifact Removing Algorithm

We restricted our analysis to invasive pressures because oscillometrically determined systolic and diastolic pressures are indirect measurements calculated from the mean pressures using various proprietary algorithms which may not be reliable.24,25  Intraoperative mean arterial pressures (MAPs) recorded in the Perioperative Health Documentation System cannot be modified by clinicians, but can be identified as artifacts. As in previous studies, we removed artifacts using the following rules, in order: (1) blood pressures documented as artifacts; (2) pressures out-of-range defined by systolic greater than or equal to 300 or less than or equal to 20 mmHg, systolic less than or equal to diastolic plus 5 mmHg, or diastolic less than or equal to 5 mmHg or diastolic greater than or equal to 225 mmHg; and (3) abrupt changes defined by systolic change greater than or equal to 80 mmHg within 1 min in either direction or greater than or equal to 40 mmHg within 2 min in either direction.2 

Baseline blood pressure components were calculated for each patient as the average of all measurements in the 6 months before surgery. Anesthesia time was defined as the interval between induction and emergence. Induction was defined by injection of the induction medications, and emergence as the period from when minimum alveolar concentration fraction decreased to 0.3 toward the end of the procedure until patients left the operating room.

Confounding Variables

Potential confounding variables are listed in table 1. We defined preexisting medical conditions using International Classification of Diseases, Ninth Revision procedure codes and included only those fulfilling at least one of the following: (1) appeared in the patient’s “problem list” with a date preceding the date of surgery; (2) appeared in procedure code list before the index surgery; or (3) were flagged as a chronic condition based on Healthcare Cost and Utilization Project definitions. Because there were many types of surgical procedures, we characterized each procedure code into one of 231 clinically meaningful categories using the Agency for Healthcare Research and Quality’s Clinical Classifications Software for Services and Procedures. We then aggregated low-frequency event or nonevent categories (N < 10) into one group and used that as the reference group (a low-risk group).

Table 1.

Patient Baseline and Intraoperative Characteristics by Postoperative MINS and AKI

Patient Baseline and Intraoperative Characteristics by Postoperative MINS and AKI
Patient Baseline and Intraoperative Characteristics by Postoperative MINS and AKI

Outcomes

(1) Myocardial injury was defined by elevation in fourth-generation troponin T ≥ 0.03 ng/ml26  or creatinine kinase–myocardial bound > 8.8 ng/ml27  during the first 7 postoperative days.

2) Acute kidney injury (AKI) was defined as an increase in postoperative serum creatinine concentration during the first 7 postoperative days by more than 1.5-fold or greater than 0.3 mg/dl.28  Preoperative concentration was defined as the most recent recorded measurement within 30 days before the surgery.

Statistical Analysis

Determining Blood Pressure Thresholds.

We used graphical and statistical methods to assess the threshold for each blood pressure component below which the risk of myocardial or kidney injury begins to increase. The exposure of interest for each component was the lowest value maintained for cumulative, but not necessarily contiguous, 5 min.

Specifically, we first assessed the univariable relationship between myocardial and kidney injury and the lowest cumulative 5 min for each blood pressure component using moving-average smoothing plots. Univariable moving average plots were constructed for each (binary) outcome variable as a way to display the relationship between a binary outcome and continuous predictor. Starting from the lowest values of the predictor variable, the proportion with the outcome was calculated and plotted for a fixed number of subjects; the bin was repeatedly moved to the right by a fixed number of subjects to create multiple overlapping bins until the proportion with outcome was estimated and plotted for the entire range of the predictor. We then used multivariable logistic regression to model the relationships while adjusting for confounding; a linearity test between each blood pressure component and response was modeled by a restricted cubic spline function with three knots, located at the 10th, 50th, and 90th percentiles, and Wald chi-square test.

Univariable moving-average plots and multivariable smoothed cubic spline curves were used to visualize threshold pressures at which the odds of poor outcome began to increase. Guided by the visual trends, we used the threshold logistic regression method developed by Fong et al. to statistically estimate the threshold pressures for each component.29  Specifically, we fitted a univariable threshold logistic regression with a hinge effect of threshold (i.e., assuming zero slope for the range of higher pressures before the threshold) to estimate component-specific threshold pressures, or change-points, and their 95% CIs. We then determined threshold pressures combining evidence from the inspection of the curves of univariable/multivariable exposure versus outcome with the change-point testing from the threshold logistic regressions.

We also used a multivariable piecewise logistic regression, adjusted for potential confounding variables, to evaluate slopes before and after the chosen thresholds. Here, the exposure of the lowest blood pressure for a patient for a cumulative 5 min (“lowest pressure”) for each of myocardial and kidney injury were partitioned into the two intervals determined by the estimated thresholds, and a separate line segment was fit to each interval (piece-wise regression). Odds ratios were estimated for each segment (i.e., odds of outcome for a decrease of a specified amount from the threshold) and Bonferroni-corrected CI were reported (significance criterion of 0.0125 [0.05/4]; 98.75% CI).

Finally, we evaluated whether the relationship between the lowest blood pressure component values and outcome relationship changed as a function of baseline blood pressures (i.e., interaction between quartiles of baseline and lowest blood pressure components).

Blood Pressure Exposures.

Based on the absolute thresholds of each blood pressure component obtained by graphical and statistical methods, we then modeled our main exposures as: (1) the area under each threshold which represents severity and duration of each component of hypotension, defined as sum of all areas (a1+a2+a3…) below a specified threshold, where areas were calculated using the trapezoid rule and interpolating between measurements; and, (2) the number of minutes under each threshold, defined as total duration (t1+t2+t3…) of time spent under each absolute threshold for a given pressure component. Supplemental Digital Content 1, figure S1 (http://links.lww.com/ALN/C93) illustrates an example of the method for calculation of the area (mmHg × min) and number of minutes under a threshold.

Univariable and multivariable logistic regression were used to assess associations between each blood pressure exposure (area and minutes under each threshold) and postoperative myocardial or kidney injury. All potentially confounding variables listed in table 1 were forced into the models regardless of statistical significance. Linearity of the relationship between each exposure and outcome was assessed.

The C-statistic (area under the receiver operating characteristic curve) from the univariable logistic regression was used to compare discriminative ability among blood pressure components. Descriptively, the component with the highest observed C-statistic was considered to have the best discriminative ability for each major outcome. However, components were also compared statistically using bootstrap resampling with replacement during 1,000 runs, accounting for the inherent correlation among the blood pressure components. Specifically, for each bootstrap run, the six differences between the four components on the C-statistic were recorded, then CIs and standard error of the differences estimated using the distribution of differences across bootstrap runs.

We further provide an intuitive interpretation of the C-statistic for each exposure as the Wilcoxon–Mann–Whitney probability, P, indicating the probability that a randomly chosen patient with myocardial or kidney injury has a worse value of the outcome than a randomly chosen patient without the outcome. We also report the Wilcoxon–Mann–Whitney odds associated with each relationship of interest, calculated as P/(1-P), which indicates the estimated odds of having a worse value in those with versus without myocardial or kidney injury.

Exposure Categories.

Since all relationships were nonlinear, we categorized patients as belonging to either a reference group in which patients spent no time under a given threshold or to one of four groups based on quartiles of time spent under the threshold. Multivariable logistic regression was used to assess the association between categorized blood pressure exposures (area and minutes under the thresholds) and postoperative myocardial and kidney injury. All potentially confounding variables listed in table 1 were forced into the models regardless of statistical significance. Bonferroni correction was used to adjust for four main comparisons within each exposure of interest, with P < 0.0125 (i.e., P < 0.05/4) considered statistically significant. Interactions between baseline blood pressure components and exposures were considered significant if P < 0.05.

Sensitivity Analyses.

In the first post hoc sensitivity analysis, we assessed whether the relative discriminative ability of the components depended on age group (younger than 65 vs. 65 or older) or American Society of Anesthesiologists Physical Status by estimating the C-statistic and 95% CI for area under the estimated thresholds within each category. In a second sensitivity analysis, we assessed the impact of including quadratic plus linear terms for continuous confounding variables (age, preoperative hemoglobin, log [surgery duration], log [estimated blood loss], and log [estimated glomerular filtration rate)] in our multivariable models for association between area under the threshold and outcome compared to the primary analysis, which only included linear terms.

Sample Size Considerations.

The original study on which this analysis was based included about 57,000 patients, and about 40% of the patients had arterial catheters. With the 23,000 available patients and overall incidence of 6% for myocardial injury and 8% for AKI, we had good statistical power to detect clinically important differences, adjusting for multiple comparisons. For example, we had 95% power to detect an absolute difference of 2% in myocardial injury between those who never went below mean pressures of 65 mmHg (4% myocardial injury; N = 3,500) and each of the groups of patients who had some exposure (roughly N = 5,000 for each quartile of exposure). Power was higher for AKI since the overall incidence was higher.

All analyzes were performed with the use of SAS software, version 9.4 (SAS Institute, USA), and R 3.4.1 software (Institute for Statistics and Mathematics, Austria).

Results

Among 164,514 patients who had noncardiac surgery between 2005 and 2014, 23,140 patients met our inclusion and exclusion criteria (fig. 1). The incidence of myocardial and kidney injury was 6.1% and 8.2%, respectively. Only 5,699 patients (25%) had postoperative troponin concentration assayed, and we assumed that patients without the test did not have myocardial injury. Patient’s baseline, preoperative, and intraoperative characteristics are shown in table 1.

Fig. 1.

Flow chart of the study. EGFR, estimated glomerular filtration rate.

Fig. 1.

Flow chart of the study. EGFR, estimated glomerular filtration rate.

Univariable analyses showed that patients having postoperative myocardial or kidney injury had higher intraoperative area under thresholds and number of minutes under threshold for each blood pressure component compared to those with no evidence of myocardial or kidney injury (all P < 0.001; Supplemental Digital Content 2, table S1 http://links.lww.com/ALN/C94). As might be expected, there were no clinically-important associations between intraoperative time-weighted average blood pressure components (i.e., without using thresholds) and having either myocardial or kidney injury.

Blood Pressure Thresholds

Visual Thresholds.

Univariable moving average and multivariable spline smoothing plots for lowest observed blood pressure components for each patient are shown for myocardial injury in figure 2, A and B and for AKI in figure 3, A and B. Blood pressure components below the thresholds of 90 mmHg for systolic, 65 mmHg for mean, 50 mmHg for diastolic, and 35 mmHg for pulse pressure were visual change-points associated with increasing odds of myocardial and kidney injury.

Fig. 2.

Relationship between lowest blood pressure values and myocardial injury. Univariable and multivariable relationship between myocardial injury and lowest blood pressure for a cumulative 5 min for each of four blood pressure components. (A) Estimated probability of myocardial injury from a univariable moving-window with a bin width of 10% of the data; (B) Multivariable logistic regression smoothed by restricted cubic spline with 3 knots at 10th, 50th, and 90th percentiles of given blood pressure component. Multivariable models adjusted for covariates in table 1. Based mainly on the multivariable plots, blood pressure component thresholds of 90 mmHg for systolic blood pressure, 65 mmHg for mean arterial pressure (MAP), 50 mmHg for diastolic blood pressure (DBP), and 35 mmHg for pulse pressure (PP) were visual change-points associated with increasing odds of myocardial injury. Histogram at the bottom of each graph shows the fraction of patients at each lowest blood pressure value. The blue lines in (A) and smoothed lines with 95% confidence bands in (B) indicate estimated probability of myocardial injury as a function of the lowest 5 min of each component. MINS, myocardial injury after noncardiac surgery; SBP, systolic arterial pressure.

Fig. 2.

Relationship between lowest blood pressure values and myocardial injury. Univariable and multivariable relationship between myocardial injury and lowest blood pressure for a cumulative 5 min for each of four blood pressure components. (A) Estimated probability of myocardial injury from a univariable moving-window with a bin width of 10% of the data; (B) Multivariable logistic regression smoothed by restricted cubic spline with 3 knots at 10th, 50th, and 90th percentiles of given blood pressure component. Multivariable models adjusted for covariates in table 1. Based mainly on the multivariable plots, blood pressure component thresholds of 90 mmHg for systolic blood pressure, 65 mmHg for mean arterial pressure (MAP), 50 mmHg for diastolic blood pressure (DBP), and 35 mmHg for pulse pressure (PP) were visual change-points associated with increasing odds of myocardial injury. Histogram at the bottom of each graph shows the fraction of patients at each lowest blood pressure value. The blue lines in (A) and smoothed lines with 95% confidence bands in (B) indicate estimated probability of myocardial injury as a function of the lowest 5 min of each component. MINS, myocardial injury after noncardiac surgery; SBP, systolic arterial pressure.

Fig. 3.

Relationship between lowest blood pressure values and acute kidney injury (AKI). Univariable and multivariable relationship between AKI and lowest blood pressure for a cumulative 5 min for each of four blood pressure components. (A) Estimated probability of AKI from the univariable moving-window with a bin width of 10% of the data; (B) multivariable logistic regression smoothed by restricted cubic spline with 3 knots at 10th, 50th, and 90th percentiles of given blood pressure component. Multivariable models adjusted for covariates in table 1. Based mainly on the multivariable plots, blood pressure component thresholds of 90 mmHg for systolic blood pressure (SBP), 65 mmHg for mean arterial pressure (MAP), 50 mmHg for diastolic blood pressure (DBP), and 35 mmHg for pulse pressure (PP) were visual change-points associated with increasing odds of AKI. Histogram at the bottom of each graph shows the fraction of patients at each lowest blood pressure value. The blue lines in A and smoothed lines with 95% confidence bands in B indicate estimated probability of AKI as a function of the lowest 5 min of each component.

Fig. 3.

Relationship between lowest blood pressure values and acute kidney injury (AKI). Univariable and multivariable relationship between AKI and lowest blood pressure for a cumulative 5 min for each of four blood pressure components. (A) Estimated probability of AKI from the univariable moving-window with a bin width of 10% of the data; (B) multivariable logistic regression smoothed by restricted cubic spline with 3 knots at 10th, 50th, and 90th percentiles of given blood pressure component. Multivariable models adjusted for covariates in table 1. Based mainly on the multivariable plots, blood pressure component thresholds of 90 mmHg for systolic blood pressure (SBP), 65 mmHg for mean arterial pressure (MAP), 50 mmHg for diastolic blood pressure (DBP), and 35 mmHg for pulse pressure (PP) were visual change-points associated with increasing odds of AKI. Histogram at the bottom of each graph shows the fraction of patients at each lowest blood pressure value. The blue lines in A and smoothed lines with 95% confidence bands in B indicate estimated probability of AKI as a function of the lowest 5 min of each component.

Statistical Change-Point Analysis.

From the threshold regressions for myocardial injury, the estimated change-point (95% CI) was 87 (85 to 90) for systolic, 65 (63 to 69) for mean, 51 (48 to 56) for diastolic, and 35 (34 to 37) mmHg for pulse pressure; for AKI, the estimated change-point (95% CI) was 87 (85 to 90) for systolic, 60 (57 to 72) for mean, 50 (43 to 65) for diastolic, and 36 (35 to 39) mmHg for pulse pressure. All thresholds were statistically significant with P < 0.001, indicating an increase in slope using the values higher than the threshold versus those below the threshold. The statistically-determined thresholds were similar for myocardial and kidney injury and agreed well with the visual change-points described above. We therefore chose to use the same thresholds for myocardial and kidney injury, specifically 90 mmHg for systolic, 65 mmHg for mean, 50 mmHg for diastolic, and 35 mmHg for pulse pressure.

Relationships between Exposures below Thresholds and Outcomes.

The results of the multivariable piecewise logistic regression (two separate line segments based on a given threshold) showed that the slopes for blood pressure values decreasing below the identified change-points were associated with increasing odds of myocardial and kidney injury; the exception was that the slope of the lowest diastolic blood pressure less than 50 mmHg was not significantly associated with AKI after Bonferroni correction (P = 0.028), Supplemental Digital Content 3, table S2 (http://links.lww.com/ALN/C95), and Supplemental Digital Content 4, table S3 (http://links.lww.com/ALN/C96).

Direct comparisons among standardized blood pressure components on the confounder-adjusted relationship between lowest value for a cumulative 5 min and outcome showed a visually stronger relationship for systolic, but not much weaker for mean and pulse pressure, as compared to diastolic for both myocardial and kidney injury, figure 4, A and B. For all relationships, the estimated probability of either outcome increased steeply at blood pressure values 1 and 2 standard deviations below the observed mean.

Fig. 4.

Multivariable association between standardized lowest blood pressure components for a cumulative 5 min and each of myocardial and kidney injury. Each blood pressure component was standardized (subtract mean, divide by SD) to have normal distribution with mean = 0 and SD = 1. Therefore, 0 on the x axis corresponds to the mean of each variable (i.e., the mean across patients of the lowest blood pressure for a cumulative 5 min). Values of −1 and −2 correspond to blood pressure values 1 and 2 SD below the mean, respectively. The observed mean ± SD was 89 ± 13 for systolic blood pressure (SBP), 61 ± 9 for mean arterial pressure (MAP), 45 ± 8 for diastolic blood pressure (DBP), and 36 ± 12 mmHg for pulse pressure (PP). For all relationships, estimated probability of outcome decreases steadily and then flattens as lowest patient blood pressure increases. AKI, acute kidney injury; MINS, myocardial injury after noncardiac surgery.

Fig. 4.

Multivariable association between standardized lowest blood pressure components for a cumulative 5 min and each of myocardial and kidney injury. Each blood pressure component was standardized (subtract mean, divide by SD) to have normal distribution with mean = 0 and SD = 1. Therefore, 0 on the x axis corresponds to the mean of each variable (i.e., the mean across patients of the lowest blood pressure for a cumulative 5 min). Values of −1 and −2 correspond to blood pressure values 1 and 2 SD below the mean, respectively. The observed mean ± SD was 89 ± 13 for systolic blood pressure (SBP), 61 ± 9 for mean arterial pressure (MAP), 45 ± 8 for diastolic blood pressure (DBP), and 36 ± 12 mmHg for pulse pressure (PP). For all relationships, estimated probability of outcome decreases steadily and then flattens as lowest patient blood pressure increases. AKI, acute kidney injury; MINS, myocardial injury after noncardiac surgery.

In general, the relationships between lowest intraoperative component pressures and either myocardial or kidney injury did not depend on their baseline pressures. However, there was some evidence of interaction between each of lowest intraoperative mean pressures and lowest intraoperative diastolic pressures with their baseline pressures on myocardial injury (P < 0.05; Supplemental Digital Content 5, table S4, http://links.lww.com/ALN/C97). Graphical displays of pressure-outcome relationships for various baseline pressure ranges showed parallel lines (no apparent interaction, not shown) below the change-point thresholds, and some differences in slopes above the thresholds. We therefore conclude that the interactions between baseline pressures and intraoperative lowest cumulative 5-min component pressures are not clinically important in terms of hypotension.

Discriminative Ability

Exposures of interest based on the four blood pressure components had weak discriminative ability for myocardial and kidney injury, with univariable C-statistics between 0.55 and 0.59, (Supplemental Digital Content 6, table S5A http://links.lww.com/ALN/C98), with no clinically important differences among them. In any case, for myocardial injury, the area under systolic less than 90 mmHg (P < 0.001) and mean pressure less than 65 mmHg (P = 0.0019) had higher C-statistic than pulse pressure. For AKI, lowest MAP for 5 min and MAP area under 65 mmHg had better discriminative ability than lowest diastolic for 5 min and area of diastolic less than 50 mmHg (both P < 0.001). None of the other comparisons was statistically significant, Supplemental Digital Content 7, table S5B (http://links.lww.com/ALN/C99).

Adding lowest blood pressure component sustained for a cumulative 5 min into the multivariable logistic regression models containing all baseline variable did not increase the discriminative ability for myocardial or kidney injury, (Supplemental Digital Content 6, table S5A http://links.lww.com/ALN/C98).

In table 2 we report the Wilcoxon–Mann–Whitney odds as an intuitive way of interpreting the C-statistic results. It describes the odds of a patient with the outcome (i.e., myocardial or kidney injury) having a longer blood pressure component exposure compared to one without the outcome. Just as the C-statistics were all significantly greater than 0.50, the Wilcoxon–Mann–Whitney odds all significantly exceeded 1.

Table 2.

Univariable C-Statistics (95% CI) and Wilcoxon–Mann–Whitney Odds (95% CI) Relationship between Exposures and Outcomes

Univariable C-Statistics (95% CI) and Wilcoxon–Mann–Whitney Odds (95% CI) Relationship between Exposures and Outcomes
Univariable C-Statistics (95% CI) and Wilcoxon–Mann–Whitney Odds (95% CI) Relationship between Exposures and Outcomes

Relationship between Exposure Categories and Outcomes

Exposure categories for area under the curve and minutes under the given thresholds were significantly associated with myocardial and kidney injury overall for mean, systolic, and pulse pressures, but not for diastolic.

For myocardial injury, when hypotension exposure was characterized by area under the identified threshold, the fourth quartile of minutes under the threshold of systolic, the third and fourth quartiles for mean, and the second, third and fourth quartile of pulse pressure were associated with poor outcome compared to patients who never had pressures below the threshold, figure 5. For minutes under the threshold, the fourth quartile of systolic, the third and fourth quartiles of mean, and second through fourth of pulse pressure had increased odds of having a myocardial injury compared to those patients who had never values of the blood pressure component below the respective threshold, Supplemental Digital Content 8, figure S2 (http://links.lww.com/ALN/C139).

Fig. 5.

Multivariable associations between myocardial injury and area under curve (AUC) under each blood pressure component threshold. Multivariable logistic regression model adjusting for covariates listed in table 1. Bonferroni correction was used to adjust for four comparisons to the reference group within each exposure of interest so that P < 0.0125 (0.05/4) was considered statistically significant. DBP, diastolic blood pressure; MAP, mean arterial pressure; MINS, myocardial injury after noncardiac injury; OR, odds ratio; PP, pulse pressure; SBP, systolic blood pressure.

Fig. 5.

Multivariable associations between myocardial injury and area under curve (AUC) under each blood pressure component threshold. Multivariable logistic regression model adjusting for covariates listed in table 1. Bonferroni correction was used to adjust for four comparisons to the reference group within each exposure of interest so that P < 0.0125 (0.05/4) was considered statistically significant. DBP, diastolic blood pressure; MAP, mean arterial pressure; MINS, myocardial injury after noncardiac injury; OR, odds ratio; PP, pulse pressure; SBP, systolic blood pressure.

For AKI, the odds of outcome were significantly higher only for the fourth quartile of minutes and areas under curve under the threshold for systolic, mean and pulse pressure compared to patients who never went below the threshold, figure 6 and Supplemental Digital Content 9, figure S3 (http://links.lww.com/ALN/C140).

Fig. 6.

Multivariable associations between acute kidney injury (AKI) and area under curve (AUC) under each blood pressure component threshold. Multivariable logistic regression model adjusting for covariates listed in table 1. Bonferroni correction was used to adjust for four comparisons to the reference group within each exposure of interest so that P < 0.0125 (0.05/4) was considered statistically significant. DBP, diastolic blood pressure; MAP, mean arterial pressure; OR, odds ratio; PP, pulse pressure; SBP, systolic blood pressure.

Fig. 6.

Multivariable associations between acute kidney injury (AKI) and area under curve (AUC) under each blood pressure component threshold. Multivariable logistic regression model adjusting for covariates listed in table 1. Bonferroni correction was used to adjust for four comparisons to the reference group within each exposure of interest so that P < 0.0125 (0.05/4) was considered statistically significant. DBP, diastolic blood pressure; MAP, mean arterial pressure; OR, odds ratio; PP, pulse pressure; SBP, systolic blood pressure.

Sensitivity Analyses

In the first post hoc sensitivity analysis, we assessed whether the relative discriminative ability of the components depended on age group (younger than 65 vs. 65 yr or older) or American Society of Anesthesiologists Physical Status. We found that the pattern of C-statistics among components for the area under the estimated threshold exposures was similar within these categories versus for the overall dataset, with diastolic pressures generally being the weakest discriminator (Supplemental Digital Content 10, table S6 http://links.lww.com/ALN/C100 and Supplemental Digital Content 11, table S7 http://links.lww.com/ALN/C101). In a second sensitivity analysis, we assessed the impact of including quadratic terms for continuous confounding variables in our multivariable models, and found very little difference when compared to models including only linear terms for these variables (Supplemental Digital Content 12, table S8 http://links.lww.com/ALN/C102 and Supplemental Digital Content 13, table S9 http://links.lww.com/ALN/C103).

Discussion

For each blood pressure component, we report significant and clinically meaningful associations between the lowest pressure sustained for 5 min and myocardial and kidney injury. Absolute population risk thresholds were similar for myocardial and kidney injury, being roughly 90 mmHg for systolic, 65 mmHg for mean, 50 mmHg for diastolic, and 35 mmHg for pulse pressures. The odds for myocardial and kidney injury was inversely related to the lowest pressures maintained for at least 5 min below each threshold, even after adjusting for potential baseline confounding factors. In contrast, there was little relationship between lowest pressures and myocardial and renal injury above the threshold for each component. We caution, though, that this result does not imply that higher pressures are safe since our analysis was restricted to the lowest values in each patient. To assess hypertensive risk, a similar analysis based on highest pressure would be required.

Our results are broadly consistent with a previous report that mean pressure less than 55 mmHg was independently associated with myocardial and kidney injury.3  The slight difference in harm thresholds may be partially related to differences in invasive and non-invasive measurements. But more importantly, the patients in this analysis—all of whom had arterial catheters—were generally sicker. Another study reported an increased risk of perioperative morbidity and mortality increases when intraoperative pressures are less than 70 mmHg for systolic, 50 mmHg for mean, and 30 mmHg for diastolic.15 

Each pressure component can be further considered in terms of its absolute level. Thus, while a given component has instantaneous values over one or several cardiac cycles, each value also varies over minutes and hours. Intraoperative levels of any blood pressure component can therefore be characterized by various curve descriptors such as mean, time-weighted average, maximum, minimum, time or area below or above thresholds. All have been used in recent analyses30  and there is no firm consensus that one is preferable to others, with optimal approaches probably being context dependent. However, it appears that myocardial and kidney injury, along with death, is more strongly associated with extreme values—especially hypotensive excursions—than overall mean values. That is, minimum values and time or area below thresholds better predicts organ injury than simple time-weighted averages. As in previous studies,2,13  we therefore considered the lowest component pressure maintained for a cumulative, but not necessarily consecutive, 5 min to be our primary exposure. Secondarily, we also considered area and time below each threshold.

Variability over time is an additional consideration. Variability refers to blood pressure dispersion around an average level, usually over a period of minutes. Variability is natural and in many biologic signals, such as heart rate, is a sign of health.31  It therefore seems unlikely that blood pressure variability per se is harmful. The challenge is that dispersion below the average level is almost surely far more harmful than values above the average. How harmful variability might be therefore depends critically on the average level from which it departs. Harm associated with variability therefore cannot reasonably be considered without fully adjusting for level, and considering whether harm apparently consequent to variability actually results simply from cumulative hypotensive exposure.13 

Using the cut-off points obtained in our initial analysis, we modeled the confounder-adjusted association between exposure (both minutes and area) under each component’s threshold and outcomes. When magnitude, as well as duration, were considered (area under each threshold), the highest quartile in each category except diastolic was associated with significantly increased odds of myocardial and kidney injury. Myocardial and kidney injury were most common when systolic and mean hypotension was well below the derived thresholds and sustained for 15 min or longer. However, about 20% of included patients had such exposures which is hardly a trivial fraction. Our findings are consistent with a recent analysis which concluded that the risk of AKI was greatest when mean pressures were below 60 mmHg for at least 20 min.5 

To evaluate which blood pressure component was most strongly associated with organ injury, we evaluated the discriminative ability of each. Discrimination was similar for each, and weak (e.g., C-statistics 0.55 to 0.59), indicating that baseline risk is far more important than intraoperative pressure. Furthermore, none of the blood pressure components added much to a multivariable model that included baseline risk factors. Pressure nonetheless remains interesting because it is modifiable and thus amenable to intervention. Systolic and mean pressures were slightly better discriminators of outcome than the other components, but only by small margins. Arterial mean pressures are defined by time-averaged instantaneous pressures during a cardiac cycle. Mean and systolic pressures are usually correlated so it is not especially surprising that each is similarly associated with organ injury.

We evaluated only two organs, both of which have similar population harm thresholds. Presumably the harm thresholds differ for other organs. For example, strokes are of considerable interest, but too rare to easily evaluate; they are also generally poorly coded in administrative datasets. It is also likely that intestines are also sensitive to hypotension, but there are currently neither good biomarkers nor strong clinical correlates for intestinal ischemia. Myocardial injury was defined by cardiac enzyme elevation, but only a fraction of our surgical patients had troponin monitored routinely rather than for cause. We therefore surely underestimated the actual incidence of myocardial injury, perhaps by a factor-of-three.26  We excluded patients who had preoperative abnormal creatinine; hence, our results cannot be generalized to patients who had chronic kidney disease or those undergoing urologic procedures.

We considered intraoperative vasopressor use to be a potential mediating variable,32  as they would presumably be given in response to hypotension and may also be associated with outcome; hence, we did not consider vasopressor use a confounder and did not adjust for vasopressor use in our analysis. Most artifacts in electronically obtained arterial pressure data were removed by algorithms. There are, however, situations in which an arterial catheter is not zeroed properly or is dampened which we would not have detected. A final limitation is that our analysis was restricted to a single center which may reduce generalizability.

A strength of our study is that we restricted our analysis to patients in whom blood pressure was measured from a radial arterial catheter. Direct pressure measurements are relatively accurate, whereas oscillometric systems estimate systolic and diastolic pressures from a measured mean value.24,25  But that said, radial arterial pressure can differ substantially from aortic pressure.33 

In summary, absolute population risk thresholds were similar for myocardial and kidney injury on the lowest intraoperative pressure maintained for at least 5 min. The thresholds were roughly 90 mmHg for systolic, 65 mmHg for mean, 50 mmHg for diastolic, and 35 mmHg for pulse pressures. Among the components, systolic and mean pressures were most predictive, but only by small margins. Baseline risk is a far better predictor of myocardial and renal injury than intraoperative blood pressure, but intraoperative pressure differs in being modifiable and thus subject to intervention.

Research Support

Support was provided solely from institutional and/or departmental sources.

Competing Interests

Drs. Sessler and Maheshwari are consultants to Edwards Lifesciences (Irvine, California). Dr. Khanna consults for Edwards Lifesciences and Retia Medical (Valhalla, New York). The other authors declare no competing interests.

References

References
1.
Convertino
VA
,
Cooke
WH
,
Holcomb
JB
: .
Arterial pulse pressure and its association with reduced stroke volume during progressive central hypovolemia.
J Trauma
.
2006
;
61
:
629
34
2.
Salmasi
V
,
Maheshwari
K
,
Yang
D
,
Mascha
EJ
,
Singh
A
,
Sessler
DI
,
Kurz
A
: .
Relationship between intraoperative hypotension, defined by either reduction from baseline or absolute thresholds, and acute kidney and myocardial injury after noncardiac surgery: A retrospective cohort analysis.
Anesthesiology
.
2017
;
126
:
47
65
3.
Walsh
M
,
Devereaux
PJ
,
Garg
AX
,
Kurz
A
,
Turan
A
,
Rodseth
RN
,
Cywinski
J
,
Thabane
L
,
Sessler
DI
: .
Relationship between intraoperative mean arterial pressure and clinical outcomes after noncardiac surgery: Toward an empirical definition of hypotension.
Anesthesiology
.
2013
;
119
:
507
15
4.
Sun
LY
,
Wijeysundera
DN
,
Tait
GA
,
Beattie
WS
: .
Association of intraoperative hypotension with acute kidney injury after elective noncardiac surgery.
Anesthesiology
.
2015
;
123
:
515
23
5.
Abbott
TEF
,
Pearse
RM
,
Archbold
RA
,
Ahmad
T
,
Niebrzegowska
E
,
Wragg
A
,
Rodseth
RN
,
Devereaux
PJ
,
Ackland
GL
: .
A prospective international multicentre cohort study of intraoperative heart rate and systolic blood pressure and myocardial injury after noncardiac surgery: Results of the VISION Study.
Anesth Analg
.
2018
;
126
:
1936
45
6.
van Waes
JA
,
van Klei
WA
,
Wijeysundera
DN
,
van Wolfswinkel
L
,
Lindsay
TF
,
Beattie
WS
: .
Association between intraoperative hypotension and myocardial injury after vascular surgery.
Anesthesiology
.
2016
;
124
:
35
44
7.
Sessler
DI
,
Meyhoff
CS
,
Zimmerman
NM
,
Mao
G
,
Leslie
K
,
Vásquez
SM
,
Balaji
P
,
Alvarez-Garcia
J
,
Cavalcanti
AB
,
Parlow
JL
,
Rahate
PV
,
Seeberger
MD
,
Gossetti
B
,
Walker
SA
,
Premchand
RK
,
Dahl
RM
,
Duceppe
E
,
Rodseth
R
,
Botto
F
,
Devereaux
PJ
: .
Period-dependent associations between hypotension during and for four days after noncardiac surgery and a composite of myocardial infarction and death: A substudy of the POISE-2 Trial.
Anesthesiology
.
2018
;
128
:
317
27
8.
McEvoy
JW
,
Chen
Y
,
Rawlings
A
,
Hoogeveen
RC
,
Ballantyne
CM
,
Blumenthal
RS
,
Coresh
J
,
Selvin
E
: .
Diastolic blood pressure, subclinical myocardial damage, and cardiac events: Implications for blood pressure control.
J Am Coll Cardiol
.
2016
;
68
:
1713
22
9.
Beddhu
S
,
Chertow
GM
,
Cheung
AK
,
Cushman
WC
,
Rahman
M
,
Greene
T
,
Wei
G
,
Campbell
RC
,
Conroy
M
,
Freedman
BI
,
Haley
W
,
Horwitz
E
,
Kitzman
D
,
Lash
J
,
Papademetriou
V
,
Pisoni
R
,
Riessen
E
,
Rosendorff
C
,
Watnick
SG
,
Whittle
J
,
Whelton
PK
;
SPRINT Research Group
: .
Influence of baseline diastolic blood pressure on effects of intensive compared with standard blood pressure control.
Circulation
.
2018
;
137
:
134
43
10.
Abbott
TEF
,
Pearse
RM
,
Archbold
RA
,
Wragg
A
,
Kam
E
,
Ahmad
T
,
Khan
AW
,
Niebrzegowska
E
,
Rodseth
RN
,
Devereaux
PJ
,
Ackland
GL
: .
Association between preoperative pulse pressure and perioperative myocardial injury: An international observational cohort study of patients undergoing non-cardiac surgery.
Br J Anaesth
.
2017
;
119
:
78
86
11.
Vaccarino
V
,
Holford
TR
,
Krumholz
HM
: .
Pulse pressure and risk for myocardial infarction and heart failure in the elderly.
J Am Coll Cardiol
.
2000
;
36
:
130
8
12.
Maheshwari
K
,
Nathanson
BH
,
Munson
SH
,
Khangulov
V
,
Stevens
M
,
Badani
H
,
Khanna
AK
,
Sessler
DI
: .
The relationship between ICU hypotension and in-hospital mortality and morbidity in septic patients.
Intensive Care Med
.
2018
;
44
:
857
67
13.
Mascha
EJ
,
Yang
D
,
Weiss
S
,
Sessler
DI
: .
Intraoperative mean arterial pressure variability and 30-day mortality in patients having noncardiac surgery.
Anesthesiology
.
2015
;
123
:
79
91
14.
Ashish Khanna
GM
,
Liu
L
,
Yang
D
,
Perez-Protto
S
,
Chodavarapu
P
,
Schacham
Y
,
Mascha
E
,
Sessler
D
: .
Hypotension increases acute kidney injury, myocardial injury and mortality in surgical critical care.
Critical Care Medicine
.
2018
;
46
:
71
15.
Monk
TG
,
Bronsert
MR
,
Henderson
WG
,
Mangione
MP
,
Sum-Ping
ST
,
Bentt
DR
,
Nguyen
JD
,
Richman
JS
,
Meguid
RA
,
Hammermeister
KE
: .
Association between intraoperative hypotension and hypertension and 30-day postoperative mortality in noncardiac surgery.
Anesthesiology
.
2015
;
123
:
307
19
16.
Sun
LY
,
Chung
AM
,
Farkouh
ME
,
van Diepen
S
,
Weinberger
J
,
Bourke
M
,
Ruel
M
: .
Defining an intraoperative hypotension threshold in association with stroke in cardiac surgery.
Anesthesiology
.
2018
;
129
:
440
7
17.
Rettig
TCD
,
Peelen
LM
,
Geuzebroek
GSC
,
van Klei
WA
,
Boer
C
,
van der Veer
JW
,
Hofland
J
,
van de Garde
EMW
,
Noordzij
PG
: .
Impact of intraoperative hypotension during cardiopulmonary bypass on acute kidney injury after coronary artery bypass grafting.
J Cardiothorac Vasc Anesth
.
2017
;
31
:
522
8
18.
Mizota
T
,
Hamada
M
,
Matsukawa
S
,
Seo
H
,
Tanaka
T
,
Segawa
H
: .
Relationship between intraoperative hypotension and acute kidney injury after living donor liver transplantation: A retrospective analysis.
J Cardiothorac Vasc Anesth
.
2017
;
31
:
582
9
19.
Schmid
S
,
Kapfer
B
,
Heim
M
,
Bogdanski
R
,
Anetsberger
A
,
Blobner
M
,
Jungwirth
B
: .
Algorithm-guided goal-directed haemodynamic therapy does not improve renal function after major abdominal surgery compared to good standard clinical care: A prospective randomised trial.
Crit Care
.
2016
;
20
:
50
20.
Albrecht
RF
2nd
,
Lanier
WL
: .
Potentiation of succinylcholine-induced phase II block by vancomycin.
Anesth Analg
.
1993
;
77
:
1300
2
21.
Poukkanen
M
,
Wilkman
E
,
Vaara
ST
,
Pettilä
V
,
Kaukonen
KM
,
Korhonen
AM
,
Uusaro
A
,
Hovilehto
S
,
Inkinen
O
,
Laru-Sompa
R
,
Hautamäki
R
,
Kuitunen
A
,
Karlsson
S
;
FINNAKI Study Group
: .
Hemodynamic variables and progression of acute kidney injury in critically ill patients with severe sepsis: Data from the prospective observational FINNAKI study.
Crit Care
.
2013
;
17
:
R295
22.
Badin
J
,
Boulain
T
,
Ehrmann
S
,
Skarzynski
M
,
Bretagnol
A
,
Buret
J
,
Benzekri-Lefevre
D
,
Mercier
E
,
Runge
I
,
Garot
D
,
Mathonnet
A
,
Dequin
PF
,
Perrotin
D
: .
Relation between mean arterial pressure and renal function in the early phase of shock: A prospective, explorative cohort study.
Crit Care
.
2011
;
15
:
R135
23.
Wu
X
,
Jiang
Z
,
Ying
J
,
Han
Y
,
Chen
Z
: .
Optimal blood pressure decreases acute kidney injury after gastrointestinal surgery in elderly hypertensive patients: A randomized study: Optimal blood pressure reduces acute kidney injury.
J Clin Anesth
.
2017
;
43
:
77
83
24.
Ribezzo
S
,
Spina
E
,
Di Bartolomeo
S
,
Sanson
G
: .
Noninvasive techniques for blood pressure measurement are not a reliable alternative to direct measurement: A randomized crossover trial in ICU.
ScientificWorldJournal
.
2014
;
2014
:
353628
25.
Babbs
CF
: .
Oscillometric measurement of systolic and diastolic blood pressures validated in a physiologic mathematical model.
Biomed Eng Online
.
2012
;
11
:
56
26.
Botto
F
,
Alonso-Coello
P
,
Chan
MT
,
Villar
JC
,
Xavier
D
,
Srinathan
S
,
Guyatt
G
,
Cruz
P
,
Graham
M
,
Wang
CY
,
Berwanger
O
,
Pearse
RM
,
Biccard
BM
,
Abraham
V
,
Malaga
G
,
Hillis
GS
,
Rodseth
RN
,
Cook
D
,
Polanczyk
CA
,
Szczeklik
W
,
Sessler
DI
,
Sheth
T
,
Ackland
GL
,
Leuwer
M
,
Garg
AX
,
Lemanach
Y
,
Pettit
S
,
Heels-Ansdell
D
,
Luratibuse
G
,
Walsh
M
,
Sapsford
R
,
Schünemann
HJ
,
Kurz
A
,
Thomas
S
,
Mrkobrada
M
,
Thabane
L
,
Gerstein
H
,
Paniagua
P
,
Nagele
P
,
Raina
P
,
Yusuf
S
,
Devereaux
PJ
,
Devereaux
PJ
,
Sessler
DI
,
Walsh
M
,
Guyatt
G
,
McQueen
MJ
,
Bhandari
M
,
Cook
D
,
Bosch
J
,
Buckley
N
,
Yusuf
S
,
Chow
CK
,
Hillis
GS
,
Halliwell
R
,
Li
S
,
Lee
VW
,
Mooney
J
,
Polanczyk
CA
,
Furtado
MV
,
Berwanger
O
,
Suzumura
E
,
Santucci
E
,
Leite
K
,
Santo
JA
,
Jardim
CA
,
Cavalcanti
AB
,
Guimaraes
HP
,
Jacka
MJ
,
Graham
M
,
McAlister
F
,
McMurtry
S
,
Townsend
D
,
Pannu
N
,
Bagshaw
S
,
Bessissow
A
,
Bhandari
M
,
Duceppe
E
,
Eikelboom
J
,
Ganame
J
,
Hankinson
J
,
Hill
S
,
Jolly
S
,
Lamy
A
,
Ling
E
,
Magloire
P
,
Pare
G
,
Reddy
D
,
Szalay
D
,
Tittley
J
,
Weitz
J
,
Whitlock
R
,
Darvish-Kazim
S
,
Debeer
J
,
Kavsak
P
,
Kearon
C
,
Mizera
R
,
O’Donnell
M
,
McQueen
M
,
Pinthus
J
,
Ribas
S
,
Simunovic
M
,
Tandon
V
,
Vanhelder
T
,
Winemaker
M
,
Gerstein
H
,
McDonald
S
,
O’Bryne
P
,
Patel
A
,
Paul
J
,
Punthakee
Z
,
Raymer
K
,
Salehian
O
,
Spencer
F
,
Walter
S
,
Worster
A
,
Adili
A
,
Clase
C
,
Cook
D
,
Crowther
M
,
Douketis
J
,
Gangji
A
,
Jackson
P
,
Lim
W
,
Lovrics
P
,
Mazzadi
S
,
Orovan
W
,
Rudkowski
J
,
Soth
M
,
Tiboni
M
,
Acedillo
R
,
Garg
A
,
Hildebrand
A
,
Lam
N
,
Macneil
D
,
Mrkobrada
M
,
Roshanov
PS
,
Srinathan
SK
,
Ramsey
C
,
John
PS
,
Thorlacius
L
,
Siddiqui
FS
,
Grocott
HP
,
McKay
A
,
Lee
TW
,
Amadeo
R
,
Funk
D
,
McDonald
H
,
Zacharias
J
,
Villar
JC
,
Cortés
OL
,
Chaparro
MS
,
Vásquez
S
,
Castañeda
A
,
Ferreira
S
,
Coriat
P
,
Monneret
D
,
Goarin
JP
,
Esteve
CI
,
Royer
C
,
Daas
G
,
Chan
MT
,
Choi
GY
,
Gin
T
,
Lit
LC
,
Xavier
D
,
Sigamani
A
,
Faruqui
A
,
Dhanpal
R
,
Almeida
S
,
Cherian
J
,
Furruqh
S
,
Abraham
V
,
Afzal
L
,
George
P
,
Mala
S
,
Schünemann
H
,
Muti
P
,
Vizza
E
,
Wang
CY
,
Ong
GS
,
Mansor
M
,
Tan
AS
,
Shariffuddin
II
,
Vasanthan
V
,
Hashim
NH
,
Undok
AW
,
Ki
U
,
Lai
HY
,
Ahmad
WA
,
Razack
AH
,
Malaga
G
,
Valderrama-Victoria
V
,
Loza-Herrera
JD
,
De Los Angeles Lazo
M
,
Rotta-Rotta
A
,
Szczeklik
W
,
Sokolowska
B
,
Musial
J
,
Gorka
J
,
Iwaszczuk
P
,
Kozka
M
,
Chwala
M
,
Raczek
M
,
Mrowiecki
T
,
Kaczmarek
B
,
Biccard
B
,
Cassimjee
H
,
Gopalan
D
,
Kisten
T
,
Mugabi
A
,
Naidoo
P
,
Naidoo
R
,
Rodseth
R
,
Skinner
D
,
Torborg
A
,
Paniagua
P
,
Urrutia
G
,
Maestre
ML
,
Santaló
M
,
Gonzalez
R
,
Font
A
,
Martínez
C
,
Pelaez
X
,
De Antonio
M
,
Villamor
JM
,
García
JA
,
Ferré
MJ
,
Popova
E
,
Alonso-Coello
P
,
Garutti
I
,
Cruz
P
,
Fernández
C
,
Palencia
M
,
Díaz
S
,
Del Castillo
T
,
Varela
A
,
de Miguel
A
,
Muñoz
M
,
Piñeiro
P
,
Cusati
G
,
Del Barrio
M
,
Membrillo
MJ
,
Orozco
D
,
Reyes
F
,
Sapsford
RJ
,
Barth
J
,
Scott
J
,
Hall
A
,
Howell
S
,
Lobley
M
,
Woods
J
,
Howard
S
,
Fletcher
J
,
Dewhirst
N
,
Williams
C
,
Rushton
A
,
Welters
I
,
Leuwer
M
,
Pearse
R
,
Ackland
G
,
Khan
A
,
Niebrzegowska
E
,
Benton
S
,
Wragg
A
,
Archbold
A
,
Smith
A
,
McAlees
E
,
Ramballi
C
,
Macdonald
N
,
Januszewska
M
,
Stephens
R
,
Reyes
A
,
Paredes
LG
,
Sultan
P
,
Cain
D
,
Whittle
J
,
Del Arroyo
AG
,
Sessler
DI
,
Kurz
A
,
Sun
Z
,
Finnegan
PS
,
Egan
C
,
Honar
H
,
Shahinyan
A
,
Panjasawatwong
K
,
Fu
AY
,
Wang
S
,
Reineks
E
,
Nagele
P
,
Blood
J
,
Kalin
M
,
Gibson
D
,
Wildes
T
;
Vascular events In noncardiac Surgery patIents cOhort evaluatioN (VISION) Writing Group, on behalf of The Vascular events In noncardiac Surgery patIents cOhort evaluatioN (VISION) Investigators; Appendix 1. The Vascular events In noncardiac Surgery patIents cOhort evaluatioN (VISION) Study Investigators Writing Group; Appendix 2. The Vascular events In noncardiac Surgery patIents cOhort evaluatioN Operations Committee; Vascular events In noncardiac Surgery patIents cOhort evaluatioN VISION Study Investigators
: .
Myocardial injury after noncardiac surgery: A large, international, prospective cohort study establishing diagnostic criteria, characteristics, predictors, and 30-day outcomes.
Anesthesiology
.
2014
;
120
:
564
78
27.
Thygesen
K
,
Alpert
JS
,
Jaffe
AS
,
Simoons
ML
,
Chaitman
BR
,
White
HD
,
Thygesen
K
,
Alpert
JS
,
White
HD
,
Jaffe
AS
,
Katus
HA
,
Apple
FS
,
Lindahl
B
,
Morrow
DA
,
Chaitman
BR
,
Clemmensen
PM
,
Johanson
P
,
Hod
H
,
Underwood
R
,
Bax
JJ
,
Bonow
JJ
,
Pinto
F
,
Gibbons
RJ
,
Fox
KA
,
Atar
D
,
Newby
LK
,
Galvani
M
,
Hamm
CW
,
Uretsky
BF
,
Steg
PG
,
Wijns
W
,
Bassand
JP
,
Menasche
P
,
Ravkilde
J
,
Ohman
EM
,
Antman
EM
,
Wallentin
LC
,
Armstrong
PW
,
Simoons
ML
,
Januzzi
JL
,
Nieminen
MS
,
Gheorghiade
M
,
Filippatos
G
,
Luepker
RV
,
Fortmann
SP
,
Rosamond
WD
,
Levy
D
,
Wood
D
,
Smith
SC
,
Hu
D
,
Lopez-Sendon
JL
,
Robertson
RM
,
Weaver
D
,
Tendera
M
,
Bove
AA
,
Parkhomenko
AN
,
Vasilieva
EJ
,
Mendis
S
,
Bax
JJ
,
Baumgartner
H
,
Ceconi
C
,
Dean
V
,
Deaton
C
,
Fagard
R
,
Funck-Brentano
C
,
Hasdai
D
,
Hoes
A
,
Kirchhof
P
,
Knuuti
J
,
Kolh
P
,
McDonagh
T
,
Moulin
C
,
Popescu
BA
,
Reiner
Z
,
Sechtem
U
,
Sirnes
PA
,
Tendera
M
,
Torbicki
A
,
Vahanian
A
,
Windecker
S
,
Morais
J
,
Aguiar
C
,
Almahmeed
W
,
Arnar
DO
,
Barili
F
,
Bloch
KD
,
Bolger
AF
,
Botker
HE
,
Bozkurt
B
,
Bugiardini
R
,
Cannon
C
,
de Lemos
J
,
Eberli
FR
,
Escobar
E
,
Hlatky
M
,
James
S
,
Kern
KB
,
Moliterno
DJ
,
Mueller
C
,
Neskovic
AN
,
Pieske
BM
,
Schulman
SP
,
Storey
RF
,
Taubert
KA
,
Vranckx
P
,
Wagner
DR
;
Joint ESC/ACCF/AHA/WHF Task Force for Universal Definition of Myocardial Infarction; Authors/Task Force Members Chairpersons; Biomarker Subcommittee; ECG Subcommittee; Imaging Subcommittee; Classification Subcommittee; Intervention Subcommittee; Trials & Registries Subcommittee; Trials & Registries Subcommittee; Trials & Registries Subcommittee; Trials & Registries Subcommittee; ESC Committee for Practice Guidelines (CPG); Document Reviewers
: .
Third universal definition of myocardial infarction.
J Am Coll Cardiol
.
2012
;
60
:
1581
98
28.
Mehta
RL
,
Kellum
JA
,
Shah
SV
,
Molitoris
BA
,
Ronco
C
,
Warnock
DG
,
Levin
A
;
Acute Kidney Injury Network
: .
Acute Kidney Injury Network: Report of an initiative to improve outcomes in acute kidney injury.
Crit Care
.
2007
;
11
:
R31
29.
Fong
Y
,
Di
C
,
Huang
Y
,
Gilbert
PB
: .
Model-robust inference for continuous threshold regression models.
Biometrics
.
2017
;
73
:
452
62
30.
Vernooij
LM
,
van Klei
WA
,
Machina
M
,
Pasma
W
,
Beattie
WS
,
Peelen
LM
: .
Different methods of modelling intraoperative hypotension and their association with postoperative complications in patients undergoing non-cardiac surgery.
Br J Anaesth
.
2018
;
120
:
1080
9
31.
Shaffer
F
,
Ginsberg
JP
: .
An overview of heart rate variability metrics and norms.
Front Public Health
.
2017
;
5
:
258
32.
Mascha
EJ
,
Dalton
JE
,
Kurz
A
,
Saager
L
: .
Statistical grand rounds: Understanding the mechanism: Mediation analysis in randomized and nonrandomized studies.
Anesth Analg
.
2013
;
117
:
980
94
33.
Hynson
JM
,
Sessler
DI
,
Moayeri
A
,
Katz
JA
: .
Thermoregulatory and anesthetic-induced alterations in the differences among femoral, radial, and oscillometric blood pressures.
Anesthesiology
.
1994
;
81
:
1411
21