Secondary Logo

Journal Logo

Original Research Articles: Original Clinical Research Report

Postoperative Hypotension and Adverse Clinical Outcomes in Patients Without Intraoperative Hypotension, After Noncardiac Surgery

Khanna, Ashish K. MD, FCCP, FCCM*,†; Shaw, Andrew D. MB, FRCPC‡,§; Stapelfeldt, Wolf H. MD; Boero, Isabel J. MD, MS; Chen, Qinyu MS; Stevens, Mitali PharmD, BCPS#; Gregory, Anne MD, MSc, FRCPC; Smischney, Nathan J. MD, MSc**

Author Information
doi: 10.1213/ANE.0000000000005374

Abstract

KEY POINTS

  • Question: What is the association between postoperative hypotension (POH) and clinical outcomes in patients admitted to the ward after noncardiac surgery without evidence of intraoperative mean arterial pressure (MAP) ≤65 mm Hg?
  • Findings: POH in patients without intraoperative hypotension (IOH) was not associated with major adverse cardiac or cerebrovascular events (MACCE) and no interaction was detected between POH and IOH for any of the outcomes investigated.
  • Meaning: In the absence of IOH, POH is not associated with MACCE in patients managed on the ward following surgery.

Major adverse cardiac or cerebrovascular events (MACCE), myocardial injury after noncardiac surgery (MINS), and acute kidney injury (AKI) are associated with substantial morbidity/mortality after noncardiac surgery.1–3 Risk factors are mostly baseline patient characteristics or surgical factors that are largely nonmodifiable.4,5 However, hypotension is 1 potential risk factor that is common during/after noncardiac surgery which could be optimized.6–9 Even short durations of intraoperative hypotension (IOH) at a mean arterial pressure (MAP) <70 mm Hg are associated with mortality, AKI, and MINS.6–17 Postoperative hypotension (POH) is also associated with major adverse events (AEs).9 This relationship is especially obvious among critically ill patients, though it seems to be dependent on IOH.16,17 However, there is limited information in the literature on the hazard of POH across different blood pressure (BP) thresholds in a cross-section of patients admitted to the ward postsurgery.9,18 Furthermore, previous studies did not exclude patients with IOH and, therefore, failed to capture the isolated association of POH with patient outcomes across varying BP thresholds.

An important factor on the ward is the limitation of hemodynamic monitoring to intermittent vital sign assessments compared to the operating room or intensive care unit (ICU) where continuous hemodynamic monitoring is standard protocol. Infrequent monitoring on the ward may, therefore, contribute to harm since it often causes a delay in the intervention.9,19

To confidently evaluate whether POH across multiple hemodynamic thresholds is associated with worse outcomes in the absence of IOH, we chose to evaluate a cohort limited to POH. Using a large population-based data set, patients admitted to the ward for at least 48 hours after noncardiac surgery were identified. Our primary outcome was 30-day MACCE while secondary clinical outcomes included all-cause mortality (30- and 90-day), 30-day AMI, 30-day acute ischemic stroke (AIS), 30-day readmissions, 7-day AKI stage II/III, and hospital-free days in the 30 days postsurgery. An additional analysis (cohort #2) included patients with IOH to assess the association of cumulative hypotension exposures with these postoperative outcomes.

METHODS

Data Source

Data were obtained from the Optum (Optum, Eden Prairie, MN) deidentified electronic health records database, which standardizes/integrates records from >2000 hospitals and 7000 clinics. The data are sourced from ambulatory and inpatient settings, and cover diagnosis/procedure codes, clinical observations (ie, vital signs), medications, and laboratory results. This study was determined exempt by the institutional review board in advance, through provision of the statistical analysis plan, by Western Institutional Review Board (Puyallup, WA). This article adheres to applicable Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.

Cohort Selection

Our initial study population was 368,222 noncardiac/non-Caesarean surgical procedures (January 1, 2008–December 31, 2017; based on data availability) with valid patient intraoperative MAP readings (calculated, detailed in exposure section) and ≥1 year of presurgical history and follow-up in the database (Supplemental Digital Content, Figure 1, http://links.lww.com/AA/D312, for initial attrition).

Exclusion criteria were subsequently applied to the 368,222 patient-procedures to obtain the final original cohort of 67,968 patient-procedures, which excluded patients: (1) not managed on the ward for ≥48 hours postsurgery; (2) with more than two 5-hour gaps between MAP readings20 within 48 hours postsurgery; (3) who died within 24 hours postsurgery; (4) who were discharged to the ICU (recorded/identified via the algorithm described below); (5) with multiple conflicting discharge locations (“invalid”); and (6) with IOH exposure (MAP ≤65 mm Hg based on the literature8,15) during qualifying surgery.

The list of noncardiac surgeries was determined using procedures in the Center for Disease Control’s National Healthcare Safety Network Surgical Site Infection monitoring program; applying the International Classification of Diseases (ICD)-9/10, and Healthcare Common Procedure Coding System codes.21,22 For patients without a documented location, we leveraged machine learning approaches to exclude patients discharged to the ICU (n = 19,052; Supplemental Digital Content, Method 1, http://links.lww.com/AA/D312, for algorithm for patients discharged to ward). The final original cohort consisted of patients discharged to the ward with no evidence of IOH (n = 67,968). For patients who had multiple procedures within 30 days, the last surgery was utilized as the index procedure. In cases where qualifying surgeries were >30 days apart, all procedures were included.

We utilized an additional cohort of POH patients with IOH (cohort #2; defined as MAP ≤65 mm Hg; 16,034 patient-procedures [9511 patients with a listed ward care area postsurgery; 6523 identified via algorithm; Supplemental Digital Content, Method 1, http://links.lww.com/AA/D312] as discharged to ward) to examine any association between POH and outcomes among patients with IOH. Finally, given previous literature,16 we conducted an additional analysis by combining the original cohort (without IOH; n = 67,968) and cohort #2 (with IOH; n = 16,034) to evaluate the association of POH exposure on outcomes in the presence or absence of IOH, by utilizing an interaction term between IOH and POH (cohort #3; n = 84,002).

Determining MAP Thresholds and Exposures

Figure 1.
Figure 1.:
Patient attrition diagram. EHR indicates electronic health record; MAP, mean arterial pressure.

MAP was calculated using the formula: ([2×Diastolic BP]+Systolic BP)/3. Artifacts were removed using previously published criteria: an invalid MAP value was defined as (1) systolic BP (SBP) ≥300 or ≤20 mm Hg, (2) SBP ≤ diastolic BP (DBP) +5 mm Hg, or (3) DBP ≤5 or ≥225 mm Hg; abrupt changes were removed, defined by SBP change ≥80 mm Hg within 1 minute in either direction or ≥40 mm Hg within 2 minutes in both directions.8 Due to missing or invalid MAP readings within the 48-hours post-surgery, 0.7% of patient procedures were excluded (Figure 1). We used 3 absolute MAP thresholds (≤75, ≤65, and ≤55 mm Hg) to assess POH, and MAP ≤65 mm Hg for IOH, defined a priori based on the literature.15 We assessed POH exposure as a binary variable (presence/absence; POH for relevant threshold was defined as a single MAP measurement below the threshold) over postoperative 48 hours (beginning from surgical stop time), given the intermittency of monitoring on the ward.9 We also accounted for the timing of binary POH exposure to ensure that the outcome occurred postexposure.

Potential Confounding Variables

Potential confounding variables that influence POH and the outcome of interest were defined a priori based on previously described methods.20 Briefly, we determined patient demographics from the database and identified comorbidities in the year presurgery using ICD-9/10 codes (present-on-admission indicators were not utilized) except for valvular disease and severity, which was identified using physician notes (Supplemental Digital Content, Table 1, http://links.lww.com/AA/D312, for ICD codes). Baseline patient severity was assessed using the Charlson comorbidity index (CCI).23 Procedures in the year presurgery were captured from ICD-9/10 and Current Procedural Terminology codes. Antihypertensive medication use in the year presurgery was assessed from patient records (Supplemental Digital Content, Method 2, http://links.lww.com/AA/D312, for hypertensive drugs).

Evidence of major bleeding (by ICD-9/10), use of antihypertensive and, separately, vasopressor agents 48 hours postsurgery were collected from records and included as time-dependent confounding variables in the analysis. To control for prior hypotension exposure in patients with IOH (n = 16,034), we used time-weighted MAP (area under the MAP threshold divided by surgery duration) to standardize the exposure by operative time.17 Potential confounding variables for primary and secondary outcomes and baseline patient characteristics are shown in Table 1 and Supplemental Digital Content, Tables 2–4, http://links.lww.com/AA/D312.

A sensitivity analysis was performed to measure the magnitude of unobserved or unaccounted confounding effects. The E value, defined as the association between the unobserved confounding variable and each exposure and outcome required to reduce the observed odds ratio for an outcome to 1.0, was calculated for outcomes for our original cohort and cohort #2 (Supplemental Digital Content, Table 5, http://links.lww.com/AA/D312).24

Primary and Secondary Outcomes

The primary outcome was 30-day MACCE,25–27 a composite of 30-day all-cause mortality (captured from Social Security Index), AMI, or AIS, captured using ICD-9/10 codes (identified from previous literature and a ICD-9 to ICD-10 medical code crosswalk).25,28,29 Additionally, the Clinical Classifications Software diagnosis code 100 was used to capture AMI.

The following secondary outcomes were assessed: all-cause mortality (30- and 90-day), 30-day AMI, 30-day AIS, 30-day readmission, 7-day AKI stage II/III, and hospital-free days in the 30 days postsurgery. We limited AKI to a 7-day window as previously defined.30 Thirty-day readmission was defined as hospital admission within 30 days postdischarge from index hospitalization. AKI stage II/III was defined as postoperative creatinine 2 times greater than the most recent preoperative value, an increase in serum creatinine ≥4 mg/dL, or initiation of dialysis therapy.31 Primary and secondary outcomes were not limited to events in the index visit; subsequent visits were also included. If a patient died in the hospital, the number of hospital-free days was counted as 0.

Statistical Analysis

The association of POH with or without IOH, for 3 POH thresholds, was evaluated in our original cohort (without IOH), cohort #2 (with IOH), and a combined cohort (cohort #3) using 2-tailed hypothesis testing. AE rates were evaluated in each POH group; the reference group was defined as all patients who did not experience POH at the specified MAP threshold (ie, for ≤65 mm Hg, this is >65 mm Hg). The associations between MAP exposure and outcomes were controlled for confounding as outlined in the potential confounding variables section. Cox proportional hazards models with POH modeled as a time-dependent covariate were used for the primary outcome and all mortality-associated end points (proportional hazards assumption was verified) to ensure only outcomes were included where the event followed the exposure. If the event occurred before the POH exposure, the patient was not included in the analysis for that specific outcome. The timing of POH was evaluated in the original cohort postsurgery without IOH, by surgery type, for the 3 POH thresholds. We also evaluate conditional hazards of POH with or without IOH exposure, and the potential interaction between IOH and POH exposures in a combined cohort (cohort #3) of patient-procedures.

For secondary outcomes that did not include a mortality or length-of-stay component, Fine-Gray regression models32 were performed to account for the competing risk of death, and subdistribution hazards are reported. A sensitivity analysis was performed excluding patients who died in the first 48 hours. The number of patients right censored for the models and censored due to death were counted (Supplemental Digital Content, Table 6, http://links.lww.com/AA/D312, for censored patients). Hospital-free days in the first 30 days were assessed via Poisson regression and a negative binominal model.

Patients who had any record of an outcome or a procedure within the 30-day presurgery were excluded from the corresponding outcome analyses. In a post hoc analysis, POH was treated as a time-dependent continuous exposure (defined as lowest MAP value on each postoperative day), and a restricted cubic spline was used to explore the nonlinear relationship between POH and outcomes.

Since we evaluated the outcomes across 3 POH thresholds, we used the Bonferroni correction33 with a P < .05/3 (.016). Hospital-free days (counted immediately postsurgery) were reported as incidence rate ratios (IRRs) with 98.4% CIs; all other outcomes were reported as hazard ratios (HRs) and 98.4% CIs. All analyses were conducted using SAS version 9.4 (SAS Institute Inc, Cary, NC) and R3.5.2.

Sample Size Considerations

A previous US study (2004–2013) found a 3% MACCE incidence rate in 10,581,621 hospitalizations.25 We hypothesized that a 1% difference in MACCE (95% CI and power) would be detectable in a sample size of 17,550. Therefore, our study with 67,968 procedures should be powered to detect a difference of at least 0.5% for MACCE.

RESULTS

POH Without IOH: Cohort Characteristics

The original cohort included 67,968 procedures (64,542 patients) managed on the ward (34,839 procedures identified by algorithm [Supplemental Digital Content, Method 1, for algorithm for patients discharged to ward; Figure 2, http://links.lww.com/AA/D312, for receiver operating characteristic] at a positive predictive value [PPV] of 0.97) with patient selection in Figure 1. Of the cohort procedures, 43,157 (63%) experienced a MAP ≤75 mm Hg within 48 hours postsurgery; 15,377 (22%) and 2417 (4%) experienced hypotension for thresholds of MAP ≤65 and ≤55 mm Hg, respectively. A large proportion of total POH events occurred within the first 4 hours for all MAP thresholds (Figure 2). The incidence of hypotensive episodes by threshold and prior surgery types are in Supplemental Digital Content, Figure 3, http://links.lww.com/AA/D312.

Figure 2.
Figure 2.:
Timing distribution for the first POH event after noncardiac surgery. Distribution provided for patients discharged to the ward without intraoperative hypotension. The proportion of patients with a first POH event of a specific threshold during the defined time frame is shown for all MAP thresholds (MAP ≤75, ≤ 65, ≤ 55 mm Hg). MAP indicates mean arterial pressure; POH, postoperative hypotension.
Table 1. - Cohort Baseline Characteristics
POH
Patient characteristics Overall (n = 67,968) MAP ≤55 mm Hg (n = 2417) MAP >55 mm Hg (n = 65,551) MAP ≤65 mm Hg (n = 15,377) MAP >65 mm Hg (n = 52,591) MAP ≤75 mm Hg (n = 43,157) MAP >75 mm Hg (n = 24,811)
Sex
 Male 26,167 (39%) 646 (27%) 40,030 (61%) 3969 (26%) 22,198 (42%) 13,770 (32%) 12,397 (50%)
 Female 41,801 (62%) 1771 (73%) 25,521 (39%) 11,408 (74%) 30,393 (58%) 29,387 (68%) 12,414 (50%)
Race
 Asian 316 (1%) 6 (0%) 310 (0%) 73 (1%) 243 (0%) 209 (1%) 107 (1%)
 Black 7011 (10%) 126 (5%) 6885 (11%) 897 (6%) 6114 (12%) 3336 (8%) 3675 (15%)
 Other 4905 (7%) 134 (6%) 4771 (7%) 1035 (7%) 3870 (7%) 3034 (7%) 1871 (8%)
 White 55,736 (82%) 2151 (89%) 53,585 (82%) 13,372 (87%) 42,364 (81%) 36,578 (85%) 19,158 (77%)
Region
 Midwest 29,817 (44%) 1143 (48%) 28,674 (44%) 7045 (46%) 22,772 (43%) 19,231 (45%) 10,586 (43%)
 North 1353 (2%) 39 (2%) 1314 (2%) 212 (1%) 1141 (2%) 761 (2%) 592 (2%)
 Other 1603 (2%) 81 (3%) 1522 (2%) 403 (3%) 1200 (2%) 1025 (2%) 578 (2%)
 South 32,943 (48%) 1018 (42%) 31,925 (49%) 7058 (46%) 25,885 (49%) 20,607 (48%) 12,336 (50%)
 West 2252 (3%) 136 (6%) 2116 (3%) 659 (4%) 1593 (3%) 1533 (4%) 719 (3%)
Age (y)
 <40 4813 (7%) 110 (5%) 4703 (7%) 929 (6%) 3884 (7%) 2805 (7%) 2008 (8%)
 40–49 6827 (10%) 157 (7%) 6670 (10%) 1153 (8%) 5674 (11%) 3748 (9%) 3079 (12%)
 50–59 13,739 (20%) 363 (15%) 13,376 (20%) 2609 (17%) 11,130 (21%) 8147 (19%) 5592 (23%)
 60–69 18,927 (28%) 647 (27%) 18,280 (28%) 4160 (27%) 14,767 (28%) 12,044 (28%) 6883 (28%)
 70–79 14,985 (22%) 613 (25%) 14,372 (22%) 3770 (25%) 11,215 (21%) 10,147 (24%) 4838 (20%)
 ≥80 8677 (13%) 527 (22%) 8150 (12%) 2756 (18%) 5921 (11%) 6266 (15%) 2411 (10%)
Charlson comorbidity index
 0 29,502 (43%) 960 (40%) 28,542 (44%) 6524 (42%) 22,978 (44%) 18,825 (44%) 10,677 (43%)
 1 12,936 (19%) 447 (18%) 12,489 (19%) 2805 (18%) 10,131 (19%) 8206 (19%) 4730 (19%)
 2 10,128 (15%) 375 (16%) 9753 (15%) 2331 (15%) 7798 (15%) 6488 (15%) 3640 (15%)
 3 5730 (8%) 256 (11%) 5474 (8%) 1428 (9%) 4302 (8%) 3642 (8%) 2088 (8%)
 4+ 9672 (14%) 379 (16%) 9293 (14%) 2289 (15%) 7383 (14%) 5996 (14%) 3676 (15%)
Year of surgery
 2008–2011a 3252 (5%) 145 (6%) 3107 (5%) 787 (5%) 2465 (5%) 2033 (5%) 1219 (5%)
 2012–2013 12,687 (19%) 561 (23%) 12,126 (19%) 3149 (20%) 9538 (18%) 8321 (19%) 4366 (18%)
 2014–2015 24,833 (37%) 833 (34%) 24,000 (37%) 5609 (36%) 19,224 (37%) 15,748 (36%) 9085 (37%)
 2016–2017 27,196 (40%) 878 (36%) 26,318 (40%) 5832 (38%) 21,364 (41%) 17,055 (40%) 10,141 (41%)
Surgery length (h)
 ≤1 57,049 (84%) 2064 (85%) 54,985 (84%) 13,102 (85%) 43,947 (84%) 36,733 (85%) 20,316 (82%)
 1–2 9384 (14%) 313 (13%) 9071 (14%) 2019 (13%) 7365 (14%) 5607 (13%) 3777 (15%)
 >2 1535 (2%) 40 (2%) 1495 (2%) 256 (2%) 1279 (2%) 817 (2%) 718 (3%)
Use of antihypertensives (year presurgery) 56,153 (83%) 2024 (84%) 54,129 (83%) 12,598 (82%) 43,555 (83%) 35,289 (82%) 20,864 (84%)
Surgery types (10 most common)
 Limb amputation 2246 (4%) 58 (2%) 2388 (4%) 398 (3%) 2048 (4%) 1253 (3%) 1193 (5%)
 Gallbladder surgery 2210 (3%) 36 (1%) 2174 (3%) 339 (2%) 1871 (4%) 1178 (3%) 1032 (4%)
 Colon surgery 3533 (5%) 78 (3%) 3455 (5%) 602 (4%) 2931 (6%) 1944 (5%) 1589 (6%)
 Craniotomy 98 (0%) 2 (0%) 96 (0%) 21 (0%) 77 (0%) 63 (0%) 35 (0%)
 Spinal fusion 3335 (5%) 122 (5%) 3213 (5%) 830 (5%) 2505 (5%) 2205 (5%) 1130 (5%)
 Spinal fusion laminectomy 1774 (3%) 63 (3%) 1711 (3%) 508 (3%) 1266 (2%) 1286 (3%) 488 (2%)
 Open reduction of fracture 4104 (6%) 214 (9%) 3890 (6%) 1264 (8%) 2840 (5%) 2861 (7%) 1243 (5%)
 Hip prosthesis 8968 (13%) 532 (22%) 8436 (13%) 2985 (19%) 5983 (11%) 6879 (16%) 2089 (8%)
 Knee prosthesis 19,212 (28%) 615 (25%) 18,597 (28%) 4051 (26%) 15,161 (29%) 12,504 (29%) 6708 (27%)
 Other 21,491 (32%) 640 (26%) 20,851 (32%) 4139 (27%) 17,352 (33%) 12,423 (29%) 9068 (37%)
 Thoracic surgeryb 797 (1%) 57 (2%) 740 (1%) 240 (2%) 557 (1%) 561 (1%) 236 (1%)
Characteristics for noncardiac surgery patients managed on the floor for 48 hours after surgery with no history of intraoperative hypotension (MAP ≤65 mm Hg). Because of rounding, categories will not always add to 100%.
Abbreviations: MAP, mean arterial pressure; POH, postoperative hypotension.
aFour years were combined due to small sample size in 2008 and 2009.
bNoncardiac and nonvascular.

Table 2. - Rate of Adverse Events Among Noncardiac Surgery Patients Discharged to the Floor Without Intraoperative Hypotension (MAP <65 mm Hg)
POH
Overall MAP ≤ 55 mm Hg MAP ≤ 65 mm Hg MAP ≤ 75 mm Hg MAP > 75 mm Hg
Adverse event Number of patients Patients with event (%) Number of patients Patients with event (%) Number of patients Patients with event (%) Number of patients Patients with event (%) Number of patients Patients with event (%)
30-d MACCE 66,591 1054 (1.6 %) 2332 62 (2.7%) 14,976 297 (2.0%) 42,214 702 (1.7%) 24,377 352 (1.4%)
30-d mortality 67,968 594 (0.9%) 2411 40 (1.7%) 15,368 193 (1.3%) 43,151 412 (1.0%) 24,817 182 (0.7%)
90-d mortality 67,968 1048 (1.5%) 2411 78 (3.2%) 15,368 338 (2.2%) 43,151 722 (1.7%) 24,817 326 (1.3%)
30-d AMI 67,317 226 (0.3%) 2372 16 (0.7%) 15,170 59 (0.4%) 42,692 143 (0.3%) 24,625 83 (0.3%)
30-d AIS 67,200 362 (0.5%) 2381 23 (1.0%) 15,178 91 (0.6%) 42,658 232 (0.5%) 24,542 130 (0.5%)
7-d AKI (II/III) 67,845 986 (1.5%) 2360 60 (2.5%) 15,214 256 (1.7 %) 42,965 607 (1.4%) 24,880 379 (1.5%)
30-d readmission 67,580 4920 (7.3%) 2387 233 (9.8%) 15,233 1211 (8.0%) 42,872 3166 (7.4%) 24,708 1754 (7.1%)
Patients are stratified by postoperative hypotensive status. AMI and AIS were captured using ICD-9/10 codes (see Supplemental Digital Content, Table 1, http://links.lww.com/AA/D312, for ICD codes). AKI stage II/III was defined using the following definition: postoperative creatinine 2 times greater than the most recent preoperative value, an increase in serum creatinine ≥4 mg/dL, or initiation of dialysis therapy.
Abbreviations: AIS, acute ischemic stroke; AKI, acute kidney injury; AMI, acute myocardial infarction; ICD, International Classification of Diseases; MACCE, major adverse cardiac and cerebrovascular events; MAP, mean arterial pressure; POH, postoperative hypotension.

The original cohort was predominantly female (n = 41,801 [61%]) with a median CCI of 1 (25th, 75th; 0, 2) and mean (standard deviation [SD]) age of 62.9 (14.5) years (Table 1). Patients ≤18 years comprised 0.4% of the population. All outcomes were more frequent among patients who experienced POH at lower thresholds (Table 2). MACCE was present in 1.7% of patients with MAP ≤75 mm Hg, 2.0% with MAP ≤65 mm Hg, and 2.7% with MAP ≤55 mm Hg.

Outcomes: POH Cohort Without IOH

Adjusted HRs and 98.4% CIs for the original cohort are reported in Figure 3A. In adjusted models for the non-IOH cohort, POH was not associated with MACCE for any MAP threshold (MAP ≤75 mm Hg, HR 1.18 [98.4% CI, 0.99-1.39], P = .023; MAP ≤65 mm Hg, HR 1.18 [0.99–1.41], P = .028; MAP ≤55 mm Hg, HR 1.23 [0.90–1.71], P = .121). However, significant associations between POH and secondary outcomes were observed at all thresholds for AKI and 30-day readmissions (Figure 3A). Furthermore, an association was also found between secondary outcomes and POH MAP ≤55 mm Hg for 90-day mortality, and for POH MAP ≤65 mm Hg and 30- and 90-day mortality; yet the association was not significant for other thresholds (Supplemental Digital Content, Table 7, http://links.lww.com/AA/D312, for P values). E values for unmeasured confounders required to reduce the odds ratio to 1.0 for MACCE and secondary outcomes in the original cohort ranged from 3.58 to 1.40 (AKI ≤55 mm Hg threshold to 30-day readmissions ≤65 mm Hg threshold) for significant outcomes (Supplemental Digital Content, Table 5, http://links.lww.com/AA/D312). Additionally, across all thresholds, POH was significantly associated with reduced hospital-free days (MAP ≤75 mm Hg, IRR 0.992 [98.4% CI, 0.988-0.996], P < .001; MAP ≤65 mm Hg, IRR 0.987 [0.982–0.991], P < .001; MAP ≤55 mm Hg, IRR 0.971 [0.961–0.981], P < .001).

Figure 3.
Figure 3.:
Adjusted hazard and subdistribution hazard ratios for ward patients with POH. Data shown for patients (A) without IOH exposure and (B) with IOH exposure (cohort #2) for various POH MAP thresholds. Panels (C) and (D) report conditional hazards of POH exposure in a combined cohort of patients with/without IOH exposure (cohort #3), controlling for the interaction between IOH and POH (n = 67,968 [A]; n = 16,034 [B]; n = 84,002 [C] and [D]). Potential confounding variables used in the models are detailed in Methods and listed in Table 1, and Supplemental Digital Content Tables 2–4, http://links.lww.com/AA/D312. Adj. indicates adjusted; AIS, acute ischemic stroke; AKI, acute kidney injury; AMI, acute myocardial infarction; CI, confidence interval; HR, hazard ratio; IOH, intraoperative hypotension; MACCE, major adverse cardiac or cerebrovascular events; MAP, mean arterial pressure; POH, postoperative hypotension; SDHR, subdistribution hazard ratio.

We observed similar results when we: (1) excluded POH events in the first 4 hours to isolate the ward from the postanesthesia care unit (PACU; Supplemental Digital Content, Table 8, http://links.lww.com/AA/D312, for hazards for the ward cohort with/without IOH); (2) evaluated the cohort of patients with a documented care unit postsurgery (Supplemental Digital Content, Table 9, http://links.lww.com/AA/D312); and (3) evaluated patients who were managed on the ward for 72 and 96 hours (Supplemental Digital Content, Table 9, http://links.lww.com/AA/D312, outcomes for patients without IOH and identified as discharged to the ward). A sensitivity analysis excluding patients who died in the first 48 hours provided similar results for MACCE (Supplemental Digital Content, Table 10, http://links.lww.com/AA/D312, for sensitivity analysis).

Outcomes: POH Cohort With IOH

Cohort #2 (with IOH) included 16,034 procedures (Supplemental Digital Content, Table 3, http://links.lww.com/AA/D312, for baseline characteristics), of which 13,529 (84%) experienced postoperative MAP ≤75 mm Hg; 7639 (48%) ≤65 mm Hg; and 1903 (12%) ≤55 mm Hg, within 48 hours postsurgery. POH with IOH was associated with MACCE for MAP ≤55 mm Hg (HR 1.53 [98.4% CI, 1.05-2.22], P = .006); however, the associations were not significant for other thresholds (MAP ≤65 mm Hg: HR 1.29 [0.93–1.80], P = .061; ≤75 mm Hg: HR 1.14 [0.73–1.81], P = .484). Forest plots of adjusted HRs for cohort #2 are presented in Figure 3B.

Associations between POH and the secondary outcome AKI were found for both MAP ≤65 and ≤55 mm Hg. For patients with IOH, we found no association between POH and the secondary outcome 30-day readmissions for any threshold investigated. Secondary outcomes 30- and 90-day mortality were associated with POH exposure only for MAP ≤55 mm Hg (Supplemental Digital Content, Table 7, http://links.lww.com/AA/D312, for P values). E values assessing unmeasured confounding for MACCE in cohort #2 with IOH for significant end points ranged from 3.87 to 2.24 (AKI ≤55 mm Hg threshold to AKI ≤65 mm Hg threshold) (Supplemental Digital Content, Table 5, http://links.lww.com/AA/D312).

Conditional HRs and Interaction Between POH and IOH

Given previous literature,16 in which a strong interaction between IOH and POH and associated outcomes was observed, we chose to perform an additional analysis in which we accounted for the interaction between IOH and POH when looking at the effects of POH in a combined cohort (cohort #3) without (n = 67,968) and with (n = 16,034) IOH exposure, resulting in 84,002 procedures (Supplemental Digital Content, Table 4, http://links.lww.com/AA/D312, for baseline characteristics). However, our analysis revealed that the interaction terms between POH and IOH exposures on primary and secondary outcomes were not significant (Supplemental Digital Content, Table 11, http://links.lww.com/AA/D312, for interaction P values).

Overall, the application of the interaction term revealed similar results, with only slight differences in the conditional HRs. When an interaction term between POH and IOH was utilized, the association between POH without IOH exposure and MACCE was significant for MAP ≤75 mm Hg (HR 1.20 [98.4% CI, 1.01-1.41]) and MAP ≤65 mm Hg (HR 1.21 [1.02–1.45]), but not MAP ≤55 mm Hg (HR 1.26 [0.90–1.74]) (Figure 3C). When the interaction term between hypotension exposures was utilized, the association between POH with IOH exposure and MACCE was significant for MAP ≤55 mm Hg (HR 1.45 [98.4% CI, 1.01-2.06]) but not MAP ≤65 mm Hg (HR 1.18 [0.88–1.59]) or MAP ≤75 mm Hg (HR 1.15 [0.76–1.74]) (Figure 3D).

Modeling to Examine Overall Risk of POH

Models examining lowest POH MAP on any given day, fitted with a cubic spline, revealed an expanding association of MACCE across thresholds, with accentuated associations at exposures to MAP ≤65 mm Hg (Supplemental Digital Content, Figure 4, http://links.lww.com/AA/D312, secondary outcomes among patients without IOH). Similar patterns were observed for secondary outcomes of 30- and 90-day mortality, AMI, AKI and 30-day readmissions. No clear pattern was observed for AIS (Figure 5, Supplemental Digital Content, Figure 5, http://links.lww.com/AA/D312, Restricted cubic spline to explore nonlinear relationship between POH and MACCE).

DISCUSSION

Death within 30 days postsurgery is a common cause of mortality in the United States.1,25,34,35 Hypotension-related events such as major bleeding and myocardial injury are common occurrences in the postoperative period.36 This is the first population-based cohort study of noncardiac surgery patients who had acceptable intraoperative BP but had POH, where an association with adverse clinical outcomes at various hemodynamic thresholds was examined. Even though the primary outcome of 30-day MACCE was not associated with POH, hypotension in the postoperative period at a MAP threshold ≤65 mm Hg was associated with a larger number of other AEs (HR >1 for AKI, 30-day readmissions, and 90-day mortality).

The fact that we found no association between POH (in the absence of IOH) and MACCE at any MAP threshold investigated might partially be due to the intermittent nature of ward BP recordings. Additionally, the small sample size for the MAP ≤55 mm Hg exposure group may have affected our ability to detect a statistical difference in this group. This may have contributed to the lack of a clear “dose-response” relationship for the associations of POH with multiple secondary outcomes (AIS/AMI). Liem et al37 demonstrated an association of POH with myocardial injury; however, they measured BP every hour in a high-dependency unit. POH is not entirely benign. The association of 90-day mortality increased in a stepwise manner below MAP ≤65 mm Hg and the association of 30-day readmission was significant for MAP ≤75 mm Hg, and further increased as MAP dropped to ≤55 mm Hg. Consistent increase in the association of kidney injury with hypotension was previously shown.6,16,17 Similarly, we demonstrated a clear and consistent association with 7-day AKI as MAP decreased to ≤75 mm Hg, which doubled as MAP dropped to ≤55 mm Hg.

Previously, postoperative surgical ICU patients were shown to be exquisitely sensitive to hypotension; however, the relationship depended on the amount of IOH.16 Other cohorts demonstrated that postoperative ward hypotension was associated with a 3-fold increased association of 30-day myocardial infarction and mortality compared with the same amount of IOH or POH on the day of surgery, after adjusting for IOH but only using a single SBP threshold (90 mm Hg).9 However, our novel work examined an exclusive patient cohort without IOH, across multiple POH thresholds and for associations with outcomes extending beyond specific cardiac end points (ie, MINS).

Using cohort #2 of patients with IOH (intraoperative MAP ≤65 mm Hg), we found slight differences in the HRs for primary and secondary outcomes, where most associations between POH and outcomes emerged only at low (MAP ≤55 mm Hg) POH thresholds. This may be because patients with IOH likely received aggressive interventions in the PACU or on the ward and those with early warnings of being at highest risk for these events (ie, with rising troponin in the PACU) were likely sent to a higher level of care for intervention, and excluded from the analysis. However, those with a “double hit” and a profound, and possibly unexpected significant POH of MAP ≤55 mm Hg, may have potentiated organ system injury. We also hypothesize that prior exposure to IOH may have rendered these patients more resistant to smaller drops in pressure (up to MAP >55 mm Hg) due to possible ischemic preconditioning.38,39 Importantly, when we utilized an interaction term between IOH and POH (P values for the interaction terms were not significant) the trends in outcome HRs remained largely similar from a clinical perspective, despite slight changes in statistical significance. The slight difference in the associations between POH and primary/secondary outcomes most likely resulted from the fact that although the interaction term was not significant, the term was also not close to 1.0 and therefore, the interaction terms had some effect on the HRs depicted in Figure 3C, D.

Hypotension is common, prolonged, and largely unpredictable in the postoperative period.9,19 Approximately 50% of noncardiac surgical patients with MAP ≤65 mm Hg for up to 30 continuous minutes are missed using standard monitoring compared with continuous monitoring on the postoperative ward.19 We demonstrated that POH was associated with adverse outcomes, including mortality and 30-day readmissions, in a novel patient population with no hemodynamic perturbations during surgery, and were likely sent directly to the ward from the operating room or PACU. Considering that these data are based on intermittent traditional ward monitoring, 1 such intervention to prevent adverse outcomes could have potentially been decreasing exposure time to hypotension had these patients been monitored continuously.

Limitations to our work include a patient cohort hospitalized ≥48 hours postoperatively; therefore, results may not be generalizable to all patients. Second, there is risk of missing/invalid BP data, for which we attempted to control by restricting the interval between MAP readings and implementing validity criteria, which may have impacted our ability to establish a “dose-response” gradient. We used strict criteria that excluded patients with frequent and longer durations of missing data as reported and published elsewhere.17 Third, because this was an observational analysis and data were subject to reporting bias/data entry errors, uncontrolled confounding may exist. E values provide the strength of an association that an unmeasured confounder must have to reduce the observed association to unity. Because calculated E values for significant outcomes (with/without IOH) ranged from 3.87 to 1.40 (with IOH: AKI MAP ≤65 mm Hg to without IOH: 30-day readmissions MAP ≤65 mm Hg), an unmeasured confounder that was associated with the exposure and outcome of interest by a risk ratio of 3.87- or 1.40-fold each, above and beyond the measured confounders, could explain it away, but weaker confounding could not. Although we controlled for all known confounders available in our data set, should hypotension simply be an indicator of the severity of the underlying illness, the estimates of risk in this study would be overestimated.9 Fourth, since patient numbers with MAP ≤55 mm Hg are fewer than for higher thresholds, further evaluation would be warranted to confirm our findings. Fifth, even though we have restricted gaps in MAP readings and controlled for antihypertensive/vasopressor use postsurgery, we had no control over specific protocols within/across hospitals. Therefore, we could not control for more aggressive therapy or more intensive and frequent BP monitoring that sicker patients may have received on the hospital ward. Sixth, we used AKI stage II/III as our outcome measure, though AKI stage I has similar long-term renal dysfunction when compared with stage II/III.40 At the time, our decision was based on previous literature showing a significant increase of hospital mortality with stage II/III renal dysfunction.41 Seventh, we utilized a Bonferroni correction of P < .016 and 98.4% CI for secondary outcomes that corresponded to our 3 a priori defined MAP thresholds, but did not take into account the number of secondary outcomes. We made inference on 84 associations all together; even with our applied conventional Bonferroni correction there is a chance that some of our results might be false positive findings. Therefore, these hypothesis-generating outcomes may help develop future prospective trials. Finally, we developed algorithms to assign patients without a known care location, to the ward (PPV for model was 0.97, thus 3% of patients could be misidentified). However, we did observe similar trends for HRs in patients with known care location.

In conclusion, our analyses revealed no association between MACCE and POH in patients without IOH. Additionally, we found no interaction between postoperative and intraoperative hypotension on postoperative AEs. Significance of these findings appears to be tempered by the vastly intermittent nature of routine MAP assessment on the ward. Future prospective randomized trials are necessary to show whether greater fidelity measurements and avoidance of POH may be beneficial.

ACKNOWLEDGMENTS

The authors thank Stuart Gander, Jason Jager, and Otto Liska of Boston Consulting Group for their support to this research; Matthew Brummond, BBA, of Edwards Lifesciences, Irvine, CA for assistance with figures; and Sibyl H. Munson, PhD, Halit O. Yapici, MD, MPH, MBA, and Francie Moehring, PhD, of Boston Strategic Partners, Boston, MA (supported by Edwards Lifesciences) for editorial support.

DISCLOSURES

Name: Ashish K. Khanna, MD, FCCP, FCCM.

Contribution: This author helped design the study, supervise the data collection, analyze the data, and prepare the manuscript.

Conflicts of Interest: A. K. Khanna consults for Medtronic, and Potrero Medical and is funded with a Clinical and Translational Science Institute (CTSI) NIH/NCTAS KL2 TR001421 award for a trial on continuous postoperative hemodynamic and saturation monitoring. The author also received consulting fees from Edwards Lifesciences for this work.

Name: Andrew D. Shaw, MB, FRCPC.

Contribution: This author helped supervise the data collection, analyze the data, and prepare the manuscript.

Conflicts of Interest: A. D. Shaw received consulting fees from Edwards Lifesciences.

Name: Wolf H. Stapelfeldt, MD.

Contribution: This author helped supervise the data collection, analyze the data, and prepare the manuscript.

Conflicts of Interest: W. H. Stapelfeldt received consulting fees from Edwards Lifesciences.

Name: Isabel J. Boero, MD, MS.

Contribution: This author helped design the study, supervise the data collection, analyze the data, and prepare the manuscript.

Conflicts of Interest: I. J. Boero is an employee of Boston Consulting Group, who received funds from Edwards Lifesciences to perform the research.

Name: Qinyu Chen, MS.

Contribution: This author helped analyze the data, and prepare the manuscript.

Conflicts of Interest: Q. Chen is an employee of Boston Consulting Group, who received funds from Edwards Lifesciences to perform the research.

Name: Mitali Stevens, PharmD, BCPS.

Contribution: This author helped analyze the data and prepare the manuscript.

Conflicts of Interest: M. Stevens is an employee of Edwards Lifesciences.

Name: Anne Gregory, MD, MSc, FRCPC.

Contribution: This author helped analyze the data and prepare the manuscript.

Conflicts of Interest: None.

Name: Nathan J. Smischney, MD, MSc.

Contribution: This author helped design the study, supervise the data collection, analyze the data, and prepare the manuscript.

Conflicts of Interest: N. J. Smischney received consulting fees from Edwards Lifesciences.

This manuscript was handled by: Tong J. Gan, MD.

    REFERENCES

    1. Semel ME, Lipsitz SR, Funk LM, Bader AM, Weiser TG, Gawande AA. Rates and patterns of death after surgery in the United States, 1996 and 2006. Surgery. 2012;151:171–182.
    2. Goren O, Matot I. Perioperative acute kidney injury. Br J Anaesth. 2015;115(suppl 2):ii3–14.
    3. Botto F, Alonso-Coello P, Chan MT, et al.; 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;120:564–578.
    4. Dalton JE, Kurz A, Turan A, Mascha EJ, Sessler DI, Saager L. Development and validation of a risk quantification index for 30-day postoperative mortality and morbidity in noncardiac surgical patients. Anesthesiology. 2011;114:1336–1344.
    5. Gupta PK, Gupta H, Sundaram A, et al. Development and validation of a risk calculator for prediction of cardiac risk after surgery. Circulation. 2011;124:381–387.
    6. Walsh M, Devereaux PJ, Garg AX, et al. Relationship between intraoperative mean arterial pressure and clinical outcomes after noncardiac surgery: toward an empirical definition of hypotension. Anesthesiology. 2013;119:507–515.
    7. 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.
    8. Salmasi V, Maheshwari K, Yang D, et al. 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.
    9. Sessler DI, Meyhoff CS, Zimmerman NM, et al. 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;128:317–327.
    10. Monk TG, Saini V, Weldon BC, Sigl JC. Anesthetic management and one-year mortality after noncardiac surgery. Anesth Analg. 2005;100:4–10.
    11. Bijker JB, Persoon S, Peelen LM, et al. Intraoperative hypotension and perioperative ischemic stroke after general surgery: a nested case-control study. Anesthesiology. 2012;116:658–664.
    12. Stapelfeldt WH, Yuan H, Dryden JK, et al. The SLUScore: a novel method for detecting hazardous hypotension in adult patients undergoing noncardiac surgical procedures. Anesth Analg. 2017;124:1135–1152.
    13. Sun LY, Wijeysundera DN, Tait GA, Beattie WS. Association of intraoperative hypotension with acute kidney injury after elective noncardiac surgery. Anesthesiology. 2015;123:515–523.
    14. Monk TG, Bronsert MR, Henderson WG, et al. Association between intraoperative hypotension and hypertension and 30-day postoperative mortality in noncardiac surgery. Anesthesiology. 2015;123:307–319.
    15. Wesselink EM, Kappen TH, Torn HM, Slooter AJC, van Klei WA. Intraoperative hypotension and the risk of postoperative adverse outcomes: a systematic review. Br J Anaesth. 2018;121:706–721.
    16. Khanna AK, Maheshwari K, Mao G, et al. Association between mean arterial pressure and acute kidney injury and a composite of myocardial injury and mortality in postoperative critically ill patients: a retrospective cohort analysis. Crit Care Med. 2019;47:910–917.
    17. Maheshwari K, Nathanson BH, Munson SH, et al. The relationship between ICU hypotension and in-hospital mortality and morbidity in septic patients. Intensive Care Med. 2018;44:857–867.
    18. Roshanov PS, Sheth T, Duceppe E, et al. Relationship between perioperative hypotension and perioperative cardiovascular events in patients with coronary artery disease undergoing major noncardiac surgery. Anesthesiology. 2019;130:756–766.
    19. Turan A, Chang C, Cohen B, et al. Incidence, severity, and detection of blood pressure perturbations after abdominal surgery: a prospective blinded observational study. Anesthesiology. 2019;130:550–559.
    20. Gregory A, Stapelfeldt WH, Khan A, et al. Intraoperative hypotension is associated with adverse clinical outcomes after non-cardiac surgery. Anesth Analg. Published online November 6, 2020. doi: 10.1213/ANE.0000000000005250.
    21. CDC. 2019 NHSN Operative Procedure Code Mappings.2019. Accessed May 8, 2019. https://www.cdc.gov/nhsn/xls/icd10-pcs-pcm-nhsn-opc.xlsx.
    22. CDC. ICD-10-PCS & CPT CodesGuidance for HPRO & KPRO Procedure Details.2019. Accessed May 8, 2019. https://www.cdc.gov/nhsn/xls/guidance-for-hpro-kpro-procedure-details.xlsx.
    23. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45:613–619.
    24. VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the E-value. Ann Intern Med. 2017;167:268–274.
    25. Smilowitz NR, Gupta N, Ramakrishna H, Guo Y, Berger JS, Bangalore S. Perioperative major adverse cardiovascular and cerebrovascular events associated with noncardiac surgery. JAMA Cardiol. 2017;2:181–187.
    26. Xia J, Xu J, Li B, et al. Association between glycemic variability and major adverse cardiovascular and cerebrovascular events (MACCE) in patients with acute coronary syndrome during 30-day follow-up. Clin Chim Acta. 2017;466:162–166.
    27. Ham SY, Song SW, Nam SB, Park SJ, Kim S, Song Y. Effects of chronic statin use on 30-day major adverse cardiac and cerebrovascular events after thoracic endovascular aortic repair. J Cardiovasc Surg (Torino). 2018;59:836–843.
    28. ICD10data.com. Convert ICD-9-CM Codes to ICD-10-CM/PCS.Published 2018. Accessed March 1, 2019. https://www.icd10data.com/Convert.
    29. National Cancer Institute SEER. ICD-9-CM to ICD-10-CM Based on FY2017 ICD-10-CM codes.Published 2016. Accessed March 1, 2019. https://seer.cancer.gov/tools/conversion/.
    30. Koeze J, Keus F, Dieperink W, van der Horst IC, Zijlstra JG, van Meurs M. Incidence, timing and outcome of AKI in critically ill patients varies with the definition used and the addition of urine output criteria. BMC Nephrol. 2017;18:70.
    31. Ostermann M, Chang R; Riyadh ICU Program Users Group. Correlation between the AKI classification and outcome. Crit Care. 2008;12:R144.
    32. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94:496–509.
    33. Armstrong RA. When to use the Bonferroni correction. Ophthalmic Physiol Opt. 2014;34:502–508.
    34. Li G, Warner M, Lang BH, Huang L, Sun LS. Epidemiology of anesthesia-related mortality in the United States, 1999-2005. Anesthesiology. 2009;110:759–765.
    35. Fecho K, Lunney AT, Boysen PG, Rock P, Norfleet EA. Postoperative mortality after inpatient surgery: incidence and risk factors. Ther Clin Risk Manag. 2008;4:681–688.
    36. Spence J, LeManach Y, Chan MTV, et al. Association between complications and death within 30 days after noncardiac surgery. CMAJ. 2019;191:E830–E837.
    37. Liem VGB, Hoeks SE, Mol KHJM, et al. Postoperative hypotension after noncardiac surgery and the association with myocardial injury. Anesthesiology. 2020;133:510–522.
    38. Zarbock A, Schmidt C, Van Aken H, et al.; RenalRIPC Investigators. Effect of remote ischemic preconditioning on kidney injury among high-risk patients undergoing cardiac surgery: a randomized clinical trial. JAMA. 2015;313:2133–2141.
    39. Ali ZA, Callaghan CJ, Lim E, et al. Remote ischemic preconditioning reduces myocardial and renal injury after elective abdominal aortic aneurysm repair: a randomized controlled trial. Circulation. 2007;116:I98–105.
    40. Turan A, Cohen B, Adegboye J, et al. Mild acute kidney injury after noncardiac surgery is associated with long-term renal dysfunction: a retrospective cohort study. Anesthesiology. 2020;132:1053–1061.
    41. Hoste EA, Bagshaw SM, Bellomo R, et al. Epidemiology of acute kidney injury in critically ill patients: the multinational AKI-EPI study. Intensive Care Med. 2015;41:1411–1423.

    Supplemental Digital Content

    Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the International Anesthesia Research Society.