Intraoperative Hypotension and Myocardial Infarction Development Among High-Risk Patients Undergoing Noncardiac Surgery

A Nested Case-Control Study

Linn Hallqvist, MD, PhD Student, DESA; Fredrik Granath, PhD; Michael Fored, MD, PhD; Max Bell, MD, PhD


Anesth Analg. 2021;133(1):6-15. 

In This Article


Study Design

In a nested case-control study, MI case patients matched with non–MI patients from the same source population.

The study registered at ClinicalTrials.Gov (NCT03974321; Intraoperative Hypotension and Perioperative Myocardial Injury) before analysis is provided in the Supplemental Digital Content 1, Registry Protocol,

Data Sources and Study Population

The source population was identified from a large surgical cohort, the Orbit cohort,[18] of patients undergoing noncardiac surgery in Sweden between 2007 and 2014. This was collected from 23 Swedish hospitals and data were linked, using the unique Swedish personal identification number, to several national registries held by the National Board of Health and Welfare; the National Patient Register (NPR)[19] using International Classification of Diseases (ICD)-10 codes, the Swedish Cause of Death Registry,[20] the Swedish Prescribed Drug Register[21] and to the National Quality Registry: Swedeheart.[22] Detailed information is found in the previous study.[18]

Study Participants

Adults undergoing noncardiac surgery at 3 Swedish hospitals, Karolinska-, Malmö-, and Lund University hospital, were eligible for inclusion. We excluded cardiac-, obstetric-, minor, and ambulatory care surgeries and if valid surgery codes or American Society of Anesthesiologists (ASA) physical status were unavailable.

Cases were patients developing MI <30 days postsurgery as registered in the NPR—and/or Swedeheart. One control was selected for each case, matched by age (5-year intervals), sex, ASA physical status, cardiovascular disease, surgical year, hospital, surgical code, acute/elective surgery, and duration of surgery (less or greater than 3 hours). The selection of matching variables was based on risk factors of MI identified in the Orbit cohort study.[18] For 10% of the sampled cases, an exact matched control could not be identified and matching on calendar year and duration of surgery was relaxed, resulting in a slight imbalance regarding these factors. Controls were sampled among patients alive without MI diagnosis at day 30, that is, cumulative incidence sampling. Description of the source population and case-control selection is found in Figure 1.

Figure 1.

Participant flowchart. ASA indicates American Society of Anesthesiologists.

Data Collection

Electronic medical records validated MI diagnoses and comorbidities. Information retrieval was performed blinded to case-control status. Preoperative history of cardiovascular disease and diabetes mellitus (DM) was registered. Baseline BP was determined as the patient's habitual value measured as an estimate of all BPs, 5 on average, documented within 2 months before surgery, obtained from the surgical ward, preoperative anesthetic consultation or documentations from the primary health care. Lowest Hb and highest creatinine values, included in routine laboratory testing within a week before surgery, and postoperative days 1–3, were registered. Intraoperative medical information was collected from anesthetic charts, including systolic blood pressure (SBP), heart rate, oxygen saturation, blood loss, and fluid balance. The predefined intraoperative events were hypotension (decrease in SBP relative to each patient's baseline >5 minutes), tachycardia (increase in heart rate to >110 beats per minute >5 minutes), blood loss (mL), hypoxemia (periferal oxygen saturation [SpO 2] < 90% >5 minutes), and cumulative fluid balance (mL). Intraoperative information in nonelectronic anesthetic charts, including BP, has previously been validated, detailed in Supplemental Digital Content 2, BP Validation,


The main exposure was IOH, defined as at least 1 event of an absolute decrease in SBP, from patient preoperative baseline, lasting >5 minutes. IOH was categorized into quartiles in accordance with incidence among controls; ≤20, 21–40, 41–50, or >50 mm Hg drop from individual baseline. Notably, IOH was further analyzed as an absolute threshold and as a relative decrease from baseline, detailed in statistics.


Acute MI, fulfilling the universal criteria,[17] subclassified as type 1 and 2, occurring within 30 days. Mortality beyond 30 days among case and control patients.

Statistical Analysis

Data were analyzed using STATA 14.2 (Stata Corp, College Station, TX). Continuous data are presented as medians with 25th–75th percentiles and categorical variables as percentages. For comparison of linear and categorical variables, Mann-Whitney U test or χ2 tests were used. Statistical tests are 2-sided, P values <.05 considered significant. Several descriptive analyses were performed; incidence of IOH at the 3 surgical sites and frequency of IOH in different surgical and anesthetic procedures were analyzed, presented in Supplemental Digital Content 3, Tables 3–6, Conditional logistic regression was used to assess associations between predefined risk covariates and perioperative MI. Confounding was, first and foremost, handled, by design, in the matching procedure. The controls were carefully selected and closely matched to the corresponding case, based on the strong risk factors of MI identified in the Orbit cohort study,[18] thus maximizing the possibility—power—to study potential confounding effect of intraoperative risk factors. In the analysis phase, confounding was further assessed using multivariable conditional logistic regression; preoperative, unmatched, risk factors; preoperative BP, DM, ischemic heart disease (IHD), and intraoperative risk factors; blood loss, low Hb value and fluid balance, were evaluated as potential confounders. Three definitions of the main exposure, IOH, were explored; relative to baseline (mm Hg), relative to baseline (%), and absolute intraoperative thresholds. All 3 definitions were subdivided into 4 categories, according to incidence among controls. The multivariable models yielded were compared using Akaike information criterion (AIC) test.

To illustrate the overall low-absolute risks, cases and controls were distributed to different risk strata, according to 5 risk groups created in the original Orbit cohort study:[18]

  1. Very low risk: age <65, ASA physical status I, low-risk surgery, no cardiovascular comorbidity, or DM.

  2. Low risk: same as group 1, but with 2 or 3 factors described in risk group 3 below.

  3. Medium risk: age 65–79, ASA physical status II, medium-risk surgery, cardiovascular comorbidity without previous MI, DM.

  4. High risk: same as group 3 but with 2 or 3 factors described in group 5 below.

  5. Very high risk: age ≥80, ASA physical status >II, high-risk surgery, cardiovascular comorbidity with previous MI.

The surgical risk groups, also obtained from the Orbit study, were low (endocrine, ear, nose, and throat [ENT], ophthalmic dental, breast, and gynecological surgery), medium (gastrointestinal [GI], neuro, urologic, orthopedic, and dermatologic surgery), and high (vascular and thoracic surgery). Absolute risks in these risk groups in relation to hypotensive events were calculated using absolute risks of MI in the Orbit study and the relative risks associated with IOH in this study. These calculations rely on the assumption that the estimated incidence of IOH events among our sampled controls corresponds to the incidence of IOH in the whole Orbit cohort; thus the estimated odds ratios (ORs) in this study apply to the source population.

Mortality among cases and controls from day 31 to 90 and day 91 to 365 was compared with stratified Cox regression, crude and adjusted hazard ratios (HRs) with 95% confidence interval (CI) are presented. The IOH-related risk of a fatal MI <30 days was analyzed with logistic regression. Controls were selected using cumulative incidence sampling; all controls were bound to be alive at 30 days; thus 30-day mortality difference between cases and controls could not be analyzed.

Sensitivity Analysis

We assessed effect modification by preoperative BP, risk group, MI day, and tachycardia. Internal stratified analyses of preoperative BP (<140 vs ≥140 mm Hg), risk group; low (1–3) versus high (4–5), postoperative day of MI diagnose (day 1–2 versus >day 2), and tachycardia (yes/no) were performed together with interaction tests.