Pain and Opioid Consumption and Mobilization After Surgery

Post Hoc Analysis of Two Randomized Trials

Eva Rivas, M.D.; Barak Cohen, M.D.; Xuan Pu, M.Sc.; Li Xiang, M.D.; Wael Saasouh, M.D.; Guangmei Mao, M.Sc.; Paul Minko, B.S.; Lauretta Mosteller, B.A.; Andrew Volio, M.D.; Kamal Maheshwari, M.D.; Daniel I. Sessler, M.D.; Alparslan Turan, M.D.


Anesthesiology. 2022;136(1):115-126. 

In This Article

Materials and Methods

Study Design

We conducted a retrospective analysis of data from two randomized trials: Effect of Intravenous Acetaminophen on Postoperative Hypoxemia After Abdominal Surgery: the FACTOR Randomized Clinical Trial (NCT02156154; Alparslan Turan; registered on June 5, 2014),[34] which evaluated the effect of intravenous acetaminophen on postoperative opioid-related complications after colorectal surgery; and Transversus Abdominis Plane Block with Liposomal Bupivacaine versus Continuous Epidural Analgesia for Major Abdominal Surgery: the EXPLANE clinical trial (NCT02996227; Alparslan Turan; registered on December 19, 2016), which compared the effect of continuous epidural analgesia and transversus abdominis plane blocks on postoperative analgesia and opioid consumption after abdominal surgery. Both trials enrolled patients who had abdominal surgery and used a continuous vital sign recording system that also captures mobilization information, and were approved by the Cleveland Clinic Institutional Review Board (IRB; Cleveland, Ohio). The current analysis was approved by the IRB with waived individual consent and was designed before completing the trial enrollment (Cleveland Clinic IRB No. 19–341; approval date March 19, 2019).

Patients were managed according to Cleveland Clinic enhanced recovery after surgery protocols. Orogastric tubes were removed before endotracheal extubation. Patients were encouraged to walk on the evening of surgery and were offered noncarbonated liquids ad libitum. On the first postoperative day, patients were encouraged to walk at least one round of the nursing floor (approximately 60 m) up to five times, to sit out of bed between walks, and to perform regular incentive spirometry. Liquids were allowed and solid food offered if tolerated. Intraoperative analgesia was provided with short-acting opioids, and intravenous patient-controlled analgesia was used during the postoperative period. Oral analgesia was started, and the Foley catheter was removed on postoperative day 1.[35–37]

Study Population

We included adult inpatients having elective open or laparoscopic abdominal surgery scheduled to last at least 2 h with general anesthesia at the Cleveland Clinic between February 2014 and September 2019 who participated in FACTOR or EXPLANE and had continuous postoperative activity monitoring. We excluded patients who had less than 12 h of continuous activity monitoring during the initial 48 postoperative hours, along with patients who had missing pain assessments or lacked important confounding variables.


Postoperative pain was recorded using the numerical rating scale, which is an 11-point Likert scale from 0 (no pain) to 10 (worst imaginable pain). Pain scores were recorded at least every 30 min while patients remained in the postanesthesia care unit, and at least every 4 h while hospitalized. All available pain scores during the monitoring period were collected from patients' electronic medical records, and time-weighted average pain score was calculated for each patient. Opioid consumption during the initial 48 postoperative hours was also collected from patients' electronic records and converted into milligram of intravenous morphine equivalents.[38,39]

Our primary outcome was duration of mobilization, defined as hours per monitoring day spent sitting or standing. Position and activity were continuously monitored and recorded at 15-s intervals from postanesthesia care unit admission until the earlier of 48 postoperative hours or hospital discharge using the ViSi Mobile monitoring system (Sotera Wireless, Inc., USA), which is cleared by the U.S. Food and Drug Administration (Silver Spring, Maryland) for noninvasive continuous vital sign monitoring.[40]

The ViSi Mobile monitoring system includes a three-axis accelerometer that characterizes patients' orientation and activity.[41] It captures posture status as upright 90 degrees, upright 45 degrees, supine, lying on the side, walking, and fallen. We defined mobilization as standing or sitting position, with standing defined "walking" posture and sitting as "upright 90 degrees" posture. When more than one posture was detected during a 15-s interval, a combined posture was recorded. For example, if a patient walked and sat upright 45° during a single 15-s interval, the posture would be recorded as upright 45 degrees and walking. Combined postures were considered as mobilization if one of the components was eligible. Neither patients nor clinicians had access to mobilization data.

The exploratory outcome was a composite of postoperative complications, including myocardial injury (defined as either a postoperative peak fourth-generation troponin T concentration 0.03 ng/ml or greater within the first 7 days after surgery, apparently of cardiac origin, or an International Classification of Diseases code for myocardial infarction);[42] stroke or transient ischemic attack; venous thromboembolism; pulmonary complications; and all-cause in-hospital mortality (appendix, Table A1). Data were extracted from patients' electronic medical records, the anesthesia record-keeping system, and pharmacy records.

Statistical Methods

Gaps in activity/posture monitoring were removed and subtracted from total monitoring time, so that each patient's mobilization was calculated as hours of mobility per day by multiplying average mobilization minutes per monitoring hour by 24 h/day. Around 4% of all postures were recorded as unknown and were treated as missing and removed. Demographic and clinical characteristics of patients were summarized and presented by time-weighted average pain score category in Table 1, only for presentation purposes.

For the primary analysis, a quantile regression model was used to assess the association between time-weighted average pain score and mobilization time, in hours per day. A similar model was used to secondarily assess the association between opioid consumption during the first 48 postoperative hours and mobilization time. The associations were adjusted for all demographics and surgical variables in Table 1.

We conducted four sensitivity analyses. The first evaluated associations between time-weighted average pain scores and opioid consumption and postoperative mobilization, restricted to patients older than 65 yr. Second, we restricted the definition of mobilization to include only standing position. Third, we restricted the analysis to daytime, defined as 7 PM. For our final sensitivity analysis, we restricted the analysis to the first 24 postoperative hours.

For our exploratory analysis, the incidence of a composite of myocardial injury, stroke or transient ischemic attack, venous thromboembolism, pulmonary complications, and all-cause in-hospital mortality was summarized by quartiles of mobilization time.

We did two post hoc analyses. First, we fitted a quantile regression model with interaction between surgical approach (open or laparoscopic) and both exposures (pain and opioid consumption) with mobilization time as the outcome; and second, we explored the relationship between the mobilization time in hours per day and the composite of postoperative complications through a logistic regression model adjusting for age, sex, race, and surgery duration.

We performed a complete case analysis, as participants with missing data were excluded. Data were assumed to be missing at random. The significance level was 0.025 for each association (pain and mobilization/opioid consumption and mobilization) after Bonferroni correction for the primary and secondary analyses. All analyses were conducted using the Statistical Analysis System (SAS) statistical software package, version 9.04.01 (SAS Institute Inc., USA).

Power Consideration

We did a simulation using quantile regression to calculate the effect size we could detect. With the current sample size, we had more than 90% power for detecting a 0.2-h reduction in mobilization per day for each unit increase in pain score, or to detect a 0.13-h reduction in mobilization per day for each doubling of morphine consumption at a significance level of 0.025.