TRACE (Routine posTsuRgical Anesthesia visit to improve patient outComE): a prospective, multicenter, stepped-wedge, cluster-randomized interventional study

The TRACE study is a multicenter prospective interventional study with a stepped-wedge design, aiming to recruit 5600 patients. In the beginning of the study all participating medical centers will provide perioperative care using the standard approach. All patients included during the standard approach period will be enrolled in the control group. Stepwise and at equal time periods, interventional postoperative anesthesia care will be implemented in the participating centers. The order in which the centers start with the intervention is randomized. The additional anesthesia care will consist of anesthesia visits on postoperative days 1 and 3 to follow up patient care and enhance rescue therapy when needed. By the end of the study, all participating centers will have implemented the intervention.

Figure 1 shows the planned study schedule. Total recruitment time covers 31 weeks. All hospitals start in week 1 in the control phase and after week 8 sequentially cross over to the intervention phase. Between the two phases there is a one-week transition period (depicted with an x). In week 24, all hospitals will have started the intervention phase.

In total, 2800 patients will be recruited in the control phase and 2800 in the intervention phase. Each hospital will recruit 700 patients. The first hospitals to cross over (A–D) will recruit more patients in the intervention phase than in the control phase and the last hospitals (E–H) will recruit more patients in the control phase than in the intervention phase.

Randomization for the order in which hospitals cross over from the control phase to the intervention phase was performed before the first patient was recruited. The names of the eight hospitals were ordered randomly using envelope drawing. This randomized order was translated into the schedule presented in Fig. 1.

Patients will be recruited by a member of study team (anesthesiologist or research assistant) preoperatively, either during the preoperative screening or directly after hospital admission. Patients receive a written patient information letter and are additionally orally informed about the study aims and the study phase they will enter (control or intervention). If they agree to participate, they will be asked to sign informed consent.

During the intervention phase, all participants will receive a postoperative visit from the anesthesiologist at postoperative days 1 and 3. The postoperative visit is standardized and based on the Modified Early Warning Score (MEWS). The MEWS includes the following measurements: respiratory rate; heart rate and rhythm; systemic oxygen saturation; systolic blood pressure; body temperature; level of consciousness; and urine output. Other measurements include: visual analogue score to assess pain during rest and movement; nausea/vomiting; defecation; and mobilization. In case of a higher MEWS score and/or other concerns, the anesthesia health professional will advise the treating physician on further diagnosis, treatment, and follow-up of the patient.

During the control phase, participants will receive standard care, without a standardized visit from the anesthesiologist. The treating physician may ask the anesthesiologist in consultation when deemed necessary. Standard of care may slightly differ across the different participating hospitals.

The study will be performed in eight Dutch hospitals, representing general hospitals, tertiary referral hospitals, and academic centers. All participating hospitals received approval from the Board of Directors to participate in the TRACE study.

Patient and clinical data will be collected at several time-points during the study period: at inclusion (baseline); intraoperatively; postoperatively at days 1 and 3; during hospital stay; from hospital discharge until 30 days; and until 12 months. Data records will be coded and transcribed by local investigators into an Internet-based electronic case record form.

Data to be collected from patient record files include patient baseline characteristics, data on surgery and anesthesia, intraoperative adverse events, the postoperative clinical course, postoperative in-hospital adverse events, and post-discharge events measured at 30 days and at 12 months after the date of surgery. During the intervention period, data collection additionally includes anesthesia interventions on postoperative days 1 and 3.

Data will also be collected from patient questionnaires, completed at inclusion, seven days, 30 days, and 12 months postoperatively. The questionnaires include questions on quality of life (EuroQol Dutch EQ-5D-5 L), pain scores (numeric rating scale, NRS), functional recovery (Functional Recovery Index), and expected/perceived recovery (Global Surgery Recovery index). See Fig. 2 for an overview of enrolment, intervention, and assessments.

The primary outcome measure will be the mortality at 30 days after surgery. Secondary outcome measures will be incidence of postoperative complications (in eight domains—infectious, cardiac/transfusion, pulmonary, venous thromboembolic, renal, neurological, surgical, and other—in line with Meguid et al. [14]), time to implement rescue therapy, length of stay in hospital, length of stay in a high-care unit, quality of recovery 30 days after surgery, and postoperative quality of life up to one year following surgery.

The calculation is based on the primary outcome parameter 30-day mortality. In an earlier Dutch study, postoperative all-cause death was 1.85% [15]. Recently, the EUSOS study observed a mortality rate of approximately 2% [16], which is in line with previous results.

We considered a difference of 50% between monitored and non-monitored patients to be clinically relevant. In order to detect a reduction from 2% to 1% in postoperative 30-day mortality, with an alpha of 0.05 and a power of 80%, and a 1:1 ratio between control and intervention group, a total of 4638 patients have to be included in a parallel randomized controlled trial (RCT) design. To compensate for inter-institutional variation, a possible time-effect inherent to the stepped-wedge design, and drop-out, we increased this sample size with 20%. The relatively uncommon approach of applying a 20% inflation factor was discussed extensively during the grant application and ethical approval processes and was finally approved by both the Netherlands Organisation for Health Research and Development and the Medical Ethical Commitee.

Thus, we estimate a requirement of 5600 patients. With eight participating hospitals, this translates into 700 patients per hospital. Figure 1 illustrates how these patients will be divided across hospitals and between the control and intervention groups.

Primary and secondary outcome measures (proportions with 95% confidence intervals or means with standard deviations) will be calculated and presented in tables. A patient flow chart will be presented to account for inclusions per hospital and drop-out.

For the primary outcome and binary secondary outcomes, multi-level mixed effect logistic regression with random time by cluster interaction will be used to compare between the cohort that has not and the cohort that has received the intervention according to the stepped-wedge schedule. Time effects will be modelled according to appropriate methods, e.g. using spline functions. For continuous secondary outcomes, multi-level mixed effect linear regression will be used and for countable secondary outcomes, multi-level mixed effect Poisson regression models will be used. Analyses will be performed using SPSS and R.

On cluster level, per-protocol analyses will be performed (i.e. if the intervention phase starts later than expected in a certain hospital, patients will be allocated to the control group until the hospital has actually crossed over). Once the intervention phase has started, individuals within that cluster will be analyzed according to the intention-to-treat principle. Adjustments will be made for temporal trends and for clusters within hospitals [17]. An economic evaluation from a hospital perspective will be performed for postoperative anesthesia follow-up of high risk surgical patients compared to standard postoperative care. The time horizon of the economic evaluation will be 30 days and one year following surgery. For reasons of efficiency, the analysis after one year will be performed in a subset of study participants. A cost-effectiveness analysis will be performed, based on the incremental costs per postsurgical visit. Standard sensitivity analysis will be performed to test whether results are robust for changing certain parameters, e.g. cost-prices. Confidence intervals surrounding the mean differential costs will be calculated by the bootstrap method. The latter will also be used to quantify the uncertainty surrounding the incremental cost-effectiveness ratio [18, 19]. Additionally, a budget impact analysis will be performed to assess the affordability of a routine postoperative visit by an anesthesia healthcare professional by relating the increase in cost due to investment in anesthesia professionals to the cost-savings due to the expected reduction in complications and hence length of hospitalization and mortality.

At the time of initial manuscript submission, recruitment had started (November 2016) but has not been completed. We expect recruitment to be completed in September 2018. The current protocol version is version 5 (3 January 2018).