Advanced Recovery Room Care II - Improved Recovery After Surgery

The problem: Surgery is essential for approximately 1/3 of healthcare needs, and its volume needs to double out to 2040. However, it has risks, with serious postoperative complications occurring frequently (5%-30% of patients) and postoperative death being the third most prevalent cause of death globally.

Complications result in increased patient suffering and mortality, late adverse events such as re-admission, and two to three-fold increases in resource use (cost). They can be considered 'the hidden pandemic' of modern healthcare and, because of an ageing comorbid population, are estimated to increase by 300% in Australia by 2027.

An Australian National Summit on Postoperative Complications (March 2020), chaired by G Ludbrook, acknowledged the importance of this problem. Its report identified the need for a Systems approach, with an evidence base, clear pathways of care matched to individual patients' risks and needs, techniques to minimize unnecessary practice variation, and relevant performance measures: outcomes (institution- and consumer-centred) and cost.

One solution: In 2017, an Advanced Recovery Room Care model (ARRC) was trialled at RAH. This utilized all the Summit Systems elements and Principles - early validated risk assessment (NSQIP), triage of moderate-risk patients (predicted 30-day mortality: 1-4%) to mid-level high-acuity postoperative care overnight (ARRC), appropriate staff and capacity, precise regular checklist-based assessment, specific physiological goals, repeated re-assessment and re-triage, formal handovers, and evaluation with relevant metrics. This trial revealed that: early serious adverse events were unexpectedly common (50% of patients); these mainly occurred for 12-18 hours postoperatively; and re-admissions were very common (34%).

Reduced complications and sequelae: ARRC resulted in fewer ward-based complications; and an 80% reduction in hospital re-admission days, p=0.1. A 10-bed unit should free up 4000 RAH bed days annually.

Overall, ARRC findings align with (i) UK findings in moderate-risk patients - removing high acuity care increased mortality and in-hospital days, (ii) RAH Health Roundtable data - moderate-risk patient groups with unplanned re-admission rates of 35%.

Iterative better risk prediction and triage: The data also suggested that by formal repeated re-evaluation of risk over time (in theatre and ARRC), based on adverse events and physiological parameters, the 50% of patients unlikely to have subsequent major complications can be identified, and triaged to lower-acuity ARRC care, further reducing costs. This aligns with RAH data showing early Recovery adverse events as a major predictor of subsequent postoperative complications, and large databases associating pre-operative, in-theatre and post-operative events with delayed complications.

Cost-effectiveness: Critically, ARRC appears cost-effective. Markov modelling of the ARRC data reveal improved outcome (Days at Home) and costs, with an Incremental Cost Effectiveness Ratio (ICER) of -$600. This aligns with Peter McCallum Cancer Centre's finding $1M annual savings from implementation of a 2-bed unit (unpublished data).

The primary hypothesis is that ARRC, through rapid detection and management of early serious postoperative adverse events, will reduce subsequent complications, thus increasing patient Days-at-Home at 30 and 90 days (DAH30/90).

The secondary hypotheses are: (i) that ARRC will be cost-effective, producing a negative Incremental Cost-Effectiveness Ratio (ICER) when compared to conventional treatment, and (ii) that continuous re-evaluation of individual patient risk during surgery and ARRC will allow accurate identification of the lower risk sub-group of patients unlikely to have adverse events overnight, allowing care to be downscaled earlier, this incrementally reducing cost and improving ICER.

This is a prospective unblinded pragmatic trial of ARRC versus conventional care. CALHN has committed to re-starting ARRC, but excellent data are required to confirm efficacy on outcomes and cost.

Cohort: Patients are moderate-risk patients (NSQIP-predicted 30-day mortality = 0.7-5%) undergoing elective or emergency Colorectal, Orthopaedic, Gynae-Oncology, Neurosurgery, and other surgeries over time, and scheduled for postoperative ward care.

System of Care: NSQIP risk is calculated early (elective surgery - outpatients; unscheduled surgery - on admission), to quantify risks and highlight and address modifiable risks. Allocation to conventional care or ARRC is pragmatically based on bed availability, thereby producing minimal selection bias in the ARRC pilot and the observational UK study by Swart et al. Intraoperative management will be consistent, guideline-based care, something successfully implemented at RAH previously.

ARRC will be the system piloted at RAH, located in close to the Recovery Room (PACU). Critical elements include regular checklist-based assessment, and defined physiological goals. ARRC involves multidisciplinary staff - anaesthesia, nursing, surgery, internal medicine, allied health.

This system includes all Summit report-recommended Principles for excellent high-value care, and its Implementation recommendations that one institution take the lead.

Data collection: Enabled throughout the perioperative journey by our EMR and our Theatre/Recovery e-record system (AIMS) and managed by the PARC Clinical Research staff who ran the ARRC pilot.

Data are an extension of the pilot. In brief, this includes: (i) Preoperatively: patient demographics / comorbidities; (ii) Intraoperatively: surgery type/duration, complications, vitals sign variation, adverse events. Advanced monitors (ANI antinociception, continuous blood pressure monitoring) aid development of the risk prediction algorithms, (iii) ARRC: vital signs, adverse events, checklist completion, MET-level events (iv) post-ARRC: ward adverse events, ICU transfers, re-operation, length of stay, re-admissions, days-at-home (30 and 90 days). Costs are from RAH Finance - ARRC, ICU, Ward, ED (re-admissions).

Data analysis Primary hypothesis: U Adelaide statisticians, who managed the ARRC pilot, will use linear regression to find the mean difference in days-at-home (DAH) at 30/90 days between the case and control groups. Power analysis from the ARRC pilot, conservatively assuming a DAH difference of 1.5 days, suggests a sample size of N=876 cases and N=438 controls (weighting 2:1). Interim analyses will be conducted to assess progress. Based on 16 patients/week, study duration is to be 82 weeks. With positive outcome trends, we expect ARRC will expand to 32 patients/week, potentially reducing study duration to 61 weeks.

Cost-effectiveness: The CIA will update the Markov model transition probabilities and ICER as data emerge. Sensitivity analysis determines uncertainty. Interim analysis will also be conducted at the half-way point. Lower costs emerging from better risk prediction (e.g. lower ARRC nursing ratios, earlier transfers from ARRC to the ward) will be incorporated.

Improved risk prediction: U Southampton Clinical Informatics Unit colleagues will use their ALEA platform. Predictive analytics expertise will be utilised to generate a hierarchical set of novel predictive algorithms (based on available data inputs categories) and start to create real-time clinical risk prediction dashboards. RAH data may ultimately be added to other pilot sites worldwide.