Advertisement

Cost-Effectiveness in Perioperative Care: Application of Markov Modeling to Pathways of Perioperative Care

  • Guy L. Ludbrook
    Correspondence
    Correspondence: Guy Ludbrook, MBBS, PhD, MSc, Royal Adelaide Hospital, 3G 395, 1 North Terrace, Adelaide, South Australia, Australia 5000.
    Affiliations
    Department of Anaesthesia, Royal Adelaide Hospital and Discipline of Acute Care Medicine, University of Adelaide, Adelaide, South Australia, Australia
    Search for articles by this author
  • Esrom Leaman
    Affiliations
    Department of Anaesthesia, Royal Adelaide Hospital and Discipline of Acute Care Medicine, University of Adelaide, Adelaide, South Australia, Australia
    Search for articles by this author
Open AccessPublished:September 10, 2021DOI:https://doi.org/10.1016/j.jval.2021.07.018

      Highlights

      • There are very limited data on formal cost-effective modeling of various pathways or model of care in perioperative medicine, despite an unmet clinical need for high value care in a cost-constrained environment.
      • This article demonstrates how application of Markov modeling to pathways of perioperative care, using data on outcome and cost from 2 published clinical trials, can be used to guide selection of high value options. Furthermore, it demonstrates the importance of inclusion of this type of approach in future clinical trials of innovations in perioperative care delivery.

      Abstract

      Objectives

      This study aimed to evaluate the application of cost-effectiveness modeling to redesign of perioperative care pathways, from a hospital perspective.

      Methods

      A Markov cost-effectiveness model of patient transition between care locations, each with different characteristics and cost, was developed. Inputs were derived from clinical trials piloting a preoperative call center and a postoperative medium-acuity care unit. The effect chosen was days at home (DAH) after surgery, reflecting quality of in-hospital care, acknowledged financially by fundholders, and relevant to consumers. Cost was from the hospital’s perspective. A model cycle time of 4 hours for 30 days reflected relevant timelines and costs.

      Results

      A Markov model was successfully created, accounting for the care locations in the 2 pathways as model states and accounting for consequences and costs. Cost-effectiveness analysis allowed the calculation of an incremental cost-effectiveness ratio comparing these pathways, providing a mean incremental cost-effectiveness ratio of −$427 per additional DAH, where incremental costs and DAH were −$644 and +1.51, respectively. Probabilistic sensitivity analysis suggested the new pathway had a 61% probability of reduced costs and a 74% probability of increased DAH and a 58% probability this pathway was dominant. Tornado analysis revealed the major contributor to increased costs as intensive care unit stay and the major contributor to decreased costs as ward stay. For the new pathway, the probability of transfer from ward to home and the probability of staying at home had the greatest impact on DAH.

      Conclusions

      These data suggest Markov modeling may be a useful tool for the cost-effectiveness analysis of initiatives in perioperative care.

      Keywords

      Introduction

       Addressing Postoperative Complications

      Perioperative care encompasses patient management before, during, and after surgery, with the elements of care at each stage important to optimize patient outcomes and minimize the risk of complications after surgery. Postoperative complications have substantial consequences to patients’ recovery and wellbeing and to healthcare resources. Complications prolong hospital stay,
      • Swart M.
      • Carlisle J.B.
      • Goddard J.
      Using predicted 30 day mortality to plan postoperative colorectal surgery care: a cohort study.
      increase the risk of delayed readmission to hospital after discharge,
      • Kassin M.T.
      • Owen R.M.
      • Perez S.D.
      • et al.
      Risk factors for 30-day hospital readmission among general surgery patients.
      and increase hospital costs by 50% to 100%.
      • Healy M.A.
      • Mullard A.J.
      • Campbell Jr., D.A.
      • Dimick J.B.
      Hospital and payer costs associated with surgical complications.
      Major postoperative complications are common
      • Nepogodiev D.
      • Martin J.
      • Biccard B.
      • Makupe A.
      • Bhangu A.
      National Institute for Health Research Global Health Research Unit on Global Surgery
      Global burden of postoperative death.
      International Surgical Outcomes Study Group
      Global patient outcomes after elective surgery: prospective cohort study in 27 low-, middle- and high-income countries [published correction appears in Br J Anaesth. 2017;119(3):553].
      • Grocott M.P.
      • Browne J.P.
      • Van der Meulen J.
      • et al.
      The postoperative morbidity survey was validated and used to describe morbidity after major surgery.
      and likely to increase substantially in many countries because of aging comorbid populations, in particular,
      • Ludbrook G.L.
      • Walsh R.M.
      The impact of age on the future burden of postoperative complications in Australia.
      which will lead to unsustainable hospital cost increases unless addressed.
      Innovations in perioperative care to reduce complications include specific therapeutic interventions
      • Myles P.S.
      • Bellomo R.
      • Corcoran T.
      • et al.
      Restrictive versus liberal fluid therapy for major abdominal surgery.
      and broader “bundled” interventions, or models or systems of care, which combine high quality elements. Examples of the latter include prehabilitation programs
      • Barberan-Garcia A.
      • Ubré M.
      • Roca J.
      • et al.
      Personalised prehabilitation in high-risk patients undergoing elective major abdominal surgery: a randomized blinded controlled trial.
      (before surgery) and Enhanced Recovery After Surgery
      • Zhang X.
      • Yang J.
      • Chen X.
      • Du L.
      • Li K.
      • Zhou Y.
      Enhanced Recovery After Surgery on multiple clinical outcomes: umbrella review of systematic reviews and meta-analyses.
      (after surgery). These new models or systems of care may provide a greater opportunity than specific interventions for measurable benefits.
      • Ludbrook G.
      Hidden pandemic of postoperative complications-time to turn our focus to health systems analysis.
      These benefits, from a hospital perspective, include “downstream” effects on patient length of hospital stay during the initial admission and postdischarge readmission rates,
      • Kassin M.T.
      • Owen R.M.
      • Perez S.D.
      • et al.
      Risk factors for 30-day hospital readmission among general surgery patients.
      ,
      • Barberan-Garcia A.
      • Ubré M.
      • Roca J.
      • et al.
      Personalised prehabilitation in high-risk patients undergoing elective major abdominal surgery: a randomized blinded controlled trial.
      ,
      • Kohlnhofer B.M.
      • Tevis S.E.
      • Weber S.M.
      • Kennedy G.D.
      Multiple complications and short length of stay are associated with postoperative readmissions.
      both associated with increased hospital costs. These can be captured in a single composite measure, overall days at home (DAH) after surgery out to 30 days after the date of surgery,
      • Myles P.S.
      • Shulman M.A.
      • Heritier S.
      • et al.
      Validation of days at home as an outcome measure after surgery: a prospective cohort study in Australia.
      ,
      • Bell M.
      • Eriksson L.I.
      • Svensson T.
      • et al.
      Days at home after surgery: an integrated and efficient outcome measure for clinical trials and quality assurance.
      which is a useful endpoint for hospitals for cost-effectiveness analysis (CEA) of changes in perioperative care because of its major implications for hospital costs.

       Cost-Effectiveness

      CEA to allow fundholders to identify high value care is a high priority in an era of increasingly strained healthcare budgets,
      • Ludbrook G.
      • Riedel B.
      • Martin D.
      • Williams H.
      Improving outcomes after surgery: a roadmap for delivering the value proposition in perioperative care.
      and the future challenges around cost and outcome in perioperative care have been outlined earlier. Nevertheless, formal cost-effectiveness calculations are relatively uncommon in the analysis of innovations in perioperative systems or models of care.
      • Zhang X.
      • Yang J.
      • Chen X.
      • Du L.
      • Li K.
      • Zhou Y.
      Enhanced Recovery After Surgery on multiple clinical outcomes: umbrella review of systematic reviews and meta-analyses.
      ,
      • Nunns M.
      • Shaw L.
      • Briscoe S.
      • et al.
      Multicomponent Hospital-Led Interventions to Reduce Hospital Stay for Older Adults Following Elective Surgery: A Systematic Review.
      Data from clinical trials from one hospital from a new model of care for preoperative management
      • Ludbrook G.
      • Seglenieks R.
      • Osborn S.
      • Grant C.
      A call center and extended checklist for pre-screening elective surgical patients - a pilot study.
      and from a new model of postoperative care
      • Ludbrook G.
      • Lloyd C.
      • Story D.
      • et al.
      The effect of advanced recovery room care on postoperative outcomes in moderate-risk surgical patients: a multicentre feasibility study.
      provided an opportunity to examine the potential to use formal cost-effectiveness modeling to estimate hospital benefits from systems change in perioperative care. Although Markov cost-effectiveness modeling is common for assessment of new specific therapies, such as cancer drugs,
      • Garrison L.P. J.r.
      • Babigumira J.
      • Tournier C.
      • Goertz H.P.
      • Lubinga S.J.
      • Perez E.A.
      Cost-effectiveness analysis of pertuzumab with trastuzumab and chemotherapy compared to trastuzumab and chemotherapy in the adjuvant treatment of HER2-positive breast cancer in the United States. [published correction appears in Value Health. 2019;22(7):843].
      and has been successfully applied to specific perioperative treatments,
      • Sceats L.A.
      • Ku S.
      • Coughran A.
      • et al.
      Operative versus nonoperative management of appendicitis: a long-term cost effectiveness analysis.
      ,
      • Casele H.
      • Grobman W.A.
      Cost-effectiveness of thromboprophylaxis with intermittent pneumatic compression at cesarean delivery.
      it is less commonly applied to healthcare systems.
      Hence, a transition cost-effectiveness Markov model of recovery after surgery was developed using locations of care as model states and with model inputs derived from these 2 trials. The general aim was to determine whether the data from clinical trials of health system improvement were adequate to allow development of a Markov model to be used for cost-effectiveness evaluation of these perioperative health system changes. The hypothesis was that creating a cost-effectiveness model was feasible and could provide information relevant to hospital decision makers when choosing to implement new models of care.

      Methods

       Clinical Trials

      Two prospective clinical trials reported on new perioperative models of care from a large tertiary public hospital, The Royal Adelaide Hospital. These trials were approved by the Queen Elizabeth Hospital Human Research Ethics Committee and prospectively registered (ANZCTRN12614000199617; ANZCTRN 2617001173381). The first trial analyzed the impact of a model of care incorporating: call center-based assessment assisted by computer-supported smart patient questioning and triage to preoperative workup remotely or in-hospital outpatient clinics. It compared this call center approach with usual care of face-to-face outpatient clinic assessment. It found no change in the quality of assessment, but a substantial (50%) decrease in-hospital costs.
      • Ludbrook G.
      • Seglenieks R.
      • Osborn S.
      • Grant C.
      A call center and extended checklist for pre-screening elective surgical patients - a pilot study.
      The second trial examined the feasibility and impact of overnight postoperative care of medium risk patients (predicted 30-day mortality of 1%-4%) in an advanced recovery room care (ARRC) unit instead of a general surgical ward.
      • Ludbrook G.
      • Lloyd C.
      • Story D.
      • et al.
      The effect of advanced recovery room care on postoperative outcomes in moderate-risk surgical patients: a multicentre feasibility study.
      ARRC uses the care medium-acuity care provided by recovery rooms, or postanesthesia care units, such as high staff-to-patient ratios, specific medical and nursing skill mix, advanced patient monitoring, and access to a wider range of treatment such as vasopressor support. It extends care from a few hours to overnight (18-24 hours) and adds elements such as checklists, protocols, and guidelines to reduce unnecessary variation in care and targets specific physiological goals by the morning of postoperative day 1. This is in comparison with usual care, which is transfer to a general surgical ward, which includes low staff-to-patient ratios, general medical and nursing skill mix, limited monitoring, and limited treatment options. The trial found ARRC was feasible, and at The Royal Adelaide Hospital, there were signals of decreased complications on the ward after leaving ARRC, possibly reduced reoperation rates, and an 80% decrease in days spent in-hospital out to 90 days postoperatively, which was significant on post hoc analysis at P=.1. All these should be associated with reduced hospital costs.

       Model Structure

      The Markov model structure was based on a common general pathway of perioperative care
      • Kain Z.N.
      • Vakharia S.
      • Garson L.
      • et al.
      The perioperative surgical home as a future perioperative practice model.
      and included specific types of care locations commonly used in hospitals for care before and after surgery. These aimed to reflect patient flow as much as possible and included locations such as outpatient clinics, wards, intensive care units (ICUs), and recovery rooms, developed as model states, each with their own capacities, effects, and costs, which could be estimated from available data from the 2 clinical trials. The model is shown in Figure 1. Innovative elements to create a new clinical pathway to be specifically examined were (1) a preoperative call center as an alternative to a conventional resource-expensive face-to-face outpatient clinic and (2) a new postoperative medium care unit (ARRC), as an alternative to conventional recovery room followed by ward care. To account for out-of-hospital locations or health status, the states reflecting any out-of-hospital care were simplified as home or supported care. Rescue from the ward by ICU and readmission to hospital from home were accounted for by new states, ICU2, and Ward2, respectively, to account for the possibility that patients with complications responded differently to treatment.
      Figure thumbnail gr1
      Figure 1Markov model of patient disposition during perioperative care, showing model states and transition probabilities. White states are the new elements being added to usual care (a preoperative call center, and a medium-acuity postoperative care unit, ARRC) to create a new clinical pathway.
      ARRC indicates advanced recovery room care; ICU, intensive care unit.

       Model Inputs and Assumptions

      Model inputs were estimated from The Royal Adelaide Hospital data from (1) the call center trial and (2) the before and after ARRC trial. The model inputs and the evidence base to inform them are displayed in Appendix 1 in Supplemental Materials found at https://doi.org/10.1016/j.jval.2021.07.018. Costs in Australian dollars were from the perspective of the hospital, derived from The Royal Adelaide Hospital’s Finance Department’s lumped daily costs at each location broken down to costs of each 4 hours of care by dividing daily costs by 6, assuming these costs are constant over a 24-hour period, and are displayed in Appendix 1 in Supplemental Materials found at https://doi.org/10.1016/j.jval.2021.07.018. No attempt at this stage was made to unpack costs further into elements such as staff time and staff ratios, laboratory costs, or costs of emergency department presentations. Daily ward bed stay costs were assumed to be constant, although they may in practice decrease with length of stay.
      • Fine M.J.
      • Pratt H.M.
      • Obrosky D.S.
      • et al.
      Relation between length of hospital stay and costs of care for patients with community-acquired pneumonia.
      Because detailed data on use of supported care were not available, this state was retained in the structure to allow for future modeling, but excluded from this analysis by assuming transition probabilities of zero.

       Clinical Effectiveness

      The effect chosen was DAH after surgery out to 30 days. Data from the ARRC trial suggested a decreased readmission rate and a decreased duration of stay after readmission. Readmission rates are clearly associated with quality of in-hospital care,
      • Kassin M.T.
      • Owen R.M.
      • Perez S.D.
      • et al.
      Risk factors for 30-day hospital readmission among general surgery patients.
      ,
      • Kohlnhofer B.M.
      • Tevis S.E.
      • Weber S.M.
      • Kennedy G.D.
      Multiple complications and short length of stay are associated with postoperative readmissions.
      but the use of this parameter alone as an effect would markedly underestimate the potential impact of ARRC. Nevertheless, DAH account for readmission rates, length of stay, and duration of readmission and have been validated against quality of in-hospital care.
      • Myles P.S.
      • Shulman M.A.
      • Heritier S.
      • et al.
      Validation of days at home as an outcome measure after surgery: a prospective cohort study in Australia.
      ,
      • Bell M.
      • Eriksson L.I.
      • Svensson T.
      • et al.
      Days at home after surgery: an integrated and efficient outcome measure for clinical trials and quality assurance.
      This parameter has value to hospitals and also to consumers and hence was selected as the primary measure of effect.

       Clinical Effectiveness Modeling

      A model cycle time of 4 hours was chosen, reflecting a time period clinically relevant to decisions on patient movement between states. A model timeline of 30 days after surgery was selected, being the most common period for which readmissions and DAH are measured and acknowledged in many systems of reimbursement. Discounting of costs for such a short-term model is probably unnecessary, but was included at a nominal rate of 3% per annum to allow for future longer-term modeling. All patient numbers at each location (state) for each cycle were summed to ensure all patients were accounted for at each of the 180 cycles.

       Model Outcomes

      A deterministic analysis was initially performed, where point estimates informed DAH, costs, and incremental cost-effectiveness ratio (ICER). For sensitivity analysis (see also Appendix 1 in Supplemental Materials found at https://doi.org/10.1016/j.jval.2021.07.018), the deterministic model was converted to probabilistic model, adding uncertainty to model parameters. The Dirichlet distribution, approximated by gamma and normal distributions, was used for all events.
      • Briggs A.
      • Claxton K.
      • Sculpher M.
      Decision Modelling for Health Economic Evaluation.
      A random number generator selected probability weighted values within each distribution, and 10 000 simulations were run. The mean cost and effects and the mean ICER were then calculated for a new clinical pathway with the addition of the 2 new interventions (call center and ARRC) compared with usual care. The mean ICER was calculated as the ratio of mean incremental costs to mean incremental effectiveness. All individual ICERs from 10 000 simulations and the mean were then plotted as cost against effect (DAH). A sensitivity analysis, summarized in a tornado diagram, was performed to evaluate the sensitivity of both cost and effect to each parameter. Parameters were varied sequentially using their lower and upper bounds based on 95% confidence intervals (CIs).
      Microsoft Excel was used for all modeling, with spreadsheets adapted from those used in London School of Economics’ Masters in Health Economics, Policy and Management to analyze the effect of a new cancer therapy.
      • Garrison L.P. J.r.
      • Babigumira J.
      • Tournier C.
      • Goertz H.P.
      • Lubinga S.J.
      • Perez E.A.
      Cost-effectiveness analysis of pertuzumab with trastuzumab and chemotherapy compared to trastuzumab and chemotherapy in the adjuvant treatment of HER2-positive breast cancer in the United States. [published correction appears in Value Health. 2019;22(7):843].

      Results

      Data from the ARRC trial were adequate to allow estimation of transition probabilities between most states. Exceptions were patients who had experienced deterioration requiring ICU admission from the ward or hospital readmission. Hence, the transition probabilities for the states ICU2 and Home2 were assumed to be the same as for ICU and home. The sum of all patient numbers at each location (state) for all 180 4-hour cycles was 100, confirming the model was balanced.

       Patient Flow

      The output from the model in terms of patient movement between states, before and after inclusion of a preoperative call center and ARRC-containing postoperative care, is displayed in Figure 2. Patients followed 1 of 2 pathways for preoperative assessment and care. Postoperatively, time was initially spent in recovery, followed by transition to ARRC or the ward, and ultimately home. Time in Ward2 reflects readmissions from home. Time in ICU reflects planned transitions from either recovery or ARRC (because of delayed recovery by the morning of day 1), and time in ICU2 reflects unplanned “rescue” from the ward. Notable observations from the new clinical pathway were the appearance of ARRC (red), increased initial ICU referrals from ARRC (blue), reduced ICU “rescue” from the wards (gray), and faster transition to home (green).
      Figure thumbnail gr2
      Figure 2Model output, with patient transition between model states (locations) for usual care (left) and the new clinical pathway (right). Lower figures display an expanded scale to better display early patient transitions. Note, with the new clinical pathway, the appearance of ARRC (red), increased initial ICU referrals from ARRC (blue), reduced ICU “rescue” from the wards (gray), and faster transition to home (green).
      ARRC indicates advanced recovery room care; ICU, intensive care unit.

       Base-Case Results

      The key findings from ICER calculation from the deterministic model are presented in Table 1. This revealed, with the addition of a call center and ARRC in a cohort of 100 patients, (1) increased overall DAH (1.51), (2) decreased overall hospital cost (−$644), and (3) an ICER of −$427 per additional DAH.
      Table 1ICER calculations from the deterministic model.
      Usual careNew clinical pathwayDifferenceICER
      Days at homeCost (A$)Days at homeCost (A$)Days at homeCost (A$)
      19.839713.2221.349069.511.51−643.71−426.56
      Note. Costs and days at home after surgery before and after the introduction of the new clinical pathway and the estimated ICER (A$ per additional day at home).
      ICER indicates incremental cost-effectiveness ratio.

       Sensitivity Analysis

      For sensitivity analysis, the ICER results from 10 000 simulations of the probabilistic model based on randomly generated transition probabilities and costs are displayed in a scatter plot in Figure 3. The mean cost was −$665.62 (CI −5527.70 to +4196.46), and the mean DAH was 1.55 (CI −3.05 to +6.15). The mean ICER (mean of the means) was −$429.43, with CIs of $898.77 increased cost per reduced DAH to $1377.18 decreased cost per additional DAH.
      Figure thumbnail gr3
      Figure 3Base-case probabilistic sensitivity analysis; scatter plot of DAH versus cost displaying 10 000 iterations, and the mean ICER of the deterministic and the probabilistic models.
      ARRC indicates advanced recovery room care; DAH, day at home; ICER, incremental cost-effectiveness ratio.
      Most simulations demonstrated a cost reduction (60.9%), an effectiveness (DAH) increase (74.3%), and both an increase in DAH and reduced cost, the standard error quadrant of the cost-effectiveness plot (58.1%).

       Two-Way Deterministic Sensitivity Analysis

      The tornado diagram (Fig. 4) shows the results of the 2-way sensitivity analysis on costs and DAH. For the new clinical pathway compared with usual care, the ward stay costs contributed the most to decreased cost, and ICU stay to increased costs. For the new model, the transition probability of transfer from ward to home and the probability of staying at home had the greatest impact on DAH.
      Figure thumbnail gr4
      Figure 4Tornado plots displaying (A) left (major contributions of state costs to incremental costs from changing from usual care to the new clinical pathway) and (B) right (major contributions of transition probabilities to DAH for the new clinical pathway).
      cICU indicates cost of initial intensive care unit; cICU2, ICU2 cost; cPrehab, cost of prehabilitation; cRec, cost of recovery; cUsual, usual cost; cWard, ward cost; cWard2, Ward2 cost; DAH, day at home.

      Discussion

      The data presented here support the hypothesis that this type of modeling can feasibly be applied to analysis of new health systems or models of care. Importantly, it also addresses cost and outcomes relevant to the perspective of the hospitals (reducing hospital costs as a result of reduced complications) and those of consumers (uncomplicated recovery leading to more time at home). In addition, the sensitivity analysis summarized in the tornado plots in Figure 4 provides insight into future hospital strategic directions and where clinical improvements can be made to further enhance the clinical pathway, something not always obvious from simple business cases. For example, in relation to this specific model of care, the major contributions of ward costs and costs of initial ICU referral from ARRC suggest exploration of strategies to address these. This might be by providing more lower cost “hospital at home” beds to avoid discharge block
      • Bryan A.F.
      • Levine D.M.
      • Tsai T.C.
      Home Hospital for Surgery.
      or enhanced after-hours staffing to allow more active treatment overnight, thus reducing the need for (expensive) transfers to ICU after ARRC, which were seen in the pilot study.
      • Ludbrook G.
      • Lloyd C.
      • Story D.
      • et al.
      The effect of advanced recovery room care on postoperative outcomes in moderate-risk surgical patients: a multicentre feasibility study.
      In terms of DAH, the large contribution of length of stay, reflected in tpWard2Home, again emphasizes the need for good discharge planning, and the major contribution of reduced readmissions, reflected in tpHome2Home, emphasizes the need for good inpatient care to reduce readmissions.
      • Kassin M.T.
      • Owen R.M.
      • Perez S.D.
      • et al.
      Risk factors for 30-day hospital readmission among general surgery patients.
      As cost and resources become increasingly sparse, detailed CEA will become more critical to guide fundholders’ decisions on the cost-effectiveness of perioperative models of care or systems. Nevertheless, analyses are frequently relatively simple and sparse. For example, in a recent umbrella systematic review of a new postoperative clinical model (Enhanced Recovery After Surgery),
      • Zhang X.
      • Yang J.
      • Chen X.
      • Du L.
      • Li K.
      • Zhou Y.
      Enhanced Recovery After Surgery on multiple clinical outcomes: umbrella review of systematic reviews and meta-analyses.
      only 13 of more than 200 studies even mentioned cost as an outcome. Those studies that did address cost and effectiveness frequently including simple estimates or formal but simple cost-effectiveness calculations. A recent systematic review of clinical interventions or models of care in the perioperative space,
      • Nunns M.
      • Shaw L.
      • Briscoe S.
      • et al.
      Multicomponent Hospital-Led Interventions to Reduce Hospital Stay for Older Adults Following Elective Surgery: A Systematic Review.
      specifically preoperative workup and early enhanced postoperative care, provided the following conclusions: “Studies were usually of moderate or weak quality. Enhanced recovery and prehabilitation interventions were associated with reduced hospital stay without detriment to other clinical outcomes. The impacts on patient-reported outcomes, healthcare costs, or additional service use are not well known.” This identifies the need for better quality studies of clinical interventions, but also a need for robust CEA in perioperative care. This was a specific recommendation of a 2020 Australian National Summit on postoperative complications.
      • Ludbrook G.
      • Riedel B.
      • Martin D.
      • Williams H.
      Improving outcomes after surgery: a roadmap for delivering the value proposition in perioperative care.
      Formal cost-effectiveness modeling, including calculation of ICERs, has been performed in some analyses of perioperative care pathways.
      • Lee L.
      • Mata J.
      • Ghitulescu G.A.
      • et al.
      Cost-effectiveness of enhanced recovery versus conventional perioperative management for colorectal surgery.
      Markov modeling has been shown to be feasible in analysis of a postoperative care model to reduce repeat fractured neck of femur,
      • Beaupre L.A.
      • Lier D.
      • Smith C.
      • et al.
      A 3i hip fracture liaison service with nurse and physician co-management is cost-effective when implemented as a standard clinical program [published correction appears in Arch Osteoporos. 2020;15(1):136].
      finding an ICER, using QALYs as an effect, below the standard cost-effectiveness threshold, although this study did not examine in-hospital processes and pathways. It concluded: “…a cohort simulation CEA suggested that the H-FLS was cost-effective with potential to become cost-saving.” In addition, the benefits of this Markov modeling were around projections of future iterations of the model of care, suggesting that removing the 9-month postoperative visit and reducing the 6-month visit would be fertile areas for change and might result in the model actually becoming cost-saving.
      In the model used in the analysis presented here, DAH was used as the measure of effectiveness, because of evidence of its validity as a measure of quality of inpatient care and because it has financial relevance to hospitals. It aligns with readmission data, used as a measure of quality improvement in a recent study of preoperative prehabilitation,
      • Barberan-Garcia A.
      • Ubré M.
      • Roca J.
      • et al.
      Personalised prehabilitation in high-risk patients undergoing elective major abdominal surgery: a randomized blinded controlled trial.
      which also found evidence of cost benefit.
      • Barberan-Garcia A.
      • Ubre M.
      • Pascual-Argente N.
      • et al.
      Post-discharge impact and cost consequence analysis of prehabilitation in high-risk patients undergoing major abdominal surgery: secondary results from a randomised controlled trial.
      The use of more broadly relevant effect measures, such as quality-adjusted life-years (QALYs), was not feasible because of a paucity of data on longer-term outcomes, but may be included in the future if further data become available.
      There are several limitations in this current modeling. First, there is the uncertainty in the data from 2 small trials producing uncertainty in the ICER. This is also evident from the simulations in Figure 3, although this sensitivity analysis included very conservative substantial uncertainty in costs, which in reality are actually known with better certainty. Second, the parameters used in this model were derived from a single site and so the external validity of these findings is uncertain. Nevertheless, such models allow other jurisdictions to apply their own knowledge on outcomes and costs. Third, the choice of DAH for an effect limits the generalizability of this model across other clinical areas. Nevertheless, it may be feasible to add longer-term impacts, such as 90-day readmissions and QALYs, if data emerge supporting longer-term effects on QALY of early enhanced care, thus allowing cost-effectiveness comparisons with nonsurgical hospital activities. Fourth, this modeling does not include detailed analysis of out-of-hospital postoperative care (such as community care), preoperative care (such as prehabilitation), or detailed within-hospital costs such as consumables and laboratory or radiology testing.
      To address some of the abovementioned limitations of this specific application of Markov modeling to perioperative care, a powered prospective clinical trial at The Royal Adelaide Hospital (NCT04769518) has commenced to examine the cost and effects of the ARRC model at scale, with more detailed data collection on costs and outcomes and with DAH and cost-effectiveness as formal trial endpoints.
      The analysis presented here provides an example of the application of Markov cost-effectiveness modeling to clinical models or systems of care and its utility in estimating the benefits of alternative care models. Furthermore, the inclusion of this type of modeling and cost-effectiveness as an endpoint has potential merit in other future trials of perioperative care.

      Article and Author Information

      Author Contributions: Concept and design: Ludbrook
      Acquisition of data: Ludbrook
      Analysis and interpretation of data: Ludbrook, Leaman
      Drafting of the manuscript: Ludbrook, Leaman
      Critical revision of the paper for important intellectual content: Ludbrook, Leaman
      Statistical analysis: Ludbrook, Leaman
      Obtaining funding: Ludbrook
      Conflict of Interest Disclosures: Dr Ludbrook and Mr Leaman reported receiving competitive grants from the Australian and New Zealand College of Anaesthetists and The Royal Adelaide Hospital and internal resources from The University of Adelaide and Central Adelaide Local Health Network. Dr Ludbrook reported receiving part funding to attend London School of Economics that was provided by Central Adelaide Local Health Network and The University of Adelaide. Dr Ludbrook is an advisory board member for Edwards Life Sciences and director of a clinical trials unit, PARC Clinical Research.
      Funding/Support: The clinical trials on which the modeling is based were supported by competitive grants from the Australian and New Zealand College of Anaesthetists and The Royal Adelaide Hospital and internal resources from The University of Adelaide and Central Adelaide Local Health Network.
      Role of Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

      Acknowledgment

      The support of academic staff from the London School of Economics Health, Economics, Policy, and Management Program and the medical, nursing, and senior executive staff from The Royal Adelaide Hospital is gratefully acknowledged.

      Supplemental Material

      • Appendix 1

        Markov model of patient disposition after surgery—estimated transition probabilities every 4 hours, the basis of their derivation, costs from the perspective of the hospital of care in each state for 4 hours, and the basis for the probabilistic model.

      References

        • Swart M.
        • Carlisle J.B.
        • Goddard J.
        Using predicted 30 day mortality to plan postoperative colorectal surgery care: a cohort study.
        Br J Anaesth. 2017; 118: 100-104
        • Kassin M.T.
        • Owen R.M.
        • Perez S.D.
        • et al.
        Risk factors for 30-day hospital readmission among general surgery patients.
        J Am Coll Surg. 2012; 215: 322-330
        • Healy M.A.
        • Mullard A.J.
        • Campbell Jr., D.A.
        • Dimick J.B.
        Hospital and payer costs associated with surgical complications.
        JAMA Surg. 2016; 151: 823-830
        • Nepogodiev D.
        • Martin J.
        • Biccard B.
        • Makupe A.
        • Bhangu A.
        • National Institute for Health Research Global Health Research Unit on Global Surgery
        Global burden of postoperative death.
        Lancet. 2019; 393: 401
        • International Surgical Outcomes Study Group
        Global patient outcomes after elective surgery: prospective cohort study in 27 low-, middle- and high-income countries [published correction appears in Br J Anaesth. 2017;119(3):553].
        Br J Anaesth. 2016; 117: 601-609
        • Grocott M.P.
        • Browne J.P.
        • Van der Meulen J.
        • et al.
        The postoperative morbidity survey was validated and used to describe morbidity after major surgery.
        J Clin Epidemiol. 2007; 60: 919-928
        • Ludbrook G.L.
        • Walsh R.M.
        The impact of age on the future burden of postoperative complications in Australia.
        Anaesth Intensive Care. 2019; 47: 480-481
        • Myles P.S.
        • Bellomo R.
        • Corcoran T.
        • et al.
        Restrictive versus liberal fluid therapy for major abdominal surgery.
        N Engl J Med. 2018; 378: 2263-2274
        • Barberan-Garcia A.
        • Ubré M.
        • Roca J.
        • et al.
        Personalised prehabilitation in high-risk patients undergoing elective major abdominal surgery: a randomized blinded controlled trial.
        Ann Surg. 2018; 267: 50-56
        • Zhang X.
        • Yang J.
        • Chen X.
        • Du L.
        • Li K.
        • Zhou Y.
        Enhanced Recovery After Surgery on multiple clinical outcomes: umbrella review of systematic reviews and meta-analyses.
        Medicine (Baltimore). 2020; 99e20983
        • Ludbrook G.
        Hidden pandemic of postoperative complications-time to turn our focus to health systems analysis.
        Br J Anaesth. 2018; 121: 1190-1192
        • Kohlnhofer B.M.
        • Tevis S.E.
        • Weber S.M.
        • Kennedy G.D.
        Multiple complications and short length of stay are associated with postoperative readmissions.
        Am J Surg. 2014; 207: 449-456
        • Myles P.S.
        • Shulman M.A.
        • Heritier S.
        • et al.
        Validation of days at home as an outcome measure after surgery: a prospective cohort study in Australia.
        BMJ Open. 2017; 7e015828
        • Bell M.
        • Eriksson L.I.
        • Svensson T.
        • et al.
        Days at home after surgery: an integrated and efficient outcome measure for clinical trials and quality assurance.
        EClinicalmedicine. 2019; 11: 18-26
        • Ludbrook G.
        • Riedel B.
        • Martin D.
        • Williams H.
        Improving outcomes after surgery: a roadmap for delivering the value proposition in perioperative care.
        ANZ J Surg. 2021; 91: 225-228
        • Nunns M.
        • Shaw L.
        • Briscoe S.
        • et al.
        Multicomponent Hospital-Led Interventions to Reduce Hospital Stay for Older Adults Following Elective Surgery: A Systematic Review.
        NIHR Journals Library, Southampton, UK2019
        • Ludbrook G.
        • Seglenieks R.
        • Osborn S.
        • Grant C.
        A call center and extended checklist for pre-screening elective surgical patients - a pilot study.
        BMC Anesthesiol. 2015; 15: 77
        • Ludbrook G.
        • Lloyd C.
        • Story D.
        • et al.
        The effect of advanced recovery room care on postoperative outcomes in moderate-risk surgical patients: a multicentre feasibility study.
        Anaesthesia. 2021; 76: 480-488
        • Garrison L.P. J.r.
        • Babigumira J.
        • Tournier C.
        • Goertz H.P.
        • Lubinga S.J.
        • Perez E.A.
        Cost-effectiveness analysis of pertuzumab with trastuzumab and chemotherapy compared to trastuzumab and chemotherapy in the adjuvant treatment of HER2-positive breast cancer in the United States. [published correction appears in Value Health. 2019;22(7):843].
        Value Health. 2019; 22: 408-415
        • Sceats L.A.
        • Ku S.
        • Coughran A.
        • et al.
        Operative versus nonoperative management of appendicitis: a long-term cost effectiveness analysis.
        MDM Policy Pract. 2019; 4381468319866448
        • Casele H.
        • Grobman W.A.
        Cost-effectiveness of thromboprophylaxis with intermittent pneumatic compression at cesarean delivery.
        Obstet Gynecol. 2006; 108: 535-540
        • Kain Z.N.
        • Vakharia S.
        • Garson L.
        • et al.
        The perioperative surgical home as a future perioperative practice model.
        Anesth Analg. 2014; 118: 1126-1130
        • Fine M.J.
        • Pratt H.M.
        • Obrosky D.S.
        • et al.
        Relation between length of hospital stay and costs of care for patients with community-acquired pneumonia.
        Am J Med. 2000; 109: 378-385
        • Briggs A.
        • Claxton K.
        • Sculpher M.
        Decision Modelling for Health Economic Evaluation.
        1st ed. Oxford University Press, Oxford, UK2006
        • Bryan A.F.
        • Levine D.M.
        • Tsai T.C.
        Home Hospital for Surgery.
        JAMA Surg. 2021; 156: 679-680
        • Lee L.
        • Mata J.
        • Ghitulescu G.A.
        • et al.
        Cost-effectiveness of enhanced recovery versus conventional perioperative management for colorectal surgery.
        Ann Surg. 2015; 262: 1026-1033
        • Beaupre L.A.
        • Lier D.
        • Smith C.
        • et al.
        A 3i hip fracture liaison service with nurse and physician co-management is cost-effective when implemented as a standard clinical program [published correction appears in Arch Osteoporos. 2020;15(1):136].
        Arch Osteoporos. 2020; 15: 136
        • Barberan-Garcia A.
        • Ubre M.
        • Pascual-Argente N.
        • et al.
        Post-discharge impact and cost consequence analysis of prehabilitation in high-risk patients undergoing major abdominal surgery: secondary results from a randomised controlled trial.
        Br J Anaesth. 2019; 123: 450-456