Economic Impact of Multidisciplinary Pre‐operative Diabetes Care Versus Usual Care in Adults Undergoing Elective Non‐cardiac Surgery in Australia: A Trial‐Based Analysis Authors: Abby Yu, Karen Whitfield, Andrew Hale, Michael D'Emden, Peter Donovan, Centaine Snoswell Published: 2026 ABSTRACT ABSTRACTAimsThe detrimental impact of perioperative hyperglycaemia on surgical outcomes is well documented. However, economic evaluations on interventions to improve perioperative blood glucose management are lacking. This study aimed to assess the cost‐effectiveness of a multidisciplinary pre‐operative diabetes service compared with usual care in adults with diabetes undergoing elective non‐cardiac surgery.Materials and MethodsAn economic evaluation was conducted using data from a 3‐year prospective interventional study involving adults with sub‐optimally managed diabetes who were scheduled for elective non‐cardiac surgery. Analysis was from a healthcare institution perspective. A decision analytic model was used to compare healthcare costs and outcomes between an intervention arm receiving multidisciplinary pre‐operative diabetes input and a simulated usual care arm. Model inputs were derived from the interventional study, National Surgical Quality Improvement Program (NSQIP) data and published literature. Univariate and probabilistic sensitivity analyses were performed to assess the robustness of the model outcome.ResultsThe intervention arm had a lower modelled incidence of 30‐day post‐operative infection‐related readmissions and lower total costs per patient compared with the usual care arm. The intervention avoided 7 readmissions per 100 patients treated and generated cost savings of A$3967 per patient (total cost intervention: A$8335 vs. total cost usual care: $12 302). Probabilistic sensitivity analysis revealed an 89.5% probability that the intervention would remain dominant, with lower costs and fewer post‐operative infection‐related readmissions than usual care.ConclusionsIn adults awaiting elective non‐cardiac surgery, a multidisciplinary pre‐operative diabetes service represents a cost‐effective strategy compared with usual care from a healthcare institution perspective. Introduction Effective perioperative management of diabetes is crucial for reducing the risk of surgery related complications and their associated healthcare burden. Perioperative hyperglycaemia is associated with increased risk of surgical site infections and other perioperative complications [1, 2, 3, 4]. Studies have reported an independent association between each 1% increase in pre‐operative haemoglobin A1c (HbA1c) and increase in major post‐operative complications, intensive care unit (ICU) admissions, acute kidney injury, surgical site infections and hospital length of stay [5, 6]. These perioperative complications incur significant financial costs for the healthcare sector and wider society. One Australian study reported the direct cost per case of surgical site infection was valued at 2018–2019 A$18 814, with the total direct cost across Australia over a 1‐year time horizon estimated at A$323.5 million [7]. This, together with indirect costs from absenteeism and premature death, was estimated to cost the Australian economy A$3.3 billion per annum [7]. Studies have also reported a strong association between perioperative hyperglycaemia and increased direct hospital costs following general surgeries and robotic prostatectomies [8, 9], ranging from US$1860 for robotic prostatectomies [9], to US$59 272 for general surgeries [8]. Improving perioperative diabetes management is crucial to reducing this economic burden. Studies in community settings show that multidisciplinary diabetes care leads to greater reductions in HbA1c, blood pressure, and lipid levels in people living with diabetes [10, 11], which are predicted to reduce diabetes‐related complications and generate substantial healthcare savings [11]. Integration of pharmacists into diabetes care teams further improves blood glucose levels, systolic blood pressure, and body mass index in people with type 2 diabetes and has been shown to be a cost‐effective intervention [12, 13, 14, 15, 16, 17]. Whilst the cost‐effectiveness of multidisciplinary diabetes care in the general population is well established [16, 18], its impact in the pre‐operative setting remains unexplored. Given the high volume of people with diabetes undergoing non‐cardiac surgery each year, determining whether multidisciplinary pre‐operative optimisation can reduce downstream healthcare costs is clinically and economically important. Therefore, the aim of this study was to assess the cost‐effectiveness of a multidisciplinary pre‐operative diabetes service compared with usual care in adults with diabetes undergoing elective non‐cardiac surgery. Methods Study Design and Setting An economic evaluation was constructed using data from a 3‐year prospective single‐centre interventional study involving adults with sub‐optimally managed diabetes (defined as pre‐operative HbA1c levels ≥ 8.0% (64 mmol/mol)) who were scheduled for elective non‐cardiac surgery at a large metropolitan teaching hospital in Brisbane, Australia [19]. Details of the study design and results have been published previously [19]. Briefly, participants referred to a multidisciplinary pre‐operative diabetes service received regular consultations with a pharmacist and diabetes nurse who provided ongoing monitoring, adjustment of medications and treatment advice under the direction of a diabetologist. Freestyle Libre continuous glucose monitoring (CGM) sensors (Abbott Diabetes Care Inc., Alameda, CA, USA) were used to assist participants with glucose monitoring and enable accurate titration of pharmacotherapy. Participants were also reviewed by a diabetologist if additional pharmacotherapy was indicated. HbA1c and serum fructosamine levels were measured at baseline and on the morning of surgery to assess glycaemic changes from the intervention. Participant Characteristics In total, 79 participants completed the interventional study. The baseline characteristics of the cohort and types of surgery performed were previously published [19] and are reproduced in Tables S1 and S2, respectively. Economic Model As the study had no control arm due to ethical concerns in excluding people with poorly managed diabetes access to specialist care, a decision analytic model built in Microsoft Excel 2026 was used to assess the cost‐effectiveness of the intervention compared with a simulated usual care arm (Figure 1). At the institution where the original study was conducted, usual care involves routine diabetes management by the patient's general practitioner (GP) or specialist diabetologist at their usual designated review interval. Following the decision to proceed with surgery, patients would be seen in the preadmission clinic, whereby if an elevated pre‐operative HbA1c or random glucose level was detected, patients would be advised to follow up with their GP for further evaluation and treatment. No further action from the hospital would be undertaken to review if HbA1c or glucose levels had improved before surgery. Decision model diagram. p, probability; SSI, surgical site infection. In addition to published literature [20, 21, 22, 23], data from the original interventional study and deidentified National Surgical Quality Improvement Program (NSQIP) aggregate outcome data for surgical patients with diabetes in Queensland, Australia [24], were used to inform the decision‐tree probabilities and cost and effectiveness outcomes for the intervention and usual care arms, respectively (see Table 1 for model parameters). The common components of the model were patients living with diabetes undergoing elective non‐cardiac surgery. The two arms then modelled either receiving care through a multidisciplinary pre‐operative diabetes service or usual care. Patients in each arm then had varying probabilities of developing a 30‐day post‐operative infection requiring hospital readmission. For the usual care arm, event probabilities were estimated from the observed incidence of infection‐related readmissions in NSQIP data collected between 1 January 2025 and 31 December 2025 available from five hospital sites (n = 686 patients; two metropolitan and three regional hospitals from Queensland, Australia). Mean length of stay was derived from NSQIP data available from seven hospital sites (n = 835 patients; three metro and four regional hospitals) and was used to estimate inpatient admission costs for the usual care arm [24]. To ensure that the two arms were comparable in terms of surgery type, data for ophthalmology surgery patients in the intervention cohort were excluded from the analysis as these surgical procedures are not captured by NSQIP and are generally day case or short‐stay admissions. TABLE 1Model parameters.ParameterBase‐caseRange for univariate sensitivity analysisPSA distribution parametersSourceCost, A$Deep surgical site infection11 740(4210, 12 135) γ (35, 334)[19, 23, 25]Organ space infection12 135(5019, 22 288) γ (8, 1596)[7, 23, 25]Urinary tract infection3311(1300, 10 542) γ (2, 1613)[19, 23, 25]Pneumonia9401(4776, 11 542) γ (31, 304)[23, 25]Sepsis16 808(8117, 35 682) γ (6, 2825)[25]Length of stay cost, intervention5423 c (1808, 7136) γ (14, 377)[19]Length of stay cost, usual care11 397(7385, 14 770) γ (4, 2626)[24, 25]Mean length of stay, intervention3.8 days(1, 5)[19]Mean length of stay, usual care4.6 days(3, 6)[24]Event probabilitiesDeep surgical site infection, intervention0.013(0.005, 0.03) β (4, 323)[19, 20, 21]Organ space infection, intervention0 a (0, 0.04) β (1, 197)[19, 20, 21]Urinary tract infection, intervention0.013(0.008, 0.03) β (5, 407)[19, 20, 21]Pneumonia, intervention0 a (0, 0.015) β (2, 403)[19, 20, 21]Sepsis, intervention0 b (0, 0.008) β (2, 745)[19, 20, 21, 22]Deep surgical site infection, usual care0.016(0.0054, 0.0506) β (2, 120)[24]Organ space infection, usual care0.032(0.0109, 0.0759) β (4, 113)[24]Urinary tract infection, usual care0.037(0.012, 0.05) β (15, 379)[24]Pneumonia, usual care0.0087(0.0049, 0.0154) β (11, 1240)[24]Sepsis, usual care0.0073(0, 0.0127) β (5, 713)[24]Abbreviation: PSA, probabilistic sensitivity analysis. a For PSA, 0.005 was used in place of the base‐case value of 0, which could not be parameterised in the assigned probability distribution. b For PSA, 0.003 was used in place of the base‐case value of 0, which could not be parameterised in the assigned probability distribution. c Hospital clinical costing team. Model parameters. Abbreviation: PSA, probabilistic sensitivity analysis. For PSA, 0.005 was used in place of the base‐case value of 0, which could not be parameterised in the assigned probability distribution. For PSA, 0.003 was used in place of the base‐case value of 0, which could not be parameterised in the assigned probability distribution. Hospital clinical costing team. Perspective The economic evaluation was conducted from a healthcare institution perspective according to the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) statement (Figure S1) [26]. Cost‐effectiveness was assessed using the incremental cost‐effectiveness ratio (ICER), defined as the incremental cost per 30‐day post‐operative infection‐related readmissions avoided. Costs of the intervention were estimated over 1 year to capture the costs incurred for the majority of patients receiving the service. No additional costs for the usual care arm were incurred, other than hospital admission and readmission costs. No discounting was applied as costs were incurred over a 12‐month period. Costing Methods Only direct costs associated with operating the multidisciplinary service and healthcare utilisation costs for both the intervention and usual care arms were included in the economic model. Indirect and non‐medical costs, such as patient travel, were not included in the analysis. Costs were reported in 2022 Australian dollars (AUD) and converted as needed [27]. Total costs for the intervention arm included direct costs of operating the multidisciplinary service plus healthcare utilisation costs incurred by participants. Total costs for the usual care arm included healthcare utilisation costs only. Costs were estimated by applying the unit costs to the number of resources used. Direct service operating costs included costs to employ a diabetologist, pharmacist, diabetes nurse and administration officer, infrastructure costs and costs associated with pathology testing and purchase of continuous glucose monitoring (CGM) sensors. Table S3 provides a summary of how costs were calculated, and Table 2 provides a summary of direct operating costs and sources of costs. TABLE 2Resource costs per participant for the intervention arm.ResourceMean cost (A$)Standard deviation (SD)Sensitivity valuesMinimumMaximumEndocrinologist consultation (0.1 FTE)544.58 a —408.44680.73Diabetes nurse educator consultation (0.4 FTE)737.49 a —553.12921.86Pharmacist consultation (0.4 FTE)853.77 a —640.331067.21Administration support (0.2 FTE)255.75 a —191.81319.68Infrastructure costs12.87 a ———HbA1c test48.69 b 29.8120.6887.72Haemoglobin test30.24 b 18.8113.0952.36Serum fructosamine test68.63 b 13.1846.5493.08Albumin test31.08 b 6.1721.4642.92Libre sensors134.53 c 130.660750.00 Cost offset Renumeration from activity based funding from clinician appointments2983.631831.443549977Abbreviation: FTE = full‐time equivalent. a Hospital finance team. b Pathology Queensland price list. c Purchase invoices (Abbott diabetes care). Resource costs per participant for the intervention arm. Abbreviation: FTE = full‐time equivalent. Hospital finance team. Pathology Queensland price list. Purchase invoices (Abbott diabetes care). Healthcare utilisation costs included costs of post‐operative hospital admission, day surgery costs, and costs associated with 30‐day post‐operative complications. Only complications incurring costs for the healthcare institution were included in the analysis. This included complications that resulted in hospital readmissions. Costs associated with superficial surgical site infections treated in the community, such as costs of GP visits and outpatient medications, were excluded from the analysis as these costs are incurred outside the healthcare institution. Costs of inpatient admissions were estimated from the unit cost per hospital bed‐day multiplied by the length of stay. Day surgery costs were estimated from the average recovery room cost per episode of care. Sensitivity Analysis Univariate sensitivity analysis was conducted with plausible ranges from the base‐case value (Table 1) to test the robustness of model estimates. Where available, sensitivity analysis ranges were based on the original interventional study, NSQIP data, or published literature. When this data was not available, we estimated sensitivity analysis ranges that were sufficiently broad to properly assess the model robustness but remained plausible. The sensitivity analysis also included government remuneration for outpatient consultations conducted by the multidisciplinary service (see cost offset in Tables 2 and S3). In Australia, an activity‐based funding (ABF) system enables public hospitals to receive federal funding based on the volume and complexity of services delivered to public hospital inpatients and outpatients [28]. Probabilistic sensitivity analysis was undertaken to assess the impact of simultaneous uncertainty across model parameters. Gamma distributions were assigned to cost parameters and beta distributions to probability parameters. Distribution parameters were derived from the interventional study, NSQIP data, or published literature. When this data was unavailable, plausible ranges consistent with those used in the univariate sensitivity analysis were applied to reflect parameter uncertainty (Table 1). Monte Carlo simulation with 1000 iterations was performed by randomly sampling values from the specified distributions for each model input. The results were plotted on a cost‐effectiveness plane. Results Direct service operating costs were A$2718 per participant for the intervention arm. In the base‐case analysis, when total costs inclusive of healthcare utilisation were considered, the intervention was cost‐saving, with a total cost of A$8335 per person compared with A$12 302 per person for the usual care arm. The modelled incidence of 30‐day post‐operative infection‐related readmission was 3.0% in the intervention arm compared with 10.1% in the usual care arm, corresponding to 7.1 readmissions avoided per 100 patients treated. The intervention was associated with lower costs and fewer readmissions than usual care and was therefore dominant. The univariate sensitivity analysis demonstrated that the results were generally robust to changes to input parameters (Figure 2). The model was most sensitive to changes in length of stay costs for both the intervention and usual care arms (Figure 2), with cost savings varying from A$2254 to $A7581 and −A$46 to A$7339, respectively. Incorporating activity‐based funding cost offsets increased the magnitude of cost savings, with maximum savings of up to A$13 943 observed in scenarios with the highest cost offsets. Tornado diagram showing results of univariate sensitivity analysis. LOS, length of stay; prob, probability; SSI, surgical site infection; UTI, urinary tract infection. Probabilistic sensitivity analysis demonstrated that the intervention remained highly likely to be dominant, with 89.5% of simulations associated with both lower costs and fewer post‐operative infection‐related readmissions than usual care (Figure 3). The mean cost saving was A$6320 per patient (95% uncertainty interval: −A$2062 to A$20 426), and the mean incremental effectiveness was 6.3 post‐operative infection‐related readmissions avoided per 100 patients treated (95% uncertainty interval: 2.0–11.5). Cost‐effectiveness scatterplot from the probabilistic sensitivity analysis. The black diamond represents the mean probabilistic sensitivity analysis result while the black triangle represents the deterministic base‐case result. Discussion The findings of this study suggest that compared with usual care, a multidisciplinary pre‐operative diabetes service is a cost‐effective intervention from a healthcare institution perspective. To our knowledge, this is the first study to evaluate the economic impact of a dedicated multidisciplinary diabetes service in a pre‐operative setting. There is mounting evidence linking elevated pre‐operative HbA1c levels and adverse post‐operative outcomes across many different surgical specialties, with recent large‐scale studies highlighting the importance of routine pre‐operative HbA1c screening and pre‐operative diabetes optimisation [20, 21]. Findings from this present study demonstrate the value of investing in services to improve pre‐operative glycaemic management, with reductions in length of stay and post‐operative infection‐related readmissions emerging as primary contributors to cost savings and improved outcomes. Numerous studies have reported a clear relationship between higher pre‐operative HbA1c levels and increased hospital length of stay [5, 6, 20, 21]. A recent study of over 500 000 surgical patients found that prolonged length of stay was most prominent in people with pre‐operative HbA1c levels ≥ 9% (≥ 75 mmol/mol), followed by those with pre‐operative HbA1c levels between 8% and 8.9% (64–74 mmol/mol) [21]. Similar findings were presented by a study of over 200 000 general surgical patients [20]. Whilst increased incidence and severity of post‐operative complications are important contributing factors [29], the need for inpatient endocrinology consultation to manage hyperglycaemia has also been cited as a reason for prolonged post‐operative hospitalisation in people with poorly managed diabetes [30]. Interventions to improve pre‐operative glycaemic management may therefore reduce the incidence of hyperglycaemic episodes during the post‐operative admission, thereby reducing length of stay. In the original interventional study, participants enrolled in the multidisciplinary pre‐operative diabetes service achieved a 1.9% mean reduction in pre‐operative HbA1c, decreasing from a baseline mean of 9.7% (83 mmol/mol) to 7.9% (63 mmol/mol) on the day of surgery [19]. This likely accounts for the shorter mean length of stay in the intervention arm compared to the usual care arm observed in the present study. The univariate sensitivity analysis demonstrated that the economic model remained robust despite substantial variations in input parameters, including significant changes in length of stay costs across both the intervention and usual care arms. A reduction in length of stay of 1.6 days in the usual care arm was required before the intervention was no longer cost saving. These findings were further supported by the probabilistic sensitivity analysis, in which 89.5% of simulations indicated lower costs and fewer post‐operative infection‐related readmissions for the intervention compared with usual care. Collectively, these analyses suggest that the economic benefits of the intervention were robust to uncertainty in model inputs. This economic analysis addresses an important gap in literature by demonstrating the cost‐effectiveness associated with a dedicated service focused on improving diabetes management in the pre‐operative setting. Previous evaluations of pre‐operative diabetes programmes have described models of care whereby patients are fast‐tracked to existing services [31, 32]. Whilst this may reduce staffing costs, it poses significant challenges in ensuring timely, effective and equitable access—particularly in public hospital settings where there are long waiting lists for specialist diabetes services [33]. Programmes that rely on existing services may struggle to achieve meaningful glycaemic improvements in the short pre‐operative timeframe due to constraints on frequency of patient reviews. Additionally, extra resources may be required to accommodate the increased demand, which would affect cost‐effectiveness. Not only did our study find that having a dedicated pre‐operative service was cost saving, but further savings were gained from remuneration from activity‐based funding. Although ABF remuneration rates are revised annually and may influence cost‐effectiveness, it still provides a strong financial incentive for other Australian hospitals and international sites with comparable remuneration through federal or insurance schemes to implement similar services. The multidisciplinary nature of the pre‐operative diabetes service was likely influential in minimising the operating costs of the intervention. The pharmacist and diabetes nurse led model enabled the endocrinologist to review any new or complex patients at a 0.1 full‐time equivalent (FTE) capacity. This enabled staffing costs to be kept to a minimum, whilst enhancing the effectiveness of the service. These findings align with previous research which found that team‐based care improved clinic efficiency by enabling physicians to focus their time on patients with more acute or complex needs [12, 16, 34, 35]. In our study, the cost of an endocrinologist was approximately 2.5 times the cost of a pharmacist and 3 times the cost of a diabetes nurse and hence increases in endocrinologist FTE would substantially increase service costs, a finding mirrored in other studies [16, 35]. Strengths of this study include its use of prospective, real‐world data and analysis from a healthcare institution perspective which makes its findings relevant for clinicians and hospital administrators considering implementation of similar services. The detailed breakdown of direct service operating costs and healthcare utilisation costs enhances the study's reproducibility, whilst the consideration of ABF offsets reflects real‐world financial mechanisms in Australian public hospitals, enhancing the relevance of the results for policy and funding decisions. Despite these strengths, there are several limitations. The model data for the intervention arm comes from a single study and institution, which may limit generalisability. Although the original interventional study was not powered for secondary outcomes such as post‐operative infections and readmissions, the broad range examined in the sensitivity analyses based on event rates reported in literature had little effect on the results of the economic model. The post‐operative complications incorporated into the economic model were based on those reported in the original study and were therefore limited to post‐operative infections and readmissions [19]. However, large multicentre studies have shown that patients with elevated pre‐operative HbA1c levels also experience higher rates of overall post‐operative and medical complications [20, 21], suggesting that the cost savings associated with a pre‐operative diabetes intervention may be even greater than those identified in this study. Similarly, emergency department presentations were not included in the analysis as comparator NSQIP data was unavailable and accurate estimates could not be derived from literature. However, as post‐operative complications are a main driver of hospital readmissions and emergency department presentations, it is feasible that the incidence of emergency department presentations would also be higher in those with higher pre‐operative HbA1c levels. Superficial incisional surgical site infections were also excluded from the current analysis, as these infections are predominantly treated in community settings and costs are generally not incurred by the health institution [36, 37]. Additionally, there were significant differences in the methods of recording superficial incisional infections in the interventional study compared to NSQIP, and therefore incidence data was not comparable. Although not a focus of the current study, improvements in pre‐operative HbA1c levels achieved through a multidisciplinary pre‐operative diabetes service are likely to confer broader patient benefits through improved health‐related quality of life (HRQoL). Previous studies have demonstrated that reductions in HbA1c are associated with improvements in diabetes‐related symptoms, daily functioning, and utility values independent of long‐term complications [38]. Future studies should therefore aim to validate the relationship between perioperative HbA1c improvement and quality‐adjusted life years in surgical populations. Likewise, the long‐term clinical and economic outcomes of pre‐operative diabetes services, including sustained glycaemic control, HRQoL, and downstream healthcare savings, warrant further investigation. Economic analyses from a societal perspective are also needed, as indirect costs associated with post‐operative complications—including reduced productivity, caregiver burden, and long‐term disability—may be substantial. Finally, given the reported benefits of proactive diabetes care by a specialist inpatient diabetes team [39], further studies should also investigate the potential cost savings of integrating an inpatient diabetes specialist team alongside pre‐operative diabetes services to support glycaemic management across the perioperative continuum of care. Conclusion In conclusion, this study provides the first economic evaluation of a dedicated multidisciplinary pre‐operative diabetes service and demonstrates that it is cost‐effective compared with usual care. The pharmacist and nurse‐led model offers potential for scalability and implementation across other hospital sites, especially in the context of increasing diabetes prevalence and its impact on surgical outcomes and healthcare costs. Incorporating activity‐based funding reimbursements further strengthened the intervention's economic value, indicating that such services may be sustainable within current funding frameworks. These findings support broader implementation of multidisciplinary pre‐operative diabetes services to improve surgical outcomes, reduce healthcare expenditure, and enhance patient care. Author Contributions Abby Yu: conceptualisation, methodology, validation, formal analysis, investigation, writing – original draft, writing – review and editing, visualisation. Karen Whitfield: conceptualisation, writing – review and editing, supervision. Andrew Hale: conceptualisation, writing – review and editing, supervision. Michael D'Emden: conceptualisation, writing – review and editing, supervision. Peter Donovan: formal analysis, writing – review and editing, supervision. Centaine Snoswell: conceptualisation, methodology, validation, formal analysis, writing – review and editing, supervision. Funding The authors have nothing to report. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Table S1: Baseline characteristics of the intervention cohort. Table S2: Types of surgical procedures performed for intervention cohort. Figure S1: Consolidated health economic evaluation reporting standards (CHEER) checklist. Table S3: Additional information on cost inclusions for intervention arm.