Cost comparison of orthopaedic fracture pathways using discrete event simulation in a Glasgow hospital.

OBJECTIVE: Healthcare faces the continual challenge of improving outcome while aiming to reduce cost. The aim of this study was to determine the micro cost differences of the Glasgow non-operative trauma virtual pathway in comparison to a traditional pathway.
DESIGN: Discrete event simulation was used to model and analyse cost and resource utilisation with an activity-based costing approach. Data for a full comparison before the process change was unavailable so we used a modelling approach, comparing a virtual fracture clinic (VFC) with a simulated traditional fracture clinic (TFC).
SETTING: The orthopaedic unit VFC pathway pioneered at Glasgow Royal Infirmary has attracted significant attention and interest and is the focus of this cost study. OUTCOME MEASURES: Our study focused exclusively on patients with non-operative trauma attending emergency department or the minor injuries unit and the subsequent step in the patient pathway. Retrospective studies of patient outcomes as a result of the protocol introductions for specific injuries are presented in association with activity costs from the models.
RESULTS: Patients are satisfied with the new pathway, the information provided and the outcome of their injuries (Evidence Level IV). There was a 65% reduction in the number of first outpatient face-to-face (f2f) attendances in orthopaedics. In the VFC pathway, the resources required per day were significantly lower for all staff groups (p≤0.001). The overall cost per patient of the VFC pathway was £22.84 (95% CI 21.74 to 23.92) per patient compared with £36.81 (95% CI 35.65 to 37.97) for the TFC pathway.
CONCLUSIONS: Our results give a clearer picture of the cost comparison of the virtual pathway over a wholly traditional f2f clinic system. The use of simulation-based stochastic costings in healthcare economic analysis has been limited to date, but this study provides evidence for adoption of this method as a basis for its application in other healthcare settings. © Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2017. All rights reserved. No commercial use is permitted unless otherwise expressly granted.

We used a simulation-based costing approach to compare the costs of a service innovation (the virtual fracture clinic or VFC) with the default standard of care, in order to quantify the cost saving associated with the VFC and support decision makers thinking of introducing this new model in their own hospitals.

The data which populates our model is based on operating data from a hospital which has piloted VFC and operated it since 2011.

Our model is explicitly designed to reflect variability in costs as not every patient will have the same needs and the same costs.

Simulated studies involve a great deal of work and necessarily involve making assumptions, both of which could have been avoided had baseline data been collected before the VFC was introduced.

Introduction

An ongoing challenge within healthcare is to improve the quality of care while reducing cost.1–3 A common response to increasing demand is to attempt to increase capacity and activity4 through additional funding.5 6 In the longer term, this is financially unsustainable.7 8 In addition, cost reduction without regard to the outcomes achieved is unsafe and potentially limits effective care.9 When health services are faced with increased demand, service redesign can be presented as the way forward. Instead of providing additional capacity, clinicians and managers should seek to identify activity which shows no demonstrable patient benefit.10 11 These activities may be the result of traditional practices that have not been scrutinised in the light of newer evidence. However, clinicians and managers may fear future cuts in essential resources if volume decreases.12 The term ‘clinical pathway’, first used in 1985, has widespread use in healthcare management.13 Essential to creating value is to understand patient care pathways and to enhance care quality by improving risk adjusted patient outcomes, promoting patient safety and increasing satisfaction with optimum use of resources.14 More recently, computerised algorithmic forms have been called ‘care coordination pathways’ which support a variety of functions.15 These would include benchmark treatment steps with an associated cost and are considered in a different way to clinical pathways.

In healthcare systems, it is relatively simple to measure metrics such as volume16 but more difficult to measure clinical outcome, patient satisfaction and ultimately, value.9 The identification of activities that represent poor clinical and financial value, and obtaining support for redesign, are challenging.17

Case setting

This paper focuses on orthopaedic outpatient fracture clinic pathway redesign which was developed at Glasgow Royal Infirmary (GRI) with the primary aim of improving patient care.18–20 GRI is a large university teaching hospital serving a metropolitan area. It is financed by the National Health Service (NHS) in Scotland. The NHS provides universal coverage, funded from general taxation and free at point of use. In 2014/15, £10.8 billion was allocated by the Scottish Government for operating costs of the NHS.21 This is shared proportionally among the 14 health boards in Scotland and 2 of the 8 special boards, based partially on the Resource Allocation Formula.22

Clinicians identified unnecessary duplication in traditional fracture clinics (TFCs), especially where patients with minor orthopaedic trauma can be managed effectively by emergency medicine clinicians. However, it is common practice to have a further face-to-face (f2f) consultation with an orthopaedic clinician23–25 often within 48 hours of the original emergency department (ED) attendance. The patient does not gain any new information or further treatment during many of these encounters.26 It can be physically arduous and financially penalising for patients to attend during this most painful and functionally restricted period of their recovery. Recent evidence supports encouraging early mobilisation without further routine review by demonstrating good functional outcomes.27–34 In healthcare systems with fee-per-patient visit,35 these patients may be considered as a source of income.

While we recognise that non-operative orthopaedic patient pathways vary and operate according to local conditions, the VFC model pioneered at GRI has attracted significant attention and interest.20 For example, other pathways use Trauma Triage Clinics where all patients with non-operative trauma are virtually reviewed by an orthopaedic consultant, with instructions transmitted electronically for nurse and administrative staff to execute, while others carry out virtual reviews with a larger team. Previously, the only evidence on costing was derived from a relatively broad brush top-down costing,36 which gives relatively little insight into the cost savings and outcome measures. One advantage of the Glasgow virtual fracture pathway (figure 1), promoted by the Scottish Government37 is that it uses agreed evidence-based standard protocols and multidisciplinary review of patients. Agreed protocols between the ED and orthopaedic department have been established for six common/frequent injuries to reduce unnecessary variation.38–40 Protocols include the use of removable Velcro splints for specific injuries and enable direct ED discharge of minor stable fractures and facilitate patient self-care without the need to attend a clinic for removal.41 Those not discharged have their records and X-rays reviewed 7 days a week, by an Orthopaedic consultant and senior nurse in a ‘virtual fracture clinic’ (VFC). Again standard agreed protocols are used during VFC for some injuries to determine further treatment, reducing further unnecessary variation. The resulting management plan is then outlined and agreed with the patient by telephone immediately afterwards. An ‘open-door’ policy is essential to respond to any patient’s concerns after discharge. Patients therefore only attend when necessary, at the optimal time point and with the most relevant specialist. The VFC pathway includes the attendance of patients for their first face to face visit. The standardisation of care along evidence-based guidelines, as applied in the Glasgow pathway, is an important way of increasing value.39 However, is a virtual fracture pathway cost effective?

Figure 1

Virtual fracture pathway process flow model using the symbols shown to define individual steps in the process. This defines the flow of information and the review process where patients are reviewed without being present. It also includes a predefined process where some patients are seen face to face in a traditional fracture clinic. ED, emergency department; MIU, minor injuries unit.

Methods

The study was performed at GRI. It is a provider of both secondary and tertiary orthopaedic care. Under the terms of the Governance Arrangements for NHS Research Ethics Committees in the UK, the research project was classified as a service evaluation and therefore did not require ethical review.

There are different methods to quantify the costs associated with a healthcare process. A ‘top-down’ approach examining high-level, administrative, financial data has already been performed.36 This does not detail how the cost savings were realised.

Mapping patient pathways assists in the redesign of healthcare systems42 and this was used to understand the VFC pathway and the TFC pathway (figure 1 and figure 2). Discrete event simulation was chosen as the modelling tool as it is often used for comparison of alternative scenarios. Capturing variability in the parameters defining the pathways was also important.43 It had also been previously used at the case hospital with a focus on improving patient flow. This mapping can then be translated to DES models which have the benefit of incorporating stochastic variability and modelling of patient flows, which has been used extensively in healthcare.44 45 DES has also been used to examine costs in time-driven activity-based costing (TDABC), a ‘bottom-up’ approach to costing. Kaplan et al 9 evaluated the time spent and costs incurred at each step of a complete patient care cycle and summed using clinical resources, material and allocated costs. A pilot project undertaken using TDABC concluded that better management of resource utilisation will ultimately enable providers to deliver the same, or better, outcomes with fewer personnel. These detailed costings can provide evidence to inform decision making46–48 as they investigate the cost of individual episodes by measuring actual resource utilisation.9 49 By using a simplified mapping and modelling approach and including personnel costs, we can report on the value of two different patient pathways. DES for this type of costing approach facilitates comparison between alternative pathways providing managers and clinicians with information to improve the patient experience. DES was also useful as it allowed us to assess costs subject to variations in demand and available resources. The role of DES is often used as a basis for experimentation, especially when there is interest in looking at different scenarios for change and improvement. It is used here retrospectively to show the cost and efficiency saving that are possible from implementing such a virtual pathway. Patient outcomes and satisfaction are also a crucial consideration.

Figure 2

Traditional fracture pathway process flow model using the symbols shown. This defines individual steps in the process for patients attending an orthopaedic outpatient appointment. It is based on the British Orthopaedic Association Standards for Trauma where all patients attending emergency department with a non-operative orthopaedic injury must attend an outpatient clinic.

Outcomes on the Glasgow pathways

This is reported based on literature published by the case hospital. Audit and publication for specific injury of patient outcomes has been ongoing at the study hospital since the new pathway began operating in 2011. There are six protocols used by ED to discharge stable fractures, with four publications thus far published relating to patient outcomes.26 50–52 Patients are assessed using patient-reported outcome measures appropriate to their injury in addition to patient satisfaction regarding discharge leaflets and the outcome of their injury. The unit has also published more generally on the VFC18–20 and reported on the new pathway from a medicolegal perspective, assessing whether patients have access to as much information as they want.

Measuring cost

This work takes a hospital perspective on costs and DES has been used to develop detailed simulation models allowing application of a stochastic costing approach to staff utilisation within the healthcare setting.53 54 Stochastic costing uses statistical distributions of model activity duration and routings through complex pathways and can measure resource utilisation of each step, and therefore the overall process. It follows three broad steps: (1) identifying relevant patient-specific resources; (2) quantifying the resources consumed and (3) determining the value of each resource consumed. Although this method shares aspects with TDABC,9 it focuses on a single step in the patient care cycle and has the advantage of yielding accurate cost data that can be subject to statistical analysis.54 Capital and consumable costs are not modelled.

An action research methodology,55–57 combined with DES, was used to investigate the Glasgow VFC pathway and compare it with a traditional British Orthopaedic Standards for Trauma pathway. Our study focused exclusively on patients with non-operative trauma attending ED or the minor injuries unit (MIU) and the next step in the patient pathway. Two DES models were developed to determine the staff utilisation and actual staff costs of a VFC model compared with a TFC model using computer simulation software (Simul8).58 These models were developed by close observation of the clinical and administrative processes involved, and interviews with clinical and managerial staff. Conceptual pathways of both the VFC (figure 1) and the TFC (figure 2) were the first steps in the model building process.

Data collection—simulation models

Multiple sources were used to develop the models. Data for 1 year of the new process (2014/15) was extracted from the hospital patient management system (TRAKcare) and orthopaedic electronic database (Bluespier,59). A total of 6291 patients were considered. Within the redesigned VFC pathway, 30% of the patients initially attending ED/MIU with a non-operative fracture injury were discharged home without follow-up at a fracture clinic (figure 1). This did not require any new resources and brought significant benefits to ED.15 A further 35% were discharged from the VFC. These patients would incur a cost if they accessed the telephone helpline and this was modelled. This resulted in only 35% of the initial cohort of patients being reviewed at f2f consultation in a TFC. Within the TFC model: 100% (n=6291) of the patients were referred onto orthopaedics and reviewed f2f in a clinic (figure 2). The time taken for each activity was collected by direct observation over a period of 4 months. This prolonged time period made the ‘Hawthorne Effect’,60 that is, behaviour modification in response to being observed, less likely as the staff became accustomed to the researcher's presence. TFC model activity timings were based on direct observation of f2f consultations over a 2-week period, with a variable case-mix in 10 clinics of 20–32 patients. As no data were available, the routings for a TFC were estimated by multiple experts with over 30 years experience in orthopaedic care. The institution’s finance department provided the mean salary for each staff group, including a standard overhead cost of 23% (table 1). Since the same costs were used in each model and run for 1 year for comparison purposes, discounting of costs and benefits was not used. There are no costs included associated with the discharge of patients from ED18 or from the VFC, as only a small proportion of these patients sought additional help (table 2) and this was beyond the scope of this work. The model includes staff costs incurred if patients contact the helpline. The patient’s journey is based on probability distributions and availability of resources resulting in patients waiting in queues. Both models were simulated for a period of 1 year and output results were collected for 20 trial runs simulated for each model. The number of trial runs was obtained by using the trials calculator.61

Table 1

Input parameters for simulation models

	Parameter	Resources	Mean and (SD)	Data source/comments
VFC Steps	VFC arrival rates (33% discharged ED)		12 patients per day	Historical data analysis- patients discharged at ED figure 1 One-third of patients discharged at ED so only 12 at VFC
	Admin 1	Admin	1.06 min (0.64)	Direct observation
	Nurse Prep	Nurse	1.61 min (0.7)
	VFC consultant review	Consultant	1.75 min (0.95)
		Nurse
	Nurse calls	Nurse	8 min (4)
	Admin 2	Admin	2.5 min (1.7)
	Admin letters and appointments	Admin	2.9 min (1.35)	Around one-third of patients will still follow the traditional pathway. All others have been discharged.
	Discharge advice letters	Admin	2.33 min (0.5)	Direct observation
	Helpline call arrival rate	Nurse	2.6 per day
	Helpline call duration	Nurse	4.5mins (0.15)
	VFC decision point Discharged		50%	Historical data analysis
	Referred to consultant clinic		40%
	Referred to N/L clinic		10%
	Staffing: Consultants		1
	Nurses		1
	Admin		2
	Typists		1
TFC Steps	ED/MIU arrival rates		18 patients per day	Historical data—all patients (figure 2)
	X-ray		14 min (5)	Historical data
	Nurse prep	Nurse	1.61 (0.7)
	Consultant consultation	Consultant	12 min	Observation
	Second consultation	Consultant	3 min	Expert opinion/observation
	Nurse consultation	Nurse	20 min
	Discharge admin	Admin	6 min
	Assessment routing: Admitted		1%	Expert opinion: Much of these routing values are based on the clinical mix of patients and therefore does not warrant sensitivity analysis. As stated in the text, one limitation of this work is that the clinical mix for the new virtual pathway is that these routings are likely to be different as the simplest injuries have been discharged.
	X-ray		3%
	Treatment		72%
	Discharge		24%
	Treatment routing
	X-ray		6%
	Second consultation		93%
	Discharge routing
	Discharged		36%
	Return appointment		64%
	Staffing: Consultants		3	Sensitivity analysis was necessary here in terms of the number of staff required to ensure all patients were seen within the allocated session time. These values represent the necessary staffing required.
	Nurses		3
	Admin		2
	Typists		4
BOTH	Shifts Admin1 (VFC)		07:30–08:15	For printing off lists only
	Admin 1 (TFC)		07:45–09:00	Longer shift as higher volume of patients
	Consultant (14)		09:00–13:00	4-hour consultant session
	Nurse (8)		08:00–16:00	Average Full Time nurse working hours
	Typists		09:00–17:00	Discharge letters to General Practitioner and patient (Mon–Fri)
	Admin 2		11:00–16:00	For issuing letters only
	Hourly rates GRI consultant		£62.91	Average for GRI orthopaedic consultants. Obtained from GGC finance dept. for 2014/15. Include 23% employer costs
	Nurse		£20.96	Based on April 2014 figures. They have 23% employer costs added and are then divided by 42 weeks. GRI staffing levels based on average of 8 nurses (B7, B6 and B5×6).
	Admin		£12.74	Average
	Admin 2		£16.22	Average
	Efficiency		85%	Sensitivity analysis shows the effects of this on cost
	Clock		Hours	1 year as we had historical data for this time period
	Warm-up period		168 hours	Tests completed to ensure model in steady state73

ED, emergency department; GRI, Glasgow Royal Infirmary; MIU, minor injuries unit; TFC, traditional fracture clinic; VFC, virtual fracture clinic.

Table 2

Summary of outcome publications from the case hospital

Injury	Where protocol used	Study year	Patients/number of appointments Response rate	PROM	PROM Score	Satisfaction	Comments	Reference
Fifth Metatarsal	ED	2009–2010	279 patients/491 appts. Before new pathway 96% attended ortho clinics	None	–	Retrospective. Satisfaction and outcome not collected prior to 2011	No added clinical value of routine follow-up of these patients.	Ferguson et al 51
		2011–2012	339 patients/102 appts. After new pathway 57% discharged ED 19.5% attend f2f	None	–	78% satisfied outcome 79.6% satisfied leaflet	3% visited another hospital or GP
Mallet finger	ED	2011–2012	47 patients 77% response rate Two patients referred to VFC rest discharged ED	Quick Dash EQ5D VAS EQ5D	Mean=2.27 (IQR 0–4.55) 90 (IQR 75–90) 0.88 (IQR 0.84–1)	100% satisfied process 100% satisfied helpline	Mean follow-up 322 days postinjury 19% visited another hospital or GP	Brooksbank et al 50
Fifth Metacarpal	ED	2012 Apr–Oct	167 patients 59% response rate	Quick Dash EQ5D	Mean=2.3 (IQR 0–6.8) 0.87 (IQR 0.74–1)	80.6% satisfied outcome 84.9% satisfied process	Retrospective. Mean follow-up time 21 months 9% visited another hospital or GP	Gamble et al 26
Radial head and neck	ED	2011–2012	202 patients 77% response	Extensively investigated in the literature. Short-term satisfaction to maximise response.		96% satisfied- (suspected fracture) 87% satisfied (fracture group) 95% satisfied with information received.	Retrospective Mason I or II fractures included 90% discharged with no f2f review 3.4% discrepancy rate based on radiologist report 10% visited another hospital or GP	Jayaram et al 52

Two paediatric fractures are also discharged at ED, the Torus/Buckle and Clavicle. PROM’s and satisfaction for trauma pathway not routinely carried out at the case hospital prior to introduction of VFC.

ED, emergency department; f2f, face-to-face PROM, patient-reported outcome measure.

TFC process steps can be unique for a specific hospital; however, our study simplifies this by investigating and simulating only the first step after ED. This model was based on both data collections for timings and expert opinion on some of the routings as this service no longer runs at GRI. Our TFC model has similar administrative preparation, but every patient attends for an f2f review as was the case historically. Several consultants and nurses were required, but there were no weekend clinics. The time for dressings or plaster cast removal was included based on expert opinion. The model included queuing prior to each activity and the FIFO rule was applied. In the TFC, patients are not turned away and additional help would be called in to deal with patients still waiting. This meant clinics could overrun, creating additional pressures on staff.

Validation and verification

The aim of model validation is to provide the modeller and stakeholders with appropriate confidence in the model such that they will use it as an aid for decision making.62 Verification and validation occurred throughout the model development and testing62 63 as a continuous iterative process. The models contained activities proceeded by queues which were based on a first in first out routing. For both models we did micro white box structural verification through regular meetings with clinical and managerial staff who authenticated the structure, assumptions and face validity.

Parameter verification and operational validity of the TFC model was provided by the process experts who delivered some of the data. We also investigated the number of patients completed as it was essential that there was no baulking or reneging within the model and that all patients were seen within allocated session time. Black box validation was challenging as costs are not currently reported based on activity. The reported cost for an outpatient appointment at the case hospital in 2011 was £113 split into £31 allocated and £82 direct. The direct cost includes overheads, allied health professionals costs, lab and pharmacy costs and does not distinguish between trauma and multiple elective pathway patients being seen in an outpatient clinic. To test the model under extreme conditions consultant and nurse-estimated timings of +/−10% and +/−50% was also undertaken. Even making the extreme assumption that the TFC patients take 50% less time for both of the consultant consultations and the treatment time, the TFC pathway still costs more (£25.31 (95% CI 24.67 to 25.09)) than the VFC pathway. The relationship among efficiency, arrival rates and cost is demonstrated in figure 3. We conclude that the models are a good representation of reality that allow questions about the process to be answered and provide outcomes which are reliable. For the TFC, there were limited historical data available to compare with.

Figure 3

Sensitivity analysis. This shows the sensitivity to cost based on arrival rates and efficiency. It clearly highlights the difference in cost between a TFC and a VFC. TFC, traditional fracture clinic; VFC, virtual fracture clinic.

Results

The output of the models was the staff (medical, nursing and administrative) utilisation and costs for each process step. This allowed a cost per patient to be derived. The reported costs are based on figures for 2013/14, with no corrections for inflation. The performance indicators are reported based on the model output and the 95% CI calculated. The cost for each resource is then summed and divided by the number of patients processed through the model to give a per-patient cost for both the VFC and the TFC. The calculated costs for 20 simulated runs for both the TFC and VFC were compared. A t-test was performed to determine if there was a statistically significant difference between the two pathways.

Effectiveness

A summary of outcome publications from the case hospital on the new pathway is provided in table 2. This includes four upper extremity injuries where patients with simple stable fractures are discharged directly from ED. They all conclude that simple upper extremity fractures can be managed with good patient satisfaction and acceptable functional results without further f2f review. More recently, published analysis of low risk upper extremity fractures64 also found equal outcomes such that these patients should have an optional rather than a scheduled second visit.

The medicolegal aspects of the VFC at the case hospital are addressed65. It states that ‘professional negligence claims can be avoided by the use of robust, up-to-date protocols that are based on national standards’. Since 2011 when the VFC has been in operation at Glasgow, over 30 000 patients have been managed with ‘no complaints or medicolegal actions arising from diagnosis or management in the VFC’ 65. This has been achieved with robust protocols, patient access to information and an open door policy, providing an important safety net giving patients reassurance that they can access expert assistance if required. Advice leaflets for specific injuries can be found on the redesign website,20 which are also provided to patients in paper form at ED.

Efficiency

There is more efficient use of resource with the VFC pathway. The resources required per day were one consultant for 51 min (95% CI 40 to 62), one nurse for 279 min (95% CI 275 to 283) and one typist for 88 min (95% CI 68 to 108). Two administration staff were used (table 1) detailed as admin 1 for 18 min (95% CI 16 to 20) and admin 2 for 28 min (95% CI: 28 to 29) (table 3).

In the TFC model, which required all patients to be seen f2f, more staff were required (table 1). Assumptions were tested by running the model and addressing the build-up of queues. This model required 540 min of consultant time (95% CI 533 to 554), 778 min of nurse time (95% CI 749 to 792), typists for 346 min (95% CI 307 to 365) and two administration staff. Admin 1 staff were used for 17 min (95% CI 15 to 18) and admin 2 for 43 min (95% CI 36 to 48) (table 3).

Cost

The cost of a patient attending a TFC was £36.81 (95% CI 35.65 to 37.97). The cost of the VFC was £14.23 (95% CI 13.23 to 15.18) per patient. In the VFC pathway fewer patients with injuries are referred (70% of TFC model), and half of these proceed to a subsequent f2f encounter. The overall cost per patient of the VFC model was £22.84 (95% CI 21.74 to 23.92) per patient compared with £36.81 (95% C: 35.65 to 37.97), a saving of 38% (table 4). Comparison of these costs using t-tests generated p values that were highly statistically significant (p<0.001).

Table 4

Summary of costs from model

Per patient costs £	−95%	Average	95%
VFC	13.23	14.23	15.18
TFC (all patients seen f2f)	35.65	36.81	37.97
Virtual pathway (35% seen f2f in TFC)	21.74	22.84	23.92
Saving per patient	13.91	13.97	14.05

f2f, face-to-face; TFC, traditional fracture clinic; VFC, virtual fracture clinic.

Discussion

A new fracture pathway for non-operative fractures instituted at GRI demonstrated clear benefits to the patient. In addition to this, the novel pathway avoids unnecessary follow-up appointments and current evidence demonstrates clinical safety and efficacy.26 50–52 A robust ‘bottom-up’ cost analysis of the VFC demonstrated a cost saving of 38%. There was reduced utilisation of medical and nursing resources: a nurse normally works up to 5 hours per day in VFC duties, whereas in a traditional clinic, several nurses are required to cope with caring for patient’s f2f. To meet demand only one consultant is required in a VFC for up to 1 hour per day, compared with three consultants each working for 3 hours to sustain an equivalent TFC model. The pressures on the outpatient service were greatly eased as the redesign reduced the volume of f2f attendances by 65%. The peaks in demand were smoothed by improving flow and facilitating better patient management. Staff were used for a shorter time in the VFC compared with a TFC, but the actual costs of our VFC process compared with TFC per patient had previously not been established.

There have been both long-term and short-term effects on the service as a result of the somewhat theoretical savings (not reported on end of year financial reports). Shorter term, they reduced pressure on the outpatient service due to the reduction in patient appointments, providing additional capacity long-term. In addition, over the intervening period these savings were gradually reinvested into subspecialist service development, which otherwise would have resulted in significant expenditure for the hospital. In fact, in the context of GRI, 30 hours of consultant-led specialist services were provided without new capital or recurring investment.

The strength of this study was that it provided a simple bottom-up micro costing method approach which analysed the next step after ED/MIU attendance. This allowed the VFC pathway to be observed and timed, providing staff costs based on actual activity. The DES approach allowed us to model the stochastic nature of process flow, incorporating steps that are not normally studied in detail. The use of DES to examine the stochastic nature of the process and model reality in a transparent and simple way was another strength. DES is time-based and takes into account resources, constraints and how the process and activities interact with each other as time passes. DES is often used in stochastic situations where the visual component is very important when planning change.66 67 It is has been widely used in healthcare66 and more recently in health economic evaluation, to gain insight into complex processes.68 69

In terms of the limitations, the TFC simulation was built retrospectively using expert opinion and current observation of a TFC, based on the case hospital, perhaps resulting in a less robust model. One could also argue about the completeness of this data although we have been transparent about how these data were collected. In contrast, the VFC was based on direct observation and historical data analysis and validated against actual performance.

The TFC simulation used the same f2f costing as for 35% of patients in the VFC. A limitation of the study was that the virtually discharged patients in the new system were assumed to use the same clinic resources as those reviewed f2f, whereas the latter may be more complex and costly. We also did not include patients returning for further visits or other cost savings, especially replacing plaster casting with removable splints,70 or the reduction in the use of hospital transport, or societal costs. However, including these factors could only make the VFC seem more attractive and so would not change our conclusion that the VFC is more cost effective. Since this work was a process change across a number of different injuries, there was unfortunately no opportunity to do any randomisation. The process change was made based on existing clinical evidence and with the agreement of experts for each specialty.

One of the biggest opportunities for healthcare improvement is to eliminate unnecessary clinical processes, freeing up time and thereby achieving higher quality of care at lower cost. Healthcare professionals can reduce clinical variation; however, one major barrier to change is that the financial benefits of successful grass-root redesign are frequently unrecognised71 and a decrease in volume may result in a reduction in funding.

Further research could look at the cost effectiveness of other fracture pathways to explore the impact of discharge rates without standard protocols in place at ED and within orthopaedics. It would also be of benefit to understand the impact of the open door policy and the longer term effects of the standardisation on patient care and costs.

Conclusion

The VFC pathway has been in operation at the case hospital for over 5 years and has spread rapidly across the UK and beyond as the evidence of quality improvement has been recognised by other hospitals.20

The financial impact of a virtual process, with patients being reviewed f2f only if necessary, has been highlighted. This opportunity to make savings and increase activity should be of interest to Managers and Clinicians especially due to the benefits for patients. The reported outcome shows that the new pathway is at least as effect as a traditional clinic for a reduced cost. Although reproducing any best practice project on a larger scale is challenging and requires multiple factors to succeed,72 this approach to standardisation and multidisciplinary team virtual review is likely to be useful in many other areas of medicine.

Table 3

Utilisation and annual cost results for the virtual and traditional fracture clinics

VFC	Utilisation per staff member				Time worked per day on VFC				TFC	Utilisation per staff member				Time worked per day on TFC
Simulation object	Lower 95%	Avg. %	Upper 95%	shift time (min)	Lower 95%	Avg. (min)	Upper 95%		Simulation Object	Lower 95%	Avg %	Upper 95%	Shift time (min)	Lower 95%	Avg. (min)	Upper 95%
Admin 1	36	40	44	45	16	18	20		Admin 1	21	23	24	75	16	17	18
Admin 2	9	9	10	300	28	28	29		Admin 2	12	14	16	300	36	42	48
Typists (1)	14	18	23	480	68	88	108		Typists (4)	16	18	19	480×4	307	346	365
Nurse (1)	57	58	59	480	275	279	283		Nurse (3)	52	54	55	480×3	998	1037	1056
Consultant ortho (1)	17	21	26	240	40	51	62		Consultant ortho (3)	74	75	77	240×3	533	540	554

VFC: Resource	Performance Measure	Lower 95% CI	Average	Upper 95% CI		TFC: Resource	Performance Measure	Lower 95% CI	Average	Upper 95% CI
Consultant ortho	Total cost	15 199.31	19 375.79	23 552.26		Consultant ortho	Total cost	145 107.76	147 754.65	150 401.54
Nurse	Total cost	34 960.53	35 482.45	36 004.38		Nurse	Total cost	42 756.66	43 785.48	44 814.29
Typists	Total cost	4764.26	6181.82	7599.38		Typists	Total cost	21 665.30	23 945.39	26 225.47
Admin 2	Total cost	1521.80	1572.91	1624.01		Admin 2	Total cost	13 230.47	15 155.64	17 080.82
Admin 1	Total cost	1257.40	1393.10	1528.80		Admin 1	Total cost	862.75	932.78	1002.80
	TOTAL COST	57 703.31	64 006.07	70 308.83			TOTAL COST	223 622.94	231 573.93	239 524.92
GRI VFC referrals	Patients	4362.21	4496.67	4631.13		ED and MIU	Patients	6273.09	6291.00	6308.91
VFC	Per patient	13.23	14.23	15.18		TFC	Per patient	35.65	36.81	37.97

GRI, Glasgow Royal Infirmary; TFC, traditional fracture clinic; VFC, virtual fracture clinic.