2.1 Study Overview

This study was conducted in three steps (Fig. 1). A base case has been established previously for the Charing Cross Hospital infusion suite, London, UK, given their capacity and resources available at the time of data collection in March 2021 [11], and was used as reference.

Fig. 1

Study overview: A three-step approach was used to conduct the study, by defining and simulating relevant clinical scenarios, analysing KPIs, and identifying mitigation measures for improving MS care. aThis scenario comprised scheduling approval time and additional measures that would result in infusion scheduling delay (see “Methods” Sect. 2.3.6 for details). bFor example, simulate switch to home-based care with non-IV MS therapy. IV intravenous, MS multiple sclerosis, SC subcutaneous, KPI key performance indicator

Five hypothetical but clinically plausible scenarios were specified by the first author of this article, Consultant Neurologist Prof. Richard Nicholas, given the experience of leading a large tertiary referral neurology unit. These scenarios, which can be viewed as sensitivity analyses to the original publication, listed the key factors expected to influence (both exacerbate and alleviate) the demand for infusions in the coming years. The study was initiated while the COVID-19 pandemic was ongoing; therefore, these scenarios reflected the social distancing measures that were implemented at that time (and that are likely to be implemented when we are confronted with a future pandemic).

The primary focus was to predict the moment in time when patients would experience high-risk treatment delays, and thus be at risk of disease progression and/or exacerbation [7, 8]. This key outcome was defined as average waiting time for the next due infusions of ≥ 30 days (compared with the approved treatment schedule indicated in the Summary of Product Characteristics [SmPC]). Scenarios were also compared using time and cost key performance indicators (KPIs).

Finally, potential mitigation measures focusing on either reducing demand or improving capacity were identified (see Fig. 1).

2.2 The Decision Support Tool

The ENTIMOS discrete event simulation model (described in detail by Lacinova et al. [11]) has been developed as a decision-support tool using information about patient flow, MS infusion care delivery pathways and site processes collected from site administrators, nurses, pharmacists and clinicians at the London Charing Cross Hospital infusion suite, as of March 2021. Data on centre and treatment settings, such as the number of infusion chairs, patients and nurses, and treatments administered, were collected to build a process flow. The main output data are patient waiting time, as well as costs. For a full list of input and output parameters, see Lacinova et al. [11]. The clinical accuracy and relevance of the tool were validated by a consultant neurologist (Prof. Richard Nicholas).

2.3 Clinical Scenarios

We defined a set of clinically plausible scenarios which addressed current or anticipated challenges at infusion suites (scenarios 3 and 4) as well as plausible strategies to mitigate them (scenarios 1, 2 and 5) without increasing the resources such as number of chairs or reducing the patient number. In cases where high-risk treatment delays were nevertheless predicted, we used the model to estimate how many more resources will be needed or how many patients would need to be discharged to maintain adequate, continuous care. If different measures result in reduced patient waiting time, labour costs can be considered as an additional parameter to select the optimal mitigation strategy.

2.3.1 Base Case

The base case scenario used real-world inputs from the London Charing Cross Hospital infusion suite, as of March 2021. Inputs, such as the number of infusion chairs, patients, nurses, treatments used, etc., were used to simulate a realistic expectation of patient number, queue size and waiting times over a simulation time of 3 years (scenario 1.2 was also simulated over 5 years). Charing Cross Hospital infusion suite is a neurology-only suite that has 12 infusion chairs, employs ten staff nurses and treats 860 patients with MS and 170 patients with non-MS conditions; further model inputs and results of the base case have been discussed in full elsewhere [11]. In this study, the base case results were used as a reference (baseline data) for assessing the potential impact of the other five scenarios on MS care.

2.3.2 Scenario 1: IV to SC Treatment Switch

Scenario 1.1 was designed to evaluate the effect of reducing the demand for infusions by switching patients from IV to SC therapy in the hospital setting. In this scenario, 50% of patients receiving a monthly IV MS HET (natalizumab) were switched to SC administration (n = 154) of the same therapeutic agent, with otherwise similar characteristics. Only 50% of patients were chosen for treatment switch because, as revealed in a recent qualitative study to determine the drivers of patient treatment preference, which involved telephone interviews with people living with relapsing MS and healthcare professionals, ~ 40% of patients with negative experiences would choose to avoid treatment that involves injection (either SC or IV) or self-administration [15]. All new natalizumab patients were assumed to be prescribed an SC version of this drug. As this drug requires supervision from the medical personnel during administration, patients were assumed to occupy an infusion chair regardless of the administration route; therefore, this scenario simulated the effect of difference in treatment duration (1–2 h for IV and 0.1–0.7 h for SC treatment) [16]. The pre-treatment for both routes of administration was assumed to be the same. The posology inputs used for the SC version of MS HET are described in Supplementary Table 2 (see the electronic supplementary material). To determine the long-term feasibility of this scenario, a sensitivity analysis of the same settings was simulated over a longer (5-year) horizon (scenario 1.2).

2.3.3 Scenario 2: Weekend Opening

In scenario 2, the infusion service included weekends (i.e. extending the opening times from 5 to 7 days a week) to model centre capacity in line with COVID-19 response measures.

2.3.4 Scenario 3: Weekend Opening and Chair Reduction

To simulate social distancing measures that were required during the COVID-19 pandemic and may be needed in future under assumed epidemic/pandemic conditions, the third scenario reduced the number of infusion chairs by 50% (from 12 to 6) on top of the weekend opening.

2.3.5 Scenario 4: Increased Demand for IV Infusions

The fourth scenario evaluated the impact of a hypothetical IV HET that might be approved and reimbursed in the UK for a chronic, progressive neurodegenerative disease (CPND) that currently lacks an HET. The demand for IV infusions from CPND patients matched the demand from patients with MS (860 existing patients at simulation start and seven new patients per week, on the top of MS patients). It was assumed, based on clinical opinion, that a high proportion (35%) of patients with CPND would experience adverse events that would require them to skip the subsequent infusion [17]. This scenario did not consider the resources needed to conduct imaging prior to infusion and monitor patients recovering from the adverse events of the CPND HET. Supplementary Table 3 describes the posology inputs used to simulate the CPND IV case (see the electronic supplementary material).

2.3.6 Scenario 5: Infusion Scheduling Delays for New Patients with MS

The fifth scenario studied the impact of infusion scheduling delays. In routine clinical practice, infusion delays are sometimes employed as a gatekeeping tactic to shorten the patient queue. The clinics postpone the start of therapy for newly diagnosed patients, allowing for more time to schedule and treat existing patients. The tactical delays can be applied to multiple steps in the continuum of patient care, namely screening of the patient’s blood results, obtaining consent for the infusion, clinical review and multidisciplinary team approval, infusion scheduling, and dose screen and approval by pharmacy. While clinical review and multidisciplinary team approval were deemed outside of control of the infusion suite administration, all other factors listed above were included in scenario 5.

Based on clinical experience from the Charing Cross Hospital in London, this scenario assumed that after 12 months of simulation, the scheduling delay for new patients will have increased from an average 12 weeks (range 4–20 weeks) to 23 weeks (range 20–26 weeks).

2.4 Simulation Time

To reflect a typical planning horizon at an infusion centre, all scenarios were run for a simulation time of 3 years. An additional sensitivity analysis over a period of 5 years was run for scenario 1.2, to interrogate the long-term feasibility of patient switch from IV to SC MS HET when treatment would still be administered within the infusion suite.

2.5 Model Outputs: KPIs

To evaluate the efficiency of the infusion centre, three KPIs were used: waiting time (days), time to high-risk treatment delays (month) and direct costs (labour) (£). Patient waiting time refers to the number of days patients must wait to receive their next due infusion and is calculated as the average on a monthly basis. Time to high-risk treatment delays refers to the point in time when patients face treatment delays that put them at risk of disease progression and/or exacerbation, and is expressed as the month in which the average patient waiting time reaches ≥ 30 days (compared with the approved treatment schedule indicated in the SmPC) [7, 8]. Reaching this point can be considered a “system compromise” [11]. Labour costs, based on studies by Franken et al., North et al., Perry, and Schmier et al. are a direct result of nurse hours spent on infusion administration and management of infusion-related adverse reactions and are expressed as the sum accrued over the entire simulation [18,19,20,21]. Fixed costs for acquiring new chairs and other operational costs (e.g. additional infusion rooms if existing rooms are at capacity) were not considered in the model for simplicity. The entire list of outputs that can be generated by the simulation tool has been described by Lacinova et al. [11].

2.6 Mitigation Measures

Mitigation measures are adjustments in the resources or patient numbers which prevent patient waiting time from reaching a threshold of 30 days, therefore avoiding high-risk treatment delays.

This analysis included three mitigation measures: discharging new patients with MS from the infusion centre, discharging existing patients with MS or adding infusion chairs. Patients discharged from the infusion centre could either be referred to the care of another MS infusion centre or switched to oral or SC treatment options that would allow self-administered and/or home-based care. Discharged patients are the ones for whom there is effectively no capacity in the infusion centre and, therefore, a new care arrangement needs to be found for them. In the technical specifications of the tool (see [11]), discharge is represented through a “switch-out” option that removes patients from the simulation. Quantification of mitigation strategies was only completed for those scenarios where high-risk treatment delays were reached within the simulation time frame.