To develop the costing framework, we first identified the different phases of the 3DP process. The 3DP manufacturing phases were based on our previous experience in pharmaceutical 3DP and reflect the general manufacturing phases. Subsequently, we used the costing framework for a micro-costing study with several scenarios on M3DICORT. The formulation of M3DICORT was developed in previous research using fused deposition modeling (FDM) 3DP, and is described in detail elsewhere [10]. The study was conducted at the academic hospital pharmacy of the Erasmus Medical Center (MC), which has a total surface area of 2654 m2, 230 full-time equivalent (FTE) personnel, and is hence the largest academic hospital pharmacy in the Netherlands.

The costing framework includes three main phases in the 3DP process: (i) pre-printing, (ii) printing, and (iii) post-printing. For each phase, cost categories were identified including personnel, materials, equipment, facility, and quality assurance (QA). These cost categories are based on an existing manufacturing framework for Advanced Therapy Medicinal Products [19].

In phase 1, in the pre-printing phase, a digital 3D model is designed and converted to a readable file for the 3D printer to produce the desired shape. 3D models are designed using generic computer-aided design (CAD) software. These models can be directly accessed through ready-to-use digital designs, available for download in the .stl file format. Once obtained, an .stl file is converted into a geometric-code (g-code) by specialized 3DP software. This g-code entails the precise movements required by the printer, guiding it to accurately reproduce the 3D model layer by layer. The design of the final product is a crucial step in pharmaceutical 3DP as it influences the release profile and supports the manufacturing of specific formulation dosages [5, 10, 20, 21]. The detailed manufacturing steps, including the design of the M3DICORT used in this study are described elsewhere [10]. Once designed, the g-code can be used over and over.

During the pre-printing phase, drug-loaded filaments, which serve as the ink for the 3D printer, are manufactured. To this end, raw pharmaceutical powders are mixed and melted by means of hot melt extrusion, resulting in a printable drug-loaded filament. Currently, drug-loaded filaments are not commercially available and have to be manufactured in-house.

In phase 2, the printing phase, the drug-loaded filament is manually fed into the 3D printer, after which the tablets are printed layer-by-layer until the final products are formed.

Finally, in phase 3, the post-printing phase, the printed tablets undergo rigorous quality checks and packaging. In pharmaceutical manufacturing lines, quality control (QC) and quality assurance (QA) systems play a crucial role to guarantee and maintain high-quality standards. In terms of QC, both pharmaceutical manufacturing lines, raw pharmaceutical products and final products, need to be validated to ensure consistent quality. Typically, this is realized by analyzing drug contents, stability, uniformity, homogeneity, and mechanical properties of the final product. Before distribution, batches need to be formally released by a qualified person. In addition, during this phase, any used equipment is cleaned, and the printed tablets are packaged, labeled, and stored before transportation. QA involves an active quality management system which proactively prevents product defects and deviations. For instance, by conducting risk assessments, developing standard operating procedures, and comprehensive documentation to ensure that final products meet with predefined quality parameters. As hospital pharmacies generally have QA systems in place, this study limits QA costs to costs related to QA personnel only.

For each of the three printing phases, the framework further distinguishes between fixed and variable costs within the following five cost categories: personnel, materials, equipment, facility, and QA. Fixed costs are defined as all costs that are not correlated with a change in the number of printed products or services, while variable costs are costs that change in proportion to manufacturing output.

To quantify manufacturing costs per identified phase and category, we applied the costing framework following an activity-based costing methodology with which costs were estimated as a function of volume and price per volume [22]. We assumed a manufacturer perspective, meaning that only costs incurred by the developer, in this study the Erasmus MC hospital pharmacy, were considered. Consequently, costs for diagnosis, medical interventions, and logistics were not included [23].

Fixed costs are defined as the sum of the costs per year divided by the amount of tablets produced annually. For variable costs, we took an opportunity cost approach to allocate variable cost items. Hence, we valued all time and resources that were spent manufacturing the product of interest and therefore could not be used for other purposes. Historic prices or costs were converted and adjusted for inflation in adherence to the guideline of the Dutch National Healthcare institute (Zorginstituut Nederland; ZIN) on economic evaluations studies [24]. All costs stated here are expressed in 2022 euros (€).

All expenses associated with materials were identified as variable costs and included raw pharmaceutical ingredients, disposable items (such as face masks and gloves), packaging, and labelling materials. Resource use frequencies for these items were based on the M3DICORT development study, with detailed quantities used for this case study provided in the supplementary appendix [10]. All these items were valued using retail prices, derived from quotations received by the hospital pharmacy.

All expenses associated with equipment were identified as fixed costs. Equipment needed to manufacture M3DICORT includes an FDM 3D printer, a hot melt extruder to produce filaments, a scale for weighing, a rotor stator mixer for mixing, and a label printer. Costs for these items were based on their initial acquisition price and an annual depreciation. This depreciation was calculated by dividing the purchase price by an annuity factor that incorporated the lifespan of the respective equipment. The latter was uniformly assumed to be 10 years for all equipment items [25]. In addition, a standard interest rate of 4.2%, and an annual maintenance markup of 5% of the purchase price was considered for all equipment [19]. The annual expenditure for each device was further adjusted on the basis of the printing time per tablet.

For the case study, we assumed that manufacturing is performed by a pharmacy technician and batch release by a qualified person (QP). It is a legal requirement for pharmaceutical manufacturers in the Netherlands to have a QP [26]. This QP is responsible for certifying the release of every batch of a pharmaceutical product. By releasing a batch, a QP ensures that the batch complies to applicable laws, requirements of the marketing authorization, and with good manufacturing practices. QPs are typically pharmacists, with additional training on quality of pharmaceutical products. Time needed for weighing, mixing, extruding, printing, cleaning, labeling, and packaging was recorded with a stopwatch. Times for batch release were based on expert opinion and experience, as 3D-printed medication is not formally released yet. Batch release is a quality check performed by a qualified pharmacist after production. If the batch is formally signed and released by the qualified person, it can be distributed, and the batch is then “released.” Assumed working hours for personnel were 1558 h per year, based on a 36-h workweek, including holidays [27]. Personnel times were either valued with their respective salaries gathered from the collective labor agreement from The Dutch Federation of University Medical Centres (Nederlandse Federatie van Universitair Medische Centra; NFU) 2022–2023 for pharmacy staff or with references prices from the Dutch costing manual [28]. A correction of 39% was applied to salaries, which included holiday payments and social funds [29]. Personnel costs related to cleaning, loading ink, and QA were considered fixed costs, as they are not influenced by the manufacturing output. Other personnel costs were considered variable. Total FTEs of the 3DP team at our site consist of 3.04% compared with the total pharmacy’s FTEs. CAD software is available license-free and 3DP software are usually delivered together with a 3D printer. Simple shapes, such as tablet designs, are typically generated for pharmaceutical 3DP. We have not applied costs for designing since the final design is identified during formulation development. Revision control of modelling and slicing software were not considered. Once dimensions and slicing parameters are established and logged, they can be generated easily with other modelling and slicing software.

For facility costs, estimates of surface areas needed for pre-printing, printing, and post-printing rooms were made, including non-production rooms. Internal financial data demonstrate general costs of production rooms of €1467.87 per m2/year, and for non-production rooms at €698.40 m2/year. Production room costs were classified as variable costs, and non-production rooms as fixed. For the facility, we assumed that a total of 26 m2 was needed for manufacturing and storage and 7.6 m2 for offices, on the basis of internal pharmacy area data. On the basis of the 3.04% FTE, the share in m2 for the non-manufacturing areas (e.g., office, hallway, bathrooms) of the 3D facility was assumed to be the same value in relation to the whole facility.

In the base-case analysis we assumed a manufacturing output of an arbitrary 75% but varied this in scenario analyses to acknowledge variability in manufacturing output. For a precise analysis, we employed a “follow the technician” approach to monitor the duration of all activities across the three phases. More specifically, we measured start and stop time of all activities using a stopwatch, following an experienced technician. The production of 250 cm filament, which can be used to print 164 tablets, took 105 min. This includes weighing, mixing, and extruding. Hot melt extrusion, used to produce the filaments, is also known as continuous manufacturing [30]. In theory, by continuously adding powder to the extruder, an endless length of filament can be produced, enabling tablet printing until the filament is exhausted. Once produced, the filament is fed to the printer, which becomes the production rate-limiting step of the production process with a rate limited to 120 tablets per h. Assuming a facility run time per year is equal to personnel working hours, (i.e., 1558 h per year), a maximum manufacturing output of 186,960 tablets per year could potentially be realized.

Cost of quality management software is not included as all hospital and compounding pharmacies already have a quality management system in place. QC is a component of QA and its incurred fixed costs were considered for validating the manufacturing process, starting material analysis, and mandatory pharmacopeial tests for oral dosage forms. The latter encompass assessments of content uniformity, dissolution, friability, hardness, impurity testing, and a stability test [31]. The FTE of the QA department reflect the QA department of the Erasmus MC hospital pharmacy.

Per category, QA costs were categorized on the basis of their association either with ink manufacturing or 3D printing the final product. QA expenses specific to ink production were allocated to ink costs, while other attributes were attributed to 3DP. For costs that were related to both categories (i.e., ink production and 3DP), such as some facility expenses, a 50/50% allocation was applied.

Several input parameters for the M3DICORT case study are summarized in Table 1, and more detailed costs of personnel, materials, equipment, facility, and QA are provided in the supplementary material for all scenarios.

In three scenario analyses we assessed how different assumptions on the operational factors from the base-case would affect M3DICORT manufacturing costs. We specifically varied loss of materials during printing, manufacturing output, batch size, and QA share dedicated to M3DICORT quality.

The worst-case scenario was conceptualized to reflect the challenging operational environment. Here, the material loss during the printing process was highest, set at 20%, indicating significant inefficiencies. The manufacturing output was set at 50%, reflecting potential constraints in demand or operational limitations. This scenario also assumed that the printer would be dedicated to a single product, highlighting limited resource adaptability. Moreover, the QA department’s engagement was extensive, with 10% of the total full-time staff focused on M3DICORT tablet testing.

Conversely, the best-case scenario assumed an ideal setting where operational processes are optimized for maximum efficiency. Material loss was reduced to just 1%, indicating highly effective resource use. The primary cause of material loss is residue in the extruder, but this is mitigated with better process control, leading to reduced waste. With higher production output, the relative material loss becomes negligible, supporting the assumption of a 1% loss rate. Manufacturing output was therefore set at 100% of the maximum printing capacity, assuming full utilization of the facility. As a result, 186,960 tablets can be produced annually. This scenario allowed for the printer to be used for two different products, showcasing an efficient allocation of equipment. In all but the best-case scenario, there is an excess capacity available on the 3D printing recourse, allowing for more products to be printed. However, in the best-case scenario, the equipment is fully utilized at 100%, which maximizes output. Consequently, fixed costs are spread across a larger number of units, reducing the per-tablet cost. Additionally, in a typical manufacturing setup, facilities are shared for the production of multiple products. By increasing the variety of products produced, economies of scale are achieved, further reducing costs per unit. In this best-case scenario, we assumed a 50% reduction in the cost per tablet for equipment, facility, and QA expenses. Furthermore, the involvement of the QA department was set at 5% of the total staff, implying a streamlined and effective quality assurance process. The total costs to produce a single M3DICORT tablet for each scenario, except the scaling scenario, was calculated by adding the costs of manufacturing the ink to the costs of 3D printing the M3DICORT tablets.

Finally, in the scaling scenario we calculated manufacturing costs under the assumption that ready-to-use pharmaceutical ink becomes commercially available. The aim was to mirror a realistic, balanced operational model with strategic procurement of filaments that fulfill quality standards and can thus directly be used for printing. We maintained the material loss during printing at 0–20%. The manufacturing output was 75%, relative to the maximum printing rate of 120 tablets per h. This increased output was driven by the assumption that pharmacies would purchase ready-to-use pharmaceutical inks, instead of manufacturing the filament materials in-house. By acquiring these pre-filled inks, pharmacists could save considerable time. Since purchased filaments would already fulfill certain quality criteria, the QA department’s involvement in this scenario was set at 5%, similar to the best-case scenario assumption. Here, we assumed that the manufacturing site is shared by 2–4 products, translating to a 50–75% reduction in costs per product in categories of QA, facility, and equipment. We assumed that due to volume discount, materials costs are reduced by 30%. When producing larger amounts of ink, weighing and mixing cost more time, while extruding and cleaning times remain fixed. We therefore assumed that due to scaling effects, personnel costs are reduced by 50%, compared with the base case scenario. The total costs for a 3D printed tablet in this scenario were calculated by adding the ink manufacturing costs to the costs of printing one tablet of the worst-case and best-case scenario.