Recipe Development Process Re-Design with ANSI/ISA-88 Batch Control Standard in the Pharmaceutical Industry

Reducing time-to-market is one of the most challenging tasks that pharmaceutical companies deal with. In this sense, the recipe development process represents one of the most critical phases. Multi-site production companies require an efficient recipe development model, with a robust modular structure, which must be appropriately shared among local laboratories and plants. To this extent, the ANSI/ISA-88 batch manufacturing standard, rising in the context of process control and automation, is rapidly becoming widely used in pharmaceutical companies. This paper presents a step-by-step approach to assessing the compliance to the ANSI/ISA-88 standard along with a BPM-oriented methodology applicable to the re-design of any generic recipe development process. Redesigning a recipe development process is a complex activity and can mask several pitfalls and criticalities. Thus, along with the methodology, some general evidence and suggestions are provided based on the experience of a project carried out in a large multinational pharmaceutical company.


Introduction
As most recent pharmaceutical companies' trends attests, the time needed for the development process of a new drug is continuously speeding up, especially concerning the post-approval of clinical phases [1,2,3]. Interest in process improvement and optimization has been a partial response to the observed trend of stagnant profit margins and the current growth of competition in this sector, regarding both general and patented drugs. Thus, reducing time-to-market and recipe development intervals are some of the principal features and challenges that pharmaceutical companies are dealing with [4,5]. Multinational corporations in particular require an efficient recipe development model, with a robust modular structure, which must be appropriately shared among local laboratories and plants. ANSI/ISA-88, in this sense, can be considered a valid contribution to reaching the so-called "golden batch", providing a mature framework where information-streams properly integrate all the key actors involved. However, redesigning such a complex process, which is crossfunctional and affects the main end-to-end development phases, can mask several pitfalls and difficulties.
ANSI/ISA-88 is rapidly becoming a widely used standard in batch manufacturing, especially in pharmaceutical companies [6,7,8,9,10]. The standard, first approved in 1995 by the International Society of Automation (ISA) and then updated at a later stage in 2010, arises in the context of batch process control and automation. However, the recent growth of interest in this standard on the part of multinational firms can be attributed to the wellestablished set of models and terminology [11,12,13,14,15] provided, which proved to be useful for managing the end-to-end recipe development processes (from R&D to commercial production).
Adopting a new product development model, according to ANSI/ISA-88 guidelines, affects both the conceptual and documental level. The American standard defines the recipe contents, differentiating between equipment independent recipes, which are managed by scientists at an enterprise/laboratory level, and equipment dependent recipes, which are enriched with specific information about plant and process cells required for commercial production [16,17,18]. Moreover, when dealing with numerous/different manufacturing sites, most difficulties in implementing the commercial production of drugs arise during the transformation of the generic recipe (which specifies product characteristics and overlooks processing equipment parameters) into a "master recipe", calibrated to the specific machinery of a given plant. The reasonable need to differentiate the form and content of the documentation concerning different phases of the development process is clear, especially when we consider the scale-up required while passing from a pilot plant to a production plant: the parameters of materials and equipment usually change when processing a 10-litre batch or a 20,000 one. Despite this, all the information must be efficiently communicated between the main actors in the recipe development-chain. The transition from one recipe to another needs to be rationalized, and the knowledge acquired in every transformation activity must be accurately managed in order to reduce the overall time spent before commercial production, thus facilitating further development processes [19,20]. While the recipes and the process model are well defined, the main area lacking in the current ANSI/ISA-88 standard is the extent and content of the guidelines provided for the implementation of the general-to-specific recipe transformation.
Thus, in order to fully achieve the benefits of ANSI/ISA-88 implementation, this paper presents an integrative methodology, applicable to the re-design of any generic recipe development process in compliance with the ANSI/ISA-88 standard. The aim is to provide a step-bystep approach to assess the standard compliance and perform a recipe development process re-design, by leveraging process oriented techniques, such as a Business Process Management approach and the Business Process Model Notation 2.0 (BPMN 2.0) [21,22]. Any effective process improvement initiative requires a clear definition of the as-is and to-be scenario, in order to analyse the gaps and identify a proper road map to reach the expected results. In addition, a standard and appropriate notation is required to physically represent the flows of activities and documentation. Hereafter, paragraph 4 and 5 will describe in detail the path of the proposed method and a case study of the implementation of the method at a large pharmaceutical multinational company. Brief comments on the ANSI/ISA-88 standard (paragraph 2) and the BPM approach (paragraph 3) are also included, in order to introduce these concepts. Finally, the results and the conclusion (paragraph 6) are presented to highlight the benefits of this methodology and to discuss some common misalignments that may arise while assessing their compliance to the ANSI/ISA-88 standard.
2. ANSI/ISA-88 standard ANSI/ISA-88 is an international standard, which widely addresses all the main features related to batch manufacturing processes, describing procedures and equipment requirements during all the development phases, from the enterprise/laboratory level to commercial production [23,24]. It is important to point out that reference models and guidelines provided by the ISA standard are not to be considered strictly normative (some clauses are informative as well). This ensures a good level of flexibility in representing the current structure in an "ANSI/ISA-88 way", giving also the opportunity to collapse and expand parts of the reference model to better suit peculiar batch manufacturing cases. Hence, considering that the models provided are sufficiently abstract that they may be applied to a wide variety of batch manufacturing implementations, and that compliance certifications are usually performed by independent organizations, the assessment of the degree of ANSI/ISA-88 compliance is essentially to be meant as an alignment on the use of the terminology and models definition. Whereby a partial compliance exists, specifications and implementations must be described, and areas of noncompliance are to be identified. Since a full description of the standard is out of the scope of this article, a brief description is provided only for those elements that are significant for the described case, ensuring a clear understanding of the contents and omitting the non-essential information.

ANSI/ISA-88 structure
The ANSI/ISA-88 standard is divided into five parts that respectively deal with different features of the batch manufacturing process and control. Part 1 (Models and Terminology) [25]  Adopting a process oriented approach and using a visual representation ensures a better control of performance and enables the detection of those existing inefficiencies that are hardly recognizable. Corporate process flows are often made of many figures at different stages, and the aggregation of several complex phases in few synthetic diagrams should be avoided. Indeed, a clear visibility of the entire stream of events is required in order to properly understand the linkages among activities, identify the strengths and weakness of a process and, eventually, facilitate interaction and integration with outsourcers [34,35] or with suppliers [36]. Additionally, process mapping and modelling can be considered one of the first steps for almost every improvement methodology that take place when analysing processes [37,38]. Both to identify opportunities and constraints, or even to handle a correct change management process, an as-is analysis and mapping are required. Having a clear understanding of the current stream of activities leads to a correct definition of an optimal to-be scenario. Once the desired future state is also clear, it is finally possible to perform a gap analysis and define an improvement implementation road map in order to achieve the expected results.
In each case, this cycle of improvement activities must be supported by a consistent graphical notation, capable of correctly representing the main subjects, events, activities and process rooting. Business Process Model Notation (BPMN 2.0) is considered one of the most efficient standards [33,39] to represent in a graphical way a varied range of business and service processes. BPMN is an intuitive notation, suitable also for non-technicians figures, but capable of either representing complex process semantics or easing B2B and internal coordination. The activities of the flow and process representations are simple to understand, and clear communication is enhanced between who develops a new model and the final users who are affected by the analysed process. Furthermore, BPMN models consists of basic diagrams (BPD, Business Process Diagrams) which are built from a limited set of graphical elements, that are flow objects, connection objects, swim lanes and artefacts [40]. The following paragraphs explain why such a process oriented approach and flow-chart technique were chosen for the proposed methodology, highlighting the advantages obtained in the case study while assessing the ANSI/ISA-88 compliance.

Proposal for an ANSI/ISA-88 compliance assessment and roadmap identification
This paragraph illustrates the proposed methodological approach to carryng out an ANSI/ISA-88 assessment and provides a roadmap for full compliance, leveraging the above-mentioned tools of the Business Process Management approach. The aim of the method is to provide a reference framework which allows practitioners to perform an ANSI/ISA-88 compliance assessment in different contexts.
The motivations for such an effort lie in the need to support the implementation of ANSI/ISA-88 guidelines with a methodological and structured approach. Proven experiences have evidenced that implementing a new product development process in pharmaceutical companies is a difficult task, hiding several inefficiencies or inconsistencies concerning the flow of activities. ANSI/ISA-88 certainly provides a strong reference model, but implementing an effective transformation from the as-is process model to a compliant to-be one can be quite challenging for a company. The introduction of a BPM method at this stage can help identify gaps and define the road map, while BPM notation is a practical tool which can effectively provide clear visibility of the process and documental flows, enabling wiser decisions. To this extent, the proposed method was applied and validated on the case of a multinational pharmaceutical enterprise. We identify four main sequential phases to perform the activities in a structured way: • Phase 1: initial scope and input definition. Moreover, two additional cross-activities must be run to support the implementation of the method itself: • Core team members' education in ANSI/ISA-88 standard. • Convergence meetings, following a project management methodology.
Phase 1: this phase covers the initial scope and input definition, where project boundaries are defined and preliminary data gathering is performed, in order to discuss the project main objectives, together with the identification of expected benefits from applying the standard.
Phase 2: the results of phase 1 are used to identify ANSI/ISA-88 applicable / not applicable areas and define specific process requirements. Indeed, as already mentioned, the ANSI/ISA-88 standard is not strictly normative and some areas of the reference model can be collapsed or expanded due to peculiarities of the recipe development process.
In order to point out the key information of the as-is process, accountable personnel, with deep knowledge of the company's process architecture and documentation, need to be involved. These core team members should belong to the corporate functions affecting the recipe development process, such as the introduction of new products, the technical integration of drug product development, process validation, manufacturing execution systems, operation quality, quality assurance, and IT/automation. Data gathered within phase 2 will feed the subsequent steps, whereby all the fragmented information must be consolidated to provide clear visibility of the end-to-end stream process and to identify the requirements of the specific case, thus restricting the focus only to applicable areas of the standard reference model. Examples of collapsible and extendable areas are: recipe types, recipe information structure, recipe graphical representation, the transformation process of equipment-independent to master recipes, and contents of the equipmentindependent recipe object model.
Moreover, as mentioned above, during this phase and the subsequent ones, training on ANSI/ISA-88 to all main figures interviewed is required, in order to ease the asis/to-be transformation process model. Phase 3: once the main data are collected and ANSI/ISA-88 applicable/ not applicable areas are defined, the process can be properly analysed and mapped in phase three. First, an as-is recipe development process analysis is required. In order to depict this information in diagrams and have a visual representation of the process, the proposed method takes advantage of the BPMN 2.0, which provides an effective process modelling tool. Considering the complexity of the activities performed by different actors, operating both at a corporate, R&D and plant level, process end-to-end visibility is essential to defining the degree of similarity to the ANSI/ISA-88 standard's expectations. The reason why BPMN notation was chosen rather than other process flow methods is that it belongs to both the specific and adaptive set of graphical elements, which can be used to represent the process. Moreover, this tool gives us the possibility of using diagrams to create simulations and automated workflows in further phases. Another important reason for this choice is the increasing adoption of BPMN as a common standard in most companies for process modelling, workflow automation and direct execution. These considerations become more valuable considering that ANSI/ISA-88 initiatives often take place in large multinational multi-location companies, where the use of standards among different production plants enhance knowledge sharing [41,42,43] and information exchanges.
This methodology allows us to identify the misalignments on the main topics of ANSI/ISA-88, concerning recipe structure, process elements, graphical representation and libraries.

Phase 4:
At the end of the previously described phases, it is possible to design the processes in a to-be version, highlighting similarities with the current model and detecting gaps to be covered. During this phase, actions to bridge the gaps and highlight priorities must be identified and formalized into a specific road map.
In more detail, the main topics to be addressed are summarized as follows: • Alignment on ANSI/ISA-88 terminology (process model, physical model and procedural model). • Process elements (process stage, operation and action) and process element standard libraries definition, following ANSI/ISA-88 data architecture. • General, site, master and control recipe contents and structure. • Graphical representation of equipment independent and dependent processes. • General-to-master recipe transformation process activities and document definition.
The identified road map represents the main deliverable of the proposed method.
Together with the explanation of the method, our proposal also provides a list of recommendations derived from the experience obtained from a real case implementation: redesigning a recipe development process is a complex activity, and can mask several pitfalls and criticalities. In order to facilitate the adoption of the proposed method and correctly implement the different phases, some general evidence and suggestions are provided. Below the most relevant considerations supporting phase 3 and phase 4 of the proposed reference model are indicated: • Generate templates for ANSI/ISA-88 compliant documents from company existing ones (Phase 3): considering that in pharmaceutical companies recipe development processes are well established by standard operative procedures, one of the most critical tasks is the effective representation of the future recipe development process, together with the transition itself. The proposed method suggests starting the redesign process by transforming existing documents related to a new product development (NPD) process of a specific pharmaceutical product, in order to generate standard ANSI/ISA-88 documents, such as a general and master recipe, to be used as a reference. By creating these new documents, a direct link between the main information of the current and future documentation is possible and the degree of similarity to the ANSI/ISA-88 standard is verified. Once the gaps between processes and documentation are clear, it is possible to outline short term and long term opportunities for improvement.
• Ease the recipe development to-be process identification with BMPN notation (Phase 3): the recipe development process can be sensibly improved by the adoption of shared documents and the use of ad-hoc tools, such as a process element library or a set of transform components. During the assessment process, a to-be recipe development process has to be identified considering the aforementioned standard documents and tools. BPMN notation and process oriented methodologies have a significant impact on re-designing the recipe development procedures, providing a clear picture of the as-is situation and helping identify steps to reach an ANSI/ISA-88 full compliance. The use of BPMN to map the recipe development process can be considered a critical step of a repeatable approach for assessing the degree of similarity to ANSI/ISA-88 standard models and procedures. • Establish an education programme on ANSI/ISA-88 (Phase 3/4): in order to raise awareness of ANSI/ISA-88 benefits across the organization, after providing education on the method, it is advisable to continue the education process even after the assessment phase, especially if the company production processes are distributed into different plants.

• Standardize process elements and share process element libraries (Phase 3/4):
the definition of standard elements to describe the recipe development process is only a starting point. These elements, used to represent the recipe flow, should be exclusively selected from a shared library with a strong modular structure. Process actions represent the main building block to start from, in order to define the to-be standard process operations and stages. Identifying all the elements and the perfect linkages between them is a significant strategy for improving the general recipe standardization. As discussed above, a great amount of effort is made to transform the general recipe into a master recipe. This difficulty is generally due to the absence of standard contents and graphical representation used in all the documents involved in the recipe development process. Knowledge and information sharing is essential to reduce the time-to-market of new products. To overcome this problem, the proposed method suggests creating a set of stencils (e.g., in Microsoft Visio), according to the notation suggested by the ANSI/ISA-88 standard. • Define and adopt the general recipe and master recipe as standard documents (Phase 4): company documents are often not aligned to the ANSI/ISA-88 standard in terms of contents and graphical representation. The creation of specific templates for the general and master recipe, with the intent of providing reference formats for new product recipe development, is recommended. Furthermore, the general recipe template should be shared across the production sites and R&D functions, facilitating their usage and improving the level of standardization of these documents. • Standardize the transform components (Phase 4): one of the main difficulties is to define standard transform components, which are a set of equipment parameters and instructions that allow the transformation of process operations, described in the general recipe, into equipment settings and controls. Usually R&D performs the product development process on small-scale pilot plants, which can use completely different technologies from those of production plants. Farther on, the transformation process needs a long time to be performed, including deep characterization studies. Moreover, when product transfers among production sites are planned, the differences between technologies used in each plant become crucial. A consistent part of the transfer process is spent in customizing equipment parameters and working instructions for the different technologies used at the receiving unit. Hence, it is highly recommended to create a modular structure-based transformation components library, in order to standardize the transformation process for the development and transfer of future products.

• Enterprise Recipe Management solution (Phase 4): in
order to reduce process variability and batch recipe development cycle time, the increased repeatability of production processes and the reverse flow of information for recipe changes, facilitate introduction of new product/machinery, the adoption of a dedicated content management system or a product life-cycle management system (PLMS) is suggested. These issues are in fact the most relevant for the adoption of ANSI/ISA-88. A PLM enterprise platform is usually able to integrate people, data, processes and business systems allowing the proper management of the entire lifecycle of a product. According to the ANSI/ISA-88 principles, recipe documents and process elements, libraries should be shared among the organization and continuously updated, following the new product introduction process. The use of this kind of system may reduce the probability of human errors in the recipe creation and management process, and it is able to guarantee effective communication, capturing all the ANSI/ISA-88 data, including flows and parameters in a single format, and providing a mechanism for content reuse (Recipe Building Blocks), which is critical to recipe normalization. PLM software could also help organizations improve compliance to ANSI/ISA-88 in terms of the standardization of procedures and notations, even allowing the transformation of general and master recipe information into equipment-executable recipes. Moreover, the BPMN notation adopted in the proposed method facilitates the implementation of such enterprise management systems.
To sum up, the proposal combines a series of structured steps to drive the assessment and the definition of a roadmap toward a to-be model, providing a list of practical recommendations to overcome the potential pitfalls and criticalities that companies may be faced during the implementation of this standard.
The next paragraph describes how effectively the proposed approach helped in deploying a project for a multinational pharmaceutical company.

The case study
A practical experience of how a process-oriented approach can help when analysing and optimizing a recipe development process is now described. The considered company is a leading firm in the healthcare sector and commercializes a large variety of products all around the world, with production plants in more than 50 different countries. The peculiarity of the organizational model is its independent way of managing the production processes distributed on the sites, where operations are locally managed, whereas coordination and control functions are handled by a central committee.
In recent years, the number of new product introductions and technical transfers between plants has noticeably increased due to marketing strategies and cost reduction programmes. As a consequence, the company decided to undertake several initiatives aimed at speeding up the time-to-market of new products and improving knowledge management within transfers. The most effective way to reach these goals was identified by top management in the adoption of a recipe management strategy, based on the ANSI/ISA-88 standard, focused on product life-cycle, which uses an end-to-end approach. The ANSI/ISA-88 initiative, discussed here, was then undertaken in the company's best-in-class manufacturing plant to define the standard recipe management strategy, in view of its worldwide application. The steps followed during the pilot project were consistent with the proposed approach and concerned the solid oral production process of pharmaceutical drugs performed by the company. The proposed method allowed the following objectives to be reached: • Map the as-is recipe development process.
• Assess the degree of compliance of the analysed processes and documentation to the indications of the standard. • Identify recipe development process challenges and criticalities. • Define an applicable to-be recipe development process model. • Outline short term and long-term improvement opportunities, defining the implementation roadmap.

Main findings on the ANSI/ISA-88 reference model
As described above, the ANSI/ISA-88 reference model is based on a clear separation of Process, Physical and Procedural Models, with a hierarchical subdivision in different layers. With regard to the Process Model, all elements are divided into processes, stages, operations and actions. In the documentation used within the company, this distinction was not clearly applied; indeed, production processes were mapped using elements that could be connected to both the ANSI/ISA-88 process operations and actions. This misalignment was not only found on a terminological level but also at a conceptual level. This lack brought a mix of logical information, which misled a correct implementation of the general and master recipes. Consequently, the as-is structure did not allow a complete standardization of the recipe development process. In order to align process representation and mapping, a list of standard elements with a univocal description was edited. This list gave also the input to implement the library of the standard process elements. Even though the analysis took place at a specific production site, the identified building blocks maintained an equipment-independent structure, allowing their application to similar plants.
The process stages, operations and actions identified are listed in Table 1.
Concerning the physical model, the logical subdivision into the levels used at the site plant was considered coherent with the ANSI/ISA-88 approach. A comprehensive distance in terminology existed, but concept and layering were aligned with the standard guidelines. Another relevant finding is related to how these documents were produced, and by whom. Initially, documents were created mainly by the collaboration between R&D and the production plant but, as time passed, these documents were continuously updated with more and more details about materials, process characteristics and the equipment, on which the production process was tested and validated. Consequently, documents were added with equipmentdependent information and, then, could not be used as a general recipe. The main consequence is a loss of generality once the recipe is developed and implemented on a specific series of equipment, which implicates difficulties in transferring the information to different plants. In order to overcome this problem, while avoiding a drastic change in the procedure used, the proposed approach consisted of creating a general recipe structure and integrating it into the company's existing documents.
Fewer considerations emerged concerning the master recipe: from the ANSI/ISA-88 perspective, the master recipe should include all the relevant information to manufacture a product on a specific set of equipment. This information was available, spread over different documents, such as the production version and equipment recipe documents. The main problem concerned the division of recipe contents that did not match the ANSI/ISA-88 categories (introduction, header, formula, equipment requirements, procedure, other information).
With regard to the site recipe and the control recipe documents, an ad-hoc solution was proposed, which is to merge both documents, considering that this action would not affect the advantages of adopting the ANSI/ISA-88 standard. This decision was based on an ongoing parallel initiative regarding the standardization of material and suppliers.

Main findings on the recipe transformation process
More considerations can be made concerning the transformation process, which allows the generation of a master recipe, equipment-dependent and site specific, starting from a general recipe, equipment-independent and valid for all production sites. As mentioned previously, the ANSI/ISA-88 master recipe should contain all the relevant information for the manufacturing process on a specific set of equipment. Thanks to the use of process parameters and procedures included in this recipe, the production site is able to start the manufacturing of the new product for characterization, validation, and registration for commercial purposes, according to the steps of the new product introduction process. Within the project, an important lack was revealed, due to the absence of a formalized and standard process. In the as-is scenario each process element of a production flow-chart was manually transformed into a series of instructions by using procedure templates or a production version of similar products. Critical process parameters were linked to equipment settings and sometimes this connection was not easy: critical parameters, indeed, may be measured in different units or refer to parameters not adjustable on the specific equipment. In order to standardize this transformation process, following the ANSI/ISA-88 guidelines, a set of transform components were created for the pilot project. These transform components consisted of a group of reusable tables, containing process parameters and instructions associated to specific equipment, under a series of conditions. These conditions established the link between the general recipe information and the master recipe contents. Once process operations and process actions are defined in the general recipe, and production equipment are selected at the site level, the production site can automatically define the set of instructions and parameters necessary to produce the product, using the transform components.

Conclusions
The aim of this paper was to describe how an ANSI/ISA-88 assessment could be easily performed with a dedicated Business Process Management approach and specific methodologies or notations such as BPMN2.0. This approach has been applied to a pilot project carried out in a large multinational pharmaceutical company. In the project, the authors managed to re-design the recipe development process and create a set of templates and libraries to be used as a reference to the ANSI/ISA-88 standard (general recipe, master recipe, process element library, transform components). The most important result of the initiative was the standardization of the NPI process and related documentation. Recently, the company applied the developed approach to the transfer of different products, obtaining a sensible reduction of time spent for the transfer and, at the same time, reducing the need for process experts travelling between the two plants. These results encouraged the company's management to introduce the developed recipe management process in similar sites by the end of 2014.
Despite these encouraging results, as described in the article, there are different pitfalls and criticalities that companies may face during the implementation of this standard. These can be summarized as follows: • Map the end-to-end as-is recipe development process with a BPM notation. • Start from defining a general recipe and a master recipe as standard documents. • Establish continuous education training on ANSI/ISA-88 contents, leveraging on the assessment phases.
• Standardize and make use of defined process elements.
• Share process element libraries throughout production plants.
• Create a library of standard transform components.
Typical misalignments and gaps in an ANSI/ISA-88 compliance are also reported in this article as a result of the experience on the specific case. The practical evidence provided can be considered, on one hand as recurrent criticalities that may arise in any ANSI/ISA-88 assessment, and on the other hand as challenges to deal with, in order to obtain the desired benefits. Most opportunities in the adoption of this standard for recipe management process still reside in the possibility of creating executable recipes, applicable to the same equipment of different plants, through a one-to one correspondence of process actions to equipment phases. To reach this ambitious objective, a structured approach to the design of recipe development processes seems crucial.