Proposal of a Disruption Scoring Model for Pharmaceutical Supply Chains

• Jacobs University Bremen

Introduction

The phenomenon of supply chain disruptions in globalized production and distribution networks has raised many concerns during the last years. In general, such disruptions are understood as unexpected temporal events, which lead to a negative deviation of the process plans [1, 2]. However, disruptions do not only lead to immediate losses of e.g. products, but may also lead to considerable reputation losses [3]. Especially pharmaceutical supply chains, but also food chains experience a high susceptibility to disruptions, due to the special nature of the products.

On the one hand, these products can be highly perishable, demanding for fast and temperature-controlled handling. Vaccines in particular require careful handling, yet were found to be frequently exposed to freezing temperatures during shipping [4]. On the other hand, they have direct impacts on human health and are therefore tightly regulated and controlled. The FDA for instance publishes import refusal reports for food and drugs, including the name of the manufacturer, to inform the public about violations of the Food, Drug and Cosmetic Act [5]. In consequence, more stringent requirements are posed on the supply chains, originating from the product characteristics, as well as from regulatory requirements and may significantly increase sophistication of process plans.

In the face of these challenges, companies have to achieve several goals at the same time: firstly, they have to comply with the demands by the specific product to avoid quality losses and product returns. Secondly, they have to comply with the guidelines set out by official authorities to avoid legal repercussions. Thirdly, they have to make sure that any deviations from these requirements are systematically identified, measured and analyzed, so that weak points are detected and future occurrence of disruptions prevented.

This article therefore proposes a tool, which helps to systematically identify, measure and analyze disruptions in pharmaceutical supply chains. The next section deals with the deduction of general criteria based on official requirements to unambiguously identify pharmaceutical products and process activities. Thereafter, a systematic approach based on methods of decision theory and statistics is proposed for the operationalization and measurement of indicators, to analyze process performance and identify disruptions. Section 4 connects the derived measures in a scoring model and illustrates how the results of analysis can be connected to strategic decision making tools. Finally, section 5 gives some concluding remarks and provides an outlook for further research. Thereby, the developed scoring model attempts to be both generally applicable and customized to consider specific demands of the individual pharmaceutical supply chain.

Starting Points for Identification of Disruptions in Pharmaceutical Supply Chains

As mentioned in the introduction, a disruption occurs when unexpected temporal events lead to negative deviations from process plans. This rather broad definition implies that in order to identify and measure disruptions, firstly one or more negative deviations from the process plans have to be identified and quantified. Therewith, several challenges are connected. For instance, how can deviations from plans be quantified, which include qualitative criteria, such as product quality? Another challenge is to use a systematic procedure for the identification, so that each deviation can be identified in an intersubjectively and inter-temporally consistent way.

Furthermore, the content of process plans might differ considerably between companies and supply chains, especially with increasing level of detail. Thus, reproducibility and comprehensiveness are imperiled, which in final consequence would lead to non-comparability of supply chain performance. For example, deviations cannot be compared if one process plan specifies room temperature based on the European Pharmacopoeia as in between 15°C and 25°C [6], whereas another one specifies it based on the Health Sciences Authority to be in between 15°C and 30°C [7]. Nevertheless, even though several challenges exist for unambiguous identification and quantification of disruptions, there are some basic characteristics, which process plans share and which can be used to describe the plans and to identify indicators for deviations.

Usually, pharmaceutical supply chain processes can be defined as a combination of logically connected activities, which transform inputs into outputs, hence raw materials such as Active Pharmaceutical Ingredients (APIs) into pharmaceutical products [8]. Thus, the process plan specifies how products and activities shall be combined to transform the input in the desired output. If pharmaceutical products and process activities can be described based on general criteria, combining these descriptions would depict broadly underlying process plans and therewith, deviations could be identified. Additionally, this would help to assess which products and which activities are especially vulnerable to disruptions.

Criteria for the General Description of Pharmaceuticals

To allow for comparison of different products, or between similar products in different supply chains, information has to be presented in standardized manner. Of course, pharmaceutical products might have diverse characteristics and not all are relevant for assessing supply chain performance. For example, the method of administration does not seem to be relevant for supply chain performance, whereas product packaging needs to be specified as it is part of the process. Here again, the challenge is to choose a suitable basis for the description, which in general, covers characteristics of the product relevant to supply chain processes, and does not require information which is not available. Therefore, as a reference scheme, the Summary of Product Characteristics (SmPC) [9] can be used, as these characteristics have to be specified by pharmaceutical manufacturers for market authorization in the EU. Those criteria which help to unambiguously identify the product and specify its handling during the commercial part of the supply chain are illustrated in Figure 1.

Figure 1. Selection of Criteria for the Description of Pharmaceutical Products as Part of Process Plans

The criteria are extracted from the SmPC, especially of point six, “Pharmaceutical Particulars”. The first four points help to unambiguously identify the pharmaceutical product from an input and production perspective, whereas the remaining six points specify those characteristics, which might be relevant to correctly handle the finished product. As the SmPC is obligatory for all pharmaceuticals awaiting market authorization in the EU, it can be assumed that it has been tested and modified for a sufficiently large number of pharmaceuticals, to claim for its general applicability. Therefore, along these ten points, it should be possible to describe every pharmaceutical product to specify this part of the process plans in a comprehensive and comparable manner.

In order to further increase comparability, the terms used for the description of products have to be also stated in a standardized way so that every product can be described by picking the appropriate attributes from a finite list. Potential attributes can be found e.g. in EU pharmacopoeia, where different categories of attributes are listed [6]. For a complete classification of pharmaceuticals, classes have to be exhaustive and mutually exclusive [10]. These classes are obtained by transferring product characteristics as stipulated in SmPC and potential attributes as listed in the pharmacopoeia into a morphological box. For each product, appropriate attributes can then be selected for each characteristic. Therewith, an exhaustive and distinctive description of products becomes possible, which helps to unambiguously identify and describe relevant product characteristics. The question is thus, how to specify the other part of process plans, namely the activities.

Criteria for the General Description of Pharmaceutical Supply Chain Processes

Regarding process characteristics, the same methodology can be applied. The important categories as well as alternative attributes can be taken from official regulations, such as Good Distribution Practices (GDP) and Good Manufacturing Practices (GMP) [11, 12]. These documents cover official requirements for every part of supply chain processes, whereby it can be assumed that relevant activities are specified. From a content analysis, the criteria and attributes listed in Figure 2 were derived.

Figure 2. Morphological Box to Specify Process Characteristics

The illustration does not demand for completion, but shall draft the methodology for more detailed analyses and further expansion. For a given product and a given process, the actual combination of attributes can then be selected from the morphological boxes of product characteristics and process activities. This combination specifies the process plans, which are the basis for identifying disruptions. To further concretize the plans, indicators can be used, which state how an activity should transform or transfer a specific product. The advantage of indicators is that they help to concretize abstract phenomena, which can then be observed and tested empirically [10]. In order to deduce indicators in a systematic approach, the structured product description and the structured process description are combined in a matrix, providing a framework for developing indicators of process plans. To deduce indicators for a given product and a given process, instead of the general set of criteria, the actual combination from the morphological boxes is chosen. Thereby, more specific indicators can be generated. Figure 3 shows the general approach to illustrate the procedure.

Figure 3. Product-Process Matrix for the Deduction of Indicators of Process Plans

For every field of the matrix, key aspects of the process plans can be identified and indicators for the plans and therewith for deviations derived. Thereby, the risk of omission is reduced and the deduction of indicators can be inter-subjectively understood. However, to make disruptions quantifiable, these indicators require operationalization as well as general rules for how to combine values.

Procedure for Measuring and Interpreting Process Behavior

Whereas the explanations above focused on the development of the framework for deducing indicators, the following focuses on their measurement and the subsequent identification and interpretation of deviations. However, as the following steps depend on product and activity specific indicators, the explanations focus on the procedure, which can be illustrated with some examples. For instance, the stability of a specific pharmaceutical product may be specified as “store in a dry place not above 25°C; protect from light”. Indicators for stability during storage could then be moisture, temperature and light intensity. Their operationalization and quantification seems to be straightforward, but for other indicators, as for example for the pharmaceutical form, qualitative measures could prevail, which could be interpreted differently by different observers. To enhance unambiguousness of interpretation, rules for interpretation are therefore required. The resulting questions are thus firstly, how to interpret the obtained values in comparison to one another, and secondly, how to interpret them regarding the question whether a disruption prevails or not.

For quantitative measures, the range of compliant values can be calculated with the help of statistical methods as used in statistical process qualification [13]. Depending on the type and distribution of values and the pursued statement, a statistical methodology is selected to calculate the mean value, which represents regular process behavior [14]. Qualitative data in turn only allows for attributing values to different classes. Thereby, the ability to interpret results is limited, as a value can only be interpreted regarding whether it matches the process plan or not. Therewith, less distinctions are possible than with quantitative measures, which should be preferred where possible.

But also quantitative measures cannot be compared across measures, as their values may lie on different scales, such as e.g. degree Celsius and kilograms. To increase comparability of measures and indicators, and to identify patterns in the results across items, it seems therefore useful to transform the different scales into one consistent scale, even though information content, e.g. regarding distances between values, might be lost.

As a first step for interpreting each obtained value for itself, a scale is required, on which all potentially possible values of the measure can be allocated. Hence, maxima and minima have to be specified and a consistent measuring unit identified. In a second step, to aggregate values obtained from different measures, a general scale needs to be created on which the values can be transferred into classes. To assure that the transformation is consistent and does not distort the original implication of a value, classes need to be of the same size, cover the entire range of transformed values and be mutually exclusive [10]. Therefore, the design of the scale should be oriented towards those measures, which have a comparably large range of potential manifestations with different implications.

For the interpretation of measures and indicators, not only their actual manifestations are relevant, but also their relative importance for the assessment of disruptions. Deviating temperatures might be more relevant to the product quality during shipping than e.g. exposure to UV light. Furthermore, product characteristics such as quantity of API or types of excipients might be more important during ordering and processing than during distribution. Therefore, each measure and each indicator needs to be weighted based on its relative relevancy, which can be done with methods from decision theory [15].

A suitable methodology for this task is the Analytical Hierarchy Process (AHP), where each variable receives a weight based on pair-wise comparison between all variables on the same level [16]. As disruptions negatively affect supply chain processes, variables should be prioritized firstly, considering robustness of the supply chain process regarding a specific deviation, and secondly, regarding to what extent the deviation is detrimental to the process outcome. Following the principles of Multi Attribute Utility Theory (MAUT) [15], the weighted values of measures are then aggregated for each indicator, which can then be again weighted and aggregated for each field of the product-process matrix. Thus, the degree of process plan fulfillment is measurable based on ranked multiple criteria and disruptions can be inter-subjectively identified based on a systematic and consistent scoring model.

Scoring Model for Supply Chain Disruptions

The procedure explained above results in a set of measures and indicators to identify and quantify disruptions in processes transforming and moving pharmaceutical products from sourcing to delivery (and return). In order to analyze which parts of the process are more vulnerable to disruptions, or which products are especially affected, each field of the matrix can be further aggregated and disruption scores compared across shipments as illustrated in Figure 4.

Figure 4. Illustration of Interpreting Process Scores in a Polarization Graph

In the picture above, the scores obtained for each process activity are listed separately for every product characteristic and are then transferred into a polarization graph. Therewith, the fulfillment of process requirements can be analyzed for every process activity on different product dimensions. The polarization graphs for different shipments could then be compared and weak points identified by analyzing differences and similarities in patterns of process performance. Alternatively, fields of the product-process matrix could be aggregated and disruptions compared across process plans from different companies along the supply chain. Thereby, decision making regarding distribution channels could be enhanced and robustness of supply chains fostered by joint optimization.

Conclusions

The methodology provided above shall serve as a starting point for identifying, analyzing and preventing disruptions in pharmaceutical supply chains. The advantage is that the approach is systematic, as the foundations are official requirements, instead of company-specific Key Performance Indicators. Furthermore, the process for how deviations are identified is transparent, as decision making is based on methods from decision theory and statistical procedures. Additionally, the resulting scores can be linked to other managerial tools, for example from quality management or marketing, and help to tackle weak points to prevent future disruptions.

Of course, some limitations exist: firstly, the preparation of its implementation can be laborious, especially if many pharmaceutical products and supply chain processes have to be considered. An option to reduce the initial complexity would be to prioritize fields of the product-process matrix to develop indicators first for most critical parts of the process plans. Secondly, comparability could be reduced, if the choice of indicators differs among companies or even employees. Therefore, a documented procedure is required, to identify where discrepancies result from. Thirdly, the documents chosen as information sources or the content taken from them may prove to be inadequate. In consequence, this proposal may need further refinement and requires empirical testing, to assess its feasibility and usability in pharmaceutical supply chains.

References

1. Craighead, C.W.; Blackhurst, J.; Rungtusanatham, M.; Handfield, R.B. The Severity of Supply Chain Disruptions: Design Characteristics and Mitigation Capabilities. Decision Science 2007, 38, 131–156.
2. Brenner, V.; Hülsmann, M.; Korsmeier, B. Analyzing Supply Chain Disruptions - Proposal of a Classification Framework. In New Horizons in Logistics and Supply Chain Management; Pawar, K., Potter, A., Eds., 2012, pp. 228–235.
3. Hendricks, K.; Singhal, V. Association between Supply Chain Glitches and Operating Performance. Management Science 2005, 51, 695–711.
4. Matthias, D.M.; Robertson, J.; Garrison, M.M.; Newland, S.; Nelson, C. Freezing temperatures in the vaccine cold chain: A systematic literature review. Vaccine 2007, 25, 3980–3986.
5. MacPherson, A. The Impact of the U.S. Bio-Terrorism Act upon Canadian Exporters of Food Products to the United States: A Firm- Level Analysis. The International Trade Journal 2008, 22, 17–38.
6. DQM. European Pharmacopoeia, 7th Edition; Council of Europe: Strasbourg, France, 2012.
7. HSA. Medical Device Guidance: GN-01: Guidance on the Application of Good Distribution Practice for Medical Devices, 3rd ed: Singapore, 2011.
8. Cooper, M.C.; Lambert, D.M.; Pagh, J.D. Supply Chain Management: More Than a New Name for Logistics. The International Journal of Logistics Management 1997, 8, 1–14.
9. European Commission, Ed. A Guideline on Summary of Product Characteristics, 2009.
10. Kromrey, H. Empirische Sozialforschung: Modelle und Methoden der standardisierten Datenerhebung und Datenauswertung, 12th ed; Lucius & Lucius: Stuttgart, 2009.
11. Commission Guidelines on Good Distribution Practices of Medicinal Products for Human Use, 2013.
12. User227. EU Guidelines for Good Manufacturing Practice for Medicinal Products for Human and Veterinary Use, 2013.
13. Box, G.E.; Luceño, A.; Paniagua-Quinones, M.d.C. Statistical control by monitoring and feedback adjustment, 2nd ed; John Wiley & Sons: Hoboken, N.J, 2009.
14. Dietrich, E.; Schulze, A. Statistische Verfahren zur Maschinenund Prozessqualifikation: Mit 61 Tabellen, 6th ed; Hanser: München [u.a.], 2005.
15. Götze, U. Investitionsrechnung: Modelle und Analysen zur Beurteilung von Investitionsvorhaben, 5th ed; Springer: Berlin [u.a.], 2006.
16. Saaty, T.L. Decision making for leaders: The analytical hierarchy process for decisions in a complex world; Lifetime Learning Publ: Belmont, Cal. [u.a.], 1982.
17. Hülsmann, M.; Grapp, J. Monitoring of Autonomous Cooperating Logistic Processes in International Supply Networks. In Proceedings of 11th International Symposium on Logistics; Pawar, K., et. al., Eds., 2006, pp. 113–120.

Verena Brenner is a Ph.D. student in International Logistics at Jacobs University Bremen. In her research and industry projects, she focuses on cold chain management in food and pharmaceutical supply chains. Before starting her Ph.D. in 2010, Verena worked for the Cool Chain Association, an international organization of logistic service providers for the food and pharma industry. She organized workshops on pharmaceutical and food logistics and coordinated an international project on quality management in food chains. Verena holds a diploma in Economics, with a specialization on logistics and sustainable management. She published and presented various papers on cold chain management in scientific and industry-relevant venues and is co-author of a study on cold chain ruptures in food supply chains.

Prof. Dr. Michael Hülsmann is Associate Professor of Systems Management in the department of International Logistics at Jacobs University Bremen. He focuses on Strategic Management of Logistics Systems. Prof. Hülsmann teaches and performs research mainly in the fields of competence and technology-based positioning, accomplishing technological change in value-adding networks and setting up interorganizational coordination mechanisms.