Application of quality by design for 3D printed bone prostheses and scaffolds

3D printing is an emergent manufacturing technology recently being applied in the medical field for the development of custom bone prostheses and scaffolds. However, successful industry transformation to this new design and manufacturing approach requires technology integration, concurrent multi-disciplinary collaboration, and a robust quality management framework. This latter change enabler is the focus of this study. While a number of comprehensive quality frameworks have been developed in recent decades to ensure that the manufacturing of medical devices produces reliable products, they are centred on the traditional context of standardised manufacturing techniques. The advent of 3D printing technologies and the prospects for mass customisation provides significant market opportunities, but also presents a serious challenge to regulatory bodies tasked with managing and assuring product quality and safety. Before 3D printing bone prostheses and scaffolds can gain traction, industry stakeholders, such as regulators, clients, medical practitioners, insurers, lawyers, and manufacturers, would all require a high degree of confidence that customised manufacturing can achieve the same quality outcomes as standardised manufacturing. A Quality by Design (QbD) approach to custom 3D printed prostheses can help to ensure that products are designed and manufactured correctly from the beginning without errors. This paper reports on the adaptation of the QbD approach for the development process of 3D printed custom bone prosthesis and scaffolds. This was achieved through the identification of the Critical Quality Attributes of such products, and an extensive review of different design and fabrication methods for 3D printed bone prostheses. Research outcomes include the development of a comprehensive design and fabrication process flow diagram, and categorised risks associated with the design and fabrication processes of such products. An extensive systematic literature review and post-hoc evaluation survey with experts was completed to evaluate the likely effectiveness of the herein suggested QbD framework.


Overview
Since the introduction of QbD into the pharmaceutical field in 2004, it has been widely used in different pharmaceutical fields to enhance formulation and process design (Fahmy et al., 2012), improve of drug manufacture (Badawy et al., 2016;Hubert et al., 2014;Zhang, Yan, Gong, Lawrence, & Qu, 2013), and to develop nano based pharmaceutical products (Cun et al., 2011;Raina, Kaur, & Jindal, 2017;Verma, Lan, Gokhale, & Burgess, 2009). Moreover, through a systematic search and a statistical analysis (S1 File) it was found that the implementation of QbD can provide remarkable results enhancing process and product understanding, and leading to a drastic (up to 90%) reduction of experimental runs.
Additionally, it was found that QbD was mainly used to achieve four distinct objectives: process understanding (PU), prediction and optimization (PO), reduction of experimental runs (RER), and development of robust manufacturing methods (DRM).
Based on the positive results that QbD has being providing for the pharmaceutical field, we believe that the adaptation of QbD for the development of 3D printed bone prostheses and scaffolds can provide similar outcomes. As a result, this study formulated and implemented the first five steps (1-5.3) of the QbD system for the development of 3D printed bone prostheses and scaffolds. However, in order to know if this adaptation of QbD will be useful for the development of 3D printed bone prostheses and scaffolds, we validated our results with several experts in the field of study. For this purpose we performed semi-structured interviews to achieve the following objectives: a) Introduce the concept of the QbD system to a selected group of experts notably researchers and surgeons. b) To compare the QbD system with business as usual practices and obtain comments and examples on how QbD system can improve current practices. c) To study the perceptions of researchers and practitioners in relation to the results of the adaptation of QbD for custom 3D printed bone implants with the four previously identified (statistical analysis) potential benefits that QbD can provide to their field of expertise.

Materials and methods
In order to validate the results of the adaptation of QbD for custom 3D printed bone implants, a qualitative research approach was employed. For this purpose, semi-structured interviews were specifically designed to capture the experiences and opinions of researchers, industry experts, and practitioners with previous experience in at least one of the main processes necessary for the design and fabrication of custom 3D printed bone implants.

Data collection
The design and fabrication of custom 3D printed bone implants requires a multidisciplinary team of experts of different fields specifically medicine, engineering, and bone biology. Therefore, qualitative semi-structured interviews were selected as the data collection instrument to be performed with each participating. For the purpose of this study, the exploratory approach was adopted following the Consolidated criteria for reporting qualitative research (COREQ) (Tong, Sainsbury, & Craig, 2007), as the aim was to validate the empirically implemented QbD system. By following this approach, 26 questions in total divided in four specific groups were designed. Moreover, an interview guide and a PowerPoint presentation were designed to guide the direction of the conversation, present the preliminary results obtained from the adaptation of QbD and the statistical analysis, and to gather new data that could not be taken into account during the adaptation of QbD system, for more details refer to Appendix A.

Study selection
The criteria to select the participants for this study was based on their experience and expertise in the field of study. Therefore, pertinent experts in the field of tissue engineering, medical product development, and orthopaedic surgeons with previous experience with 3D printed bone implants, were selected.
The sample size is limited by the nature of the research field, which is characterized by small samples, but detailed and extensive work (Anderson, 2010). Therefore, the snowball sampling method was selected to cover as many people the researcher can gain access to.
Snowball sampling is when the research benefits from participants network, and others participants are suggested or introduced to the researcher (Harrell & Bradley, 2009). Eight invitations to participate in this study were send via email to different experts of different universities in Australia and USA.

Data extraction and analysis
To facilitate the data extraction an interview guide was used to maintain the direction of the conversation and the relevant lines of enquiry, whilst probing into the issues of interest.
To obtain a complete and accurate description of the interviewee's responses and comments, all interviews were recorded and transcribed for further analysis. Moreover, at the beginning of each interview a consent form was delivered to each participant, explaining that the information that they will provide will be considered confidential and the summary of results of this study may be used for presentations and publications. Consequently, each participant signed the consent form and approved the interview be audio recorded. The types of questions that this research addressed were descriptive and structural. Descriptive questions are asked to get descriptions of things and processes in order to get insights, or to check validity or accuracy about something (Harrell & Bradley, 2009). Structural questions help the researcher to categorize groups of things and processes and to understand its relationships (Harrell & Bradley, 2009).
Qualitative data analysis required to examine, categorize, tabulate, test and combine evidence to address the initial propositions of a study (Yin, 2014). The data analysis for the semi-structured interviews followed two key steps recommended by Eisenhardt (1989): withincase and cross-case analyses. In this study the within-case analysis was concerned with the evaluation of the collected data, as well as the reporting of the findings of each individual case study. A systematic reading through each transcript was performed, to then assign codes to tag segments with similar content to sort them into separate categories for a final distillation into major themes (Appendix B). The codes were pre-designed using the deductive coding technique based on the five groups of questions designed for the interviews (Stemler, 2001). It is worth mentioning that due to the open ended nature of the interview questions, the answers for a particular question were be mixed up with another question. As result, part of the data gathered was of an "unstructured" nature consisting of long paragraphs which were organized in a structured and evidentiary-based manner to be able to draw conclusions as the study progresses. The information obtained from each interview provided an insight into how different factors of the adapted QbD system were perceived by experts from different fields of expertise. This was presented in the form of evidentiary tables containing the classified evidence from each interview using Miles, Huberman, and Saldaña (2014) tabular approach, for more details refer to Appendix C.
Following this, the cross-case analysis was performed to find patterns, agreements, and disagreements in opinions between the interviewees (McCutcheon & Meredith, 1993). To facilitate the cross-case analysis, the information in the form of evidence extracts, was categorized in tabular manner based on the coding used, for more details refer to Appendix D.
Once all the evidence had been organised, the results of this analysis were used to enhance and validate the preliminary results obtained from the adaptation of the QbD approach, and to report participants' opinions and concerns. It has to be noted that the pattern-matching procedure involves no precise comparisons; there may not be quantitative or statistical criteria involved on which to judge the pattern, thus allowing for some interpretive discretion on the part of the researcher (Yin, 2003).

Results
A total of six semi-structured interviews were performed in October 2017. To maintain the participants' confidentiality and anonymity a code was assigned to each of them as A, B, C, D, E, and F. Overall, the participants' expertise comprised a good mix of researchers, industry experts, and medical practitioners of different fields related to medical device development, 3D bone printed implants, motion capture, tissue engineering, orthopaedic surgery, bone biomechanics, computational neuromuscular modeling, and nano engineered implants. Participants' details such as years of experience, and area of expertise and research are summarized in Table 1. The duration of each interview ranged from 30 to 90 minutes. Four sets of face to face interviews were conducted in Australia, and two via video conference in Australia and USA. All interviews were carried out by the first author of this paper on a oneto-one basis. Previously in the statistical analysis (S1 File) it was identified that the main objectives to use the QbD system in the pharmaceutical sector were: process understanding (PU), prediction and optimization (PO), reduction of experimental runs (RER), and development of robust manufacturing methods (DRM). Comparing this results with the interviews responses, it was found that there is a positive common consensus between participants' responses in relation to the main four reasons that the adaptation of the QbD system can be used in the QbD by design also "can act as an insurance for innovation to be truly examined and analysed" (Participant A). Moreover, "having the backing of QbD it will definitely assist innovation towards commercialization" (Participant E). All this can be possible due to the fact that QbD is a tool that facilitates a complete understanding of the product and its fabrication processes, allowing to short test products and ideas to find optimal ranges of operation that then can be extrapolated to invent new products (Participant D).
Other potential benefits were perceived by the participants of this study, for example: Participant B stated that QbD can facilitate communication between experts from different fields, cutting out work throughout the project and saving a lot of trouble and cost. Moreover, participants B and C mentioned that the use of QbD and computational modelling will be critical to develop personalised biological implants, because they "does not allow destructive testing, and will required millions of hours of physical testing just to one product" (Participant C). Furthermore, participants D, E, and F identified that there is a need for regulations in this area, and QbD can pave the way the way for future enhancement in this research field and act as a guidance for regulatory bodies across countries such as FDA (USA), and TGA (Australia) to develop new quality standards for these emerging customized products, thus "they can function better, brake less, and give better outcomes" (Participant C).
In regards to the participants' concerns, three of them said that the challenge is to convince industry and researchers to use the QbD system. Moreover, they point out that the biggest challenge could be the integration of the system, and the start-up time and cost to implement it. Additionally, some participants suggested few improvements to the adaptation of QbD, which were accordingly addressed in order to enhance our preliminary results.

Conclusions
Overall, the interview results show a positive common consensus between participants' responses agreeing that adapted QbD system of this study can be used to achieve the same benefits that the QbD has been providing to the pharmaceutical sector. These four benefits where previously identified in the statistical analysis (S1 File), they are: process understanding (PU), prediction and optimization (PO), reduction of experimental runs (RER), and development of robust manufacturing methods (DRM). Moreover, during the interviews the participants commented about different potential benefits that QbD can provide to this field, including some concerns and suggestions.
According to the opinions of participating experts, one of the potential benefits that QbD can offer for the development of such products is the direction of the product development process clearer goals by providing a depth understanding of the numerous factors involved in it, allowing a better definition of product boundaries, targets, and potential modifications.
Moreover, according to participants' comments the results of the adaptation of the QbD can enhance current design practices and accelerate product development by breaking down the whole development process into ease understandable sections that can independently be developed under the same goal. The results of it is a more efficient streamline process towards the reduction of product development time frames, and the help to reduce a high risk industry in a situation that is really low risk. Furthermore, QbD by design also "can act as an insurance for innovation to be truly examined and analysed" (Participant A). Moreover, "having the backing of QbD it will definitely assist innovation towards commercialization" (Participant E).
All this can be possible due to the fact that QbD is a tool that facilitates a complete understanding of the product and its fabrication processes, allowing short testing of products and ideas to find optimal ranges of operation that then can be extrapolated to invent new products (Participant D).
Other potential benefits that were perceived are that QbD can facilitate communication between experts, because it breaks down the whole development process into coherent sections cutting out work throughout the project and saving a lot of trouble and cost. Moreover, it was mentioned that the use of QbD and computational modelling will be critical for the development of personalised biological implants. Furthermore, three participants identified that there is a need for regulations in this area, and QbD can pave the way for future enhancement in this research field, acting as a guide for regulatory bodies across countries such as FDA (USA), and TGA (Australia) to develop new quality standards for emerging customized medical products, thus "they can function better, brake less, and give better outcomes" (Participant C).

Appendix A Interview Guide
Topic Potential benefits, improvements, and challenges of the Quality by Design (QbD) system in research and industry.

Objectives
1. Introduce the concept of the QbD system to a selected group of experts such as researchers and medical doctors.
2. To compare the QbD system against business as usual practices and obtain comments and examples on how QbD system can improve current practices.
3. To study the perceptions of researchers and medical doctors in relation to the four previously identified (statistical analysis) potential benefits that QbD can provide to their field of expertise.

Introduction of the QbD system adapted for custom 3D printed bon prostheses and scaffolds
General (objective 1): In the following minutes I will introduce and explain the concept of the QbD system, including the results of the adaptation of this system for the development of 3D printed bone prostheses and scaffolds (refer to page 4).  for instance in a project that I'm looking at virtual correction of children who had hip fracture so they had a deformity in the femur and we are looking cutting guides for the surgeons, so QbD will ensure that the final product more close to optimal at the time of printing (IMPRV-PRAC), but also for the down track of surgical blade plates to hold the correction and loading environment they are going to undertake. So I think QbD would definitely give us a lot more confidence to proceed with that kind of projects at least in terms of prototype that can add up (P-BENF). I think of having the backing of QbD it will definitely assist innovation towards commercialization Appendix C Within-case evidence extracts O-BENF  I think sounds really interesting and exciting. This is going to be a really powerful tool, and if is presented well to industries I think it will be a huge interest on this (Q 3.3)  Also I can see how QbD can be used to reduce the scrap rate in the manufacturing process (Q 3.4)  QbD sounds as a requirement when you are talking about customized manufacturing (Q 3.9)  This system is much easier to implement (Q 5.4)

IMPRV-PRAC
 I can see that this method can be the replacement to the Six Sigma method, which is really effective for conventional manufacturing. But in the case of 3D printing this is totally different, that's why I think that QbD can work here (Q 3.9) ENHAN  Yes, because this system is much easier to implement (Q 5.4)

INNOV
 I think QbD can act as an insurance for innovation to be truly examined and analysed. The systematic approach that QbD has, can encourage innovation (Q 5.3)

CON-IMPROV
 I did not see in your presentation the severity of the risks and consequences to the patient (Q 3.7)  The challenge here is to convince industry and get the 3D printers' qualification (Q 3.10)  Yes, it need to include design validation and verification, traceability matrix, and design history file (Q 5.5)

O-BENF
 What you just show me is quite critical to aid new companies in their product development process (Q 3.3)  Is not only about understanding the whole process, but all these different processes need different people who don't necessary know what to speak with each other. So introducing the different sections of QbD to these different people will cut-out work throughout the project saving a lot of troubles and cost (Q 3.5)  If we look it for 3D printing biological material, I think at the moment our technology is not quite there… However, in the future we need to head for biological products, and that's where this process (QbD) will be critical to develop that sort of technology (Q 3.9)

IMPRV-PRAC
 Yes (Q 3.10)  Nothing like this ever existed like this before, so look it from this perspective, this is an enhancement in streamlining the whole product development process (Q 4.1)  There is not a streamline process that we can really rely yet. So the fact that now the QbD process very systematically brakes down into sections which later people can take on and then develop independently, this will be a big step towards making the process more efficient and improve time frames (Q 4.5) ENHAN  That depends on the government. If you are able to take it to the people that can make those decisions and convince them that the process is the one that they need to adopt, which I believe you should, then if there are able to implement it and marketed and make it available to developers, then yes absolutely (Q 5.4) INNOV  It alternately lead to better outcomes for those projects that people embark on, yes. Certainty the implementation yes (Q 5.3)

CON-IMPROV
 you should include also patient's characteristics such as biology, health condition, because patient infection and defect can change etc (Q 3.7)  It needs to be simplified in some way… for somebody that is just starting and does not have an overall picture of what QbD is, it may be look it and don't understand it. It will be nice to have a single page summary of the idea and concept before you break it down (Q 5.5)

PU
 What you just show me is quite critical to aid new companies in their product development process, and giving to them a framework to understand the development process (Q 3.3)  Nothing like this ever existed like this before, so look it from this perspective, this is an enhancement in streamlining the whole product development process (Q 4.1) PO  Yes (facilitate optimization of experiments), can make the development process quicker (Q 4.3) RER  certainly will reduce the number of experiments that need to be conducted (Q 4.2) DRM  There is not a streamline process that we can really rely yet. So the fact that now the QbD process very systematically brakes down into sections which later people can take on and then develop independently, this will be a big step towards making the process more efficient and improve time frames (Q 4.5)

O-BENF
 With QbD we will produce personalized products that function better, fit the function of the person, brake less, and give better outcomes (Q 3.2)  Personalised implants does not allow destructive testing, because that will required millions of hours of physical testing just to one product. You do not do that in a personalised implant. It has to be by QbD there is no other way to do it (Q 3.9)

IMPRV-PRAC
 In my practice in aeronautical industry I used it. You can see how much the aircraft industry succeeded in reducing a high risk industry to something that is really low risk (Q 3.5)  I think this risks assessment is similar to the one used in aeronautics (Q 3.6)  Yes, I think it can. For example: A knee implant from "A" company can have between 100 to 1000 combinations, between implant components, how do you select the right component for an individual. You don't, surgeons rely upon on most go to set of components and apply it to a person. But having a digital model of that person will give better outcomes (Q 3.10) ENHAN  Definitely, especially now with personalization of implants with 3D printing. I think that QbD and computational modelling will help to design better biological implants (Q 3.4)  Certainty, I think that the whole process is innovation ( Q 5.3)  I think generally yes (Q 5.4) INNOV  Certainty, I think that the whole process is innovation. It actually allows you to play around with simulations and 3D printing, and really rapidly test different ideas, and that is absolutely critical for innovation in this space (Q 5.3)

CON-IMPROV
 I think this risks assessment is similar to the one used in aeronautics, but needs to be integrated with computational simulations (Q 3.6)  How we put QbD and BIM, PLM into a process. How we actually do that? That's I think is the problem at the moment, it doesn't exist in the field (Q 5.2)  There are some details there that need to be amended to get this other concept together in put in proper simulation in creating 4D models, I think needs to be a central part of what you are proposing. Also rehabilitation is an important component of the success of the implant in the person (Q 5.5)

Most used reasons for the use of QbD
PU  Firstly, this will allow to simulate possible variations on design and see what is important and what is not important (Q 4.1)  QbD requires feedback loops so the outcomes of the product go back into the design process in a rigorous way to analyse why something failed (Q 4.1) PO  Yes, I think that if QbD can include multiscale models, we can predict the outcomes of an implant design and treatment (Q 4.2)  Yes, it will tell us the most crucial things that we should do in experiments on (Q 4.3) RER  Definitely, because allow us to look the most sensitive parameters for the required outcomes. For example, wind design used to be empirical, in the transition from subsonic to supersonic. Design engineers use to think that they can find a wind design that allow for a smooth transition with laminar flow, but actually it was shown mathematically that can never exist, and that stop them for looking the solution for this problem. So using the same analogy we can also use predictions that will tell us what we can do and what we cannot do (Q 4.4) DRM  QbD requires feedback loops so the outcomes of the product go back into the design process in a rigorous way to analyse why something failed. Moreover, this feedback loops can ensure that the fabrication produces what we are expecting (Q 4.1)  In its complete element it (QbD) will improve biomedical engineering products, yes, I think there is enough evidence. I think this is a rigorous detailed process that have been suggested here (Q 4.5)

O-BENF
 Generally there is a need for regulations, at the moment there is no regulatory agreement in that space and also I think we need to try to have that consistency across different countries and universities, because then you can compare the differently produced products. So I think having a document that is a guideline of what you should be doing, is like when you are doing mechanical testing and there is the American's standard and you take that as your standard testing, so if somebody have done that then I should be able to reproduce it and roughly have the same results. In the implant space we don't have that and in order to compare we need something like that, is just a matter of what will people accept overall (Q 3.9) IMPRV-PRAC  Probably (Q 3.10) ENHAN  So depends on the application, for a long term product you can test all things and maybe is not the question of those short term projects (Q 5.4)

INNOV
 I guess, in terms that you develop something that is short of tested in certain range and then you just extrapolate a little bit, just by definition the extrapolation still like a new innovation, and if you had thought about that, then it will be relatively easy to extend it without having to go again through the whole application process. I think that is the interesting part, it is a good way of thinking, because you may better try to combine the approved products and invent a new one (Q 5.3)

CON-IMPROV
 (ACCEL) Is a tricky question, because all the principle that you said all make sense, the big question is going through all the principles will really improve what is currently done? So we just can hope that it will improve (Q 3.4)  As you see the standards are relatively simple generally, but yours is like a Pandora box, very complex (Q 3.9)  I think you would put a timeframe on how long some of those things will take. Because I guess if you present it to a surgeon for somebody that has a big tumour you have two options, you cut his leg off which is you normally do, or you basically 3D printing a prototype in next five days and save this limb. So for me the question is with your concept I took that to some extend or it is necessary to go the way that surgeons and engineers go at the moment (Q 5.4)

D-GOAL
 By checking all the boxes in QbD, I would suggest you can better define the direction of your project and also the research question, in particular in determining the boundaries of the product may encompass, you can determine where your product is to be targeted then also how much modifications you can see within that process. In the case for regulatory approvals and so forth because all that hasn't been stablished as supposed to. Getting half way in a project and then realizing that you are not going where should be (Q 3.3)  whereas QbD we are trying to do from the outside is to determine what is scopeable possible and have the process setup from that point of view (Q 3.8)

ACCEL
 I think product development yes. I think once is stablished, so if you have a specialized person within the research team to ensure the quality or at least investigate and to learn the process, I think it will be quite valuable (Q 3.4)  For product design, commercialization and innovation I think would be valuable, an approach like that should be implemented if the point of the research is to move to commercialization (Q 3.4)  So I think making sure that you have all the information beforehand definitely accelerate the product and will lead to the client, the surgeon in that case, and accelerate the use of the technology (Q 3.5) O-BENF  But the advantage that I suggest is that once the system is setup is very easy to ensure that all the processes are correct and followed, and you are able to identify where the errors are, and quickly fix the contribution of that errors (Q 3.2)  I think in designing the process where you have multiple check points along the way to say: it is my data is good at this point? Can I proceed yes or no? That's something of advantage for (Q 3.5)  I think QbD provides probably more a template to ensure that the quality of the data is as precise as possible (Q 3.7)  Trust is probably the main one (potential benefit), if the clinician and the regulatory approval bodies consider that there is a body of evidence of work showing that a comprehensive risk assessment has been undertaken they are more likely to have confidence that they don't going to get significant issues from using 3D printing (Q 3.9)  So I think QbD would definitely give us a lot more confidence to proceed with that kind of projects at least in terms of prototype that can add up (Q 5.1) IMPRV-PRAC  Definitely, I think it can improve current practices and improve efficiency within the service. I mean knowing all the issues and when and why those errors occurred (Q 3.10) ENHAN  Definitely will enable us to match current design and give confidence to produce products as good or even better that current designs (Q 5.4)

INNOV
 I think of having the backing of QbD it will definitely assist innovation towards commercialization (Q 5.1)  Definitely will assist with translation out of innovation (Q 5.3)

CON-IMPROV
 In terms of stablishing the QbD regime it will be difficult to start because there are a lot of check points to go through and you have to investigate a lot of steps within the process (Q 3.2)  I think product development yes (ACCEL). But in terms of research especially in my field, there is such a pressure to continuously publish that definitely corners are cut. my perception would be that a lot of research would not take the effort to learn the whole process and implement all the procedures (Q 3.4)  It Is just the start-up time and cost to implement it (Q 3.10)  Maybe, is the useability of the results, how would someone actually going to implement the whole process would say that this is a comprehensive outline which involves multiple professions, so I think along that process you are going have lots of different teams of people working in stablishing the risk management strategy (Q 5.5)

Most used reasons for the use of QbD
PU  Looking at the process, the QbD process enable you to have all the major areas, and knowing the settings on the printers and software are the key (Q 4.1) PO  I think many aspects you can predict base on it, but depends on how good your premeasures are in terms of looking it from a loading environment, so ensuring that we actually understand dynamically what forces will be experienced by the 3D printed prosthesis, so I think we can be confidence that we can predict it reliability and longevity if adequately understood the environment to which the implant will be exposed to. So my response is that QbD will enable us to predict outcomes assuming we know the input data (Q 4.2)  Yes (Q 4.3)

RER
 Yes (Q 4.4) DRM  I thing 3D printing in medicine is definitely worthwhile, so I think it will benefit the field (Q 4.5)  QbD will ensure that the final product more close to optimal at the time of printing (Q 5.1) INNOV  Yes definitely, if you avoid the trial and error, the yield and the productivity will increase overall (Q 5.3)

CON-IMPROV
 I'm new in this area of optimizing using modelling and other things, but I think the biggest challenge is how you integrate into a system where there are numerous possibilities to go wrong, how you can define how is this the correct way and this is a kind of dice way to do thinks. How can you say that this factor is the real contributing factor and this factors are allowed to a little bit change. So again I think the integration will be challenging (Q 5.5)

Most used reasons for the use of QbD
PU  I think if you have a number of factors that can influence the quality of the product and write them down in a more flowchart fashion then it is easier to find out what is the step that is missing or if it needs more control (Q 4.1)  Because I know if you ask someone that is working in this area of someone that is just starting to work in this area there are endless possibilities to go wrong there are numerous factors that people ignore (Q 4.1)  As a person who expended hours in lab trying to do different things, but if you know what needs to be done exactly following a path, so I think this will help researchers that are looking for a path towards optimization of it (Q4.  Following protocols will minimise the risk, which will be end product of a total investigation starting from modelling to the numerous mechanisms used in others models and areas and apply it to this area of fabrication of 3D printed implants with nanotubes can pave the way for future enhancement in this area (Q 4.5)  So if we really follow this mechanisms we can reduce the error bar as we generally observe. Just reduce the variation in the size or in the release profile that can be achieve and that is a big achievement (Q 5.1)

Agree:
 There is not a streamline process that we can really rely yet. So the fact that now the QbD process very systematically brakes down into sections which later people can take on and then develop independently, this will be a big step towards making the process more efficient and improve time frames (Q 4.5) Agree:  I think it will improve our success rates. I think if we can harbor this process we can do better (Q 3.3)  QbD requires feedback loops so the outcomes of the product go back into the design process in a rigorous way to analyse why something failed. Moreover, this feedback loops can ensure that the fabrication produces what we are expecting (Q 4.1)  In its complete element it (QbD) will improve biomedical engineering products, yes, I think there is enough evidence. I think this is a rigorous detailed process that have been suggested here (Q 4.5)

Agree:
 Yes I believe it can be applied (Q 4.5) Agree:  I thing 3D printing in medicine is definitely worthwhile, so I think it will benefit the field (Q 4.5)

Agree:
 Following protocols will minimise the risk, which will be end product of a total investigation starting from modelling to the numerous mechanisms used in others models and areas and apply it to this area of fabrication of 3D printed implants with nanotubes can pave the way for future enhancement in this area (Q 4.5)  So if we really follow this mechanisms we can reduce the error bar as we generally observe. Just reduce the variation in the size or in the release profile that can be achieve and that is a big achievement (Q 5.1) 5 6  But the advantage that I suggest is that once the system is setup is very easy to ensure that all the processes are correct and followed, and you are able to identify where the errors are, and quickly fix the contribution of that errors (Q 3.2)  I think in designing the process where you have multiple check points along the way to say: it is my data is good at this point? Can I proceed yes or no? That's something of advantage for (Q 3.5)  I think QbD provides probably more a template to ensure that the quality of the data is as precise as possible (Q 3.7)  Trust is probably the main one (potential benefit), if the clinician and the regulatory approval bodies consider that there is a body of evidence of work showing that a comprehensive risk assessment has been undertaken they are more likely to have confidence that they don't going to get significant issues from using 3D printing (Q 3.9)  So I think QbD would definitely give us a lot more confidence to proceed with that kind of projects at least in terms of prototype that can add up (Q 5.1)  I think it comes to the rules and regulations by the authorities to be able to be successfully be integrated into the current implant market, everything has to be reproducible, scalable, and customised for all the specific patient's needs (Q 3.9) 8

IMPRV-PRAC
 I can see that this method can be the replacement to the Six Sigma method, which is really effective for conventional manufacturing. But in the case of 3D printing this is totally different, that's why I think that QbD can work here (Q 3.9)  Yes (Q 3.10)  Nothing like this ever existed like this before, so look it from this perspective, this is an enhancement in streamlining the whole product development process (Q 4.1)  There is not a streamline process that we can really rely yet. So the fact that now the QbD process very systematically brakes down into sections which later people can take on and then develop independently, this will be a big step towards making the process more efficient and improve time frames (Q 4.5)  In my practice in aeronautical industry I used it. You can see how much the aircraft industry succeeded in reducing a high risk industry to something that is really low risk (Q 3.5)  I think this risks assessment is similar to the one used in aeronautics (Q 3.6)  Yes, I think it can. For example: A knee implant from "A" company can have between 100 to 1000 combinations, between implant components, how do you select the right component for an individual. You don't, surgeons rely upon on most go to set of components and apply it to a person. But having a digital model of that person will give better outcomes (Q 3.10)  Probably (Q 3.10)  Definitely, I think it can improve current practices and improve efficiency within the service. I mean knowing all the issues and when and why those errors occurred (Q 3.10)  Yes, this can improve current practices (Q 3.10) ENHAN  Yes, because this system is much easier to implement (Q 5.4)  That depends on the government. If you are able to take it to the people that can make those decisions and convince them that the process is the one that they need to adopt, which I believe you  Definitely, especially now with personalization of implants with 3D printing. I think that QbD and computational modelling will help to design better biological implants (Q 3.4)  So depends on the application, for a long term product you can test all things and maybe is not the question of those short term projects (Q 5.4)  Definitely will enable us to match current design and give confidence to produce products as good or even better that current designs (Q 5.4)  so if we have a protocol written down were describes step by step what could go wrong and what are the chances of other factors which are robust otherwise to go wrong, that will should, then if there are able to implement it and marketed and make it available to developers, then yes absolutely (Q 5.4)  Certainty, I think that the whole process is innovation ( Q 5.3)  I think generally yes (Q 5.4) really help the researchers (Q 5.2)  Yes, because overall this field holds great promise, but there are numerous challenges that needs to be concord before enter to the implant market, and this path can help to overcome those challenges (Q 5.4) INNOV  I think QbD can act as an insurance for innovation to be truly examined and analysed. The systematic approach that QbD has, can encourage innovation (Q 5.3)  It alternately lead to better outcomes for those projects that people embark on, yes. Certainty the implementation yes (Q 5.3)  Certainty, I think that the whole process is innovation. It actually allows you to play around with simulations and 3D printing, and really rapidly test different ideas, and that is absolutely critical for innovation in this space (Q 5.3)  I guess, in terms that you develop something that is short of tested in certain range and then you just extrapolate a little bit, just by definition the extrapolation still like a new innovation, and if you had thought about that, then it will be relatively easy to extend it without having to go again through the whole application process. I think that is the interesting part, it is a good way of thinking, because you may better try to combine the approved products and invent a new one (Q 5.3)  I think of having the backing of QbD it will definitely assist innovation towards commercialization (Q 5.1)  Definitely will assist with translation out of innovation (Q 5.3)  Yes definitely, if you avoid the trial and error, the yield and the productivity will increase overall (Q 5.3)

CON-IMPROV
 I did not see in your presentation the severity of the risks and consequences to the patient (Q 3.7)  The challenge here is to convince industry and get the 3D printers' qualification (Q 3.10)  Yes, it need to include design validation and verification, traceability matrix, and design history file (Q 5.5  you should include also patient's characteristics such as biology, health condition, because patient infection and defect can change etc (Q 3.7)  It needs to be simplified in some way… for somebody that is just starting and does not have an overall picture of what QbD is, it may be look it and don't understand it. It will be nice to have a single page summary of the idea and concept before you break it down (Q 5.5)  I think this risks assessment is similar to the one used in aeronautics, but needs to be integrated with computational simulations (Q 3.6)  How we put QbD and BIM, PLM into a process. How we actually do that? That's I think is the problem at the moment, it doesn't exist in the field (Q 5.2)  There are some details there that need to be amended to get this other concept together in put in proper simulation in creating 4D models, I think needs to be a central part of what you are proposing. Also rehabilitation is an important component of the success of the implant in the person (Q 5.5)  ACCEL) Is a tricky question, because all the principle that you said all make sense, the big question is going through all the principles will really improve what is currently done? So we just can hope that it will improve (Q 3.4)  As you see the standards are relatively simple generally, but yours is like a Pandora box, very complex (Q 3.9)  I think you would put a timeframe on how long some of those things will take. Because I guess if you present it to a surgeon for somebody that has a big tumour you have two options, you cut his leg off which is you normally do, or you basically 3D printing a prototype in next five days and save this limb. So for me the question is with your concept I took that to some extend or it is necessary  In terms of stablishing the QbD regime it will be difficult to start because there are a lot of check points to go through and you have to investigate a lot of steps within the process (Q 3.2)  I think product development yes (ACCEL). But in terms of research especially in my field, there is such a pressure to continuously publish that definitely corners are cut. my perception would be that a lot of research would not take the effort to learn the whole process and implement all the procedures (Q 3.4)  It Is just the start-up time and cost to implement it (Q 3.10)  Maybe, is the useability of the results, how would someone actually going to implement the whole process would say that this is a comprehensive outline which  I'm new in this area of optimizing using modelling and other things, but I think the biggest challenge is how you integrate into a system where there are numerous possibilities to go wrong, how you can define how is this the correct way and this is a kind of dice way to do thinks. How can you say that this factor is the real contributing factor and this factors are allowed to a little bit change. So again I think the integration will be challenging (Q 5.5) 38 to go the way that surgeons and engineers go at the moment (Q 5.4) involves multiple professions, so I think along that process you are going have lots of different teams of people working in stablishing the risk management strategy (Q 5.5)