Team Symmetric - SHDC20 Full Solution by SCAN Health

www.symmetrichealthsolutions.com

Page 1 of 45

Prepared: November 9th, 2020 For: SCAN Health Judges Mr. Ron Johnson, Vice President and Chief Information Officer, Eastern Health Mr. David M. Marcelletti, SHDC20 Chair of Judging Panel; Vice Chair Supply Chain Management, Mayo Clinic Dr. Alex Mitchell, Clinician, Nova Scotia Health Authority Dr. Jayson Myers, CEO, NGen – Next Generation Manufacturing Canada Mr. René Picard, Vice President, Supply Chain Transformation, McKesson Canada Mr. Jitendra Prasad, Chief Program Officer, Contracting, Procurement and Supply Management, Alberta Health Services Mr. Mike Schiller, Senior Director, Supply Chain, Association for Health Care Resource & Materials Management (AHRMM) Mr. Altaf Stationwala, President and CEO, Mackenzie Health Dr. Natalia Wilson, Adjunct Faculty Member, Arizona State University Mr. Steve Wretling, Chief Technology and Innovation Officer, HIMSS

www.symmetrichealthsolutions.com

Page 2 of 45

From: Symmetric Health Solutions contact@SymmetricHS.com Dear Judges, We are honored to present our detailed solution for solving one of the most pressing challenges facing the healthcare industry, creating a detailed single source of truth for product data to determine product equivalence, particularly for pandemic related supplies. Our company was formed by industry professionals after witnessing inadequate item data for healthcare professionals, and an opportunity to apply the latest data technologies to age-old problems. We appreciate your time, consideration, and feedback while reviewing our solution. Thank you, Symmetric Health Solutions

www.symmetrichealthsolutions.com

Page 3 of 45

Table of Contents About Symmetric ........................................................................................................................................ 6 Symmetric’s Vision ...................................................................................................................................... 7 Personal Protective Equipment (PPE) and Pandemic Style Supplies ......................................................... 10 Design Principles ................................................................................................................................... 10 Anticipating Shortages – Item Manufacturing Locations ...................................................................... 10 Feedback and Community – Expanding the List.................................................................................... 11 Maintaining Pandemic Inventory and Item Master Information .......................................................... 12 Vetting New Suppliers During Pandemic Chaos .................................................................................... 12 How UDIs can help COVID-19 response ................................................................................................ 13 Background on Symmetric Solution .......................................................................................................... 13 Data Inputs ........................................................................................................................................... 13 1. Device Approval Documents ......................................................................................................... 16 2. Supplier Marketing and Sales Data ............................................................................................... 18 3. Issuing Agencies ............................................................................................................................ 19 4. GUDID and Future Formal Registries............................................................................................. 19 5. Data Consumers Creating New Data ............................................................................................. 19 6. Post Market Surveillance .............................................................................................................. 19 7. Manufacturer Innovates or Obsolesces product ........................................................................... 20 Data Accuracy ....................................................................................................................................... 20 Symmetric Difference on Key Data elements from GUDID data ....................................................... 21 Data Field Consumption Requirements – Categorizations ................................................................ 22 Global Master Data Reconciliation .................................................................................................... 25 Web and Mobile Application Workflows .............................................................................................. 27 Current Symmetric Workflows .......................................................................................................... 27 Data Synchronization ............................................................................................................................ 34 Implementation & Integration .................................................................................................................. 36 Primary Objective of the Implementation ............................................................................................ 37 Value-Driving Reports and Workflows .................................................................................................. 37 Feedback in the Symmetric App ........................................................................................................... 38 Ease of use for any hospital .................................................................................................................. 39 COPYRIGHT 2020 © SYMMETRIC HEALTH SOLUTIONS. ALL RIGHTS RESERVED

www.symmetrichealthsolutions.com

Page 4 of 45

Symmetric’s Business Model .................................................................................................................... 42 Eastern Health Vision & Closing Remarks ................................................................................................. 42 Exhibit A – ROI: Boston Medical Center Case Study .................................................................. 42 1. Identifying procurement opportunities ............................................................................................ 44 2. Saving time with a single source of truth .......................................................................................... 44 3. Mitigating supplier risks .................................................................................................................... 45 4. Increasing reimbursements from the supply chain ........................................................................... 45

www.symmetrichealthsolutions.com

Page 5 of 45

About Symmetric Symmetric Health Solutions is mapping the healthcare supply chain by aggregating and synchronizing product information from approval through patient use, recalls, and eventual obsolescence. Each night Symmetric refreshes, audits, cleanses and combines data from several data sources, including user feedback from our application workflows or changes in our customers’ data feeds. Core to our company’s mission, we continue to ensure our solutions are applicable and useable for all hospitals. This is represented by the fact our subscribed customers include rural community hospitals, children’s hospitals, academic hospitals, regional health systems, and large multi-state IDNs. Symmetric’s application is built on cloud platforms that provide data volume and user scalability. Utilizing the latest in machine learning, natural language processing, and optical character recognition, we have built thousands of processes to cleanse and combine medical product data sets. These create master records for each data field, shown alongside primary source data. Symmetric’s databases integrate via data feeds with ERP systems (e.g., Lawson, Infor CloudSuite, Workday, PeopleSoft, SAP), enterprise Master Data Management Applications (e.g. Oracle Product Hub, Informatica), and EHRs. Symmetric provides an online web application where users can upload supply data sets in .csv or Excel file formats of any file size for ad-hoc analysis of item adds and new contracts that may not yet exist in any system. Within minutes, our supervised machine learning matching process aligns user data to ours to show what matched and what requires human validation, for which there is a review workflow. These reviews teach both our algorithms and users about the quality and contents of the datasets, which then informs further analysis and reporting. Users can customize reports and data feeds with their items alongside our 400+ product attributes, which includes approval/clearance documents, technical specifications, instructions for use, and both functionally and clinically verified equivalent products to update their systems or download to Excel or csv for analysis. The ROI on accurate and connected data is substantial. Included in Exhibit A is a case study at Boston Medical Center, a Symmetric customer, where over $3 million in annual cost reduction and revenue capture was identified within the first year of using Symmetric. This study did not include additional labor time-savings stemming from shortened UDI-DI data implementation, faster day-to-day master data maintenance, and impacts on downstream processes from improved data access and accuracy.

www.symmetrichealthsolutions.com

Page 6 of 45

Symmetric’s Vision Creating a detailed source of truth of all medical supplies is needed to fully solve the global supplies sourcing and substitute relationships challenge. We believe that data standards for medical supplies, Pandemic Response speed, and item substitution are all interconnected. It is hard to argue that the shift in data standards in the medical device industry would have happened without the establishment of the Unique Device Identification (UDI) system by the Food and Drug Administration Amendments Act of 2007 and subsequent publication of the UDI regulation in 2013. The UDI system’s mission was to improve patient safety through public information on medical products, linked to recalls, post-market surveillance, adverse events, counterfeit identification, national security, and more. Accomplishing these goals requires the use of consistent identifiers for medical devices by manufacturers, distributors, GPOs, healthcare providers, the FDA, researchers, healthcare payers, solutions providers, and even U.S. citizens. To truly realize the mission of the UDI system, such identifiers must be in publicly accessible databases like the U.S. FDA’s GUDID. For example, say a manufacturer selling products in the U.S. registers a GTIN with GS1 for a pacemaker, making it available in the GDSN but not in the GUDID. This poses several problems. Since the FDA is unaware of this item, they cannot facilitate recall enforcement, track adverse events, or tie premarket approval with post market surveillance. They are unable to deliver on the original purpose of the UDI system. Additionally, industry adoption is delayed due to data completeness and quality issues. For example, industry adoption by payers and hospitals was delayed due to inaccuracies in implantable device identification. These inaccuracies resulted in ambiguities in defining national lists for clinical documentation and insurance claims processes. The Association for Health Care Resource & Materials Management (AHRMM) Learning UDI Community (LUC) and FDA High Risk implant workgroup was tasked with identifying solutions to this problem. As part of our company’s mission, we assisted the workgroup in finding an FDA/NLM implementable algorithm that increased implant UDI-DI flagging accuracy from 86% to 98%. Possibly more significant, we can identify the remaining 2% inaccuracy by comparing other data elements in the GUDID. This 2% can be corrected by having the data consumers and regulators inform manufacturers of the errors, then enforcing that corrections occur, and finally continuously tracking new UDI-DI inputs. This generates a feedback loop. Creating a source of truth in this complex system requires using a design (Figure 1) familiar to those in science, programming, and engineering disciplines. COPYRIGHT 2020 © SYMMETRIC HEALTH SOLUTIONS. ALL RIGHTS RESERVED

www.symmetrichealthsolutions.com

Page 7 of 45

Figure 1 – Feedback System

Symmetric adopts this approach by continuously expanding data inputs, creating more status measurements, collecting data consumers’ feedback, and then reporting it back to the data owners. We do not believe one organization, whether private or a regulator like the FDA, can create an accurate source of truth on their own. Additionally, we know the faster we can iterate on data points, the better the source of truth becomes. Our business model is facilitating workflows and advanced reporting to increase overall accuracy and insights for those using the data. We admit that errors exist and believe achieving data accuracy is a continuous process. However, we need to create mechanisms of repair that fit the ultimate goals of the system. GDSN does help facilitate this endeavor for 70-80% of the system and could allow for proper workflows to rapidly change elements not required or included in GUDID, but if this is truly a patient safety initiative, the source data keys and critical attributes need to live publicly in GUDID (in the U.S.) for devices falling under their regulation. This means continuously engaging government bodies (e.g. the FDA), standards agencies (e.g. GMDN), manufacturers, terminology organizations (e.g. NIH/NLM’s SNOMED CT), and industry associations (e.g. AHRMM). Much to our delight, we have found successful improvement of data points after providing feedback from our data analysis with each of the organizations given in example above. Additionally, after sharing analysis and voicing concern in the data feedback process to the FDA and AHRMM, a new LUC workgroup was quickly assembled, “UDI-DI Change Communication Process.” Concrete examples of change instigated by this feedback have included FDA facilitating manufacturer updates in GUDID, the GMDN agency creating new terms for classifications that were needed but did not exist, and manufacturers’ changing their regulatory data submission behavior to include clinically relevant sizes. To accurately source and identify alternatives, many attributes in addition to an accurate category must be known about an item and the vendor. Symmetric has built data pipelines and algorithms to combine information, including human feedback, on an item from any data source. Clinically relevant sizes are critical to identifying substitute products. In this example, Symmetric each night identifies discrete missing or incomplete sizes by category, then updates that value with a known quantity from any of the following data inputs:

www.symmetrichealthsolutions.com

Page 8 of 45

•

Manufacturer or vendor updating a publicly available data source (regulatory data sets, websites, documents, etc.)

•

A user scanned a label that had additional size information included

•

A user uploaded a contract for analysis that has this information in the item description

•

A user provided feedback on an item in the application

www.symmetrichealthsolutions.com

Page 9 of 45

Personal Protective Equipment (PPE) and Pandemic Style Supplies The best way to benchmark whether our solution is performant for sourcing and substituting PPE and Pandemic style supplies is to dive into how we helped health systems during the COVID-19 pandemic, but also NGOs and the community at large. To recap, these are some of the design principles for our solution that enabled us to quickly address questions as they popped up at hospitals.

Design Principles •

It should be fast at matching datasets, so that hospitals can personalize findings, and add new information as quickly as it comes up

•

It should be comprehensive in item coverage, pulling from Global Data Sources

•

It should be broad in scope, including manufacturing information, safety events, reordering catalog numbers, UDI-DIs, standardized descriptions, suggested substitutes, storage and handling, lists of manufacturers and parent company M&A information, regulatory approval documents, standardized item categories and nomenclatures

•

It should identify and rank substitutes broadly, with the ability to filter down based on attributes, such that it is flexible during emergency situations requiring off-label or unusual usage – which is a pragmatic lens into the substitution problem

The Background on Symmetric Solution section provides further details on the Symmetric Solution.

Anticipating Shortages – Item Manufacturing Locations In February 2020, as COVID-19 was emerging as a global threat, one of the first questions asked by health systems was – Where are my items made? We were able to turnaround an industry wide report of items made in Wuhan, China. The report was expanded to the rest of China as lockdown was imminent. How were we able to turn around a report like this so quickly? We leveraged UDI-DIs. We had done a previous exercise where we mapped UDI-DIs to FDA registration listing datasets, enabling us to pull in manufacturing locations. COPYRIGHT 2020 © SYMMETRIC HEALTH SOLUTIONS. ALL RIGHTS RESERVED

www.symmetrichealthsolutions.com

Page 10 of 45

Upon further analysis, it became immediately apparent that the same types of supplies necessary for pandemic response were predominantly made in pandemic-affected areas that were shutting down.

Feedback and Community – Expanding the List We continued to help our customers and some health systems determine which supplies were manufactured in shut down areas. As we did so, we realized the importance of creating a free, public resources for health systems to explore substitutes by manufacturing location of categories of critical supplies. The list of “COVID Key Supplies” categories which we provide publicly has grown organically based on community feedback, and was first published in early April: List of Medical Supplies at Risk of Shortage During COVID-19. We distributed this list to US health systems via AHRMM, and sent it to SCAN Health, the FDA, and WHO as a free resource to be shared with health systems. We updated and maintained the list every month until August of this year. An example of the community benefit was a non-customer health system being able to determine that 1/3rd of the items they purchase were being manufactured in China. They were able to do a quick comparison due to having adopted UDI-DIs internally, and mapping to the listed UDI-DIs in our file. The list ended up including the following categories of medical supplies: Ventilators, BIPAP/CPAPs, Nebulizers, Circuits, HMEs, Bronchoscopes, Oxygen Masks, Nasal Cannulas, Stethoscopes, Pulse Oximeters, Thermometers, Gowns, Masks, Face Shields/ Goggles/ Eye Protections, Swabs, Bouffant, Shoe Covers, COVID-19 Test IVDs, PAPRs, Respirators. As we received feedback from WHO, we incorporated Global Databases, including USA, Canada, UK, and Italy data sources. Our goal was and continues to be the expansion of our datasets and solution capabilities by incorporating additional item data Global Data Sources, including Canadian data sources. In addition to the Item categorical and manufacturing information, the Key Supplies file included: 1. Supply List Addresses: All supply chain addresses for manufacturing, repacking, sterilizing, etc., by item and UDI-DI. 2. All Companies in Supply Chain: All manufacturers in the supply chain of the high-risk categories’ supplies. 3. Suspect Companies: Over 1,300 companies that have used false U.S. Agent information while registering with the FDA in the wake of COVID-19. 4. NIOSH CEL Listing: All supplies approved by National Institute for Occupational Safety and Health which has a small overlap with FDA approved supplies for masks, respirators, PAPRs, etc.

www.symmetrichealthsolutions.com

Page 11 of 45

5. Additional Symmetric fields like standard descriptions, GMDN nomenclature codes, catalog numbers, and categorical Substitutes. For further details on the contents of the COVID Key Supplies file, here is a video demonstrating How to use the COVID Key Supplies File. During these times, we used our Newsletter to keep our community informed on changes in the Healthcare Supply Chain.

Maintaining Pandemic Inventory and Item Master Information The next challenge that health systems encountered was keeping track of the Pandemic Inventory for items that they were newly purchasing for the first time, in a coordinated manner. Symmetric worked with health systems to quickly research and enrich their item master information for hundreds of new Item-adds. The goal was to enrich information necessary for smooth Inventory and Purchasing processes. This Symmetric in 60 Seconds video demonstrates what an item add looks like, and the information we were capable of adding quickly.

Vetting New Suppliers During Pandemic Chaos In the chaos and confusion that became the next few months as new companies entered the healthcare supply chain space, it became clear that the next major question was “How do I know that this new company I have never done business with is selling me legitimate supplies?”. We then put together a public resource on Vetting New Suppliers in May, in order to help health systems do quick checks on Suppliers. The steps included: 1. 2. 3. 4. 5. 6. 7.

Is the supplier registered with the FDA? Is the supplier NIOSH approved? Look up the supplier’s website owner on ICANN. Do the owner and website look legitimate? Does AHRMM or the AHA have information on this supplier? Ask for references from suppliers. Ask for referrals to new suppliers from industry colleagues. Look up the supplier’s Form 5500. Verify the supplier’s TIN by logging in with your IRS account credentials. 8. Does the supplier claim to be registered in the System for Award Management? Look up their entity registration. 9. Verify the supplier’s products using the GUDID. 10. Consider spreading out key orders across multiple vendors, avoiding sole-sourcing.

www.symmetrichealthsolutions.com

Page 12 of 45

How UDIs can help COVID-19 response Finally, it became clear how other health systems needed resources to better understand the relationship between UDI-DIs, Item enrichment information, and Global Data sources (Exhibit A). These pieces of information were critical in meeting our Design Principles for the Emergency Preparedness solution. To help others better understand the importance of UDI-DIs, we created a public resource on How UDIs can help the COVID-19 response, which we would recommend to Eastern Health. These include recommendations like using UDI-DIs to track medical supplies and testing IVDs.

Background on Symmetric Solution Data Inputs

Figure 2 â&#x20AC;&#x201C; Feedback System - Inputs

Medical product data is created and consumed at a variety of points in the market, but to start understanding it all, we must work to understand why medical products exist in the first place. The creation of a medical product starts from the recognition of a disease state, or real-world problem impacting the health of persons. As people recognize these problems, they try a variety of solutions involving a mixture of existing and developing products and processes to address or prevent the disease state. Over time, knowledge is gained around which products and processes are better than others by individual persons. Some knowledge is scientifically validated to a degree of statistical certainty using the scientific method. Other knowledge is learned heuristically with potential risk of falsehood. These varied products and processes also have competing costs, environmental impacts, perceived benefits, and a range of circumstantial outcomes based on who is using them and on whom they are used. They also have varied geographic and manufacturing production throughput availability, regulatory requirements, legal protections, and material requirements impacted by sustainability. All these pieces COPYRIGHT 2020 ÂŠ SYMMETRIC HEALTH SOLUTIONS. ALL RIGHTS RESERVED

www.symmetrichealthsolutions.com

Page 13 of 45

work in tandem in one big system. In building a source of truth of medical supplies data, we must consider each data consumer’s requirements and the ultimate goals of the system. In fact, by starting with all data consumers’ requirements at the outset, we build better measuring elements, controller error checks, and ultimately value in the system by creating processes to encourage feedback. To start, let us list all the identifying “Keys” for connecting data curation and the associated consumer requirements.

List of Keys Pertinent to Medical Products (Data Curation) •

Registration Listing Numbers – Establishments that are involved in the production and distribution of medical devices.

•

National Approval Document Codes – Regulatory grouping of a family of products which has been evaluated for approval to market the product within the regulator’s jurisdiction. This may be handled by the original equipment manufacturer or a re-labeler. Examples include: o

USA FDA: Premarket Approval Number, 510(k) K number, HDE code, Biologics approval code.

Health Canada MDALL: License Number with corresponding license type and class. Contains a “device identifiers” field, which in the case of the License Number are catalog numbers.

•

UDI-DI – Identification of the smallest package unit containing a finished product.

•

Serial numbers – Specific individual products – UDI-DI + Production Identifiers.

•

Lot numbers – Batches of specific individual products – UDI-DI + Production Identifiers.

•

Unit of Use DI – Identification of each units of a consumable product.

•

Packaging DI – Identification of the available packaging levels and contained units for a given product.

•

NPI or National Provider Identifiers – Identification of healthcare providers which can be linked to products. preference or usage.

www.symmetrichealthsolutions.com

Page 14 of 45

•

EIN / DUNS– Identification of legal entities for companies and to establish their relationship to products.

•

*Supplier and manufacturer order or catalog numbers – Internally created item numbers assigned by manufacturers or suppliers to systematically track the products that they sell. o

*Currently these are keys for transitioning the system to UDI-DI

List of Fields for Data Consumer Requirements (Data Consumption) •

Supplier to supply relationships – Identification of organizations to buy a supply from.

•

Brand Names – Marketing grouping of a family of products, which can have varied shapes, sizes, colors, and configurations. Often used to refer product lines.

•

Clinically Relevant Attributes – Identification of relevant sizes, materials, and limits for products.

•

Categorical feature groupings of devices at differing levels of detail and intended for differing usage. Examples of these include: o

GMDN (Global Medical Device Nomenclature) – use to group devices by name, intended use, and common attributes, usage frequency, and whether they are implantable. Has roughly 22,000 codes of which only ~9,800 are represented to-date in GUDID.

UNSPSC – used to group devices into sourcing categories. Has roughly 3,500 codes for medical devices.

HCPCS codes – used to group devices into categories for outpatient reimbursement, has roughly 4,000 codes total of which 300 or so are the most relevant for reporting (CCodes).

•

Recalled devices – Recall identifier stored in a market impact-based system like the US market’s FDA RES system. These are identified at the UDI, UDI-DI, Premarket-approval code, or Brand name level depending on the data maturity of manufacturer reporting the recall.

•

Adverse events for devices – Adverse event identifier stored in a market impact-based system like the US market’s FDA FAERS system. These are identified at the UDI, UDI-DI, Brand name, Premarket-approval code, or order catalog number level depending on the detail provided by the reporter. Usually has poor signal on exactly which device it pertains to depending on who is reporting the issue.

•

Product Efficacy – Critical piece of information in calculating the total cost of care and comparing products and procedures.

•

www.symmetrichealthsolutions.com

Page 15 of 45

•

Labeler to manufacturer & parts component relationships – Identifications of nodes in the supply chain to make the parts themselves down to material components.

•

National approval device groupings – common schemas utilized by regulatory agencies to segment devices to regulations for enforcement. (e.g., FDA Product Code)

•

Product usage and market share – Key to determine the statistical significance of reported events, malfunctions, and outcomes, as well as market demand for groups of products.

•

Registry event relationships to real-world events & de-duplication.

•

And many more, such as device function and cost.

With these keys and requirements in mind, let us look at the lifecycle of medical supplies data in the US.

1. FDA approval 7. Device innovation or obsolescence

2. Manufacturer creates details

6. Post market surveillance

3. Issuing agency UDI-DI

5. Data consumption & creation

4. GUDID

Figure 3 – U.S. Medical Device Data Life Cycle

1. Device Approval Documents In the U.S., the first input of medical supplies data is the request and subsequent FDA approval records. For roughly 80% of FDA approved medical devices, this first piece of information is found on the 510(k)summary approval document (Figure 4) and the Premarket Approval (PMA) document. Symmetric

www.symmetrichealthsolutions.com

Page 16 of 45

regularly downloads, stores, reads, and analyzes this text using Optical Character Recognition (OCR) and Natural Language Processing (NLP). An example of this can be seen below with a medical device approval document. The wealth of information included in such documents includes: • Detailed device attributes, features, material, etc. • Intended use for procedural mappings • Product functional equivalence, testing, research, or lack thereof by mapping the 510-k predicate chain of all approvals and their relationships • Approval numbers are often used within the FDA to identify supplies in recalls, adverse events, manufacturer / sterilization / import / export registration listings.

Figure 4 – FDA 510(K) Medical Device Approval

Figure 5 – FDA 510(K) Medical Device Approval

The approval number (e.g. K061808 Figure 4) is the first product specific key and to no surprise, it includes many data consumer requirements. One of the less transparent numbers is the fact that the device was approved based on a stated equivalent predicate device K022325. If we then pull up the predicate device approval (Figure 5), we see it is based on a device that was never tested since it predates FDA medical device approval regulations.

www.symmetrichealthsolutions.com

Page 17 of 45

One aspect of our nightly refreshed substitute product algorithm is to utilize the FDA’s substantially equivalent review of products matched to item level products and attributes in a graph database. This allows for real time querying and traversing that would not be possible in a relational database. Figure 6 below is a pictorial representation of a graph database of item relations.

Figure 6 – Symmetric Item Substitute Graph Database

This Predicate Chain Analysis enables us to use the FDA’s own data, currently siloed in approval documents, to identify clinically equivalent items. During the pandemic, we have used this analysis to provide health systems with lists of substitute products for critical PPE. Beyond the details, approval codes contain important meta-data that both categorizes devices and informs data consumers of the rigor to which it was studied before coming to market. For example, devices linked to Humanitarian Device Exemption (HDE) codes, indicate little clinical efficacy was gathered for approval, but there was some evidence to support the safety of the device. 2. Supplier Marketing and Sales Data Here we have many data consumption requirements first curated by manufacturers and suppliers including catalog numbers, descriptions, and brand names. To varying degrees, some of these are passed through the system in a controlled process, but today most of these attributes live on websites and pdfs, if at all. COPYRIGHT 2020 © SYMMETRIC HEALTH SOLUTIONS. ALL RIGHTS RESERVED

www.symmetrichealthsolutions.com

Page 18 of 45

Where possible and permissible, Symmetric collects this information. 3. Issuing Agencies Due to regulation requirements, manufacturers and suppliers first work with UDI issuing agencies to create UDI-DI keys (GTIN, HIBC, ICCBBA). These are most critical to be mapped to manufacturer catalog numbers which are how the healthcare supply chain system has referenced these products historically. 4. GUDID and Future Formal Registries Again, due to regulation requirements, this is the first place where data keys come together with critical consumption requirements data. These include GMDN and FDA Product Code categorization, Premarket Submission codes, and pieces of clinically relevant attributes, catalogs, and brand names. 5. Data Consumers Creating New Data Supply chain professionals, clinicans, medical coders, and researchers all start consuming data. It’s quite difficult since the sources usually include GPOs, Distributors, GUDID, and third party data repositories. However, by actively working with their inputs, they create additional data consumption attribute requirements. Today, this is most easily seen in staff correcting master data in ERPs and EHRs, including catalogs, UNSPSC, HCPCS, clinically relevant attributes, brand names, and supplier to supply relationships. There’s a wealth of consumption requirements work continously happening here that Symmetric captures as inputs. In fact, much of this work and subsequent analysis is where the idea for Symmetric orginated. 6. Post Market Surveillance Post market surveillance is the ultimate goal of the system. If we use the current quality of recalls, adverse events, registry data availability, and longitudal studies as a measuring stick of success of our current system, it is not working. Symmetric inputs recalls and adverse events data to meet consumption requirements, but this is perhaps one of our most difficult matching processes, given the inconsistent use of product identifiers in post-market surveillance data sources.

www.symmetrichealthsolutions.com

Page 19 of 45

Additional data from implant and outcomes registries, clinical studies from publications like Pubmed, and claims analysis could be used as additional signals, but are often inaccessible for open use. Implant registries within the U.S. are notoriously difficult to sync-up and effectively create incentives for the creation of IDN-specific registries. By aligning to UDI-DIs, and leveraging a common data infrastructure, registries could close this synchronization gap. 7. Manufacturer Innovates or Obsolesces product In the final step of the cycle, manufacturers can prioritize different initiatives based on their customer’s feedback. These include the increased production of efficacious products as evidenced by outcomes, the obsolescence and decreased production of non-efficacious or unsafe products as seen from outcomes and adverse events, the recall of unsafe products, the research and development of new products to address problems in the market, and the correction of data elements and “Keys” to support the communications of all of their products. At this stage there is also activity around merger, acquisition, and product-line divestiture that needs to be reflected in the data. This is a challenge for the transferal of ownership of UDI-DIs, EINs, and DUNS “Keys”. One way to address them is to store a lineage of previous device identifiers keys for devices, with a status code of divestiture acquisition, and could be presented to varying stakeholders.

Data Accuracy

Figure 7 – Feedback System - Inputs

To know if your data is accurate and the system is working, we need to create as many checks as possible. They can be as simple as a missing required value, or as complicated as a missing GMDN categorization to accurately describe the submitted descriptions. Internally, Symmetric continously builds error checks between all of the inputs. In fact, by using the widest net of requirements (400+ COPYRIGHT 2020 © SYMMETRIC HEALTH SOLUTIONS. ALL RIGHTS RESERVED

www.symmetrichealthsolutions.com

Page 20 of 45

fields and growing) in building inputs, mappings, and connections in the system, we enable thousands of status and error checks. The faster you can see these statuses and errors in the system, the faster you can respond and improve the data if there is a standard change submission process. Hence, our data sources are input as often as possible (which today is daily, if the source updates at that frequency). Part of this ingestion is rerunning all of our thousands of algorithms that create a daily report stating the accuracy of our data sources. Staying true with our vision of the system, we’ve focused substantial effort on the GUDID. Figure 8 shows an example high level report of metrics comparing GUDID source data quality to our own. Symmetric Difference on Key Data elements from GUDID data

• Thousands of rules, generating additional rules, to correct and populate our attributes. • Machine Learning combined with advanced text parsing focused on medical device terminology. • Automated daily report sent out on data quality & completeness metrics.

Figure 8 – GUDID Error Identification & Correction

For example, within the GUDID dataset (which as of this writing spans roughly 2.8 million medical devices), we pull catalog number information from the version model number field, descriptions, and other data sources like manufacturer listed catalogs from their websites. Additionally, we work to continuously map approval numbers not included in GUDID to UDI-DI’s. Lastly, we tie back supplier to supply entity relationships and hierarchies in the form of Entity Registration Numbers, DUNs, and TINs. All of these data points allow us to collect our keys in the system to map to additional data sets more accurately, such as recalls and adverse events (Figures 9 and 10):

www.symmetrichealthsolutions.com

Page 21 of 45

Figure 9 – Recall UDI-DI Mapping

Figure 10 – Adverse Event UDI-DI Mapping

Data Field Consumption Requirements – Categorizations Our analysis currently shows that the most common inaccurate GMDN assignations in GUDID stem from systemic mismatches between how a device is approved and how the device is packaged or labelled. For example, surgical trays & systems containing both orthopedic implants and instruments are usually approved together by the FDA and other regulators. Manufacturers then use this initial FDA approval code as their categorization to assign GMDN terms at the UDI-DI level. This results in instruments, accessories, and implants all falling under a single GMDN term code (e.g., “Bone-screw internal spinal COPYRIGHT 2020 © SYMMETRIC HEALTH SOLUTIONS. ALL RIGHTS RESERVED

www.symmetrichealthsolutions.com

Page 22 of 45

fixation system, sterile”). We can easily spot these issues from all the inputs gathered and measured at the UDI-DI level. Why does this matter? Accurate UDI-DI categorizations fulfill both a critical consumer data requirement as well as expand user engagement in the feedback process. For over two years, AHRMM’s LUC categorization and implant workgroups stalled in providing a solution to data inaccuracies and errors. Without the proper inputs, measuring elements, and status checks of the system they were unable to find a pathway forward. Symmetric became involved with the implant identification workgroup and helped drive the successful resolution by synthesizing the data inputs, measuring elements, and reporting on the status of the discrepancies to the workgroup participants, for feedback on resolution. See Figures 11 on the current state, and figures 12 and 13 on the two possible future state measurements for creating an accurate implant flag.

FDA Implant List: Class 1: Class 2: Class 3: Unclassified: HDE: Not in Current FOI: Instrument (only)/Accessories: Missing Implant: Accuracy:

FDA+GMDN Implant List: Class 1: Class 2: Class 3: Unclassified: HDE: Not in Current FOI: Instrument (only)/Accessories: Missing Implant: Accuracy:

FDA Product Code Implant List Accuracy 781,895 % Total *Missing Implant: 6,774 % Total Missing SHS 815 0.10% Class 1: 2,397 0.31% 732,462 93.83% Class 2: 4,222 0.54% 38,116 4.73% Class 3: 126 0.02% 9,861 1.25% Unclassified: 21 0.00% 500 0.06% HDE: 6 0.00% 141 0.02% Not in Current FOI: 2 0.00% 103,793 6,774 Figure 11 – FDA Product codes Implant Accuracy (current state, March 2019) 86%

FDA Product Code (Implant True) + GMDN (Concept Code Implantable) 690,786 % Total *Missing Implant: 7,209 % Total Missing SHS 473 0.07% Class 1: 2,397 0.35% 648,787 93.92% Class 2: 4,401 0.64% 33,433 4.84% Class 3: 382 0.06% 7,700 1.11% Unclassified: 21 0.00% 334 0.05% HDE: 6 0.00% 59 0.01% Not in Current FOI: 2 0.00% 13,119 7,209 Figure 12 – FDA Product codes and GMDN Term Accuracy (option 1, March 2019) 97%

www.symmetrichealthsolutions.com

Page 23 of 45

GMDN Implant List: Class 1: Class 2: Class 3: Unclassified: HDE: Not in Current FOI: Instrument (only)/Accessories: Missing Implant: Accuracy:

GMDN (Concept Code Implantable) 706,881 % Total *Missing Implant: 441 % Total Missing SHS 10,340 1.46% Class 1: 0.00% 654,859 92.64% Class 2: 185 0.03% 33,559 4.75% Class 3: 256 0.04% 7,627 1.08% Unclassified: 0.00% 334 0.05% HDE: 0.00% 162 0.02% Not in Current FOI: 0.00% 22,446 441 Figure 13 –GMDN Term Accuracy (option 2, March 2019) 97%

Symmetric’s recommendation was to use the GMDN Explorer term for Implantable devices. The top arguments for this approach included: 1) The GMDN agency change control process which provides flexibility to select which categories are implantable based on user feedback and create new categories where a distinction is necessary. 2) Its ability to meet many user requirements by incorporating other Explorer terms for regulators, insurers, accreditation agencies, manufacturers, and hospital workflows. (e.g., Attribute: Bioabsorbable, Surgical Invasiveness: Short-term vs. Long-term, Name: Suture, Staple, etc…) 3) It is a regulated required field on UDI-DI assignations, and the responsibility to correct lies with the manufacturers. 4) It does not require changes to the FDA’s product codes and impact on their regulatory processes. GMDN Explorer terms pertinent to Implantable devices

Figure 14 –GMDN Explorer Terms

Unsurprisingly, through this analysis we crossed over into discussions pertinent to the categorization workgroup and a deep collaboration with the GMDN agency, where we continue to have discussions on behalf of our customers on improving data accuracy.

www.symmetrichealthsolutions.com

Page 24 of 45

Global Master Data Reconciliation The implant flag example shows how we take multiple stakeholders’ inputs and produce derived fields for consumption. For the purposes of establishing global master data on product information, we extend this concept to every field. The steps of this process include: 1) Creating field-level definitions with input from multiple stakeholders. 2) Generating business rules to create record-level guard-rails and error checks based on other fields in the record which have already been assigned. 3) Establishing a feedback loop where reviewers can push data change requests and consumers can pull them as soon as approved. There are two additional examples of global master data fields provided below. Both examples demonstrate Symmetric’s data cleansing processes. The first on a simple True or False flag for the single-use item field and the second is a slightly more complicated True or False flag derived from text descriptions for the kit field. The first step for properly flagging a single-use item in this first example is to create the definition for the field. We can pull existing definition from various stakeholders to jump-start this process. For example: 1) GMDN Explorer Term: CT981 Single Use - Devices intended to be used only once, or for only one patient during one medical procedure or short term, and then discarded if not already rapidly absorbed. 2) GUDID For Single-Use Flag: Whether the device is intended for one use or on a single patient during a single procedure. These definitions are similar enough to assume that if the assigned GMDN is flagged Single-Use “True” and the record-level Single-Use flag is “False”, we measure an error that requires feedback validation. By pulling in additional datasets from customers, and allowing them to push change requests to the data owners (e.g., manufacturers in the case of GUDID, and GMDN agency plus manufacturers in the case of errors that impact both), we close a feedback loop and arrive at the accurate value for these conflicting flags on the single-use field. In this second example, we introduce the concept of workflows as feedback loops. We have an internal workflow for comparing information stored in unstructured fields like product descriptions, to

www.symmetrichealthsolutions.com

Page 25 of 45

structured fields. We use trained natural language processing text parsers to pull the descriptions from multiple data fields and compare the application of categorizations for further review. In the case of the product with UDI-DI of 00382830015790 there are three data inputs, a measuring stick that compares current vs. new nouns, and a workflow to ask for feedback in validating whether the new noun is correct. Data inputs: •

GUDID description: “Natus Photic Stim Kit”.

•

The manufacturer assigned GMDN in GUDID: “11474 - Electromyograph”.

•

The FDA categorization product code: “GWE - Stimulator, Photic, Evoked Response”.

•

GUDID Kit Flag: “False”.

Figure 15 –Nouns Review Process

This record was chosen because the field “description” was added to the noun parsing workflow, and it determined that the previous noun, selected based off the assigned GMDN, was likely incorrect. The algorithm prompts for intervention necessary on this noun, and the review shows that the product record has an issue with both the Kit flag and the assigned GMDN, since Photic Stimulators are not Electromyographs. Each time records are reviewed; we reduce the amount of manual work required by running our supervised learning algorithms nightly to adapt the assignation and error checking of unreviewed records.

www.symmetrichealthsolutions.com

Page 26 of 45

This was an example of a data quality workflow. Users interact with it by providing feedback on individual fields and uploading datasets. Symmetric Health Solutions and data owners then establish curation processes to guide user feedback. It is our vision to provide these discrepancies to data consumers at the point of data use either with value-added embedded API integrations or within our application’s opportunity review workflows. There are thousands of fields, millions of product records, and many data consumers. No one party can address all these discrepancies. Therefore, crowdsourcing reviews intelligently by working them into existing workflows and providing feedback to the dataset owners provides ROI for all consumers of the data.

Web and Mobile Application Workflows Symmetric Health Solutions’ application consists of data related workflows required in hospitals for sourcing, standardization, and value analysis savings opportunities as well as ERP and EHR master data management (e.g. UDI-DI synchronization). By streamlining these complicated data processes and providing a venue to quickly identify priority errors for review, we systematically encourage users to provide data feedback that is then shared with the applicable data owners. In summary, as hospital users review data in these workflows, their reports themselves become better. Current Symmetric Workflows First and foremost, the “matching workflow” consists of uploading any supply related file (e.g. item master, preference cards, purchases orders, etc.) to our application for the purposes of matching to UDIDI keys or our 400+ fields. In minutes, the file is automatically matched to UDI-DI keys with a confidence field showing whether further review is required. This algorithm, similar in nature to our data aggregation techniques, consumes all user designated upload fields (e.g. manufacturer name, vendor name, manufacturer catalog number, vendor catalog number, descriptions, UNSPSC, HCPCS, etc.). The first report generated contains information around match improvement and indicates to the user both where data errors may have existed prior to input and where to focus efforts on reviews. Users are then guided through a process to auto-approve matches with high confidence and review those with lower confidence. The items are displayed with a few key attributes side-by-side and the user can cycle through similar items to select the matching product. The attributes for the user reviewed matches are then incorporated into the mappings for future matching processes, creating our first feedback loop.

www.symmetrichealthsolutions.com

Page 27 of 45

Figure 15 – Match review summary in minutes

Figure 16 – Supervised learning match review

The search workflow contains the global master of Symmetric data and can be easily searched, filtered, or drilled into by categories. After matching, the uploaded items can take advantage of these same processes alongside the enriched Symmetric fields pertaining to the items which were engineered by Symmetric with customers feedback, or were pulled from the data sources in Exhibit A. The results can be downloaded as csv or xlsx files or could be accessed via API calls to directly interface with client ERP, EHR, or MDM solutions. The easily searchable format allows user to find items or groups of items quickly, with all pertinent information and offers detailed product pages for each item. Examples of enriched fields include the COPYRIGHT 2020 © SYMMETRIC HEALTH SOLUTIONS. ALL RIGHTS RESERVED

www.symmetrichealthsolutions.com

Page 28 of 45

adverse events and recalls fields associated at the product level, which could be passed on to clinicians through the EHR. We offer a second feedback mechanism in the search workflow, where consumers of the data can push issues with specific fields and, upon negotiating an approval process with the data owner, the feedback will be incorporated into the data and available for everyone. This secondary feedback mechanism applies to the shared Global master, hosted by Symmetric, and data sources which are shared between customers of Symmetric using our sharing mechanisms. Currently these are set-up for sharing between teams within an organization and could extend to publishing and sharing datasets with other users of the Symmetric Platform. Users are incentivized to provide accurate data and participate in the further cleansing of this data to increase ROI generated from the resulting reports and workflows or to resolve mission critical supplies shortages (e.g. PPE during pandemic). Some of these additional workflows currently include: • • • • • •

Standardization Savings Opportunities Backordered Supplies Alternatives HCPCS Missed Reimbursement Contract Overspend Against PO VA Contract Benchmarking Categorical Spend Analysis

www.symmetrichealthsolutions.com

Page 29 of 45

Figure 17 – HCPCS Missed Reimbursement

Figure 16 – Standardization Savings Opportunities

As an example, the Standardization workflow provides another instance of an excellent feedback loop with data review. This workflow asks clinical users to provide feedback on groups of products that Symmetric identifies are substitutable, prioritized by category, PO back orders, or savings opportunity. It is also flexible enough to consider contracted statuses and pricing from multiple data sources. The value of this report is twofold, first reducing spend and secondly, sharing clinically approved alternatives to add an additional data input into the system. We utilize a custom algorithm to generate lists for substitution of medical devices, incorporating many of our data mappings and attributes which learns from the Standardization workflow results. We plan to expand the type of feedback clinicians can provide during the Standardization reviews and the information available to them to include things like procedural mappings, reimbursement opportunities, and potentially substitutable procedures. The Standardization and Backorder workflows are ones where we plan to soon add the ability for clinical users to write comments and add standardized feedback that is then shared with other users of the Symmetric Platform. Figures 17 and 18 show how one can search for substitutes during backorders or look for standardization opportunities.

www.symmetrichealthsolutions.com

Page 30 of 45

Figure 18 – Search

Figure 19 – Add additional information

www.symmetrichealthsolutions.com

Page 31 of 45

Figure 20 – Click document links to see more specifications

Figure 21 – Confirm specifications and tests

www.symmetrichealthsolutions.com

Page 32 of 45

Figure 22 – Click substitutes to compare products

Figure 23 – Compare products, with all Symmetric details and documents, then provide structured feedback for all customers

www.symmetrichealthsolutions.com

Page 33 of 45

Figure 24 – Compare testing and specifications documents side by side as part of product comparison review

Data Synchronization From the LUC’s workgroup output it was clear that better formal change control processes for data source updates of GUDID were going to be necessary to accelerate the adoption and usage of UDI-DIs and their accompanying data for usage. To start with this endeavor, one must establish processes for: •

Submitted contact information capture & updates

•

Change communications

•

Emails & APIs for submission of changes

We’ve started this process by pressing the FDA and AHRMM during the May 2019 UDI conference, leading to the LUC UDI-DI Change Communication Process Workgroup, as well as collecting and recording contact information when reaching out to manufacturers on the most egregious of UDI-DI errors (e.g. duplicate records). Additionally, one must enable ERP and EHR data consumption specific requirements for all users in this process. Our application has been designed to allow for •

User-driven schema mappings.

•

Pulling from the same source of truth – updated daily.

•

Provisioning of all data fields for the usage and consumption of all customers.

www.symmetrichealthsolutions.com

Page 34 of 45

Our philosophy to schema and data mapping is to internally handle the storage of the data inputs for a field and generate fields that are ready for consumption based on the needs of our customers. If one customer needs a field mapped as true/false and another as Y/N we can apply transformations to that field and expose both within our application. Device Packaging strings are an example of this. We provide the following packaging string options for the same supply, making it flexible for whomever the consumer of data is.

Figure 27 – Multiple ways to access and upload packaging information into hospital systems depending on their configuration.

While it might seem like different users are consuming different data, the core information is stored in the most flexible format internally to support customer requests. Other fields are then derived from this information accordingly. For example, we could create a data field for color that could store color in HEX, RGB, or other color formats instead of with text like “BROWN” depending on the needs of our users. These could all be mapped to each other in logical crosswalks or be defined differently, but internally these would all be stored at the highest level of detail required to satisfy our customer’s needs. We start small, focusing on what users care about consuming. This involves enabling users to map their Key fields to the Symmetric schema, which is informed by the GUDID, HL7, GS1 attribute schema, NDC schema, and user requests for information. If users request that information be provided in some other derived way, we engineer the fields such that they are all derived from the same core data but are provided for consumption as our users need them. Each mapped Key field unlocks additional analytics access to our databases and unlocks additional workflows pertinent to Medical Devices. For example, we currently ask for vendor, manufacturer, vendor catalog, and manufacturer catalog number fields. These are standard fields that most ERP and COPYRIGHT 2020 © SYMMETRIC HEALTH SOLUTIONS. ALL RIGHTS RESERVED

www.symmetrichealthsolutions.com

Page 35 of 45

MMIS systems contain across the board. As another example, to create a report to view missing HCPCS data from an item master, users need to mark which of their item master attributes contains their current HCPCS data Normalizing each customer data field to a standard schema is difficult without defining the purpose of each field. Therefore, we provide field definitions accessible to all users around what a field means, and plan to potentially extend these to be public facing for comments and clarifications. As a baseline, these are just text fields with a variety of modifications and checks applied to them such as removal of superfluous characters and mappings of merger and acquisition activity. For each field we perform custom logic to help align the data. For example, we use a data model based on the assignations of DUNS numbers submitted to GUDID that are not exactly perfect for identifying enterprises. You would ideally want a combination of DUNS, and Employer Identification Numbers (EINs / TINs / FEINs). But this would only be applicable within the US, and to extend globally would require further incorporation of regional registrations and mastering to a single parent, division, company record hierarchy. Our transformations and modifications handle such subtle issues. Lastly, for the purposes of establishing consensus of what the manufacturer and vendor catalog numbers are for a given supply, one needs to embed the usage of UDI-DI’s into the manufacturing process as well as the sales and ordering process to be able to get feedback on whether they are sufficient to support transactions or require modification. The entire purpose of packaging UDI-DI’s is to be able to support all configurations of how one would order a part. There would be things like identification of pallets, cases, etc., which could be extended with a supplier – distributor – to UDI-DI data model for enhancing the packaging identifiers.

Implementation & Integration A goal of an implementation with Symmetric is to avoid any big-bang changes in data consumers workflows, but still be able to collect user input to create the feedback loops needed for a continuously improving source of data. The below steps attempt to aim to achieve that goal by aligning themselves with existing user workflows. Ideally, the only real implementation work would be for data consumers to create upload and download flows via data feeds or manual uploads to pull and push data from their existing systems to Symmetric, without changing their current workflows for reviews and updates. The

www.symmetrichealthsolutions.com

Page 36 of 45

average consumer would only need to know about Symmetric for additional application tools they find helpful (for example Symmetric dashboards).

Primary Objective of the Implementation Fix Source fields and keys first, address additional attributes second. To start an implementation with Eastern Health, Symmetric Health Solutions would: 1. Gather all the errors and review queues that have been built up over time to flag where there are issues in the GUDID data, and other data sources as a starting point. 2. Go through a process to map data fields and records that are pertinent to Eastern Health by collaborating with Eastern Health data owners and data stewards. 3. Flag additional errors based on the Eastern Health’s datasets which would create review queues to bring-in manufacturers on approving or rejecting suggested changes to records based on the highest impact, defined by any number of variables which could include: a) implantability, b) packaging ambiguity, c) volume, and d) unit cost. 4. Work as part of the UDI-DI Change Communications Process workgroup to establish a channel for centralized delivery of feedback to the manufacturers. 5. Generate RESTFUL API’s to provide these change logs for action and receive input back into our application, as necessary. This approach provides the most benefit to the entire ecosystem, as it focuses on correcting primary data sources where possible while providing data accuracy synchronization value to member participants.

Value-Driving Reports and Workflows The current report offerings include: • • • • • •

Item Matching Standardization Backorders and Pandemic Supplies Substitutes Contract Overspend VA Contract Benchmarking Categorical Spend Analysis COPYRIGHT 2020 © SYMMETRIC HEALTH SOLUTIONS. ALL RIGHTS RESERVED

www.symmetrichealthsolutions.com

Page 37 of 45

• • • • •

Packaging String Matching UOM Optimization Recall and Adverse Event Notification Item Master Data Quality (e.g. standardized Item Descriptions) Item Additions Data Enrichment

An implementation with Eastern Health would focus on the reports that provide the most value, or any additional reports related to items. Each report would be specific to job roles and project-specific team members at Eastern Health. Each of these reports requires accurate information to be valuable, encouraging users to verify the most impactful fields and providing feedback to Symmetric and upstream data curators on the most important data.

Feedback in the Symmetric App To start, data management personnel commonly use spreadsheet software such as Excel for their daily work. Being able to make changes there and post them back to Symmetric may be an easier workflow for these users in some cases than using the Symmetric app UI. While they would not be getting the full ease-of-use benefits of our UI, it would be an easier change in workflow and easier to adopt. Additionally, we expect that these reports, while most easily used in the Symmetric app, will need to be exposed in other forms more easily digested by users to increase adoption and thereby collect more feedback. For example, Symmetric data may power downstream applications, such as an ERP. If an error is found by an ERP user in the ERP software, it would be easier for the user to just correct the field there and submit a change request from within the ERP itself, and have that change propagated back up through the system. Sometimes Symmetric can capture this indirect feedback automatically by comparing uploaded user data to older versions of itself and scan for changes that users have made without them explicitly noting a change. There are multiple steps to this implementation that Symmetric is currently undertaking: •

For manually uploaded data, verifying that the data is the same field as previously uploaded data. For example, uploading an updated item master we can scan for changes, but an uploaded vendor master cannot be compared to an old item master. This is easier for automated data feeds since the same feed will be for the same data.

•

Requires small changesets to help with accuracy. For example, if the changes between two versions of the data include removing item A and adding Item B, and Item B is a substitute for COPYRIGHT 2020 © SYMMETRIC HEALTH SOLUTIONS. ALL RIGHTS RESERVED

www.symmetrichealthsolutions.com

Page 38 of 45

Item A based on our substitution algorithm, there is a high chance this was an intentional substitution and that Item B was validated as a substitute for Item A, and that feedback can be incorporated into the system. However, if there are many changes between data versions and there is overlap between substitutes, we cannot confidently make statements about substitution. Smaller changesets between data versions helps alleviate this issue (and data feeds help create these smaller changesets), but there is continuing development of the heuristics for this kind of inferred feedback.

Ease of use for any hospital Subscribers to our application do not need to go through any implementation process to start using our software and deriving value. As a cloud-based solution, there is not even a requirement for a Symmetric representative to go on site unless it is expressly desired by a customer. Subscribers can upload Excel and CSV files off any size and download the data elements that they need as they see fit. The process is: 1) User uploads a supply file

Figure 28 â&#x20AC;&#x201C; Upload files and match

2) We align their data to ours, exposing additional attributes that can be appended to the original, uploaded supply file 3) The user selects which additional attributes they are interested in

www.symmetrichealthsolutions.com

Page 39 of 45

Figure 29 – Select 400+ Attributes

4) The user downloads the supply file with the additional attributes appended or use the reports and web application with their uploaded data.

Figure 30 – Download any size mapped file & fields

This workflow allows users to get immediate value from the product and see some benefits of aligning data by giving them access to fields that might be immediately relevant (for example, humidity or temperature handling conditions for items they are currently analyzing).

www.symmetrichealthsolutions.com

Page 40 of 45

Lastly, anyone using any piece of the Symmetric web applications can provide feedback on data elements to improve their accuracy. An example of this process is depicted in Figure 31.

Figure 31 – User feedback on data quality https://youtu.be/oCQnYscMapc

www.symmetrichealthsolutions.com

Page 41 of 45

Symmetric’s Business Model Symmetric Health Solutions operates on a subscription-based SaaS model. There are several tiers of subscriptions with access to various workflows and they are billed on a per user or per data feed basis based on the needs of the customer. We also offer services for integrations or one-time alignment of customer data. Whether it is a small hospital interested in the full reporting and analytics platform or a larger system just interested in clean data feeds, we have a solution. Symmetric’s core competency is data, analysis, and the continuous collaborative improvement framework. We want to continue improving the industry’s data and finance this endeavor through application subscriptions where we have shown ROI for subscribing hospitals through both time and cost savings identification. A case study showcasing Symmetric’s success at Boston Medical Center is presented in Exhibit A.

Eastern Health Vision & Closing Remarks Our vision for Eastern Health is to continue building out our contextual product substitution methodology and healthcare provider community for data quality and product substitution feedback. The intent for this communal approach is to not only for use within the healthcare organization but also with peers and data creators on sourcing quality item data efficiently. Our current, single-page reports combine unmatched user data with extremely pointed, well-explained corrective actions (e.g. impossible packaging strings, item data values on manufacturer websites not in GUDID, duplicate identifiers for the same supply, possible vendors, product substitute comparisons, etc.) and are viewable in excel or our application. We believe that a collaboration with Eastern Health would establish demand on both data suppliers and consumers to fill-out and map the contextual product substitution graph for healthcare. All this work can then be combined with supply chain “signal” data, such as real time scans from Eastern Health and other Symmetric community member systems to provide unmatched visibility & prioritization of industry-wide data integrity, interoperability, and safety. Symmetric Health Solutions can change this industry, but not alone. If Eastern Health were to elect to work with us on this proposal, they would join a network of hospitals, regulators, standards agencies, world health organizations, and manufacturers collaborating with us towards generating a source of truth for product data that is detailed enough and structured to accomplish healthcare mission critical tasks like predicting clinically substitutable products. We thank you for your time and consideration in reviewing this proposal.

www.symmetrichealthsolutions.com

Page 42 of 45

An academic medical center identifies $2 million in supply chain savings by using Symmetric Boston Medical Center powered by Symmetric Health Solutions, the new standard for supply chain data and intelligence.

About the Health System Boston Medical Center (BMC) is a non-profit, 514-bed academic medical center. Located in the historic South End of Boston, Massachusetts, BMC provides medical care for infants, children, teens, and adults. It is the largest safety-net hospital and Level I trauma center in New England. *Update as of November 2020: •

•

Successful implementation of Contract Management, transitioning from an On-Prem to a Cloud ERP solution (Lawson to Infor Cloudsuite), which included 600k+ GPO and Local contracted items matched to UDI-DIs. Utilization of UDI-DI information and Symmetric associated data on Manufacturing locations, Manufacturers, and Suppliers to help BMC during 2020's COVID PPE supply research. Standardization of Item Data informed by manufacturer submissions to global regulatory datasets like GUDID, on topics including Standard Item Descriptions, Item Manufacturers, and Label information. Ongoing maintenance of providing packaging and duplicate UDI-DIs to Epic via an interface for the purposes of Clinical Documentation and Implant tracking.

FOUR AREAS OF IMPROVEMENT The projects listed below were implemented from March 2019 – July 2019.

1. Identifying procurement opportunities ................................................................................................ 44 2. Saving time with a single source of truth .............................................................................................. 44 3. Mitigating supplier risks ........................................................................................................................ 45 4. Increasing reimbursements from the supply chain ............................................................................... 45 COPYRIGHT 2020 © SYMMETRIC HEALTH SOLUTIONS. ALL RIGHTS RESERVED

www.symmetrichealthsolutions.com

Page 43 of 45

1. Identifying procurement opportunities CHALLENGE

The Health System had decided to transition to a new enterprise resource planning (ERP) system, but the existing contract management module was in a half-implemented state. The lack of a functional system made it difficult for procurement to keep track of spending under management and maintain contract files. SOLUTION & HIGHLIGHTS

The Health System engaged with Symmetric to cleanse and master its disparate contract files. After cleansing, the health system loaded contracts and purchase orders to Symmetric to analyze a variety of opportunities. • • •

Identified $1M in pricing credits from contract/purchase order/invoice misalignments. Mastered four data sources into one contract file. Identified up to $2M in medical/surgical supplies standardization opportunities.

2. Saving time with a single source of truth CHALLENGE

The Health System was looking to explore point of use scanning of UDI-DI barcodes in procedural areas, but limited I.T. resources made it challenging to prioritize. Matching their item master to millions of GUDID UDI-DI records was not going to be a simple task to tackle. “Symmetric is on the cutting edge of standardizing data within healthcare.” – Joseph D’Amore MDM Manager, Boston Medical Center

SOLUTION & HIGHLIGHTS

The Health System used Symmetric Health Solutions’ match tool and medical device database to automatically map their item master, contracts, and purchase history to UDI-DIs. The Health System’s value analysis team was then able to compare products with one source of truth. • • •

87% of the 60,000-record item master to UDI-DIs and interfaced to EHR in under a week. Improved item & contract lines addition process while simplifying maintenance of files. Achieved automatic identification of standardization opportunities.

www.symmetrichealthsolutions.com

Page 44 of 45

3. Mitigating supplier risks CHALLENGE

The Health System was looking to get definitive answers to alternate suppliers during backorders. They were not getting the information they needed quickly enough from their current suppliers and distributors. SOLUTION & HIGHLIGHTS

The Health System used Symmetric Health Solutions’ medical device and pharmaceutical databases and exploration tools to understand their supplier risks better. • • •

Assisted in resolving five backorders to date. Building automated substitute list into ERP for future backorder resolution. Identified over 1K recalled items purchased in the last year.

4. Increasing reimbursements from the supply chain CHALLENGE

The Health System was looking to expand its diversity supplier program. The expansion of a diverse supplier program would help the health system qualify for a higher insurance reimbursement tier, but it was spending too much time researching supplier certifications. SOLUTION & HIGHLIGHTS

The Health System engaged with Symmetric to confirm their current diverse supplier list and, in the process, found areas of opportunity to improve supplies reimbursements. • • •

Validated diverse suppliers list and suggested vendor master updates in under a week. Provided a target list of diverse suppliers to help rank future procurement efforts. Identified $600K in missing supplies outpatient reimbursement from missing HCPCS.

“It is my firm belief that many healthcare organizations would find Symmetric’s software and expertise extremely valuable. The collection of information they have built can dramatically impact an organization’s efficiencies and bottom line.” – Joseph D’Amore MDM Manager, Boston Medical Center

www.symmetrichealthsolutions.com

Page 45 of 45