By Alan Gresch
The data-driven concepts of evidence-based medicine that are constantly improving healthcare delivery also apply to improving outcomes and optimizing performance in managing technology and facilities.
However, a prerequisite is having complete and reliable data to mine. The old adage, “you can’t improve what you can’t measure,” always holds true. Without clean data and standards for keeping the data clean, you’ve eliminated opportunities to improve your processes and bring value to your organization. Below is key information that will greatly assist in laying the groundwork for cleaning your data, creating and implementing rules to ensure the data is complete and accurate, and establishing processes and policies to keep your data clean.
Most healthcare technology management (HTM) and healthcare facilities management (HFM) departments have been collecting information in their Computerized Maintenance Management System (CMMS) for decades with no rules around how that data should be captured and maintained. The resulting mess ends up providing little or no value for reporting and analytics. Even if the data was cleaned at one point, without a process to sustain cleanliness, the data can quickly lose its integrity and value. It is critical to first stop the garbage from going in.
Data integrity policy
Before doing any cleanup, first create and adopt a data integrity policy. This policy should establish rules for how staff should populate every field of the database, with clear definitions, nomenclature standards, and format. The policy should outline system security, assigning the lowest level of access possible for each role to effectively complete their work. Request, Failure Fault, and Result codes should be minimized and clearly defined. Workers, work centers, facilities, accounts, sources, device categories, models, manufacturers, assets, schedules, materials, procedures, and work orders all need to be defined and adopt an organizational and/or national standard whenever possible. One policy I created was 16 pages long and included a compliance commitment form that every staff member was required to sign. I’ve shared the policy with many in the industry and can make it available to others upon request to firstname.lastname@example.org. Note: for the policy to be effective, you’ll will need to extensively train staff to ensure there is clear understanding of the process and expectations.
Data cleanup process and standards
Once you have mechanisms in place to prevent the input of additional bad data, begin the process of cleanup. I have seen numerous examples of data issues, which I’ll illustrate through the example of a GE AMX 4+ Portable X-Ray Machine. Of the databases we examined in one project, GE was represented 19 different ways, the device model name was represented 8 different ways, and the device category was represented 17 different ways. As a result, we had a grand total of 2,584 different permutations for representing a single device! You can see why it is important to adopt a standard and stick with it. Whatever system you decide on, the goal is to get rid of duplicates and enable easy reporting. It is recommended to rely on one of the two widely-adopted available standards:
• ECRI Universal Medical Device Nomenclature System (UMDNS) — This is currently the most complete list our industry has as a standard, and has been a great resource to many who use it entirely or to adopt their own version. The format starts with the most general functional description and adds more specific descriptors separated by commas. FDAs Global Medical Device Nomenclature (GMDN) has used this as the basis for its system.
• FDA Global Medical Device Nomenclature (GMDN) and Global Unique Device Identification Database (GUDID) — The purpose of the GMDN and GUDID is to provide a single, global nomenclature system by which authorities can regulate medical devices. Each listing has an associated code which represents the generic descriptor. The structure is regulated by ISO 15225 and is like the VIN number structure utilized by the auto industry that can track recalls and user information. The Unique Device Identifier is made up of both a Device Identifier and a Production Identifier.
There are several companies, including Accruent, who can do data cleanups for you. If you decide to tackle the data cleanup yourself, start with exporting a report to Excel with just manufacturer, device category and model.
For manufacturers, remove things like Inc., Co., LLC, Ltd, etc., as they provide opportunities for variance without adding value. Similarly, remove all divisions, unless you have a strong business case for not doing so. Fully spell out words like “products” or “instruments”, or just remove them if they aren’t necessary. Check for acronyms against spelled out names. Lastly, decide your manufacturer rule — to keep up with acquisitions or add in the same model with the new manufacturer list. You can also create an Also Known As (AKA) field to capture these variations.
For device categories, group them together by risk groups to start. Then, prioritize based on top issues such as AEM, PM Procedures, or those items that technicians can’t even identify based on the information captured. Choose whether to spell out or use abbreviations, then be consistent throughout (i.e., O2 versus Oxygen). Lastly, filter by key words to group common names together for easy standardization (i.e., “pump”, “anesthesia”, “defib”, “analyzer”, etc.). This allows you to easily identify and fix outliers.
For model cleanup, place your newly cleaned device categories and manufacturers with the model names/numbers for easier comparison using the VLOOKUP feature in Excel. Sort by manufacturer and device category to get them grouped. Then, strip out all special characters to find duplicates. Focus on models you know your staff has trouble with due to many brand names. Lastly, create a model name field if you don’t have one.
Importantly, go out and physically capture information that can’t be discerned by what’s been entered into the database. Remember, you’re likely dealing with decades of garbage.
Once you have your manufacturer, model, and device categories cleaned up, go through your CMMS codes and clean them out to the bare minimum, and clearly define each code and how it is to be used. You do not want to have a situation where multiple people are doing the exact same work and are coding it different ways due to ambiguity in your system. Once again, there will need to be extensive staff training to eliminate any lack of understanding.
Audit the data regularly
Once you reach the point where you can trust the data you have, make sure you have an audit process in place to keep things clean. The resulting value will include the ability to calculate true cost of service ratios, provide your leadership with valuable capital planning information, defend your alternative equipment maintenance (AEM) program, be able to do effective benchmarking, and overall, increase the value you bring to your organization.
About the author: Alan Gresch is vice president, customer success, healthcare at Accruent, the world's leading provider of physical resource management solutions. Accruent serves over half of all U.S. hospitals, and its healthcare clients include Tenet Healthcare and Mayo Clinic. Prior to joining Accruent, Alan was vice president of capital technology management at Alpha Source, and held similar roles at TriMedx Healthcare Technology Management and Alexian Brothers Health System. Alan is a member, and past chair, of AAMI's Healthcare Technology Leadership Committee, and has recently published the Healthcare Technology Management Manual for AAMI.