A Project US@ Deep Dive: Part One

By Rachel Podczervinski, MS, RHI, Vice President, Professional Services, Harris Data Integrity Solutions

I have been fortunate over the last year to participate in a significantly important project that is advancing efforts toward ensuring the future of quality patient matching efforts spearheaded through the Project US@, established by the Office of the National Coordinator for Health Information Technology (ONC). I’ve written previously about our involvement with the project, but with release of the Final Version 1.0 of the Project US@ Technical Specification and ONC-AHIMA Companion Guide on Jan. 7, 2022, it’s worth taking a deeper dive into each, given the level of detail included.

The first part of this two-blog series examines the Companion Guide, developed by AHIMA and ONC. An important aspect of the Project US@ collaboration, the goal of which was to issue a unified, cross-standards specification for patient addresses that will improve patient matching throughout the healthcare ecosystem.

The Companion Guide describes how address numbers, street names and other elements of a patient address should be formatted and abbreviated in a patient’s EHR and describes what special characters can be used. It also contains operational guidance and best practices related to accurate and timely capture and management of patient addresses that support conformance to the Project US@ Technical Specification and improved patient matching.

The guide helps technical experts in healthcare prepare for data capture, although work remains to be done to unify all technology vendors behind the standards. Even so, this moment means we can better navigate outlier cases for address types outside the regular home address, with helpful examples of proper formatting.

The problem defined
Patient matching – the ability to identify and link records for the same patient within and across systems and organizations – is a significant problem for health organizations. Standardized patient matching protocols support health information exchange and interoperability.

These data linkages are more manageable when the patient has interacted with a single health group or system. However, those with information residing in different health organizations face more demanding challenges for coordinating their data. In these instances, patient data can be fragmented or duplicated, contributing to inefficiencies in care coordination.   

Before ONC’s Project US@, there were no data standards for this type of disparate patient data. Our goal was to begin to rectify this problem by establishing a structure for collection of patient address data.

Standard address collection improves patient matching
Project US@ focuses specifically on addressing lack of standards in the collection of patient addresses, which ONC notes “has long been viewed as a critical component of nationwide interoperability and the nation’s health IT infrastructure.”

According to Steve Posnack, deputy national coordinator for health IT, standardizing the addresses that are collected and used to match patient records across organizations, systems, and applications, “…might seem like a small thing, but that’s precisely why this work was important. Improving the accuracy and consistency of addresses will have a big impact if implemented at scale.”

Essentially, Project US@ encourages the adoption and alignment of patient address collection consistently across the industry. The specifications offer instruction for United States domestic and military patient addresses and for international addresses of patients who frequently cross boarders to access care, primarily from Canada and Latin America. The Companion Guide shows standardized ways to collect patient addresses and provides examples.

It's worth noting that the proposed standards are not mandates; Project US@ does not obligate healthcare systems and others in the ecosystem to modify or update existing data.

Health IT Audience
The technical specifications outlined in the guide are primarily designed to support health IT developers and implementers who will be implementing standards and technologies to improve data standardization and improve patient matching. However, several other stakeholders may also benefit by adopting the specification, including providers and health information professionals.

“This specification is not a database design document,” note Companion Guide authors. “We encourage health IT developers to tailor patient registration, scheduling, and other health IT applications to conform to the specification and support health information professionals to follow the guidelines and best practices outlined in the AHIMA Companion Guide. We also recommend systems that exchange patient demographic data with other systems to standardize patient address information according to the specification before exchange and matching to limit information loss.”

The specification applies to current and historical patient addresses. As such, there may be no limit to the number of addresses systems could maintain. For those that can leverage these data, historical addresses may be valuable for patient matching.

Data display
If parts of a patient’s address are unknown, the guide recommends leaving those fields blank. This helps avoid misclassification, which can happen when patient matching algorithms attempt to match on the unknown value

 Per the guide, if patient address data are captured and stored in a single string field, where elements such as street address and city are not in separate fields for patient matching, the data should be parsed uniformly according to the following format: 

• Business/Firm

• Name

• Street Address

• Primary Address Number

• Street Name

• Suffix

• Secondary Address

• City

• State

• Zip+4

Diacritics and special characters
Finally, and quite importantly, there are more than 1,300 individual languages and language groups that can impact patient matching efforts. The most common include Albanian, Catalan, Croatian, Czech, Dutch, French, Hungarian, Icelandic, Polish, Portuguese, Serbian, Slovak, Slovene, and Spanish. Other languages represented include Catalan, Dutch, French, Italian, Estonian, and Portuguese.

Some, though not all health IT systems, can capture diacritics. The ability to meaningfully exchange them relies on several factors, including the capacity of the receiving system to read and accurately match records containing diacritics.

For example, diacritic marks that do not successfully convert to code will often display as an inverted question mark. If patient matching algorithms are not designed to identify and disregard these or any other unrecognizable character, additional matching errors may occur.

Closing thoughts
Ultimately, the best way to accurately match patient records to the patient is through technology, processes, people, and standards. The ONC-AHIMA Companion Guide came into being because of a recognized need for actionable standards to guide capture of critical data.

Accurate and timely capture and management of patient addresses that support conformance for improved patient matching is part of this new standard protocol. Time will tell how pervasive its use becomes and how well Project US@ facilitates patient matching for all our benefit.