As Federal, State, and local public health agencies mobilize to respond to the COVID-19 crisis, they are looking to sift through the noise of data available to them to determine how best to plan for the acute management of infected individuals. While data infrastructures across the country handle a variety of different types of data that can potentially be helpful, the single piece of data that is most useful by far is the result of approved COVID-19 testing – the lab result. No other piece of information is as timely, definitive, or specific (from a data quality and completeness standpoint) as a simple laboratory test result. Lab test results are the place where documentation of a positive COVID-19 case will originate in a definitive way – they contain demographic and geolocated information about the individual being tested, and also where that person is seeking treatment (in the form of the ordering provider for the lab test).
Unfortunately, this fact is a major problem for a coordinated, data-driven response to COVID-19 in the United States, given significant gaps in the sharing of lab results between key players in the US healthcare system. The implementation of electronic lab data interfaces by public health departments and key data aggregators such as Health Information Organizations (HIOs) is uneven across the national landscape. With the exception of the major reference labs like Quest Diagnostics and Labcorp, the majority of lab results delivery is still done on paper, via fax, a fact that is especially true for public health labs – which are performing the majority of COVID-19 tests to date. Related gaps and bottlenecks in the routing of lab data make it challenging for public health agencies and health care providers to gain access to accurate information on who has COVID-19 and who does not.
Diagnosis codes (generally in the form of ICD-10 codes), entered by health care providers into their electronic health records when they make a diagnosis, would seem to present an alternative to lab results for monitoring COVID-19 cases, due to the more universal health care focus on electronic sharing of diagnoses. However, there are three primary obstacles to this approach. First, diagnosis codes currently in use for COVID-19 range from temporary CDC guidance to use a combination of existing ICD-10 codes, to a special World Health Organization ICD-10 code that is being adopted on a region-by-region basis, since it has not gone through the official balloting process for the US healthcare system. Thankfully, an emergency update to ICD-10 with a single diagnosis code for COVID-19 was recently announced for April 1 by the CDC that should help to alleviate this problem.[1] Second, a diagnosis depends on a health care provider actually encountering an individual patient and entering a code (or a combination of codes) – diagnoses rarely if ever are automatically triggered in EHR systems based on the status of a COVID-19 test result. This means that even when they diagnoses are specific enough to be used for syndromic surveillance or other types of monitoring, they are often not timely and skew toward availability for patients that have been hospitalized. Third, many public health departments don’t have ready access to clinical diagnoses anyway, and they often don’t fully trust the data given the issues above.
In contrast, they do trust lab data, as it has been largely consistent for COVID-19 testing across the country and stands to be reinforced as the large reference labs come online and implement their LOINC code-sets (which look to conform to what is already in wide use).[2] Unfortunately, lab data is not as cleanly routed through the health data ecosystem as diagnoses are. This is partially due to the fact that electronic interfaces to route these data have generally not been prioritized by States and HIOs, but also due to the fact that key parts of the laboratory ecosystem remain paper-based. Most COVID-19 tests have so far been processed by public health labs, which exist either at the State or local level in most cases – although some States are much more complex and may have many more public health labs than others (New York has two, California had 32[3] listed as of 2018…all at the local or regional level). The vast majority of these labs remain paper-based, and report lab results back to the ordering provider via fax (who then keys the result into the patient’s electronic health record). In some cases, as in both California and New York, the public health labs even report test results on paper or via a secondary data entry method[4] to the public health departments responsible for tracking and managing disease outbreaks like COVID-19. This situation is partially a result of our historical underfunding of public health systems, who have not prioritized electronic laboratory results delivery.
Typical “Flow” of Laboratory Results Data, Public Health Labs
It is important to understand that in both the commercial laboratory world and also the public health laboratory world, test results are mainly intended to be delivered to the ordering provider – the individual (or facility) that actually ordered the test. Reporting to public health departments for syndromic surveillance sometimes falls on the labs themselves, and sometimes on the ordering provider – something they are supposed to do once they get back the lab result. In some cases, lab results can be sent to a “CC’ed” entity (maybe a public health lab, maybe an HIO) – but this rarely happens outside of the commercial reference labs, since public health labs often lack the resources to send out multiple copies of an individual lab result (especially if they are pushing results out via fax). So in order for public health lab test results to get to State and local public health agencies as well as to aggregators like HIOs, the data is often circuitously routed to ordering providers via fax, who then need to key the information into their EHRs for it to get picked up and reported out to other systems.
Reference labs like Quest Diagnostics and LabCorp, on the other hand, often send their results to the ordering provider electronically, and are in some geographies CC’ing HIOs on lab results. While these and other reference labs are still working on implementation of COVID-19 testing, their entry into the space will vastly increase the degree to which electronic lab results for COVID-19 testing are being delivered to both providers and HIOs, in places where an electronic results delivery connection already exists.
Typical “Flow” of Laboratory Results data, Reference Labs
The key gap, even with the reference labs, is that HIOs and public health departments are not always on the CC “List” for lab results originating from the reference labs – impeding their ability to utilize this extremely valuable data resource to manage the COVID-19 crisis. In cases where they are, such as in Indiana[5] and New York, HIOs are playing a significant public health role as a de-facto repository of COVID-19 test results, or in the Indiana case, as a centralized alerting service that triggers when a new positive result comes in. But in places where they are not (e.g. many parts of California) they may only be doing so on a comparatively small scale.
While we cannot go back in time and fix this issue, we can work to remedy it in the present. HIOs, States, labs and public health agencies should all be building and enhancing current lab data interfaces now. Focusing on less high-fidelity data like diagnosis codes or admit, discharge, transfer data just because it is currently available is not a substitute for getting access to the gold standard for COVID-19 cases. Models like Indiana’s already exist for integrating this information more broadly, and should be replicated wherever possible.
[1] Centers for Disease Control and Prevention, March 2020: https://www.cdc.gov/nchs/data/icd/Announcement-New-ICD-code-for-coronavirus-3-18-2020.pdf
[2] LOINC Codes: 31208-2, 75325-1, 94309-2
[3] California Association of Public Health Laboratory Directors, May 2018: http://websites.networksolutions.com/share/scrapbook/74/745076/Laboratory_Listing_Report.pdf
[4] These systems are often a basic web portal that require some degree of demographic information to be entered along with a positive test result LOINC code. They do not transmit a copy of the lab result in any kind of structured, encoded data format and are often very light on the level of information that is required in order to actually submit a case.
[5] Raths, David “Some HIEs Stepping Up to Play Key Role in COVID-19 Response,” March 2020: https://www.hcinnovationgroup.com/interoperability-hie/health-information-exchange-hie/article/21130588/some-hies-stepping-up-to-play-key-role-in-covid19-response