Data Management for Large-scale COVID-19 Immunization: This is all not as simple as it seems

Noam H. Arzt, Ph.D.There is a global race for the development of a vaccine for the SARS-CoV-2 virus that causes COVID-19. Finding a vaccine that works and receives approval is only part of the process. There are a series of other steps that need to be taken so that the vaccine can be delivered. These include the mass production of the vaccine, shipment, administration and record-keeping. This may be even more complex as there may be several vaccines.

In this article we review some of these issues with a particular focus on the United States.

One historical example is the polio epidemic in the early 1950s. A large-scale trial of the initial Salk vaccine involved over 1.3 million children, and when concluded it took more than a year to analyze the results (see the History of Vaccines timeline for more details).

The first vaccine for influenza was developed in the mid-1930’s, though it was not until nearly ten years later that it was approved first for military and then civilian use. Today annual flu shots are delivered at some scale annually, largely through pharmacies with no doctor’s order required and as a single dose. But flu vaccination typically includes only about half of adults in the US.

There have been many changes and much uncertainty regarding COVID-19 testing, and according to media reports, CDC is preparing states and other jurisdictions to immunize for COVID-19 as early as the beginning of November should a viable vaccine be available. The Centers for Medicare & Medicaid Services (CMS) has just issued an interim final rule that establishes new requirements for long-term care (LTC) facilities, hospitals, and for clinical laboratories to report certain COVID-19-related testing data.

Approval for a vaccine may take some time. One COVID-19 vaccine trial was halted due to a reported adverse event. Vaccine manufacturers are concerned about being rushed into requesting approval from the Food and Drug Administration (FDA) for their vaccines and a number of them have signed a pledge to withhold their request for approval unless a rigorous clinical trial process has been followed. Some states are reviewing their legislation to determine what steps may be needed in administering a vaccine and sharing data under an emergency use authorization (EUA) rather than through routine channels/timelines.

Data Management Challenges

The clinical challenges to the large-scale administration of one or more new COVID-19 vaccines will be difficult. And they will be particularly difficult under a rushed timetable. CDC has established requirements for near real-time information about the progress of the immunization campaign (data submission within twenty-four hours of administration of a vaccine dose) and the inventory level of vaccines. Here I discuss some of the more subtle data management realities that result from these plans.

This is all not as simple as it seems. Legal authority for public health resides with state and local public health agencies — there is no federal public health law per se. This is fundamental to how government in the US is organized through the Constitution. The Centers for Disease Control and Prevention (CDC) wields influence over state and local public health agencies through its expertise, and especially its funding, since almost all public health activities are federally funded. With that funding often come requirements from CDC which are conditions for accepting the funds. State and local law, usually related to patient consent for information disclosure to third parties, almost always supersedes even these CDC funding requirements.

In terms of COVID-19 pandemic preparedness and response, consider:

  • There are 64 immunization information systems, or IIS (registries), in the US today covering 50 states, US territories, and large cities. They are locally controlled, though they largely conform to common standards. As already mentioned, the CDC funds most of their acquisition, operations, and upkeep.
  • Data sharing laws differ widely from jurisdiction to jurisdiction. Some view this as an obstacle to interoperability; some view this as a necessary protection of patients'/citizens' right to privacy.
  • The administration is pushing hard for "real-time," patient-identified data about immunization for COVID-19 once it starts up. Jurisdictions have not historically delivered much identified data to CDC and some jurisdictional law/policy prevents it.
  • There are currently three major immunization information system software platforms in use across the country: Envision’s WebIZ, STC’s STC | One, and a product called WIR developed by the State of Wisconsin and primarily supported by DXC. To make matters more complicated, CDC is funding the rewrite of the WIR software over the next few years.
  • More recently Deloitte Consulting, a major federal contractor, was awarded a contract to develop a new software platform for the CDC called the Vaccine Administration Management System (VAMS) to support mass vaccination campaigns. The platform is still under development and has yet to be rolled out. CDC claims that it embarked on this rapid, $16+ million project based on an assessment that existing systems could not perform the needed tasks, though no such clear evidence was ever circulated.
  • The initial vaccination campaign will most likely deploy the first wave of immunizations to hospital workers, first responders, and other "critical" workers. These groups will be defined by the Advisory Committee on Immunization Practices (ACIP) which historically recommends clinical guidelines related to immunization to CDC for the US.
  • Jurisdictions have other options, including functionality built into (or being added to) their existing IIS and third party products like Vaxigo, PrepMod, and EMTrack. Many jurisdictions have not yet decided on their approach, but await further guidance and decisions to finalize their strategy. It is uncertain whether the new Deloitte-built Vaccine Administration Management System will be ready on time or will perform as planned.
  • Physical vaccine distribution on a scale that will be required for a substantial part of the US population will be a whole challenge in and of itself. CDC supports existing mechanisms for jurisdictional public health agencies to order, receive, distribute, and account for the use of vaccines, especially those that are Federally-supplied to certain populations under the Vaccines for Children (VFC) program.
  • Inventory management capabilities that are needed to implement these distribution systems are increasingly becoming part of core IIS functionality that jurisdictions can rely upon.
  • Though historically vaccine ordering has not been constrained by issues of supply (other than perhaps annual flu vaccine), there have certainly been occasions where "orders" could not be fulfilled because vaccines were not available. Certainly, we expect that when the COVID-19 vaccine is initially available its supply will be constrained. It is unclear how existing ordering processes might be affected, and how vaccine allocations are to be made and by whom.

And all of this is being managed by public health agencies whose workers are just as affected as everyone else. With similar challenges of working from home or kids around, public health agency staff are trying to keep routine immunizations from continuing to plummet and further expose the population to other infectious diseases, while trying to simultaneously manage what will likely be an active flu season. Public Health is facing what has become known as the “triple threat” of promoting routine immunizations, a hard hitting influenza season and imminent COVID-19 vaccine. Adding to this fight is an anti-vaccination movement that has now focused its general concerns about vaccines onto the rush to vaccinate for COVID-19.

CZU lightning complex fire burns along Butano Ridge, in the Santa Cruz Mountains, California. CC BY-SA 4.0On top of this the Western part of the United States is facing major disasters. As of the time of writing this article, there are major wildfires in eleven Western states with a total of more than five million acres of forests in fires. The combined area of wildfires across these states is larger than the state of New Jersey. More than half-a-million people in Oregon are under evacuation orders. That is 12% of the entire population of the state. Many more have been evacuated or displaced in California. Dozens of hospitals, medical facilities, and retirement homes have been evacuated in California. More than a million California residents have been affected by rolling blackouts. And people living along the East Coast and the Gulf of Mexico, are facing a severe hurricane season. On top of this, while hospital chains are making historic profits, small medical practices, where most individuals would normally be vaccinated, are going out of business at a rapid pace. This is a chaotic situation.

A More Sensible Solution

But the important thing is that many of the pieces are in fact present for successful management of mass vaccination if the government relies on its existing infrastructure and shores it up where it needs help.

For immunization data management, that means:

  • The federal government should rely on existing immunization information systems (IIS) as much as possible to be the initial collection point for immunization data. This is the bread and butter for public health systems across the country. They have been preparing for this moment for decades. With most of the IIS based on just three platforms, CDC can rely on existing vendor relationships and existing jurisdiction technical support contracts to make sure the implementation of these products is consistent with COVID-19 data requirements. The remainder of IIS are either migrating to one of these three products or are technically sound as they are; proper instructions will allow them to adapt to new requirements as well as the other three platforms. It is likely, however, that some IIS may not be prepared to fully support COVID-19 immunization data requirements in time for the start of the campaign.
  • As state and local jurisdictions have the primary authority to maintain the health and well-being of their populations, that should include providing for their immunization and tracking the results. Federal policy and strategy should work within this reality.
  • The Federal government should also resist the temptation to implement new requirements and infrastructure to collect new information in new and different ways. The process of collecting COVID-19 testing data continues to be problematic. Let’s not have the process for collecting COVID-19 immunization data follow suit. There will be so many people to immunize - even in the initial high priority groups — that immediate access to their data should not come at the expense of accurate data.
  • The US should leverage existing mechanisms, contracts, and processes for vaccine distribution and accountability which leverage public health agencies and relationships. Once again, IIS are a key leverage point for this activity and have been honing their capabilities for many decades.
  • Public health agencies should be allowed, and even encouraged, to rely on existing, well-developed plans for mass vaccination and pandemic response, including defined data management strategies. Of course, public health cannot mount a successful response by itself; numerous other agencies are typically involved and jurisdictions organize themselves differently to respond.

Our recommendations here are based on extensive work with public health systems and nearly two decades of experience in emergency preparedness and response. HLN has been active in health informatics for more than twenty years, and the author of this article has been involved in the design, development, and support of immunization information systems since their inception more than twenty-five years ago (see this article for a history of IIS technology written by this author).

HLN works closely with its public health clients and partners to support them in rising to address and solve major challenges. In cooperation with multiple public health offices across the country we have developed an open source immunization decision support platform with extensive capabilities. Called Immunization Calculation Engine (ICE), this open source evaluation and forecasting software will be modified to support COVID-19 vaccines and their related clinical decision support rules (see related article).

Since ICE is open source individual public health facilities can modify the software to meet their needs. ICE evaluates a patient’s immunization history and generates the appropriate immunization recommendations for the patient. ICE can be deployed in diverse technical environments and easily integrates with third-party clinical systems.

Key differentiators for ICE include:

  • No license fee to use this open source software
  • Supports all routinely administered vaccines for children and adults
  • Compliant with the Advisory Committee on Immunization Practices (ACIP) clinical guidelines
  • No vendor lock-in through full access to source code
  • It is based on open standards and it is interoperable with other health IT systems
  • Maintained through an open and collaborative process
  • Flexible deployment options including cloud hosting
  • Scalable for high performance

ICE will be updated as soon as the new rules are defined by the Advisory Committee on Immunization Practices (ACIP) which establishes clinical guidelines for immunization on behalf of the CDC.

NT Cheung - CMIO Hong Kong Hospital Authority. Credit HASome Key Lessons Learned

There are several critical lessons that need to be learned. First is the need for a strong and functional public health system. And second such a system needs to run on open source software. There are examples from countries around the world that have done the best in their response to the coronavirus pandemic. As I detailed in this article, I attended the recent Office of National Coordinator for Health IT (ONC) Tech Forum. As I wrote in the article, the highlight of the meeting was the presentation by NT Cheung, MD, Chief Medical Informatics Officer of Hong Kong’s Hospital Authority. Dr. Cheung detailed how Hong Kong has built an open source digital health platform over the past 20 years that was critical in their success in addressing the coronavirus outbreak.

Hong Kong is not the only city focused on open source and open standards. Many countries around the world that successfully responded to the outbreak have public health platforms built on open source. These include Singapore, Taiwan and Thailand in Asia. Same in Europe, where Norway, Finland, Estonia, Slovenia and other countries leveraged OpenEHR as a critical component of their public health system (see here).

The potential for open source solutions in the US is similar. Several public health jurisdictions in the US have collaborated in the development of ICE. For example New York City collaborated with HLN in the development of ICE and contributed all their code as open source so that other public health jurisdictions across the country could use the code for their own solutions. Since then, a number of other jurisdictions, including Michigan, New Jersey, Rhode Island, Vermont, and the Indian Health Service have all integrated ICE into their systems.

And the countries leveraging open source solutions for their public health systems, also provide the lesson that investments in public health infrastructure are essential. The chronic underfunding of public health systems across the United States has led at least in part to our seeming inability to respond reliably even when we know what we have to do. The current influx of funding from Congress through CDC will certainly help today, but when the urgency passes we need to prepare for the next likely event. More consistent investment in systems, training, standards development, and greater reliance on open source software and its benefits, can ensure a more reliable and affordable response in the future to pandemic disease outbreaks as well as natural and man-made disasters.