Did Vaccination Campaigns Help Mitigate the Recent Hepatitis A Outbreak in Michigan?

By Lina Sorg

Hepatitis A is a contagious viral infection of the liver. Common symptoms include jaundice, stomach pain, and fatigue, and clinical illness usually lasts about two months. From 2016 to 2018, the state of Michigan experienced a severe outbreak of hepatitis A. The epidemic began in Southeast Michigan—specifically Detroit and the surrounding Wayne, Oakland, and Macomb counties—before spreading via transportation corridors to other areas of the state. Cases first appeared in 2016, with the bulk of infections occurring in the latter half of 2017 before petering off in 2018 and 2019 (see Figure 1).

Most previous outbreaks of hepatitis A were largely foodborne and did not lead to secondary infections. However, the Michigan cases were driven by person-to-person contact among the following high-risk groups:

People who use drugs (both injection and non-injection)
People experiencing homelessness or unstable/transient housing situations
People who are or were recently incarcerated or institutionalized
People with chronic liver disease
Men who have sex with men.

Roughly 50 percent of cases in Michigan were associated with substance use.

Figure 1. Confirmed hepatitis A case onset by week for the Michigan outbreak of cases from August 1, 2016 to December 18, 2019.

As soon as an outbreak of any disease begins, public health officials typically respond with vaccinations. The hepatitis A vaccine was introduced in the U.S. in 1996 and included in pediatric vaccination cycles shortly thereafter. As a result, cases steeply dropped off until the recent Michigan epidemic. In 2017 and 2018, vaccinations increased in the entire state of Michigan — particularly at public clinics. “There was a large-scale response in terms of vaccination, and also a public hygiene campaign,” Andrew Brouwer of the University of Michigan said.

During a minisymposium presentation at the 2021 SIAM Conference on Applications of Dynamical Systems, which took place virtually last week, Brouwer used mathematical modeling to connect epidemiological theory to vaccination data for Michigan’s hepatitis A outbreak. More specifically, he investigated the impact of vaccination efforts on spatiotemporal patterns of spread by examining person-to-person transmission and disease progression (see Figure 2). This type of information allows scientists to supply financial stakeholders with concrete evidence that confirms the importance and efficacy of public health endeavors like vaccine clinics.

Brouwer accounted solely for the “at-risk” population in his model. “What’s tricky about modeling the at-risk population is that we don’t really know who’s at risk or how big that population is,” he said. “It’s just this abstract notion of people who could have gotten hepatitis A had they been in the wrong place at the wrong time.” The relative level of risk for the various at-risk groups is also unknown, which serves as one of the largest sources of uncertainty in the analysis.

Although researchers cannot truly know the exact number of at-risk individuals, the outbreak’s shape does offer some clues. For instance, the basic reproduction number \((R_0)\) represents the expected number of people that a single infectious person is likely to infect near the start of an outbreak. Previous pre-vaccination epidemics indicate that \(R_0\) likely falls between 1.1 and 1.6. “If we look at how \(R_0\) is a function of that risk population, and then look at the reasonable ranges of this reproduction number, we can get a sense that the at-risk population in the Detroit area was probably somewhere between 3,000 and 6,000 people,” Brouwer said. “For the rest of the state, we’re talking 150,000 to 300,000 people.”

Figure 2. The impact of vaccination efforts on spatiotemporal patterns of hepatitis A in Michigan.

He also made several additional assumptions:

The at-risk population in Southeast Michigan is distinct from the at-risk population in the rest of the state
Only a fraction of doses of the hepatitis A vaccine were given to people who were actually at risk
Only a fraction of people who are jaundiced/symptomatic seek medical care and are included in the total case count
Transmission may be seasonal, especially for people experiencing homelessness.

When building his model, Brouwer employed data from the Michigan Department of Health and Human Services; he specifically utilized case data from the Michigan Disease Surveillance System and vaccine dose information from the Michigan Care Improvement Registry. This data allowed him to estimate a few key model parameters:

Transmission rates within Southeast Michigan, outside of Southeast Michigan, and from Southeast Michigan to the rest of the state
Vaccine coverage parameters, which transform the number of doses in the data into vaccination rates
Reporting parameters, which connect reported cases to the modeled fraction of infected people.

Using these parameters, Brouwer approximated the epidemic trajectories that capture the average peaks, growth, and decay of the epidemic. “Our results suggest that the implementation of these vaccine campaigns probably did not make that much of a difference in the southeast, but likely made a large difference in the rest of the state,” Brouwer said. Roughly 44 percent of at-risk individuals actually got hepatitis A in Southeast Michigan, while only five percent of at-risk individuals outside of Southeast Michigan became ill (see Figure 3).

Figure 3. The cumulative incidence for at-risk individuals in Southeast Michigan was 44 percent, while the cumulative incidence for at-risk individuals outside of Southeast Michigan was only five percent.

Brouwer’s model indicates that while vaccine campaigns only averted about three percent of cases in the southeast, they prevented approximately 91 percent of potential cases in the rest of Michigan. This disparity occurred because a large portion of the epidemic in the Detroit area—the origin of the outbreak—transpired before vaccinations began, while infections and vaccinations were happening simultaneously outside of the southeast region. Brouwer estimated that the absence of vaccines would have extended the outbreak in the rest of the state by another three years. “This suggests that vaccination has the potential to have a larger impact,” he said.

Although vaccination efforts had relatively little impact in southeast Michigan because the epidemic was already well underway, the vaccine’s effectiveness and longevity mean that the campaign will help prevent future outbreaks in the area. This result also confirms the larger public health implication that reactive control strategies are always less constructive than proactive control strategies. “This is sort of a perennial problem in public health,” Brouwer said. “If public health is doing its job right, it’s almost invisible because this outbreak wouldn’t have happened if we had that proactive vaccine campaign.”

Finally, Brouwer stressed the importance of vaccination within high-risk groups, especially in urban centers that often serve as hubs for disease transmission. Of course, better-targeted vaccines to individuals who are most in need—when combined with proactive implementation—would make vaccine regimens even more effective.

Lina Sorg is the managing editor of SIAM News.