In a small real-world US cohort, two doses of the MVA–BN vaccine were associated with eighty-nine percent effectiveness against mpox. But the same study also shows why that impressive number needs careful interpretation. Mpox disease, previously known as monkeypox, is caused by the zoonotic human monkeypox virus and is akin to the smallpox virus.

Historically, clade II MPXV was endemic to West Africa and self-limited in human-to-human transmission. Previously reported mpox cases in the United States were associated with international travel to endemic regions. However, in 2022, the United States experienced thirty thousand three hundred forty-four cases and forty-two deaths, approximately one-third of the total 2022 outbreak global cases.

The majority of United States cases occurred in persons who identify as men or transgendered persons who have sex with men, and approximately forty percent of these MSM had concomitant human immunodeficiency virus infection, putting them at increased risk of severe disease.

Although the multi-country outbreak was no longer a Public Health Emergency of International Concern in May 2023, more than ninety-one thousand cases since 2022, infections among the unvaccinated, and a 2024 resurgence in the Democratic Republic of the Congo demonstrate a continued need for preventive vaccination efforts.

The Modified Vaccinia Ankara-Bavarian Nordic vaccine is a third-generation smallpox vaccine approved by the United States Food and Drug Administration in 2019 for subcutaneous administration to prevent smallpox and mpox disease. A full vaccination regimen includes two separate doses administered four weeks, or twenty-eight days, apart.

The vaccination is considered effective fourteen days after the second vaccine dose, though evidence suggests some protection after the first dose. MVA–BN may also be used as a single-dose booster in individuals previously vaccinated against smallpox, although there is inadequate data to determine the appropriate timing of booster doses.

This retrospective, observational study uses administrative United States healthcare data to evaluate MVA–BN effectiveness and safety among vaccinated MSM and transgender women. USMVAc is also part of a larger research program that includes two German prospective and retrospective studies, SEMVAc and TEMVAc, to support decision making in the European Union about the MVA–BN benefit and risk profile.

This retrospective, observational comparative cohort study used adjudicated medical and prescription claims, laboratory data, and beneficiary demographic information from a de-identified administrative United States healthcare dataset curated by HealthVerity.

The dataset covered subjects with available vaccination and mpox diagnosis information between April first, 2021, and December thirty-first, 2022, within a healthcare ecosystem that includes more than two hundred million individuals from all United States states. HealthVerity applied initial inclusion criteria to create a therapeutic-area-specific subset of people considered most at risk of mpox in the United States.

The MSM and transgender women population was identified using a proxy definition grounded in previous and ongoing mpox research. Subjects had to be male according to sex assigned at birth, at least eighteen years of age, and fulfill at least one criterion: a diagnosis code related to high-risk sexual behaviour, a diagnosis code indicating HIV infection, or a drug or procedure code indicating HIV pre-exposure prophylaxis among those without a substance use disorder diagnosis during the study period.

The study population was selected between August first, 2022, the MVA–BN United States emergency use authorization date, and September thirtieth, 2022, to capture the peak of the outbreak and allow a minimum of three months of follow-up. For the primary vaccine effectiveness assessment, subjects entered the exposure group if they had a second record of MVA–BN vaccine.

Because there was no active comparator, coarsened exact matching was used for each vaccinated subject to select five similar unvaccinated subjects and assign them a matched index date. Comparator subjects had to match an exposed subject on calendar date, age group, United States region, and insurance type, and they could not have a record of MVA–BN vaccine administration before or on the matched index date.

The primary outcome, mpox disease, was defined by a positive orthopoxvirus PCR laboratory test indicating mpox or by an mpox diagnosis using ICD-ten-CM code B04 in an inpatient or outpatient medical claim, or a PCR result of detected. Follow-up began on the index date, fourteen days after the second MVA–BN dose, and ended at the earliest occurrence of mpox, disenrollment, death, or the end of data on December thirty-first, 2022.

Comparator unvaccinated subjects were also censored if they later received an MVA–BN vaccine dose after the matched index date. The secondary safety outcomes were adverse events of special interest: myocarditis, pericarditis, encephalitis, and anaphylaxis. These events were defined by ICD-ten-CM codes and selected based on clinical relevance.

Each safety outcome required a specified risk window: one day for anaphylaxis, fourteen days for myocarditis and pericarditis, and twenty-eight days for encephalitis in the primary analysis. In sensitivity analyses, the risk windows were extended to three days for anaphylaxis, twenty-eight days for myocarditis and pericarditis, and forty-two days for encephalitis.

Figure One traces cohort attrition from an at-risk MSM population of sixty-eight thousand three hundred eighty-one through inclusion and exclusion criteria. After removing subjects who did not meet vaccination timing, enrollment, insurance, or mpox-history requirements, the cohort retained one hundred sixty-three fully vaccinated patients.

Chronological matching without replacement then produced eight hundred fifteen matched unvaccinated patients, establishing the analytic groups used to assess vaccine effectiveness. After inclusion and exclusion criteria and coarsened exact matching, the primary cohort included one hundred sixty-three fully vaccinated subjects who received two doses at least twenty-eight days apart, matched to eight hundred fifteen unvaccinated subjects.

Only six point one percent of subjects were aged eighteen to twenty-five years. The remaining subjects were split across ages twenty-six to thirty-five, thirty-six to forty-five, and at least forty-five years. The majority of the study population was commercially insured and largely represented the Western United States region.

Baseline evidence of a sexually transmitted infection was low but similar between groups, at approximately five percent. After standardised mortality ratio weighting, the one hundred sixty-three fully vaccinated subjects were compared with a weighted pseudo-population of one hundred fifty-nine point eighty-five, with an effective sample size of three hundred three.

All propensity-score covariates were balanced, with absolute standardised differences no greater than zero point ten. Table one compares baseline characteristics for 163 fully vaccinated participants and 815 unvaccinated comparators before standardized marginal ratio weighting, then reports the weighted comparator distribution.

Before weighting, age, region, and insurance were closely aligned, while race and ethnicity, HIV status, PrEP use, and missing race information showed larger absolute standardized differences. After weighting, the reported differences are smaller across these characteristics, supporting a more comparable basis for evaluating vaccine effectiveness.

Compared with unvaccinated subjects, fully vaccinated subjects had a decreased risk of mpox disease after propensity-score adjustment. The resulting vaccine effectiveness was eighty-nine percent, with a ninety-five percent confidence interval from twelve percent to ninety-nine percent.

In the subgroup of people living with HIV under antiretroviral treatment, the results suggested similar vaccine effectiveness after propensity-score adjustment, but the confidence interval was wide and included zero percent. Because the vaccinated and unvaccinated sample sizes were small in subgroups without HIV, there were no reportable differences in vaccine effectiveness between PrEP users and non-users.

Table two reports vaccine effectiveness against mpox among adult men who have sex with men and transgender women, comparing fully vaccinated participants with matched unvaccinated comparators. In the overall cohort, adjusted vaccine effectiveness was eighty-nine percent, with a ninety-five percent confidence interval from twelve percent to ninety-nine percent.

Subgroup estimates varied by HIV, antiretroviral, and PrEP status, while some small subgroups had zero events or results marked not reportable, underscoring the uncertainty of those estimates. In sensitivity analyses, vaccine effectiveness was attenuated to sixty-four percent among subjects with any dose, with a ninety-five percent confidence interval from forty percent to seventy-eight percent.

Among subjects with only a single dose, vaccine effectiveness was seventy percent, with a ninety-five percent confidence interval from forty-four percent to eighty-four percent. There was no evidence of vaccine effectiveness for single-dose vaccination among subjects older than fifty years who were assumed to have had a prior smallpox vaccine.

Table three tests how sensitive the vaccine-effectiveness estimate is to different definitions of vaccination and prior smallpox vaccination. Adjusted effectiveness was sixty-four percent, with a ninety-five percent confidence interval from forty percent to seventy-eight percent, for at least one MVA–BN dose, and seventy percent, with a confidence interval from forty-four percent to eighty-four percent, for a single dose only.

Among people older than fifty years assumed to have prior smallpox vaccination, the adjusted estimate was zero percent, with a confidence interval from minus two hundred forty-two percent to seventy-one percent, showing no evidence of effectiveness in that analysis. The initial safety cohort included nine hundred forty-seven subjects receiving any dose, matched by coarsened exact matching to four thousand seven hundred thirty-five unvaccinated subjects.

A one-to-one propensity-score matching process selected a final model that included all covariates except history of myocarditis and history of encephalitis. Those two histories were removed because no patients in either the exposed or comparator group had them, violating positivity.

All covariates included in the propensity-score model were effectively balanced, and fewer than five subjects in either group had a history of adverse events of special interest during baseline. There were no adverse events of special interest among subjects who received at least one MVA–BN dose or their matched comparator subjects within the specified risk windows for each safety outcome.

When the pericarditis risk window was expanded to twenty-eight days, one pericarditis event was identified among vaccinated subjects after propensity-score adjustment, with a risk of one hundred nine point twenty-nine per one hundred thousand persons. Table five compares vaccine AESIs in a propensity-score-matched cohort of nine hundred fifteen vaccinated subjects and nine hundred fifteen comparator unvaccinated subjects.

For anaphylaxis and myocarditis, the table reports zero events in both groups across the primary and sensitivity risk windows. Pericarditis also had zero events in the primary fourteen-day window; with the twenty-eight-day sensitivity window, one vaccinated subject had an event, corresponding to a risk per one hundred thousand persons of one hundred nine point two nine, with a ninety-five percent confidence interval from fifteen point three eight to seven hundred seventy-six point six nine.

Using administrative United States healthcare data, the primary analysis observed eighty-nine percent vaccine effectiveness among a small cohort receiving two doses, compared with a matched sample of unvaccinated subjects considered to have a similar risk. The interpretation may be limited by the small sample size, but the result is consistent with existing United States studies reporting vaccine effectiveness of sixty-six percent, eighty-six percent, and seventy-six percent in different study designs and populations.

Systematic reviews and a meta-analysis of real-world evidence reported vaccine effectiveness ranging from sixty-six percent to ninety percent, and recent CDC reports show that breakthrough mpox infections after two MVA–BN doses are rare. HealthVerity claims originate for insurance billing purposes, creating a financial incentive to confirm accuracy, but misclassification remains inherent in claims-based data because of provider coding practices or coding errors.

HealthVerity does not have information on sexual orientation and gender identity, clinical notes, or other documentation to validate the MSM and transgender women algorithm used to select the cohort. Misclassification may therefore be more likely among unvaccinated subjects.

If some unvaccinated men were not MSM or transgender women, their mpox risk could be underreported, thereby overestimating vaccine effectiveness. The proxy definition was based on previous mpox studies and selected subjects presumed to be MSM or transgendered women using high-risk criteria, while German prospective and retrospective studies with self-reported gender identity and sexual behavior are intended to complement this work.

The study found high estimated effectiveness after two MVA–BN doses and no adverse events of special interest in the primary safety windows, while small samples and possible cohort misclassification limit certainty.