Clinical decision support systems (CDSSs) have been shown to increase adherence to clinical guidelines¹ and augment diagnostic and management behavior in several infectious syndromes.^2–4 CDSSs assist in diagnosis by allowing correct application of disease-specific criteria, serving as educational tools about unfamiliar syndromes, improving the appropriateness of laboratory testing,⁵ and assisting with the application of isolation precautions.⁶

Emerging infectious diseases place a high cognitive burden on frontline clinicians for several reasons: the clinical presentation is unfamiliar, testing algorithms may change rapidly, detailed epidemiologic history is crucial for identifying at-risk patients, and unfamiliar infection control protocols make applying isolation precautions challenging. The mpox outbreak of 2022–2023 exemplified all these conditions. Since the eradication of smallpox, few clinicians were familiar with poxvirus infections. At the beginning of the epidemic, testing was scarce. And, since infections were concentrated in gay, bisexual, and other men who have sex with men,⁷ it was critical that history taking focus on Centers for Disease Control and Prevention (CDC) epidemiologic criteria for a person under investigation (PUI). These questions, centering on a patient’s sexual health and behaviors, are not universally asked.⁸

To support the identification, isolation, and diagnosis of people presenting with a rash and possible mpox, the “Evaluate for Mpox” (EvalMpox) CDSS was incorporated into the electronic health record (EHR) of a large integrated healthcare system.

Based on experience developing CDSS for coronavirus disease 2019 (COVID-19),⁶ a team of infectious diseases, infection control, and information technology experts constructed EvalMpox. Toward quick dissemination at the beginning of the outbreak, we previously communicated a rapid report on the initial CDSS applied to the first 55 evaluated patients ending July 20, 2022.⁹ This manuscript analyzes the performance of an enhanced CDSS and all 668 encounters through April 12, 2023.

EvalMpox assists clinicians in identifying patients with mpox by guiding the collection of information regarding epidemiologic criteria for CDC PUI status in patients with a new, unexplained rash (Figure 1). Epidemiologic criteria were updated throughout the epidemic to conform to evolving CDC criteria. If a clinician inputs both clinical and epidemiologic criteria for mpox, EvalMpox classifies the patient as a PUI and recommends testing. If the patient does not meet clinical and epidemiologic criteria for mpox, EvalMpox recommends against testing unless clinical suspicion is high. EvalMpox then generates a risk assessment note in the EHR, coordinates the application of mpox-related infection statuses in the patient’s electronic chart, and orders appropriate isolation (Supplemental Figure 1 online).

Data on EvalMpox encounters exported from the EHR (Epic) included patient demographics, encounter date/time, practice location/setting, clinician-user role, and PUI/non-PUI status. Mpox testing results performed in our system were separately exported. Data were inspected for duplicate encounters, and charts were manually reviewed to ensure data integrity. Categorical data were analyzed with χ² testing and continuous data by t test. Negative predictive value (NPV) and positive predictive value (PPV) were calculated over the total analyzed period. Data were collected under MGB IRB protocol 2012P002359.

We describe the performance of EvalMpox, a novel CDSS for the identification, evaluation, and management of patients meeting CDC PUI criteria for mpox. There was widespread adoption of EvalMpox across our large, integrated healthcare system among diverse provider roles and in all care settings. The CDSS performed well; our patients classified as PUI had similar patient demographics compared with CDC mpox case demographics, and PUI were more likely to test positive for mpox than non-PUI. The NPV of EvalMpox was high.

There are several limitations to the conclusions that can be drawn from our report. First, this study was conducted in a single health system, potentially limiting generalizability. Second, though we performed extensive education prior to and during implementation, uptake was not universal. If patient characteristics influenced clinician decisions of whether to use EvalMpox, this utilization behavior may have biased the observed test characteristics of the CDSS. However, EvalMpox was used in most mpox testing encounters. Third, though we find that the NPV for EvalMpox was high, despite this being the largest worldwide mpox outbreak, low overall community prevalence certainly contributes to this result. Fourth, as in any CDSS that relies on provider data entry, errors in tool use can lead to inappropriate recommendations. A focused chart review of the three patients designated non-PUI by EvalMpox who tested positive by PCR revealed that one of those patients reported epidemiologic risk factors for mpox that were not input correctly into EvalMpox.

Finally, clinician judgment remains necessary when interpreting the recommendations from any CDSS. Two individuals who tested positive for mpox reported no epidemiologic risk factors to multiple interviewers and so, following CDC PUI criteria, were designated non-PUI. EvalMpox does direct users to additional clinical resources, and these patients eventually underwent mpox testing due to repeat presentations to care.

In conclusion, our data support the potential for CDSS to assist in the identification, evaluation, and management of patients with emerging infectious diseases, supporting laboratory stewardship and appropriate implementation of transmission-based precautions. Our findings lay the groundwork for future investigations, including into which factors influence healthcare workers (HCW) toward using or not using an available CDSS. It will also be useful, during future outbreaks of emerging infections, to consider randomizing HCW to CDSS use or standard of care, to allow rigorous interrogation of the ability of CDSS to improve diagnostic accuracy and disease-specific knowledge.