📄 medicine

Freedom From Infection in Cabo Verde: Supporting Evidence-Based Decision-Making for Malaria Elimination

This study applies the Freedom From Infection (FFI) framework to Cabo Verde's post-certification malaria surveillance data, demonstrating that while 72% of facilities currently meet elimination criteria under standard definitions, maintaining robust surveillance capacity is critical to sustaining freedom from infection amidst ongoing importation risks.

Original authors: Luca Nelli, Adilson DePina, Isabel Byrne, Aderitow Goncalves, Davidson Monteiro, Elin Dumont, Monica Hill, Lynn Grignard, Antonio Moreira, Kevin Tetteh, Silvânia Leal, Chris Drakeley, Jason Matthiopou

Published 2026-08-03

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Luca Nelli, Adilson DePina, Isabel Byrne, Aderitow Goncalves, Davidson Monteiro, Elin Dumont, Monica Hill, Lynn Grignard, Antonio Moreira, Kevin Tetteh, Silvânia Leal, Chris Drakeley, Jason Matthiopoulos, Gillian Stresman

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the world of public health as a giant, invisible game of "Where's Waldo?" but instead of a cartoon character, you are hunting for a microscopic parasite that causes malaria. For decades, the goal was simply to find as many "Waldos" as possible and get rid of them. But now, some places have reached a new level: they claim the game board is empty. They say, "We found zero Waldos for three years straight!" However, the tricky part is knowing if the board is truly empty or if the "Waldo" is just really good at hiding. This is where the science of "Freedom From Infection" comes in. It's not just about counting the cases you see; it's about measuring how good your search team is. If your team has bad flashlights or misses spots, finding zero cases might just mean you didn't look hard enough, not that the enemy is gone. This paper dives into that exact problem: how do we know a place is truly safe, especially when the enemy might try to sneak back in from the outside?

The story takes place in Cabo Verde, a beautiful island nation off the coast of West Africa. In January 2024, this country was officially certified as "malaria-free" by the World Health Organization. It was a massive victory after years of fighting the disease. But here's the plot twist: just because you won the championship doesn't mean you can stop training. The islands still face a risk of "importation," which is like a traveler bringing a hidden "Waldo" in their suitcase from a place where malaria is still common. The scientists in this study wanted to answer a very specific question: Now that Cabo Verde has the trophy, is their surveillance system (their "search team") actually good enough to catch any new intruders immediately, or are there blind spots where the disease could hide and start a new outbreak?

To solve this mystery, the researchers used a clever statistical tool called the "Freedom From Infection" (FFI) framework. Think of this framework as a super-smart detective that doesn't just look at the number of people who got sick. Instead, it builds a digital twin of the entire health system. It asks: "If a malaria parasite did show up in a specific town, what are the odds that our system would actually find it?" It combines two main stories: the "State Process" (what is actually happening in the population, like invisible infections) and the "Observation Process" (how well the health system catches them, from a person feeling sick to getting a test and a report). By running this model on data from 25 different health facilities over 14 years, the team could calculate a "Probability of Freedom" for each location. It's like giving every town a score from 0% to 100% on how confident we can be that they are truly free of the disease.

The results revealed a fascinating split personality across the islands. When the scientists used a standard definition of elimination (fewer than one infection per 10,000 people) and required a high level of confidence (95%), they found that 18 out of the 25 facilities (72%) had a strong case for being malaria-free as of December 2024. However, the other 7 facilities were a bit more shaky. Some had scores that hovered near zero, meaning the data suggested their surveillance system might not be sensitive enough to guarantee they are truly safe. The study also tested how "strict" the rules were. If you made the definition of "free" even harder to meet (like demanding fewer than one infection per 30,000 people), fewer facilities passed the test. This shows that the label of "eliminated" depends heavily on how strict you are with your rules.

One of the most important findings was about stability. The researchers compared the period before the official certification (the last 36 months of 2023) with the year after (2024). They found that the status of the facilities didn't change much. In fact, no facility suddenly became "safer" after the certification; if anything, a couple of places looked slightly less secure under stricter rules. This suggests that getting the certificate didn't magically fix the surveillance gaps. The "search team" was doing about the same job before and after the award. The study explicitly ruled out the idea that the country was suddenly safer just because they got the title; instead, it highlighted that the risk of reintroduction remains real, and the ability to detect it depends on specific local factors.

The paper also pointed out exactly where the "search team" was struggling. It wasn't necessarily that people weren't getting sick or that the mosquitoes were gone. The main bottleneck was often the "testing" step. In some facilities, even if a person with a fever showed up, the chance of them actually getting a malaria test was very low. It's like having a great detective who never picks up the phone. The study found that facilities with a proven ability to detect imported cases (cases brought in by travelers) had much higher confidence scores. This suggests that the key to staying malaria-free isn't just waiting for local cases, but having a system that is hyper-vigilant about travelers and ensures that anyone with a fever gets tested immediately.

In the end, this paper doesn't declare Cabo Verde's victory as "solved" or "permanent." Instead, it offers a transparent, data-driven map of where the country stands. It suggests that while the country has achieved a historic milestone, maintaining that status requires constant vigilance. The "Freedom From Infection" framework acts like a stress test, showing that if the rules get stricter, some areas might not pass. The authors conclude that to keep the disease away, Cabo Verde needs to focus on strengthening the weak links in their chain—specifically, making sure that febrile patients are actually tested. It's a reminder that in the game of public health, the trophy is just the starting line, and the real work is keeping the searchlights bright so that no "Waldo" ever gets a chance to hide again.

Technical Summary: Freedom From Infection in Cabo Verde

Problem Statement
Cabo Verde achieved World Health Organization (WHO) certification as malaria-free in January 2024 after four consecutive years of zero indigenous cases. However, the country remains vulnerable to malaria reintroduction due to continued importation of cases and ecological receptivity (the presence of Anopheles arabiensis vectors) across multiple islands. Traditional surveillance mechanisms, while effective for monitoring trends, face limitations in quantitatively proving the absence of transmission or assessing the robustness of elimination conclusions against varying operational definitions. There is a need for a rigorous, probabilistic framework to support post-certification decision-making, specifically to quantify the evidence for "freedom from infection" and to stress-test how sensitive these conclusions are to definitional assumptions (e.g., infection thresholds and confidence levels).

Methodology
The study applied and extended the Freedom From Infection (FFI) framework, a Bayesian state-space modeling approach, to Cabo Verde's archipelago context. The methodology integrated three primary data streams:

Passive Case Detection (PCD): 14 years (2014–2024) of routine surveillance data from 25 health facilities, including attendance, suspected cases, testing rates, and confirmed infections (distinguishing between locally acquired and imported).
Active Case Detection (ACD): Targeted screening in Praia and Boa Vista (2021) using respondent-driven sampling for travelers returning from Africa, utilizing PCR and serological markers (antibodies) to detect latent infections and historical exposure.
Health System Surveys: Data on facility capacity, diagnostic availability, and catchment populations.

Model Structure:

State Process: Modeled the latent (unobserved) number of malaria infections, decomposing them into endemic (locally acquired) and incoming (imported) components. Endemic transmission was modeled using a temporal autoregressive process, while incoming infections were estimated via a spatial kernel based on distance between source and receiver sites.
Observation Process: Modeled the surveillance detection cascade as four sequential stages: symptom development, care-seeking, testing of febrile attendees, and correct diagnosis/reporting. This allowed for the estimation of surveillance sensitivity (SSe) and facility-specific bottlenecks.
Extensions: The model incorporated risk stratification (grouping facilities into high, mid, and low-risk strata based on historical transmission and entomological data) and facility-specific capacity to detect imported infections.

Key Contributions

Quantification of $P_{FREE}$ : The study generated facility-level time series of the posterior probability of freedom from infection ( $P_{FREE}$ ), defined as the probability that true infection prevalence lies below a pre-specified threshold (e.g., <1 infection per $d$ people).
Stress-Testing Definitions: The authors systematically varied the elimination denominator ( $d$ , ranging from 100 to 30,000) and the probability cut-off (0.50 to 1.00) to assess the robustness of elimination classifications.
Pre- vs. Post-Certification Comparison: The study compared facility classifications in the 36 months leading up to certification (pre-2024) against the certification year (2024) to evaluate stability.
Surveillance Sensitivity Analysis: By jointly modeling transmission and detection, the study identified specific bottlenecks in the surveillance cascade, particularly the probability of testing febrile patients ( $P_{TEST}$ ).

Results

Surveillance Performance: There was significant heterogeneity across the 25 facilities. While the probability of care-seeking ( $P_{SEEK}$ ) was generally high (median 0.998), the probability of testing febrile attendees ( $P_{TEST}$ ) showed a dichotomy, with a subset of facilities having very low testing rates (<0.10).
Freedom from Infection ( $P_{FREE}$ ): Using a reference definition (<1 infection per 10,000 people; 95% confidence cut-off), 18 of 25 facilities (72%) met the elimination criterion in December 2024. However, a small subset of facilities remained below the cut-off, with two facilities showing $P_{FREE}$ of 0.00.
Sensitivity to Thresholds: The proportion of facilities meeting elimination criteria was moderately sensitive to the definition of elimination. Stricter thresholds reduced the proportion of facilities classified as meeting the criterion. Specifically, at a 0.95 probability cut-off, the proportion dropped to 40% only when the denominator was increased to 30,000; at a 0.99 cut-off with a denominator of 10,000, the proportion also fell to 40%.
Stability: Comparisons between pre- and post-certification periods revealed that facility classifications were largely stable. No facilities improved their status from "not meeting" to "meeting" the criterion in 2024; instead, a small number of facilities experienced a decrease in status, primarily under stricter definitions.
Risk Stratification: High-risk strata showed a credible increase in endemic force of infection compared to low-risk settings. However, historical risk strata did not clearly explain variations in testing probability; rather, facilities with demonstrated capacity to detect imported infections had higher inferred surveillance sensitivity.

Significance and Claims
The paper claims that the FFI framework provides a robust, probabilistic tool for sustaining malaria elimination in settings with ongoing importation risks. Its primary significance lies in:

Moving Beyond Binary Certification: It shifts the focus from a time-bound certification milestone to a continuously evidenced status that can be monitored and audited.
Explicit Uncertainty: It quantifies uncertainty in elimination conclusions, making the trade-offs between stringency (thresholds) and observed performance transparent to policymakers.
Targeted Intervention: By identifying specific bottlenecks (e.g., low $P_{TEST}$ in certain facilities), the framework offers actionable evidence for strengthening surveillance systems to prevent reintroduction.
Generalizability: While applied to malaria in Cabo Verde, the authors note the approach is generalizable to other infections where decision-making depends on demonstrating freedom relative to an explicit threshold.

The study concludes that maintaining strong surveillance capacity is essential, particularly in receptive foci, and that the FFI framework enables programs to "stress-test" their elimination assumptions to ensure the likelihood of sustained transmission interruption.

Technical Summary: Freedom From Infection in Cabo Verde

More like this