A transformer-based sequential analysis methodology for trajectory tracking at safety surveillance in manufacturing

Power-free and sensitive on-site detection of a foodborne bacterial pathogen via confinement effect activated nanomachines-mediated amplification (CEANMA).

Development of an electronic patient safety surveillance model to efficiently identify potential adverse events for gastrointestinal endoscopies.

ABSTRACT The Chinese National Adverse Events Following Immunization Information System (CNAEFIS) is the national-level passive vaccine safety surveillance system. This study aims to identify post-marketing vaccine adverse reaction signals in Chongqing through systematic analysis of 15-year AEFI data, verify the safety profile of locally recommended vaccines, and provide localized evidence for clinical practice and immunization policy formulation. The study included all 36 marketed vaccines in mainland China and collected all rare adverse reactions (RAR) reported in Chongqing through CNAEFIS from 2006 to 2021. Proportional Reporting Ratio (PRR), Reporting Odds Ratio (ROR), Bayesian Confidence Propagation Neural Network (BCPNN), and Empirical Bayes Gamma-Poisson Shrinker (GPS) were employed to mine potential safety signals within the database. From 2006 to 2021, approximately 90.1 million vaccine doses were recorded in Chongqing, China, and 20,195 individual cases of Adverse Events Following Immunization (AEFIs) were reported. Of these, 3,110 cases (15.4%) were linked to RAR, at an incidence rate of 32.2 per 100,000 doses. Allergic reactions were the most prevalent, with 2,743 cases (88.2% of the RAR), and the Measles-Rubella Vaccine (MR) had the highest IR of allergic reactions, reaching 267.1 cases per 100,000 doses administered. A total of 23 suspicious signals were identified through the combined use of PRR, ROR, GPS, and BCPNN methods. The corresponding clinical diagnoses were mainly non-serious diseases such as angioedema, Henoch-Schönlein purpura, and local abscess. The warning signals of severe diseases such as anaphylactic shock and laryngeal edema, which could be life-threatening due to improper clinical treatment, were weak. The overall safety of vaccines in Chongqing is satisfactory. Vaccine manufacturers should optimize the production process to enhance vaccine tolerability.

ABSTRACT The Canadian National Vaccine Safety Network conducted active safety surveillance for adverse events following COVID-19 and influenza immunizations. This analysis evaluated the association between the administration of mRNA COVID-19 vaccines, influenza vaccines, or co-administration of both, and health events that prevented daily activities, caused work absenteeism, or necessitated medical consultation among individuals with autoimmune diseases. Between September and December 2022, vaccinated and unvaccinated participants from seven provinces and territories self-reported health events within 7 days post-vaccination or over a 7-day period for unvaccinated individuals. This analysis focused on individuals self-reporting autoimmune diseases. Surveys were completed by 6,506 individuals: 1,743 received co-administered vaccines, 2,986 received COVID-19 vaccines alone, 491 received influenza vaccines alone, and 1,286 were unvaccinated. Health event rates were 9.5% for co-administration, 9.3% for COVID-19 alone, 5.9% for influenza alone, and 6.1% for unvaccinated controls. Compared to unvaccinated individuals, the risk of health events was higher for COVID-19 and influenza co-administration [adjusted relative risk (aRR): 1.89, 95% confidence interval (95% CI) 1.41–2.52], and COVID-19 alone [aRR: 1.86, 95% CI, 1.40–2.47], but not for influenza alone (aRR: 1.16, 95% CI, 0.76–1.78). No significant change in emergency department visits or hospitalizations was observed in any vaccine group compared to unvaccinated controls. In individuals with autoimmune diseases, mRNA COVID-19 vaccination increases mild to moderate health events compared to unvaccinated individuals. However, the rate of these events was similar when COVID-19 vaccines were administered alone or concomitantly with influenza vaccines, indicating no additional risk associated with co-administration.

Despite advances in blood safety, emerging infectious agents continue to pose risks to the blood supply. Traditional nucleic acid testing assays primarily target known pathogens, limiting the detection of novel microbes. Nanopore metagenomic sequencing enables agnostic identification of diverse pathogens, potentially enhancing transfusion safety surveillance. We assessed the analytical performance and limit of detection (LoD) of a nanopore MinION metagenomic sequencing workflow for pathogen detection in clinical plasma samples. DNA/RNA was extracted from 14 archived samples from individuals with confirmed infections (HIV, HBV, HCV, WNV), followed by cDNA synthesis, barcoded library preparation, and nanopore sequencing. Quantitative reference panels were tested, and data were analyzed using a comprehensive bioinformatics pipeline. Metagenomic sequencing generated over 2 million reads, with 2.3% of reads mapping to microbial sequences. Key transfusion-transmissible viruses, including West Nile virus (WNV), HIV, and hepatitis C virus (HCV), were reliably detected, while protocol modifications enabled HBV identification. Strain-level characterization identified HIV-1 subtype B, HCV genotypes 1a, 2b, 4a, HBV genotype C, and WNV lineage 1A. High viral loads produced extensive genome coverage, while lower loads yielded limited recovery. Co-infections, including human pegivirus (HPgV-2) and torque teno virus (TTV), were identified. The workflow detected viral targets at 103 genome copy equivalents (GCEs)/mL, with > 50% genome coverage achieved at 104 GCE/mL. Nanopore metagenomic sequencing enables comprehensive detection of blood-borne pathogens in plasma samples. This approach offers a promising complementary strategy for enhancing transfusion safety surveillance, and the demonstrated strain-level characterization supports its potential utility for blood safety applications.

Aspergillusflavus is a globally important human pathogen and agricultural contaminant, while its domesticated relative A.oryzae is widely used in food fermentation and biotechnology. Despite their importance, the evolutionary relationship, population structure and domestication history of these fungi remain unresolved. Here, we present the first global population genomic analysis of 639 A.flavus and A.oryzae isolates from clinical, environmental and food-fermentation sources across multiple continents. Our analyses reveal a complex evolutionary landscape comprising well-separated clades interspersed with highly admixed mosaic groups and potential evidence for multiple independent domestication events giving rise to A.oryzae. Clinical A.flavus isolates are distributed across several clades and mosaic groups, some overlapping with fermentation strains, highlighting an apparent role of domestication and admixture in shaping pathogen diversity. These results challenge current species boundaries and provide a framework for understanding evolutionary history, taxonomy and pangenomic architecture in these fungi, with broad implications for pathogenicity, food safety, biocontrol and metagenomic surveillance.

The rapid evolution of drone technology has expanded its applications across collaborative control, public safety, and aerial imaging, yet reliable object detection remains a challenge due to small target sizes and complex backgrounds in drone-captured imagery. To address these limitations, this paper introduces MFA-YOLO, a high-precision network specifically optimized for small-object detection in drone imagery. The proposed approach integrates three innovations: the Local Feature Mapping (LFM) unit for enhanced fine-grained feature extraction, the Progressive Shared Atrous Pyramid (PSAP) for efficient multi-scale feature integration, and the Dynamic Decoupling Head (DDH) for improved adaptive task alignment. Through these components, MFA-YOLO enhances representational capacity while preserving real-time inference efficiency. Experimental evaluations on the VisDrone benchmark demonstrate a 3.6% increase in AP50, a 2.4% increase in AP, and a 17% reduction in model parameters compared to YOLOv8n. Additional experiments on UAVDT further indicate the model's promising generalization across similar drone datasets. These results highlight MFA-YOLO's potential to advance drone-based perception systems, making them more effective and efficient for safety-critical and real-time applications in resource-constrained UAV environments, such as public safety monitoring, surveillance, and autonomous aerial operations.

The concomitant use of multiple drugs increases the risk of adverse events (AEs) due to drug-drug interactions (DDIs), which remain challenging to identify since clinical trials primarily focus on individual drugs, necessitating postmarket safety monitoring through spontaneous reporting systems. Although several statistical methodologies have been proposed to detect DDI signals with disproportionately high reporting rates, existing methods inadequately account for the hierarchical structure of AEs and potential reporting bias. To address these limitations, we developed a statistical methodology that incorporates the hierarchical structure of AEs using tree-based scan statistics while mitigating reporting bias by assuming DDIs follow a multiplicative interaction model. In simulation studies, our proposed method effectively controlled type I error rate at prespecified significance levels across all simulation scenarios and demonstrated consistent performance in power, sensitivity, and false discovery rate, even with reporting bias present. This novel tree-based scan statistic methodology for detecting DDI signals that accounts for both hierarchical AE structure and potential reporting bias can serve as a valuable tool for postmarket drug safety surveillance.

Background and Aim: Listeria monocytogenes remains a major foodborne pathogen globally, with mortality rates ranging from 20%–40%. The increasing incidence of listeriosis and the limitations of culture-based and polymerase chain reaction-based diagnostics highlight the need for rapid, cost-effective, and highly specific immunoassays. This study aimed to develop and validate a regionally adapted sandwich enzyme-linked immunosorbent assay (ELISA) based on monoclonal and polyclonal antibodies (pAbs) raised against a recombinant p60 antigen derived from a Kazakhstani L. monocytogenes field isolate. Materials and Methods: The p60 gene lacking its N-terminal signal peptide was amplified from a regional L. monocytogenes isolate, cloned into the pET28c(+) vector, and expressed in Escherichia coli arctic express (DE3). Recombinant p60 protein was purified by Ni2⁺-affinity chromatography and used to immunize BALB/c mice and Chinchilla rabbits for monoclonal antibodies and pAbs antibody production. Hybridoma clones were screened for specificity using indirect ELISA and Western blot. A sandwich ELISA was assembled using mAb 1H8 as the capture antibody and horseradish peroxidase-conjugated rabbit pAbs as detection antibodies. Analytical sensitivity and diagnostic performance were evaluated using serial dilutions of recombinant p60 and culture supernatants of L. monocytogenes isolates recovered from 507 food samples. Results: The recombinant p60 antigen (50.3 kDa) was successfully expressed and purified at 5.9 mg/L yield. Among seven stable hybridoma clones, mAb 1H8 exhibited the highest affinity (Ka = 2.5 × 1010 M⁻1) and specificity without cross-reactivity to non-Listeria bacteria. The optimized sandwich ELISA achieved a detection limit of 1.5 ng/mL, corresponding to approximately 103 colony-forming units/mL. All six L. monocytogenes field isolates tested positive in the assay, with results strongly correlating with viable cell counts (R2 = 0.89). The assay demonstrated comparable sensitivity to commercial kits while offering shorter assay time (2 h) and substantially lower production cost. Conclusion: The developed sandwich ELISA provides a sensitive, specific, rapid, and regionally tailored diagnostic platform for detecting pathogenic L. monocytogenes in food samples. By integrating locally produced recombinant antigens and immunoreagents, the assay offers a cost-effective alternative to imported kits, supporting national food safety programs and One Health surveillance initiatives. Keywords: food safety surveillance, foodborne pathogens, hybridoma technology, immunodiagnostics, Listeria monocytogenes, monoclonal antibodies, One Health, p60 antigen, polyclonal antibodies, recombinant protein, sandwich enzyme-linked immunosorbent assay.

ABSTRACT Combined mycotoxin contamination patterns with fungal sources were studied in maize, rice, and peanuts from Beninese markets, to evaluate population exposure by staple foods and to characterise its Fusarium species. An initial survey of 95 maize samples in 2022 showed fumonisins contamination up to 9900 μg/kg, with 20% exceeding the FDA advisory level and the EU maximum level of 2000 μg/kg for direct human consumption. Subsequent analysis in 2024 of another 69 samples also on aflatoxins confirmed the previous findings, with co-occurrence in many samples. Fungal characterisation from maize, sorghum and millet identified F. verticillioides predominantly in maize and F. andiyazi in sorghum. Metabolomic analysis demonstrated that F. verticillioides isolates produced multiple fumonisins, while F. andiyazi produced only fusarins. These findings connect contamination patterns to their fungal sources in cereals and highlight the importance of ongoing surveillance for regional food safety, particularly given climate change impacts on fungal growth.

The 2007 US Food and Drug Administration (FDA) Amendments Act (FDAAA) expanded its safety-related regulatory authorities, including enhanced postmarketing safety surveillance and new clinical study requirements. However, whether FDAAA has been associated with differences in the frequency and timing of postmarket safety-related actions remains poorly understood. To assess whether FDAAA was associated with differences in time to first FDA postmarket drug safety-related action, and to assess whether therapeutic and regulatory characteristics were associated with differences in time to these actions post-FDAAA implementation. This was a cross-sectional study of all novel therapeutics approved by FDA between January 1, 2001, and December 31, 2019, and followed up through December 31, 2024. Approvals were categorized as pre- or post-FDAAA (before or after March 25, 2008). Post-FDAAA therapeutic and regulatory characteristics included drug class, therapeutic area, orphan status, special regulatory pathway, and presence of a boxed warning or FDAAA-mandated postmarket study requirement at approval. Time to first FDA postmarket drug safety-related action, a composite of withdrawals due to safety concerns, incremental boxed warnings, and safety-related communications. Of the 560 novel therapeutics approved, FDA took postmarket safety-related actions for 130 (23.2%) during a median (IQR) follow-up of 12.1 (7.8-18.2) years. These comprised actions within 5 years of approval for 34 of 164 therapeutics (20.7%) approved pre-FDAAA, and 57 of 396 (14.4%) approved post-FDAAA (rate ratio, 0.69; 95% CI, 0.47-1.02; P = .06). Compared to pre-FDAAA approvals, after accounting for therapeutic and regulatory characteristics, there was no statistically significant difference in time to first postmarket safety-related action for post-FDAAA approvals (time ratio, 0.40; 95% CI, 0.15-1.07; P = .07). However, among therapeutics with postmarket safety-related actions within 5 years of approval, median time to first action was shorter post-FDAAA (median [IQR], 3.1 (2.0-4.2) years for pre-FDAAA vs 1.8 [1.3-2.5] years post-FDAAA; P = .004). Among 260 novel therapeutics approved post-FDAAA, after the October 2013 enactment of breakthrough therapy designation, the following therapeutic and regulatory characteristics were associated with time to first postmarket safety-related action: small molecule type (time ratio, 0.24; 95% CI, 0.07-0.81; P = .02), orphan designation (time ratio, 8.29; 95% CI, 2.43-28.27; P < .001), fast track (time ratio, 0.22; 95% CI, 0.08-0.64; P = .005), breakthrough therapy designation (time ratio, 0.10; 95% CI, 0.03-0.32; P < .001), prolonged regulatory review time (>400 days; time ratio, 0.16; 95% CI, 0.03-0.73; P = .02), and FDAAA-mandated postmarket study requirements at approval (time ratio, 0.35; 95% CI, 0.15-0.80; P = .01). This cross-sectional analysis found that the enhanced safety-related regulatory authorities of FDAAA were not associated with differences in time to first postmarket safety-related action. However, among therapeutics with postmarket safety-related actions within 5 years of approval, median time to first action was shorter post-FDAAA implementation.

Glucagon-like peptide-1 receptor agonists (GLP-1RAs)-along with the dual glucose-dependent insulinotropic polypeptide (GIP)/GLP-1RA tirzepatide-are widely acknowledged for their efficacy in managing both type 2 diabetes mellitus and obesity, with expanding indications in cardiometabolic risk reduction. While their glycemic, weight-lowering, and cardiovascular benefits are well established through randomized trials and meta-analyses, concerns remain regarding their safety and tolerability across diverse populations and clinical settings. Gastrointestinal (GI) adverse events-particularly nausea, vomiting, diarrhea, and constipation-are the most common side effects, generally emerging during dose escalation and resolving over time. However, accumulating evidence has identified additional GI complications, including cholelithiasis, cholecystitis, gastroparesis, and bowel obstruction, which may warrant caution in susceptible individuals. Injection-site reactions and worsening of pre-existing diabetic retinopathy are also relevant clinical concerns. Although rare, associations with nonarteritic anterior ischemic optic neuropathy, pancreatitis, medullary thyroid carcinoma, and acute kidney injury (AKI) have been reported, primarily through pharmacovigilance and case-based evidence. Importantly, large-scale randomized trials, meta-analyses, and observational studies suggest that GLP-1RAs do not significantly increase AKI risk and may even confer renal benefits in high-risk populations. There is no confirmed elevated risk of suicidality, but surveillance remains warranted. Safety data in special populations-such as pregnant or lactating women, pediatric patients, and those with advanced renal or hepatic impairment-remain limited and require further study. This state-of-the-art narrative review synthesizes current evidence from clinical trials, pharmacovigilance databases, and real-world cohorts to provide a comprehensive evaluation of the safety and tolerability of GLP-1RAs and tirzepatide. We present clinical strategies for adverse event mitigation, monitoring recommendations, contraindications, and practical considerations for treatment discontinuation or switching. Although these agents offer transformative therapeutic potential, their optimal use requires individualized care, careful patient selection, and ongoing safety surveillance. Future research should prioritize long-term safety in underrepresented populations and strategies to mitigate lean mass loss during therapy.

IntroductionTherapeutic ranking (TR) is a process used in health technology assessment (HTA) by the Institute for Health Technology Assessment (IETS) in Colombia to optimize drug prescription through analyses of effectiveness, safety, costs, and post-marketing surveillance. However, decision-making relying on clinical expert preferences may introduce biases, since they are often influenced by subjective values. This study presented an alternative methodology by IETS that emphasizes a structured, evidence-based approach to improve decision-making in TR.MethodsThe standard TR process involved forming an interdisciplinary group to assess the effectiveness, safety, costs, and post-marketing surveillance of medications, followed by evidence synthesis and informed voting using the Borda method to prioritize technologies based on consensus-driven preferences. To enhance informed voting, a methodological strategy was introduced that incorporated structured comparative tables and relative position scales derived from specific quantitative analyses for each parameter (e.g., surface under the cumulative ranking curve [SUCRA], costs, number of adverse events). This approach visually displayed performance patterns for integrated analysis by clinical experts.ResultsThis adaptation to TR consolidated the results for effectiveness, safety, costs, and economic evaluations of the assessed technologies. The structured format facilitated the identification of performance patterns by technology and enabled the integration of these results into a visual framework that supported deliberations during informed voting. This methodology promoted the integration of multiple evaluated dimensions, ensuring that inputs from epidemiology, pharmacology, health economics, and other specialized analyses were not overshadowed by the clinical experience of the expert panel. Instead, each input contributed to the collective decision-making process.ConclusionsThis methodological approach enhanced TR as an HTA process by fostering evidence-based decisions. Optimizing the process could involve assigning weights to each analysis component, including clinical expertise, to balance their influence on prioritization. Explicit weighting would improve transparency and reproducibility. This model is replicable in other settings, maximizing HTA’s impact on global health systems.

BackgroundPrevious research has demonstrated the effectiveness of lecanemab in treating early Alzheimer's disease (AD). However, its validation in the Hunan region of China remains limited. Traditional evaluation methods lack precision, particularly in brain imaging technologies. To address this gap, this study utilizes 7.0‐tesla ultra‐high‐field magnetic resonance imaging (MRI) and its sensitive susceptibility‐weighted imaging (SWI) sequence, which effectively detects microbleeds. Additionally, diffusion tensor imaging (DTI) analysis, in combination with the along the perivascular spaces (ALPS) index, is employed to evaluate glymphatic function. This study aims to assess the efficacy and safety of lecanemab in AD treatment, focusing on its impact on brain microstructure, glymphatic function, and amyloid‐β (Aβ) clearance.MethodBeginning on June 28, 2024, 27 patients diagnosed with AD were enrolled in the study and commenced treatment with lecanemab. Comprehensive baseline data collection, clinical screening, and cognitive assessments were completed. Each patient underwent SWI and DTI imaging using a 7.0T Siemens Terra MRI scanner equipped with a 32‐channel head coil and parallel transmission (pTX) system, both pre‐ and post‐treatment.ResultOf the 27 participants (mean age 61.67±7.30 years; 9 male), two patients (7.4%; 1 male) experienced flu‐like symptoms following their initial lecanemab injection. No new cortical or subcortical microhemorrhages were detected via 7.0T MRI SWI sequences during treatment. Additionally, there were no instances of ARIA‐E or ARIA‐H among participants. Cognitive function, evaluated through the Mini‐Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA), showed no significant decline during the observation period (p >0.05). Among the 27 patients, 9 (mean age 63.56±8.90 years; 4 male) completed observation for their seventh medication dose. Paired t‐tests revealed a statistically significant increase in left DTI‐ALPS index values post‐treatment (t = ‐2.781, p = 0.024), suggesting improvements in glymphatic function.Conclusion1. Lecanemab treatment in AD patients from Hunan, China, is generally safe, though its long‐term efficacy requires further investigation. 2. The use of 7.0T MRI, particularly its SWI sequence, enhances monitoring of microbleeds and improves safety surveillance. Furthermore, DTI‐ALPS, which reflects the glymphatic function, presents as a potential imaging marker for assessing treatment efficacy.

Retrospective analysis of one year of passive safety surveillance data following implementation of the dengue vaccine, Qdenga® (TAK-003) at private vaccination centers, in Buenos Aires, Argentina

Preparation of zwitterionic covalent organic framework for high throughout extraction of fluoroquinolones from lake water and milk.

Adenoviral vector vaccines were pivotal for COVID-19 control, but postmarketing safety surveillance has identified venous-predominant thrombotic risks not fully explained by platelet-centric mechanisms. We tested an RBC-associated hypothesis using an Ad5 vector-rAd/HA(PR8) rat model within a predefined sub-hemolytic window (<10% hemolysis). Ex vivo, we quantified RBC surface phosphatidylserine (PS) exposure, morphology remodeling by scanning electron microscopy, and microvesicle generation, all aligning with increased procoagulant activity. RBCs also exhibited dose-dependent increases in thrombin generation 4 h after intravenous exposure (108–109 OPU/Rat). In vivo, an inferior vena cava thrombosis model showed a pronounced, dose-responsive rise in thrombus burden, consistent with increased thrombogenic potential. Together, these integrated data provide experimental evidence consistent with RBC involvement under adenoviral exposure, supporting a biologically plausible link to the venous-predominant epidemiology observed during the COVID-19 vaccination era. Reported clinical adenoviral vaccine doses are of the same order of magnitude as the exposures tested here, supporting translational relevance while not implying inter-species or product equivalence. Incorporating RBC-focused endpoints, including PS exposure, morphology indices, microvesicle counts, and thrombin generation, into preclinical and early clinical assessments may enhance safety evaluation and inform vector design to mitigate venous thrombotic risk.

Fine-scale detailed estimation of sea ice concentration (SIC) is pivotal for maritime safety, scientific exploration, and environmental surveillance. However, current datasets frequently present challenges due to their limited resolution, thereby hindering fine-scale analysis of sea ice conditions. This paper introduces a novel Deep Self-Attention Residual U-Net (Deep SARU-Net) architecture to address the limitations inherent in existing super-resolution estimation techniques. By harnessing distinctive multi-stage self-attention mechanisms, orthogonal rectangular convolutional kernels, and residual modules, this architecture significantly augments both the precision and generalizability of SIC super-resolution estimation tasks. Experimental results demonstrate that in the vicinity of the Chukchi Sea, the Deep SARU-Net method exhibits superior performance in terms of both RMSE and SSIM values compared to other models, showcasing its efficacy. Furthermore, generalization analyses across diverse sea regions confirm the model’s universality.

Functional active safety surveillance is essential for post-authorization assessment and life-cycle safety evidence generation of vaccines and medicines. Data collected through active surveillance methods are routinely used for post-approval surveillance of novel vaccines and medicines. Implementing these methods in low- and middle-income countries remains challenging. This systematic review identified and assessed existing active safety surveillance in low- and middle-income countries, including key features, strengths, and limitations. A search of EMBASE and PubMed Google Scholar databases was conducted. The protocol was registered with PROSPERO and adhered to Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) guidelines. Findings were synthesized narratively and categorized by surveillance systems characteristics. Of 13,027 records identified, 423 publications met inclusion criteria. Articles spanned 96 low- and middle-income countries, with India (96), China (57), Brazil (30), Iran (26), Ethiopia (21), Indonesia (20), Uganda (18), Kenya (17), and Ghana (16), most represented. The majority focused on vaccines (211). A total of 127 articles utilized mobile technologies for follow-up, online data collection, and/or adverse event reporting. Fifteen percent of vaccine surveillance systems described in articles demonstrated flexibility to incorporate new vaccines, 34% reported multi-sectoral collaborations, and 10% involved multiple countries. Gaps identified include small sample sizes, lack of sustainability, limited flexible surveillance, staffing challenges, and limited use of standard case definitions and digital technologies. Active safety surveillance in low- and middle-income countries has made progress but still faces challenges. The capture and management of safety data through harmonized digital tools that promote consistency in recording and reporting and cross-country collaborations are crucial in further strengthening active safety surveillance in low- and middle-income countries.