Pwani University

Sustainability scientists have engaged in extensive discussions on ethical ways of doing research and argued on the importance of co-production approaches to counter knowledge extractivism. The specific issue of research fatigue, often associated with knowledge extractivism, and the possible methods to counter it, have however received less attention. This paper seeks to contribute to discussions on ethical ways of doing research by focusing on our experience of using theatre, specifically, Forum Theatre, to investigate divergent perceptions of environmental change and related tensions among selected coastal communities in Kenya and Mozambique. We argue that Forum Theatre constitutes an ethical method for sustainability scientists for four reasons: (i) it allows to co-produce knowledge with participants; (ii) it facilitates horizontal exchange; (iii) it creates joyful moments; and (iv) it enables the transmission of skills that remain with participants beyond project durations. The paper engages with these four themes, first theoretically and then proposing a reflection based on our project experience. In the last section, we warn against some of the limitations of the approach.

Background Long-lasting insecticidal nets are the primary malaria vector control measure in coastal Kenya. In 2018, phenotypic resistance to pyrethroids and low frequency of L1014S kdr mutation were reported in the Anopheles gambiae complex. Since then, additional pyrethroid-treated nets were distributed in 2021. The objectives of this study were to determine the insecticide resistance profiles of An. gambiae and Anopheles funestus from Kwale County and evaluate potential resistance mechanisms. Methods From July 2023 to May 2024, adult and larval collections of An. funestus and An. gambiae mosquitoes were done with the aim of conducting insecticide susceptibility bioassays using WHO protocol for permethrin, deltamethrin, bendiocarb, DDT and pirimiphos-methyl. Species found resistant to pyrethroids were subjected to synergism testing by pre-exposure to piperonyl-butoxide (PBO). This was followed by genotyping of resistance-associated mutations in An. funestus (CYP6P9a, CYP6P9b, GSTe2-L119F and 6.5kb S.V) and An. gambiae (kdr L1014S and L1014F). Sibling species identification was done using PCR. The association between genetic markers and phenotypic resistance was explored using logistic regression. Results A total of 1826 An. gambiae and 715 An. funestus were used in insecticide susceptibility bioassays. Both An. gambiae and An. funestus were resistant to permethrin (mortality, 58.7% and 57.1, respectively) and deltamethrin (mortality 51% and 76%, respectively), but susceptible to DDT, bendiocarb and Pirimiphos-methyl. Pre-exposure to PBO increased susceptibility to deltamethrin in both species. Both kdr west and east were detected in Anopheles arabiensis (L1014S freq = 0.083, L1014F freq = 0.063) and Anopheles quadriannulatus (L1014S freq = 0.074, L1014F freq = 0.043) at low frequencies. Anopheles funestus sensu stricto and Anopheles rivolurum had the presence of CYP6Pa, CYP6Pb, 6.5kb S.V and GSTe2-L119F, with low allele frequencies. There were no significant associations between the genotypes and phenotypic profile. Conclusions Malaria vectors in Kwale are resistant to pyrethroids. PBO fully restored susceptibility, indicating this resistance could be caused by metabolic mechanism. The presence of kdr and metabolic resistance alleles suggests a recent selection on Anopheles mosquitoes. Pyrethroid-only nets may not fully ensure community protection against malaria in coastal Kenya due to resistance. Operational failure remains uncertain, requiring further studies. Net distribution programs should consider pyrethroid-PBO nets to enhance malaria control effectiveness.

This study analyzes the distribution, genetic diversity, and spread of Anopheles stephensi in Kenya following initial detection in December 2022. A total of 114 larval and 33 adult An. stephensi samples were confirmed in 7 of 18 surveyed counties majorly along transportation routes. Genetic analyses revealed three distinct genetic compositions with different levels of genetic diversity, suggesting multiple introductions into the country. The genetic composition of mosquitoes in most counties resembled southern Ethiopian populations, while those from Turkana showed a unique haplotype. A species distribution model predicts a more extensive range than currently observed, with low precipitation and minimal seasonal temperature variations as key factors influencing distribution. Challenges in adult sampling were noted, with larval sampling revealing co-occurrence with native Anopheles species. The findings have implications for surveillance and control strategies, emphasizing the need for continued monitoring, refined sampling techniques to inform bionomics, and cross-border collaboration. Supplementary Information The online version contains supplementary material available at 10.1038/s41598-025-04682-1.

Background Clinical diagnosis of tuberculosis (TB), referring to diagnosis without bacteriological confirmation, is common and may affect an individuals’ outcomes. We undertook a systematic review to assess the proportion of people with TB who were diagnosed clinically, and mortality compared to those with bacteriologically confirmation in the published literature. Methods We searched Medline, Embase, Web of Science and Cochrane Library from January 2010 to December 2024 using terms for ‘TB’ and diagnostic studies. We excluded studies with participants aged <15 years, not reporting clinical and bacteriologically confirmed TB, not conducted in high TB burden settings, and studies that were not trials, cohort or cross-sectional in design. Published summary data was extracted and risk of bias assessed. Summary estimates for proportion of diagnoses that were clinical were calculated overall and by pre-specified subgroups. Risk ratio for mortality of clinical compared to bacteriological diagnosis was evaluated by random effects meta-analysis. This review was prospectively registered (PROSPERO CRD42023404419). Findings Our search identified 5693 records, of which 53 datasets were included. 12 studies were rated as low risk of bias. Median proportion of TB diagnosed clinically (n = 85,623) was 40% (95% CI: 31–46%, interquartile range 27%–53%). The proportion of TB diagnosed clinically was higher in people living with HIV and extrapulmonary TB. Clinical diagnosis did not differ by diagnostic modality available or by study year. The pooled risk ratio for mortality (n = 20,523, 10 studies) was 1·5 (95% CI: 1·0–2·2, I² = 78·7%) indicating higher mortality in people diagnosed clinically. Interpretation Clinical diagnosis of TB remains common and was associated with higher mortality risk than bacteriologically confirmed TB, suggesting conditions other than TB that are not being adequately treated. Better understanding of reasons for clinical TB diagnosis and investment in improved diagnostics for TB and non-TB conditions is needed. Funding UK 10.13039/501100000272National Institute for Health and Care Research and 10.13039/501100000691Academy of Medical Sciences; US 10.13039/100000002National Institutes of Health.

Microbes inhabiting the banana rhizosphere are varied and mediate essential functions that enhance plant growth and increase crop productivity. Their abundance in soil habitats is a potential indicator of soil health and quality. Despite the well-known benefits of rhizosphere microorganisms in banana cultivation, their genomic and functional diversity remains largely unexplored within smallholder agroecosystems. In this study, we characterized the community composition and functional potential of bacteria in banana rhizospheric soils from Gituamba, Mangu and Ngenda constituencies in Kiambu County, Kenya. Using Illumina Novaseq sequencing, we analyzed 16S rRNA gene amplicons and shotgun metagenomic profiles to explore these microbial communities. Variations of soil physicochemical parameters across the study sites were assessed. The parameters varied across the sites, with soils in Gituamba and Ngenda depicting better soil fertility characteristics than Mangu. Amplicon sequencing profiles revealed higher bacterial diversity in Gituamba compared to Mangu, while the single sample from Ngenda exhibited moderate diversity. The dominant phyla across the study sites were Proteobacteria, Actinobacteria, and Cyanobacteria. Functional profiling of 16S rRNA gene amplicons showed a higher enrichment in Gituamba compared to Mangu. Overall, the functional profiling indicated that predicted metabolic pathways across the study sites were linked to genes encoded by the members of the most abundant bacterial phyla in the soil environments, majorly contributing to beneficial roles for soil health and crop yield. This study offers methods to reveal the banana rhizosphere as a rich reservoir for potential microbes of agricultural and biotechnological significance, which can promote sustainable agriculture.

Neurodevelopmental disorders (NDDs) are a group of conditions with their onset during the early developmental period and include conditions such as autism and intellectual disability. Occurrence of NDDs is thought to be determined by both genetic and environmental factors, but data on the role of environmental factors for NDD in Africa is limited. This study investigates environmental influences on NDDs in children from Kenya. This case-control study compared children with NDDs and typically developing children from two studies on the Kenyan coast. We included 172 study participants from the Kilifi Autism study and 151 from the NeuroDev study who had a diagnosis of at least one NDD and 112 and 73 with no NDD diagnosis from each study, respectively. Potential risk factors were identified using unadjusted univariable analysis and adjusted multivariable logistic regression. Univariable analysis in the Kilifi Autism study sample revealed hypoxic-ischaemic encephalopathy conferred the largest odds ratio (OR) 10.52 [95%CI: 4.04, 27.41] for NDDs, followed by medical complications during pregnancy (gestational hypertension & diabetes, eclampsia, maternal bleeding) (OR=3.17 [95%CI: 1.61, 6.23]). In the NeuroDev study sample, labour and birth complications (OR=7.30 [95%CI 2.17, 24.61]), neonatal jaundice (OR=5.49 [95%CI 1.61,18.72]) and infection during pregnancy (OR= 5.31 [95%CI 1.56, 18.11]) conferred the largest risk associated with NDDs. In the adjusted analysis, seizures before age 3 years in the Kilifi Autism study and labour and birth complications in the NeuroDev study conferred the largest increased risk. Higher parity, the child being older and delivery at home were associated with a reduced risk for NDDs. Recognition of important risk factors such as labour and birth complications could guide preventative interventions, developmental screening of at-risk children and monitoring progress of these children. Further studies examining the aetiology of NDDs in population-based samples, including investigating the interaction between genetic and environmental factors, are needed.

Piscine myocarditis virus (PMCV), a double‐stranded RNA virus, is the causative agent of cardiomyopathy syndrome (CMS) in Atlantic salmon Salmo salar , which was first reported in Ireland in 2012. PMCV continues to be a disease threat to Atlantic salmon aquaculture in Ireland, and as such, it is of growing importance to understand how the virus spreads. Genetic sequences allow for transmission pathways to be examined, with whole genomes providing the most accurate information. In this study, whole genome sequencing has been applied to Irish strains of PMCV, in doing so revealing how > 80% of the genetic diversity of the virus lies outside the commonly sequenced open reading frames (ORFs). Second, this paper examines the effectiveness of incorporating a MinION sequencing approach into routine diagnostics by comparing a MinION generated genome to the corresponding sequence generated via Illumina MiSeq. The results showed the MinION genome shared 99.59% identity with the Illumina genome, and while this accuracy may be sufficient for studies such as pathogen identification and deeper evolutionary questions, it was shown to be insufficient for accurately tracking viral transmission pathways. Finally, comparing Irish and Faroese sequences reveals that some strains of PMCV in Ireland may originate from wild fish.

Aedes aegypti is the primary vector of arboviral diseases such as dengue, chikungunya, yellow fever, and Zika, posing significant global health and economic challenges. The effective control of this mosquito species requires understanding its seasonality, feeding behavior, and ecological dynamics. Modern molecular techniques, such as amplicon metagenomic sequencing, provide insights into vector–host interactions and feeding patterns. This study investigated the temporal variation of Ae. aegypti abundance and its blood meal sources in coastal and western Kenya over 16 months. A total of 64,360 mosquitoes were collected, with Ae. aegypti comprising 10.9% (7035/64,360). Coastal sites had a higher proportion (64.7%) of Ae. aegypti than western Kenya. Seasonal variation in abundance was observed, with peaks during the long rainy season and decline during the dry season. Blood meal analysis identified 15 vertebrate hosts, with humans being the primary source (86.6–95.9%). Other hosts included domestic animals such as turkey, sheep, cow, goat, and chicken. These findings highlight the role of rainfall in arboviral disease transmission and Ae. aegypti’s strong preference for human hosts. Additionally, this study demonstrates the cost-effectiveness of group testing for identifying blood meal sources, with implications for public health interventions.

Introduction Whereas Sustainable Development Goal 2 seeks to achieve zero hunger by 2030, food security remains a global challenge. Therefore, SDG 2 aims to achieve sufficient food access through sustainable agriculture. This is in line with the tenets of SDG1 that endeavor to end poverty in its diverse forms and, among other targets, secure land tenure rights. However, disparities in land tenure security still exist, especially in developing societies, and the prevalence of food insecurity is rising. Consequently, this study sought to establish the influence of land use patterns on food security. Methods A sample of 301 households was selected using proportionate sampling techniques. Accordingly, we used a cross-sectional research design, whereas data was collected using semi-structured interview guides. The Malthusian theory was used as an interpretive theoretical framework for this study. Results and discussion Furthermore, using the Household Food Insecurity Access Scale (HFIAS), it was established that the study site was food insecure, whereby only 22.6% of the households were food secure. Moreover, 39.5% of households experienced severe food insecurity, 16.3% had moderate food access, and 21.3% had mild food security. Results further revealed that agriculture characterized by subsistence farming was the primary land use pattern. Additionally, findings revealed agricultural expansion and increased demand for settlement spaces, resulting in deforestation, clearing of bushes, and general land degradation. Consequently, soil erosion and loss of soil fertility necessitating the use of pesticides and fertilizers were reported. Using a 5% significance level, a chi-square analysis revealed that these land use changes significantly influenced food security. Based on these findings, we recommend adopting sustainable agriculture to boost land productivity and diversify livelihood sources to reduce overreliance on mono-crop subsistence farming. Moreover, we draw the need to increase awareness of environmental conservation through agriculture extension services.

Background Nurses comprise the dominant cadre of healthcare workers yet there remains an acute shortage of nurses globally with Africa most affected. However, access to higher nursing education in sub-Saharan Africa remains limited. We aimed to i) Assess the need for a Master of Nursing (MScN) program among graduate nurses in Kenya ii) Identify preferred MScN program options among graduate nurses intending to enrol for MScN in Kenya iii) Identify skills mismatch among graduate nurses in Kenya iv) Assess the relationship between intention to enrol in MScN program and job satisfaction among graduate nurses in Kenya. Methods A cross-sectional design employing an online survey was used. Consenting nurses with a first degree in nursing were included. Socio-demographic indicators, job satisfaction and skills mismatch were assessed. The need for an MScN program was assessed by determining the proportion of participants who expressed the desire to pursue an MScN program. Preferred MScN program options were determined as frequencies and proportions. Skills mismatch was computed as frequencies and proportions. The relationship between the need for MScN and job satisfaction was assessed using the point biserial-correlation. Results Of the 355 volunteers enrolled, (n = 337, 94.9% [95% CI: 92.1–96.9]) expressed the desire to pursue MScN training with the critical care/renal specialty (n = 84, 24.9% [95% CI: 20.3–29.9]) being the most preferred. A majority of the volunteers (n = 319, 89.9%) felt their skills were inferior to their responsibilities (under skilled). We found no significant correlation between the need for MScN and job satisfaction (r = 0.058; p = 0.269). Conclusion Our findings suggest a strong desire by graduate nurses in Kenya to pursue MScN with a preference for critical care specialization. There is a need to establish more MScN programs in Kenya coupled with the deployment of nurses as per the scope of practice.

Biodiversity monitoring using metabarcoding is now widely used as a routine environmental management tool. However, despite the rapid advancement of third‐generation high‐throughput sequencing platforms, there are limited studies assessing the most suitable tools and approaches for environmental metabarcoding studies. We tested the utility of Oxford Nanopore Technologies MinION sequencing for short‐read amplicon sequencing of mitochondrial COI mini‐barcodes from a known composition of arthropod species and compared its performance with more commonly used Illumina NovaSeq sequencing. The mock arthropod species assemblage allowed us to optimise a bioinformatic filtering pipeline to identify arthropod species using MinION long reads. Using this pipeline, we identified host species and diet composition by sequencing droppings collected from five individual Irish brown long‐eared bats (Plecotus auritus) roosts. We showed that MinION data provided a similar taxonomic assignment to NovaSeq but only if the reference species barcode database was accurate and comprehensive. The P. auritus diet inferred was as expected based on previous morphological and Illumina metabarcoding studies. We showed that less sequencing depth, but a higher number of biological samples were necessary for complete species composition detection by MinION. A relatively simple bioinformatic filtering tool such as NanoPipe could adequately retrieve both host species and diet composition. The biggest standing challenge was the reference database format transferability and comprehensiveness. This pipeline can be used to guide future metabarcoding studies using nanopore sequencing to minimise the cost and effort while optimising results.

Background Self‐help groups offer an approach to empowering the lives of caregivers and their children with disabilities in settings of limited resources and support. A study was conducted over a 5‐year period (2018–23) to assess the sustainability of 11 self‐help groups in Kilifi, Kenya, during which there was the COVID‐19 pandemic. Methods An integrated framework of action research and mixed methods was carried out over three stages. Stage 1: pre‐pandemic, three self‐help groups participated in focus group discussions. Template analysis structured around the five pillars of the WHO community‐based rehabilitation matrix (CBR: health, education, livelihood, social, empowerment) was carried out. Stage 2: inter‐pandemic, a bespoke questionnaire was administered to monitor each group. Descriptive statistics were reported (Questions 1–6) and the CBR template was applied to free‐field responses (Questions 7–8). Stage 3: post‐pandemic, a quality of life (QoL) questionnaire was administered to 21 caregivers of children with disabilities and a control group of 11 parents of typically developing children in the same geographical area and 8 caregivers pre‐ and post‐pandemic. Descriptive statistics were applied. Results Pre‐pandemic, there was food security, medicine availability, school attendance, social connections and livelihood. Group plans involved livelihood and social inclusion developments. However, member commitment, community attitudes and environmental conditions were ongoing challenges. Inter‐pandemic, some socially distanced group meetings focus on COVID‐19 prevention, livelihood and social support. Livelihood activities were affected variously with reported difficulties including food insecurity, school closures and reduced meeting frequency. Post‐pandemic comparison between caregivers and a control group revealed overall significantly higher caregiver QoL scores. Pre‐ to post‐pandemic evaluations demonstrated overall significantly improved caregiver QoL. Conclusions Despite the changing context brought by the COVID‐19 pandemic, the self‐help groups appear to have afforded some protection against the worst psychosocial and economic effects and helped to sustain the caregivers.

Plain language summary In early 2024, there was a significant increase in cases of acute hemorrhagic conjunctivitis (AHC), commonly known as "red eye" or “pink eye’ disease, in coastal Kenya, which led to the Ministry of Health issuing an outbreak alert. To understand the cause of this outbreak, we collected eye swabs from 13 patients at an eye clinic in Mombasa City, coastal Kenya. Initial testing of these samples was performed by checking for growth of pathogenic bacteria, and presence of a commonly known viral agent of AHC, adenovirus, using a polymerase chain reaction. The results showed no presence of harmful bacteria in the samples, and all tested negative for adenovirus. However, on further analysis of these samples by characterizing all non-human genetic material in the samples using an approach known as metagenomic sequencing, we detected the presence coxsackievirus A24 variant (CA24v) genetic material, a well-recognized cause of AHC, in three patients. The presence of this virus in the samples was confirmed through its specific in-house molecular test and a second genetic analysis approach. The CA24v strains detected in the samples were determined to be closely related to a known CA24v genotype assigned IV. Our findings suggest that the AHC disease outbreak observed in coastal Kenya in early 2024 was likely caused by CA24v. We highlight the usefulness of metagenomic sequencing as a tool for identifying the causes of new disease outbreaks, crucial for public health responses.

Background Nurses comprise the dominant cadre of healthcare workers yet there remains an acute shortage of nurses globally with Africa most affected. However, access to higher nursing education in sub-Saharan Africa remains limited. We aimed to i) Assess the need for a Master of Nursing (MScN) program among graduate nurses in Kenya ii) Identify preferred MScN program options among graduate nurses intending to enrol for MScN in Kenya iii) Identify skills mismatch among graduate nurses in Kenya iv) Assess the relationship between intention to enrol in MScN program and job satisfaction among graduate nurses in Kenya. Methods A cross-sectional design employing an online survey was used. Consenting nurses with a first degree in nursing were included. Socio-demographic indicators, job satisfaction and skills mismatch were assessed. The need for an MScN program was assessed by determining the proportion of participants who expressed the desire to pursue an MScN program. Preferred MScN program options were determined as frequencies and proportions. Skills mismatch was computed as frequencies and proportions. The relationship between the need for MScN and job satisfaction was assessed using the point biserial-correlation. Results Of the 355 volunteers enrolled, (n = 337, 94.9% [95% CI: 92.1–96.9]) expressed the desire to pursue MScN training with the critical care/renal specialty (n = 84, 24.9% [95% CI: 20.3–29.9]) being the most preferred. A majority of the volunteers (n = 319, 89.9%) felt their skills were inferior to their responsibilities (under skilled). We found no significant correlation between the need for MScN and job satisfaction (r = 0.058; p = 0.269). Conclusion Our findings suggest a strong desire by graduate nurses in Kenya to pursue MScN with a preference for critical care specialization. There is a need to establish more MScN programs in Kenya coupled with the deployment of nurses as per the scope of practice.

Background Health systems in low and middle-income countries (LMICs) face chronic Human Resources for Health (HRH) shortages. This is especially worse in rural and primary healthcare settings. The Cuban government since 1960s has been implementing a policy strategy for producing healthcare workers for export, to boost their economy. Several LMICs have since established health cooperation programs with Cuba to import health workers to address their shortages. This review aimed to examine the emergence, design, utility, outcomes, and lessons learned from the implementation of these programs. Methods We conducted a scoping review using the Joanna Briggs Institute (JBI) methodology and searched for literature across four databases. Two independent reviewers screened the articles and selected relevant articles based on pre-defined criteria. We extracted data and synthesized findings using thematic analysis. Results We included 71 articles after screening 3509 articles. Cuban health cooperation programs have been implemented in many LMICs in South America, Africa, Southeast Asia, and the Pacific region. These programs are formalized primarily through bilateral agreements and implemented as exchange initiatives. This involves importing Cuban healthcare workers and sending collaborating country students to study in Cuba. These programs aimed to address HRH shortages and maldistribution, inadequate training capacity, and respond to medical emergencies in the host countries. Cuban healthcare workers, primarily family physicians, within the host countries; are deployed in primary healthcare settings, increasing the rural health workforce, and improving healthcare access and outcomes. These programs have faced several challenges including opposition from local medical professionals, underutilization due to poorly coordinated recruitment, and language barrier. Conclusion Cuban health cooperations in LMICs have shown diverse results based on their structures. Long-term comprehensive programs have proven to be more successful in boosting the healthcare workforce and enhancing health outcomes. Key factors for optimizing HRH health cooperation include effective collaborative decision-making and need-based deployment.

Background The surgical field in Africa has long grappled with a gender imbalance, with women being significantly underrepresented. Despite global efforts to foster gender diversity in healthcare practices, African women pursuing surgical careers still face substantial hurdles. This paper investigates these women’s experiences and challenges, aiming to raise awareness of these issues and propose strategies for improving gender equity. Objective To describe contextual aspects of barriers affecting women in surgery in Africa. Methods A cross-sectional survey was conducted, targeting female medical students interested in surgical careers, interns, trainees, and surgical consultants across Africa. The survey was distributed in November-December 2021. Data were analysed using descriptive statistics for quantitative data and a simplified thematic analysis for qualitative data. Results A total of 105 participants from 17 countries, aged 20 to 50 years and with various training levels, completed the survey. General surgery was the most common speciality among the respondents. Notably, 63% reported gender-based discrimination, with many (74%) attributing societal and familial discouragement and financial commitments as major barriers to pursuing surgical careers. Participants also shared experiences of gender-based inequity, underestimation of their skills, sexist comments, and even instances of sexual harassment during training or work. Conclusion This study sheds light on the complex barriers African women face in pursuit of surgical careers. To enhance diversity in the field, fundamental change is required. This necessitates recognising the underlying causes hindering women’s progress in surgery and the implementation of interventions to promote gender equity.

Tuberculosis (TB) remains a leading cause of death globally. Patients who get lost to follow-up (LTFU) during TB treatment have high risk of relapse, mortality, treatment failure and developing Multidrug resistant TB. Empirical data to monitor long-term TB treatment outcomes in low-and-middle income countries (LMICs) are sparse. We determined proportion of TB patients who die or are LTFU during six months of treatment and identified factors independently associated with mortality or LTFU. A retrospective cohort using data from routine Meru County TB surveillance system was conducted. We included 38020 records of TB patients aged ≥15 years on treatment 2012‒2022. TB treatment outcomes of interest were LTFU or death within six months of treatment. Survival analyses accounting for competing events were performed. Among the 38020 patients, 27608 (73%) were male and the median (IQR) age was 32 [25‒42] years. 26599 (70%) had bacteriologically confirmed TB while 11421 (30%) were clinically diagnosed. During 16531 person-years of follow-up, 2385 (6.3%, 95%CI 6.0‒6.5) and 1942 (5.1%, 95%CI 4.9‒5.3) patients were LTFU and died respectively. In the multivariable model, patients on re-treatment after LTFU or after failure and those previously treated compared to new TB patients were positively associated with LTFU. Patients coinfected with HIV and those with unknown HIV status were positively associated with LTFU. In contrast, females, clinically diagnosed, extra-pulmonary TB and older patients were negatively associated with LTFU. HIV infected patients on ARVs, not on ARVs and unknown HIV status compared to not infected were positively associated with mortality. Other comorbidities (not HIV), clinically diagnosed, undernourished and older patients were positively associated with mortality. More than 10% of TB patients either die or are LTFU before completing treatment. Targeted interventions are needed to improve treatment outcomes for TB patients who are at high risk of death or being LTFU.

Objectives Approximately 80% of people with epilepsy live in low- and middle-income countries (LMICs), where limited resources and stigma hinder accurate diagnosis and treatment. Clinical machine learning models have demonstrated substantial promise in supporting the diagnostic process in LMICs by aiding in preliminary screening and detection of possible epilepsy cases without relying on specialised or trained personnel. How well these models generalise to naïve regions is, however, underexplored. Here, we use a novel approach to assess the suitability and applicability of such clinical tools to aid screening and diagnosis of active convulsive epilepsy in settings beyond their original training contexts. Methods We sourced data from the Study of Epidemiology of Epilepsy in Demographic Sites dataset, which includes demographic information and clinical variables related to diagnosing epilepsy across five sub-Saharan African sites. For each site, we developed a region-specific (single-site) predictive model for epilepsy and assessed its performance at other sites. We then iteratively added sites to a multi-site model and evaluated model performance on the omitted regions. Model performances and parameters were then compared across every permutation of sites. We used a leave-one-site-out cross-validation analysis to assess the impact of incorporating individual site data in the model. Results Single-site clinical models performed well within their own regions, but generally worse when evaluated in other regions (p<0.05). Model weights and optimal thresholds varied markedly across sites. When the models were trained using data from an increasing number of sites, mean internal performance decreased while external performance improved. Conclusions Clinical models for epilepsy diagnosis in LMICs demonstrate characteristic traits of ML models, such as limited generalisability and a trade-off between internal and external performance. The relationship between predictors and model outcomes also varies across sites, suggesting the need to update specific model aspects with local data before broader implementation. Variations are likely to be particular to the cultural context of diagnosis. We recommend developing models adapted to the cultures and contexts of their intended deployment and caution against deploying region- and culture-naïve models without thorough prior evaluation.

Objective First unprovoked seizures and acute seizures are common and can develop into epilepsy. The risk of epilepsy following these seizures in community samples is not well established, and it is unclear whether the probability of subsequent unprovoked seizures following these seizures reaches the International League Against Epilepsy's threshold of 60%. Methods We followed participants initially classified as having first unprovoked seizures, having acute seizures, or without seizures in a community‐based survey conducted in 2003 to estimate the subsequent risk of epilepsy in 2008 and 2021. The diagnosis of epilepsy in 2008 and 2021 was based on data from a community survey and health care visits to Kilifi County Hospital and the epilepsy clinic. Poisson regression models were used to compute incident risk ratios (IRRs) for epilepsy and population‐attributable risk (PAR); population‐attributable risk fractions (PAFs) were computed from contingency tables. Results In the 5‐year follow‐up (censored in 2008 survey), the IRR for epilepsy was 23.3 (95% confidence interval [CI] = 14.2–38.2) for first unprovoked seizures and 10.4 (95% CI = 5.6–19.5) for acute seizures compared to the no‐seizure group. By 2021 (including 2008), the IRR was 18.4 (95% CI = 11.9–28.5) for first unprovoked seizures and 7.9 (95% CI = 4.3–14.5) for acute seizures compared to the no‐seizure group. The PAR for first unprovoked seizures and acute seizures was 29.0 and 8.0/1000 persons in the long‐term follow‐up. The PAF was 56.3% for first unprovoked seizures and 26.3% for acute seizures in the long‐term follow‐up. There was a high probability that a person with acute seizures (72%) or first unprovoked seizures (92%) developed epilepsy earlier than a person from the comparison group. Significance First unprovoked seizures and acute seizures are associated with high risk for developing epilepsy. Neurological correlates for epilepsy risk following first unprovoked seizures should be investigated to inform epilepsy diagnosis and treatment.