Summary In this paper, we set out general principles and develop geostatistical methods for the analysis of data from spatio-temporally referenced prevalence surveys. Our objective is to provide a tutorial guide that can be used in order to identify parsimonious geostatistical models for prevalence mapping. A general variogram-based Monte Carlo procedure is proposed to check the validity of the modelling assumptions. We describe and contrast likelihood-based and Bayesian methods of inference, showing how to account for parameter uncertainty under each of the two paradigms. We also describe extensions of the standard model for disease prevalence that can be used when stationarity of the spatio-temporal covariance function is not supported by the data. We discuss how to define predictive targets and argue that exceedance probabilities provide one of the most effective ways to convey uncertainty in prevalence estimates. We describe statistical software for the visualisation of spatio-temporal predictive summaries of prevalence through interactive animations. Finally, we illustrate an application to historical malaria prevalence data from 1 334 surveys conducted in Senegal between 1905 and 2014.

Summary In this paper, we set out general principles and develop geostatistical methods for the analysis of data from spatio-temporally referenced prevalence surveys. Our objective is to provide a tutorial guide that can be used in order to identify parsimonious geostatistical models for prevalence mapping. A general variogram-based Monte Carlo procedure is proposed to check the validity of the modelling assumptions. We describe and contrast likelihood-based and Bayesian methods of inference, showing how to account for parameter uncertainty under each of the two paradigms. We also describe extensions of the standard model for disease prevalence that can be used when stationarity of the spatio-temporal covariance function is not supported by the data. We discuss how to define predictive targets and argue that exceedance probabilities provide one of the most effective ways to convey uncertainty in prevalence estimates. We describe statistical software for the visualisation of spatio-temporal predictive summaries of prevalence through interactive animations. Finally, we illustrate an application to historical malaria prevalence data from 1 334 surveys conducted in Senegal between 1905 and 2014.

Background: Diarrhoea causes many deaths in children younger than 5 years and identification of risk factors for death is considered a global priority. The effectiveness of currently recommended fluid management for dehydration in routine settings has also not been examined. Methods: For this observational, association study, we analysed prospective clinical data on admission, immediate treatment, and discharge of children age 1-59 months with diarrhoea and dehydration, which were routinely collected from 13 Kenyan hospitals. We analysed participants with full datasets using multivariable mixed-effects logistic regression to assess risk factors for in-hospital death and effect of correct rehydration on early mortality (within 2 days). Findings: Between Oct 1, 2013, and Dec 1, 2016, 8562 children with diarrhoea and dehydration were admitted to hospital and eligible for inclusion in this analysis. Overall mortality was 9% (759 of 8562 participants) and case fatality was directly correlated with severity. Most children (7184 [84%] of 8562) with diarrhoea and dehydration had at least one additional diagnosis (comorbidity). Age of 12 months or younger (adjusted odds ratio [AOR] 1.71, 95% CI 1.42-2.06), female sex (1.41, 1.19-1.66), diarrhoea duration of more than 14 days (2.10, 1.42-3.12), abnormal respiratory signs (3.62, 2.95-4.44), abnormal circulatory signs (2.29, 1.89-2.77), pallor (2.15, 1.76-2.62), use of intravenous fluid (proxy for severity; 1.68, 1.41-2.00), and abnormal neurological signs (3.07, 2.54-3.70) were independently associated with in-hospital mortality across hospitals. Signs of dehydration alone were not associated with in-hospital deaths (AOR 1.08, 0.87-1.35). Correct fluid prescription significantly reduced the risk of early mortality (within 2 days) in all subgroups: abnormal respiratory signs (AOR 1.23, 0.68-2.24), abnormal circulatory signs (0.95, 0.53-1.73), pallor (1.70, 0.95-3.02), dehydration signs only (1.50, 0.79-2.88), and abnormal neurological signs (0.86, 0.51-1.48). Interpretation: Children at risk of in-hospital death are those with complex presentations rather than uncomplicated dehydration, and the prescription of recommended rehydration guidelines reduces risk of death. Strategies to optimise the delivery of recommended guidance should be accompanied by studies on the management of dehydration in children with comorbidities, the vulnerability of young girls, and the delivery of immediate care. Funding: The Wellcome Trust.

Background: Diarrhoea is a major cause of child mortality. Although oral rehydration solution (ORS) is an efficacious intervention for correcting dehydration, inadequate monitoring may limit its effectiveness in routine settings. We evaluated the effect of using a caregiver-administered chart to monitor oral fluid therapy on hydration status among children with some dehydration. Methods: An open-label randomized controlled trial was conducted among children 2-59 months of age. ORS fluid monitoring charts were given to caregivers in the intervention arm to record the hourly intake of ORS. ORS was administered without charting in the control arm. The primary outcome was dehydration defined by the presence of clinical signs of some dehydration, severe dehydration or shock assessed 4 h after initiation of treatment. We also assessed the acceptability of the charts among caregivers. Results: We evaluated 252 patients for the primary endpoint. Among those who received the intervention, 7/122 (5.7%) were still dehydrated following 4 h of ORS administration vs 20/130 (15.4%) in the control group (risk ratio 0.37 [95% confidence interval 0.16-0.85]). Caregivers in the intervention arm reported positive experiences using the fluid charts. Conclusions: The use of fluid monitoring charts reduced the frequency of dehydration and was well accepted by caregivers, representing a promising innovation for the management of diarrhoea and dehydration in resource-limited settings.

Background: Although thousands of clinical isolates of Plasmodium falciparum are being sequenced and analysed by short read technology, the data do not resolve the highly variable subtelomeric regions of the genomes that contain polymorphic gene families involved in immune evasion and pathogenesis. There is also no current standard definition of the boundaries of these variable subtelomeric regions. Methods: Using long-read sequence data (Pacific Biosciences SMRT technology), we assembled and annotated the genomes of 15 P. falciparum isolates, ten of which are newly cultured clinical isolates. We performed comparative analysis of the entire genome with particular emphasis on the subtelomeric regions and the internal var genes clusters. Results: The nearly complete sequence of these 15 isolates has enabled us to define a highly conserved core genome, to delineate the boundaries of the subtelomeric regions, and to compare these across isolates. We found highly structured variable regions in the genome. Some exported gene families purportedly involved in release of merozoites show copy number variation. As an example of ongoing genome evolution, we found a novel CLAG gene in six isolates. We also found a novel gene that was relatively enriched in the South East Asian isolates compared to those from Africa. Conclusions: These 15 manually curated new reference genome sequences with their nearly complete subtelomeric regions and fully assembled genes are an important new resource for the malaria research community. We report the overall conserved structure and pattern of important gene families and the more clearly defined subtelomeric regions.

Background: Although thousands of clinical isolates of Plasmodium falciparum are being sequenced and analysed by short read technology, the data do not resolve the highly variable subtelomeric regions of the genomes that contain polymorphic gene families involved in immune evasion and pathogenesis. There is also no current standard definition of the boundaries of these variable subtelomeric regions. Methods: Using long-read sequence data (Pacific Biosciences SMRT technology), we assembled and annotated the genomes of 15 P. falciparum isolates, ten of which are newly cultured clinical isolates. We performed comparative analysis of the entire genome with particular emphasis on the subtelomeric regions and the internal var genes clusters. Results: The nearly complete sequence of these 15 isolates has enabled us to define a highly conserved core genome, to delineate the boundaries of the subtelomeric regions, and to compare these across isolates. We found highly structured variable regions in the genome. Some exported gene families purportedly involved in release of merozoites show copy number variation. As an example of ongoing genome evolution, we found a novel CLAG gene in six isolates. We also found a novel gene that was relatively enriched in the South East Asian isolates compared to those from Africa. Conclusions: These 15 manually curated new reference genome sequences with their nearly complete subtelomeric regions and fully assembled genes are an important new resource for the malaria research community. We report the overall conserved structure and pattern of important gene families and the more clearly defined subtelomeric regions.

BACKGROUND: Severe acute malnutrition (SAM) remains a major cause of admission and inpatient mortality worldwide in children aged less than 5 years. In this study, we explored SAM prevalence and outcomes in children admitted in 13 Kenyan hospitals participating in a Clinical Information Network (CIN). We also describe their immediate in-patient management. METHODS: We analyzed data for children aged 1-59 months collected retrospectively from medical records after discharge. Mean, median and ranges were used to summarize pooled and age-specific prevalence and mortality associated with SAM. Documentation of key signs and symptoms (S/S) and performance of indicators of quality of care for selected aspects of the WHO management steps were assessed. Logistic regression models were used to evaluate associations between documented S/S and mortality among SAM patients aged 6-59 months. Loess curves were used to explore performance change over time for indicators of selected SAM management steps. RESULTS: 5306/54140 (9.8%) children aged 1-59 months admitted with medical conditions in CIN hospitals between December 2013 and November 2016 had SAM. SAM prevalence identified by clinicians and case fatality varied widely across hospitals with median proportion (range) of 10.1% (4.6-18.2%) and 14.8% (6.0-28.6%) respectively. Seventeen variables were associated with increased mortality. Performance change over time of management steps varied across hospitals and across selected indicators but suggests some effect of regular audit and feedback. CONCLUSION: Identification of SAM patients, their mortality and adherence to in-patient management recommendations varied across hospitals. An important group of SAM patients are aged less than 6 months. Continued efforts are required to improve management of SAM in routine settings as part of efforts to reduce inpatient mortality.

Background: The World Health Organization (WHO)/UNICEF Baby-Friendly Hospital Initiative step number five of the "Ten steps to successful breastfeeding" states "Show mothers how to breastfeed and how to maintain lactation even if they should be separated from their infants." Urban mothers in Nairobi have low rates of exclusive breastfeeding after returning to work but there are no published data on rural Kenya mothers' infant feeding practices when working or schooling away from home. Methods: We explored knowledge of, and attitudes to, the practice of giving expressed breastmilk in a mixed methods observational study of breastfeeding in rural Kenyan mothers. Fifty mothers with newborns, identified by nurses and community health workers, were asked questions about their experiences of breastfeeding and who they had sought or received advice from on breastfeeding. Focus group discussions, one with community health workers, and four each with mothers and their named advisers were held. Recordings were analyzed using a thematic framework approach. Results: The main themes were: the baby's right to feed from the breast, lack of knowledge about expressing and giving breastmilk, negative attitudes towards expressed breastmilk, and traditional customs for disposing of expressed breast milk. Most participants did not have any experience of giving expressed breastmilk to infants. They described practices of expressing and discarding milk when the mother or baby was ill, to relieve discomfort from engorgement or after the baby had died. Conclusions: Feeding expressed breastmilk to infants is a new concept in this context. Promotion of, and training in this practice would help mothers to maintain their milk supply when away from their babies and benefit the infants of working and schoolgirl mothers.

BACKGROUND: In sub-Saharan Africa, severely immunocompromised HIV-infected individuals have a high risk of mortality during the first few months after starting antiretroviral therapy (ART). We hypothesise that universally providing ready-to-use supplementary food (RUSF) would increase early weight gain, thereby reducing early mortality compared with current guidelines recommending ready-to-use therapeutic food (RUTF) for severely malnourished individuals only. METHODS: We did a 2 x 2 x 2 factorial, open-label, parallel-group trial at inpatient and outpatient facilities in eight urban or periurban regional hospitals in Kenya, Malawi, Uganda, and Zimbabwe. Eligible participants were ART-naive adults and children aged at least 5 years with confirmed HIV infection and a CD4 cell count of fewer than 100 cells per muL, who were initiating ART at the facilities. We randomly assigned participants (1:1) to initiate ART either with (RUSF) or without (no-RUSF) 12 weeks' of peanut-based RUSF containing 1000 kcal per day and micronutrients, given as two 92 g packets per day for adults and one packet (500 kcal per day) for children aged 5-12 years, regardless of nutritional status. In both groups, individuals received supplementation with RUTF only when severely malnourished (ie, body-mass index [BMI] <16-18 kg/m(2) or BMI-for-age Z scores <-3 for children). We did the randomisation with computer-generated, sequentially numbered tables with different block sizes incorporated within an online database. Randomisation was stratified by centre, age, and two other factorial randomisations, to 12 week adjunctive raltegravir and enhanced anti-infection prophylaxis (reported elsewhere). Clinic visits were scheduled at weeks 2, 4, 8, 12, 18, 24, 36, and 48, and included nurse assessment of vital status and symptoms and dispensing of all medication including ART and RUSF. The primary outcome was mortality at week 24, analysed by intention to treat. Secondary outcomes included absolute changes in weight, BMI, and mid-upper-arm circumference (MUAC). Safety was analysed in all randomly assigned participants. Follow-up was 48 weeks. This trial is registered with ClinicalTrials.gov (NCT01825031) and the ISRCTN registry (43622374). FINDINGS: Between June 18, 2013, and April 10, 2015, we randomly assigned 1805 participants to treatment: 897 to RUSF and 908 to no-RUSF. 56 (3%) were lost-to-follow-up. 96 (10.9%, 95% CI 9.0-13.1) participants allocated to RUSF and 92 (10.3%, 8.5-12.5) to no-RUSF died within 24 weeks (hazard ratio 1.05, 95% CI 0.79-1.40; log-rank p=0.75), with no evidence of interaction with the other randomisations (both p>0.7). Through 48 weeks, adults and adolescents aged 13 years and older in the RUSF group had significantly greater gains in weight, BMI, and MUAC than the no-RUSF group (p=0.004, 0.004, and 0.03, respectively). The most common type of serious adverse event was specific infections, occurring in 90 (10%) of 897 participants assigned RUSF and 87 (10%) of 908 assigned no-RUSF. By week 48, 205 participants had serious adverse events in both groups (p=0.81), and 181 had grade 4 adverse events in the RUSF group compared with 172 in the non-RUSF group (p=0.45). INTERPRETATION: In severely immunocompromised HIV-infected individuals, providing RUSF universally at ART initiation, compared with providing RUTF to severely malnourished individuals only, improved short-term weight gain but not mortality. A change in policy to provide nutritional supplementation to all severely immunocompromised HIV-infected individuals starting ART is therefore not warranted at present. FUNDING: Joint Global Health Trials Scheme (UK Medical Research Council, UK Department for International Development, and Wellcome Trust).

BACKGROUND: Scale-up of insecticide-based interventions has averted more than 500 million malaria cases since 2000. Increasing insecticide resistance could herald a rebound in disease and mortality. We aimed to investigate whether insecticide resistance was associated with loss of effectiveness of long-lasting insecticidal nets and increased malaria disease burden. METHODS: This WHO-coordinated, prospective, observational cohort study was done at 279 clusters (villages or groups of villages in which phenotypic resistance was measurable) in Benin, Cameroon, India, Kenya, and Sudan. Pyrethroid long-lasting insecticidal nets were the principal form of malaria vector control in all study areas; in Sudan this approach was supplemented by indoor residual spraying. Cohorts of children from randomly selected households in each cluster were recruited and followed up by community health workers to measure incidence of clinical malaria and prevalence of infection. Mosquitoes were assessed for susceptibility to pyrethroids using the standard WHO bioassay test. Country-specific results were combined using meta-analysis. FINDINGS: Between June 2, 2012, and Nov 4, 2016, 40 000 children were enrolled and assessed for clinical incidence during 1.4 million follow-up visits. 80 000 mosquitoes were assessed for insecticide resistance. Long-lasting insecticidal net users had lower infection prevalence (adjusted odds ratio [OR] 0.63, 95% CI 0.51-0.78) and disease incidence (adjusted rate ratio [RR] 0.62, 0.41-0.94) than did non-users across a range of resistance levels. We found no evidence of an association between insecticide resistance and infection prevalence (adjusted OR 0.86, 0.70-1.06) or incidence (adjusted RR 0.89, 0.72-1.10). Users of nets, although significantly better protected than non-users, were nevertheless subject to high malaria infection risk (ranging from an average incidence in net users of 0.023, [95% CI 0.016-0.033] per person-year in India, to 0.80 [0.65-0.97] per person year in Kenya; and an average infection prevalence in net users of 0.8% [0.5-1.3] in India to an average infection prevalence of 50.8% [43.4-58.2] in Benin). INTERPRETATION: Irrespective of resistance, populations in malaria endemic areas should continue to use long-lasting insecticidal nets to reduce their risk of infection. As nets provide only partial protection, the development of additional vector control tools should be prioritised to reduce the unacceptably high malaria burden. FUNDING: Bill & Melinda Gates Foundation, UK Medical Research Council, and UK Department for International Development.