Conference Material > Slide Presentation
Wardley T, West KP, Tesfay B, Robinson N, Parry L, et al.
MSF Scientific Day International 2024. 16 May 2024; DOI:10.57740/EQ5OG2MuMi
Conference Material > Abstract
Wardley T, West KP, Tesfay B, Robinson N, Parry L, et al.
MSF Scientific Day International 2024. 16 May 2024; DOI:10.57740/a4TlzIISm
INTRODUCTION
Climate and environmental conditions are critical factors in malaria transmission. Médecins Sans Frontières (MSF) teams in South Sudan have seen changes in the timing and intensity of malaria seasonal peaks over the past decade. The Malaria Anticipation Project (MAP) aims to develop predictive early warning systems to better predict and act upon any expected rise in malaria cases, through routine surveillance.
METHODS
Predictive models were developed using environmental data collected from climate and space agencies and weekly outpatient department (OPD) malaria case count in Lankien hospital (Jonglei State, South Sudan) as the epidemiological input and output. An ensemble modelling approach was developed using linear regression and extreme gradient boosting (XGBoost) models in a recursive modelling framework. The models were developed using data from 2012–2020, verified with data from 2020–2022, and then monitored in real time in the 2022/23 season. To assess model performance, observed OPD malaria cases were compared with the monthly average cases and classified into categories to assess how often the model prediction was in the same category as the observed number of cases. We
also conducted a qualitative survey to explore community understanding of malaria and its relationship to climate.
RESULTS
During model development, the predictive performance was very high at 2 weeks’ lead time (75% classification accuracy). Model performance remained satisfactory at up to 8 weeks’ lead time (70% classification accuracy), while beyond this, it became increasingly susceptible to large prediction errors. In the 2020/21 and 2021/22 malaria seasons, the predictive performance at 2 weeks’ lead time was good, but it overpredicted for both seasons at 4–8 weeks. The 2022–23 season saw the lowest number of malaria cases of any year in the data used to train the model. The models predicted that the number of cases would be below the long-term average for Lankien hospital, but overpredicted the burden. Across all models, the shorter the lead time of the models, the greater their predictive performance.
CONCLUSION
This modelling approach has the potential to inform anticipatory action within an operationally useful timeframe. Given the models are trained on historical data and cannot include all factors affecting malaria transmission, if relationships between malaria and other conditions change over time, this will impact model performance, demonstrating the limits of forecasting approaches. The next stage of the MAP project will focus on replicability in other settings and pilot implementation to understand operational feasibility and improve performance.
Climate and environmental conditions are critical factors in malaria transmission. Médecins Sans Frontières (MSF) teams in South Sudan have seen changes in the timing and intensity of malaria seasonal peaks over the past decade. The Malaria Anticipation Project (MAP) aims to develop predictive early warning systems to better predict and act upon any expected rise in malaria cases, through routine surveillance.
METHODS
Predictive models were developed using environmental data collected from climate and space agencies and weekly outpatient department (OPD) malaria case count in Lankien hospital (Jonglei State, South Sudan) as the epidemiological input and output. An ensemble modelling approach was developed using linear regression and extreme gradient boosting (XGBoost) models in a recursive modelling framework. The models were developed using data from 2012–2020, verified with data from 2020–2022, and then monitored in real time in the 2022/23 season. To assess model performance, observed OPD malaria cases were compared with the monthly average cases and classified into categories to assess how often the model prediction was in the same category as the observed number of cases. We
also conducted a qualitative survey to explore community understanding of malaria and its relationship to climate.
RESULTS
During model development, the predictive performance was very high at 2 weeks’ lead time (75% classification accuracy). Model performance remained satisfactory at up to 8 weeks’ lead time (70% classification accuracy), while beyond this, it became increasingly susceptible to large prediction errors. In the 2020/21 and 2021/22 malaria seasons, the predictive performance at 2 weeks’ lead time was good, but it overpredicted for both seasons at 4–8 weeks. The 2022–23 season saw the lowest number of malaria cases of any year in the data used to train the model. The models predicted that the number of cases would be below the long-term average for Lankien hospital, but overpredicted the burden. Across all models, the shorter the lead time of the models, the greater their predictive performance.
CONCLUSION
This modelling approach has the potential to inform anticipatory action within an operationally useful timeframe. Given the models are trained on historical data and cannot include all factors affecting malaria transmission, if relationships between malaria and other conditions change over time, this will impact model performance, demonstrating the limits of forecasting approaches. The next stage of the MAP project will focus on replicability in other settings and pilot implementation to understand operational feasibility and improve performance.
Conference Material > Slide Presentation
Tremblay LL, Wardley T, Tesfay B, Galban-Horcajo F, West KP, et al.
MSF Scientific Day International 2023. 7 June 2023; DOI:10.57740/9wsa-v278
Journal Article > ResearchFull Text
Public Health Nutr. 1 June 2023; Volume 26 (Issue 6); 1210-1221.; DOI:10.1017/S1368980023000149
Briend A, Myatt M, Berkley JA, Black RE, Boyd EM, et al.
Public Health Nutr. 1 June 2023; Volume 26 (Issue 6); 1210-1221.; DOI:10.1017/S1368980023000149
OBJECTIVE
To compare the prognostic value of mid-upper arm circumference (MUAC), weight-for-height Z-score (WHZ) and weight-for-age Z-score (WAZ) for predicting death over periods of 1, 3 and 6 months follow-up in children.
DESIGN
Pooled analysis of twelve prospective studies examining survival after anthropometric assessment. Sensitivity and false-positive ratios to predict death within 1, 3 and 6 months were compared for three individual anthropometric indices and their combinations.
SETTING
Community-based, prospective studies from twelve countries in Africa and Asia.
PARTICIPANTS
Children aged 6–59 months living in the study areas.
RESULTS
For all anthropometric indices, the receiver operating characteristic curves were higher for shorter than for longer durations of follow-up. Sensitivity was higher for death with 1-month follow-up compared with 6 months by 49 % (95 % CI (30, 69)) for MUAC < 115 mm (P < 0·001), 48 % (95 % CI (9·4, 87)) for WHZ < -3 (P < 0·01) and 28 % (95 % CI (7·6, 42)) for WAZ < -3 (P < 0·005). This was accompanied by an increase in false positives of only 3 % or less. For all durations of follow-up, WAZ < -3 identified more children who died and were not identified by WHZ < -3 or by MUAC < 115 mm, 120 mm or 125 mm, but the use of WAZ < -3 led to an increased false-positive ratio up to 16·4 % (95 % CI (12·0, 20·9)) compared with 3·5 % (95 % CI (0·4, 6·5)) for MUAC < 115 mm alone.
CONCLUSIONS
Frequent anthropometric measurements significantly improve the identification of malnourished children with a high risk of death without markedly increasing false positives. Combining two indices increases sensitivity but also increases false positives among children meeting case definitions.
To compare the prognostic value of mid-upper arm circumference (MUAC), weight-for-height Z-score (WHZ) and weight-for-age Z-score (WAZ) for predicting death over periods of 1, 3 and 6 months follow-up in children.
DESIGN
Pooled analysis of twelve prospective studies examining survival after anthropometric assessment. Sensitivity and false-positive ratios to predict death within 1, 3 and 6 months were compared for three individual anthropometric indices and their combinations.
SETTING
Community-based, prospective studies from twelve countries in Africa and Asia.
PARTICIPANTS
Children aged 6–59 months living in the study areas.
RESULTS
For all anthropometric indices, the receiver operating characteristic curves were higher for shorter than for longer durations of follow-up. Sensitivity was higher for death with 1-month follow-up compared with 6 months by 49 % (95 % CI (30, 69)) for MUAC < 115 mm (P < 0·001), 48 % (95 % CI (9·4, 87)) for WHZ < -3 (P < 0·01) and 28 % (95 % CI (7·6, 42)) for WAZ < -3 (P < 0·005). This was accompanied by an increase in false positives of only 3 % or less. For all durations of follow-up, WAZ < -3 identified more children who died and were not identified by WHZ < -3 or by MUAC < 115 mm, 120 mm or 125 mm, but the use of WAZ < -3 led to an increased false-positive ratio up to 16·4 % (95 % CI (12·0, 20·9)) compared with 3·5 % (95 % CI (0·4, 6·5)) for MUAC < 115 mm alone.
CONCLUSIONS
Frequent anthropometric measurements significantly improve the identification of malnourished children with a high risk of death without markedly increasing false positives. Combining two indices increases sensitivity but also increases false positives among children meeting case definitions.
Journal Article > ResearchFull Text
Public Health Nutr. 3 February 2023; Volume 26 (Issue 4); 803-819.; DOI:10.1017/S136898002300023X
Khara T, Myatt M, Sadler K, Bahwere P, Berkley JA, et al.
Public Health Nutr. 3 February 2023; Volume 26 (Issue 4); 803-819.; DOI:10.1017/S136898002300023X
OBJECTIVE
To understand which anthropometric diagnostic criteria best discriminate higher from lower risk of death in children and explore programme implications.
DESIGN
A multiple cohort individual data meta-analysis of mortality risk (within 6 months of measurement) by anthropometric case definitions. Sensitivity, specificity, informedness and inclusivity in predicting mortality, face validity and compatibility with current standards and practice were assessed and operational consequences were modelled.
SETTING
Community-based cohort studies in twelve low-income countries between 1977 and 2013 in settings where treatment of wasting was not widespread.
PARTICIPANTS
Children aged 6 to 59 months.
RESULTS
Of the twelve anthropometric case definitions examined, four (weight-for-age Z-score (WAZ) <−2), (mid-upper arm circumference (MUAC) <125 mm), (MUAC < 115 mm or WAZ < −3) and (WAZ < −3) had the highest informedness in predicting mortality. A combined case definition (MUAC < 115 mm or WAZ < −3) was better at predicting deaths associated with weight-for-height Z-score <−3 and concurrent wasting and stunting (WaSt) than the single WAZ < −3 case definition. After the assessment of all criteria, the combined case definition performed best. The simulated workload for programmes admitting based on MUAC < 115 mm or WAZ < −3, when adjusted with a proxy for required intensity and/or duration of treatment, was 1·87 times larger than programmes admitting on MUAC < 115 mm alone.
CONCLUSIONS
A combined case definition detects nearly all deaths associated with severe anthropometric deficits suggesting that therapeutic feeding programmes may achieve higher impact (prevent mortality and improve coverage) by using it. There remain operational questions to examine further before wide-scale adoption can be recommended.
To understand which anthropometric diagnostic criteria best discriminate higher from lower risk of death in children and explore programme implications.
DESIGN
A multiple cohort individual data meta-analysis of mortality risk (within 6 months of measurement) by anthropometric case definitions. Sensitivity, specificity, informedness and inclusivity in predicting mortality, face validity and compatibility with current standards and practice were assessed and operational consequences were modelled.
SETTING
Community-based cohort studies in twelve low-income countries between 1977 and 2013 in settings where treatment of wasting was not widespread.
PARTICIPANTS
Children aged 6 to 59 months.
RESULTS
Of the twelve anthropometric case definitions examined, four (weight-for-age Z-score (WAZ) <−2), (mid-upper arm circumference (MUAC) <125 mm), (MUAC < 115 mm or WAZ < −3) and (WAZ < −3) had the highest informedness in predicting mortality. A combined case definition (MUAC < 115 mm or WAZ < −3) was better at predicting deaths associated with weight-for-height Z-score <−3 and concurrent wasting and stunting (WaSt) than the single WAZ < −3 case definition. After the assessment of all criteria, the combined case definition performed best. The simulated workload for programmes admitting based on MUAC < 115 mm or WAZ < −3, when adjusted with a proxy for required intensity and/or duration of treatment, was 1·87 times larger than programmes admitting on MUAC < 115 mm alone.
CONCLUSIONS
A combined case definition detects nearly all deaths associated with severe anthropometric deficits suggesting that therapeutic feeding programmes may achieve higher impact (prevent mortality and improve coverage) by using it. There remain operational questions to examine further before wide-scale adoption can be recommended.
Journal Article > ResearchFull Text
Health Policy Plan. 7 November 2019
Elston JWT, Danis K, Gray NSB, West H, West KP, et al.
Health Policy Plan. 7 November 2019
Sierra Leone has the world’s highest estimated maternal mortality. Following the 2014–16 Ebola outbreak, we described health outcomes and health-seeking behaviour amongst pregnant women to inform health policy. In October 2016–January 2017, we conducted a sequential mixed-methods study in urban and rural areas of Tonkolili District comprising: household survey targeting women who had given birth since onset of the Ebola outbreak; structured interviews at rural sites investigating maternal deaths and reporting; and in-depth interviews (IDIs) targeting mothers, community leaders and health workers. We selected 30 clusters in each area: by random GPS points (urban) and by random village selection stratified by population size (rural). We collected data on health-seeking behaviours, barriers to healthcare, childbirth and outcomes using structured questionnaires. IDIs exploring topics identified through the survey were conducted with a purposive sample and analysed thematically. We surveyed 608 women and conducted 29 structured and 72 IDIs. Barriers, including costs of healthcare and physical inaccessibility of healthcare facilities, delayed or prevented 90% [95% confidence interval (CI): 80–95] (rural) vs 59% (95% CI: 48–68) (urban) pregnant women from receiving healthcare. Despite a general preference for biomedical care, 48% of rural and 31% of urban women gave birth outside of a health facility; of those, just 4% and 34%, respectively received skilled assistance. Women expressed mistrust of healthcare workers (HCWs) primarily due to payment demanded for ‘free’ healthcare. HCWs described lack of pay and poor conditions precluding provision of quality care. Twenty percent of women reported labour complications. Twenty-eight percent of villages had materials to record maternal deaths. Pregnant women faced important barriers to care, particularly in rural areas, leading to high preventable mortality and morbidity. Women wanted to access healthcare, but services available were often costly, unreachable and poor quality. We recommend urgent interventions, including health promotion, free healthcare access and strengthening rural services to address barriers to maternal healthcare.
Protocol > Research Protocol
West KP, Greig J, Lokuge K, Caleo GNC, Stringer B, et al.
1 July 2018
Aim: To reduce suffering, morbidity and mortality by containing and reducing the spread of Ebola Virus Disease (EVD), while preserving human dignity for the affected population in Sierra Leone.
Purpose: To reduce and ultimately eliminate the transmission of EVD in a defined catchment population in Freetown.
Objectives:
• Provide epidemiological technical support to intensify surveillance, supervision of the alert response and enhanced case investigation in the defined area.
• Assess and respond to current gaps in infection prevention and control, water and sanitation, and triage in health facilities within the defined area.
• Assess community social mobilisation, health promotion, contact tracing and quarantine interventions in the defined area and respond to any gaps through advocacy towards the relevant pillar/organization and/or through direct MSF intervention.
• Prioritise MSF and health staff safety & biosecurity at all times
• Medical (non-Ebola) and humanitarian needs of the population are monitored, recorded, analysed and responded to through advocacy or MSF action.
Purpose: To reduce and ultimately eliminate the transmission of EVD in a defined catchment population in Freetown.
Objectives:
• Provide epidemiological technical support to intensify surveillance, supervision of the alert response and enhanced case investigation in the defined area.
• Assess and respond to current gaps in infection prevention and control, water and sanitation, and triage in health facilities within the defined area.
• Assess community social mobilisation, health promotion, contact tracing and quarantine interventions in the defined area and respond to any gaps through advocacy towards the relevant pillar/organization and/or through direct MSF intervention.
• Prioritise MSF and health staff safety & biosecurity at all times
• Medical (non-Ebola) and humanitarian needs of the population are monitored, recorded, analysed and responded to through advocacy or MSF action.