Subjective modeling choices and the robustness of impact evaluations in conservation science.

Arbitrary modeling choices are inevitable in scientific studies. Yet, few empirical studies in conservation science report the effects these arbitrary choices have on estimated results. I explored the effects of subjective modeling choices in the context of counterfactual impact evaluations. Over 5000 candidate models based on reasonable changes in the choice of statistical matching algorithms (e.g., genetic and nearest distance mahalanobis matching), the parametrization of these algorithms (e.g., number of matches), and the inclusion of specific covariates (e.g., distance to nearest city, slope, or rainfall) were valid for studying the effect of Virunga National Park in Democratic Republic of the Congo on changes in tree cover loss and carbon storage over time. I randomly picked 2000 of the 5000 candidate models to determine how much and which subjective modeling choices affected the results the most. All valid models indicated that tree cover loss decreased and carbon storage increased in Virunga National Park from 2000 to 2019. Nonetheless, the order of magnitude of the estimates varied by a factor of 3 (from -4.78 to -13.12 percentage points decrease in tree cover loss and from 20 to 46 t Ce/ha for carbon storage). My results highlight that modeling choices, notably the choice of the matching algorithm, can have significant effects on point estimates and suggest that more structured robustness checks are a key step toward more credible findings in conservation science.

Selecciones Subjetivas de Modelos y la Contundencia de las Evaluaciones de Impacto en las Ciencias de la Conservación Resumen Las selecciones arbitrarias de modelos son inevitables en los estudios científicos. Sin embargo, pocos estudios empíricos en las ciencias de la conservación reportan los efectos de estas selecciones arbitrarias sobre los resultados estimados. Exploré los efectos de las selecciones subjetivas de modelos en el contexto de las evaluaciones de impacto contrafactuales. Más de 5000 modelos candidatos basados en cambios razonables en la elección de los algoritmos de emparejamiento estadístico (p. ej.: emparejamiento genético y de distancia más cercana mahalanobis), la parametrización de estos algoritmos (p. ej.: número de parejas) y la inclusión de covariados específicos (p. ej.: distancia a la ciudad más cercana, inclinación, precipitación) fueron válidos para estudiar el efecto del Parque Nacional Virunga en la República Democrática del Congo sobre la pérdida de cobertura arbórea y el almacenamiento de carbono a través del tiempo. Escogí al azar 2000 de los 5000 modelos candidatos para determinar cuántos y cuáles selecciones subjetivas de los modelos afectaron más al resultado. Todos los modelos válidos indicaron que la pérdida de la cobertura arbórea disminuyó y el almacenamiento de carbono incrementó en el Parque Nacional Virunga desde entre el año 2000 y el 2019. No obstante, el orden de magnitud de las estimaciones varió con un factor de 3 (una disminución de -4.78 hasta -13.12 puntos porcentuales en la pérdida de la cobertura arbórea y de 20 hasta 46 t Ce/ha para el almacenamiento de carbono). Mis resultados resaltan que la selección de los modelos, notablemente la elección del algoritmo de emparejamiento, puede tener efectos significativos sobre las estimaciones de puntos y sugieren que las revisiones más estructuradas de la contundencia son un paso importante hacia descubrimientos más creíbles en las ciencias de la conservación.

Onchocerciasis control in the Democratic Republic of Congo (DRC): challenges in a post-war environment.

OBJECTIVE: To evaluate onchocerciasis control activities in the Democratic Republic of Congo (DRC) in the first 12 years of community-directed treatment with ivermectin (CDTI). METHODS: Data from the National Programme for Onchocerciasis (NPO) provided by the National Onchocerciasis Task Force (NOTF) through the annual reports of the 21 CDTI projects for the years 2001-2012 were reviewed retrospectively. A hypothetical-inputs-process-outputs-outcomes table was constructed. RESULTS: Community-directed treatment with ivermectin expanded from 1968 communities in 2001 to 39 100 communities by 2012 while the number of community-directed distributors (CDD) and health workers (HW) multiplied. By 2012, there were ratios of 1 CDD per 262 persons and 1 HW per 2318 persons at risk. More than 80% of the funding came from the fiduciary funds of the African Programme for Onchocerciasis Control. The cost of treatment per person treated fell from US$ 1.1 in 2001 to US$ 0.1 in 2012. The therapeutic coverage increased from 2.7% (2001) to 74.2% (2012); the geographical coverage, from 4.7% (2001) to 93.9% (2012). Geographical coverage fell in 2005 due to deaths in loiasis co-endemic areas, and the therapeutic coverage fell in 2008 due to insecurity. CONCLUSIONS: Challenges to CDTI in DRC have been serious adverse reactions to ivermectin in loiasis co-endemic areas and political conflict. Targets for personnel or therapeutic and geographical coverages were not met. Longer term funding and renewed efforts are required to achieve control and elimination of onchocerciasis in DRC.

Community-supported models of care for people on HIV treatment in sub-Saharan Africa.

OBJECTIVES: Further scale-up of antiretroviral therapy (ART) to those in need while supporting the growing patient cohort on ART requires continuous adaptation of healthcare delivery models. We describe several approaches to manage stable patients on ART developed by Médecins Sans Frontières together with Ministries of Health in four countries in sub-Saharan Africa. METHODS: Using routine programme data, four approaches to simplify ART delivery for stable patients on ART were assessed from a patient and health system perspective: appointment spacing for clinical and drug refill visits in Malawi, peer educator-led ART refill groups in South Africa, community ART distribution points in DRC and patient-led community ART groups in Mozambique. RESULTS: All four approaches lightened the burden for both patients (reduced travel and lost income) and health system (reduced clinic attendance). Retention in care is high: 94% at 36 months in Malawi, 89% at 12 months in DRC, 97% at 40 months in South Africa and 92% at 48 months in Mozambique. Where evaluable, service provider costs are reported to be lower. CONCLUSION: Separating ART delivery from clinical assessments was found to benefit patients and programmes in a range of settings. The success of community ART models depends on sufficient and reliable support and resources, including a flexible and reliable drug supply, access to quality clinical management, a reliable monitoring system and a supported lay workers cadre. Such models require ongoing evaluation and further adaptation to be able to reach out to more patients, including specific groups who may be challenged to meet the demands of frequent clinic visits and the integrated delivery of other essential chronic disease interventions.

Effectiveness of blood transfusions and risk factors for mortality in children aged from 1 month to 4 years at the Bon Marché Hospital, Bunia, Democratic Republic of the Congo.

OBJECTIVE: To assess the effectiveness of blood transfusions in a hospital of north-eastern Democratic Republic of the Congo. METHODS: Prospective study of children admitted for severe anaemia. During admission, data were collected on clinical condition and haemoglobin levels, before and after blood transfusion. A linear regression model was built to explore factors associated with haemoglobin level after transfusion. Risk factors for mortality were explored through multivariate logistic regression. RESULTS: Haemoglobin level (Hb) was below 4 g/dl in 35% (230/657), between 4 and 6 g/dl in 58% (348/657) and at least 6 g/dl in another 6% (43/657) of the transfused children. A transfusion of 15 ml/kg of whole blood increased the Hb from 4.4 to 7.8 g/dl. Haemoglobin level after transfusion was associated with baseline Hb, quantity of delivered blood and history of previous transfusions. Overall case-fatality rate was 5.6% (37/657). Risk factors for deaths were co-morbidities such as chest infection, meningitis or malnutrition, Hb ≥ 6 g/dl, impaired consciousness or jugular venous distention on admission, and provenance. CONCLUSION: Transfusion was a frequent practice, the use of which could clearly have been rationalised. While indications should be restricted, quantities of transfused blood should be adapted to needs.