Development and external validation of a clinical prognostic score for death in visceral leishmaniasis patients in an area of Ethiopiawith a high burden of HIV co-infection
*Charles Abongomera1,2,Koert Ritmeijer3,Florian Vogt2, Jozefien Buyze2, Zelalem Mekonnen1,Henok Admassu1, Robert Colebunders2, Rezika Mohammed4, Lutgarde Lynen2, Ermias Diro4,Johan van Griensven2
1Médecins Sans Frontières (MSF), Abdurafi, Ethiopia; 2Institute of Tropical Medicine, Antwerp, Belgium; 3MSF, Amsterdam, Netherlands;4University of Gondar, Gondar, Ethiopia
*
Introduction
In Ethiopia, case fatality rates among subgroups of visceral leishmaniasis (VL) patients are high. No validated clinical prognostic score is available to predict death in clinical practice or VL programmes. The optimal management and classification of VL severity remains poorly defined, and is highly variable across physicians and treatment sites. The aim of this study was to identify predictors of death from VL, and to develop and externally validate a clinical prognostic score for death in VL patients, in a high HIV co-infection burden area in Ethiopia.
Methods
We conducted a retrospective cohort study in northwest Ethiopia.Data on potential predictors were collected at admission to the health facility. The whole derivation data set was used to develop the score. Five-fold cross validation and external validation were performed to evaluate the performance of the score. Predictors with an adjusted likelihood ratio ≥1.5 or ≤0.67 were retained to calculate the predictor score.
Ethics
This study was approved by the following Ethics Review Boards (ERBs): the Institute of Tropical Medicine, Antwerp, Belgium; and the Institute of Public Health, Gondar University, Ethiopia. This research fulfilled the exemption criteria set by the MSF ERB for a posteriori analyses of routinely collected clinical data and thus did not require MSF ERB review. It was conducted with permission from Sidney Wong, Operational Centre Amsterdam, MSF.
Results
The derivation cohort consisted of 1686 VL patients treated at an upgraded health centre, and the external validation cohort consisted of 404 VL patients treated in a hospital. There were 99 deaths in the derivation cohort and 53 in the external validation cohort. Predictors of death were: age >40 years (score +1); HIV seropositive (score +1); HIV seronegative (score -1); haemoglobin <6.5 g/dL (score +1); bleeding (score +1); jaundice (score +1); oedema (score +1); ascites (score +2), and tuberculosis (score +1). The total prognostic score per patient ranged from -1 to +5. A score of -1indicated a low risk of death (1.0%), a score of 0 an intermediate risk (3.8%), and a score of +1 to +5 a high risk (10.4-85.7%). In the derivation cohort, 873 patients (51.8%) had a low risk of death, 369 (21.9%) an intermediate risk, and 444 (26.3%) a high risk.The area under the receiver operating characteristic curve was 0.83 (95%CI 0.79-0.87) in derivation, and 0.78 (95%CI 0.72-0.83) in external validation.
Conclusion
The scoring system showed good performance in identifying VL patients at high risk of death. Classification of patients according to their prognosis for death could help reduce case fatality rates by enabling them to be triaged towards a unit with the highest level of care. However, more evidence is needed on the impact of the score when applied for such strategies. Alimitation of this study is its retrospective design, as we could study predictors from only the collected variables.
Conflicts of interest
None declared.