Decision-model estimation of the age-specific disability weight for schistosomiasis japonica: a systematic review of the literature

Abstract

Schistosomiasis is among the most prevalent parasitic infections worldwide. However, current Global Burden of Disease (GBD) disability-adjusted life year estimates indicate that its population-level impact is negligible. Recent studies suggest that GBD methodologies may significantly underestimate the burden of parasitic diseases, including schistosomiasis. Furthermore, strain-specific disability weights have not been established for schistosomiasis, and the magnitude of human disease burden due to Schistosoma japonicum remains controversial. We used a decision model to quantify an alternative disability weight estimate of the burden of human disease due to S. japonicum. We reviewed S. japonicum morbidity data, and constructed decision trees for all infected persons and two age-specific strata, <15 years (y) and > or =15 y. We conducted stochastic and probabilistic sensitivity analyses for each model. Infection with S. japonicum was associated with an average disability weight of 0.132, with age-specific disability weights of 0.098 (<15 y) and 0.186 (> or =15 y). Re-estimated disability weights were seven to 46 times greater than current GBD measures; no simulations produced disability weight estimates lower than 0.009. Nutritional morbidities had the greatest contribution to the S. japonicum disability weight in the <15 y model, whereas major organ pathologies were the most critical variables in the older age group. GBD disability weights for schistosomiasis urgently need to be revised, and species-specific disability weights should be established. Even a marginal increase in current estimates would result in a substantial rise in the estimated global burden of schistosomiasis, and have considerable implications for public health prioritization and resource allocation for schistosomiasis research, monitoring, and control.

Figure 1. Literature search strategy.

Shown is a diagrammatic representation of the retrieval strategy used for identifying and selecting studies for inclusion in the final analysis.

Figure 2. Schematic representation of the model.

Shown is the branch of the model depicting liver pathology, which may or may not be present. If present, there may be hepatomegaly of varying degrees. Regardless of the degree of hepatomegaly, fibrosis may exist. Cirrhosis could only occur when fibrosis was present. Other comorbidities did not depend on the presence of other conditions.

Figure 3. One-way sensitivity analysis.

The horizontal bars depict the effect of re-evaluating the disability weight (shown on the X-axis) after changing the value of the specified parameter from the low to the high end of its range. The number of parameters that resulted in the greatest variation are shown.

Figure 4. Probabilistic sensitivity analyses.

Boxplots depicting the results of 5,000 simulation Monte Carlo analysis for each age-group model. Boxes represent the median, 25^th and 75^th percentiles, and error bars extend to the 2.5^th and 97.5^th percentile. Means are depicted by the circles.