Fig 1.
Overall decision-tree model structure.
Squares designate decision points. Circles designate probabilistic events. Triangles designate terminal nodes. Ellipses (…) indicate a symmetrical sub-tree that is not shown due to space constraints.
Fig 2.
Influenza sub-tree structures.
(A) Series of events that determine the monetary and utility costs of influenza infection during pregnancy. (B) Series of events that determine the monetary and utility costs of influenza infection during the first 6 months of life.
Table 1.
Baseline parameters and ranges used in the model.
Fig 3.
Changes to influenza sub-trees when accounting for decreased access to care.
(A) Changes to the sub-tree for influenza infection during pregnancy. (B) Changes to the sub-tree for influenza infection during the first 6 months of life.
Table 2.
Costs of laboratory-confirmed influenza (LCI) and influenza-like illness (ILI) episodes in stratified by population and vaccination status in the setting of a maternal influenza vaccine trial in Bamako, Mali.
Table 3.
Changes in the cost-effectiveness ratio of maternal influenza immunization from the baseline from limited access to healthcare, changing the attack rate of influenza and the vaccine efficacy to rates reported in other randomized trials, and adjusting the costs of illness based on per-capita healthcare spending.
Fig 4.
One-way sensitivity analysis of the base case.
Each input variable is evaluated on a separate curve. The X-axis shows the percent change in the input variable from its baseline value (set at 100%), and the Y-axis shows the cost per DALY saved (left) and the relative change in the cost-effectiveness ratio from its initial value (right). Moving down in the Y-axis indicates a lower cost per DALY, i.e. a more efficient intervention. The greater the slope of each curve, the more sensitive the baseline model is to changes in that variable. The 7 most sensitive variables are depicted here.
Fig 5.
Regression tree analysis of 10,000 Monte Carlo simulations sampling across the uncertainty in all parameters.
The leaves of the tree end in ovals that show the mean cost per DALY of maternal influenza vaccine among all simulations whose parameters follow the paths described in the proximal branches. Further partitioning of this pruned, cross-validated tree did not reduce generalization error. Parameters affecting infant influenza are shown in shades of blue, while parameters affecting influenza in women are shown in purple. Cost-effectiveness ratios < 1x per-capita GDP in Mali are shown in green, those between 1x and 3x per-capita GDP in Mali are shown in yellow, those >3x per-capita GDP in Mali are shown in red. LCI: Lab-confirmed influenza; CFR: case-fatality ratio; VE: vaccine efficacy.
Fig 6.
Cost (US$) per DALY saved of a maternal influenza immunization program varied by cost of vaccination program per pregnant woman vaccinated.
(A) Baseline model. (B) Comparison across 4 scenarios: Baseline, Poor Access to Care, Severe Disease, and Severe Disease + Poor Access to Care.
Fig 7.
Effect of the cost of a maternal immunization program on the probability that the program will be cost-effective.
Probability that the cost per DALY saved < Mali GDP per capita by programmatic costs per pregnant woman vaccinated across 4 scenarios: Baseline, Poor Access to Care, Severe Disease, and Severe Disease + Poor Access to Care.
Fig 8.
Impact of changing attack rates of influenza during pregnancy and in the first 5 months of life on the cost-effectiveness of maternal influenza immunization.
AR: Attack Rate.