Development and use of the Lives Saved Tool: a model to estimate the impact of scaling up proven interventions on maternal, neonatal and child mortality

Robert Steinglass et al.1 highlight that ‘how’ the Lives Saved Tool (LiST) gets used is a critical issue and we fully agree. LiST is to be used to assess the potential mortality impact of improving the coverage of interventions whether they are long-standing or new ones. They are not correct in saying that it is primarily being used to advocate for new or low-coverage interventions. It estimates the additional effects of this increased coverage giving first priority to preventive interventions, such as immunizations. What Steinglass et al. are calling ‘advocacy’ we understand as promoting evidence-based decision making for planning the appropriate expansion of interventions. Of course, interventions that are already at high coverage will have little marginal benefit with small changes in implementation, but that does not reduce their importance if they are still addressing sizable causes of death. In fact, achievement of high coverage of interventions like immunization increases the plausibility of rapid scale-up of other vaccines or interventions that can be added to the immunization platform, and these effects can be modelled in LiST. In relation to the past mortality reduction associated with established interventions, we have clearly stated that one of the criteria used for the inclusion of the interventions in LiST was to incorporate those that are not likely to have an impact on mortality because of its current high coverage, but that would probably result in an increase in mortality where coverage levels are not sustained (e.g. measles and Diphtheria, Pertussis and Tetanus (DPT) vaccines).2 Moreover, we believe that the two figures on ‘estimated lives saved’ and ‘estimated lives lost’ are misleading. The latter shows that lives lost reduces from 97 to 0% in 1 year, while the figure on lives saved presents a slow increase, from 97 to 100% in 5 years. It is unrealistic to model a drop from 97 to 0% coverage of any well-established intervention especially immunizations. In the current version of LiST, if users do lower coverage of an intervention, there is an explicit warning that this should not be done, although the users can ignore the warning and continue with their model run. Users can obviously make use of the tool in both ways: looking at future and past gains as Steinglass et al. have done. If a lower coverage for a preventive intervention is entered, LiST still gives priority to preventive interventions but some of the possible deaths that have not been prevented are subsequently averted by therapeutic interventions. There is no question that successful interventions that have already reached high coverage and saved an important number of lives not only need to at least maintain but also increase their levels of coverage. However, there are still about 9 million children under-5 years, including almost 4 million babies <28 days of life, dying every year, even with the high-level coverage of such interventions. There is an imperative need to focus on what more can be done to save these millions of lives. For example, it is urgent to set clear and achievable targets, estimate the likely impact of different interventions and delivery channels, implement effective strategies and monitor progress of their achievement, according to the needs and availability of interventions in different settings. There is need for flexibility and specificity; not for the one-size-fits-all models or packages. This is what LiST is designed to facilitate. Consideration of the costs of interventions is also important. As stated in at least two papers included in the series,2,3 work is underway to include costing modules in the tool, which will allow for the estimates of cost-effective interventions as well as for the costs of increased coverage of specific interventions in different scenarios. In fact, in the current version of LiST one can output data files that can be read into the Marginal Budgeting for Bottlenecks software (MBB) that allows one to link the coverage and deaths averted estimates from LiST to MBB for costing and bottleneck analysis.