NorthWest Biotherapeutics Inc (NWBO)

[flipper44]

That passage does not state what you think it states.

Try this passage.

As described in previous sections, the approach to incorporating external control data is a multistep process (see Figure 3), where the first efforts must be to define study purpose and determining whether an external control data source is fit for purpose. This process would include considerations of data quality, data comprehensiveness and completeness, and comparability to a potential experimental arm with respect to characteristics such as (but not limited to) underlying patient population, temporality, and key data ascertainment. Major barriers to establishing an appropriate external control data source that is fit for purpose in oncology often include challenges with non-comparable endpoints, differences in selection by or availability of information regarding biomarkers, and poor capture of information that may be generally prognostic such as standard of care or supportive treatment. Importantly, the granularity of the data with respect to measurement or categorization of important variables is not trivial, and must be similar between experimental arm and external control data to ensure comparative analyses.

Furthermore, the choice of external control comparator must be rigorously supported by the pre-specification of a detailed protocol and robust statistical analysis plan, including a comprehensive plan to address sources of bias. Bias, including confounding, selection bias, and survivor or lead-time bias, cannot be totally eliminated in any non-randomized comparison. However, as previously mentioned, study design elements can help to minimize bias, and statistical methods may address the influence of bias on the estimation of treatment effect. However, if there are major issues with the validity or accuracy of the data, such as inability to establish audit trails or concerns of measurement error and misclassification, or substantial concerns of bias, statistical methods cannot salvage the comparison. Development of additional methods and bias quantification are needed to appropriately characterize use of external control arm as substantial evidence in regulatory applications.