Periodic Reporting for period 1 - Dynamic CER (Dynamic Comparative Effectiveness Research for health care interventions)
Reporting period: 2021-11-01 to 2023-10-31
This fellowship was aimed to provide two main contributions to the current state-of-the-art in comparative effectiveness research (CER) of healthcare interventions, with a special focus on so-called complex interventions (typical examples of complex interventions include psychological interventions for mental illness or non-pharmacological interventions to support behavioural change in digital health). First, the fellowship investigated dynamic regimes of complex just-in-time adaptive interventions (JITAIs) in primary research. Specifically, novel use of data from JITAIs to dynamically evaluate response to treatment are proposed. Second, the fellowship contributed to the field of evidence synthesis by developing novel methodology for both complex and non-complex interventions, as well as software to dynamically update network meta-analyses results in a user-friendly and timely manner. All this addresses important limitations of current CER methodologies, which are to date not well developed to take into account the temporal evolution of treatment effects. In turn, this can also enable a more effective and precise health decision and policy making in the near future.
Overview of main results achieved:
- novel statistical methodologies for performing outlier-detection and disentangle effects of components of complex interventions in network meta-analyses were developed.
- a novel software tool was developed, in the form of a flexible and user-friendly web-application and python package, freely-available in open-source.
- a novel methodology for predicting the optimal and personalised timing of treatment in mobile health (e.g. when to send a notification/medication reminder to promote physical activity via wearable technologies) is in late-stage development.
All methods are accompanied by open-source software, thus facilitating their use from other researchers in the field. Results were disseminated at major international conferences in the field of biostatistics, including the International Society for Clinical Biostatistics conference and the Cochrane Colloquium.
- novel statistical methodologies for performing outlier-detection and disentangle effects of components of complex interventions in network meta-analyses were developed.
- a novel software tool was developed, in the form of a flexible and user-friendly web-application and python package, freely-available in open-source.
- a novel methodology for predicting the optimal and personalised timing of treatment in mobile health (e.g. when to send a notification/medication reminder to promote physical activity via wearable technologies) is in late-stage development.
All methods are accompanied by open-source software, thus facilitating their use from other researchers in the field. Results were disseminated at major international conferences in the field of biostatistics, including the International Society for Clinical Biostatistics conference and the Cochrane Colloquium.
The output of the fellowship with more direct and wider societal implications is represented by the novel open-source software for producing and updating network meta-analyses in a dynamic fashion. The software eases presentation of all the available evidence and interpretation of findings from large networks of interventions and is well-suited to continually synthesise new evidence. This has direct potentials to impact clinical decision-making and be used by public health stakeholders for health policy-making. The software aids to reduce needs on resources, time needed for analysis and allows for results to be continuously updated, thus improving up-to-date evidence-based decisions. In addition, the method developed to detect outlying studies in network meta-analysis can also serve as a robust tool to allow for less-biased evidence-based decisions.