All findings must be considered in the context of potential limitations related to the research process itself. The most important ones are summarised below.
Work package 1
- Owing to the nature of the emergency pathway developed, specific training was required for the intervention paramedics, but uptake was approximately 55% (i.e. 453/817). This self-selection may have influenced intervention delivery and the views expressed during the process evaluation. It was not possible within the remit of the programme to explore any implications for interpretation of the results or to understand the reasons for the reluctance to train.
- Challenging operational conditions impeded objective confirmation of intervention fidelity and approximately half of the study ambulance data forms were not returned, despite efforts to encourage completion. It is unclear whether these forms were completed and accidentally lost within clinical services or not completed, and reasons for the possibility of them not being completed have not been explored.
- As the PASTA intervention sought to directly influence clinician behaviour, it was not possible to mask group allocation. However, the lack of imbalance in baseline characteristics makes selection bias an unlikely explanation for the results.
- It is important to recognise that the post hoc association found between the PASTA intervention and specialist availability was hypothesis-generating and mechanisms remain unclear for any influence on treatment decisions, health and economic outcomes.
- The main limitation of the economic analysis is that utility values were estimated using published algorithms for mapping mRS scores on to the EuroQol-5 Dimensions, rather than being based on responses collected directly from participants. Although the QALY difference found was small and, therefore, potentially prone to measurement error, the value reflects the entire trial population, whereas only a proportion of patients received thrombolysis, which is a treatment that benefits or harms only a proportion of those who are treated.
- It was not possible to obtain views from patients receiving the PASTA pathway intervention; however, the WP2 surveys strongly indicated that rapid specialist care is valued from the start of the emergency stroke pathway.
Work package 2
- Delphi exercises and surveys reflect the views of the individuals involved only. In particular, only 150 respondents completed the BWS survey, although it is reassuring that the responses agreed with the earlier simpler survey of 103 volunteers.
- Simulation models assumed that populations were consistent with published meta-analyses,23,24 included limited parameters reflecting underlying data sets and did not consider capital costs for setting up new services. Consequently, the results may not be completely replicated post implementation.
- The modelling work also cannot account for unforeseen developments in future services or technologies. For example, currently, there are few published data about how IAT treatment effect varies for older patients, which could influence future stakeholder views about changes to the emergency stroke pathway.