Improving the use of data to support management of piscirickettsiosis in Chile

Abstract

A shadow to the success of the salmonid aquaculture industry in Chile is the fish health burden caused by salmonid rickettsial septicaemia (SRS), an infection caused by Piscirickettsia salmonis. The work presented in this thesis builds on a research platform developed to integrate data routinely collected on salmonid aquaculture farms, an initiative named PIISAC. The overall objective of the work conducted during this PhD was to investigate how data may be used more effectively to improve the management of fish health in salmonid farming, in particular for salmonid rickettsial septicaemia in Chilean sea farms. In the first part of the thesis, the PIISAC platform was used to further examine the risk factors for SRS and evaluate the effectiveness of various interventions to control the disease in farmed salmon and trout in Chile. In chapter 2, routine production and health data were analysed to assess the effectiveness of antimicrobial treatment of infected fish, identifying practices which impact positively or negatively on the management of mortality outbreaks in sea farms. In chapter 3, the analysis focused on the interplay between sea lice burden, sea lice management using bathing treatments and SRS mortality. The work highlighted the challenge in managing ectoparasites while not increasing the stressors applied to the fish. After considering interventions to control SRS at the farm-level, the correlation in SRS mortality between neighbouring sea farms was assessed in chapter 4. The results presented in the first part of the thesis contribute to the growing evidence base available for veterinarians and fish health managers to make better decisions regarding the management of SRS. These results also show that analysis of routine production and health data can provide useful insights into the effectiveness of within-farm interventions to control SRS. The use of data in this manner requires durable data infrastructure and governance arrangements. In this context, the second part of the thesis looks at how epidemiological research may be conducted in the future in a sustainable manner. The barriers to the adoption of the PIISAC platform were analysed using a qualitative approach, presented in chapter 5. Such data platforms may provide a means to share and re-use data for the benefits of the data providers, while minimizing the costs and efforts of repeated rounds of data collection. This study showed that more work is needed to identify the most appropriate arrangements for data curation and governance and support future data integration platforms for fish health. An alternative approach to improving the use and re-use of data in scientific communities was proposed in 2016, under the form of the FAIR (Findable, Accessible, Interoperable, Reusable) principles. Chapter 6 investigates the progress towards implementing these principles among the aquaculture epidemiology community. The final chapter addresses questions relating to statistical modelling, including causal analysis in veterinary epidemiology and the use of different approaches for making inferences from statistical models. Last, an outline of what qualitative approaches can contribute to veterinary epidemiology is presented, based on the experience gained during the course of this PhD.