Impact of implementing a computerised quality improvement intervention in primary healthcare

Health systems worldwide experience large evidence practice gaps with underuse of proven therapies, overuse of inappropriate treatments and misuse of treatments due to medical error. Quality improvement (QI) initiatives have been shown to overcome some of these gaps. Computerised interventions, in particular, are potential enablers to improving system performance. However, implementation of these interventions into routine practice has resulted in mixed outcomes and those that have been successfully integrated into routine practice are difficult to sustain. The objective of this thesis is to understand how a multifaceted, computerised QI intervention for cardiovascular disease (CVD) prevention and management was implemented in Australian general practices and Aboriginal Community Controlled Health Services and assess the implications for scale-up of the intervention. The intervention was implemented as part of a large cluster-randomised controlled trial, the TORPEDO (Treatment of Cardiovascular Risk using Electronic Decision Support) study. The intervention was associated with improved guideline recommended cardiovascular risk factor screening rates but had mixed impact on improving medication prescribing rates. In this thesis, I designed a multimethod process and economic evaluation of the TORPEDO trial. The aims were to: i. Develop a theory-informed logic model to assist in the design of the overall evaluation to address study aims (Chapter 3). ii. Conduct a post-trial audit to quantify changes in cardiovascular risk factor screening and prescribing to high risk patients over an 18-month post-trial period and understand the impact of the intervention outside of a research trial setting (Chapter 4). vi iii. Use normalisation process theory to identify the underlying mechanisms by which the intervention did and did not have an impact on trial outcomes (Chapter 5). iv. Use video ethnography to explore how the intervention was used and cardiovascular risk communicated between patients and healthcare providers (Chapter 6). v. Conduct an economic evaluation to inform policy makers for delivering the intervention at scale through Primary Health Networks in New South Wales (Chapter 7). vi. Use a new theory to explain the factors that drove adoption and non-adoption of the intervention and assess what modifications may be needed to promote spread and scale-up (Chapter 8). I found variable outcomes during the post-trial period with a plateauing of improvements in guideline recommended screening practices but an ongoing improvement in prescribing to high risk patients. The group that continued to have the most benefit was patients at high CVD risk who were not receiving recommended medications at baseline. The delay in prescribing recommended medication suggests healthcare providers adopt a cautious approach when introducing new treatments. Six intervention primary healthcare services participated as case studies for the process evaluation. Qualitative and quantitative data sources were combined at each primary healthcare service to enable a detailed examination of intervention implementation from multiple perspectives. The process evaluation identified the complex interaction between several underlying mechanisms that influenced the implementation processes and explained the mixed trial outcomes: (1) organisational mission; (2) leadership; (3) the role of teams; (4) technical competence and dependability of the software tools. Further, there were different ‘active ingredients’ vii necessary during the initial implementation compared to those needed to sustain use of the intervention. In the video ethnography and post-consultation patient interviews, important insights were gained into how the intervention was used, and its interpretation by the doctor and patient. Through ethnographic accounts, the doctor’s communication of cardiovascular risk was not sufficient in engaging patients and having them act upon their high-risk status; effective communication required interactions be assessed, discussed and negotiated. The economic evaluation identified the cost implications of implementing the intervention as part of a Primary Health Network program in the state of New South Wales, Australia; and modelled data looked at the impact of small but statistically significant reductions in clinical risk factors based on the trial data. When scaled to a larger population the intervention has potential to prevent major CVD events at under AU$50,000 per CVD event averted largely due to the low costs of implementing the intervention. However, the clinical risk factor reductions were small and a stronger case for investment would be made if the effects sizes could be enhanced and sustained over time. The findings from chapters 4-6 provide insight into the intricacy of the barriers influencing implementation processes and adoption of the intervention. Taken together, these studies provide a detailed explanation of the processes that may be required to implement such an intervention at scale and the factors that might influence its impact and sustainability. The findings are expected to assist policy makers, administrators and health professionals in developing multiple interdependent QI strategies at the organisational, provider and consumer levels to improve primary healthcare system performance for cardiovascular disease management and prevention