A multifaceted quality improvement intervention for CVD risk management in Australian primary healthcare: a protocol for a process evaluation

Non-communicable diseases (NCDs)—including cardiovascular diseases (CVD), cancers, respiratory diseases, and diabetes mellitus—are the leading cause of death worldwide [1],[2]. It is predicted that by 2030, NCDs will account for 75% of all deaths with the largest proportion of deaths attributed to CVD [3]. Globally, approximately 17 million (30%) deaths per year and 151 million disability-adjusted life years (DALYs) are caused by CVD [4].

In Australia, CVD is responsible for 34% of deaths and 18% of the burden of disease and injury, making it the largest contributor to health system expenditure [5]. Aboriginal and Torres Strait Islander people experience a greater CVD burden than other Australians, and this is a major contributing factor to the 10-year gap in life expectancy [6],[7]. The majority of CVD is caused by modifiable risk factors, which include blood pressure, lipids, diabetes, body mass index (BMI), tobacco smoking, alcohol use, unhealthy diet, physical inactivity, and psychosocial stress [8]. Recent modelling suggests that a 25% reduction in the prevalence of six NCD risk factors alone (tobacco, alcohol, salt, blood pressure, obesity, and glucose) could reduce global disease burden by 25% in the next 10 years [9].

International clinical guidelines recommend that assessment for CVD prevention and management should be based on a combination of risk factors (the ‘absolute risk’ approach) rather than treating risk factors such as elevated blood pressure and cholesterol in isolation [10]-[13]. Absolute risk calculation estimates an individual’s risk of a CVD event over time based on modifiable and non-modifiable risk factors such as age and gender. This enables early identification, management, and primary prevention of CVD for individuals at high risk. In combination with well-established secondary prevention recommendations for people who have experienced a previous CVD event, the absolute risk approach offers considerable potential for reducing CVD burden [14].

Despite the widespread availability and consistency of these guidelines and the availability of validated risk prediction equations, there are large evidence practice gaps. Health professionals tend to use these guidelines and equations sporadically and inconsistently [15],[16]. In the Australian context, studies have found that only 50% of adults attending primary healthcare have been screened for CVD risk in accordance with guideline recommendations, and only 40% of those identified as high risk have been prescribed recommended medications [17],[18]. International studies have similarly demonstrated that a minority of people are being provided with appropriate screening measures and preventive treatments [19]. The majority of CVD events can potentially be averted with adequate implementation of established treatments and interventions that are effective, efficient, and universally accessible in Australian primary healthcare settings. Primary healthcare is the ‘front-line’ for effective prevention and management of the increasing burden of chronic diseases. With 88% of Australians visiting a general practitioner each year, the opportunities to improve primary and secondary prevention of CVD at this level are great [20].

Given the magnitude of these evidence practice gaps in CVD prevention, effective quality improvement (QI) innovations that support health services to improve their outcomes are urgently needed. QI is a multidimensional concept that focuses on improving the efficiency and process of a program, service, or organisation, resulting in improved health outcomes [21],[22]. The Agency for Healthcare Research and Quality (AHRQ) has identified nine broad QI strategies to improve quality of care: provider reminder systems, facilitated relay of clinical data to providers, audit and feedback, provider education, patient education, promotion of self-management, patient reminder systems, organisational change, and financial incentives, regulation, and policy[23]. An increasingly important component of QI strategies is an adoption of health information technologies (HITs). Meaningful use of electronic health records (EHRs), computerised provider order entry systems, and electronic decision support (EDS) are increasingly recognised as key enablers to improvements in quality and delivery of healthcare [24]. However, the overall impact of computerised QI interventions to improve CVD burden has been limited, and effects on patient outcomes remain unclear [25]-[27].

Whilst QI strategies have been well characterised, there is relatively little knowledge translation research to help guide how these strategies can be optimally implemented into routine practice [28],[29]. The Canadian Institute of Health Research defines knowledge translation as ‘the exchange, synthesis and ethically sound application of knowledge—within a complex system of interactions among researchers and users—to accelerate the capture of the benefits of research for patients through improved health, more effective services and products and a strengthened health care system’ [30]. Whilst it is critical that QI interventions are robustly assessed for effectiveness, equally important is a detailed understanding of how those QI interventions are implemented, using process and economic evaluations. This will allow a deeper understanding of how the intervention worked/did not work, in which contexts was it most effective/ineffective, and why. Given QI interventions are inevitably complex in nature, robust evaluation requires the use of multiple theories and frameworks to answer these questions. By using knowledge translation frameworks, we are able to better identify active ingredients of the intervention that change behaviour, causal mechanisms of change, effective modes of delivery, and the intended population or target [31]. This will enable promotion and integration of the interventions into clinical practice, health systems, and policy.

There have been few randomised evaluations of QI interventions in the Australian primary healthcare setting. We designed a multifaceted QI intervention for CVD risk management in Australian primary healthcare. The intervention drew on two established QI mechanisms: 1) electronic decision support and 2) audit and feedback (summary of the clinical performance over a specified period of time) [32]-[37]. The intervention was evaluated in the TORPEDO study, a cluster-randomised controlled trial (cRCT) involving 60 health services. Details of the trial are published elsewhere [38]. In brief, the system was integrated with the healthcare provider’s EHRs, and included (1) a real-time decision support interface using an algorithm derived from several evidence based national guidelines (Figure 1); (2) a patient risk communication interface which included ‘what if scenarios’ to show the benefits from particular health risk factor improvement during a consultation (Figure 2); (3) an automated clinical audit tool for extraction of data and review of health service performance (Figure 3); and (4) a web portal where services can view peer-ranked performance over time (Figure 4). Healthcare providers could use the point-of-care tool as part of a routine clinical consultation. For the audit and feedback component, quality indicators were developed for patients who had visited the health service at least three times in the preceding 2 years and once in the preceding 6 months. The population studied was based on national guideline recommendations for CVD risk screening and included Aboriginal and Torres Strait Islander people over 35 years and all others over 45 years [38].

Data were collected for 38,725 people from the 60 health services. Overall, when compared with control, the intervention was associated with a 25% relative (10% absolute) improvement in CVD risk factor screening. Overall, there was no significant difference in the prescribing rates of recommended medicines to people at high CVD risk. The intervention was, however, strongly associated with improvements in the sub-group at high CVD risk that was not prescribed with recommended medicines at baseline. There were also improvements in intensification of existing, recommended medication regimens. There were modest improvements in attaining blood pressure targets but no differences in other clinical outcomes. The improvements in recommended prescriptions to high risk patients were not accompanied by increased prescribing rates for patients at low risk [39].

Although the intervention exhibited significant improvements in some outcomes, it remains unclear how this intervention was actually used in practice and what factors promoted and hindered its use. Answers to these questions are critical in order to inform future directions for its implementation and for implementation of similar interventions in other settings.

In this paper, we outline our protocol for a theory-based process evaluation of the TORPEDO intervention. The broad objectives are to understand how, why, and for whom the intervention produced the observed outcomes and to develop effective strategies for translation and dissemination. The evaluation will identify which intervention components promoted or had minimal impact at the provider, patient, and system levels and the mechanism of change. It will also identify the contextual influences on the delivery of the intervention and its outcomes. The specific objectives are the following:

Drawing on the RE-AIM framework, a logic model was developed to assist in the planning, conduct, and evaluation of the research components (Figure 5) [41]-[44]. The model assesses five dimensions of the intervention at different levels (individual, health service/clinic or organisation, and community/population): (1) participant R each; (2) E ffectiveness of the intervention; (3) A doption by the target health service; (4) I mplementation fidelity, costs, and adaptations made during delivery; and (5) M aintenance of intervention effects over time. The model identifies and describes inputs, activities, outputs, and outcomes of the intervention [45]. The four objectives have been mapped onto the relevant components of the RE-AIM framework.

The logic model will be underpinned by three theoretical perspectives—realist evaluation [46], the theoretical domain framework (TDF) [47],[48], and normalisation process theory (NPT) [49],[50].

Realist evaluation seeks to answer the question what works, for whom, and in what circumstances?[46]. This is accomplished by identifying and understanding the underlying mechanisms by which the intervention succeeds or fails in varying contexts to produce the patterns of outcomes. Therefore, unpacking of the underlying generative mechanisms of the intervention and its effects is contingent on understanding the features of the context (i.e. roles and relationships of personnel at health services, IT infrastructure, economic conditions, demographic, motivation and skills of health professionals, etc.) [46],[51],[52].

Successful implementation of evidence-based guidelines and QI interventions depends largely on changing the behaviour of healthcare professionals and patients, who are influenced by external (i.e. organisational, environmental, resources) and internal (i.e. motivation, capability) factors. TDF is a consensus of numerous behaviour change models and comprises 14 domains derived from psychological and organisational theory (knowledge, skills, social/professional role and identity, beliefs about capabilities, optimism, beliefs about consequences, reinforcement, intentions, goals, memory attention and decision process, environmental context and resources, social influences, emotion, and behavioural regulation) [47],[48],[53].

In order for healthcare innovation and technology to become routinely embedded in every day work, we need to understand how the innovation is integrated within the existing practices of a healthcare organisation. NPT assists in understanding how health professionals implement an intervention. NPT identified four main components: (1) coherence (sense making by participants), (2) cognitive participation (commitment and engagement by participants), (3) collective action (the work participants do to make the intervention function), and (4) reflexive monitoring (the degree to which participants reflect on or appraise the intervention [49],[54],[55].