Developing an LHS requires infrastructure (access to data and analytical power) and links to policy and practice. Such infrastructure needs to be developed as a platform to support multiple learning health cycles in order to maximise the return on investment
Effectiveness and value of health care is an increasing challenge to WA with rising health costs, increasing burden of chronic disease and a challenging fiscal environment where governments are faced with large and increasing budget deficits. Measuring the effectiveness of health care to ensure services are valuable and produce beneficial outcomes in a timely and appropriate way is a major challenge to the research community and health departments implementing clinical service planning.
Research into what is effective in health care is rapidly expanding, yet there is an estimated 17-year lag between publication of research findings and translation into practice and/or policy. Learning Health Systems (LHS) have been proposed as a system-level innovation to help address these challenges.
LHS have been developed both within and between many successful institutions in the US, Europe, and Asia.
The LHS set up by the Collaborative for Healthcare Analysis and Statistical Modelling (CHASM) at UWA, already supports a number of learning health cycles for WA Health.
The LHS for dental services has enabled monitoring of regions with high demand and low access to services. This research led to the Chief Dental Officer being able to target increased funding to areas of need, reducing rates of caries and hospitalisations and generally improving dental health.
Another health cycle for management of high-risk foot (HRF), including diabetic foot and osteomyelitis, predicts five-years into the future those suburbs which will have high rates of hospitalisations. This allows time for the development and implementation of interventions to avoid HRF hospitalisations. Similar learning health cycles have been developed for mental health and other chronic diseases such as COPD, diabetes and heart failure.
Interventions (both complex and simple) following implementation can then be evaluated using statistical techniques that use the large population-based datasets to identify the counterfactual, that is, a matched cohort who have not been exposed to the intervention and measure the effectiveness of the intervention.
As interventions and changes to service provision are enacted, the resulting outcomes are monitored, refined and improved. This ensures clinical services are optimally placed and provide the right service at the right time, improving efficiency and patient outcomes.
In an LHS, the characteristics, interventions and experiences of every patient in the health system are available to improve knowledge. The rapid escalation in ‘big data’ computing power and electronic health records with linkage of administrative datasets, have made this approach more timely and achievable. Best-practice knowledge from analysis of such data can then be quickly available to support decision making by clinicians and care providers, as well as service planners and policy makers.
LHS support learning health cycles
These multiple and iterative cycles include a ‘data to knowledge’ step, which represents the standard research approach. However, there is also a ‘knowledge to data’ step that closes the learning health cycle, allowing research findings to be turned into policy and services, the impact of which is monitored creating new data for a new learning cycle to begin.
Developing an LHS requires infrastructure (access to data and analytical power) and links to policy and practice. Such infrastructure needs to be developed as a platform to support multiple learning health cycles in order to maximise the return on investment.
Finally, stakeholders within an LHS should view its ongoing activity as an intrinsic part of their culture, and for the process to be trusted and valued. LHS have been described as cyber-social eco-systems involving large-scale, human-intensive, computer-supported information processing systems.
One of the fathers of LHS, Charles Freidman, described current research as ‘individual rowboats’ being built to answer individual questions which deliver research analysis across the river to the health system where there is a 17-year wait before translation to policy. An LHS is a multi-laned bridge quickly delivering knowledge, analysis and policy change back and forth to drive effective and efficient health outcomes.
In WA, CHASM has developed the infrastructure and analytic power to support LHS, funded and sponsored by the WA Department of Health. Its vision is to improve access to appropriate health care, improve equity in health, and optimise health outcomes.
CHASM is a collaboration of the Schools of Medicine, Population Health, Earth and Environment, and Mathematics and Statistics, in order to ensure access to the necessary clinical, statistical, health economic, and geo-spatial modelling expertise. It is actively improving and developing ways to analyse health care data, including better methods to calculate the impact of behaviours and risk factors on health outcomes and to measure outcomes in the long-term.
Its research activities are governed and directed by Department of Health senior executive from all area health services, Aboriginal health and clinical service planners and led by the WA Chief Medical Officer. It relies on access to one of the largest and most comprehensive linked datasets in the world, the WA Data Collections, which include de-identified, encrypted, and access restricted clinical and demographic data for all hospital admissions, emergency department presentations, state-funded mental health services, cancer registrations, births, and deaths in WA since the 1970s.
Access to a whole-of-population dataset maximises the analytical power, provided by dedicated high-end servers, to allow more ‘real-time’ analysis to respond to the research and evaluation priorities of the WA government and the health department of WA.
If you have any comments or questions about LHS in WA, please contact the authors.
References on request