Focus on real-world data in diabetes
Real-world data is changing the landscape of research into diabetes outcomes. The opportunities and challenges offered by this new approach are discussed in a recent article in Diabetes Care. Dr Susan Aldridge reports.
Despite advances in research, technology and care, diabetes remains an increasing burden around the world. Addressing this burden, with its associated costs, calls for new efforts across the spectrum of research disciplines to improve diabetes care.
In recent years, there has been a rapid growth in systems for accumulating real-world data (RWD) and using it to identify real-world evidence (RWE) outside of research settings. Collection and analysis of big data are increasing the options for evidence-based guidance of diabetes management and prevention. Although these new data are not collected for research, they do have the potential to enrich the body of health-related information. This RWD is being applied to a number of disciplines, including clinical epidemiology, health services research, population surveillance and implementation research.
These developments have the potential to improve diabetes prevention and treatment by increasing the range of populations, settings, interventions, outcomes and clinical settings that can be explored. However, it must be borne in mind that the level of evidence that may be derived from RWD is ultimately a function of the quality of that data and rigour of study design and analysis.
In a new report, Edward Gregg of RCSI University of Medicine and Health Sciences, Dublin, and colleagues elsewhere, look at the current landscape and applications of RWD in clinical effectiveness and population health research in diabetes. They then summarise opportunities and best practises in the conduct, reporting and dissemination of information derived from RWD to optimise its value and limit its drawbacks.
Applying RWD
Randomised controlled trials (RCTs) have long been the gold standard behind treatments, health services and policies. They take a linear path, starting with a research question and study design, then data collection and analysis. Real-world evidence studies may also start with a question, but can expand the scope of research objectives because they include a broader range of variables.
There are four ways in which RWD can complement traditional approaches. First, it seeks to understand the effectiveness of evidence-based practices as they are necessarily modified for delivery outside the research setting. Second, it extends evaluation to population subgroups that may be under-represented in trials. Third, RWD also aims to assess the effectiveness of interventions and policies where conventional experimentation is too impractical, costly or unethical. Finally, it permits the examination of outcomes that are beyond the scope of traditional survey, trial, cohort or other measurement methods.
Clinical diabetes research
While RCTs remain the gold standard for determining the efficacy of diabetes treatments and establishing guidelines, they do have a number of drawbacks. They study effects under ideal healthcare practice circumstances and often with highly selected populations, thereby resulting in findings that are not really generalisable to routine clinical care. They are also costly, lengthy, expose participants to potential risk and are usually designed to answer a narrow hypothesis. Higher risk older people with multiple comorbidities and polypharmacy are often excluded, although these will be the very recipients of the medications on trial when they reach the mainstream of diabetes care. For instance, in the PROactive, ADVANCE and EMPA-REG OUTCOME trials (of pioglitazone, preterax and diamicron, and empagliflozin, respectively), the clinical characteristics of the study population would have represented only 3.5%, 15.7% and 35% of the real-world population, according to one analysis. And the protocols of RCTs, which standardise treatment dosage and timing, may not be representative of management in a broader range of settings. Nor are issues relating to concurrent illness and medications and other aspects of care addressed.
These limitations call for RWE to complement the findings of RCTs. Current healthcare systems produce a large amount of longitudinal, patient-level, electronic data that assess exposures of interest and associated health outcomes in clinical practice. These RWD can create large databases that can be used for research to improve diabetes care.
Most published applications of RWD have compared the effectiveness of pharmacological interventions. For instance, recent cardiovascular outcome trials of SGLT-2 inhibitors have demonstrated improvements in cardiorenal outcomes. These findings have been confirmed and extended by non-interventional studies based on RWD through the inclusion of unrestricted populations as treated in actual clinical settings. These studies have resulted in effectiveness estimates for previously unstudied populations in the context of clinically varying treatments, dosages, comorbidities and dilutions of effect that can occur due to suboptimal adherence. In this setting, RWD may offer more useful evidence to complement RCT findings for those involved in planning a treatment strategy. In another example, comparative effectiveness analyses based on RWD provided support for the broader relevance of the CAROLINA study, which compared the efficacy and cardiovascular safety of a DPP-4 inhibitor and a sulphonylurea in type 2 diabetes.
Thanks to the very large populations represented by RWD collected during routine care, RWE is also being used increasingly by regulatory agencies to assess safety of medical products, particularly with respect to rare adverse drug events that RCTs are not powered to examine. For example, diabetic ketoacidosis among users of SGLT-2 inhibitors was identified as a safety concern post-marketing by RWD studies.
Finally, RWD studies can provide insights into the natural history of diabetes and these have extended our understanding of the risks associated with hypoglycaemia beyond what was found in RCTs. For example, the Hypoglycaemia Assessment Tool (HAT) study has shown rates of overall, nocturnal and severe hypoglycaemia that were much higher than seen in RCTs. Further exploration of these findings could allow earlier and more accurate identification of those at risk of hypoglycaemia.
Surveillance and monitoring
Surveillance has many applications in diabetes care – including gathering data on risk factors, incidence and prevalence, complications and death among those with the condition. It can also help monitor levels of care, disparities in care access and inform policy, resource allocation and clinical decision-making. Traditionally, surveillance has relied on surveys and registries, but now we can leverage various sources of RWE, such as electronic health records in well-defined healthcare delivery systems databases (for example, Veterans Administration and Kaiser Permanente in the US). These data, collected during the routine provision of care to large populations, provide passive surveillance that can complement the traditional active surveillance carried out by the long-running National Health and Nutrition Examination Survey and other well-established studies in the US and elsewhere.
However, surveillance based on RWD requires a good understanding of the data source. For instance, surveillance based on electronic health records works well in acute settings, such as the emergency room or hospital, but less so for outcomes that may not even come to medical attention. Surveillance of severe hypoglycaemia based only on emergency room visits or hospitalisation has been estimated to miss around 95% of episodes because they were cared for at home or by ambulance staff. And, despite the expansion of the range of RWD sources for diabetes surveillance, they remain limited in their ability to assess behaviours and patient-centred outcomes.
Much of the precedence and proof of concept for national health systems-based registries comes from early work in Scotland, the UK and Scandinavia. Here, and elsewhere, the presence of single-payer systems has facilitated the creation of comprehensive, linked registries covering primary care, lab, pharmacy, hospitalisation and mortality data. These data systems are regularly used to track trends in risk factors, care and outcomes. They also provide a basis for the study of effectiveness of interventions. This is in stark contrast to the situation in low- and middle-income countries, and in high-income countries without a single-payer system, where national or population-level systems data are rare – so they cannot yet benefit from the application of RWE.
Challenges in RWD use
Since RWD are not usually collected for research, their use in generating evidence does present some challenges. For instance, even posing the research question can be difficult because the population being studied is less under the control of the investigator than in an RCT. Then the choice of study design will depend upon the study question and the availability of data. In comparative effectiveness studies and health services research, the goal is often to emulate an RCT and often uses a new-user, active-comparator cohort design, which focuses on new users of alternative treatments with similar indicators. This study design is analogous to an RCT, but without the baseline randomisation. Other study designs – for instance, using propensity score to ‘stand in’ for randomisation – are also being explored.
When it comes to population monitoring, compared with cohort studies based on primary data collection, RWD-derived cohorts are usually more representative and have larger numbers of participants. However, drawbacks include missing data, non-standardised exposures and lack of linked behavioural, genetic or patient-centred outcomes. RWD tend to be dominated still by diagnostic and service data related to the processing of payments and generally lack patient-reported data on behaviour, risk, function and quality of life that could dramatically improve its utility.
Furthermore, individual-level social determinants of health, including income, education, social care and voluntary-sector interventions, can impact surveillance outcomes, but these may not be routinely collected in RWD sources. This means there are large gaps in understanding the drivers of, and progress in, achieving health equality.
Data harmonisation across different sources of RWD can be challenging. For example, in some settings, race and ethnicity are merged whereas elsewhere they may be treated as separate variables. Validated instruments are needed to facilitate collection of patient-reported outcomes in healthcare settings to improve future surveillance efforts.
Meanwhile, new sources of RWD are emerging. For instance, some healthcare settings are initiating remote glucose monitoring programmes and cloud-based continuous glucose monitoring downloaded to the electronic health record. This opens up the possibility of new models of care and new research opportunities – but, again, work will be needed on data harmonisation.
The final challenge lies in the reporting of RWD to ensure transparency and credibility. The Professional Society for Health Economics and Outcomes Research (ISPOR) and the International Society for Pharmacoepidemiology have created a task force to make recommendations for good practices that would increase confidence in RWE. In their report, they look at the planning, implementation and dissemination of hypotheses evaluating treatment effectiveness in RWD studies.
In conclusion
RCTs address a narrow hypothesis in a highly selected population, so we can’t be sure that the intervention under study will be effective and safe in practice in the wider population. RWE can help provide additional complementary evidence of efficacy and safety. In diabetes, RWE has become increasingly relevant and important in recent years in various areas, including clinical effectiveness, long-term surveillance and monitoring and regulatory affairs.
However, there are concerns around real-world studies, including poor data quality and inconsistent methodology, which affect the reliability of findings. Standardising the structure, registration and reporting of such research is essential to improve the quality of RWE. There are other challenges, such as facilitating broader data linkage and access, as well as a need for analytical methods to keep up with newer RWD sources, such as patient apps. Notwithstanding, RWE has the potential to revolutionise patient care in diabetes as its acceptance and credibility increase.
To read this paper, go to: Gregg EW, Patomo E, Karter AJ, Mehta R, Huang ES, White M, Patel CJ, McElvaine AT, Cefalu WT, Selby J, Riddle MC, Khunti K. Use of real-world data in population science to improve the prevention and care of diabetes-related outcomes. Diabetes Care 46(7):1316–1326. https://doi.org/10.2337/dc22-1438
To learn more, enrol on the EASD e-Learning course ‘Real-world evidence’.
Any opinions expressed in this article are the responsibility of the EASD e-Learning Programme Director, Dr Eleanor D Kennedy.
