Role of Cohort Studies and Registries in Musculoskeletal Healthcare and Research

Patricia D. Franklin, MD, MBA, MPH

Jeffrey K. Lange, MD

Dr. Franklin or an immediate family member has received research or institutional support from Zimmer. Neither Dr. Lange nor any immediate family member has received anything of value from or has stock or stock options held in a commercial company or institution related directly or indirectly to the subject of this chapter.

INTRODUCTION: COHORTS AND REGISTRIES IN MUSCULOSKELETAL HEALTHCARE AND RESEARCH

Researchers and regulators (eg, the FDA) are prioritizing the use of real-world evidence (RWE) to generate new knowledge of best practices to treat the growing population of adults afflicted by chronic musculoskeletal conditions.¹ In addition, the FDA has called for enhanced postmarketing surveillance and safety monitoring of implantable devices using RWE.² Although electronic health records (EHRs) and administrative billing data are standard RWE sources, disease-specific registries and cohort studies are fundamental to evaluating outcomes of diverse treatments and longitudinal device surveillance. For example, in Sweden, both the osteoarthritis and total knee and hip replacement registries have informed clinical practice for decades.³ For musculoskeletal outcomes, the Swedish, United Kingdom, and Australian arthroplasty registries, among others, provide data to compare outcomes across health systems, comparative effectiveness information to assess the impact of diverse treatments, and device performance and safety statistics for surgical interventions.⁴^,⁵

A growing number of US registries and cohort studies are emerging in response to the need to evaluate both treatments and outcomes for various musculoskeletal conditions. Professional societies are taking the lead to create registries to guide clinical practice. The American Association of Orthopaedic Surgeons is developing a centralized registry infrastructure for total joint replacement, spine, foot and ankle, sports, and other procedures.⁶ The American College of Rheumatology⁷ and the American Physical Therapy Association⁸ are developing parallel registries to capture clinical interventions, diagnoses, and outcomes. In addition, private health associations (eg, Kaiser Permanente⁹), state-wide associations (eg, Michigan arthroplasty registry¹⁰), and national research programs (eg, the FORCE-TJR¹¹ and the SPORT¹² and MOON trials¹³) are being created/used to evaluate treatments for ligament pathology and to collect longitudinal outcome data for patients with a variety of musculoskeletal conditions and treatments. Longitudinal implant survivorship, postprocedure adverse events, and patient-reported health outcomes are typically measured and monitored.

In the United States today, EHRs are the primary data repository capturing the details of clinical care as well as information to support billing and administrative functions, such as visit scheduling. Thus, the EHR is the source of important RWE. In addition, EHR data are often submitted to registries or research cohorts. However, data sharing between the EHR and registries or research cohorts—including the ability to return registry data to the EHR—has the potential not only to enhance the value of EHR data but to expand the utility of registries and research to serve quality monitoring, regulatory evaluations, and research. This bidirectional sharing is often called a “learning health system (LHS).” In the LHS, electronic data collected for
clinical care also serve research, and vice versa. In this way, the surgeons and health systems that participate in the registry or cohort are simultaneously contributing to research, as well as monitoring the outcomes of their own care as compared with others in the registry.

Califf and co-authors, who represent major federal research and health policy organizations (FDA, HHS, NIH, AHRQ, PCORI, CDC, etc.), have called for research to develop new methods to reliably answer questions and for new approaches to study design and conduct that will optimally collect and use RWE.¹⁴ They list the principles of a learning health system as requiring at least two features that are often seen in well-designed cohorts and registries:

1. Operational practice networks that generate research evidence within a framework for autonomy, privacy, confidentiality, and security.
2. Common definitions, collection, and storage of digital data with informed consent, as possible, and appropriate privacy protection.

Although US public health and safety data are beginning to emerge from registries, the secondary uses of registry data for research and feedback to clinicians and patients in learning health systems are currently limited. In 2017, a National Academy of Medicine paper on engaging clinicians in the learning health system highlighted FORCE-TJR as a model for engagement. FORCE-TJR both supports federally funded research and provides real-time quality feedback to clinicians.¹⁵ Internationally, Scandinavian arthroplasty registries have a long history of engaging in research while serving the public health reporting needs of their countries.¹⁶

WHAT ARE THE BENEFITS OF REGISTRIES AND COHORT STUDIES TO CLINICIANS AND PATIENTS?

Although the randomized controlled clinical trial (RCT) is the standard method for understanding the effect of a medical intervention on select groups of patients, extrapolating RCT results to clinical practice may be difficult because of the diversity of patients and healthcare factors as compared with the controlled factors of the RCT. Variation in patient and healthcare system factors will influence the intervention effects in daily practice. Cohorts and registries that include diverse patients across “real-world” clinical settings may narrow the gap between treatment intervention effects in RCTs and practice. Thus, clinicians should value the outcome analyses that emerge from cohorts and registries.

Although EHR data serve healthcare delivery within a single health system, EHR data elements are rarely consistently defined or structured to support aggregated use across health systems. Thus, well-designed registries and cohort studies with consistent patient, surgical, and outcome data are the keys to comparative effectiveness and outcomes research.

Predefined measurement timing and capture across settings.

Inconsistent timing and scheduling of clinical visits is a major challenge to using EHR data for longitudinal monitoring. Thus, data-capture strategies are needed to improve the completeness of data and the consistency of timing and intervals between measures. To complement EHRs, registries can collect measures across different healthcare delivery systems and associated EHRs. Links to national death indices and administrative data are key to assessing longitudinal endpoints. Registries that integrate data submitted by patients across time and location can enhance the quality and usability of data for RWE research. Outreach to engage patients in longitudinal monitoring, especially for patients who do not return to the clinic, can improve data completeness.

Consistent core measure sets to meet research standards.

Well-designed registries ensure consistent structure and format of data elements. For example, smoking history can be captured using a single, consistent measure (eg, pack years) or the presence or absence of relevant medical comorbid conditions may be explicitly documented rather than assuming that the absence of documentation for the condition means the absence of the condition, as can occur when using EHR data.

Complete data automation systems to minimize bias.

Well-designed registries are programmed to prompt data collection at key intervals from all patients—both those with and without office visits. Automated data-capture protocols can be implemented through multiple capture modes (eg, web portal, phone, or in-office computers) to increase the likelihood of complete data capture. Similar automation can notify clinicians about patients who need to be reevaluated through an in-office visit based on results of certain patient-reported outcome measures.

Integration of clinical and patient-generated data in registries.

Until recently, patient-reported outcomes (PRO) and other patient-generated data were not routinely recognized as valuable for evaluating musculoskeletal treatment outcomes. Registries can include PRO data and play a leadership role in refining relevant PRO metrics, evaluating the strengths and limitations of patient-reported data in comparative effectiveness research and public health and quality analyses. As passive data emerge from smart phones and other wearable devices, further research to define “clinically” meaningful metrics and change over time are needed.¹⁷

WHAT ARE THE CLINICAL BENEFITS OF DATA COLLECTED FROM COHORTS AND REGISTRIES?

Although local EHRs store clinical information for patients treated within a health system, they do not house data from sites across the nation. Thus, hospital administrators and clinicians charged with managing population outcomes for quality or bundled payment initiatives must seek external data sources against which to benchmark outcomes and identify opportunities to improve. Registries designed to integrate data across health systems can readily calculate national or regional benchmarks. Such data can be valuable for comparison in individual patient care as well as at the health system or population level.

Because registries aggregate data across multiple healthcare settings, the number of patients can accumulate faster than data from a single health system. As a result, signals and trends within care patterns can be more quickly identified. Faster accumulation allows for recognition of both positive and negative outcomes, and these lessons can be readily returned to participating health systems. Furthermore, EHR software is not designed to perform sophisticated statistical analyses, whereas registry data can be easily programmed using advanced statistical techniques. As the prevalence of APIs (Application Programming Interfaces) expands, EHRs can readily submit single-site data to a registry with the expectation of real-time statistical processing and the return of risk-adjusted or other refined metrics.

Patient care and treatment decisions can be informed by real-time reports to integrate the patient’s health status with national norms. APIs and other technical interfaces are available to link population analyses of real-world data with individual patient records. Collaborations between Dartmouth researchers and rheumatology registries are testing this “co-production” model.¹⁸ A second example of translating national registry data to real-time shared-decision reports is the FORCE-TJR registry that provides real-time predictive analytics adjusted for individual patient risk factors to patients and surgeons considering total joint surgery. With PCORI funding, the impact of these shared-decision reports on decision quality and outcomes is being evaluated now in the Arthritis care through Shared Knowledge (A.S.K.) study.¹⁹

EXISTING COHORT AND REGISTRY RESOURCES FOR ORTHOPAEDIC SURGEONS

During the past decade, great focus has been placed on the development of high-quality data resources that can inform best practices for orthopaedic clinical care delivery. Well-designed longitudinal cohort studies help to provide answers to specific clinical questions regarding patient factors, surgical factors, and outcomes associated with particular intervention strategies. Well-established and developing registries have given us insight into orthopaedic surgical trends and basic quality and performance measures. Some of this decade’s most impactful musculoskeletal cohorts and registries are summarized below. Table 1 summarizes key resources from existing cohort studies and registries to inform best practices in high-volume orthopaedic practices.

SPINE

The Spine Patient Outcomes Research Trial (SPORT) was a multicenter randomized controlled trial with a concurrent observational cohort, studying the comparative effectiveness of surgical management and nonsurgical management of lumbar spine disease. Thirteen medical centers in 11 US states participated. With respect to the treatment of lumbar disk herniation, long-term results from the randomized cohort (501 patients) showed no statistically significant differences in bodily pain (BP), physical function (PF), or modified Oswestry Disability Index (ODI) between surgical and nonsurgical management in the intent-to-treat analysis.¹² However, improvements in secondary outcomes including sciatica bothersomeness, satisfaction with symptoms, and self-rated improvement were statistically significantly greater in the surgical group as compared with the nonsurgical group in the intent-to-treat analysis. Of note, there was significant nonadherence to treatment, and the as-treated analysis showed statistically significant and clinically meaningful greater improvements in symptoms for those patients treated surgically as compared with those treated nonsurgically in both randomized and observational cohorts (501 and 743 patients, respectively). The ultimate result of this cohort study indicated that for appropriately selected patients, surgical intervention for lumbar disk herniation can confer greater benefit than nonsurgical management at 8-year follow-up.¹² Similar midterm results were reported for surgical versus nonsurgical management of lumbar stenosis and degenerative spondylolisthesis.²⁰^,²¹

Only gold members can continue reading. Log In or Register to continue