The development of clinical registries in medicine has resulted in numerous advancements in a wide variety of health care domains. Such domains include, but are not limited to, understanding the course of particular diseases and various treatments affecting outcomes; identifying specific factors that influence quality of life or prognosis; monitoring safety; evaluating socioeconomic effect; and studying quality of care, quality improvement, and more recently, patient-reported outcome measures (PROMs). Modern clinical registries play an increasingly important role in orthopaedic surgery and have the potential to further advance the field.

Clinical registries, also referred to as patient registries, are organized bodies of information collected in a uniform manner to evaluate specific outcomes in a patient population. Registries allow collection, storage, retrieval, and analysis of information on individuals based on a disease, condition, or exposure of interest. Data may be collected from a specific geographic region or nationwide, and only a limited amount of information that reflects the purpose of the registry is captured. Clinical registries have evolved from the pooled data of single surgeon’s series or small clinical trials to large databases of patients, which provide valuable epidemiologic data that can be analyzed, disseminated, and compared throughout global orthopaedic communities.

The components of a clinical registry follow a specific design and rationale to collect data on particular outcomes of interest. The data collected are scrutinized regarding internal and external validity, compliance, and generalizability during collection, analysis, and reporting. Implementing and maintaining a registry can be complex and requires substantial effort and resources. This chapter focuses on
the history, purposes, composition, application, and implications of clinical registries in orthopaedics. Specific examples will show how clinical registries can affect the delivery of value-based health care. A thorough discussion of all significant registries in orthopaedics is beyond the scope of this chapter; however, a few select registries are highlighted to provide an overview of this topic.

Multiple facets are required to create a successful registry. At initiation, there must be a clearly articulated purpose or objective of the registry, an identified target population to study, and specifically identified data that can be collected accurately and efficiently. An established governance is imperative to ensure proper guidance and decision making on behalf of the registry, clear communication with the stakeholders involved, transparent funding proposals and acquisition, and data collection, reporting, and dissemination of results. This requires a variety of participating teams, with representative individuals who have expertise in multiple domains.¹,²

A committee involved in project management is necessary for registry coordination, time management, expense budgets, funding, and communication with both stakeholders and participating data collection sites. Clinical experts must delineate the appropriate target population to study and data collected aligning with registry goals. An infrastructure for data collection and management serves as the core of the clinical registry; information collected from multiple target sources at participating units must be accurately documented, stored appropriately, and accessible for extraction. This database management requires trained personnel to ensure data validity and completeness: as technology has enabled much of this work to be done electronically, management of specific registry-driven programs and algorithms require regular assessment for quality assurance and validation. Additionally, specialists in the field of epidemiology, statistics, and health outcomes are vital in data analysis. Legal counsel is necessary to provide guidance on patient eligibility within the registry and protection of identifiable information. Finally, the clinical registry team must be able to effectively communicate and corroborate results with other organizations, the orthopaedic community, and patients to increase knowledge and quality of care.¹,²,³

The components of a clinical registry are a reflection of the defined focus of the registry; therefore, the rationale, design, and goals of the registry must align to assess a variety of related issues regarding epidemiology, safety, efficacy, and practice policy, among others. Registries at various levels, from institutional to national, may be different regarding the type and amount of data collected. For example, total joint arthroplasty registries record specific patient identification and corresponding demographic, surgical, implant, and clinical outcome information. There are four recognized levels of data that can be collected by registries that, in turn, determine potential application⁴ (Table 1).

In total joint arthroplasty registries that record data on primary and revision arthroplasty, level I data include identifiers on patients, surgeons, and hospitals as well as procedural data to monitor rates of revisions. Level II data include patient
factors and comorbidities, surgical information, perioperative care, and complications; this allows for assessment in the types and rates of complications associated with a particular standard of care, and respective changes in these patterns with time. Level III data include PROMs by using questionnaires focusing on the patient’s perceived health, function, pain, and satisfaction. These data have implications in the identification of factors driving poor or successful patient outcomes, as well as socioeconomic implications of the procedures performed that inherently affect cost analysis and policy making. The addition of radiographs for further assessment of implants denotes level IV data; such information, collected and stored for a large number of patients, can be influential in analyzing technically driven alignment and component positioning, as well as implant wear and osteolysis over time.⁴,⁵

Two examples of dataset recommendations to meet the specific aims of the organization include those proposed by the International Society of Arthroplasty Registries (ISAR) and American Joint Replacement Registry (AJRR). The ISAR has developed a minimum dataset recommended for collection by national arthroplasty registries. This dataset represents the core minimum required to effectively compare specific prosthesis and patient outcomes, limited in an attempt to increase coverage, accuracy, and efficiency of recorded information. The ISAR minimum dataset includes prosthesis, patient, surgery, and hospital details⁶ (Table 2). In comparison, elements currently collected by the AJRR focus on three categories:
procedural, postoperative, and PROMs. Within the procedure category, data are collected on the patient, site of service, surgeon, specific procedure performed, patient comorbidities, and surgical complications. The postoperative category includes data on postoperative complications and 90-day readmissions. A variety of patient-reported outcomes are collected, with recommendations on using a measure of health-related quality of life and specific hip- or knee-related surveys.⁷ Registries may elect to add data elements that become clinically relevant (such as robotics or computer navigation) or sunset the collection of those that are no longer of interest.

The foundation of clinical registries relies on the input of quality data. Five dimensions, as reported by Malchau et al,⁵ that are fundamental to clinical registry reporting are coverage, completeness, response rate, missing values, and validation. In an attempt to maximize internal and external validity of the studies conducted, clinical registries must control for potential bias influencing results.⁸

Clinical registries are subject to accurate and comprehensive reporting. Coverage, as defined by the ratio of the number of participating units to the total number of units producing data for a procedure of interest, is particularly relevant to achieve more representative data collections. Participating institutions are also subject to the completeness of data reporting, as underreporting at the singular level may deleteriously affect analysis into misleading conclusions.⁴,⁵ As
previously mentioned, the ISAR attempts to achieve such quality by requiring accurate data collection, more than 80% contribution of national hospitals, with at least 90% procedural reporting from each site.⁶ Similarly, the use of PROMs risk incomplete or unanswered variables prior to analysis—this poses a requirement to be included in statistical analysis, as well as response rates recorded at individual follow-up times. At each step of the process, from defining the purpose of the registry to data acquisition, completeness of data reporting, and finally, analysis, efforts are required to adequately structure, staff, and fund the registry to enhance overall compliance.⁴,⁵

The concept of validity is essential to understand. Internal validity, in essence, is a measure of bias influencing results of a study. Reducing potential bias, or systematic errors influencing results, increases the internal validity of a study in that there are fewer implications due to unmeasured variables not controlled for in reported associations between exposure and outcomes. Registry data are subject to random and systematic error, notably during data collection and storage. Validation of registry data to clinical records or an external dataset representative of criterion validity and regular assessment of new data are methods to resolve such error. Registry data should be cross-compared to clinical data and even a similar database that has been recognized as criterion validity (cross comparing to a gold standard database) can help to reduce errors; consistent assessment of incoming or new data decreases unrecognized error.⁴ Randomized controlled trials, for example, attain a high degree of internal validity given the process of treatment randomization among groups of similar measured or unmeasured characteristics. Resultant outcome differences are therefore attributed to differences in the efficacy and safety of the different treatments and less subject to bias. Clinical registries, however, tend to focus on external validity instead of a more homogenous patient population of randomized controlled trials that inherently limits generalizability.⁸

External validity refers to the generalizability of the inferences made of a study, translating to a wider population beyond the population under study. Clinical registries often achieve high external validity given the population heterogeneity under study. As such, registry data may be more applicable and realistic of disease epidemiology, treatment, and outcomes. An argument can therefore be made that the inferences made from observational studies in clinical registries are more representative of the diverse patient population in current medical practices, and more relevant in driving decision making and policymaking to improve outcomes.⁸ Clinical registries are inherently different from one another regarding processes for estimating internal and external validity, but such processes should be publicly and clearly available for review.⁵

The current progression of clinical registries to utilize PROMs requires special consideration. As discussed later in the chapter, PROMs are derived from questionnaires filled out by patients regarding aspects of generalized health, pain, function, and quality of life, among others. These tools are subject to thorough assessment to ensure standards of validity, reliability, and responsiveness are met. Validity, the ability to measure an intended outcome, requires specific content to address the concept of interest; the tool should provide comparable measurements to known
standards and among different groups of interest. Reliability is the consistency and reproducibility of the tool’s ability to produce similar measurements in different scenarios in which an element is unchanged. Finally, the responsiveness of the outcome measure is the actual ability to detect change in a particular area of interest in the patient. Intrinsic to standards of responsiveness are principles of minimal clinically important difference and minimal detectable change. The minimal clinically important difference is the minimal change in outcome scoring that is clinically important or significant to the patient; the minimal detectable change is the minimal change necessary to ensure the resultant score is outside the scope of standard error within the outcome measure, and that the change in score is true and not due to internal error.⁹,¹⁰,¹¹

Clinical registries have been central to the understanding of disease epidemiology and treatment patterns. As one of the earliest registries, the Swedish Hip Arthroplasty Registry (SHAR) has documented an increase in incidence and corresponding change in prevalence of total hip arthroplasties (THAs) over time; by the end of 2019, 3.6% of the population older than 40 years underwent THA, of whom 27% underwent bilateral THA with a higher prevalence in women (4.1%) than men (3.0%). In men and women, primary osteoarthritis was the primary diagnosis for THA, but a greater proportion of women underwent THA secondary to acute trauma (hip fracture) during this same timeframe. In addition, there was a higher rate of cemented femoral stems in women compared with men.¹² This registry has been able to trend multiple epidemiologic and demographic data points with corresponding treatment patterns, which is representative of how clinical registries serve as a valuable tool for analysis and future projections.

Similarly, registries have expanded understanding regarding etiologies of primary arthroplasty failure. The recent Australian Orthopaedic Association National Joint Replacement Registry (AOANJRR) 2020 report identified the revision burden (defined as the ratio of implant revisions to the total number of arthroplasties in a specific period) of primary total knee arthroplasty (TKA) and THA at 8.0% and 8.4%, respectively¹³; the recent AJRR 2020 report identified revision rates of 4% in TKA and 3.1% in THA from years 2012 through 2019.¹⁴ The registries have identified major etiologies for failure as well as risk factors for revision surgery including patient factors, implant factors, and surgeon factors such as surgical approach. Registries are uniquely situated to follow failure rates of primary procedures and help identify areas where improvement is needed for survivorship.

Registries have also been instrumental in the surveillance of specific implants used in orthopaedic surgery, providing critical information on implant performance, survivorship, and adverse events. Currently, orthopaedic devices comprised more than 16% of all class I and II medical device recalls in the United States from the years 2015 through 2019.¹⁵ There has been a focus on revision as an endpoint as it can serve as an indicator for the quality of an implant, poses considerable burden to the patient and healthcare system (time and expense), and is a reproducible and comparable data point. The United Kingdom National
Joint Registry (NJR) provided early insight on the poor implant survivorship of metal-on-metal (MoM) hip resurfacing with more than 13% requiring revision after 10 years, recognized by other registries such as the AOANJRR, notably in women, irrespective of femoral head size.¹⁶ These two registries also identified similar findings regarding MoM bearing surfaces in THA, for which 1 in 5 of these articulations needed revision 10 years after the index procedure because of MoM wear-related issues. In a similar mechanism of failure, femoral stems with a modular neck in the 2012 AOANJRR report showed a 7.4% revision rate at 5 years across all similar featured stem designs, 10.6% at 10 years, which was twice the failure rate of other contemporary stems.¹⁷ These results are just a few examples of the value of clinical registries in implant surveillance, which has led to corroboration of results within the orthopaedic community, resulting in improvement of quality, safety, and efficacy of care.

Fundamental to value-based health care is the idea that an intervention should be measured by the outcome achieved for the patient and whether or not it successfully meets their needs. Substantial progress has been made in the past decade to measure and understand clinical parameters in orthopaedics more relevant to patients through use of PROMs, and PROMs are increasingly being collected in clinical registries: to date, approximately 18 current orthopaedic arthroplasty registries are collecting PROMs.¹⁸ Measurements that include the patient’s perspective, often in the form of surveys/questionnaires, have become more relevant as the focus of health care systems transitions to improving quality, value, and outcome-based patient care. Generalized PROMs focus on assessing physical, mental, and social qualities of health to gauge overall health and quality of life. More specific PROMs aim to measure additional features related to a specific disease or intervention, such as osteoarthritis in a patient who underwent TKA¹⁸ (Table 3). To the patient, PROMs increase communication to clinicians regarding treatment outcomes and effects on quality of life, facilitating open decision making between both patient and clinician on future decision making together. Clinicians utilize PROMs to compare performance with established standards of care and regulate changes in the health of patients, again facilitating improved communication with the patient. Healthcare organizations also use PROMs collectively to monitor and compare performance with other organizations; this allows recognition in areas of deficit and subsequent vital feedback to quality improvement initiatives. At the highest level, PROMs allow health system policymakers to understand outcomes at local to international levels over time. By using this information, the advantages and disadvantages of the different models of care can be compared, prompting necessary changes to increase value-based health care.¹⁰,¹¹,¹⁸