Data Sharing and Data Registries in Physical Medicine and Rehabilitation

Carmen E. Capó-Lugo; Abel N. Kho; Linda C. O'Dwyer; Marc B. Rosenman

doi:10.1016/j.pmrj.2017.04.003

Data Sharing and Data Registries in Physical Medicine and Rehabilitation

Carmen E. Capó-Lugo^*, Abel N. Kho, Linda C. O'Dwyer, Marc B. Rosenman

^*Corresponding author for this work

Research output: Contribution to journal › Review article › peer-review

1 Scopus citations

Abstract

The field of physical medicine & rehabilitation (PM&R), along with all the disciplines it encompasses, has evolved rapidly in the past 50 years. The number of controlled trials, systematic reviews, and meta-analyses in PM&R increased 5-fold from 1998 to 2013. In recent years, professional, private, and governmental institutions have identified the need to track function and functional status across providers and settings of care and on a larger scale. Because function and functional status are key aspects of PM&R, access to and sharing of reliable data will have an important impact on clinical practice. We reviewed the current landscape of PM&R databases and data repositories, the clinical applicability and practice implications of data sharing, and challenges and future directions. We included articles that (1) addressed any aspect of function, disability, or participation; (2) focused on recovery or maintenance of any function; and (3) used data repositories or research databases. We identified 398 articles that cited 244 data sources. The data sources included 66 data repositories and 179 research databases. We categorized the data sources based on their purposes and uses, geographic distribution, and other characteristics. This study collates the range of databases, data repositories, and data-sharing mechanisms that have been used in PM&R internationally. In recent years, these data sources have provided significant information for the field, especially at the population-health level. Implications and future directions for data sources also are discussed.

Original language	English (US)
Pages (from-to)	S59-S74
Journal	PM and R
Volume	9
Issue number	5
DOIs	https://doi.org/10.1016/j.pmrj.2017.04.003
State	Published - May 2017

ASJC Scopus subject areas

Physical Therapy, Sports Therapy and Rehabilitation
Rehabilitation
Neurology
Clinical Neurology

Access to Document

10.1016/j.pmrj.2017.04.003

Cite this

@article{b904662543114556a00a0e0b12ba189c,

title = "Data Sharing and Data Registries in Physical Medicine and Rehabilitation",

abstract = "The field of physical medicine & rehabilitation (PM&R), along with all the disciplines it encompasses, has evolved rapidly in the past 50 years. The number of controlled trials, systematic reviews, and meta-analyses in PM&R increased 5-fold from 1998 to 2013. In recent years, professional, private, and governmental institutions have identified the need to track function and functional status across providers and settings of care and on a larger scale. Because function and functional status are key aspects of PM&R, access to and sharing of reliable data will have an important impact on clinical practice. We reviewed the current landscape of PM&R databases and data repositories, the clinical applicability and practice implications of data sharing, and challenges and future directions. We included articles that (1) addressed any aspect of function, disability, or participation; (2) focused on recovery or maintenance of any function; and (3) used data repositories or research databases. We identified 398 articles that cited 244 data sources. The data sources included 66 data repositories and 179 research databases. We categorized the data sources based on their purposes and uses, geographic distribution, and other characteristics. This study collates the range of databases, data repositories, and data-sharing mechanisms that have been used in PM&R internationally. In recent years, these data sources have provided significant information for the field, especially at the population-health level. Implications and future directions for data sources also are discussed.",

author = "Cap{\'o}-Lugo, {Carmen E.} and Kho, {Abel N.} and O'Dwyer, {Linda C.} and Rosenman, {Marc B.}",

note = "Funding Information: These findings provide a unique resource for those interested in databases and data sharing as mechanisms to advance the field of PM&R. This study collates, for the first time, the range of databases, data repositories, and data-sharing mechanisms that have been used in PM&R internationally. Although the development of initiatives to gather high-quality data in large sharable databases started a couple of decades ago, in recent years these databases have provided significant information especially at the population-health level. From a clinical perspective, many studies have used databases to assess quality of care, describe current practices, and assess new interventions through efficacy trials. An in-depth look at databases that have been established for more than 15 years, like the Model Systems Knowledge Translation Center (MSKTC) [50] , show that these endeavors have resulted in significant advances that have had or will have the potential to modify clinical practice. The MSKTC has contributed to the development of: (1) practice guidelines in multiple areas, (2) recommendations on best clinical practices regarding personalized rehabilitation interventions, and (3) mechanisms to better measure physical function and quality of life, among many other clinically relevant contributions [50] . In fact, the MSKTC has resulted in about 1900 publications in the areas of spinal cord, traumatic brain, and burn injuries. Although these staggering numbers support the idea that large longitudinal datasets effectively and efficiently support high-quality, clinically relevant research, the high volume also presents an interesting and pressing difficulty. The accumulation of scientific knowledge is increasing dramatically, and clinicians, as well as researchers, struggle to keep up with findings that are relevant to their practices. High-quality research that lacks a knowledge-translation component does not in and of itself create tangible and sustainable changes in clinical practice. It seems reasonable then to expect—and for publicly financed institutions to demand—that in the future these large databases not only will serve as repositories of scientific data and publications but also will summarize and disseminate findings in ways that increase uptake and usability of scientific knowledge. For clinicians, patients, caregivers, and policy makers, it would be ideal to have access to interactive tools that present and summarize the scientific findings, the level of evidence that supports the variety of available interventions, and potentially the relevancy of data to different ethnic, cultural, or racial groups. Also, new implementation/dissemination systems could be developed to integrate EHR and research databases to alert health care providers about the best-available treatments supported by evidence in their field. In research, as well as in education and training, these same interactive tools could be used to present a “network of knowledge” that details research findings (eg, with regard to the outcomes of interventions), level of evidence that supports such findings, gaps in knowledge, and even connections across research areas that would be otherwise missed [52] . We support the idea that large databases and data-sharing endeavors should aim to become “learning systems” that are integrated with both health care delivery and research. Although efforts to develop these learning systems are underway, health care systems, policies, and regulations continue to change. For example, within the United States there is increasing pressure from the Federal Government to measure patient outcomes, as seen with PQRS as well as the upcoming Quality Payment Program [53] . Health care providers are expected to justify the services they provide to their patients. Physicians and rehabilitation providers will need to demonstrate that their interventions positively impact patient outcomes while at the same time containing health care costs. PM&R, as a specialty, needs to use these databases to determine which rehabilitation-specific outcomes or assessments should be used to quantify the care given for their patients and to justify intensity of service for varying levels of rehabilitation care settings. Hence, practitioners, researchers, and organizations must get involved in data-sharing endeavors, either at institutional, local, national, or international levels, as users and/or direct contributors of data, as well as developers of quality and value measures. The opportunities to contribute to these endeavors are vast: the American Physical Therapy Association launched in February 2017 its Physical Therapy Outcomes Registry to gather discipline-specific data, and it currently is enrolling facilities [49] . Similarly, the American Academy of Physical Medicine and Rehabilitation and the American Association of Neurological Surgeons are actively seeking participant sites for their Spine and Quality Outcomes Database [54] . Also, organizations have formed task forces to (1) determine the best outcome measures according to setting of care, patient acuity, and training purposes; (2) establish optimal standards of evidence and research methodologies; and (3) develop value-based interventions and knowledge translation [55-59] . Other opportunities include participation of PM&R practitioners and clinician-scientists at national and international events such as those organized by AcademyHealth [60] . AcademyHealth holds 2 annual events, Health Datapalooza and a research meeting, in which data are used in innovative ways that inform policy and practice [60] . From a research perspective, there are many data sources that can be queried to address PM&R-related questions. The online resource Disability Statistics is an example [61] . It is a Web site that condenses several disability data sources, research tools, and reports into a single interactive repository. The National Institutes of Health funded an initiative in 2016, the Archive of Data on Disability to Enable Policy and Research, to support disability and rehabilitation research [62] . This archive is hosted by the Center for Large Data Research and Data Sharing in Rehabilitation (CLDR) [63] and the Interuniversity Consortium of Political and Social Research (ICPSR) [64] , the largest social science data archive in the world. The collaboration between CLDR and ICPSR will leverage CLDR{\textquoteright}s expertise on evidence-based rehabilitation practices and ICPSR{\textquoteright}s expertise on data repository infrastructure. This archive not only will serve as a hub for data sharing but also will provide education and training opportunities [62] . To date, however, there are few data sources that can be used to address PM&R-related questions across the translational continuum, from foundational or preclinical studies to population and global health, or across settings of care. The use of the same cohort throughout the translational continuum (ie, from bench to bedside), as well as across settings of care, could provide insights into the impact of interventions at both the individual and population levels. Although foundational or preclinical studies tend to be conducted in animal models, there are molecular and genetic components relevant to rehabilitation interventions that are rarely addressed through large sharable datasets. An example is the growing number of studies regarding exercise and brain-derived neurotrophic factor (BDNF). BDNF is a protein that has critical roles in neuroplasticity of the brain (learning and memory) in healthy and disease states, in young and as well as in older populations [65-68] . Interestingly, the concentrations of BDNF can be increased by exercise and certain rehabilitation interventions [66,69] . This line of research has moved from preclinical studies to randomized controlled trials, but there has been a slow progression toward efficacy and effectiveness trials of exercise. This example illustrates a scenario in which PM&R could benefit from and contribute to precision medicine endeavors to determine the right rehabilitation treatment at the right time that leads to the best-possible outcomes. In this case, as in many others, precision medicine and large, sharable databases might result in more efficient and efficacious care. To accomplish this goal, there are several issues that must be addressed, such as the incorporation and interpretation of molecular data in clinical care, the systematic input and/or analysis of clinical data, and gathering and analysis of such data in a continuous way across settings of care [70] . This article pinpoints several limitations of the use and generalizability of currently available databases. The most notable limitation is that the geographical distribution of published articles is not representative of that of the world population. The majority of studies arise from the United States, followed by Australia, New Zealand, and the United Kingdom. Hence, the results of population-level studies might not be generalizable globally. The effects of location, environment, and ethnic/cultural practices on health and recovery or maintenance of function require additional explication. There is also variability in access to rehabilitation and other health care services across (and within) countries. Some developed countries are exploring new health care-delivery models such as tele-medicine and tele-rehabilitation. Some of these models of care involve minimal-to-no contact with health care providers; these models might limit the availability of systematically measured outcomes [71] . Hence, large databases, as well as guidelines for CDEs, will require new methods and updates to account for these changes, including the addition of patient-reported outcome measures. Creating methods to collect data from new health care delivery approaches could help move large data-gathering efforts toward a more patient-centered approach but also may hinder comparisons with data obtained previously. Finally, there are several limitations surrounding data sharing, which include efficient storage of data, sustaining data security, ease of accessibility and transparency regarding data transformation and metadata, and granting credit for shared data, to name a few [72,73] . There are perhaps 44 petabytes (44 million gigabytes) of data in the Kaiser Permanente EHR alone [74] . These large volumes of data have created challenges for storage and security, especially for smaller organizations. Data storage and security has been discussed extensively elsewhere [14,74] . Inability or reluctance to store health care data could hinder patient care and research (eg, by limiting longitudinal analyses of outcomes), but inappropriate or unsafe storage of data also could harm patients (eg, via security breaches) or health care delivery (eg, via ransomware attacks) [75] . When data sharing is added to this equation, the challenges expand. For example, how should data be shared between institutions or countries? Where should the shared data be stored and what are the most secure mechanisms? How are patients{\textquoteright} rights protected (eg, was consent provided for international transfer of patient information)? In recent years, answers to these questions are slowly being developed and tested [76] . These issues also affect accessibility and transparency in data sharing and usage. From this narrative review, it is evident that there are a great many datasets, representing different areas of interest (professional fields, clinical conditions, settings of care), and almost every one of them has a distinct mechanism for data access or sharing, as well as unique codebooks, data dictionaries, and CDEs. This lack of uniformity makes it difficult, and sometimes impossible, to merge, analyze, and interpret data in meaningful ways. Barriers caused by cost, time, and limited technical support may prevent the use of these datasets as envisioned initially. These difficulties helped stimulate new endeavors to share data and methods across the United States, as in PCORnet. Data sharing is based on the premise that researchers can access previously collected data to conduct secondary analyses. Hence, it is imperative that the PM&R field develop standards for data use agreements that enable timely research while protecting the rights of investigators and patients. These standards should include provision of credit for generating and sharing datasets, and for envisioning novel ways of using the data [45] . Publisher Copyright: {\textcopyright} 2017 American Academy of Physical Medicine and Rehabilitation",

year = "2017",

month = may,

doi = "10.1016/j.pmrj.2017.04.003",

language = "English (US)",

volume = "9",

pages = "S59--S74",

journal = "PM and R",

issn = "1934-1482",

publisher = "Elsevier Inc.",

number = "5",

}

TY - JOUR

T1 - Data Sharing and Data Registries in Physical Medicine and Rehabilitation

AU - Capó-Lugo, Carmen E.

AU - Kho, Abel N.

AU - O'Dwyer, Linda C.

AU - Rosenman, Marc B.

N1 - Funding Information: These findings provide a unique resource for those interested in databases and data sharing as mechanisms to advance the field of PM&R. This study collates, for the first time, the range of databases, data repositories, and data-sharing mechanisms that have been used in PM&R internationally. Although the development of initiatives to gather high-quality data in large sharable databases started a couple of decades ago, in recent years these databases have provided significant information especially at the population-health level. From a clinical perspective, many studies have used databases to assess quality of care, describe current practices, and assess new interventions through efficacy trials. An in-depth look at databases that have been established for more than 15 years, like the Model Systems Knowledge Translation Center (MSKTC) [50] , show that these endeavors have resulted in significant advances that have had or will have the potential to modify clinical practice. The MSKTC has contributed to the development of: (1) practice guidelines in multiple areas, (2) recommendations on best clinical practices regarding personalized rehabilitation interventions, and (3) mechanisms to better measure physical function and quality of life, among many other clinically relevant contributions [50] . In fact, the MSKTC has resulted in about 1900 publications in the areas of spinal cord, traumatic brain, and burn injuries. Although these staggering numbers support the idea that large longitudinal datasets effectively and efficiently support high-quality, clinically relevant research, the high volume also presents an interesting and pressing difficulty. The accumulation of scientific knowledge is increasing dramatically, and clinicians, as well as researchers, struggle to keep up with findings that are relevant to their practices. High-quality research that lacks a knowledge-translation component does not in and of itself create tangible and sustainable changes in clinical practice. It seems reasonable then to expect—and for publicly financed institutions to demand—that in the future these large databases not only will serve as repositories of scientific data and publications but also will summarize and disseminate findings in ways that increase uptake and usability of scientific knowledge. For clinicians, patients, caregivers, and policy makers, it would be ideal to have access to interactive tools that present and summarize the scientific findings, the level of evidence that supports the variety of available interventions, and potentially the relevancy of data to different ethnic, cultural, or racial groups. Also, new implementation/dissemination systems could be developed to integrate EHR and research databases to alert health care providers about the best-available treatments supported by evidence in their field. In research, as well as in education and training, these same interactive tools could be used to present a “network of knowledge” that details research findings (eg, with regard to the outcomes of interventions), level of evidence that supports such findings, gaps in knowledge, and even connections across research areas that would be otherwise missed [52] . We support the idea that large databases and data-sharing endeavors should aim to become “learning systems” that are integrated with both health care delivery and research. Although efforts to develop these learning systems are underway, health care systems, policies, and regulations continue to change. For example, within the United States there is increasing pressure from the Federal Government to measure patient outcomes, as seen with PQRS as well as the upcoming Quality Payment Program [53] . Health care providers are expected to justify the services they provide to their patients. Physicians and rehabilitation providers will need to demonstrate that their interventions positively impact patient outcomes while at the same time containing health care costs. PM&R, as a specialty, needs to use these databases to determine which rehabilitation-specific outcomes or assessments should be used to quantify the care given for their patients and to justify intensity of service for varying levels of rehabilitation care settings. Hence, practitioners, researchers, and organizations must get involved in data-sharing endeavors, either at institutional, local, national, or international levels, as users and/or direct contributors of data, as well as developers of quality and value measures. The opportunities to contribute to these endeavors are vast: the American Physical Therapy Association launched in February 2017 its Physical Therapy Outcomes Registry to gather discipline-specific data, and it currently is enrolling facilities [49] . Similarly, the American Academy of Physical Medicine and Rehabilitation and the American Association of Neurological Surgeons are actively seeking participant sites for their Spine and Quality Outcomes Database [54] . Also, organizations have formed task forces to (1) determine the best outcome measures according to setting of care, patient acuity, and training purposes; (2) establish optimal standards of evidence and research methodologies; and (3) develop value-based interventions and knowledge translation [55-59] . Other opportunities include participation of PM&R practitioners and clinician-scientists at national and international events such as those organized by AcademyHealth [60] . AcademyHealth holds 2 annual events, Health Datapalooza and a research meeting, in which data are used in innovative ways that inform policy and practice [60] . From a research perspective, there are many data sources that can be queried to address PM&R-related questions. The online resource Disability Statistics is an example [61] . It is a Web site that condenses several disability data sources, research tools, and reports into a single interactive repository. The National Institutes of Health funded an initiative in 2016, the Archive of Data on Disability to Enable Policy and Research, to support disability and rehabilitation research [62] . This archive is hosted by the Center for Large Data Research and Data Sharing in Rehabilitation (CLDR) [63] and the Interuniversity Consortium of Political and Social Research (ICPSR) [64] , the largest social science data archive in the world. The collaboration between CLDR and ICPSR will leverage CLDR’s expertise on evidence-based rehabilitation practices and ICPSR’s expertise on data repository infrastructure. This archive not only will serve as a hub for data sharing but also will provide education and training opportunities [62] . To date, however, there are few data sources that can be used to address PM&R-related questions across the translational continuum, from foundational or preclinical studies to population and global health, or across settings of care. The use of the same cohort throughout the translational continuum (ie, from bench to bedside), as well as across settings of care, could provide insights into the impact of interventions at both the individual and population levels. Although foundational or preclinical studies tend to be conducted in animal models, there are molecular and genetic components relevant to rehabilitation interventions that are rarely addressed through large sharable datasets. An example is the growing number of studies regarding exercise and brain-derived neurotrophic factor (BDNF). BDNF is a protein that has critical roles in neuroplasticity of the brain (learning and memory) in healthy and disease states, in young and as well as in older populations [65-68] . Interestingly, the concentrations of BDNF can be increased by exercise and certain rehabilitation interventions [66,69] . This line of research has moved from preclinical studies to randomized controlled trials, but there has been a slow progression toward efficacy and effectiveness trials of exercise. This example illustrates a scenario in which PM&R could benefit from and contribute to precision medicine endeavors to determine the right rehabilitation treatment at the right time that leads to the best-possible outcomes. In this case, as in many others, precision medicine and large, sharable databases might result in more efficient and efficacious care. To accomplish this goal, there are several issues that must be addressed, such as the incorporation and interpretation of molecular data in clinical care, the systematic input and/or analysis of clinical data, and gathering and analysis of such data in a continuous way across settings of care [70] . This article pinpoints several limitations of the use and generalizability of currently available databases. The most notable limitation is that the geographical distribution of published articles is not representative of that of the world population. The majority of studies arise from the United States, followed by Australia, New Zealand, and the United Kingdom. Hence, the results of population-level studies might not be generalizable globally. The effects of location, environment, and ethnic/cultural practices on health and recovery or maintenance of function require additional explication. There is also variability in access to rehabilitation and other health care services across (and within) countries. Some developed countries are exploring new health care-delivery models such as tele-medicine and tele-rehabilitation. Some of these models of care involve minimal-to-no contact with health care providers; these models might limit the availability of systematically measured outcomes [71] . Hence, large databases, as well as guidelines for CDEs, will require new methods and updates to account for these changes, including the addition of patient-reported outcome measures. Creating methods to collect data from new health care delivery approaches could help move large data-gathering efforts toward a more patient-centered approach but also may hinder comparisons with data obtained previously. Finally, there are several limitations surrounding data sharing, which include efficient storage of data, sustaining data security, ease of accessibility and transparency regarding data transformation and metadata, and granting credit for shared data, to name a few [72,73] . There are perhaps 44 petabytes (44 million gigabytes) of data in the Kaiser Permanente EHR alone [74] . These large volumes of data have created challenges for storage and security, especially for smaller organizations. Data storage and security has been discussed extensively elsewhere [14,74] . Inability or reluctance to store health care data could hinder patient care and research (eg, by limiting longitudinal analyses of outcomes), but inappropriate or unsafe storage of data also could harm patients (eg, via security breaches) or health care delivery (eg, via ransomware attacks) [75] . When data sharing is added to this equation, the challenges expand. For example, how should data be shared between institutions or countries? Where should the shared data be stored and what are the most secure mechanisms? How are patients’ rights protected (eg, was consent provided for international transfer of patient information)? In recent years, answers to these questions are slowly being developed and tested [76] . These issues also affect accessibility and transparency in data sharing and usage. From this narrative review, it is evident that there are a great many datasets, representing different areas of interest (professional fields, clinical conditions, settings of care), and almost every one of them has a distinct mechanism for data access or sharing, as well as unique codebooks, data dictionaries, and CDEs. This lack of uniformity makes it difficult, and sometimes impossible, to merge, analyze, and interpret data in meaningful ways. Barriers caused by cost, time, and limited technical support may prevent the use of these datasets as envisioned initially. These difficulties helped stimulate new endeavors to share data and methods across the United States, as in PCORnet. Data sharing is based on the premise that researchers can access previously collected data to conduct secondary analyses. Hence, it is imperative that the PM&R field develop standards for data use agreements that enable timely research while protecting the rights of investigators and patients. These standards should include provision of credit for generating and sharing datasets, and for envisioning novel ways of using the data [45] . Publisher Copyright: © 2017 American Academy of Physical Medicine and Rehabilitation

PY - 2017/5

Y1 - 2017/5

N2 - The field of physical medicine & rehabilitation (PM&R), along with all the disciplines it encompasses, has evolved rapidly in the past 50 years. The number of controlled trials, systematic reviews, and meta-analyses in PM&R increased 5-fold from 1998 to 2013. In recent years, professional, private, and governmental institutions have identified the need to track function and functional status across providers and settings of care and on a larger scale. Because function and functional status are key aspects of PM&R, access to and sharing of reliable data will have an important impact on clinical practice. We reviewed the current landscape of PM&R databases and data repositories, the clinical applicability and practice implications of data sharing, and challenges and future directions. We included articles that (1) addressed any aspect of function, disability, or participation; (2) focused on recovery or maintenance of any function; and (3) used data repositories or research databases. We identified 398 articles that cited 244 data sources. The data sources included 66 data repositories and 179 research databases. We categorized the data sources based on their purposes and uses, geographic distribution, and other characteristics. This study collates the range of databases, data repositories, and data-sharing mechanisms that have been used in PM&R internationally. In recent years, these data sources have provided significant information for the field, especially at the population-health level. Implications and future directions for data sources also are discussed.

AB - The field of physical medicine & rehabilitation (PM&R), along with all the disciplines it encompasses, has evolved rapidly in the past 50 years. The number of controlled trials, systematic reviews, and meta-analyses in PM&R increased 5-fold from 1998 to 2013. In recent years, professional, private, and governmental institutions have identified the need to track function and functional status across providers and settings of care and on a larger scale. Because function and functional status are key aspects of PM&R, access to and sharing of reliable data will have an important impact on clinical practice. We reviewed the current landscape of PM&R databases and data repositories, the clinical applicability and practice implications of data sharing, and challenges and future directions. We included articles that (1) addressed any aspect of function, disability, or participation; (2) focused on recovery or maintenance of any function; and (3) used data repositories or research databases. We identified 398 articles that cited 244 data sources. The data sources included 66 data repositories and 179 research databases. We categorized the data sources based on their purposes and uses, geographic distribution, and other characteristics. This study collates the range of databases, data repositories, and data-sharing mechanisms that have been used in PM&R internationally. In recent years, these data sources have provided significant information for the field, especially at the population-health level. Implications and future directions for data sources also are discussed.

UR - http://www.scopus.com/inward/record.url?scp=85044760677&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85044760677&partnerID=8YFLogxK

U2 - 10.1016/j.pmrj.2017.04.003

DO - 10.1016/j.pmrj.2017.04.003

M3 - Review article

C2 - 28527505

AN - SCOPUS:85044760677

SN - 1934-1482

VL - 9

SP - S59-S74

JO - PM and R

JF - PM and R

IS - 5

ER -

Data Sharing and Data Registries in Physical Medicine and Rehabilitation

Abstract

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this