Using a Learning Health System to Improve Physical Therapy Care for Patients With Intermittent Claudication: Lessons Learned From the ClaudicatioNet Quality System.

Routinely collected outcomes data can be used to improve physical therapy care through benchmarking, personalization, continued education, and treatment optimization. This article describes how a nationwide infrastructure to routinely collect data from daily practice was created and how these data were used through a support system (called the ClaudicatioNet Quality system) to improve physical therapy care for patients with intermittent claudication in the Netherlands. ClaudicatioNet is a nationwide network of 2100 specialized physical therapists, providing high-quality supervised exercise therapy in combination with lifestyle counseling. The ClaudicatioNet Quality system uses a large national registry in which specific relevant health outcomes have been routinely collected since 2015. These data have then been used in turn to assess quality of care and provide transparency to therapists and other stakeholders. The Quality system is intended to serve as a learning health system, to support continuous learning at the therapist, practice, and network level. In this approach, individual patients and physical therapists are provided with opportunities to personalize, benchmark, and evaluate (and possibly alter) a treatment plan using routinely collected data from historical patients. The Quality system is described based on the essential elements of a learning health system. The challenges and lessons learned in developing the Quality system also are described. IMPACT: The use of routinely collected health outcomes can, if implemented correctly, facilitate continuous learning among physical therapists and contribute to person-centered care. This example of a learning health system might serve as a blueprint for physical therapists on how to optimally implement and distill meaning from routinely collected clinical data.

Introduction

Routinely collected outcomes data in physical therapy offer opportunities to improve the quality of care. Improvements may be achieved directly, by using these data to aid therapists in personalizing and optimizing their treatment plans with patients, or indirectly, by using these data to help therapists reflect on their practice through benchmarking of performance or by using data to develop new educational courses to stimulate continuous learning cycles.

To date, a number of large physical therapy outcomes registries have already been established; for instance: the Physical Therapy Outcomes Registry of the American Physical Therapy Association, and the National Register for Physical Therapy of the Royal Dutch Society for Physical Therapy (KNGF). However, to fully harness their potential for optimizing care, we believe it is essential to integrate these registries with health care networks. Such an integration transforms a regular health care network into a learning health system (LHS). An LHS is defined as a system in which (routinely collected) information is used for continuous improvement and innovation.,

Becoming an LHS can be quite challenging. There are numerous steps involved with creating a data acquisition infrastructure, such as deciding which outcomes to measure, implementing these outcomes in daily practice, dealing with the wide variety of electronic health records (EHRs), and continuously stimulating therapists to collect accurate and complete data alongside their regular duties.,, Synthesizing the data from daily practice in a relevant manner is also a substantial task, requiring at least time and analytical expertise. Finally, feeding the results of the data synthesis back to practitioners in a such a way as to stimulate learning and subsequent behavior change is additionally challenging and also benefits from expertise in implementation science and behavioral design, as well as systems-level incentives (eg, the requirement of practitioners to engage in continuing education). In this article we aim to describe how we transformed the ClaudicatioNet care network into an LHS with the goal of further improving physical therapy care for patients with intermittent claudication in the Netherlands. Furthermore, we aim to share our own insights regarding this transformation process. Through this specific example, we are hoping to help readers appreciate the complexities involved in the transformation into an LHS.

Usual Physical Therapy Care for Patients With Intermittent Claudication

In the Netherlands, progress has been made in routinely gathering outcomes data of patients with intermittent claudication. Intermittent claudication is defined as walking-induced discomfort or pain, which disappears after a brief period of rest. Intermittent claudication is the most common presentation of lower extremity peripheral arterial occlusive disease, a chronic disease caused by atherosclerotic narrowing of the arteries in the lower limbs., Supervised exercise therapy combined with lifestyle modification is recommended in international multidisciplinary guidelines as the primary treatment for patients with intermittent claudication., In the Netherlands, these guideline recommendations are realized in a so-called stepped-care approach. This approach aims to initially refer patients for supervised exercise therapy and provide invasive treatments only to nonresponders.,

Steps Toward a Learning Health System Based on Routinely Collected Data

The ClaudicatioNet Quality system is an export system for collecting patient, process, and outcomes data of all patients with intermittent claudication who receive supervised exercise therapy within ClaudicatioNet. Pseudo-anonymized data are gathered based on the National Register for Physical Therapy of the KNGF. The aim of the ClaudicatioNet Quality system is to provide transparency and guarantee quality of care for patients with intermittent claudication using these routinely collected data. The ClaudicatioNet Quality system is part of ClaudicatioNet, a network of specialized physical therapists to provide accessible and evidence-based care for all patients with intermittent claudication in the Netherlands. Nowadays, ClaudicatioNet comprises more than 2100 specialized therapists who treat over 10,000 patients with intermittent claudication annually. To participate in the network, physical therapists have to meet certain criteria. Criteria are related to their knowledge of exercise and lifestyle interventions and include a baseline training on treating patients with intermittent claudication.

To describe ClaudicatioNet’s transformation from a guideline-based physical therapy care network into a transparent, data-driven, personalized physical therapy care network, we use the LHS framework.,,, LHSs go beyond data collection for the purpose of policy making and/or research, because they strive to use data for optimizing the care processes within the health care network. To do so, 5 attributes are essential: (1) collaborating with people who are intrinsically driven to improve health care; (2) creating a data infrastructure; (3) gathering data on health outcomes; (4) using knowledge derived from these data; and (5) initiating a continuous process of health care improvement.,,, For each attribute, challenges and lessons learned will be discussed.

Collaborating With People Who Are Intrinsically Driven to Improve Health Care

According to Friedman et al, people with intrinsic motivation to improve health care are crucial to successfully operate an LHS. ClaudicatioNet was originally founded by a vascular surgeon motivated to improve health care for patients with intermittent claudication. The scope of ClaudicatioNet extended beyond the vascular surgery discipline. The aims of ClaudicationNet actually had potential negative consequences for vascular surgery output, because programmatic success would result in fewer invasive interventions performed. Besides the single intrinsically driven founder, a dedicated team in collaboration with different stakeholders was necessary to set up ClaudicatioNet as an LHS. Furthermore, knowhow on nationwide collection of relevant health outcomes was required, as well as expertise in implementation of resulting knowledge for health care providers using, for instance, information and communications technology solutions.

ClaudicatioNet is run by a team comprised of a variety of people with a broad range of knowledge, expertise, and education. The team includes project managers who have knowledge of translating research into practice, researchers who provide best practice evidence, and physical therapists with practical experience who know and understand the problems from daily practice (knowledge brokers).

Physical therapists are not only represented on the board and team, but also are deployed as trainers in different courses. The nationwide network of ClaudicatioNet is subdivided in 55 regional networks. Each regional network has at least 1 senior physical therapist, responsible for the distribution of knowledge and organization of regular meetings to exchange knowledge. Senior physical therapists serve as knowledge brokers and are able to pass on knowledge and new insights from clinical practice to the ClaudicatioNet team and the other way around. Besides the physical therapists with a specific operational function within the network, all therapists affiliated with the network as providers are indispensable to its function as an LHS. Without a sufficient number of therapists, nationwide coverage is not possible and the network would not be able to make supervised exercise therapy available and accessible for all patients with intermittent claudication.

To create a network like ClaudicatioNet, collaboration with different stakeholders is important. In our experience, this has included patients, professional bodies, and web-development companies. Patients have been involved in the network, for example, through their physical therapists, specific patient surveys, and focus group meetings to assess the performance of various programmatic initiatives. Also, collaboration with the Dutch patient federation for people with cardiac and vascular disease was important to incorporate the patients’ perspective. For example, the collaboration with the KNGF resulted in a collective update of clinical practice guidelines for the treatment of intermittent claudication, sufficient reimbursement of supervised exercise therapy for patients with intermittent claudication, and an adequate and efficient development of the Quality system by use of the KNGF infrastructure to collect data from EHRs. Close collaboration with a web-development company was also important to create and regularly update a website and digital platform, including all specific functionalities like individual online portfolios for therapists to make use of an online referral system and the Quality system.

Intrinsic motivation to improve the care for patients of collaborating stakeholders is important and often self-evident, because patient care is the core business of health professionals and patient associations. To enhance efficient collaboration, ClaudicatioNet learned to define common grounds in early stages of collaboration and discuss the added value of creating an LHS for all stakeholders.

Creating a Data Infrastructure

Routinely measured and documented data of sufficient quantity and quality are a prerequisite to successfully build and operate an LHS. Some important lessons were drawn from our experiences with gathering data via spreadsheets, which was the initial practice before the development of the Quality system. Though laudable in the effort, manual data entry into spreadsheets proved cumbersome, error-prone, and distracting to many of the important daily tasks of care providers. With regard to the data infrastructure, 2 points will be discussed: (1) the importance of integrating data collection into daily practice, and (2) the need for uniform data collection.

First, to implement routine data collection, it should be embedded as invisibly as possible in daily practice. Figure 1 is a schematic overview of the LHS, including data collection via the Information and Communications Technology infrastructure. ClaudicatioNet receives data from EHRs, which physical therapists use to register patient information and health outcomes. For therapists to adhere to data collection, the administrative workload needs to be as low as possible, because the required documentation burden is already quite high and arguably distracts from the primary goal of providing quality care. A major advantage of data collection via EHRs is the potential for minimal additional workload for the therapist.

Figure 1

Data infrastructure of the ClaudicatioNet Quality system. Data are gathered by ClaudicatioNet physical therapists into their Electronic Health Records (EHRs). Data from all physical therapy EHRs in the Netherlands are collected by KNGF in the National Register for Physical Therapy. ClaudicatioNet receives this pseudo-anonymized data from KNGF to support continuous learning. The National Registry for Physical Therapy collects data from Electronic Health Records (EHR) of not only ClaudicatioNet therapists. However, ClaudicatioNet receives only data delivered by therapists affiliated with the network. KNGF = Royal Dutch Society for Physical Therapy.

Second, data collection via EHRs enables uniform data collection. The set of health outcomes (see “Gathering Data on Health Outcomes” below) is available for all therapists via standardized measures and questionnaires. Limitations in scoring options prevent documentation errors due to variation in interpretations of the measures and questions.

ClaudicatioNet has collaborated with KNGF since 2015 to implement uniform measures and questionnaires in all different EHR systems available for therapists. Thereby, ClaudicatioNet is able to access routinely collected health data from different EHRs. In 2015, KNGF started the implementation of an infrastructure to collect data from all physical therapy EHRs in the Netherlands (18 in total), called the National Register for Physical Therapy. This register was initiated to collect data from EHRs to improve patient-centered care and effectiveness of physical therapy care. For patients that provide informed consent, pseudo-anonymized data from EHRs are sent to the National Register for Physical Therapy. Therapists enter patient information and outcomes into the EHRs. The variables selected as valuable for the National Register for Physical Therapy are pushed from the EHRs to the register by the therapists. Technical specifications describe how data can uniformly be transferred from the EHRs to the National Register and processed by ClaudicatioNet. Challenges that may arise when collecting data through physical therapists’ EHRs have been described by Meerhoff et al; the 2 most prominent challenges are user-unfriendly EHR systems and lack of integration of outcome measures with patient records.

Gathering Data on Health Outcomes

Apart from establishing a data collection and management infrastructure, it is essential to decide what data are relevant to collect. Over several years, ClaudicatioNet has undertaken a challenging endeavor to select health outcomes that are most relevant to the Quality system’s goals. Despite a well-intentioned urge to gather comprehensive data to inform research or policy questions, ClaudicatioNet reduced the number of measures and questionnaires to a bare minimum and aligned the data collection efforts primarily toward what could be useful to inform real-time clinical decisions (eg, improving clinicians’ ability to monitor therapy progress).

Based on expert opinion and in line with current guidelines, relevant health outcomes for the Quality system were selected and refined over time based on user feedback. This resulted in the 3 major categories of clinical data: patient characteristics, outcome results, and process data. Patient characteristics include sex, age, weight, and height. Outcome results include certain patient-reported outcome measurements (PROMs), as well as assessments of smoking behavior, willingness to change lifestyle behavior, walking distance (measured as functional and maximal walking distance using the Gardner-Skinner protocol), quality of life (measured with the Vascular Quality of Life Questionnaire), and patients’ perspective on recovery, executing activities of daily living, and treatment. PROMs are valuable to improve health care by incorporating the patient’s perspective and create more personalized health care. Process data include duration of treatment, number of treatment sessions, and whether or not the treatment goal has been achieved. Measurements are administered and documented for each patient every 3 months, beginning at the start of the treatment, up to a maximum of 12 months.

Even though this measurement protocol has been in practice since 2015, it remains challenging to collect data from all ClaudicatioNet therapists (Fig. 2). In 2017, data were received from only approximately half of all associated therapists. Because collection of at least the minimal set of outcomes data is a prerequisite for ClaudicatioNet therapists, several actions were undertaken to increase the completeness of the data. Data collection optimization was initiated using strategies at the group level, followed by strategies at the individual level. Strategies to optimize data collection were mainly driven by knowledge and experience of the network, observations from daily practice, and feedback of therapists. At the group level, data collection was stimulated by:

Figure 2

Timeline of the development of the registry used by the Quality system. The horizontal axis represents the time divided into years, ranging from 2015 to 2020. The left vertical axis represents the number of therapists providing data to the Quality system, and the right vertical axis represents the number of patients for whom data were available at that time. Important milestones over time are described underneath the graph.

Increasing therapist understanding of the purpose of data gathering. A lack of awareness of the overall purpose of data gathering can contribute to anxiety on the part of therapists. Knowledge on the purpose of data gathering is increased through online training and presentations, provided by the network.

Providing information on how therapists can use data in therapy sessions and how health care data can be used to improve quality of care. The goal was to increase therapists’ intrinsic motivation to gather data. We provided information on the usefulness of data gathering through the online portfolios, oral presentations, emails, and regular news letters.

Making therapists aware of the minimal administrative workload to send the data to the national register. We aim to keep administrative workload as low as possible. However, therapists were generally unaware of the ease with which the requested data could be entered into EHRs. The misconception that multiple onerous steps were needed was a barrier to data entry.,, We attempted to increase awareness through emails and regular newsletters, sent by the network.

To reduce missing data at the individual (therapist) level, information on personal data collection was communicated with individual therapists. Such feedback was initially provided via personalized emails on whether data were received or not. Thereafter, we provided feedback through a signaling system in their personal account on the ClaudicatioNet website and later we added personalized feedback on the completeness of the received data. The signaling system was designed to automatically send reminders to individual physical therapists who did not transfer any data in the previous 2 months. Additionally, the system will send a warning in case no data have been collected for a period of 6 months and will temporarily suspend therapist participation after 3 warnings. The personalized feedback includes information on whether data have been received or not, as well as information on data completeness and the content of collected data. The feedback is provided via tables as well as in email form. We learned that feedback on data collection could be a driver for therapists to initiate data collection or improve data completeness (Fig. 2).

Besides unawareness of the added value of data collection, missing data may have other causes, including: lack of follow-up of patients, not entering the follow-up data into the EHRs, or lack of adherence to the measurement protocol. The measurement protocol recommends follow-up measurements and data collection at least every 3 months for a 1-year period and describes the relevant measures., Substantial efforts have been made to draw attention to these guidelines and to achieve implementation of the recommendations that therapists routinely collect data for the patient, process, and outcomes domains. The message conveyed by the network was that performing these measurements and documentation should be part of daily clinical practice and is valuable for the treatment process.

Using Knowledge Derived From These Data

A sufficient amount of routinely collected data is necessary to initiate learning cycles consisting of: transformation of data into knowledge, transformation of knowledge into performance, and ultimately the transformation of new performance into new data.

ClaudicatioNet Quality system data are used to initiate learning cycles to answer specific questions from physical therapy practice. Data are used to learn from and improve individual performance, as well as to influence policy making. Data from improved practice are gathered in the Quality system, which completes the initial learning cycle and creates the opportunity to start a new one. There are several examples of how data in the Quality system are used to improve practice:

Monitor the overall quality at a network level, in terms of therapy outcomes and cost-effectiveness of the treatment.

Create benchmarks at the national and regional level. This allows for benchmark comparisons, whereby the variations in therapy outcomes over time and among therapists can be readily monitored. The characteristics and circumstances of therapists under- or overperformance may be explored. These national and regional benchmarks contribute to learning at all levels of the system: individual therapist, practice, and network.

Visualize pseudo-anonymized data on individual therapist treatment results. These so-called visuals can be used to learn at both group and individual therapist level. Therapists may be stimulated to evaluate their own data with respect to the benchmarks and to discuss their evaluations with other ClaudicatioNet therapists. For example, mean outcomes results over time of individual therapists, as well as regional and national averages are shown (Fig. 3). Collaboration with a web-development company has been essential to create these visuals and enable quick, easy, and meaningful benchmark comparisons.

Figure 3

Example of a visualization of an individual therapist’s results (in terms of walking distance) benchmarked to the average regional and national results.

Provide insight into patients’ individual prognosis. Routinely collected data are used to provide insight into patients’ individual expected outcome of the supervised exercise therapy using personalized outcome forecasts (Fig. 4). Although evidence for optimal treatment content is still derived from guideline recommendations, insight into an individual prognosis may support patients and therapists to align the treatment plan to the needs of the patient. Moreover, these personalized outcome forecasts can elicit shared decision-making and improve clinical reasoning of the therapist, potentially resulting in more realistic and personalized treatment goals and interventions, as well as (importantly) improved monitoring of progression., To further improve physical therapy care for people with claudication intermittens, we recently initiated a project aimed to integrate the personalized outcome forecasts with guideline recommendations. Guidelines can have the negative side effect that they not only reduce the unwanted practice variation, but also the wanted treatment variation in practice. By integrating the personalized outcome forecasts with the guideline recommendations, we aim to facilitate therapists to make their treatment plans more personalized and participatory, whilst still following the guideline recommendations. Not only will this potentially result in better outcomes, it will also set the stage for continually improving and updating the guideline recommendations themselves. If we succeed, our learning health system will be transformed even further, namely into an evidence ecosystem. This innovative project is only made possible through collaborations with relevant stakeholders.

Figure 4

Two examples of personalized outcome forecasts for 2 different patients. Personalized outcomes forecasts are individual estimates of a patient’s outcome over time, visualized as plots. For people with intermittent claudication, we developed personalized outcomes forecasts, which estimate the walking distance during a trajectory of supervised exercise therapy (blue shade and black median line). Outcomes forecasts are based on historic data of patients similar to the index patient (dark-blue lines). Actual patients’ outcomes are also included in the graphs (orange lines) to enable monitoring therapy progress. In this specific example the left graph (A) represents a personalized outcome forecast for a 75-year-old female patient who smokes and has a strongly impaired functional walking distance at baseline; the graph on the right (B) represents a 62-year-old male who smokes and has a moderate impaired walking distance at baseline.

Initiating a Continuous Process of Health Care Improvement

Routine data collection and use of these data to improve health care should be continuous processes. Data on every new patient added to the Quality system should contribute incrementally to improvements in the registry, representation of the experiences of patient population writ large, and enhanced knowledge and perception on data gathering. For example, data on new patients in the Quality system should, over time, result in more accurate personalized outcome forecasts for individual patients.

To continuously improve, new insights and knowledge from data should be made accessible to therapists, patients, and relevant stakeholders. Over the past years, ClaudicatioNet has learned that communication on new insights or innovations is important to successfully transfer knowledge and implement change. ClaudicatioNet communicates weekly on the progression of projects to create early awareness through newsletters, websites, and social media. Furthermore, ClaudicatioNet uses instructional videos to make information easily accessible, for both patients and therapists. With regard to the personalized outcome forecasts, patient-therapist interaction videos have also been developed based on observations and input from interactions in daily practice. Besides these attempts at good communication, usage of therapists’ portfolios, which are known platforms, might enhance the success of implementation. Therefore, personalized outcome forecasts are embedded in the online portfolios of therapists.

The ClaudicatioNet website, social media, newsletters, annual congresses, trainings, and regional meetings are important ways to distribute knowledge. The ClaudicatioNet website contains information on the network: the organizational structure, quality criteria, and regular project updates. Additionally, the website provides news on exercise and lifestyle interventions from other resources, which are accessible for therapists, referral sources, and other interested people. Regional projects are used to transfer knowledge at a group level, for example, peer assessment meetings. Peer assessment meetings are organized by ClaudicatioNet and include a small group of physical therapists aiming to exchange experience and knowledge. During these meetings, expertise is developed in assessing physical therapist behavior through self-assessment and peer feedback.,

To continuously learn, questions and obstacles from (ever-improving) daily practice should be a direct input for new impulses to improve, including training and research. Questions or needs from daily practice should create inputs for new scientific research. For example, a historical lack of reimbursement for physical therapy services (resulting in low referral rates of patients with intermittent claudication for supervised exercise therapy) spurred scientific research about the overall cost efficiency of supervised exercise within ClaudiocatioNet. This resulted in increased reimbursement for supervised exercise therapy for all patients in the Netherlands who are referred to specialized ClaudicatioNet physical therapists, because ClaudicatioNet data informed the cost efficiency argument. Besides the input from daily practice for new impulses to improve, all results from collaborating, data gathering, data management, and generating and distributing knowledge represent continuous inputs for new learning cycles. For example, recent data have illuminated certain barriers to gathering PROMs in routine practice. Ideally, PROMs should be reported directly by the patients and gathered as part of daily practice. However, misplaced incentives (ie, an organizational desire to show patient improvement) as well as a lack of time and sufficient technology are barriers to obtaining accurate PROMs data in registries., These barriers should be addressed in the future to improve data gathering and the efficacy of the LHS.

Further Development

The ClaudicatioNet Quality system tries to continuously improve and expand data collection. Data collection could be expanded in several different areas: remote data, other diseases, and other disciplines. First, it would be valuable to include remote data, measured by the patients themselves. There is growing interest in remote patient monitoring within physical therapy. Activity trackers or smartphone applications have great potential in physical therapy in general to remotely monitor patients and support therapists in personalized coaching. However, remote patient monitoring specifically for patients with intermittent claudication is not available yet in the Netherlands. Remote data could be useful to further optimize and personalize treatment for patients with intermittent claudication. Second, data collection could be expanded by collecting data of patients referred for physical therapy with other types of chronic diseases than intermittent claudication. To do so, Chronic CareNet was introduced in March 2020. Chronic CareNet builds on the lessons learned from the ClaudicatioNet network and transfers obtained knowledge to set up new networks for other noncommunicable, chronic diseases. The core value of Chronic CareNet is to provide “the right care in the right place,” meaning that all patients with noncommunicable chronic disease for whom physical therapy care is indicated should receive evidence-based care., This introduction of Chronic CareNet now makes it possible to extend the nationwide data collection to other chronic patient groups. Finally, data collection could be expanded to other allied health professions. Data gathering of other allied health professionals could be valuable and stimulate interdisciplinary treatment.

Conclusion

In this article we have discussed how physical therapy care for patients with intermittent claudication in the Netherlands has shifted from generalized guideline-based physical therapy care toward transparent, personalized, evidence-based physical therapy care, using routinely collected data. ClaudicatioNet aims to continuously educate specialized therapists to provide optimal supervised exercise therapy for patients with intermittent claudication in the Netherlands. The initiation of the Quality system has enabled the use of routinely collected data to improve and personalize care. Several lessons can be drawn from initiating ClaudicatioNet and the process of routinely collecting data. An intrinsically motivated team, with a broad range of knowledge and expertise is required. Furthermore, collaboration with intrinsically motivated stakeholders can be beneficial and more efficient to achieve the goals of an LHS, as well as the alignment of the LHS with existing initiatives. To use routinely collected data to continuously learn and improve, data of sufficient quality and quantity are prerequisite. Therefore, data collection should be uniform with minimization of missingness and errors. Finally, data should be transformed into knowledge, leading to new performances in daily practice and new data. To do so, knowledge derived from data should be made applicable for therapists, patients, and stakeholders.