Improving Clinical Practice with Person-Centered Outcome Measurement

Abstract

Measurement choices play a critical role in improving healthcare systems. As this book describes, improved measurement can promote excellence in person-centered outcomes through health policy, methodology, theory, and clinical practice. This chapter focuses on measurement choices within clinical practice that can guide decision-making for improved outcomes. The overall objective is for clinical practice to deliver high-quality and equitable healthcare tailored to the unique situation of each specific person. Separate sections discuss the context surrounding measurement choices, specific dilemmas or competing priorities that affect measurement choices, and recommendations for improving person-centered outcome measurement.

4.1 Introduction

“Are we there, yet?” Does clinical practice achieve consistently brilliant results, yet? No. Occasional brilliance shines through: high-quality person-centered outcomes do occur [48], possibly due to brilliant people making extraordinary use of imperfect healthcare systems. These imperfect systems require improvement so that every person in healthcare gets brilliant results. With better systems, ordinarily competent clinicians and informed patients may consistently achieve excellent person-centered outcomes. Achieving excellence requires examining current practice, identifying possible barriers, and bridging remaining gaps. This chapter addresses these issues with a focus on the measurement choices made in clinical practice.

We begin by introducing person-centeredness and the context within which measurers make measurement choices. Three subsequent sections discuss dilemmas or competing priorities that clinicians and clinical researchers must navigate when choosing measures: personalization versus standardization, satisfaction versus effectiveness, and scientific rigor versus practical convenience. Throughout, examples promote understanding or provide evidence of improved clinical practice with person-centered measurement. Discussion includes evaluation of measurement choices that may limit interpretation of the evidence. Case A (below) illustrates the application of measurement choices when managing a multifaceted health condition. The case continues within each section to demonstrate the process of choosing measures while balancing competing interests. This chapter’s final section summarizes and provides recommendations for improving clinical practice with person-centered outcome measurement.

4.1.1 Case A [8]

A person diagnosed with amyotrophic lateral sclerosis (ALS) 14 months ago meets with the multidisciplinary rehabilitation team again to make some clinical decisions. So far in this progressive and incurable disease, the person with ALS is undergoing functional losses in multiple systems. He has lost the ability to get up and down stairs, and needs help to walk on level ground. He gets short of breath after walking 10 feet. He can no longer type well enough to keep working at his software job. He has lost 10% of his body weight because of difficulty swallowing and impaired ability to manipulate a fork or spoon. What outcome measures would be most appropriate to assess his status and guide clinical practice?

Team members from different disciplines can track lab values, caloric intake, and vital capacity to document declining bodily functions; together, the team might use the ALS-Functional Rating Scale-Revised [32] or ALS-FRS-R to record the functional losses, using 12 items scored 0–4 that address different components of everyday function. From the team’s perspective, this standardized scale reliably and validly identifies current capabilities; repeated measures can document the rate of decline over time [8]. From the person’s perspective, the ALS-FRS-R does not ask about his home life, social support, environment, or emotional state; in short, this scale of functional deficits does not reflect the person’s health-related quality of life, which can be quite variable in ALS [94, 142]. The team needs measures that address what matters most to this person. Discussions of declining function and end of life care can challenge both the patient and clinicians [46, 109]; they need measures that facilitate short- and long-term goal-setting and decision-making to optimize quality of life at each stage as the disease progresses.

4.1.2 Person-Centeredness

Person-centeredness has emerged as a critical principle underlying best practice in healthcare. No longer does the physician or other healthcare professional dictate patient management in authoritarian prescription; the individual’s preferences and values take precedence; patients must be engaged as much as possible in the decision-making for their own care [36]. Further, person-centeredness, a more holistic version of patient- or client-centeredness, incorporates the total person and their life apart from their role as a patient or client. In person-centered practice, clinicians must ask people receiving care what they feel, believe, and prefer, and how they perform and respond to the activities they value in their own context, not just in a controlled clinical environment. “Asking” people can involve a number of different methods and types of measurement tools, including interview, self-report questionnaires or rating scales, performance-based measures focusing on valued tasks, biometric equipment, or wearables. Ideally, engaging the whole individual promotes attainment of the common objective: high quality and equitable healthcare tailored to the unique situation of each specific person.

Although person-centeredness underlies best practice, many individuals may never have experienced it. Healthy individuals may go to a healthcare provider based on policy recommendations: an employer or prospective travel may require a physical exam, tuberculosis test, or vaccination. The person only decides on the appointment time. When the person is very ill or has a medical condition that requires surgery or long-term care, emotional distress can affect cognition, cause anxiety, and inhibit the person’s participation in decision-making. Despite that problem, clinicians and the healthcare system interested in person-centered care must diligently strive to elicit the patient’s preferences. Sometimes that means involving a caregiver in the conversation or recording the information so the person can review the details after absorbing an initial diagnosis. The OpenNotes framework initiatives provide an example of greater transparency of communication when clinicians share their notes with patients, with evidence that patients find them valuable [58].

Despite occasional difficulties in extracting accurate information from patients, patient preferences contribute one of the three key tenets of evidence-based medicine or clinical practice along with research literature and clinician judgment [86]. Specifically, in evidence-based decision-making, a systematic review of high quality randomized controlled trials and the clinician’s expert opinion do not take precedence over the preferences of the individual receiving care. A patient cognizant of the literature and clinician judgment may agree to proceed, negotiate modifications, request additional information or a second opinion, or decide against a procedure. Evidence-based practice presumes, however, that the three tenets work in conjunction with each other rather than in opposition, which means that each informs the others. The clinician should know the patient’s preferences before examining applicable literature; researchers should have drawn from patients’ experiences and outcomes in designing data collection and results reported in the literature; and the patient should be fully informed of the applicable literature, their own status, and the clinician’s judgment before making decisions. The best outcome measures to advance person-centered care draw together these three components of evidence-based practice.

The United States (U.S.) formally legislated the importance of person-centeredness and all three tenets of evidence-based practice by creating the Patient-Centered Outcomes Research Institute (PCORI) through the Affordable Care Act of 2010. PCORI has since granted hundreds of millions of dollars to translational healthcare research that engages patients and clinicians at every stage: from conception to research design that includes outcomes meaningful to patients, oversight of data collection, and analysis of patient-centered changes. Further, PCORI mandates that patients contribute to every grant review panel evaluating the merit of proposed research; governmental funding agencies have followed suit in including patients in their grant review panels.

PCORI’s mission, to improve outcomes important to patients, has contributed to the increasing emphasis in healthcare to keep individual patients front and center, first and always. ‘Important to patients’ means that assessment and outcome measurement must include not only objective measures of centimeters, grams, and number of repetitions, but also patients’ direct responses on questionnaires and rating scales focusing on “patient-reported outcomes.” Sometimes denigrated as subjective, in contrast to traditionally-preferred objective measures, questionnaires and rating scales have gained more universal popularity, credibility, and technical support since 2004, when the U.S. National Institutes of Health funded the creation of the Patient-Reported Outcomes Measurement Information Systems (PROMIS®). PROMIS supports development of patient-reported outcomes. It also provides a repository for standardized item banks and scales that can reliably collect patients’ responses in any of 70 domains of physical, mental, and social well-being.

By 2009, the U.S. Food and Drug Administration (FDA) was publicizing Guidance for Industry [159] documents and presentations of Clinical Trial Endpoints [156] supporting patient-reported outcomes as “intuitively desirable” and “a reasonable goal of therapy.” Although such outcome measures are not strictly required in FDA standards when reviewing new drugs or medical devices, these documents indicate an expectation: primary endpoints of phase 3 clinical trials should directly measure something that is important to the patient: survival, perceptible benefit, or decreased incidence of adverse events. With legislated elevation of importance and rigor of self-report measures, and impetus to consider the person’s perspective throughout episodes of care, clinicians and healthcare researchers have gradually included both quantitative and qualitative measures when assessing whether interventions effectively change health or function. In addition, clinicians infused by person-centered priorities make a conscious effort to fully engage patients in all aspects of their care, from diagnosis to rehabilitation and wellness, and in all of the ways by which patients can participate in clinical decisions.

4.1.3 Context for Choosing Measures

The current emphasis on person-centeredness when choosing measures in healthcare has not evolved in isolation, but within a context that includes the clinicians and healthcare environment. Before focusing on the choice of appropriate measuring tools, the clinician and environmental contexts deserve attention.

Clinicians

Clinicians make diagnostic, treatment, and rehabilitation decisions about measurement based on their healthcare setting and what they choose to assess about the patient’s healthcare needs. If one clinician treats one patient one time, then asking the patient directly for their preferences may yield the person-centered information on which to negotiate a medical decision. If the clinician sees the patient a second time, changes in the wording or circumstances of the question may result in a different response unrelated to any effects of the intervention. Keeping the words and circumstances (e.g., lighting, distractions, time since symptom-reducing medication) as similar as possible--standardizing the questions--helps focus attention on the actual treatment effects. Perhaps the clinician draws the question from the scientific literature to support evidence-based practice; in that case, changing the wording of the question may also alter the response expected. If the clinician hopes to leverage the experience with this one patient to improve practice with other patients, the clinician must standardize the question and question-delivery to minimize the effects of differences in the measure and maximize the meaningfulness of the response. Unfortunately, most clinicians do not have updated expertise in measurement along with content expertise in their specialty areas of practice. In addition, most practicing clinicians lack the time to evaluate the hundreds of new measurement tools published yearly. Thus, they may revert to common assumptions about measurement that can limit the selection of instruments and interpretation of data and thus restrict person-centered care. Five of these common clinical assumptions are described below.

Clinical Assumption 1: Standardized Measures Can Be Home-Grown and Altered

In a 2009 study, only 218 of 456 physical therapists claimed they used standardized outcome measures; and the second most frequently employed “standardized” measures were identified as intra-facility “home-grown” measures [84]. Home-grown measures may facilitate communication within a clinical department that all agrees on the definitions and uses of that measure, but do not translate easily to people outside of that group. Further, psychometric testing for reliability of scoring and validity of interpretation and meaning frequently gets overlooked in home-grown measures. Any alterations made to standardized measures require that the clinician document the departure from the standardized measure to allow comparison across persons, testing occasions, and settings. For alterations to become standardized as in the ALSFRS-R [32] or Modified Fatigue Impact Scale [52], they, too, must undergo psychometric testing [99, 118] to ensure sound measurement.

Clinical Assumption 2. Items on a Measure Are All Equal

Clinicians familiar with quantitative measures such as meters, kilograms, or heart rate may be tempted to treat numbers obtained from qualitative measures the same way: interpret ordinal scales as continuous measures, sum the items, and report mean and standard deviation as if on an interval scale. However, when taking a test of knowledge in a particular subject, most students would agree that some test items are harder than others. Likewise, some items on an attitude questionnaire are more difficult to endorse than others [166]. Counting the total number of items correct or endorsed does not reveal whether the easiest or hardest items are contained in that count. Examining a person’s knowledge or attitude more granularly requires that measurers account for the differences among items in a measurement tool.

Clinical Assumption 3. Latent Variables Are Directly Observable

People may assume that a score on an IQ test equals intelligence, but the responses only document how well the person reads, interprets, and marks an answer on the particular test. Constructs such as intelligence, attitude, pain, mood, and perception of one’s own quality of life are not directly observable; they are latent constructs [7]. The full construct remains unknowable from the observer’s standpoint; careful definition and calibration of representational measures are required to record status or effects of the construct in the person observed. In addition, the circumstances and context of the measurement dictate which aspects of each latent construct get assessed or described. Some of the differences in context include the patient’s diagnosis or condition, generic measures versus those specific to specialty clinical settings (i.e., oncology versus orthopedics), and whether the measurer is a primary care provider or a specialist. Constructs such as fatigue, mood, and pain can change by the minute, hour, or day, so they require repeated measures or assessment of the impact of these person-centered latent constructs. The integrity of measuring latent variables depends on continued patient engagement; if patients do not engage in revealing this aspect of themselves, responses will be missing or inaccurate in representing the construct of interest to the observer.

Clinical Assumption 4. A Change in Score Means the Intervention Was Effective

A change in score may merely reflect variability in the measure’s ability to record the construct of interest, or variability in the person’s performance due to factors such as time of day, mood, and concentration [62]. Measures such as effect size, minimal detectable change (MDC), minimal clinically important difference (MCID) [164, 167], and patient acceptable symptom state (PASS; defined as the symptom score beyond which patients consider themselves to be well) [47] can help determine if a particular change in score means that the intervention was effective and important in changing a person’s life [19].

Clinical Assumption 5. Competing Measures Are Interchangeable

Just as a change in wording may influence the respondent’s interpretation of the meaning of a question, a different measure may reveal a different aspect of the construct to be measured [128]. For clinicians who recognize differences in measures, lack of comparability can stymie incorporation of new evidence into their practice or documentation of changes a particular person experiences. Choosing which measure to use requires examination of the available evidence supporting the measures and matching measure characteristics to the appropriate characteristics of the person to be measured [35]. Equating [108] and meta-analysis [80] procedures can help interpret results across multiple measures and healthcare systems while noting the differences among them.

Clinician and clinical researcher knowledge levels vary regarding measurement assumptions and their resources differ for reviewing current literature. Even updated measurers likely compromise terminology and processes when collaborating with other stakeholders who have less-informed assumptions. In addition, measurers may choose legacy assessment tools based on their prior “state of the art” status or past endorsement; literature and practice in many areas have not yet caught up to person-centered best practices. The examples in the current chapter utilize a range of assessment tools and methodologies; caveats are provided when limitations in procedures limit the interpretation.

Environment in Which Measurement Choices Are Made

Whatever their measurement capabilities, all clinicians and clinical researchers function within a healthcare environment that can both promote and hinder best practices in person-centered measurement. Each measurer makes choices that fit their own environment and clinical setting.

One pervasive environmental factor stems from the need to apportion limited resources: prioritization of healthcare delivery can result in health inequities. The ideal of health equity seems unimaginable when pitted against the constraints of power, greed, and competing self-interests. Governmental leaders and people with wealth or the “right” mix of racial/ethnic/religious/national/gender characteristics receive healthcare that is not available to the rest of the world. Current tools and healthcare systems were created for the benefit of people with employment, health insurance, and social advantages, thus continuing systemic racism and injustices, and propagating inequities [106, 161]. Too many existing outcome measures were developed by isolated groups with a university education and extramural funding and tested on convenience samples rather than samples representing the applicable population [43]. Well-meaning measurers may be blind to a tool’s built-in cultural biases and the differences in perception that individual respondents bring to the assessment. The result is that many person-centered outcome measures may not apply to a diverse set of people or may mislead measurers in their interpretation of data [161].

For example, physical function items in questionnaires tend to assume that individuals have similar sets of daily tasks performed in similar environmental conditions. On the SF-36 [112, 163], a well-studied and common measure of health-related quality of life, two questions ask respondents whether their health limits them in walking several blocks or lifting/carrying groceries. Blocks can differ in size between urban and rural areas; the terrain can differ from clear, level sidewalks to trash-filled, pot-holed pavement; groceries may consist of a drink and chips for one or a week’s supply for a family of 5. Thus, the same response to these SF-36 questions could mean widely different functional abilities. Statistical techniques such as examination of differential item functioning (DIF) can determine if groups of people, otherwise at the same level of attitude on a questionnaire, respond differently to specific items [16], but personalized differences remain unaccounted for. Cognitive debriefing after a person completes the measure might help clinicians understand patients’ mindset and interpret the responses to inform clinical decision-making. Although interviewing a respondent after the questionnaire may personalize the responses, the interview also removes the advantages of brevity, consistency, and comparability that a standardized measure provides.

Clinician and environmental contexts influence measurement choices in multiple ways; conflicting expectations can add to the stress of clinical practice. The clinician may want to improve practice, but policy, access, and resource limitations may constrain measurement choices. For example, larger health organizations may have resources that enable access to references and electronic record keeping, thus lowering data collection and analysis hurdles compared to an organization with a manual entry for paper files. On the other hand, larger organizations also tend to have layers of procedural complexities in a workflow that restrict measurement choices. Further tensions arise when hospital beds or appointment slots are limited: how will measurers assess who will benefit most from care or how much care each person receives? Such dilemmas abound in clinical practice.

Questions regarding the value of uniformity and individuality also add to tension: utilizing a common terminology in measurement tools may force uniformity on persons of widely different experiences. Likewise, tools with global comparability can promote consistency across healthcare, but may also restrict individualized medicine. On the other hand, while treatment of one patient as an n-of-1 study may be very individualized, the need for economies of scale and training of new clinicians requires that pattern recognition and uniformity of measurement also develop. Each measurer must navigate such competing priorities when choosing the right measurement tools.

Healthcare administrators also face dilemmas regularly, specifically those who work in population health or quality teams alongside clinical champions, domain experts in patient-reported outcome measures (PROMs), and the information technology (IT) support team. Their priority may be to stretch the budget to pay for healthcare while showing payers that patients reliably get what they need. Such administrators may benefit from the governance of PROMs in organizations to collect PROMs more uniformly and use them to individualize care (e.g., https://epros.becertain.org/governance/guidelines/governance-structures). Other team members also must confront the challenge of minimizing resource expenditure while optimizing clinical benefit and sometimes profitability. Not all of them consider person-centered care above the financial rewards of a successful business model.

The next sections describe some of the more common dilemmas or competing priorities measurers encounter and some solutions that might help in decision-making. Examples provide common situations and evidence of clinically meaningful measurement that can improve communication and consistency in person-centered healthcare management. The ideal solutions typically balance competing priorities in a “both/and” approach.

4.2 Competing Priorities: Personalization Versus Standardization

This section discusses personalization versus standardization. Personalization tends to support person-centered care; standardization tends to support uniformity. Each has merit, but they can conflict when measurers evaluate measurement methods. This section first relates these concepts to equality and equity, then describes theoretical frameworks that can maximize the benefits of standardized measurement for equitable person-centered care.

4.2.1 Equality and Equity

Personalized medicine tailors healthcare based on the individual’s pathology, genes, preferences, comorbidities, social determinants of health, and numerous other characteristics known to affect responses. Standardization of healthcare ideally means that everyone has equal access to quality care, using the same measures and interventions. However, the literature on health disparities distinguishes between equality and equity (Fig. 4.1). Equality results from all persons with the same condition receiving the same (standardized) procedures and treatment: the same surgery or the same number of rehabilitative visits. Equity results from all persons with the same condition receiving what they need (personalized) to achieve the same results: the ability to perform the same functions. Equality tends to focus on the delivery or supply of healthcare; equity tends to focus on what the person can do once receiving healthcare. This distinction has profound implications on the measures to choose for personalized and standardized care.

Fig. 4.1

Equality and Equity. The left image shows equal supports for all three people, but the function is not the same. The right image shows different supports that allow all three people to watch the game. Equity might also be systemically incorporated by making the fence transparent at all levels. (Image: [2])

For example, with standardization, one unit of surgery for replacing a hip equals any other unit of the same surgery; and one physical therapy visit to improve function post-stroke is equal to any other unit of the same type of visit. This concept underlies the philosophy of healthcare reimbursement known as “fee for service,” in which clinicians and institutions request payment for the number of units of different procedures delivered. Frequently identified as work relative value units (wRVUs), institutions base these on the provider’s work, the institution’s expenses, and the cost of malpractice insurance premiums associated with the healthcare procedures delivered. Institutions may also weight wRVUs by the severity level of the patient’s condition and use them to record a provider’s productivity.

However, while counted as the same, each wRVU disguises differences in the characteristics of all individuals involved, both providers and patients, who affect the experience and effectiveness of each unit for persons receiving care. Comparison of units (or standardized payment for units) across settings and regions will always have limitations to some degree because of differences in provider expertise and patient responsiveness.

In contrast, with personalization, differences among clinicians and patients can be addressed, with measures recording the individual patient’s condition, context, and risk factors that may route the patient to clinicians with appropriate specialist expertise. Comparison of care across such personalized episodes may not reveal equality because different individuals receive different numbers of visits or procedures. However, measures of overall outcomes may reveal equity if all individuals attain similar levels of functional ability. This concept underlies the reimbursement philosophy of “value-based care,” in which clinicians and institutions request payment based on persons with specific conditions achieving particular outcomes.

Note that identifying the effects of a condition for a person or population requires measurement; further, comparing overall outcomes across groups requires measurement that is standardized. Thus, equitable healthcare does not preclude standardization but mandates that person-centered outcomes form a major part of the constructs measured. Personalization then relies on grouping individuals into meaningful categories of those needing less and more attention or additional units of standardized procedures. The grouping depends on appropriate assessment of individual physical and psychological characteristics that affect outcomes from specific health conditions. The meaningfulness of the groupings determines the effectiveness of the additional procedures in contributing to equitable outcomes.

Clinical prediction rules (CPRs) can help categorize persons based on physical or psychological characteristics associated with better outcomes from prescribed intervention [111]. One example of a clinical prediction rule utilizes the 9-item questionnaire called Subgroups for Targeted Treatment (STarT) Back Screening Tool (https://www.physio-pedia.com/STarT_Back_Screening_Tool), in which patients in rehabilitation for low back pain answer Agree or Disagree that back pain has bothered them or limited certain activities, or they have worrying thoughts about it. The STarT Back Tool categorizes respondents into those with low, medium, or high risk of developing persistent, disabling low back pain. Without this tool, wide variations in patient experiences and rehabilitation have challenged resource utilization and resulted in disparate outcomes.

A meta-analysis examined studies comparing risk-stratified care based on the STarT Back tool versus standard (non-stratified) care [168]. In risk-stratified care, all patients received support and enablement of self-management. Persons classified as low risk received one treatment session; persons classified as medium risk received individualized physiotherapy with a focus on functional improvements and reduced disability; persons classified as high risk received the same plus additional training. The additional training addressed psychosocial barriers to recovery with a focus on cognitive, emotional, and behavioral responses to pain and dysfunction. In 2788 persons across 4 separate studies, applying intervention based on the STarT Back categorization compared to standard care resulted in significant and clinically important reductions in pain-related disability for each group of stratified patients; comparison of patients receiving risk-stratified care versus standard care revealed a significant although small reduction in pain-related disability plus a cost saving of more than £34 per patient.

The cost savings may reflect avoidance of unnecessary care in persons with a low risk of developing persistent, disabling low back pain [168]. Despite cost savings and positive outcomes from using the STarT Back tool, implementation of this standardized way to categorize patients with low back pain has encountered barriers [115]. In a process evaluation across 33 clinics, the rate of utilization of risk stratification was only 37.8% (range: 14.7–64.7%); author-identified barriers included staff members’ knowledge and beliefs, patients’ needs, technology issues, lack of clinician engagement, and lack of time [115]. Thus, standardization of a tested improvement to personalize care has had only partial success: strategies for implementing this and other improvements require additional work before being definitively recognized as improving equitable clinical practice.

Grouping patients for more equitable and person-centered healthcare must include differences in persons’ previous experiences and social risk factors. Standardized measures might collect personalized information to reveal these inequities. For example, a meta-analysis reveals that homeless adults can experience accelerated aging, showing an incidence of activity limitations and fall rates usually associated with adults 4–20 years older [155]. Standardized measures across the primary studies include self-report data regarding basic (bathing, dressing, toileting, transferring, eating) and instrumental activities of daily living (taking transportation, managing medications, managing money, applying for benefits). In these studies, data were analyzed based on the number of people indicating difficulty performing at least one of the tasks listed; differences in difficulty among tasks (items) were not addressed. Fall rates were similarly compared based on self-report of fall frequency in the last year.

Frailty did not reveal significant differences in incidence between homeless and housed adults, likely because of heterogeneity in measurement choices [155]. Two studies used a 5-question survey and respondents were considered frail if they affirmed at least 3 of the 5 questions. Another study used a 42-item index of symptoms, signs, and disease classification, with a stated cut-off value that indicated frailty. The difference in granularity of the two frailty tools potentially contributed to the 4 times larger rate-ratio of frailty in homeless versus housed adults when using the 42-item index compared to the 5-question survey [155].

While additional work on measuring frailty in this population is needed, this research line explores a potentially meaningful categorization of people based on their past or present housing stability. Suppose clinicians standardize assessment by age-group, only targeting patients over 60 for routinely checking activities of daily living and fall risk. If so, they may miss physical function losses in a 45-year-old homeless individual with a premature geriatric syndrome. Similarly, for clinics to target the person-centered healthcare needs of people experiencing homelessness, resources must address physical function losses as much as psychosocial disorders. Improving clinical practice would require that past or present housing stability be considered as a risk factor for decreased functional abilities and increased falls.

Additional personalized grouping variables may either increase or decrease equitable healthcare. Population-level measures have documented differences in health and experience of pathology across height, weight, age, race, ethnic group, sexual identity and orientation, education, and insurance status. For example, in the U. S., Black patients have a higher incidence of hypertension, infant mortality, and lower life expectancy and were more likely to succumb in the first year of the COVID-19 pandemic than White patients [106]. Erroneous interpretation of these differences in health as genetic differences among races have led to appalling discrimination; inclusion of race as a factor in medical decision-making has typically resulted in under-treatment and inferior healthcare for Black patients [161]. Corrected interpretation considers such differences in health as the effects of racism, frequently resulting in crowded and unsafe living conditions, employment in essential positions that cannot be performed remotely, and delays in healthcare.

Ensuring equitable healthcare does not mean that measurers should not collect data on race, ethnicity, and socio-economic status. Examining the research on population-level differences in treatment across groups can guide improvements in clinical practice. For example, Black and Latinx patients, post-stroke, have experienced significantly greater morbidity and disability compared to White patients [54]. Armed with this data point, clinicians and patients can institute mitigating protocols and confirm their effectiveness when a repeated population measure improves. As another example, working-aged people on Medicaid (public insurance) have 14–24% lower odds of being discharged after a stroke to an inpatient rehabilitation facility and are twice as likely to get discharged to a skilled nursing facility compared to those on private health insurance plans [114]. Policymakers and administrators can use such inequities as a baseline from which to mark improvement after changes are made.

Both standardization and personalization can contribute to or detract from equitable healthcare. The choice of constructs to measure and their interpretation makes the difference. Each person collaborating in their own person-centered healthcare wants to know: “What is the effect of this proposed treatment on people like me?” Standardized measures can reveal typical effects of an intervention, but if the measures are not created or interpreted in a way that accounts for personal differences, the literature will underestimate intervention effects on one group and overestimate effects for other groups. Thus, person-centered care requires that researchers and clinicians keep equity and diversity in mind as they design and review studies with patients. Further, clinicians must discuss clinical decisions with every patient, ideally based on documented response differences in people similar to them compared to randomized controls [43].

4.2.2 Frameworks for Assessing Effects of Health Conditions

Theoretical frameworks can guide measurers in determining when to consider personalized variables and standardized measures in healthcare. One of the most globally recognized frameworks was drafted in 2001 by the World Health Organization (WHO): the International Classification of Functioning, Disability, and Health [79]. The ICF framework shifts the focus of medical attention from disability [121, 160] to health, thereby philosophically considering individuals and wellness as relevant constructs when mapping the course of diseases. The ICF framework allows for both standardization and personalization in healthcare across conditions, settings, and professional disciplines.

In this framework (see Fig. 4.2 showing the ICF framework for Case A), health conditions--whether wellness or pathology-- affect body structures and body functions along with the activities and participation required to engage in the multitude of roles an individual plays in life (e.g., work, leisure, self-care, relationships). Activities such as walking may become limited when function diminishes in body structures such as lower extremity muscles; participation in work or leisure activities may become restricted when walking deteriorates. The effects of health conditions are personalized by environmental and personal factors that influence how the individual responds to the impairments of body structure and function within the activity limitations and participatory restrictions that individuals encounter. Environmental factors refer to the surroundings in which activities or participation occur, including effects of physical and social conditions, and attitudes of people around whom the person functions. Personal factors refer to sex, age, coping styles, prior history with disease or disability, social background, education, and overall behavior patterns that can serve as facilitators or barriers to function. As clinical management of pathology proceeds, person-centered care requires that assessment occurs at each point of the framework to facilitate collaborative decision-making.

Fig. 4.2

Framework for relating the effects of ALS in Case A to the person’s function based on the International Classification of Functioning, Disability, and Health. ALS amyotrophic lateral sclerosis. (Adapted from: International classification of functioning, disability and health: ICF. Geneva, Switzerland: World Health Organization [79])

Empirical data has confirmed the construct validity of the ICF framework; the framework components are distinct, although related [49]. Cross-sectional data were collected from 89 rehabilitation centers in 32 countries. Over 3000 persons diagnosed with one of five health conditions completed the SF-36 to enable comparison among the different conditions: low back pain, rheumatoid arthritis, osteoarthritis, obesity, or stroke. Their healthcare professionals graded the functioning of each person 0 (no problem) to 4 (complete problem) along with core sets of ICF categories (items) that had been developed previously for each individual diagnosis and across diagnoses. Multidimensional techniques were used to assess whether Rasch family models fit better with 1, 2, 3, or 4 dimensions associated with the components of body structure, body function, activities, and participation. Correlations among the dimensions ranged from 0.36 to 0.93, confirming that these components are related, but the model fit was best with all four components as distinct dimensions [49]. The findings confirmed that the ICF framework not only has construct validity but also is interpretable by healthcare professionals across different disciplines and countries.

The ICF framework facilitates clinical consideration of the interlinkages among components and the influence of contextual factors such as environmental conditions and personal factors on a person’s function. These linkages can help direct a clinician in assessment and intervention planning. For example, a patient with a degenerative neurological disorder may note difficulty going to a friend’s house to play bridge (restricting participation in a leisure role). The clinician can then focus assessment on the activity limitations, body structure/function impairments and personal and environmental factors that contribute to this restriction. Suppose the limiting factor is environmental, such as steps to the friend’s front door or low height of the toilet seats. In such cases, interventions will involve different assessments and decision-making than if the limiting factors are inability to manipulate the cards or fatigue during an afternoon’s bridge tournament. Once the direction of investigation has focused on the person’s specific limitations in valued activities, standardized measures may be employed to document the person’s current status and progress made with intervention.

Note that the task the person wishes to perform interacts with the environmental and personal contexts in a systems model of functional activity or movement [146]; tasks may be easier or harder based on facilitators or barriers in the environment or in personal factors. The clinician must ensure that the person is tested or asked about tasks that they want to do, at the intensity and with the stamina desired, within the sport or home/community environment where they must complete that task when participating in their preferred life roles. Personalized performance measures may either use or simulate the conditions and environmental conditions for the tasks of interest; self-report measures can ask patients what impact their health condition has on performance of valued tasks that are either pre-specified or patient determined.

The ICF framework and the systems model underscore potential differences in effects that a similar health condition might have on various people. Two people with the same amount of lower extremity weakness may continue working for different lengths of time, depending on the job requirements (sedentary or active), employer’s support, aptitude and liking for the work, and the financial incentive for maintaining it. As another example, two people may report knee pain from identical pathology, but their pain experience does not simply indicate the threat or real-time occurrence of tissue damage [66]. Instead, the pain experience reflects each person’s assessment of how dangerous, intense, and frequent the pain is, and the effect of pain on their activities and participation. Successful person-centered healthcare relies on consideration of the person’s tasks, the environment in which they perform those tasks, and the personal factors of anxiety, depression, fear, belief system, and prior experience that affect how they view any activity that induces pain [66].

Thus, the ICF framework and systems model together provide a theoretical foundation for personalizing healthcare while utilizing standardized measures to assess targeted components. The standardized measures can contribute needed gradations for prescribing treatment and comparing outcomes across time and persons.

4.2.3 Case A, as Informed by the ICF Framework (Fig. 4.2) and Systems Model

In the initial description of this case, the patient with ALS is meeting with the rehabilitation team, indicating the team’s interest in engaging the patient’s preferences along with clinician judgment in person-centered decision-making. The ICF health condition in this case is ALS. The pathology and identified activity limitations suggest that impairments of body structures and functions include neurological structures and function, along with progressive weakness in all extremities and some oro-facial muscles. The patient can no longer participate in his prior work roles and has become more dependent in the role of self-care; other participation roles and restrictions have not been specified. The ALS-FRS-R assesses the activity component of the ICF framework, indicating the severity of limitations in activities of daily living and some restrictions in his participation in self-care.

The literature shows that quality of life for people with ALS varies, depending more on perceptions of control and communication than actual severity of the disease [94, 142]. To improve person-centered care, the team will need a greater understanding of the patient’s additional participation roles and restrictions to know what tasks the patient values, and the contextual factors related to person and environment. The choice of measures should focus on the quality-of-life indicators of importance to the patient, standardized to grade goal attainment or further restriction, personalized to apply within the patient’s context. Critical decisions regarding use of a feeding tube or ventilatory assistance must be addressed at some point; the team and patient will need to consider the implications for extension of life versus quality of that life. Such decisions must take into account the person’s attitude and circumstances as well as progression of the disease.

4.3 Competing Priorities: Satisfaction Versus Effectiveness

This section contrasts satisfaction and effectiveness. Satisfaction means contentment with the process or a particular status. Effectiveness means the person made progress toward some goal. Person-centeredness does not in itself dictate sole assessment of either construct. However, measures of satisfaction do not substitute for measures of effectiveness when preferred outcomes include more than satisfaction with the experience. This section first differentiates between these constructs, then examines several measures related to person-centered effectiveness: biomarkers and wearables, disability, discomfort, and personal factors that include perceived quality of life.

4.3.1 Distinguishing Satisfaction from Effectiveness

Gauging satisfaction and person-centered effectiveness both likely require some version of self-report questionnaires. The distinction comes in the questions asked and the comparisons made. Instruments assessing satisfaction typically focus more on a customer/business relationship than the effects of the intervention and are completed at the end of the individual’s experience. Questions about available parking, helpfulness of staff making the appointment, and cleanliness of the restrooms hopefully lead healthcare businesses to improve the next customers’ experience with regard to infrastructure, personnel, or maintenance. Healthcare businesses, like other businesses, may want high net promoter scores, meaning that patients indicate more top ratings than low ratings on questions about likeliness that the person will recommend this business to others [3].

On the other hand, unlike retail or food service businesses, healthcare businesses may be financially penalized for repeat visits, particularly if the return (e.g., to the emergency room or re-hospitalization within 30 days) indicates previous ineffectiveness of care or premature discharge for the same health condition. Thus, in addition to interest in patient satisfaction with the interaction, healthcare businesses also have an interest in the effectiveness of interventions. Questions about patient-perceived effectiveness might relate to curing or minimizing an adverse condition, slowing degeneration, stabilizing or improving activity, and increasing participation in valued life roles. Such questions presume that a condition or activity level has changed; thus, the measurement must occur at least twice, for a baseline and post-intervention data point. Differences in responses across repeated measures could lead clinicians to improve clinical practice: identifying gaps in practice, implementing best practices, and engaging patients in shared decision-making. Measurers interested in effectiveness must differentiate what construct they value and wish to improve, then measure in a way that can show status and change in that construct. Most satisfaction measures include very few items related to effectiveness and their single time point does not address change. Measuring effectiveness may require separate tools that specifically address the constructs that the clinician and patient hope will change with intervention.

Cott et al. [38] attempted to merge the constructs of satisfaction and person-centered effectiveness by developing a client-centered measure of perception of care received for clients who had undergone inpatient rehabilitation services. The authors defined their concepts and the component domains through literature review, focus groups with clients, and review by content experts. The seven resultant domains included: participation in decision-making and goal setting, education, evaluation of client-centered outcomes, family involvement, emotional support, coordination/continuity, and physical comfort. The authors drafted five-six questionnaire items for each domain and tested item statements for clarity and relevance through cognitive interviews with patients; all items had response choices on a 5-point Likert scale with 5 being strongly disagree and 1 being strongly agree. The 33-item Client-Centred Rehabilitation Questionnaire (CCRQ) was then tested for internal consistency, test-retest reliability, and discriminative construct validity in a mailed survey to patients discharged from two rehabilitation facilities. Cronbach’s alpha and test-retest reliability proved acceptable based on the authors’ stated standards, and the subscales differentiated as expected among patients with different primary diagnoses [38].

In evaluating the resultant CCRQ, the measure’s success is mixed with regard to capturing satisfaction and effectiveness. Only one of the 7 CCRQ domains asks about outcomes, with 4 statements for respondents to endorse [38]: “I was kept well-informed about my progress in areas that were important to me”; “I accomplished what I expected in my rehabilitation program”; “The program staff and I discussed my progress together and made changes as necessary”; and “I learned what I needed to know in order to manage my condition at home.” Other categories of questions focused more on whether respondents felt they were treated well, not especially if they were treated effectively: shared clinical decision making, inclusion of family, and whether valued symptoms and conditions were addressed in a personalized manner. Although a literature review examining the CCRQ confirmed the 7 domains as important in person-centered rehabilitation literature [131], a German study conducting an exploratory and confirmatory factor analysis could not confirm the original 7 dimensions of the CCRQ [95]. Körner et al. [95] proposed a 3-factor structure across 20 items: decision-making/communication; self-management/empowerment; and psychosocial well-being. Their proposed 20 and 15-item versions no longer include the item “I accomplished what I expected in my rehabilitation program,” the one item that comes closest to answering whether the respondent got better with treatment. Another study of a slightly modified CCRQ [51] concluded that person-centered goal setting (determining which outcomes mattered) varied based on numerous factors: respondents agreed (90%) that they were encouraged to participate in goal setting but agreed less strongly (73%) that decisions about what would help them were made in collaboration with program staff [31]. In general, the literature advocates use of the CCRQ for assessing whether a rehabilitation program is person-centered, and what aspects of the program might require improvement to increase person-centeredness, but the CCRQ is not specifically an outcome measure of intervention effectiveness.

A person’s satisfaction with care could conceivably influence the effectiveness of that care. Studies have shown relationships between satisfaction and the likelihood of adherence to treatment [77]. In other words, people who express satisfaction with care may also adhere better to their clinicians’ recommendations. Further, if barriers in access to the facility or the person’s communication with program staff prevent a person from receiving or following through with recommendations, then reduced effectiveness may follow. Satisfaction may also relate to health-related quality of life [77], a common construct typically associated with the ICF’s participation component and measured to assess restrictions that might ensue from the particular health conditions. However, characteristics that influence a person’s positive or negative outlook on life can also influence responses to questionnaires about satisfaction and quality of life. The positive or negative bias on these measures further restricts their usefulness in conveying information about the effectiveness of healthcare. In a systematic review of studies examining patient satisfaction with musculoskeletal physical therapy, a meta-analysis across 7 of the 15 studies revealed that patients generally marked “satisfied” to “very satisfied” with care, with clinician’s attributes as the most consistent determinant of patient satisfaction [77]. Only 3 of the 15 studies identified treatment outcome or symptom improvement as important to satisfaction. The authors of the review concluded that positive response bias may have affected outcomes, and overall satisfaction was more related to interactions with the therapist and the process of care than the outcome of the treatment. The implication is that any measure of effectiveness likely requires multifactorial assessment to distinguish between symptom improvement and satisfaction with overall care [77].

Clinicians may hesitate to put effectiveness to the test: measuring satisfaction seems less risky because patients might be satisfied with their experience even if the healthcare has not been effective. In addition, clinicians may fear that patients know what they want but not what they need; “allowing” patients their say requires trust that they will receive, apply, and appreciate recommendations even when not exactly the solutions that patients thought they wanted. Researchers advocate transparency to build greater bilateral trust: clinicians must elicit relevant information from and listen to patients, share information about the health condition, and collect relevant data regarding the patient’s status. Clinicians must then ensure that patients see the data indicating their own changes with intervention. With guidance to interpret visualized data, patients can determine for themselves when their outcomes have improved. When clinicians and patients both agree that person-centered outcomes show improvement, bilateral trust may ensue.

4.3.2 Person-Centered But Not Person-Reported: Health and Disease Biomarkers

Health and disease biomarkers can help measure healthcare effectiveness. They are person-centered when they measure a construct important to the patient and patients can see the data themselves. Usually these instruments record fluctuations in body structure or function while a person engages in meaningful activity; as such, biomarkers can function as a type of biofeedback. Biomarkers such as heart rate, blood pressure, and body weight can help assess health, wellness, and disease progression. Heart rate and blood pressure can be ascertained through biometric equipment that ranges from Intensive Care Unit monitoring to smart-watches. An overall increase in blood pressure monitors can mean that the person is working hard or straining homeostatic systems; a decrease might mean that the body is more relaxed, has greater vascular efficiency (as in trained athletes), or is about to pass out from decreased oxygen to the brain (as in orthostatic hypotension). Thus, biomarkers must be monitored and interpreted before using them for clinical decision-making; patient education regarding the fluctuations in these measures can help individuals use the data more effectively in controlling their own activities.

Walking speed has emerged as an effective biomarker for multiple conditions; some researchers call it a functional “sixth vital sign” [55]. Gait velocity, or distance covered in designated time, is measured across various distances and with equipment as simple as a stopwatch or complex as instrumented gait mats or computerized motion analysis systems. The starting instructions may specify either self-selected gait speed or “as fast as possible while staying safe.” Slower gait velocities have been associated with fall risk, functional decline, hospitalization, and discharge destination in several populations. Gait velocity is a person-centered measure in the sense that slow velocities can limit activities such as walking across a busy street before the light changes. Cut-off values have been proposed to indicate that a person walking at velocities below 0.4 m/s likely has ambulatory ability restricted to household use. A community ambulator typically requires a gait velocity of 0.8 m/s or greater [55]. Interventions that can progress gait speed from one category of function to the next make an important difference in a person’s life.

Physical activity or step count as recorded by wearable equipment continue to evolve as potential biomarkers of disease progression or intervention effectiveness [22]. Research grade and commercially available technology typically employ accelerometers with carefully calibrated algorithms to count steps or movements, usually summed and averaged by the day. The advantage to wearable equipment is that clinicians and patients do not have to rely on one-time snapshots of walking behavior in the clinical environment to make decisions; they can monitor and assess behavior changes during everyday participation in life roles [23]. For everyday use to become feasible, however, devices must be inexpensive enough for each individual to have one, and person-friendly enough for individuals to wear with any clothing options, and remember to resume wearing after recharging the batteries. Person-centeredness argues for commercially-available technology worn continuously long-term rather than a returnable device worn for a few days or a week [23]. Persons in many diagnostic groups have used step count monitors for healthcare and personal data collection [22]. In multiple sclerosis (MS), continuous step count monitoring adds value to clinical assessment because the wide range of step count levels within each level of disability on the clinician-reported Expanded Disability Status Scale (EDSS) [98] can potentially alert patients and clinicians to disease progression or goal attainment more quickly [24].

Nonlinear Variability

When measuring biomarkers continuously, it is tempting to characterize the results with an overall mean and standard deviation. However, recurring bodily functions such as heartbeat, postural sway, and stepping during gait occur dynamically, varying over time: the next heartbeat, postural sway direction and amplitude, or step length and width does not exactly match the previous one [74, 90]. Patterns of variation have importance and can be measured [150]. For example, postural sway forward and back and side to side can be recorded as the displacement path of the person’s center of pressure while standing quietly on a force plate [67, 75, 150). The pattern of repetition from one forward to backward sway cycle to the next is neither robotically repeating nor randomly varying, but normally depicts a certain amount of complexity necessary for “optimal movement variability.” This Goldilocks pattern ensures that the body is not too rigid to respond quickly to changes in muscle contraction or environmental alterations and not too variable to accomplish the task of staying upright. Nonlinear measures of structural variability over time include approximate entropy (a measure of unpredictability) [126, 127] and Lyapunov exponent (a measure of divergence) [73, 75, 149], among others. Optimal movement variability is person-centered in the sense that individuals function from a particular growing and aging body structure within a changing environment that presents various obstacles to movement. From this perspective, variability can present information about the state of the body and environment from which to make adjustments, not simply errors in movement program selection or execution. Each individual must then adapt each goal-oriented movement appropriately to accomplish the task as desired. Skill does not mean freedom from errors but greater variability so the individual can flexibly adapt.

Hunt et al. [75] demonstrated a critical implication that supports the examination of optimal movement variability in person-centered care. In a health condition such as MS, problems with balance and excessive postural sway can result in falls and fear of falling. Traditional reports of postural sway have used linear measures of center of pressure displacement such as range and root mean square. However, patterns of postural sway over time can present as either abnormally rigid (low approximate entropy or Lyapunov exponent) or abnormally random (high ApEn or LyE) in their variability compared to the movement patterns of healthy controls. People with abnormally rigid patterns might respond best to interventions that introduce variability: perturbation training or supported practice correcting induced errors in balance. People with abnormally random patterns might respond best to interventions that restrict variability: repeating the same task while constraining part of the movement or augmenting sensory feedback. Mixed groups of patients receiving the same intervention might wash out differences in both the linear and nonlinear measures of variability, resulting in the erroneous conclusion that the intervention had no effect [75]. Meaningful categorization of persons is necessary to reveal the actual effectiveness of some interventions, and then to apply the appropriate intervention that matches the needs of each person. Nonlinear measures of variability can distinguish important differences that may guide clinicians in determining when to repeat the same task to minimize error and when to provide variations similar to those the person encounters daily to practice adjusting responses.

4.3.3 Person-Centered Disability as a Measure of Effectiveness

Measuring disability depends on the definition of the term and purpose of the measurement [10]. The classic medical model of disability [78] defines it as the effect of trauma, congenital factors, or disease. In such cases, the purpose of measuring is to assess how different the person is from normal health or function. A disability may be defined as the inability to work because of a medical condition that is expected to last one year or more or result in death [147], and measures would determine who will receive disability insurance benefits. Measurement based on traditional medical or governmental definitions can serve business or societal purposes but typically lack appropriate information to guide person-centered clinical practice. For the latter purpose, “meaningful disability” may be defined as that which affects abilities most critical to a particular patient’s health-related quality of life [141]. Measures would then focus on patients’ preferred bodily functions, activities, and participation roles, as the ICF components that patients most want to change.

To define meaningful disability, Mitra proposed that “an individual is disabled if he or she cannot do or be the things he or she values doing or being” [119]. When aligning this statement with the ICF framework, “doing” implies bodily function applied toward the performance of an activity; “being” implies the roles the individual assumes for participation; and what the individual “values” is a personal factor that influences preferences and satisfaction when engaging in activities in the individual’s environment. From the individual’s perspective, disability is thus the gap between doing and being, between having and wanting with regard to the ability to do or be. Measurement of meaningful disability must include both the having and wanting components [10]. While assessment of current ability the individual has can be accomplished with performance-based or disease impact measures, assessment of preferred ability that the individual wants requires observation of behavior choices during daily living or self-report to determine what values the individual places on various functions.

Standardized functional assessment tools ask respondents to record their current ability across tasks that the creators or clinicians deem valuable in activities of daily living; the patient’s preferences typically do not get concurrently assessed. For example, most functional measures include walking items, but wheelchair athletes may prefer the speed and agility of their wheelchairs to slow, clumsy and non-functional walking with braces. Common measures of disability also focus on what people can do currently, using “normal” function as the standard by which to gauge the effects of the health condition on the individual [50, 63, 83, 98]. Measures that assess self-perceived disease impact come closer to addressing the doing-valuing gap Mitra identified [119]. For instance, an MS impact scale asks respondents how much their disease has affected their ability to do specific everyday tasks [70]. With this phrasing, respondents compare current abilities to their own remembered normal rather than to a hypothetical societal norm.

Another way to examine effectiveness related to disability is through quality-adjusted- or disability-adjusted life years (QALYs or DALYs, respectively). QALYs indicate the effectiveness of medical treatment regarding increases in the length and quality of life the patient experiences. The quality-of-life estimation has sometimes come from the clinician’s perspective, but association with patient-reported quality of life measures ensures the person-centeredness of the QALYs [134]. A systematic review of studies basing QALYs on patient-reported quality of life measures before and after treatment indicated that almost half of the 70 studies in their review had what the original authors deemed acceptable cost per QALY. The authors of the systematic review qualify their findings because of differences in QALYs when based on differences in the quality-of-life measures used, and the variability of costs per QALY in different regions [134]. In contrast, disability-adjusted life years or DALYs indicate the number of life years while living with the effect of a disease or disability [33, 57]; thus, instead of years and quality of life gained as recorded in QALYs, DALYs focus on the years of life lost and years spent in ill-health [14, 33]. People’s perceptions of the effects of disability differ among countries and populations. In one developing country, healthcare providers and mothers deem severe and profound disabilities as far worse than death; mothers who have a child with a disability that has required hospitalization deem even moderate disability as worse than death [148]. Factors influencing these differences include societal stigmas, environmental accommodations for particular disabilities, access to rehabilitation, and financial support. Measuring with QALYs or DALYs assumes that all persons value life years that are free of disability, but that people might tolerate a certain amount of disability if healthcare could extend the total years of life. However, attaining the individual person’s perspective requires individual-level measures that show what they value most in their lives.

The literature provides several options for aligning individual abilities with the person’s preferences. The Patient-Specific Functional Scale (PSFS) [154], Goal Attainment Scale [92], and others [101] utilize goal setting to indicate patient preferences [76]. The patient chooses the tasks toward which the patient works in rehabilitation in consultation with the clinician. Both measures show evidence of reliability and validity [20, 76, 88, 105, 151, 152]. The patient-generated tasks in both measures facilitate repeated measures in individual patients at the beginning and end of care but limits the usefulness of each measure for comparing across patients. Further, neither scale explicitly records gaps remaining at the end of an episode of care between outcomes achieved and the patient’s preferred abilities. A Movement Ability Measure was generated that addresses these deficits [6].

Movement Ability Measure

The Movement Ability Measure was developed based on measurement principles [165] to operationalize Mitra’s [119] latent construct of disability specifically for rehabilitation. The Movement Ability Measure assesses self-reported current and preferred movement capabilities [7] based on the Movement Continuum Theory (MCT) of physical therapy [37]; the interval logit (log of the odds) scale allows calculation of the resultant current-preferred gap. According to the MCT, the focus of therapy is to minimize the gap between preferred and current abilities within the person’s maximum capabilities [37]. To extend the MCT and facilitate testing of the theory, Allen [4] subdivided movement into six dimensions: flexibility, strength (force production), accuracy (including timing and direction), speed, adaptability (ability to make dynamic adjustments to movement based on sensory input), and endurance [4]. The six dimensions evolved from clinical practice, literature review, and the criteria needed for robot development; each dimension subsumes related terms to result in a set of dimensions that fit predetermined criteria: descriptive, efficient, distinct, measurable, and understandable. Once the dimensions were identified, structured interviews of prior patients informed the clinical hypotheses about behaviors and performers that align with less and more ability to move. The 24-item Movement Ability Measure was generated using six ordered statements regarding self-perceived current and preferred movement abilities for each item, with four items in each dimension. The six statements or response choices for each item target abilities across the range of the construct from less to more (Fig. 4.3).

Fig. 4.3

One item on the Movement Ability Measure associated with the movement dimension called flexibility. (Adapted from Allen et al. [12])

Testing reveals the strong link retained among the Movement Ability Measure, the six-dimensional movement ability construct, and the extended Movement Continuum Theory [7]. When testing “current ability” responses to the Movement Ability Measure, the one-parameter multidimensional model aligning items with their appropriate dimensions fit best, and 52% of the 318 community-dwelling respondents showed differences rather than a uniform average across dimensions [4].

The Movement Ability Measure also shows evidence of reliability, content and construct validity [6], and responsiveness of the current ability responses to physical therapy intervention [5]. Examination of the current-preferred gaps indicated that respondents discriminated between current and preferred abilities, and most indicated a preferred ability other than the top statement for each item [9]. Patients starting physical therapy had a larger gap than healthy controls, and their perceived gaps significantly narrowed after 2 weeks in physical therapy [9].

Testing also indicates that this person-centered outcome instrument can improve clinical practice. A computer-adaptive test version of the Movement Ability Measure (MAM-CAT) was developed and used for a two-part pragmatic study [144]. The first part of the study assessed differences in patient priorities as revealed by the movement dimensions showing the largest gaps on the MAM-CAT (Fig. 4.4) and clinician emphases as revealed by the health records of assessments and interventions during the episode of physical therapy [12].

Fig. 4.4

Movement Ability Plot (MAP) from the self-report MAM-CAT depicting the gaps between the movement ability the person in Case A likely has now and the ability the person would like to have. Dimensions with the largest gap (endurance at −2.6, with strength and flexibility closely following) reflect the person’s priorities for care. The original logit scale has been transformed to a scale in which 0 means cannot do even with assistance, and 6 means moves competitively. (Figure adapted from Allen [8])

Although both the MAM-CAT and physical therapy notes indicated that patients progressed, comparison showed poor or slight agreement on the movement dimensions to prioritize (Fig. 4.5). The second part of the study compared the first results with patient outcomes when MAM-CAT responses were reviewed and discussed between therapist and patient at the beginning of the episode of physical therapy [13]. The average decrease in total gap size at the end of care was significantly greater when therapists and patients had the opportunity to view the MAM-CAT responses together.

Fig. 4.5

One patient with a diagnosis of low back pain in which the emphases from the physical therapy notes do not match the movement dimensions with the largest gaps according to MAM-CAT results. Flexibility and strength improve by discharge, and the current-preferred gaps narrow, but the largest gaps remain in endurance and speed at discharge. (Adapted from Allen et al. [12])

4.3.4 Person-Reported Effectiveness: Minimizing Adverse Conditions

Satisfaction and effectiveness may come closest to merging when the health condition targeted by healthcare is self-perceived pain, fatigue, dizziness, other discomfort, or fear. From the patient’s perspective, the adverse condition is the reason they are seeking healthcare; effectiveness in managing the condition will likely weigh heavily in their satisfaction. These conditions, however, can change rapidly: an overall improvement with intervention may be masked by a current spike in discomfort. Thus, a person-centered assessment must consider the multiple facets of the targeted condition so that both the immediate condition and the impact of the condition on everyday life can be revealed.

Numeric or visual analogue scales have been used to assess the level of a person’s current pain, fatigue, exertion, or dizziness. Numeric scales typically use the form: “on a scale of 0 to 10, with zero meaning no pain and 10 meaning the worst possible pain, what number would you say your pain is right now?” Some numeric scales have qualifying phrases attached to several of the numbers rather than just at the ends of the scale; Borg’s Rating of Perceived Exertion has respondents put a number on how hard they think they were working during the target activity. On a 6–20 scale, 7 is very, very light, 13 is somewhat hard, and 19 is very, very hard. The scale has been shown to correlate with exercise heart rates, with 15 on the scale approximately equal to 150 beats per minute [26]. A category ratio scale is also used, with numbers 0–10 indicating increasing exertion. Visual analogue scales generally have patients mark the distance on a line or a more severely wincing face that represents their discomfort. Studies have shown that a VAS measure of the impact of fatigue on daily life has moderate reliability and can rapidly screen individuals for severe fatigue impact on their life [96].

The impact of pain on function has been assessed in well-documented measures such as the Roland-Morris Disability Questionnaire [139] and the Oswestry Low Back Pain Disability Questionnaire [50] for patients with low back pain. Dizziness has been assessed more generally in the Dizziness Handicap Inventory [81], which can be used across diagnostic groups.

Fatigue measures have been devised in general or for specific use in a particular diagnostic group. For example, the 9-item Fatigue Severity Scale (FSS) [97, 103, 117], has been used across multiple populations with neurologic disorders, including MS, stroke, and ALS [59]. Unfortunately, researchers disagree on the scale’s usefulness; in a Rasch analysis of the FSS in people post stroke, 2 of the 9 items did not fit their model and a 7-item FSS showed better psychometric properties [103]. In contrast, a Rasch analysis of the FSS in people with MS showed that 4 of the 9 items did not fit a unidimensional construct of the social impact of disease; the authors recommend using a 5-item version [117]. Some measures subdivide fatigue into cognitive and physical dimensions [118, 157] and others report that factor analyses do not support such a distinction on the MFIS [132]. Hudgens et al. [72] developed the Fatigue Symptoms and Impacts Questionnaire—Relapsing Remitting Multiple Sclerosis using the best available guidance regarding instrument creation and testing with Rasch and exploratory factor analyses. The resulting instrument has multiple types of items to address both current severity and impact. The authors report good reliability and validity that align with the patient-derived constructs from which the measure was generated [72].

4.3.5 Person-Reported Mediators and Measures of Effectiveness

Personal contextual factors such as the person’s self-efficacy or confidence, and their coping capacity, mood, happiness, or quality of life may reveal mediating variables that affect the person’s responses to treatment [137], or may be targeted as outcomes themselves.

For example, perceived self-efficacy as recorded in scales such as the 14-item Self Efficacy Scale for exercise/physical activity provides respondents with various adverse conditions that might conceivably hinder them from partaking in the activity [56]. Examples of three items include “I could exercise … when tired; when my schedule is hectic; when I haven’t reached my exercise goals.” The level of a person’s self-efficacy may mediate their ability to change behaviors such as smoking, over-eating, or a sedentary lifestyle [122].

As another example, stronger coping capacity as documented by the 13-item Sense of Coherence scale has been linked with greater health-related quality of life in healthy populations and people with diseases such as ALS [142]. An example of a mediating variable that may also become an outcome in itself is confidence in one’s ability to perform various tasks without falling, e.g., the Activities-Specific Balance Confidence scale [130] which is both related to fall risk and an indicator of reduced fear of falling. Likewise, health-related quality of life may mediate the impact of the person’s health condition on their body structure, function, and activities, or may become the outcome that healthcare specifically targets.

Health-related quality of life measures may apply specifically to those with a particular diagnosis or across diagnoses, as the commonly cited Sickness Impact Profile (SIP) [21] and SF-36 do. The SIP is a 136-item questionnaire covering 12 categories that patients might perceive to have been impacted by their adverse health condition: (a) sleep and rest; (b) emotional behavior; (c) body care and movement; (d) home management; (e) mobility; (f) social interaction; (g) ambulation; (h) alertness behavior; (i) communication; (j) work; (k) recreation and pastimes; and (l) eating. SIP scores are represented as percentages of disease effect.

In contrast, the SF-36 incorporates the concept of wellness along with disease impact into the range of quality of life. The SF-36 is a 36-item short form of a much longer health-related questionnaire in the Medical Outcomes Study; the eight sub-scales include “eight of the most frequently represented health concepts” [163, p. 906]. The sub-scales consist of general health, physical functioning, role physical, role emotional, social functioning, bodily pain, vitality, and mental health. In addition, one question asks about the person’s health transition, comparing their health to one year ago. Multidimensional analysis of the functioning of the SF-36 across diagnostic groups in the U.S. [112] confirms the 8-dimensional construct and hypothesized order [133] of the items and response levels. Treatments that change physical morbidity have the most effect on sub-scales relating most to the physical summary scale, while treatments that target mental health mostly affect sub-scales correlated more highly with the mental summary scale [163].

Comparison of the SF-36 across populations and with scales specific to particular diagnoses can inform researchers of characteristics of the scales. For example, SF-36 responses in people with ALS are generally lower than in age-matched controls across all dimensions except for bodily pain which is nearly the same in the two groups [94]. However, functional deficits as recorded in the ALS-FRS-R are significantly associated with deficits in the physical functioning and vitality subscales of the SF-36, not the other aspects of health-related quality of life. In contrast, increasing depression as recorded in the Beck Depression Inventory is associated with all SF-36 subscales except for physical functioning. Although not an inevitable consequence of incurable disease, moderate to severe levels of depression have been identified in 30% of people with ALS [94].

4.3.6 Case A, as Informed by Satisfaction and Effectiveness Measures

In the initial description of this case, the patient with ALS is meeting with the rehabilitation team. The team prioritizes the person-centeredness of care and has each patient complete the Client-Centred Rehabilitation Questionnaire (CCRQ) at the end of an inpatient stay to gauge the team’s success in that realm. Based on the underlying constructs of person-centeredness, the team emphasizes seven domains: patient participation in decision-making and goal setting, education, evaluation of person-centered outcomes, family involvement, emotional support, coordination/continuity, and physical comfort. In the team meeting with the patient, they discuss the person’s social and home environment, lifestyle, and priorities. They determine together which family members and healthcare specialists will participate in the next team meeting, the information the patient and family still need before making upcoming decisions, and the outcomes and mediating variables most important to assess.

Recommended mediating and effectiveness measures include continuation of the ALS-FRS-R, and inclusion of the SF-36, MAM-CAT (see Fig. 4.4), Sense of Coherence (SOC) scale, and Beck Depression Inventory. Assessment of current-preferred gaps in movement ability with the MAM-CAT will help determine the patient’s current priorities for addressing body function impairments. Assessment of coping capacity (with the SOC) and depression will help determine the usefulness of psychotherapy as part of the multidisciplinary emotional support. The most relevant adverse condition currently impacting the patient is fatigue; the Fatigue Severity Scale will be added to the outcome assessments to document changes with instruction on movement efficiency or use of energy-saving devices. Education regarding the use of a feeding tube or non-invasive versus tracheostomy mechanical ventilation at home will incorporate assessment of QALYs to assist the patient and family in analyzing the pros and cons of different decisions [138]. In addition to the measures recommended here, each discipline will use their own specific instruments. The team will need to consider the patient-burden of all these measures going forward.

4.4 Competing Priorities: Scientific Rigor Versus Practical Convenience

This section contrasts scientific rigor and practical convenience. In the ideal world, stakeholders in healthcare all have scientifically rigorous and technically sound measures at their fingertips to meet all their measurement needs. However, as much as clinicians and researchers desire scientific rigor, constraints in time, personal knowledge, availability, and clinical usability remain determinants when choosing instruments. Practical convenience, therefore, must be considered so that the average clinician treating both the everyday and unusual patient has access to brilliant tools to promote brilliant results. This section briefly reviews types of criteria associated with rigor in measurement, then examines ways that measurers have attempted to make measurement more convenient, accessible, and usable.

4.4.1 Ideal Measurement

Ideally, developing scientifically rigorous tools involves collaboration between those who see the need for a construct to be measured and those who create and assess the instrument. Best practice for generating new person-centered measures requires collaboration among patients and clinicians to define the construct of interest and base the questions or tasks on theory, so the items represent a scale of the construct as intended (for examples, see the CCRQ, MAM-CAT, and the Hudgens fatigue measure already described in this chapter). Metrologists and psychometricians can ensure that the resulting items fit the underlying construct and function as designed. Complex constructs may require complex instruments with multiple dimensions or question types to better align with the underlying theory [6, 72]; measurement experts can both develop and test such complexity using factor analysis and multidimensional procedures. Building instruments based on item response and Rasch measurement principles ensures that differences in an item or task difficulty factor into the assessment of people’s abilities or attitudes, and each individual respondent gets located on a unidimensional or multidimensional construct with an individualized standard error of measurement. Psychometricians can ensure that resultant measures reduce the uncertainty around the individual’s measurement.

Sometimes multiple published measures already assess some portion of the construct of interest, such as functional abilities while in an acute hospital or in outpatient clinics but not across settings [64]. Measuring a construct using a single metric across settings has person-centered advantages because patients (and their clinicians, if the settings share data) who have seen their data in one setting can track their progress as they move to the next setting in an episode of care. Instead of developing a measure from scratch, measurers might pool relevant items from multiple measures, hypothesize the order of the items on the construct, and test all of the items in a sample representing the population to be tested. Measurers can thereby create an item pool. For example, the PROMIS database banks items across many different constructs. Using Rasch measurement techniques, measurers can determine the usefulness and difficulties of individual items and response sets and refine the pool.

Once an item pool has been generated, the next step likely involves creating subsets or groups of items because fewer items take less time and decrease patient burden. However, indiscriminate cutting of items in a pool or on an established test can diminish reliability and sensitivity [11]. One technically complex solution uses computer adaptive tests, in which the computer program adapts the test by choosing which items get presented to the respondent from the item bank [144]. The programmed algorithm selects the next item based on responses to prior items; the objective is to fine-tune the location of the respondent on the latent construct.

Another way to expedite test-taking is to create fixed sets of items as short forms, using items that span the range of interest on the construct so that floor and ceiling effects are minimized in the population for whom the measure is created. For example, the Activity Measure for Post-Acute Care (AM-PAC), based on the activity component of the ICF framework, originated by gathering a pool of 222 items across 8 original sources/instruments. The creators wanted clinician- or patient-reported short forms for each of three domains: physical/movement, applied cognition, and daily living [64]. They tested items relevant to each domain with a Rasch one-parameter model, and whittled down to 10 items in each domain, with overlapping sets for inpatient and community (outpatient) use, all on the same metric. The originators simultaneously created a CAT version [65]; the CAT version has the advantage that estimates for individual scores are more precise than when using the fixed short forms. Subsequently, a version for acute care, the AM-PAC 6 Clicks mobility measure [85], has been widely used to alert providers to functional needs and predict discharge disposition [124].

When clinicians do not have access to item pools for the constructs they want to measure, they may want to understand how one measure compares with other measures, particularly measures that address the same or similar construct. Equating methodologies can inform stakeholders of overlap and differences among various measures, using common items to help calibrate items that are distinct on multiple measures to ascertain what areas of the latent construct are well-covered [108].

When equating methodologies have not yet compared measures of interest, measurers may look to the experts to ascertain the best measures to use in research or practice. Systematic reviews and meta-analyses gather related published evidence to compare the usefulness of one intervention or measure over another in a particular population. The advantage is that meta-analyses can gather data across measures of a similar construct; when combining controlled trials of interventions, meta-analyzers typically use Cohen’s d, which expresses the difference in means in standard deviation units [80]. The drawback is that many of the measures in these meta-analyses are reported in and analyzed as continuous scales despite their ordinal scale origins. Readers of these studies should interpret findings with caution.

Published clinical practice guidelines [29, 120], consensus standards, or recommendations [89, 110, 123, 129] can direct readers to the measures that have the most literature support to date. Their advantage is that the review typically comes from a panel of content experts, covers much more literature than the individual practitioner has available, and summarizes designated aspects of the various measures. Their drawback is that the recommendations do not commonly guide clinicians in choosing the right measure for a particular patient [35]. Also, panels typically choose the aspects of reliability and validity they will review and prioritize measures with the most literature, so newer, less-utilized measures may receive lower recommendations despite the use of more rigorous development criteria and better alignment with the latent construct.

While guidelines and recommendations generally specify the need for continual updates, resources may not keep up with the rapidly changing literature. Additional resources may come from task forces that review sets of measures and attempt to standardize the best set for research [44] or practice (e.g., a consensus statement on core outcome measures for persons with Covid-19 released by the American Physical Therapy Association, June 2020) in particular populations.

While the judgment of experts matters, the clinician still must make an individual choice for each patient. Assessing effectiveness requires that the selected outcome measures be responsive to the changes expected with the proposed intervention. If the clinician expects the intervention to improve or delay deterioration in function, then the function must be measured using tools that are sensitive to detect changes from the intervention. Further, if the patient wants to get better at function, then the measure should assess functional ability on the most interesting tasks to the patient. While many different statistics have been used to assess instrument responsiveness [5], those that align change with patient-assessed clinical importance or quality of life likely have a better person-centered focus [140]. The types of statistics can include minimal important difference, minimal clinically important difference, or PASS.

4.4.2 Accessibility and Convenience

With a scientifically rigorous and responsive measure or recommendations in hand, the measurer then requires access to the instrument and ease of use for each patient. Access to the instrument may include fees paid to use a copyrighted instrument along with convenient methods for patients or subjects to answer the questions or perform the tasks, clinicians or researchers to collect and enter data, processors to analyze data and interpret results, and clinicians to share visualized, interpretable data with the patient. Convenience is critical because different patients will need different sets of instruments, so measurers cannot make measurement choices just once, but newly for every diagnostic group and patient.

Large data sets can assist in making decisions even at the level of the individual patient because they show associations and results that can be expected with various treatments. The next two subsections discuss data repositories and electronic health records, and the promises, challenges, and future possibilities in relation to person-centered outcome measurement.

Data Repositories

The National Library of Medicine defines a data repository as “a place that holds data, makes data available to use, and organizes data in a logical manner” [42]. Just as person-centered care must attend to a patient’s multifaceted characteristics and context, a system that holds person-centered data must provide for multifaceted data types collected from various sources. Unlike historical use of a data repository as a siloed, single storage unit limited in data sharing, the concept of “Information Commons” [158] fosters collection from multiple data sources such as electronic health records (EHR) and biomedical research data to collaborate and accelerate research, advance clinical care, and improve person-centered outcomes [15, 25, 162]. For example, Donado et al. developed a pediatric pain data repository integrated with EHR data to collect longitudinal patient-reported outcome measures [45]. Using the 6 core areas recommended by the International Associations for the Study of Pain, the authors implemented validated questionnaires from EHR to capture patients’ quality of life (e.g., physical function, emotional functioning, sleep, mood) and assess their progress and response to treatments [45].

The rehabilitation domain has traditionally been slow to adopt data repositories despite the potential benefits when addressing shared challenges in rehabilitation research. Challenges include rarity of some health conditions which limits sample sizes for determining evidence of efficacy and effectiveness [30], and the “black-box” phenomenon meaning that variability in symptom presentation hinders characterization of the mechanism of effectiveness of interventions or treatments used across research studies [68, 82]. Other challenges include inconsistent use of standardized outcome measures [84], and insufficient collection of pertinent patient attributes (e.g. comorbidities, gender, age, geocodes, etc.) that mediate their response to treatment.

However, early stages of the development of data repositories in some populations denote notable progress in meeting these challenges. A Multiple Sclerosis Rehabilitation Repository (MSRehabRep) can potentially store, retrieve, and share rehabilitation information in MS to support clinicians and researchers [27]. Focus groups from various stakeholders prioritized key features to “envision” retrospective and prospective data use in the MS population [27]. Desired characteristics for the repository included security and accessibility for data use and data sharing, and most essential, high-quality data standards [27]. Further, addressing the ongoing challenges of establishing common data elements and standardized outcome measures across diverse MS populations was an important step identified in data repository development [27].

The American Physical Therapy Association (APTA) has created the Physical Therapy Outcomes Registry [125] to collect clinical data from therapists on a registry platform, a type of data repository. A registry is defined as “an organized system that uses observational study methods to collect uniform data (clinical and other) to evaluate specified outcomes for a population defined by a particular disease, condition, or exposure, and that serves one or more predetermined scientific, clinical, or policy purposes” [60]. Registry data has several key advantages for researchers and clinicians. Along with providing opportunities for comparative effectiveness research [153], the combined use of clinical EHR data and other uniform data of interest (i.e. patient outcomes data relevant to a specific population) can establish benchmarking outcomes data and fulfill regulatory quality reporting requirements [125]. Desired characteristics overlap between data repositories and the PTOR features: enabling commonly used patient-reported outcome measures, establishing a core set of de-identified patient data, and collecting longitudinal data from an episode of care in physical therapy [125, 153]. While enrolling in PTOR is optional for users, organizations that currently participate in quality payment programs (e.g., Merit-based Incentive Payment System or MIPS) or other quality initiatives will likely benefit from using registries like PTOR because the Centers for Medicare & Medicaid Services have designated it as one of the Qualified Clinical Data Registries to collect and report on behalf of therapists [125]. This designation makes the PTOR particularly valuable if an organization does not have Office of National Coordinator (ONC)-certified EHR systems nor sufficient resources to collect, manage, and report outcomes data from their native EHR systems.

Ongoing hurdles hinder the utilization of data repositories to accurately reflect person-centered care. First, patients and their caregivers should know that having a data repository means that their data could appear online, even if de-identified, which may not preclude deduction of personal data if few people have the target condition. In addition, a clinical repository of data will have less ability to help direct person-centered care if it does not clearly identify the patient’s priorities and values. Thus, patient engagement is needed throughout the development cycle of a data repository as well as future data use. Second, strong data governance is needed to address privacy and data sharing policies and procedures. Even with de-identified patient data, HIPAA privacy rules and data sharing agreements need to comply with federal and local organizational guidelines. Third, development and maintenance costs could be a challenge for smaller health systems; they may need to consider outsourcing solutions to third party entities or clinical registries. The outsourcing itself can present challenges when attempting to select from several options without deep knowledge of the criteria to optimize. Fourth, curating different data sources to ensure that a repository remains reusable and standardized while continually updating with new clinical data types could be very resource-intensive [162]. Fifth, despite current cloud computing power and its robust nature to handle big data, barriers to accessibility and data use hinder administrators, researchers, and clinicians, let alone patients, from taking full advantage of refined and standardized data to inform clinical decisions.

As healthcare reimbursement practices pivot from fee-for-service to value-based care, harnessing person-centered data to reduce variations in practice, along with any unnecessary treatment costs, becomes non-negotiable. Data repositories in rehabilitation can provide an entry point to critically reevaluate person-centered rehabilitation, potentially uncovering ways to improve clinical practice at an aggregate level. However, to discover and deliver value-based care for our patients, healthcare needs a robust system that facilitates the collection of the right patient-reported outcomes for the right types of patients so that providers and patients have the ability to access the relevant data to make shared decisions. The fundamental groundwork of standardizing core data elements and outcome measures must continue if data repositories will succeed in promoting the use of person-centered outcomes to improve clinical practice.

Electronic Health Record Systems

The Federally funded Health Information Technology for Economic and Clinical Health (HITECH) Act incentivized adoption of electronic health record (EHR) systems and the Affordable Care Act (ACA) promoted their interoperability so that patients might “carry” their health records from one provider to another. These changes represented an evolution of person-centered data collection and processing methods across US healthcare.

Even so, the integration of patient-reported outcome measures (PROMs) with EHR systems has been moderately slow, partly due to EHR vendors’ priorities to meet the early stages of Meaningful Use (MU) regulatory requirements. Specifically, under Stage 1 and 2 of MU requirements, the emphasis was primarily on adopting EHR systems, capturing and advancing process measures, and promoting interoperability, not explicitly prioritizing PROMs.

However, from stage 3 of the MU requirements to the Quality Payment Programs in Merit-based incentive payment system (MIPS), the focus shifted to evaluate “person and caregiver-centered experience and outcomes” [69]. This shift mandated the collection of PROMs to achieve MIPS quality measure requirements. Concurrently, different specialty practices were shifting toward implementing PROMs in real-time clinical care, including oncology, orthopedics, and pediatrics as well as primary care practices [17, 18, 45, 145]. Increased usage of PROMs put additional pressure on vendors to include common PROMs in their EHR systems. Currently, a growing number of health organizations have implemented two of the largest EHR vendors in the market, Epic Systems and Cerner Corporation, and benefited from its certified EHR systems integrated with a suite of PROMs such as the PROMIS measures.

Several essential areas continue to evolve with the integration of PROMs in EHR technology: accessibility for rapid point-of-care data collection and visualization as well as remote administration of PROMs, and data integration into routine clinical management. The objectives of data integration into routine management is for the clinician to be alerted to changes and patient priorities, and clinicians and patients to review person-centered data together so that the measures play a part in shared decision-making [12].

Timeliness and convenience of electronic patient-reported outcomes (ePRO), when present, facilitate data completeness and utilization for aligning practice with person-centered changes. For example, using a patient portal tethered to the EHR system allows clinicians or staff to assign (or automatically assign based on preset organizational rules such as diagnosis, visit types, language, etc.) and send ePRO questionnaires securely to patients. Patients then have the flexibility to complete their assigned questionnaires at their convenience through available communication devices such as a web-based portal, smartphone app, or kiosk.

Specifically, PROMIS CAT questionnaires can be made readily available and exchanged electronically through a patient web portal. As soon as the patient completes a questionnaire online, it gets processed in real-time and allows assessment of results to be routed to the care team or specific referring clinician. If further action is warranted due to severe symptoms, then automatic triggering of notifications to different care teams or clinicians allow them to make necessary referrals and medical decisions in a timely manner [28].

For instance, if a new patient in an outpatient physical therapy clinic hits a cutoff point for depression in the PHQ-9 (9- item Patient Health Questionnaire), then the electronic notification may help alert the clinician of the need for psychiatric consultation. However, since the questionnaire results are tied to the EHR, the clinician can first see the patient’s record of long-standing depression and active treatment by the patient’s psychiatrist or care team, then this information can be taken into account and collaboration may ensue. Rapid communication among EHR users facilitates improved data completeness and the ability to make timely clinical decisions while reducing administrative burdens and costs to healthcare organizations.

The integration of PROMs into EHR systems also means PROMs data may be harnessed quickly and meaningfully shared between patients and clinicians. For example, native integration of PROMs in the EHR, in contrast to use of PROMs located on third-party sites, minimizes disruption in clinical workflow. The all-in-one integrated platform collects data through the patient portal, tablets, kiosks, or direct entry into the EHR system during patient encounters, which facilitates the creation of reports on PROM data with less hindrance from complexities of third-party PROM tool integration. The capacity for on-demand and trending data visualization for potential sharing with patients has also improved due to native EHR integration and prioritization of ePROs for clinicians. For example, a visualization tool such as Epic Synopsis Report from Epic Systems (Fig. 4.6) depicts both up-to-date and trended PROMs data that reflects a patient’s progress and response to treatments.

Fig. 4.6

Trended patient’s outcome data in a Synopsis report from Epic Systems (© 2021 Epic Systems Corporation)

From a patient’s perspective, responding to questionnaires before visits means the PROM data can be summarized and viewed immediately prior to seeing providers at the time of appointment [53]. Actual usage of this protocol requires planned and ongoing training for administrators and clinicians to understand not only the interpretation of the scores, but also visualization from the PROMs data to discuss with patients. The data must be communicated in a way that will engage the patient’s interest and seem relevant to their context. Otherwise, completing the PROM will likely result in non-use by clinicians and poor rate of data completeness by patients when confronted with other questionnaires in the future.

Another key advancement in EHR systems is the technology that allows the integration of patient-generated health data (PGHD) from smartphones or wearable devices with PROMs. An industry-standard, interoperable SMART (Substitutable Medical Applications and Reusable Technologies) on FHIR (Fast Health Interoperability Resources) technology, can provide additional context to patient’s health status such as changes to vital signs or blood sugar level potentially impacting physical function captured in their own devices or apps.

Integration with the EHR allows for synchronized, timely remote monitoring of patient’s symptoms and their health status following treatment or, as in a rehabilitation setting, independent home exercise programs prescribed by physical therapists. Perhaps the biggest vantage point for integrating patient-generated data from other devices is that it forces the EHR to solve problems of interoperability, therefore, providing the ability to communicate and exchange pertinent patient data between different health organizations. Moreover, the rapid, scalable data collection also allows management of population health outcomes as data are collected, analyzed, and exchanged with different health systems or entities using the same EHR systems.

Additional benefits from using PROs that are integrated into EHR systems include: assisting patients in remembering their symptoms to report to providers based on questionnaires asked (Fig. 4.7) [143]; monitoring patient’s status and responses to interventions and treatments over time [145]; improving patient engagement and shared decision-making by allowing patients to actively control their health data [100]; increasing patients’ likelihood of compliance and self-management of their health status [61]; and providing a platform for patients to tell their side of the story using PROMs in clinical settings. The timeliness of data allows clinicians to make relevant decisions based on the latest information and likely to increase patient engagement and data completeness. Patients are more likely to complete the data and stay engaged when clinicians reference and share the outcome data [61, 100].

Fig. 4.7

Example of Patient-Reported Outcome Measures. (Taken from Sayeed et al. [143])

While the standardization of PROMs and their collection methods have traditionally been a challenge for multiple stakeholders in health systems or private clinics, EHR-related technology has taken a giant step forward in integrating PROs into EHR systems and interoperability. However, EHR systems are not without limitations. Selecting and administering the ‘right’ PROMs are more difficult for primary care physicians (PCPs) compared to specialists (or allied health professionals such as PTs) due to the wide range of medical conditions PCPs treat. Adding a measure to the EHR system requires multiple layers of bureaucracy and extensive time even after convincing the vendor to make the change. Choosing the most relevant measure among disparate PROMs is challenging and makes personalized care more difficult [71]. An individual’s primary conditions cannot always be assessed in a timely fashion or generalized for comparisons with further data analysis [71].

Currently, PROM incorporation has no cookie cutter template that allows a ‘one-size-fits-all’ approach to integrating PROs electronically in clinical settings. Their use depends on several factors such as patient context, chief complaints, and psychosocial factors. Perhaps a more concerning matter is the potential inequity of access to technology for certain racial and ethnic groups despite all the technological advancements for enhancing communication between patients and clinicians. Li et al reported that African American and Hispanic patients were least likely to have online patient portal accounts (50%, 57.3%, respectively, compared to 83.9% of white patients) among the rheumatology patient population in a large urban medical center [104]. And while there is no “best of breed” PROM technology due to the ongoing challenges of interoperability and standardization, advancement in the functionality of EHR integration with PROM, and its accessibility, reporting, and analytics of patient data, all have made progress toward person-centered care.

4.4.3 When Practical Convenience Means Telehealth

Person-centered care sometimes means that technical complexity and practical convenience must merge: the clinician must take scientifically rigorous measurement and intervention to the patient rather than having the patient come get treatment on the clinician’s sophisticated equipment. Electronic platforms for collecting PROMs have helped because the patient with internet access can take PROMs before a clinical visit. Home healthcare remains a viable option for clinicians to gather performance data in the patient’s everyday home environment. However, home health still requires a burden of transportation time and resources for the clinician. The growing use of technology to deliver services remotely can improve access and convenience for both the clinician and patient.

Telehealth technology is defined as the use of electronic communications to deliver healthcare services, support, and information remotely to improve patient care [136]. No longer is remote rehabilitation treatment a fringe idea mainly for patients in underserved regions or with limited access to high-quality care. For decades, patients in rural areas relied on medical services from “critical access” hospitals, which are eligible rural hospitals federally funded to improve population health [1]. Patients that required increasingly specialized services had greater difficulty getting it. Underserved populations are further disadvantaged because of hospital closures and decreasing retention of clinicians, including physical therapists.

Telehealth holds potential solutions to address health disparity among those needing specialized services such as neurological or orthopedic rehabilitation. Evidence has shown the efficacy of telehealth when providing physical therapy in different clinical conditions such as low back pain, total joint arthroplasty, and chronic conditions compared to physical therapy in other health settings (e.g., SNFs, home health, inpatient, outpatient) [34, 41, 87, 102]. The advantages for patients are numerous: receiving equitable care even when distance is a barrier for patients (i.e., in rural areas), reducing costs of travel and time, receiving education and training for multiple caregivers or family members, and checking in remotely to monitor critical health conditions.

Patient satisfaction from telehealth shows positive feedback regarding different aspects, including technical setup and overall user experience. For example, Miller et al. used a patient satisfaction survey that contained ten items across multiple domains (e.g., connectivity to telehealth, using telehealth, hearing, seeing, feeling safe, comfort, experience with a physical therapist, meeting expectations, and overall satisfaction). Patients rated their satisfaction for each item on a 5-point Likert scale [116]. For instance, to assess a level of effectiveness of telehealth with regard to the patient’s need, a question such as “how did the telehealth physical therapy session(s) meet your needs/expectations” was asked on a 5-point Likert scale (1: Not at all satisfied, 3: Satisfied, 5: Very satisfied) [116].

The authors reported that 94% of 307 patients were satisfied with their telehealth sessions during the early stages of the COVID-19 pandemic [116]. While the target population was primarily from an outpatient rehabilitation environment, specialty areas also included neurologic and pelvic health [116]. And though a vast majority of patients (92%) responded that they were willing to have additional telehealth sessions [116], additional supporting evidence reveals that synchronous or real-time telerehabilitation effectively improves patient outcomes in quality-of-life measures such as physical function and disability, and pain [39]. Further, a hybrid approach to telehealth, combining it with in-person care, is also perceived more favorably by patients than a completely remote episode of care [39].

One of the perceived challenges to rehabilitative assessment when using telehealth is the validity and reliability of remote versions of measures. A systematic review revealed that telerehabilitation assessments demonstrated good concurrent validity for a wide range of outcome measures used in physical therapy practice: pain, gait, balance (e.g., Tinetti test, Berg test), muscle strength, range of motion, functional assessment (e.g., Timed Up and Go), and pain [107]. However, Mani et al. also reported that other special orthopedic tests, neurodynamic tests (e.g., Straight Leg Raises), and postural assessment in the lumbar spine had low to moderate concurrent validity [107].

A potential loss of agreement between telehealth and in-person assessment with respect to clinical management may hinder decision-making in physical therapy. For example, after collecting objective measures and other clinical information about a patient, accurate screening and diagnosis must be reliable regardless what “mode” of treatment is being delivered. Supporting evidence shows a high level of clinical decision agreements between telehealth and in-person assessments among chronic neurological and orthopedic populations [40]. While a high level of agreement (e.g., 83%) in clinical screenings and diagnosis between telehealth and in-person appear favorable, the study had a relatively small sample. A more advanced level of clinical expertise by providers may also be an important factor to deliver consistently high-quality telehealth care.

The rapid adoption rate for telehealth as a response to the COVID-19 pandemic deserves attention. During the early pandemic period, local and state policies were enacted in conjunction with the Centers for Disease Control (CDC) recommendations for healthcare providers to offer a virtual platform including telehealth to slow the spread of the virus that might occur if patients came in for treatment [93]. Surveillance data showed a 154% increase in telehealth visits compared to the previous year during the last week of March 2020, which coincides with the early shelter-in-place orders [93]. A similar study was reported on the surge in telehealth utilization, specifically in an outpatient physical therapy setting at an urban medical center, accounting for 84% of all follow-up visits during early periods of the pandemic [116]. Such evidence of telehealth adoption may indicate that it is a safe and effective platform for patients, clinicians, and the public to minimize exposure to infection. Telehealth also supports the continuity of care from in-person to remote clinical management. For example, the high-risk populations with underlying medical conditions susceptible to infection may find it particularly valuable to continue to rely on rehabilitation and medical care facilitated and accessible by telehealth.

However, telehealth may also exacerbate potential health inequities among racial and ethnic minorities. For example, despite patient-centered, value-added benefits from telehealth, some patients living in lower socioeconomic areas may not own basic smart devices or phones, or have access to stable internet connections. Consequently, a loss of opportunity to receive quality healthcare puts those who may be most vulnerable at risk of further adverse impacts on their quality of life. Even the basic limitations of Internet access due to inadequate infrastructures in their neighborhoods will increase the difficulty of accessing telehealth sessions [113, 135].

One study showed patients living in lower socioeconomic status neighborhoods were significantly less likely to choose telehealth visits [135]. Further, patients with non-English preferences may not choose telehealth given the option of in-person office visit [135]. Those unfamiliar with the use of technology or unable to afford it due to financial constraints will carry the greatest burden of telehealth barriers [135]. In addition, higher acuity level for the person’s condition may require an in-person physical examination or diagnostic testing [93], especially when ongoing remote monitoring of critical conditions is time-sensitive.

While telehealth technology continues to be adopted worldwide, the innovation of wearable devices (WD) and phone apps for remote patient monitoring could complement holistic care for patients. Wearable devices can be worn, carried in a pocket, or attached to clothing, to monitor and track health information with minimal to no disruption to patients. The Fitbit is an example of a fitness and wellness promotion device worn as a wrist-watch. Different versions provide wearers with the possibility of setting their own goals for activity, and monitor their progress toward clinically established goals.

This valuable information can be communicated with clinicians or exchanged between health partners or different health organizations using EHR interoperability. The convenient nature of wearing a non-invasive device is that patients can be mobile in their natural environments, whether they are at home or work without significant barriers of physical restrictions using WD. A systematic review reported several clinical indications for WD use with telehealth in chronic populations such as cardiac disease, diabetes, chronic obstructive pulmonary disease [91]. For example, uses might include self-management support for patients, clinical decision support for clinicians, and integration with clinical information systems (i.e. part of EHR systems) in healthcare organizations and outpatient settings [91].

The advancement of digital modalities, wearable devices, and telehealth have facilitated innovative ways to treat different neuromusculoskeletal conditions in rehabilitation. Devices like smart devices, apps, mobile devices and the more affordable wearable devices have pushed the envelope from asynchronous remote monitoring of patients to real-time communication. However, clinicians and healthcare teams still need to consider important implications before prematurely advocating for technology. Person-centered care must always involve the patients early and throughout the process to ensure practical and equitable healthcare with or despite technology.

4.4.4 Case A, Navigating Scientific Rigor and Practical Convenience

The rehabilitation team opens the patient’s chart from the electronic health record together with the patient and the team reviews the blood oxygen levels, vital capacity, weight changes, and MAM-CAT scores indicating body structure and function effects of the ALS course in the last month. They detect the trend for the ALS-FRS-R and, based on the literature and information from an ALS data repository (https://nctu.partners.org/MNDS/mnds_als_data_repository), they can project a possible time course for the next few months. They include information from the additional coping and fatigue measures and discuss what the patient and his family deem most important both for the short-term and long-term.

Based on the information shared, the patient and family choose to wait until the next quarterly meeting before starting non-invasive respiratory assistive devices, initially just for better sleep at night. Some re-arrangement of living areas will ultimately become necessary for easier mobility because the patient and family want him to be home for as long as possible. They expect to take advantage of telerehabilitation to ease the burden of transport for some of the upcoming meetings. For the time being, however, they want to optimize mobility and participation in social activities. Thus, they endorse rehabilitation to maximize flexibility and efficiency of movement and recommend assistive devices to increase function with less energy expenditure. The patient and family express their appreciation for the team’s emphasis on meaningful outcomes and the patient’s preferences throughout the experience.

4.5 Summary and Recommendations

This section summarizes both the challenges and promises for improving clinical practice through person-centered outcome measurement. Although the dilemmas facing measurers present real challenges when choosing clinically relevant measures, the ideal of brilliant results for every person should motivate stakeholders to continue striving. Success will mean that consistently high quality and equitable healthcare tailored to each specific person’s unique situation becomes the norm.

However, no one person or stakeholder role can meet the challenges alone. Many individuals will need to participate. The first section of this chapter discussed the contributions of both the clinician and environment to measurement choices. The clinician and healthcare environment must both evolve. The clinician must move past outdated assumptions about measuring; the healthcare environment must effectively emphasize health equity with the diversity of measures and experiences required in person-centered practice. The healthcare environment shows some promise: it has progressed toward valuing patient-reported outcomes with policy support from the legislature and processes in place through PROMIS and PCORI. Patients now participate in planning and funding research focused on person-centered outcomes. The dissemination of this type of participation throughout healthcare may help bridge the gap between adopters and the rest of society.

The second section discussed personalization and standardization and their potential association with equality and equity. Patient groupings along meaningful categories can help direct care more appropriately for both efficiency and effectiveness. The section also presented theoretical constructs such as the ICF and systems frameworks that can help stakeholders keep standard measurement personalized by choosing measures according to the person and environmental context in which they live and work.

The third section discussed satisfaction and effectiveness. While assessing the person’s experience in healthcare through client-centered measures such as the CCRQ can inform practice, satisfaction with the experience does not substitute for measures of effectiveness. Multiple types of outcome measures can help determine if healthcare effectively meets the person’s needs, including health and disease biomarkers and wearables, attention to nonlinear variability within time-sequenced data, disability as the gap between having and valuing, the use of QALYs and DALYS, and assessing both current and preferred abilities. Effectiveness measures may also include assessing adverse conditions that an intervention may minimize, mediators of patient’s response to treatment, and health-related quality of life.

The fourth section discussed scientific rigor and practical convenience, both of which require extensive expertise and diligence to get technically sound measures practically available, so clinicians have all they need at their fingertips. This requires collaboration at multiple levels. Best practice in measurement development calls for patients and clinicians to work together to define constructs and the scales on which they will be measured. Psychometricians can ensure that resulting measures meet the standards for scientific rigor. Researchers must contribute comparative studies to support recommendations of measures for common constructs. Computer experts may program a computer-adaptive test or provide a platform for access to the measures of interest or a data repository. Healthcare administrators must navigate a budget for servers and electronic health record vendors. Vendors must attend to the needs of the clinicians who demand that their preferred measure gets supported while simultaneously optimizing profitability for shareholders. Policymakers must continue to strive to make systems interoperable so that patients can move through a course of care or between providers and carry their excellent healthcare along with them.

These discussions lead to four recommendations for improving practice.

• 1. First, collaborations among content specialists and psychometricians must support measure development and modification; they must ensure that rehabilitation teams have measures for both qualitative and quantitative data that have a theoretical and person-centered foundation.

• 2. Second, scientifically rigorous measures must also meet criteria for accessibility and convenience, which means that developers must collaborate with computer programmers and platform creators who then test ease of use by patients.

• 3. Third, data managers and vendors must work together to build and support systems that are both flexible and durable while also secure and easy for patients and clinicians to view.

• 4. Fourth, administrators will need both satisfaction and effectiveness data to demonstrate clinical practice quality and improvement over time.

The ideal team, then, would comprise the care team and patient, IT support, domain experts, psychometricians, and third-party vendors associated with electronic health records. The objective is to leverage outcome measures to optimize equitable and person-centered healthcare and improve clinical practice. Every link in the chain must be person-centered if brilliant processes can be used to achieve brilliant results when used by everyday people.

Are we there, yet? No. But the evidence shows areas where we have gotten closer. The next steps in the process involve collaborations at all levels. Only then will we improve clinical practice with brilliant person-centered outcome measurement.