Why Hold Several Rulers Against One Person — An Introduction to the ICF Framework

When you open a rehabilitation paper, you will find three or four outcome measures attached to a single patient. One test measures muscle strength, another looks at walking, and another is a questionnaire. This article answers why one is not enough. The short answer is that each measure assesses a different layer of a person.

The framework that divides those layers is the ICF. The ICF was already introduced in the primer article "Rehabilitation — Starting to Plan a Livable Home as the Fire Is Brought Under Control". It is a framework for describing a person's functioning through three layers—body functions and structures, activities, and participation—along with the surrounding environment and personal factors. This article deals with the next step. It explores how the ICF assigns values to each layer, and what is revealed when the measures we used to rely on are hung on this framework. The later pieces in this series will continue to use this language.

Why You Need to Know — When What You Can Do and What You Actually Do Diverge

We can ask a person two different things: whether they are able to do a task, and whether they are actually doing it. The two answers are frequently different.

The area where this gap widened the most was housework. Among housework tasks, 40.5% of people actually did less than their self-assessed ability. For learning and knowledge tasks it was 28%, and for mobility it was 17% (PMID 41127427). This was the result of evaluating both what people could do and what they actually did, side by side, in a randomly selected sample of 129 people out of 1,707 community residents aged 50 and older in Aragon, Spain.

The gap also ran in the opposite direction. In community life (13.3%) and interpersonal relationships (5.5%), what people actually did exceeded what they could do (PMID 41127427). The authors read performance falling short of ability as a trace of environmental barriers, especially barriers in the home and learning spaces. They saw performance exceeding ability in the social domain as a sign that conditions existed to help with those tasks (PMID 41127427).

Even within the same group, the direction of the gap split depending on the domain. Writing down function as a single number erases this picture.

Three Layers and Context — One Step Further Than the Primer

The ICF (International Classification of Functioning, Disability and Health) is a classification published by WHO in 2001, which is a framework that describes functioning and its limitations within a biopsychosocial model (PMID 42423744). The biopsychosocial model is a perspective that views health and functional problems as being forged together by the body, mind, and society-environment. The ICF captures health and disability in terms of body functions and structures, activities and participation, and environmental and personal factors (PMID 24850572).

Laying out the three layers again line by line, they are as follows. The names applied when a problem arises appear as they are in research papers, so the English terms are provided alongside them.

Surrounding these three are contextual factors. Environmental factors are things outside that person, such as housing, assistive devices, people nearby, and systems, while personal factors are the conditions of the person themselves, such as age, personality, and life history. Therefore, in this framework, disability does not reside solely inside a person.

The primer article showed these layers through an illustration comparing them to a house. What is needed for this article is the subsequent fact that the ICF is a 「classification」. This means that a person's functioning is split into a very large number of items, with numbers assigned to each one. A category is each of these numbered items. The letter in front of the number tells you the layer. b stands for body functions, s for body structures, d for activities and participation, and e for environmental factors. "Carrying objects", which will appear later, is also a category on the activity and participation side starting with d.

Activities and participation share a single list of categories. Even for the same walking, if you look at whether it is accomplished as a task, it is activity, and if you look at whether it is done in the places of life, it is participation. This is why "activity and participation" are counted as a single bundle in research papers.

How the ICF Assigns Values — A Single Scale, Plus Capacity and Performance

A qualifier is a scale attached to each category to record the degree of the problem. The ICF rates every category on the same 0–4 scale (0 = no problem, 4 = complete problem) (PMID 17225041). Whether it is walking or memory, it is recorded on the same scale.

Two such scales are attached to the activity and participation categories. One is capacity, and one is performance. Capacity is what a person can do, and performance is what a person actually does (PMID 42181832). The ICF records capacity as the level achievable in a standard environment with no help and no hindrance, such as a laboratory. Performance is the level actually done in the environment where the person lives. Therefore, the difference between the two values captures what the person's environment has done.

If performance is lower than capacity, it means something is blocking the way (PMID 41127427). If performance exceeds capacity, it means something is helping that person (PMID 41127427). In the earlier Aragon survey, housework was on the blocked side, and community life was on the helped side.

The method for moving already used indicators into this framework is also established. Linking rules are standard procedures that map already used indicators to ICF categories. For each question in an indicator, what is being asked—namely, the concept—is extracted and linked to a category. Even the perspective from which information was collected and how response choices are divided into categories are determined by rules (PMID 26984720). Because the rules are set, health information from different indicators can be compared side-by-side in the same language (PMID 26984720).

The last is the abbreviated version. A core set is an abbreviated version created by selecting only essential categories because the entire ICF is too large for clinical use. There are core sets created separately as a basis for recording the functioning of specific diseases (PMID 42423744). The stroke core set that will appear later is an example of this. There is also a minimum set used for rehabilitation regardless of the disease. It consists of 30 categories (9 body functions + 21 activities and participation) drawn from data of 9,863 people, along with 12 environmental factors used together (PMID 26827829). It is a proposal to report functioning using the same framework regardless of the disease type, timing, clinical setting, and country (PMID 26827829).

How It Actually Appears — Which Layer Is This Tool Measuring

When we hang familiar measures on the ICF, it reveals which layer those measures have been capturing all along.

The first example is arm loss, or upper limb amputation research. From 86 tools used in this field, 1,115 concepts were extracted and linked to categories using connection rules. The concepts were split into body functions 29.2%, activity and participation 57.7%, environmental factors 8.6%, body structures 1.7%, and personal factors 0.53% (PMID 39692753). The tools in this field widely measure the body and everyday tasks, but almost never measure personal factors, which capture who that person is.

The second example is the mRS. The mRS (Modified Rankin Scale) is a scale that grades how independently a person can live in daily life after a stroke, and it is the flagship measure in acute stroke trials. The meaningful concepts extracted from the mRS were 12, which covered 9% of the stroke core set. The Structured Interview version of the mRS (mRS-SI), which asks using the same scale in a prescribed order of questions, covered 32% with 40 concepts (PMID 26006081). The mRS score moved most strongly together with the number of activity and participation problems, among which self-care, mobility, and housework were the most prominent (PMID 26006081). The authors note that while the content of the mRS links to the ICF, it may not neatly fit into any single outcome of the terminology proposed by the WHO, and caution is needed when interpreting mRS grades through the ICF framework (PMID 26006081). Read in our terms, the mRS moves closest to the activity and participation side, but it is not a ruler that cleanly measures just one layer, and the breadth it captures changes depending on how the same scale is asked.

The third example is hemiplegia. Hemiplegia is a state where one side of the body's limbs loses strength or cannot move at all. When tools measuring hemiplegic arms and hands were divided into three branches and counted, there were 18 tools measuring capacity, 9 measuring perceived performance, and 3 measuring actual performance (PMID 22498041). Perceived performance asks how much the person feels they use it, while actual performance measures how much they actually use it in daily life. There are many rulers that measure what can be done, but rare are the rulers that measure whether it is actually used (PMID 22498041). "Able But Unused — An Introduction to Learned Non-Use and CIMT" deals with the problem of not using a paralyzed arm in daily life even though one is able to. This is why the rulers to measure that problem are so thin.

Common Misconceptions

There is a saying that "ICF is paperwork." At least Brazilian therapists welcomed the ICF. In a survey of 604 physical and occupational therapists, 98% said they intended to introduce it into clinical practice, and 93% answered that it was essential. The barrier cited most often was that it is difficult to actually apply to patient assessments (PMID 40575875). The authors noted that most respondents did not receive sufficient education in school and had low confidence in using the ICF in clinical practice (PMID 40575875).

There is also the idea that "disability lies within the person." In a literature review on geriatric rehabilitation, environmental factors contributed to rehabilitation outcomes (PMID 41563587). The authors of the aforementioned Aragon survey read areas where performance exceeded capacity as a signal that there are conditions assisting with that task (PMID 41127427).

There is also the idea that "participation is the same thing as quality of life." Participation records whether that person is engaged in the settings of life, while quality of life is how that person feels about their own life. It has been pointed out that the ICF does not capture how an individual perceives that experience (PMID 20441436). The authors who made this point suggested expanding the framework by putting individual perception inside personal factors (PMID 20441436).

Honest Unresolved Questions

• Patients and the team do not see the same picture. In a study tracking 118 stroke patients, at 1 month after discharge, patients rated their own function higher than the rehabilitation team did, especially in cognition. By 12 months, the ratings were reversed, with patients rating themselves lower than the team did (PMID 41649090). The authors interpreted the initial high self-ratings as resulting from a lack of awareness of their condition and the influence of emotions, and noted that a framework combining self-reports and team assessments is needed (PMID 41649090).
• The reliability of qualifiers is uneven. Reliability is the degree to which two people get the same score when evaluating the same patient. In a study where two observers scored d430 (carrying objects), kappa—the agreement excluding what would match by chance—was 0.36, which is a "fair" grade and thus on the lower side (PMID 17225041). On the other hand, in a Polish study where a multidisciplinary team determined scores by consensus, the inter-rater reliability (ICC) was very high (PMID 40764554). Who scores and through what procedure determines the results. However, the latter study calculated reliability after the team determined scores by consensus, so it is not a number that can be directly compared to the kappa of two independent observers.
• Methods that separately score capacity and performance are actually rarely used. In literature from 2001 to 2025, clinician-reported assessments (assessments scored by clinicians through observation rather than by patients themselves) that used the original definition numbered only 5 papers, and the categories were limited to walking and upper limb use (PMID 42181832). The authors noted that internationally agreed-upon criteria are needed for each category (PMID 42181832). A commentary at the 25-year mark also points out that a gap exists between pilot projects and routine use in research. The complexity and effort revealed by pilot projects make them difficult to use consistently in everyday practice (PMID 42423744).
• Personal factors are not yet in the official coding structure. There is no numbered category list for personal factors, unlike the b, s, d, and e seen earlier. That is why personal factors did not appear at all in a geriatric rehabilitation literature review (PMID 41563587).

To Go Deeper

When reading a paper, if you write down the tier next to each outcome measure, it becomes clear what that paper is saying has changed. Whether it is impairment, activity, or participation, and if it is activity, whether it is capacity or performance.

Discrepancies between measures are not signs of failure, but signs of different tiers. The point of this framework is that different tiers have different answers.

Outcome measures are covered next in "How Do We Measure Recovery — An Introduction to Rehabilitation Outcome Measures", the boundary between compensation and recovery in "How Deep Is the Recovery — Between Compensation and True Recovery", and the problem of being able to use something but not doing so in "Able But Unused — An Introduction to Learned Non-Use and CIMT". The answers to why the papers appearing in those pieces apply multiple rulers to a single person lie in this framework.

Source Papers

• Disparity between capacity and performance in the ICF: implications for functionality in older adults. PMID 41127427 (2025)
• Exploring clinician-reported assessments of capacity and performance qualifiers in the ICF. PMID 42181832 (2026)
• [Twenty-five years after publication of the "ICF"—what research and practice can learn from each other]. PMID 42423744 (2026)
• ICF as a framework for change: revolutionizing rehabilitation. PMID 24850572 (2014)
• Refinements of the ICF Linking Rules. PMID 26984720 (2019)
• Toward the ICF Rehabilitation Set: A Minimal Generic Set of Domains for Rehabilitation. PMID 26827829 (2016)
• Validation of the ICF Rehabilitation Set: a Polish clinical perspective. PMID 40764554 (2025)
• Comparing the contents of outcome measures in upper-limb amputation using the ICF. PMID 39692753 (2025)
• Analyzing the modified Rankin Scale using concepts of the ICF. PMID 26006081 (2016)
• Valid and reliable instruments for arm-hand assessment at ICF activity level in hemiplegia. PMID 22498041 (2012)
• Factors associated with the clinical use of the ICF by physical and occupational therapists. PMID 40575875 (2025)
• Personal perception and personal factors: incorporating health-related quality of life into the ICF. PMID 20441436 (2010)
• Significant differences in evaluation of ICF core sets between stroke rehabilitants and the rehabilitation team. PMID 41649090 (2026)
• Assessing observer agreement when classifying functioning with the ICF. PMID 17225041 (2007)
• Rehabilitation outcomes of older persons within the context of the ICF. PMID 41563587 (2026)