Invited Commentary on the ICF and Physical Therapist Practice

Escorpizo and colleagues¹ present a thoughtful argument for the well-defined utilization of the International Classification of Functioning, Disability and Health (ICF)² in physical therapist practice documentation. To accomplish this goal, they advocate for the systematic use of assessment tools based on the ICF, specifically, the ICF Core Sets.^2,3 They propose that an ICF Core Set, a short and manageable list of categories applicable to a health condition or event, can be used in clinical practice to describe and monitor the most salient aspects of the disability experience related to a patient's health condition or event. They advocate that each ICF Core Set, the product of a systematic development process that involved extensive input from experts, patients, and review of the literature, be used as the documentation template in physical therapy clinical encounters. In a companion article, Rauch et al⁴ illustrate how the ICF Core Sets might be used as a documentation template in a case report of a patient with spinal cord injury.

I strongly endorse the use of the ICF framework as a common language to facilitate communication among and within health care professions.^5,6 Furthermore, I applaud the authors' stated goal that the physical therapy profession adopt systematic and concrete assessments based on the ICF framework for use in clinical encounters as a means of advancing physical therapist practice.⁷ The ICF classification of function and disability information, in principle, offers several attractive advantages.^8–11 The ICF framework establishes a common language for describing health-related states and provides a systematic classification scheme for health information systems. The comprehensiveness of ICF's classification system could facilitate the widespread and systematic recording of functional status data in medical and rehabilitation patient records (at individual and institutional levels) and could influence universal design, public education, and legislation. Measures built from the ICF framework could permit comparison of health and health-related states across patients, studies, and countries, as well as across clinical services. Such measures could be used to compare the distribution and determinants of health-related states of different populations, predict health system usage and costs, and provide evidence for social policies and laws.

However, I am skeptical as to whether the ICF Core Sets, given their current level of development, are worthy of being adopted as a comprehensive classification system of functioning and disability in physical therapist practice. I believe that without further research that clearly demonstrates the psychometric adequacy of the ICF Core Sets for such an application, it is premature to advocate that the ICF Core Sets be integrated into physical therapist practice documentation. My reasons for this skepticism are as follows.

The ICF coding, which consists of a system for categorizing or classifying people's health, functioning, and disability states, meets the basic definition of measurement: the assignment of symbols to objects, individuals, or events according to systematic rules. Measurements have 4 major purposes: categorization (or classification), discrimination between groups, prediction, and evaluating change.¹² As measures, ICF Core Sets must meet basic psychometric standards, including reliability, validity, and responsiveness to clinically meaningful change.

The ICF's unit of classification is discrete categories within health and health-related states, defined as the situation of each person within an array of health and health-related domains. Using ICF coding, health-related states of an individual may be recorded by selecting the appropriate category code (a nominal measure) and then adding qualifiers (ordinal measures), which are numeric codes that specify the extent or the magnitude of the functioning or disability in that category. The qualifiers identify the presence and severity of a decrement in functioning for each ICF component (ie, body functions and structures, activities and participation, and environmental factors). Each ICF code denotes a component, domain, category, and at least one qualifier. The codes are alphanumeric, beginning with a letter and followed by up to 4 or 5 digits. Qualifiers are mostly coded as one or more numbers after a decimal point. For example, one could use the ICF coding approach to classify the walking ability of Mrs Jones, an 80-year-old woman with a hip fracture, as ICF code d450.32. This specific ICF code denotes that walking is in the activities and participation domain (alphabetical code d), the mobility domain (chapter 4), and the walking category (50) and that the patient has severe performance difficulty (3) and moderate difficulty with capacity (2).

From a methodological perspective, the magnitude of measurement error inherent in using the ICF Core Sets is currently unknown. As the authors themselves conclude, “One important caveat is that neither the ICF, as a classification, nor the ICF generic Core Set, as an extraction from it, should be considered a substitute for standard clinical outcome measures. Psychometric evidence (reliability and responsiveness to change) for ICF-derived measures is lacking.”¹ Is now the time to advocate using the ICF Core Sets as an evaluation and documentation system for use in physical therapist clinical practice? At the present stage of development, do we know the degree to which the ICF Core Set classifications are reliable when used by patients or by different therapists? Do we know the magnitude of measurement error when used over time to gauge a patient's level of improvement? How is a clinician to evaluate changes in ICF codes such as those reported in Rauch and colleagues' case report?⁴ Have the ICF Core Sets been subjected to thorough psychometric evaluation and testing in clinical applications such as those being advocated by the authors?

The ICF single-item coding approach to categorization with qualifiers consists of nominal and ordinal level of measurement.¹³ I fear that single-item measures such as these are fraught with considerable measurement error. For example, Okochi et al,¹⁴ in one of the few studies of test-retest reliability of the ICF Core Sets, found ICF qualifiers of body function and activities and participation items to have low test-retest reliability within a 1-week interval with a sample of stable elderly individuals using long-term care services in Japan. With low test-retest reliability, the user cannot be confident in the ability of ICF coding to distinguish clinically meaningful change in heath-related states from measurement error. The ICF coding may have limited utility for classifying individual states of health, discriminating ICF health-related states between groups of people, or evaluating meaningful change over time.

Might it not be more prudent to advocate the careful evaluation of the ICF Core Sets first to determine whether the ICF coding approach is psychometrically adequate for use in clinical practice? In the absence of psychometric evaluation, how can a clinician interpret the clinical relevance of the ICF Core Sets for use in physical therapist practice?

• © 2010 American Physical Therapy Association