Subjective measures and clinical decision making.

Subjective Measures and Clinical Decision Making In the keynote address of this issue of Physical Therapy, Magistro mentioned the fear that medical care is becoming increasingly dehumanized. This unfortunate trend is most likely the result of many influences, some of which have been described previously. Feinstein sees an inordinate dependence on paraclinical methods (eg, anatomical, microbiological, electrophysiological) at the expense of what he calls distinctively clinical phenomena to be at least partially causing this dehumanizing trend. [1] The distinctively clinical phenomena include type and severity of symptoms, rate of progression of illness, functional capacity, and "other aspects of the physical activities, joys and sorrows of daily life." These "soft" data are usually lowly regarded and even intentionally repudiated among researchers and clinicians. This leads to little regard for the patient's description of his or her illness when making clinical decisions.

Rather than ignore or deliberately reject these phenomena, Feinstein promotes their quantification. Clinimetrics involves the quantifying of the realm of clinical data that are observed, judged, and decided during clinical examination by clinicians themselves. [1] Feinstein advocates the development of clinimetrics as an expansion of the scientific basis of clinical practice. Furthermore, he sees clinimetrics as a tool that will rehumanize the management of patients in the clinical setting.

Although we, as physical therapists, acknowledge the importance of a patient's verbal description of his or her illness, these aspects of the assessment are rarely quantified. In fact, the term "subjective" at the beginning of the problem-oriented medical record's SOAP (subjective, objective, assessment, plan) format dictates these data are soft. Thus, the patient's symptoms, such as pain, disability, and difficulty, are relegated essentially to anecdotes documented at the beginning of the examination. Everyone would agree that information regarding subjective phenomena of illness serves as an important base for clinical decision making. This base could be strengthened, however, if these data were made "hard." Making "soft" data "hard" requires quantifying the clinically important subjective phenomena. Transforming soft information to hard, quantifiable measurements is the basis for Feinstein's clinimetrics.

Previous attempts at quantifying soft data pertinent to physical therapy are best illustrated by the functional indexes described for various disorders by various authors. Functional indexes for patients with arthritis, [2,3] low back syndrome, [4,5] and knee injuries [6] are most common. These indexes, however, are not in widespread use in everyday clinical care. Instead, most clinicians note subjective complaints and then proceed to the "meat" of the examination involving the physical examination, or the "objective" findings of the assessment. At times, it appears that our effort is directed at further objectifying our physical examination. We measure, arguably to the point of scrutiny, joint range of motion to the nearest 5 degrees, muscle strength to the nearest footpound, and so on. In a knee examination, for example, instruments are now available that measure joint laxity in millimeters. In isokinetic muscle strength assessments, the addition of computers offers the clinician a multitude of numbers that represent torque production on the part of the patient. In gait assessment, we teach students to place inkblots on the bottom of a patient's shoes before having the patient walk on strips of paper, thus obtaining a "quantitative gait analysis." The point here is not to downplay objective measures. Some "hard" measures, however, need not be made harder, especially when there is such a void of quantifiable measures of the subjective phenomena.

The question I would pose to physical therapists is, Do these measurements, commonly referred to as the "objective" portion of patient assessment, become analogous to medicine's "paraclinical" measurements, as described by Feinstein? In orthopedic physical therapy, for example, there is a tremendous emphasis on isokinetic muscle strength assessments. To some clinicians, these measures are a major determinant in the decision for an athlete to return to his or her sport. Given the above, how much are we contributing to the dehumanizing of clinical management by seemingly encouraging the further use and emphasis on traditionally "objective" findings?

My purpose in this short discussion is to clearly outline criteria necessary for sound clinimetric indexes in physical therapy. Furthermore, I will illustrate an easily administered clinimetric index that attempts to quantify what we, as physical therapists, traditionally note as "subjective" under the typically used "problem-oriented" framework.

"Sensible" Clinimetric Indexes

Feinstein describes five basic features of "sensible" clinimetric measures. [1] First, the clinimetric index must be suitable for its clinical purpose and setting. In the case of clinical purpose, an example would be the clinimetric index that is intended to be useful for determining a treatment, but that is of little use in diagnosis. Many potential clinimetric measures would fit this description in physical therapy. For example, consider the patient with an anterior cruciate deficient knee. A few clinical tests confirm the diagnosis, but considerations such as activity level the patient aspires to, incidence of instability, and degree of present disability are all potential clinimetric indexes that clinicians will use to guide treatment. In the case of suitability in the clinical setting, imagine an index needs to be designed to help predict functional capacity in the hemiplegic patient. One approach may be to administer the test at admission; another choice may be to test the patient at discharge. It is easy to see how these two tests may be composed of entirely different elements. Thus, both indexes, even though designed to predict outcome of patients with the same diagnosis, are not parallel because of their suitability in different clinical settings.

The second principle of sensible clinimetric indexes is face validity. Reliance on face validity in psychometric measurement is now downplayed. In fact, in the most recent edition of the American Medical Association's Standards for Educational and Psychological Testing, the term face validity is not mentioned. Unlike other requisites for sensible clinimetric indexes that can be assessed quantitatively, face validity refers to "scientific" common sense. In fact, the inability to quantify face validity comprises a major reason face validity, as applied to psychometric measurement, is disavowed.

Basic common sense applies to clinimetric indexes, and Feinstein admits that quantitatively assessing this entity is difficult because it cannot be measured statistically and is difficult to define judgmentally. [1] Specifically, he points to the following areas: 1) the focus of the interpersonal exchange, 2) the focus of basic evidence, 3) the biologic coherence of components, and 4) the attention to personal corroboration. During the focus of the interpersonal exchange, problems can arise when the subject misunderstands or misinterprets the questions given. We have found that in low back assessment, for example, the terms "worsen" and "increase" have specific and very different definitions when describing symptomatology. Misinterpreting the definitions by the clinician or patient will result in deleterious effects on the quality of the decisions made that are based on this index. Additionally, does the index take into account the patient's cooperation? For example, how do we handle the clinimetric index of disability that assesses a patient receiving workman's compensation we suspect of not cooperating to the fullest?

In describing the "unsuitable displacement" of focus of basic evidence, Feinstein points toward the inappropriate extrapolation of the focus of the evidence (eg, functional capacity) from a clinical test (eg, isokinetic assessment) that will comprise the clinimetric index. [1] Here, common sense appears to prevail. In the above example, muscle strength would logically be related to functional capacity, so an isokinetic test of muscle strength would logically belong in a clinimetric index of functional capacity. Others may argue with this "logic" (or common sense) by stating that isokinetic assessment is flawed in that functional strength is not assessed by an isokinetic dynamometer, perhaps using reasoning such as "people do not move isokinetically." As can be seen, the evaluation of such an item will be an act of qualitative judgment.

The biological coherence of the components that make up the clinimetric index must be apparent, according to Feinstein. [1] Again, common sense prevails here. In the example above dealing with isokinetic muscle assessment, biological coherence can be argued to be high, mostly based on the accommodating resistance component of isokinetic exercise. Because maximal resistance is applied throughout the ROM, maximal muscle strength is assessed. Thus, the test appears to have biological coherence. Summarizing, although face validity is problematic in psychometric theory, defining face validation as merely judging whether a clinimetric measure makes "scientific common sense," as Feinstein does, is well taken.

The third principle of sensible clinimetric indexes described by Feinstein includes content validation. [1] This aspect deals with the component elements of the clinimetric index. Content validity pitfalls in clinimetric indexes include omission, inappropriate inclusions, and improper weighing of component parts. For example, nobody will deny that sexual dysfunction accompanying low back syndrome comprises a major component of the resultant disability seen in these patients. Yet the sexual dysfunction of a patient with a backache may be omitted from the assessment because of a clinician's inability to address the issue with the patient or the patient's unwillingness to discuss sexual dysfunction with the clinician. By not taking sexual dysfunction in the patient with low back syndrome into account, however, an index of disability may be severely lacking.

The fourth principle of sensibility in clinimetric indexes deals with the formal expression of the index. Here, Feinstein describes six features: 1) comprehensibility, 2) as few variables as possible, 3) a relatively transparent set of contents, 4) replicability, 5) a suitable scale of expression, and 6) a scale of expression with suitable sensitivity to discriminate. [1] Feinstein often uses the Apgar score for newborns as an example of a clinimetric index that fulfills all six of the above criteria. In physical therapy, Cibulka et al propose an index of sacroiliac joint dysfunction. [7] This index is based on the patient testing positive in three of four common measures of sacroiliac joint dysfunction.

The last principle of sensibility in clinimetric indexes involves ease of usage. Here, we are focused on minimal time to administer the index while still adequately measuring the desired phenomena. The chronology of pain in the patient with low back syndrome may be important for both diagnostic as well as treatment purposes. We can call the patient hourly, in which case we will have the desired data. Or we can administer a questionnaire daily or every two days. In the latter case, obvious time savings are involved, and we will probably sufficiently measure the desired phenomena. The difficulty here lies in drawing the line between time savings and data loss. Another example involves outcomes in patients with low back syndrome. For example, with acute backache, most patients are clearly close to normal after 10 days, despite what is done to them. If our intervention is to have any effect, we are usually talking in terms of getting better sooner. To appropriately assess this, however, requires almost daily appraisal of patients symptomatology. It would be an obvious hurdle for a patient to be seen daily for this assessment, so a home questionnaire is a viable alternative here. Of course, this requires compliance on the part of the patient.

Examples

Functional assessment indexes exist in various forms and are relevant for patients commonly seen in physical therapy. For example, there are well-described forms for patients with arthritis, low back syndrome, and knee injuries. [2-7] Are these various assessments undiscovered solutions to the abovementioned lack of documentation of a patient's "subjective" complaints? Or are they clinically unusable because they are not "sensible" clinimetric indexes?

I will offer what I believe are possible answers to these questions. These are:

1. Perhaps some functional indexes, although they appear on the surface to fulfill most criteria for clinimetric assessment, remain difficult to administer. They may require too much time on the part of the patient and clinician or too much understanding on the part of the patient. Or perhaps they fail to provide the clinician with meaningful "numbers." Because of these and probably other reasons, therefore, and because they fail to meet Feinstein's criteria for "sensible" clinimetric indexes, [1] they are unused in most everday clinical situations.

2. Perhaps clinicians are reluctant to rely on data that on the surface appear "too soft to rely on," despite appropriate experimental validation. If clinicians, for some reason, cannot force themselves to use measures that on the surface appear soft despite good evidence that these measures fulfill all criteria for a hard measure, then perhaps the problem lies somewhere in our teaching. I have noticed that in an effort to model our teaching after medicine, we are increasingly trying to "objectify" already objective measures. In addition, with increased focus on measurement in the physical therapy field, I do not see many clinicians seeing functional assessment or development of clinimetric indexes as part of this needed trend.

In physical therapy, examples I feel fit Feinstein's criteria include the Oswestry Low Back Pain Disability Questionnaire. This questionnaire was originally described by Fairbanks et al. [5] The actual questionnaire is short and requires no supervision because it is self-administered (Figure). The areas covered are pain intensity and use of pain medication, personal hygiene, walking, lifting, sitting, standing, sleeping, sexual activity, social activities, and traveling. All of these activities are believed to include the most relevant problems suffered by patients with low back syndrome. Each area contains six statements that begin with "Pain prevents me from. . . ." Scores for each area range from 0 to 5 points, following the ordered scale. Patients mark the statement in each area that most accurately identifies their limitations. The questionnaire is scored by adding all scores and creating a percentage of the total number of points possible (50). Fairbanks et al suggest that scores of 0% to 20% represent minimal disability, scores of 20% to 40% represent moderate disability, scores of 40% to 60% represent severe disability, and scores of 60% to 80% represent the patient is "crippled." [5] Patients with scores greater than 80% are considered bedridden.

On the surface, this test appears to fulfill the criteria for sensibility in that it is suitable for the purpose of documenting the patient's perceived disability; it is easy to administer; it has "face validity"; and its formal expression, or score, is easily comprehensible. The test has also shown good reliability. Lacking, however, is empirical evidence of content validation.

Summary

I have attempted to use Feinstein's model of clinimetric indexes and his criteria as a focus for further development of measures that in physical therapy are currently considered "soft" or "subjective." I feel this development will enhance the body of knowledge by objectifying a portion of clinical assessment (eg, the patient's complaints, "subjective" portion of the POMR's SOAP format) that is in tremendous need of quantification. By making these "soft" data "hard," I feel we will enhance the decision-making power of clinicians.

References

[1] Feinstein AR: Clinimetrics. New Haven, CT, Yale University Press, 1987, pp 1-5, 141-66

[2] Meenan RF, Gertman PM, Mason JH: Measuring health status in arthritis: The arthritis impact measurement scales. Arthritis Rheum 23:146-152, 1980

[3] Jette AM: Functional status index: Reliability of a chronic disease evaluation instrument. Arch Phys Med Rehabil 61:395-401, 1980

[4] Rose SJ, Shulman AD, Strube MJ: Functional assessment of patients with low back syndrome. Topics in Geriatric Rehabilitation 1:9-30, 1986

[5] Fairbanks JCT, Couper J, Davies JB, et al: The Oswestry Low Back Pain Disability Questionnaire. Physiotherapy 66:271-273, 1980

[6] Tegner Y, Lysholm J: Rating systems in the evaluation of knee ligament injuries. Clin Orthop 173:43-49, 1984

[7] Cibulka MT, Delitto A, Koldehoff RM: Changes in innominate tilt after manipulation of the sacroiliac joint in patients with low back pain: An experimental study. Phys Ther 68:1359-1363, 1988

A Delitto, MHS, PT, is Instructor, Program in Physical Therapy, Washington University Medical Center, and Doctoral Candidate, Department of Psychology, Washington University, St Louis, MO 63110.

Address correspondence to PO Box 8083, 660 Euclid Ave, St Louis, MO 63110 (USA).