Abstract
Keywords
I feel good!
They called him the Godfather of Soul and the Hardest Working Man in Show Business, and when he stepped onto the concert stage, it didn’t take long for the audience to understand why. In performance, James Brown emitted the energy of a barely contained thermonuclear reaction. The moment he appeared, spotlights flashing off his luxuriant shock of hair, a lightning bolt of electricity crackled through the auditorium. The band would get things going: a driving beat, punctuated with staccato blasts of saxophone and brass. Then James would unleash the song in a distinctive style somewhere between singing and shouting, with an occasional scream thrown in for added excitement. Scantily clad go-go dancers streamed in from the wings, with Brown himself soon joining in their frenetic gyrations. The dancing frenzy would quickly spread to the backup singers and even the band itself, until the audience could barely resist jumping to their feet when James commanded, “Get funky!”
Brown’s signature number was “I Feel Good.” He performed it with such energy and conviction that whatever tribulations the artist might be enduring, whatever worries might be troubling the audience, they couldn’t help but feel good, if only for the duration of the number. Although he passed away in 2006, his definitive rendition of “I Feel Good” continues to explode onto the soundtrack of motion pictures and television shows whenever the feeling of exuberant, unbounded and, yes, funky, happiness is called for. When the saxophone riffs and James Brown shouts, “I feel good,” you do too.
Patients seek us out in the hope that we will be able to relieve their pain and improve their function. They are appreciative if we can help them feel better, but ideally they don’t just want to feel better, they want to feel good. 7,14,16 If possible, they probably want to feel as funkadelicly good as James Brown seemed to feel onstage. Numerous patient-reported outcome (PRO) measures have been developed to gauge the effect of our treatments on the way patients feel. The challenge for researchers is to report these outcomes in a way that is interpretable and relevant to individual patients.
Most commonly, researchers report the mean or median value of a group of patients. They may compare the PRO scores before and after treatment and proudly proclaim that this difference was statistically significant. They may compare 2 groups of patients or 2 different treatments and tell us whether there was a statistically significant difference between them. While such analyses may be scientifically sound, they do not tell us whether the significant differences are large enough that patients will perceive an improvement or, more desirably, feel good.
A number of techniques have been proposed to evaluate the clinical importance of changes in PROs. 5,10,11,16 Names given to these measures include the minimal important difference (MID), the minimal clinically important difference (MCID), the minimal clinically important improvement (MCII), the subjectively significant difference (SSD), the clinically important difference (CID), and the minimally important change (MIC). Some of these reflect patient perception, some take into account the judgment of the physician, and some are purely statistically based. Some of them measure the smallest perceptible change and others the smallest desirable change. 6
As summarized in an excellent review by King, 10 these changes are most commonly established with 1 of 4 methods. Two of these are statistical in nature and thus, strictly speaking, only proxies for a clinically important difference. The standard error of measurement (SEM) is a property of a PRO scale that is calculated mathematically from the standard deviation (SD) and reliability (R). The SEM identifies the amount of difference that is greater than the uncertainty produced by the scale’s inherent measurement error. The effect size may be thought of as demonstrating the signal-to-noise ratio of a PRO; it is calculated by dividing the mean change in the PRO by the variability among individuals as indicated by the SD. The utility of the effect size is that, as a ratio, it is unitless and can be used to compare studies that report their results with different PROs. Like the SEM, the effect size does not directly address the clinical importance of a demonstrated outcome. However, an arbitrary convention proposed by Cohen is often used to describe the magnitude of the effect size, with 0.2, 0.5, and 0.8 chosen to represent small, medium, and large effect sizes, respectively. 10
A third way to establish a meaningful change in a PRO is to correlate it with a clinical anchor considered by clinicians to reflect quality of life. This might be an activity scale or a clinically important measurement, such as serum hemoglobin 3 or level of pain on a visual analog scale (VAS). 2 The fourth and final method for measuring the clinical importance of a change in PRO is the one most directly connected to the patient’s experience: the use of a global transition question. With this method, patients are asked to compare their current state with their baseline and classify the difference according to a 5-category scale, ranging from “much worse” to “much better.” (Funkadelicly better is not usually offered as an option.) Typically, the mean change in the PRO score among patients who feel “a little better” (or worse) is declared the minimal important difference. Some commonly used variants incorporate an element of the clinician’s judgment with the patient’s perspective, such as the MID of Guyatt et al 8 : “The smallest difference in score in the domain that patients perceive as important, either beneficial or harmful, and that would lead the clinician to consider a change in the patient’s management.”(p377)
These measures are useful for putting statistically significant findings into perspective, telling us whether they are large enough to be clinically relevant. They are particularly valuable when used to calculate the sample size needed for a prospective study, which assures us that the study is large enough to detect clinically meaningful differences. However, when used to report the findings of a study, they are usually applied to the average change for an entire cohort of patients, which still may be difficult to relate to the expectations of an individual. Reporting the percentage of subjects who achieve the minimally important change is helpful because this tells us the chances of an individual experiencing a meaningful improvement, of feeling better. It does not, however, indicate the odds of feeling good.
If we want to know how often a given treatment will allow a sick or injured person to feel good, we first need to define what good means. Once upon a time, it was common to group PRO scores into categories, with 90-100 classified as excellent, 80-89 as good, and so on. Such a classification may have been arbitrary and unscientific, but it flourished because it translated a numerical score into terms that seemed meaningful to patients. In this issue of The American Journal of Sports Medicine, Chahal and colleagues 4 determine a similar but more scientifically derived concept, the patient acceptable symptomatic state (PASS), for 2 commonly used PROs for femoroacetabular impingement (FAI) of the hip: the modified Harris Hip Score (mHHS) and the Hip Outcome Score (HOS).
The concept of the PASS has been discussed extensively in the rheumatologic literature, 17 but is relatively new to orthopaedic surgery. 13 It is derived by asking patients an anchor question designed to determine whether they are satisfied with their current state of health. Patient response options are limited to “yes” or “no.” The question that Chahal et al used is typical: “Taking into account all the activities you have during your daily life, your level of pain, and also your functional impairment, do you consider that your current state is satisfactory?”4(p1846),15
One of 2 methods is then used to derive the PASS. 11 In the first method, the 75th percentile score of patients who responded “yes” is declared to be the PASS for that PRO. 16 In the second method, a receiver operator characteristic (ROC) curve is constructed to identify the cutoff score that optimally defines the PASS, based on sensitivity and specificity. Chahal and colleagues queried 130 patients, 42% male and 35.6 ± 11.7 years of age, who were 12 months after surgical treatment of FAI. Using the ROC method for their primary analysis, they derived a PASS of 74 for the mHHS, 87 for the HOS activities of daily living (HOS-ADL) subscale, and 75 for the HOS sport subscale; these scores were attained by 71%, 66%, and 61% of their patients, respectively.
The PASS for a PRO is intended to establish a benchmark for therapeutic success, although it may vary with the condition being treated and the population studied. In the study of Chahal and colleagues, the PASS was not affected by the patients’ baseline scores. Not surprisingly, however, patients with higher baseline scores were more likely to achieve the PASS, although age and sex were not significantly related to the odds of achieving it. In other studies, age, baseline PRO score, duration of symptoms, or length of follow-up have been found to influence the PASS. 13,15 For example, in a study deriving the PASS for pain in patients being treated for rotator cuff disease, Tashjian et al 13 reported that a higher pain level was acceptable to older patients, whereas the amount of pain that was acceptable diminished as the time from the initiation of treatment increased. Tubach et al 15 found that the PASS for the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) in patients with osteoarthritis of the knee and hip varied moderately according to the baseline score, but not with age, disease duration, or sex. It is reasonable to presume that the PASS might vary according to individual patient expectations and activity level: The PASS for knee function would probably differ in professional athletes compared with a group of weekend warriors.
In the sports medicine literature, other metrics besides the PASS have been used to determine whether patients have regained a satisfactory level of function. The numerous studies that report the percentage of patients who return to competition and their prior performance level would be one example. Using an approach similar to the PASS, in this issue of AJSM, Ingelsrud et al 9 report the percentage of patients who agreed that their knee function was satisfactory after ACL reconstruction. In a 2013 study from the Swedish National Ligament Register, Barenius et al 1 reported the percentage of ACL reconstruction patients who achieved a functional recovery, which they defined as regaining the lower limit of the 95% confidence interval previously established for normal 18- to 34-year-old Swedes on the Knee injury and Osteoarthritis Outcome Score (KOOS). Yet another method for defining a successful outcome after ACL injury was recently proposed by Lynch et al. 12 These authors surveyed 1779 members of a number of international sports medicine associations and obtained a consensus on 6 measures deemed to indicate a successful recovery.
Reporting the percentage of patients who achieve a therapeutic benchmark such as the MID or the PASS in clinical studies is a useful feature that will help patients understand the likelihood of a successful outcome following surgery or other treatment. Standards of clinically important difference such as the MID or MCID illustrate the chances of feeling better, while the PASS and other measures of successful recovery indicate the chances of feeling good. However, it is important to remember that these metrics are still based upon the opinions of the particular population used to derive the value, and that individual patients will have different expectations and definitions of success. What constitutes an acceptable state of health for most patients may be a disappointing result for the individual who expects a more funkadelic outcome.