To determine the most appropriate means to assess the response to treatment in terms of pain and functional impairment in a chronic rheumatic condition (knee osteoarthritis [OA]) and an acute rheumatic condition (rotator cuff syndrome [RCS]).
Two prospective studies were conducted consisting of 1,019 outpatients with knee OA and 271 patients with acute RCS. The minimal clinically important improvement and the patient acceptable symptom state were determined for knee OA pain using a visual analog scale, and for knee OA function using the Western Ontario and McMaster Universities Osteoarthritis Index function subscale; for acute RCS pain, a numeral rating scale was used, and the Neer function subscale was used for RCS function.
The minimal clinically important improvement was shown to be the change required to achieve the patient acceptable symptom state, whatever the baseline level of symptom, the outcome (pain or function), or type of condition (chronic or acute). This acceptable state for pain was higher for chronic (27.0–36.4 across the baseline score) than acute (16.7–24.1) conditions. The level of functional impairment considered satisfactory by patients with knee OA was higher for more disabled patients (43.1) than for less disabled patients (20.4).
Patients consider that they experienced an important improvement only if this improvement allowed them to achieve a state they consider satisfactory. The most appropriate means to assess the response to therapy seems to be to assess whether patients feel good (i.e., achieve the patient acceptable symptom state).
In chronic diseases, the aim of treatment is to improve patient symptoms, function, and well-being. Consequently, patient-reported outcomes (PROs) are widely used in assessing the result of care in clinical practice, clinical trials, and longitudinal epidemiologic studies. PROs are valid and reliable measures, and are sensitive to clinically important changes (1, 2). PROs range from reported symptoms such as pain to the impact of physical function, social function, mental health, and general health perception. Because these measures describe what the patient has experienced as a result of medical care, they are useful and important supplements to traditional physiologic or biologic measures of health status.
Several tools are available to assess PROs, but their results, usually expressed as continuous data, are not meaningful to most people. How can clinicians determine what constitutes a clinically relevant treatment success when using PROs? Two concepts could help in interpreting PRO scores at the individual level: 1) the minimal clinically important difference (3–5) or minimal clinically important improvement (MCII) (6), defined as the smallest change in measurement that signifies an important difference/improvement in symptom; and 2) the patient acceptable symptom state (PASS) (7), an intermediate state between activity of the disease and complete remission, defined as the score below which patients consider themselves well. Inference at the individual level allows for describing the proportion of responders or nonresponders (patients who have improved or achieved an acceptable state or not) in addition to mean effects. Such inference will add useful information and aid in the interpretation of trial and longitudinal results to decide whether a treatment should be used. These thresholds also allow for monitoring individual response to therapy over time and adapting treatment at the individual level (e.g., determining whether to modify or interrupt a treatment). However, it is not known how the MCII and PASS are interrelated, and is unclear which one should be recommended.
The aim of this study was to evaluate how the concepts of MCII and PASS relate to each other in terms of pain and function scores in a chronic rheumatic condition (knee osteoarthritis [OA]) and an acute condition (rotator cuff syndrome [RCS]), and therefore to determine the most appropriate means to assess the response to treatment in these conditions in terms of pain and functional impairment.
MATERIALS AND METHODS
Role of the funding source.
Merck Sharp and Dohme Chibret participated in the study design, data collection, agreement to submit the manuscript, and approval of the content of the manuscript. They did not participate in the data analysis or in the writing of the manuscript.
Data from 2 previous studies were used. The first study was a prospective cohort of 4 weeks' duration, involving 1,019 outpatients with painful knee OA (pain score ≥30 mm on a 0–100-mm visual analog scale [VAS]), as defined by the American College of Rheumatology criteria (8). The design was described in detail elsewhere (6, 7). Briefly, after an initial baseline visit, patients received nonsteroidal antiinflammatory drugs (NSAIDs) for 4 weeks. At both the baseline visit and final visit (week 4), patients assessed pain on movement measured on a VAS, and functional impairment measured with the Western Ontario and McMaster Universities (WOMAC) Osteoarthritis Index function subscale (Likert 5 version, where higher scores indicate higher degree of functional impairment) (9). All scores were normalized to 0–100 scale. At the final visit, a random two-thirds (n = 688) sample of patients were asked to assess their response to NSAID treatment using a 5-point Likert scale (possible responses were none, poor, fair, good, excellent). In the total sample, patients' opinion of their state was recorded. Patients were asked to respond yes or no to the question “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?”
The second study was a 3-arm randomized, double-blind, double-dummy, 7-day trial comparing treatment with 2 different NSAIDs or placebo in 271 patients with an acute painful flare (pain on movement ≥5 on a 0–10 numeral rating scale [NRS]) with clinically diagnosed RCS lasting <7 days (10). Patients in all 3 arms were considered together in the analysis (whether they received NSAIDs or placebo). After a baseline visit, the patients received NSAID therapy for 7 days. At the baseline and final (day 7) visits, patients assessed their pain on movement, as measured on an NRS; and functional impairment, as measured on the Neer rating function subscale (where higher scores indicate higher degree of functional impairment) (11). All scores were normalized to 0–100 scores. At the final visit, the patients assessed their response to NSAID treatment on a 15-point Likert scale (where –7 = a very great deal worse, 0 = no change, +7 = a very great deal better), and the patients' opinion of their state was also recorded by their answering “Yes” or “No” to the question “Taking into account your level of pain and also your functional impairment, if you were to remain for the next few months as you are today, would you consider that your current state is satisfactory?”
Both studies were conducted in compliance with the Good Clinical Practice and the Declaration of Helsinki principles. In accordance with French national law, all patients gave written informed consent.
An anchoring method based on the patient's assessment of response to therapy was used to determine the MCII and the PASS for each outcome criteria. As described previously (6), the MCII was estimated by constructing a curve of cumulative percentages of patients as a function of the change in score (e.g., difference in pain score) among patients whose final evaluation of response to therapy was “good.” This procedure was used because we wanted to focus on the improvement that was clinically important. Logistic regression was used to model the observations. We targeted the point at the flattening of the curve at which most subjects stated they had improved. To determine the change in score corresponding to this point, we first looked at the 2-parameter logistic model that best fit the data. Then we determined the square root of the third derivative of this logistic function that corresponded to the MCII. One can demonstrate that this point corresponds by construction to the 78.9th percentile of the change in score, and thus we propose to define the MCII as the 75th percentile of the change in score, because it is very close to the point defined above and easier to derive. The model permitted us first to determine that the target point was correctly approached by the 75th percentile, and second to estimate the 95% confidence intervals (95% CI).
The MCII was therefore defined as the 75th percentile of the change in score between the baseline and final visits among patients whose final evaluation of response to therapy was good (i.e., in the knee OA study, those who answered “good,” and in the acute RCS study those who answered “a great deal better” [+5 score] or “a good deal better” [+6 score]). This means that this change (the MCII) or a higher change was experienced by 75% of these patients.
Following the same methods as for the MCII (by constructing a curve of cumulative percentages of patients as a function of the score of interest at final visit among patients who considered their state satisfactory) (7), the PASS was defined as the 75th percentile of the final score (at the final visit) for patients who considered their state satisfactory. This means that this level of symptom (the PASS) or a lower level was achieved by 75% of these patients. Logistic regression was used to model the observations and compute the 95% CI.
Because the baseline score (pain or functional impairment) was previously demonstrated to have a modifying effect on the MCII and PASS (6, 7) for each study we stratified the analysis on the baseline score of interest (i.e., for pain the WOMAC function score and Neer rating function score) divided into tertiles (i.e., low, intermediate, and high baseline score) to assess whether the baseline scores affected the PASS estimates. Statistical analyses were performed with SAS software, release 9.1 (SAS Institute, Cary, NC) and S-Plus, version 6.2 (Insightful, Seattle, WA).
Clinical and demographic characteristics of the patients are shown in Table 1. A total of 914 (90%) patients with knee OA and 252 (93%) with acute RCS completed the final visit. Patients lost to followup were excluded from the analysis; but their baseline characteristics did not differ from those who completed the study (data not shown).
Table 1. Baseline characteristics of patients in knee osteoarthritis (OA) study and acute rotator cuff syndrome (RCS) study*
Knee OA (n = 914)
Acute RCS (n = 252)
Values are mean ± SD, unless otherwise indicated. Change = final score minus baseline score; WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index.
Female, no. (%)
67.8 ± 10.2
47.6 ± 10.4
Disease duration, years
4.8 ± 5.8
Pain score (scale 0–100)
58.3 ± 16.9
67.9 ± 14.7
−24.5 ± 22.1
−36.8 ± 24.9
WOMAC function score (scale 0–100)
42.9 ± 16.6
−11.6 ± 14.4
Neer function score (scale 0–100)
42.1 ± 15.0
−18.3 ± 19.4
The interrelationship between MCII and PASS is shown in Figures 1 and 2. For example, in the pain analysis in patients with knee OA in Figure 1, the low tertile initial mean score (TIMS; i.e., the mean pain score in the lower tertile of the baseline pain score) was 40.0 mm, the MCII was −10.8 mm (i.e., patients considered themselves clinically improved if the decrease in pain exceeded 10.8 mm on the VAS), and the PASS was 27.0 mm (i.e., patients considered their state satisfactory if the pain score at the final visit was below 27 mm on the VAS). Whatever the tertile, when subtracting the MCII from the TIMS, the obtained value is very close to the PASS (in this example, 40.0 minus 10.8 = 29.2 mm; PASS = 27.0 mm). The minimal change considered important to the patients seems to be the change that makes the pain decrease from baseline to the PASS.
The MCII (95% CI) for pain in patients with knee OA varied from –10.8 (−12.7, 8.7) to –36.6 (−38.3, −34.7), and the MCII (95% CI) in patients with acute RCS varied from −34.3 (−38.2, −29.3) to −62.5 (−68.3, −54.0). The PASS (95% CI) for pain in patients with knee OA varied from 27.0 (24.6, 29.9) to 36.4 (33.2, 40.0), and the PASS (95% CI) in patients with acute RCS varied from 16.7 (13.4, 20.7) to 24.1 (19.3, 30.0).
For functional impairment in knee OA, the MCII (95% CI) varied from –5.3 (−6.5, −3.8) to −20.4 (−22.5, −18.1), and from –15.1 (−16.9, –12.9) to –37.3 (−39.4, –34.8) in acute RCS. The PASS (95% CI) for functional impairment in knee OA varied from 20.4 (19.1, 21.8) to 43.1 (40.8, 45.7), and from 12.7 (11.0, 14.8) to 21.1 (18.4, 24.3) in acute RCS.
Medical decision making relies increasingly on PROs. Therefore, providing physicians with PRO-based practical tools that are comprehensive, and can meaningfully influence clinical decision making is crucial. The importance of defining a minimal important difference or an acceptable state is widely accepted (12, 13). This article adds to existing knowledge about the MCII and PASS by explicitly conceptualizing the relationship between the 2 concepts using 2 types of PROs (pain or function) and 2 diseases (chronic or acute). It also improves our knowledge concerning the most appropriate means to assess the response to treatment regarding pain and functional impairment in daily practice or in clinical trials.
The MCII varied greatly across tertiles of baseline scores in both pain and function assessment and in chronic and acute conditions. For patients with more severe disease, a greater change in score was required to consider themselves improved than for patients with less severe disease. The PASS was more constant across tertiles of the baseline score for the assessment of pain in both types of diseases. When subtracting the MCII from the TIMS, the obtained value is very close to the PASS. This result has been replicated in another independent study of patients with knee OA (data not shown). The MCII clearly appears to be the change needed to achieve the PASS, whatever the baseline level of symptom, patient-reported outcome (pain or function), or type of disease (chronic or acute). Patients possibly consider that they experienced an important improvement only if this improvement allows them to achieve a satisfactory state, a state in which they feel good. Consequently, it seems that what is important to patients is to feel good (concept of PASS) rather than to feel better (concept of MCII). This finding suggests that the outcome measure in clinical trials and in daily practice should be a sense of well-being on the part of the patient rather than a sense of improved health.
In a chronic condition (knee OA), the PASS for functional impairment varied across tertiles of the baseline score. This pattern was not observed for functional impairment in an acute condition (RCS), nor for pain in either condition. It seems that a satisfactory level of pain in the patients' perspective is approximately the same, whatever the initial level of pain or type of disease. Patients' expectations about pain relief do not vary with time; one cannot get used to a high level of pain. On the contrary, for functional impairment in a chronic condition such as knee OA, the most impaired patients seem to judge as satisfactory a higher level of functional impairment than the least impaired patients. Patients seem to be resigned to functional impairment and can adapt to it with time. In an acute condition such as RCS, the PASS for functional impairment was, for the most part, stable, whatever the initial level of impairment, as compared with chronic knee OA. This discrepancy between acute and chronic conditions could be explained by the fact that patients with chronic diseases have time to adapt to their limitation or, perhaps because high pain intensity in acute RCS (unlike that in knee OA) is associated with high functional impairment and low pain with low impairment, because the limitation is largely due to pain intensity. Therefore, the observed pattern for functional impairment in acute RCS may copy the pain pattern, not the impairment pattern.
The PASS for pain was higher for the chronic condition (knee OA) than the acute condition (RCS). Taking into account that the baseline level of pain was higher in the acute RCS study, the intermediate tertile in the knee OA study (TIMS 58.0, PASS 34.5) corresponds to the low tertile in the acute RCS study (TIMS 54.9, PASS 16.7) and the high tertile in the knee OA study (TIMS 77.1, PASS 36.4) corresponds to the intermediate tertile in the acute RCS study (TIMS 70.0, PASS 23.4). The PASS for functional impairment scores also seem to be higher in the chronic condition study than the acute condition study; however, the tools used to assess functional impairment were disease-specific and therefore differed between the 2 studies. Although scores were normalized, such comparisons between the 2 must be made cautiously. Patients' expectations seem to be different in chronic and acute conditions.
Our study had several limitations. We studied only 2 conditions (knee OA and acute RCS) both in the rheumatologic field, and musculoskeletal conditions have been shown to have a great impact on quality of life. Among patients with a wide range of chronic diseases, those with musculoskeletal diseases reported the lowest level of physical functioning and the highest level of pain (14). Otherwise, knee OA and acute RCS are believed to be 2 relevant pain and function models, one being an acute condition and the other a chronic condition. The definition of a good response to therapy for pain or functional impairment may not be specific to the disease, and therefore these results could be applied to other medical conditions in which patients have pain and/or functional impairment.
Another study limitation is that the wording for assessing patients' opinion of their state (to determine the PASS) differed slightly between the knee OA and RCS studies. It has been demonstrated in a study of asthma patients using an asthma-specific quality of life questionnaire (15) that the external anchor's wording has an impact on the clinically meaningful change data. Future studies should evaluate the impact of wording on evaluating the PASS.
Because the MCII seems to be the change required to reach the PASS, the MCII and PASS might reveal the same information. However, the importance of the baseline score on the MCII justifies the use of 1 cutoff per tertile of baseline score if using the MCII, which is less convenient than using the same PASS for all patients.
In our opinion, the PASS rather than the MCII should be recommended to assess response to treatment, first, because it is least sensitive to baseline levels of symptoms, and second, because it is a simple and intuitively reasonable concept that seems to better reflect what is important to patients. In daily practice, patients should be asked if they feel good rather than if they feel better. This clearer understanding of how the MCII relates to the PASS could have important implications for defining treatment success and understanding what matters to patients. Our results might encourage the evaluation of treatment strategies as goal-directed therapy (treating and intensifying the treatment until the patient achieves the PASS) in clinical practice and determining the proportion of patients achieving the PASS in clinical trials, in addition to the difference between baseline and final scores. This finding could help the physician better understand the expectations for treatment care of chronic patients with pain and functional impairment and therefore adapt their treatment to their objectives.
In conclusion, the results of this study suggest that a change is considered an important improvement to patients with knee OA or acute RCS if patients can achieve an acceptable symptomatic state, and the most appropriate means to assess response to treatment is to ask the patient if they feel good rather than if they feel better. Patients seem to accept higher levels of pain and functional impairment in chronic than in acute conditions. In chronic conditions, patients who are more disabled seem to accept a higher degree of functional impairment than patients who are less disabled, but not a higher level of pain.