Thresholds of patient-reported outcomes that define the patient acceptable symptom state in ankylosing spondylitis vary over time and by treatment and patient characteristics

Authors

  • Walter P. Maksymowych,

    Corresponding author
    1. University of Alberta, Edmonton, Alberta, Canada
    • 562 Heritage Medical Research Building, University of Alberta, Edmonton, Alberta, Canada T6G 2S2
    Search for more papers by this author
    • Dr. Maksymowych has received consultant fees, speaking fees, and/or honoraria (less than $10,000) from Abbott Laboratories.

  • Katherine Gooch,

    1. Abbott Laboratories, Abbott Park, Illinois
    Search for more papers by this author
    • Ms Gooch, Drs. Wong and Kupper, and Mr. Chen own stock and/or hold stock options in Abbott Laboratories.

  • Maxime Dougados,

    1. Hôpital Cochin, University of Paris, Paris, France
    Search for more papers by this author
    • Dr. Dougados has received consultant fees, speaking fees, and/or honoraria (less than $10,000 each) from Abbott Laboratories and Merck Sharpe & Dohme-Chibret Laboratories.

  • Robert L. Wong,

    1. Abbott Laboratories, Parsippany, New Jersey
    Search for more papers by this author
    • Ms Gooch, Drs. Wong and Kupper, and Mr. Chen own stock and/or hold stock options in Abbott Laboratories.

  • Naijun Chen,

    1. Abbott Laboratories, Abbott Park, Illinois
    Search for more papers by this author
    • Ms Gooch, Drs. Wong and Kupper, and Mr. Chen own stock and/or hold stock options in Abbott Laboratories.

  • Hartmut Kupper,

    1. Abbott GmbH & Co. KG, Ludwigshafen, Germany
    Search for more papers by this author
    • Ms Gooch, Drs. Wong and Kupper, and Mr. Chen own stock and/or hold stock options in Abbott Laboratories.

  • Désirée van der Heijde

    1. Leiden University Medical Center, Leiden, The Netherlands
    Search for more papers by this author
    • Dr. van der Heijde has received consultant fees, speaking fees, and/or honoraria (less than $10,000 each) from Abbott Laboratories, Centocor, BMS, Schering-Plough, Roche, UCB, and Wyeth.


Abstract

Objective

The patient acceptable symptom state (PASS) is a single-question outcome tool to assess the level of symptoms at which patients with rheumatic diseases consider themselves well. We evaluated whether ankylosing spondylitis (AS) patient characteristics were associated with attaining the PASS and whether these characteristics influenced PASS thresholds for patient-reported outcome (PRO) tools.

Methods

The Adalimumab Trial Evaluating Long-term Efficacy and Safety for Ankylosing Spondylitis was a randomized, placebo-controlled study that evaluated the efficacy and safety of adalimumab in treating AS. The PASS and PROs were assessed over 24 weeks. PASS thresholds for PROs were set as either the 25th or 75th percentiles of the PRO response score. Logistic regression analyses were conducted to determine the associations of particular patient characteristics with the PASS and other response outcomes at 12 weeks (ASessment in Ankylosing Spondylitis International Working Group criteria for 20% improvement [ASAS20], ASAS40, ASAS5/6, ASAS partial remission, and Bath Ankylosing Spondylitis Disease Activity Index 50% improvement).

Results

Age >40 years, disease duration >10 years, female sex, placebo treatment, and English-speaking site were consistently associated with greater PASS thresholds for PROs. Age, male sex, disease duration, and treatment were each independently associated with attainment of the PASS at 12 weeks. Only age and treatment were independently associated with other response outcomes. PASS thresholds also decreased over 24 weeks.

Conclusion

PASS thresholds for PROs changed over time. These thresholds, as well as the attainment of the PASS, were affected by covariates unrelated to treatment. If confirmed in other studies, these results cast doubt on using the PASS to assess absolute health status in clinical research.

INTRODUCTION

Clinicians have expressed increasing interest in reporting health status in readily understandable and relevant terms to the individual patient. Such a tool may be particularly helpful in monitoring the impact of changes in treatment and in interpreting the response to new therapeutic interventions in clinical trials (1). Ankylosing spondylitis (AS) constitutes one of the most common arthritides, with a significant impact on patient function and quality of life (QOL). Because most validated outcomes evaluating these domains are assessed as continuous variables, study results are often presented as changes in mean values for treatment groups. Expression of results in such a manner may be of uncertain significance to the clinician in assessing the impact of treatment on individual patients. Consequently, study findings now include not only the mean change data for continuous variables but also the percentage of responders based on the ASessment in Ankylosing Spondylitis International Working Group (ASAS) criteria (2). Related concepts undergoing further validation as an approach to the assessment of absolute health status include the low disease activity state (or minimal disease activity) and the patient acceptable symptom state (PASS) (3, 4).

The concept of the PASS primarily reflects the overall health state at which patients consider themselves to be feeling well. One of the objectives of treatment is to achieve the PASS as soon as possible and then to remain in the PASS. This is somewhat different from achievement of a response because the patient may be better after treatment yet still may not consider him/herself to be in a state of well-being. In the setting of a clinical trial of AS, a patient may be considered an ASAS20 responder as defined by a relative improvement of at least 20% and an absolute change of 1 unit in at least 3 of 4 domains (patient's global assessment, function, pain, and stiffness), but still may not achieve a state of well-being; this highlights the attractiveness of assessing the PASS as a measure of absolute health status. Assessment of the PASS requires only that the clinician ask the patient a single question requiring a yes or no answer: “Considering all the different ways your disease is affecting you, if you would stay in this state for the next months, do you consider that your current state is satisfactory?”

Although the PASS is a relevant concept, there have been relatively few reports addressing the validation of the PASS and its application in rheumatology and AS in particular. One cross-sectional study described the validation of the PASS in AS according to reported flare status, expressed need to see the physician, attainment of ASAS partial remission, and PASS contribution to QOL (5). A second study showed that more patients with active AS who were ASAS20 responders following treatment with adalimumab achieved the PASS compared with ASAS20 nonresponders to adalimumab (6). Significantly more adalimumab-treated patients reported achievement in the PASS compared with placebo-treated patients, indicating excellent discriminant capacity. There has also been considerable interest in defining the PASS concept according to thresholds for the patient self-reported outcomes (PROs) in AS (5–8). However, these estimates have varied in different reports and it is unclear whether these variations reflect study design variability (cross-sectional versus prospective) or other unknown factors. Variability in PRO thresholds defining the PASS is potentially problematic with respect to its use in clinical practice and research, and suggests the influence of important confounders that are unrelated to treatment and could affect the attainment of PASS. A previous study in osteoarthritis (OA) identified disease duration and severity as potential confounders in the ascertainment of minimal clinically important improvement (9). In this study, we defined PASS thresholds for the AS-relevant PROs in a randomized clinical trial in which patients received either adalimumab or placebo. We then evaluated the effects of several covariates on both PASS thresholds and AS-specific response indices; finally, we evaluated the stability of the PASS thresholds over time for each PRO.

PATIENTS AND METHODS

Patient population.

The Adalimumab Trial Evaluating Long-term Efficacy and Safety for Ankylosing Spondylitis (ATLAS) was the pivotal trial of adalimumab for the treatment of patients with active AS. Detailed safety and efficacy data from this trial have been reported previously (10). Patients were age ≥18 years with a diagnosis of active AS based on the modified New York criteria (11). Active AS was defined as fulfillment of 2 of the following criteria: a Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) (12) score ≥4, a total back pain score ≥4 measured by a 0–10-cm visual analog scale (VAS), or morning stiffness ≥1 hour in duration. All of the patients had an inadequate response or intolerance to at least 1 nonsteroidal antiinflammatory drug (NSAID) and may have experienced failure of 1 or more disease-modifying antirheumatic drugs.

Study design.

ATLAS was a multicenter, randomized, double-blind, placebo-controlled study with a 24-week double-blind period followed by an 80-week open-label period. The 315 enrolled patients were randomized in a 2:1 ratio to self-administer adalimumab (40 mg every other week) or placebo by subcutaneous injection. The patients were assessed at scheduled clinic visits at weeks 2, 4, 8, 12, 16, 20, and 24 during the double-blind period. An early escape, open-label treatment option of 40 mg of adalimumab every other week was available for patients who did not achieve an ASAS20 response at weeks 12, 16, or 20. Patients were enrolled at 43 centers in the US and Europe following approval from an independent ethics committee and provision of written, informed consent. The percentage of patients who were ASAS20 responders at week 12 was the primary efficacy end point for ATLAS (2). Several PROs were assessed during the study, including the BASDAI (0–100-mm VAS), the Ankylosing Spondylitis Quality of Life questionnaire (ASQoL; 0–18 scale) (13), the Bath Ankylosing Spondylitis Functional Index (BASFI; 0–100-mm VAS) (14), total back pain (0–100-mm VAS), nocturnal back pain (0–100-mm VAS), patient's global assessment of disease activity (0–100-mm VAS), and the Short Form 36 health survey (SF-36). In addition to assessment of all other efficacy and PRO measures, the PASS question, “Considering all the different ways your disease is affecting you, if you would stay in this state for the next months, do you consider that your current state is satisfactory?” was asked at each scheduled clinic visit (except the baseline visit) and a dichotomous answer (yes/no) was recorded; a nearly identical PASS question was subsequently endorsed by the Outcome Measures in Rheumatology Clinical Trials (OMERACT) Conference 8 Special Interest Group (15). The 24-week double-blind period of ATLAS was initiated and completed in 2004.

Statistical analyses.

At each visit, PASS thresholds for each PRO were defined as either the 25th or 75th percentiles for patients who considered their state to be satisfactory depending on whether the PRO had a direct or inverse relationship with disease severity. This approach to defining PASS thresholds for pain, patient's global assessment, and function has been described for patients with knee OA (9) and is based on construction of a curve of the cumulative percentage of patients reporting achievement of the PASS as a function of the outcome of interest. Logistic regression was used to model the observations. The 75th (or 25th) percentile is very close to the point at the flattening of the curve at which most patients reported achievement of the PASS. We calculated 75th percentiles with 95% confidence intervals (95% CIs) for all of the PROs, in which greater scores represent more disability. We calculated 25th percentiles with 95% CIs for the SF-36, in which lesser scores represent more disability. PASS thresholds were estimated at 4, 8, 12, 16, 20, and 24 weeks to assess their stability. Thresholds were recalculated for the data that were stratified according to treatment allocation, age, sex, disease duration, and primarily English-speaking versus non–English-speaking country. Although the PASS question has been translated from English into several languages, it is possible that patient understanding of the concept and response to the question may differ between English-speaking and non–English-speaking countries because of subtle differences in the meaning of the words used in the translation or unspecified cultural factors that influence the concept of health-related well-being (16). These covariates also were examined in multivariate logistic regression analyses using week 12 data to determine their independent contribution to attainment of either the PASS or clinical response as defined by AS-specific composite response measures (ASAS20, ASAS40, ASAS5/6, ASAS partial remission, and BASDAI 50% improvement [BASDAI50]) as the dependent variables. All of the patients (n = 315) were included in the multivariate logistic regression analyses, and all of the variables were entered into the model at the same time. As a complementary analysis, a logistic regression analysis was completed in which age and disease duration were defined as categorical variables (i.e., age <40 years versus age ≥40 years and disease duration <10 years versus disease duration ≥10 years).

RESULTS

PASS thresholds by treatment group.

Of the 315 patients who enrolled, 208 were randomized to receive adalimumab and 107 were randomized to receive placebo. Detailed patient disposition data through week 24 have been reported previously (10). A total of 112 patients (88 in the adalimumab group and 24 in the placebo group) attained the PASS at 12 weeks. PRO scores defining PASS thresholds were consistently worse for patients randomized to placebo, with the exception of minor differences for some of the SF-36 domains (global health, vitality, and mental health) (Table 1). There were substantial treatment group differences in the 75th (or 25th for the SF-36) percentile estimates for PRO change scores from baseline to 12 weeks such that attainment of the PASS was reported in a minority of placebo-treated patients despite there being minimal improvement in PROs from baseline.

Table 1. PASS thresholds for patient-reported outcomes in patients who attained the PASS at 12 weeks following randomization to either adalimumab 40 mg every other week or placebo (2:1 ratio)*
VariablePASS thresholds at 12 weeksChange from baseline to 12 weeks
Adalimumab (n = 88)Placebo (n = 24)Adalimumab (n = 88)Placebo (n = 24)
  • *

    Values are the PASS threshold (95% confidence interval) defined by either the 75th or 25th percentile of the cumulative distribution of scores at 12 weeks and change scores from baseline to 12 weeks. PASS = patient acceptable symptom state; BASDAI = Bath Ankylosing Spondylitis Disease Activity Index; VAS = visual analog scale; BASFI = Bath Ankylosing Spondylitis Functional Index; ASQoL = Ankylosing Spondylitis Quality of Life questionnaire; SF-36 = Short Form 36 health survey; PCS = physical component summary; MCS = mental component summary.

75th percentile    
 BASDAI (0–100-mm VAS)26.7 (21.8, 33.6)44.8 (36.5, 61.1)−20.7 (−30.4, −15.8)−12.1 (−21.8, 18.7)
 BASFI (0–100-mm VAS)31.9 (23.7, 40.2)45.1 (34.6, 70.4)−12.2 (−15.6, −6.5)1.2 (−5.4, 6.5)
 ASQoL (0–18 points)7.0 (5.0, 8.0)10.0 (7.0, 14.0)−2.0 (−3.0, −1.0)−0.5 (−2.0, 2.0)
 Total back pain (0–100-mm VAS)28.5 (24.0, 35.0)53.5 (38.0, 87.0)−23.5 (−33.0, −15.0)2.5 (−8.0, 15.0)
 Nocturnal pain (0–100-mm VAS)28.0 (17.0, 35.0)60.0 (33.0, 88.0)−21.0 (−29.0, −9.0)2.5 (−4.0, 19.0)
 Patient's global assessment  (0–100-mm VAS)29.5 (20.0, 39.0)53.0 (32.0, 81.0)−18.0 (−31.0, −13.0)2.0 (−5.0, 18.0)
25th percentile    
 SF-36 PCS39.9 (34.0, 44.3)33.0 (25.3, 36.5)4.7 (1.8, 7.3)−0.9 (−6.6, 2.0)
 SF-36 MCS47.0 (42.4, 50.8)43.0 (34.0, 53.1)−0.7 (−2.2, 1.4)−2.1 (−6.1, 1.5)
 SF-36 physical function60.0 (50.0, 70.0)45.0 (20.0, 66.7)5.0 (0.0, 10.0)−10.0 (−30.0, 0.0)
 SF-36 bodily pain52.0 (42.0, 62.0)41.0 (31.0, 51.0)19.0 (11.0, 21.0)5.0 (0.0, 19.0)
 SF-36 global health45.0 (37.0, 52.0)42.0 (20.0, 45.0)5.0 (0.0, 10.0)−7.5 (−15.0, 0.0)
 SF-36 vitality45.0 (35.0, 50.0)42.5 (20.0, 50.0)5.0 (0.0, 10.0)−2.5 (−30.0, 10.0)
 SF-36 social function75.0 (62.5, 75.0)62.5 (50.0, 62.5)0.0 (0.0, 0.0)0.0 (−25.0, 0.0)
 SF-36 mental health64.0 (56.0, 72.0)64.0 (44.0, 80.0)0.0 (−4.0, 4.0)−2.0 (−8.0, 8.0)

PASS thresholds stratified by covariates.

Stratification by age, sex, disease duration, and English-speaking versus non–English-speaking geographic location resulted in large variations in PASS thresholds (Table 2). PASS thresholds stratified by age were consistently greater (lesser for the SF-36 domains) for all of the PROs in patients age ≥40 years compared with those age <40 years. This threshold pattern was also evident when patients were further stratified by treatment category (data not shown). The 75th (or 25th for the SF-36) percentile estimates for PRO change scores from baseline to 12 weeks were also consistently less for all of the PROs in patients age ≥40 years; therefore, older patients reported being in PASS despite receiving less benefit from treatment (Table 3).

Table 2. PASS thresholds stratified by age, sex, disease duration, and English-speaking versus non–English-speaking geographic location*
VariableAll patients (n = 112)AgeSexDisease durationSite location
<40 years (n = 58)≥40 years (n = 54)Male (n = 93)Female (n = 19)<10 years (n = 59)≥10 years (n = 53)US/UK (n = 61)Non-US/UK (n = 51)
  • *

    Values are the PASS threshold (95% confidence interval) for patient-reported outcomes in patients who attained the PASS at 12 weeks following randomization to either adalimumab 40 mg every other week or placebo (2:1 ratio). PASS = patient acceptable symptom state; BASDAI = Bath Ankylosing Spondylitis Disease Activity Index; VAS = visual analog scale; BASFI = Bath Ankylosing Spondylitis Functional Index; ASQoL = Ankylosing Spondylitis Quality of Life questionnaire; SF-36 = Short Form 36 health survey; PCS = physical component summary; MCS = mental component summary.

BASDAI (0–100-mm VAS)32.5 (25.8, 41.0)25.1 (19.7, 32.5)41.4 (28.3, 61.1)29.8 (25.5, 41.4)41.0 (18.6, 64.8)27.2 (22.5, 41.4)38.4 (26.4, 52.3)40.3 (28.3, 58.9)26.6 (21.3, 39.4)
BASFI (0–100-mm VAS)37.2 (28.8, 42.4)28.7 (18.9, 37.4)45.2 (37.1, 61.2)34.6 (28.7, 42.3)48.0 (20.1, 61.1)29.2 (18.9, 42.3)40.5 (34.5, 55.3)41.1 (34.6, 58.5)29.9 (23.9, 40.2)
ASQoL (0–18 points)7.5 (6.0, 10.0)5.0 (3.0, 8.0)10.0 (7.0, 12.0)7.0 (5.0, 10.0)10.0 (5.0, 15.0)8.0 (5.0, 10.0)7.0 (4.0, 10.0)6.0 (4.0, 12.0)8.0 (6.0, 10.0)
Total back pain (0–100-mm VAS)33.5 (27.0, 41.0)30.0 (24.0, 36.0)38.0 (29.0, 68.0)33.0 (27.0, 48.0)35.0 (12.0, 58.0)34.0 (24.0, 48.0)33.0 (28.0, 56.0)41.0 (32.0, 63.0)28.0 (17.0, 35.0)
Nocturnal pain (0–100-mm VAS)32.5 (23.0, 41.0)28.0 (14.0, 35.0)39.0 (28.0, 72.0)32.0 (22.0, 41.0)39.0 (12.0, 72.0)35.0 (17.0, 47.0)30.0 (23.0, 58.0)37.0 (26.0, 72.0)29.0 (15.0, 38.0)
Patient's global assessment (0–100-mm VAS)33.0 (27.0, 45.0)30.0 (18.0, 39.0)45.0 (31.0, 71.0)32.0 (28.0, 49.0)39.0 (18.0, 57.0)31.0 (24.0, 54.0)39.0 (27.0, 49.0)44.0 (34.0, 54.0)27.0 (18.0, 31.0)
SF-36 PCS36.1 (33.8, 40.4)42.5 (34.6, 45.6)34.2 (29.1, 36.5)36.8 (33.5, 42.5)35.4 (28.1, 43.0)36.0 (31.4, 42.5)36.8 (32.7, 42.1)36.1 (29.1, 39.7)36.1 (33.7, 44.5)
SF-36 MCS46.3 (42.1, 50.6)47.6 (42.1, 51.3)44.9 (34.1, 51.1)47.6 (41.9, 51.1)42.5 (33.8, 53.2)44.4 (34.1, 50.6)47.6 (41.0, 51.1)48.8 (38.2, 53.0)45.4 (40.2, 49.7)
SF-36 physical function57.5 (50.0, 65.0)70.0 (50.0, 75.0)50.0 (35.0, 60.0)60.0 (45.0, 70.0)50.0 (35.0, 66.7)65.0 (50.0, 70.0)50.0 (40.0, 65.0)50.0 (40.0, 66.7)60.0 (45.0, 70.0)
SF-36 bodily pain51.5 (41.0, 52.0)52.0 (42.0, 62.0)41.0 (32.0, 52.0)52.0 (41.0, 61.0)41.0 (22.0, 62.0)41.0 (41.0, 52.0)52.0 (41.0, 62.0)51.0 (32.0, 61.0)52.0 (41.0, 62.0)
SF-36 global health42.0 (40.0, 47.0)45.0 (40.0, 52.0)42.0 (35.0, 45.0)42.0 (40.0, 47.0)37.0 (35.0, 57.0)45.0 (40.0, 52.0)42.0 (35.0, 47.0)45.0 (37.0, 57.0)42.0 (35.0, 47.0)
SF-36 vitality45.0 (35.0, 50.0)50.0 (45.0, 50.0)40.0 (30.0, 50.0)45.0 (35.0, 50.0)40.0 (25.0, 55.0)45.0 (35.0, 50.0)45.0 (35.0, 50.0)40.0 (25.0, 50.0)45.0 (40.0, 50.0)
SF-36 social function62.5 (62.5, 75.0)75.0 (62.5, 75.0)62.5 (50.0, 62.5)62.5 (62.5, 75.0)62.5 (50.0, 75.0)62.5 (50.0, 75.0)75.0 (50.0, 75.0)62.5 (50.0, 75.0)62.5 (62.5, 75.0)
SF-36 mental health64.0 (56.0, 72.0)64.0 (56.0, 72.0)64.0 (48.0, 72.0)68.0 (56.0, 72.0)52.0 (48.0, 76.0)60.0 (48.0, 68.0)68.0 (55.0, 76.0)68.0 (52.0, 76.0)60.0 (48.0, 68.0)
Table 3. The 75th (or 25th for the SF-36) percentile estimates for change scores from baseline to 12 weeks for PROs in patients who attained the PASS at 12 weeks, stratified by age, sex, disease duration, and English-speaking versus non–English-speaking geographic location*
VariableAll patients (n = 112)AgeSexDisease durationSite location
<40 years (n = 58)≥40 years (n = 54)Male (n = 93)Female (n = 19)<10 years (n = 59)≥10 years (n = 53)US/UK (n = 61)Non-US/UK (n = 51)
  • *

    Values are the percentile estimate (95% confidence interval) for PROs in patients who attained the PASS at 12 weeks following randomization to either adalimumab 40 mg every other week or placebo (2:1 ratio). SF-36 = Short Form 36 health survey; PRO = patient-reported outcome; PASS = patient acceptable symptom state; BASDAI = Bath Ankylosing Spondylitis Disease Activity Index; VAS = visual analog scale; BASFI = Bath Ankylosing Spondylitis Functional Index; ASQoL = Ankylosing Spondylitis Quality of Life questionnaire; PCS = physical component summary; MCS = mental component summary.

BASDAI (0–100-mm VAS)−18.8 (−25.6, −14.4)−21.8 (−31.6, −18.2)−14.7 (−21.9, −3.4)−18.2 (−26.6, −12.7)−21.4 (−40.8, −9.4)−19.9 (−30.7, −12.9)−16.9 (−25.2, −4.5)−16.9 (−22.8, −4.5)−21.8 (−30.4, −12.7)
BASFI (0–100-mm VAS)−9.5 (−14.1, −4.2)−11.1 (−15.1, −4.2)−6.1 (−14.8, −0.2)−9.5 (−14.1, −2.7)−9.4 (−22.1, −3.8)−14.1 (−17.8, −6.5)−5.3 (−11.1, −0.2)−6.1 (−12.1, −0.2)−13.7 (−18.1, −4.8)
ASQoL (0–18 points)−1.2 (−2.0, −1.0)−2.0 (−3.0, 0.0)−1.0 (−2.0, 0.0)−1.4 (−2.0, −1.0)−1.0 (−5.0, 1.0)−1.0 (−2.0, 0.0)−2.0 (−3.0, −1.0)−1.0 (−2.0, 0.0)−2.0 (−3.0, −1.0)
Total back pain (0–100-mm VAS)−17.0 (−25.0, −8.0)−24.0 (−36.0, −8.0)−10.0 (−22.0, 3.0)−20.0 (−26.0, −6.0)−10.0 (−43.0, 0.0)−21.0 (−34.0, −6.0)−15.0 (−23.0, −4.0)−15.0 (−21.0, −4.0)−24.0 (−33.0, −2.0)
Nocturnal pain (0–100-mm VAS)−10.0 (−20.0, −3.0)−20.0 (−30.0, −5.0)−7.0 (−17.0, 4.0)−17.0 (−24.0, 0.0)−7.0 (−29.0, 6.0)−17.0 (−26.0, −3.0)−9.5 (−24.0, 2.0)−5.0 (−11.0, 3.0)−24.0 (−30.0, −7.0)
Patient's global assessment (0–100-mm VAS)−15.0 (−20.0, −5.0)−19.0 (−31.0, −9.0)−8.0 (−18.0, 11.0)−14.5 (−23.0, −3.0)−15.0 (−34.0, −1.0)−15.0 (−24.0, −3.0)−12.0 (−23.0, 0.0)−11.0 (−21.0, 3.0)−16.0 (−31.0, −1.0)
SF-36 PCS3.1 (0.6, 5.0)4.7 (−0.3, 6.7)1.8 (−1.6, 4.2)3.1 (1.2, 5.0)−0.2 (−4.2, 9.9)3.1 (−0.2, 6.6)2.3 (−1.6, 6.2)1.8 (−1.6, 3.9)4.7 (1.3, 7.3)
SF-36 MCS−0.8 (−2.5, 0.6)0.2 (−2.6, 2.6)−2.0 (−5.4, 0.5)−1.0 (−2.8, 0.6)−0.2 (−3.6, 6.5)−2.2 (−3.6, 0.2)1.0 (−2.5, 3.5)−2.0 (−5.4, 0.3)0.4 (−2.6, 3.5)
SF-36 physical function5.0 (0.0, 5.0)5.0 (0.0, 10.0)0.0 (−10.0, 5.0)5.0 (−5.0, 5.0)5.0 (−25.0, 20.0)5.0 (−5.0, 10.0)0.0 (−5.0, 5.0)0.0 (−10.0, 5.0)5.0 (−5.0, 10.0)
SF-36 bodily pain11.5 (10.0, 20.0)19.0 (11.0, 22.0)10.0 (0.0, 19.0)12.0 (10.0, 21.0)11.0 (−9.0, 22.0)11.0 (1.0, 21.0)12.0 (10.0, 21.0)11.0 (0.0, 12.0)21.0 (11.0, 23.0)
SF-36 global health0.0 (0.0, 5.0)5.0 (0.0, 10.0)0.0 (−7.0,0.0 (0.0, 5.0)0.0 (−7.0, 22.0)0.0 (0.0, 8.0)0.0 (−10.0, 5.0)0.0 (−7.0, 0.0)5.0 (0.0, 12.0)
SF-36 vitality5.0 (0.0, 10.0)5.0 (−5.0, 15.0)5.0 (−5.0, 5.0)5.0 (−5.0, 10.0)0.0 (−5.0, 15.0)0.0 (−5.0, 10.0)5.0 (−5.0, 10.0)5.0 (−5.0, 10.0)5.0 (−5.0, 10.0)
SF-36 social function0.0 (0.0, 0.0)0.0 (0.0, 0.0)0.0 (−12.5, 0.0)0.0 (0.0, 0.0)0.0 (−12.5, 12.5)0.0 (−12.5, 0.0)0.0 (0.0, 12.5)0.0 (−12.5, 0.0)0.0 (0.0, 12.5)
SF-36 mental health0.0 (−4.0, 4.0)0.0 (−4.0, 8.0)0.0 (−4.0, 0.0)0.0 (−4.0, 4.0)0.0 (−4.0, 12.0)0.0 (−4.0, 4.0)0.0 (−4.0, 4.0)0.0 (−8.0, 0.0)4.0 (−4.0, 8.0)

PASS thresholds stratified by sex were consistently greater (lesser for the SF-36 domains) for most of the PROs in women (Table 2). Change score 75th/25th percentile estimates for PROs were similar for the BASDAI, BASFI, and ASQoL, but greater beneficial change was noted for men for total and nocturnal back pain, patient's global assessment of disease activity, SF-36 physical component summary (PCS), and SF-36 vitality assessments (Table 3). Therefore, female patients reported being in PASS despite receiving less benefit from treatment.

Differences in PASS thresholds for PROs also were evident when patients were stratified by disease duration. The primary differences were greater PASS thresholds for the BASFI, BASDAI, and patient VAS, and a lesser PASS threshold for the SF-36 PCS in patients with longer disease durations (Table 2). Differences in change score 75th/25th percentile estimates were primarily evident for the BASFI, total back pain, and nocturnal back pain such that for these domains, patients with a longer disease duration reported being in PASS despite receiving less treatment benefit (Table 3).

Substantial differences in PASS thresholds also were evident when patients were stratified according to English-speaking versus non–English-speaking geographic location of investigator site. PASS thresholds for the BASDAI, BASFI, total and nocturnal back pain, and patient's global assessment VAS were greater in patients from English-speaking countries (US and UK investigator sites) compared with patients from non–English-speaking countries (Table 2). The corresponding change score 75th/25th percentile estimates also were lesser for English-speaking patients such that the PASS was reported by patients in English-speaking countries despite receiving less treatment benefit for the corresponding PROs as well as for the SF-36 PCS, bodily pain, and global health subscales (Table 3).

Patient characteristics associated with the PASS, ASAS, and BASDAI50 responses.

Results of the multivariate logistic regression model for the PASS, ASAS response measures, and BASDAI50 are shown in Table 4. Age, male sex, disease duration, and treatment group had independent, statistically significant associations with attainment of the PASS at week 12. In a further logistic regression analysis, age <40 years was independently associated with attainment of the PASS (odds ratio 1.85, P = 0.02), but not a disease duration of <10 years. Site location was not independently associated with attainment of the PASS, but patients were significantly older at sites in the US and UK as compared with non-US/UK sites (data not shown). Results of the logistic regression analysis for the ASAS response measures and the BASDAI50 at week 12 showed that treatment with adalimumab was significantly associated with meeting these response outcomes. In addition, age was independently associated with attaining the ASAS20 (P = 0.011), ASAS5/6 (P = 0.021), ASAS partial remission (P = 0.003), and BASDAI50 (P = 0.006), with younger patients significantly more likely to achieve these response outcomes. However, disease duration and sex were not independently associated with these response outcomes.

Table 4. Patient characteristics independently associated with attainment of the PASS, ASAS response measures, and BASDAI50 at 12 weeks*
VariableAge, yearsMale sexDisease duration, yearsSite locationTreatment
  • *

    PASS = patient acceptable symptom state; ASAS = ASsessment in Ankylosing Spondylitis International Working Group; BASDAI50 = Bath Ankylosing Spondylitis Disease Activity Index 50% improvement; OR = odds ratio; 95% CI = 95% confidence interval; ASAS20 = ASAS criteria for 20% improvement in disease activity.

PASS     
 OR (95% CI)0.97 (0.94, 0.99)1.95 (1.07, 3.55)1.04 (1.01, 1.07)1.21 (0.74, 1.99)2.34 (1.35, 4.03)
 P0.0060.0290.0160.4490.002
ASAS20     
 OR (95% CI)0.97 (0.95, 0.99)1.33 (0.75, 2.34)1.00 (0.97, 1.03)0.92 (0.56, 1.52)5.36 (3.08, 9.35)
 P0.0110.3300.9640.751< 0.001
ASAS40     
 OR (95% CI)0.98 (0.96, 1.01)1.67 (0.91, 3.08)1.00 (0.97, 1.03)0.91 (0.55, 1.53)4.17 (2.25, 7.75)
 P0.1450.0980.8500.732< 0.001
ASAS5/6     
 OR (95% CI)0.97 (0.95, 1.00)1.35 (0.75, 2.46)1.00 (0.97, 1.03)0.74 (0.45, 1.24)6.34 (3.36, 11.98)
 P0.0210.3100.9780.256< 0.001
ASAS partial remission     
 OR (95% CI)0.94 (0.91, 0.98)1.68 (0.72, 3.92)1.03 (0.99, 1.08)1.08 (0.55, 2.10)6.27 (2.16, 18.19)
 P0.0030.2270.1400.827< 0.001
BASDAI50     
 OR (95% CI)0.97 (0.94, 0.99)1.36 (0.76, 2.44)1.00 (0.97, 1.03)1.01 (0.61, 1.67)4.29 (2.36, 7.80)
 P0.0060.3080.9320.972< 0.001

Stability of PASS thresholds for PROs over 24 weeks of treatment with adalimumab.

There was a consistent decline in PASS thresholds for most of the PROs over 24 weeks of treatment with adalimumab (Figure 1). Such a decline was not evident for the SF-36 assessments at the 12- and 24-week time points during which PASS thresholds were stable, with the exception of the threshold for the bodily pain subscale, which decreased from week 12 to 24 (data not shown).

Figure 1.

Patient acceptable symptom state (PASS) thresholds defined by the 75th percentile of the cumulative distribution of scores at several time points from 4 to 24 weeks for patient-reported outcomes in adalimumab-treated patients who attained the PASS. BASDAI = Bath Ankylosing Spondylitis Disease Activity Index; BASFI = Bath Ankylosing Spondylitis Functional Index.

DISCUSSION

Our data describe several novel observations relevant to an understanding of the PASS concept. First, attainment of the PASS depended not only on treatment but also on certain demographic characteristics, specifically, younger age and male sex. The effect of age, but not sex, was also observed on most of the ASAS response measures (with the exception of the ASAS40) and the BASDAI50 response. Second, treatment, age, and male sex also affected the thresholds for pain, function, and QOL at which patients considered themselves to be in PASS. In order to report being in PASS, adalimumab-treated patients, younger patients (age <40 years), and/or male patients required lesser scores for pain and function and greater scores for QOL as compared with placebo-treated, older, and/or female patients. Regression analysis showed that these demographic variables acted independently. Third, thresholds for pain, function, and global well-being associated with being in PASS declined over a time period of 24 weeks from the start of treatment.

Overall, PASS thresholds for PROs were similar to those reported in AS, OA, and other settings (6, 7, 9, 17). The increased likelihood of attaining PASS and other response outcomes at a younger age is intuitive and likely reflects an early and reversible phase of disease. Older patients are less likely to achieve satisfactory health status because the disease becomes increasingly more irreversible with limitation in spinal mobility and impairment of function. The effect of disease duration was somewhat unclear in this analysis. Its inclusion as a continuous variable in the regression analysis indicated that longer duration was independently associated with an increased likelihood of attaining PASS, whereas inclusion as a dichotomous variable with a prespecified 10-year cutoff indicated no independent association with the PASS when age was included in the model. In addition, disease duration was not independently associated with the ASAS response outcomes. In this trial, disease duration was estimated from the time of diagnosis; further research focusing on symptom duration is needed. The greater PASS thresholds for PROs with increasing age and disease duration are also consistent with the increasing irreversibility of disease. In addition, qualitative research has shown an increasing age-related adaptation to chronic pain focused around increasing personal and societal acceptance, resourcefulness in seeking distraction, pacing, and viewing others as worse off (18).

The increased likelihood of attaining PASS in men has not been reported, and it is unclear why sex should impact attainment of the PASS but not the other response outcomes. However, it has been shown that ASAS measures contribute only 37% of the variance in patient reporting of the PASS (5). Previous research indicates that men typically tolerate more pain in experimental settings than women and highlights the influence of social norms on behavior (19). Greater scores are also generally reported for the BASDAI in women compared with men (20, 21). Moreover, pain reporting is influenced by the sex of the experimenter, with men reporting lower scores for pain in the presence of a female experimenter (22). This may be relevant to clinical trials in which clinical coordinators are typically female. The greater PASS threshold in women has been reported previously in AS for the BASDAI and may reflect a greater diversity of coping mechanisms and adaptation that has been reported in women (23).

A previous study of hip and knee OA did not show any impact of age, sex, or disease duration on PASS thresholds for PROs (9). However, the PASS question posed in the OA study was focused on the experience of pain and impairment of function, and the PASS was shown to depend on the initial values of pain, patient's global assessment, and in particular, function score. The PASS question used in our report is almost identical to the PASS question formulated and endorsed by patients and researchers at the OMERACT 8 Conference. It reflects a broader concept aimed at measurement of the impact of the disease on overall satisfaction with health status. Inherent to this approach is the potential for greater divergence in the interpretation of the meaning and the response required of the question between patients and researchers as a consequence of extraneous factors modulating health status unrelated to the underlying disease (e.g., comorbid illness as well as the patient's adaptation to the disease [e.g., self-efficacy]). These factors, in turn, may vary according to the covariates examined in this report, such as age, sex, and country of origin. Beyond differences in the construct of the question, the discrepancy with the prior study assessing the PASS in OA of the hip and knee also raises the potential concern that PASS estimates may vary according to the disease and the specific musculoskeletal problems being studied. Adaptation to spinal disease, generally regarded as lacking effective treatment, may differ from adaptation to hip and knee OA, for which a definitive resolution in the form of arthroplasty is available.

The decline in PASS thresholds for PROs over the 24 weeks of the trial, especially in the first 12 weeks, is contrary to a previous report from a randomized controlled trial in which patients with AS received NSAID therapy for 12 weeks and a different PASS question that focused on pain and function was used (7). Patients recruited into the adalimumab trial had already failed NSAID therapy and were therefore more treatment refractory, which may have affected adaptation to disease and treatment expectations. Having failed several established therapies, patients may have had lowered their expectations for treatment efficacy at the start of the trial. Once patients were started on the treatment, however, those patients who received adalimumab experienced improvement, and treatment expectations may have changed as patients realized that perhaps there is an additional treatment approach that is effective for this disease. In contrast, patients receiving placebo experienced no such improvement and continued to harbor lower expectations for treatment efficacy, accounting for the divergence in PASS thresholds for PROs that are therefore greater at 12 weeks compared with adalimumab-treated patients. One would therefore expect that this divergence in PASS thresholds would be far more evident with a highly effective treatment. This might also account for the relative stability of PASS thresholds after introducing a modestly efficacious treatment such as NSAIDs as compared with the decline in PASS thresholds shortly after introducing a highly effective therapy such as an anti–tumor necrosis factor antagonist.

These observations highlight the plasticity of the concept of an acceptable symptom state. The term response shift has been applied to the observation that internal standards, values, and the conceptualization of QOL may change over the course of the disease and that these changes are inherent to the process of accommodating the illness (24–26). For example, it has been shown that patients undergoing radiotherapy and experiencing diminishing levels of fatigue retrospectively amplified their pretreatment level of fatigue, whereas patients whose fatigue had increased retrospectively minimized their initial level of fatigue (27). Response shift may attenuate estimates of treatment effects if patients adapt to treatment toxicities or disease progression over time. Methods that assess response shift have therefore been developed in an attempt to reveal unbiased treatment effects. In the retrospective pretest approach, the test is administered after treatment in which patients are invited to provide a renewed judgment of their baseline level of QOL. The comparison of the baseline and retrospective measures is hypothesized to provide an indication of the amount and direction of response shift effects (28). A large number of studies have used this approach to estimate response shift for various QOL domains, and a recent meta-analysis aimed at estimating the magnitude and clinical significance of these effects showed that the largest effect sizes were observed for fatigue, followed by global status, physical role limitation, psychological well-being, and pain (29).

The subgroup differences in self-reported outcomes, either PROs or the PASS, may also reflect item bias in these PRO scales known as differential item functioning (DIF). Psychometric properties of PROs may vary across cultures, countries, and demographic categories, owing to differences in the meaning of scale items and the fact that differences in PRO scores may be reported that are unrelated to the quantitative value of the latent trait. The importance of culture in the expression of depression as primarily somatic symptoms unexplained by comorbidities has been previously documented as one example of DIF (30). DIF can be tested between different groups using item-response theory (IRT) models, which estimate the latent trait level (severity level) of an individual responding to a test item, the severity of the test item (or symptom), and the accuracy to which the item measures the latent trait (31). Recent formulations of IRT models also allow for accurate assessment of dimensional constructs, which is of particular relevance to health-related QOL and function instruments that are fundamentally multidimensional. Although most of the PROs assessed in our study have been translated and validated in many countries, we are unaware of studies that have assessed DIF across countries and demographic subgroups using IRT. Consequently, even though our analysis did not support patient country as an independent variable in attaining the PASS, we cannot exclude DIF as a potential factor accounting for subgroup differences in either PROs or the PASS. Our data further support the desirability of IRT studies of these PROs.

In conclusion, our results demonstrate that PASS thresholds for PROs in this placebo-controlled trial of adalimumab for AS vary not only according to treatment (placebo versus adalimumab) but also according to demographic variables such as age, sex, and disease duration, and that PASS thresholds can progressively decrease over the 24 weeks in the placebo-controlled phase of the study. Moreover, both younger age and male sex are independently associated with attainment of the PASS. These observations suggest that further study is required prior to the use of the PASS approach as currently framed to assess absolute health status in clinical research.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Maksymowych had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Maksymowych, Gooch, Dougados, Wong, Chen, Kupper, van der Heijde.

Acquisition of data. Maksymowych, Gooch, Wong, Chen.

Analysis and interpretation of data. Maksymowych, Gooch, Dougados, Chen, van der Heijde.

ROLE OF THE STUDY SPONSOR

The decision to evaluate the patient acceptable symptom state in an ankylosing spondylitis population was made jointly by the Adalimumab Trial Evaluating Long-term Efficacy and Safety for Ankylosing Spondylitis advisory committee, which was composed of individuals from academic institutions and Abbott Laboratories. Abbott Laboratories provided the study drug and was responsible for collecting the clinical data; data analyses were planned and reviewed by the advisory committee. All authors contributed to manuscript development and review, agreed to submit the manuscript, and approved the content of the submitted manuscript.

Acknowledgements

The authors are grateful to all ATLAS investigators, study coordinators, and patients who participated in this study. Arbor Communications, Ann Arbor, Michigan, and Michael A. Nissen, ELS, of Abbott Laboratories, edited the manuscript on behalf of the authors.

Ancillary