Dr. Maksymowych is a Senior Scholar of the Alberta Heritage Foundation for Medical Research.
Evaluation and validation of the patient acceptable symptom state (PASS) in patients with ankylosing spondylitis
Article first published online: 31 JAN 2007
Copyright © 2007 by the American College of Rheumatology
Arthritis Care & Research
Volume 57, Issue 1, pages 133–139, 15 February 2007
How to Cite
Maksymowych, W. P., Richardson, R., Mallon, C., van der Heijde, D. and Boonen, A. (2007), Evaluation and validation of the patient acceptable symptom state (PASS) in patients with ankylosing spondylitis. Arthritis & Rheumatism, 57: 133–139. doi: 10.1002/art.22469
- Issue published online: 31 JAN 2007
- Article first published online: 31 JAN 2007
- Manuscript Accepted: 24 APR 2006
- Manuscript Received: 3 FEB 2006
- Ankylosing spondylitis;
- Health status;
- Patient acceptable symptom state
The Patient Acceptable Symptom State (PASS) constitutes an absolute level of patient well-being and represents an ambitious target for disease management. We explored contributors to PASS, validated the PASS concept, and assessed thresholds of self-reported outcomes below which patients considered themselves in PASS.
Patients with ankylosing spondylitis completed a questionnaire that included self-reported assessments of pain, fatigue, disease activity, function, patient global, quality of life (QOL), and whether they considered their current disease state satisfactory or not. Stepwise logistic regression was used to assess contributors to PASS. PASS was validated by analyzing proportions of patients reporting need for a rheumatologist and who were in current flare. PASS thresholds for self-reported outcomes were estimated using an anchoring method based on the patient's opinion and targeting the 75th percentile of the cumulative distribution.
PASS data were available for 291 patients, of whom 169 (58%) were in PASS. Significant contributors were age (Exp[B] 1.05; P = 0.003), patient global disease activity (Exp[B] 0.79; P = 0.008), and function (Bath Ankylosing Spondylitis Functional Index [BASFI]; Exp[B] 0.72; P < 0.001). PASS reflected need to consult the rheumatologist and current flare (71% and 73% correctly classified, respectively) and significantly contributed to QOL (B = −5.99; 95% confidence interval −7.16, −4.08). PASS thresholds were 5.0 for patient global disease activity, 5.0 for total back pain, 22.8 for fatigue, 4.8 for disease activity (Bath Ankylosing Spondylitis Disease Activity Index), and 4.0 for function (BASFI).
A majority of patients (58%) reported being in PASS. PASS thresholds for pain and function were unexpectedly high, possibly suggesting adaptation to the consequences of the disease.
There is increasing interest in reporting health status in concepts that are relevant to the individual patient and readily understood by physicians. This is of particular importance when considering a change in treatment management and when interpretation of response to therapeutic interventions is compared between treatment groups in clinical trials. For example, presentation of study findings is often more meaningful to clinicians when the proportion of patients achieving a minimal clinically important improvement (MCII) is reported (1–3). In addition to clinical relevant improvement, 2 complementary concepts have been proposed in rheumatology to define absolute health status, namely, the Low Disease Activity State (LDAS; or Minimal Disease Activity) and the Patient Acceptable Symptom State (PASS) (4, 5). The concept of the LDAS is primarily anchored to physician-based measures of disease activity. The concept of the PASS primarily reflects the overall health state at which patients consider themselves well. As an example, one might consider the Health Assessment Questionnaire (HAQ) in rheumatoid arthritis. The MCII has been assessed to be 0.22 (6). A patient who experienced an improvement in HAQ from 1.85 to 1.55 (>0.22) might consider this a relevant improvement but might still experience significant functional limitation with a HAQ score of 1.55. It is therefore desirable to describe change in health status according to both the MCII as well as the PASS. The objective of treatment would therefore be to achieve both an MCII and a PASS condition. For evaluation of clinical trials in ankylosing spondylitis (AS), the ASsessment in Ankylosing Spondylitis (ASAS) international working group developed the concept of an ASAS20 response, as defined by improvement of at least 20% and 1 unit in at least 3 of 4 domains (patient global, function, pain, stiffness), as well as the ASAS partial remission criteria, as defined by a score <2 on all 4 domains. These data were derived by analyzing which composite measures best discriminated between nonsteroidal antiinflammatory drugs (NSAIDs) and placebo in controlled clinical trials (ASAS20 response) and by expert opinion (ASAS partial remission).
Although a relevant concept, the approach to the estimation and validation of the PASS and its application constitute an important subject of discussion and research. In rheumatology, there has been only 1 publication that has addressed this issue, and that report focused on defining PASS estimates for patient self-reported outcomes in osteoarthritis following a 4-week intervention with NSAIDs but did not validate these estimates (5). A recent report has also focused on methodologic issues with respect to determination of thresholds for the PASS (7). As yet, there have been no published reports describing PASS estimates for the main outcomes in patients with AS other than in abstract form (8). In the present cross-sectional study, we asked patients directly whether they considered themselves to be in an acceptable symptom state. Next, we validated the concept of PASS according to reported flare status, expressed need to see the physician, attainment of remission, and PASS contribution to quality of life (QOL). We further analyzed the levels of several patient-reported outcomes below which patients considered themselves in PASS.
PATIENTS AND METHODS
We conducted a cross-sectional postal survey of patients with AS in the province of Alberta, Canada. All the academic- and community-based rheumatology centers in northern Alberta participated in the study. The names and addresses of all 795 patients with AS who had been in contact at least once with the rheumatology sites in northern Alberta for the past 15 years and fulfilled the modified New York criteria for AS (9) were selected. These included patients who visited both community- and academic-based rheumatologists. Questionnaires were mailed in April and May 2003. Questionnaires were fully anonymous and 1 reminder was sent to all patients ∼4 weeks after the initial mailing to encourage patients to mail the completed questionnaire in a prepaid envelope to the co-coordinating University of Alberta rheumatology center. Data were entered at the co-coordinating center on an ongoing basis and the database was locked 12 weeks after the mailing. All patients signed a consent form and the study received ethical approval from the University of Alberta Bioethics Committee.
We collected data on patient demographics, education, and disease-related and non–disease-related comorbidity, as well as validated patient self-reported outcomes describing disease severity. The latter included the ASAS patient-reported outcome measures such as total spinal pain (numerical rating scale [NRS], range 0–10), nocturnal spinal pain (NRS, range 0–10), spinal stiffness (Bath Ankylosing Spondylitis Disease Activity Index [BASDAI] question 6 by NRS, range 0–10), patient global disease activity (NRS, range 0–10), fatigue (Functional Assessment of Chronic Illness Therapy [FACIT], range 0–52; available at www.facit.org), BASDAI question 1 (NRS, range 0–10), and functioning (Bath Ankylosing Spondylitis Functional Index [BASFI] by NRS, range 0–10) (10). In addition, the entire BASDAI (NRS, range 0–10) (11); the Bath Ankylosing Spondylitis Patient Global Score (NRS, range 0–10) (12); and the AS Quality of Life (ASQoL) index (13), a disease-specific QOL scale (yes/no answers, range 0–18, higher values indicate worse QOL), were completed. Patients' opinions of their symptom state were recorded by a “yes” as opposed to a “no” answer to the question, “Considering all the different ways your disease is affecting you, if you were to stay in this state for the next few months, do you consider that your current state is satisfactory?” We also estimated disease status by asking the patient to answer yes or no to the following 2 questions: “Are you currently experiencing a flare of your AS?” and “Is your AS sufficiently active to require an assessment and examination from your rheumatologist?” Also, disease remission was assessed according to the ASAS definition of partial remission (patient global disease activity, total spinal pain, BASFI, and mean score for items 5 and 6 of the BASDAI, all <2 on a 0–10 scale) (14).
Descriptive statistics were used to describe the study sample. We compared patients in a PASS condition versus those who reported not being in a PASS condition for demographic and disease severity characteristics by descriptive statistics (mean ± SD, median, interquartile range) and using 2-tailed t-test. To assess the construct of the PASS concept, the contribution of demographic variables (age, sex, disease duration, educational status) and disease activity variables (ASAS core measures: total back pain, fatigue [item 1 of the BASDAI], stiffness [item 6 of the BASDAI], patient global disease activity, BASFI) to attainment of a PASS condition was analyzed by stepwise conditional logistic regression. In the first block we included age, sex, disease duration, and educational level (dichotomized by completion of high school), and in the second block we added the AS core measures. We further examined the construct validity of the PASS using 3 approaches. First, we analyzed the proportions of patients correctly classified as being in a flare and/or requiring consultation with their rheumatologist according to whether they were PASS positive or PASS negative. Second, we compared proportions of PASS positive and PASS negative patients who met the ASAS criteria for remission. Third, we analyzed the contribution of being PASS positive to QOL (ASQoL) in a linear regression analysis controlled for QOL. The threshold at which patients considered themselves in PASS for each of the patient self-reported ASAS outcome measures and for the BASDAI and patient global well-being was estimated using 2 approaches. In the first, we constructed a curve of cumulative percentages of patients who considered their state satisfactory as a function of the score of interest. The PASS was defined as the 75th percentile of the cumulative distribution for each outcome for patients who rated their condition as PASS positive (5). We also calculated the percentages of patients who rated their condition as PASS negative but whose outcome scores actually fell below the 75th percentile threshold and therefore would be false positively misclassified as being PASS positive according to this definition. In the second approach, we estimated PASS thresholds by plotting receiver operating characteristic (ROC) curves and identifying cutoffs that yielded the smallest number of false positives and false negatives.
Description of PASS status.
Within 12 weeks, 302 (38.0%) completed questionnaires of 795 mailed questionnaires were received, 11 of which were excluded from the study sample because the PASS question had not been completed. There were no significant differences between questionnaire participants and nonparticipants for age, sex, and disease duration (data not shown). Of the patients who completed the PASS question, 169 (58.1%) indicated that they considered themselves in PASS. The characteristics of the entire sample and of the PASS positive and PASS negative patients separately are shown in Table 1. Patients in PASS reported less total and nocturnal back pain, less spinal stiffness, lower BASDAI scores, better function (BASFI), less fatigue (FACIT), and better QOL (ASQoL; P < 0.0001 for all comparisons with PASS negative patients). Patients in PASS were also significantly older (P < 0.0001) and had a somewhat longer disease duration (P = 0.02).
|Parameter||PASS + (n = 169)||PASS − (n = 122)||All patients (n = 291)|
|Male sex, no. (%)||121 (71.6)||97 (79.5)||218 (74.9)|
|Age, mean ± SD years||48.0 ± 11.8||42.5 ± 12.1||45.7 ± 12.2|
|Disease duration, mean ± SD years||22.9 ± 11.5||19.6 ± 12.1||21.5 ± 11.9|
|Comorbidity, no. (%)||61 (36.1)||36 (29.5)||97 (33.3)|
|Completed high school, no. (%)||146 (89)||101 (84.9)||239 (84.5)|
|ASAS core measures|
|Nocturnal pain (0–10 scale)||2.9 ± 2.3 (1–4)||4.7 ± 2.6 (2–7)||3.6 ± 2.5 (2–6)|
|Total back pain (0–10 scale)||3.3 ± 2.5 (1.5–5)||5.3 ± 2.3 (4–7)||4.0 ± 2.5 (2.0–6.0)|
|Patient global (0–10 scale)||3.1 ± 2.2 (1–5)||5.5 ± 2.4 (4–7)||4.0 ± 2.6 (2–6)|
|Morning stiffness (0–10 scale)†||3.2 ± 2.5 (1–5)||5.6 ± 2.8 (3–8)||4.2 ± 2.9 (2–6)|
|Fatigue (0–10 scale)‡||4.2 ± 2.6 (2–6)||6.5 ± 2.4 (5–8)||5.1 ± 2.7 (3–7)|
|BASFI (0–10 scale)||2.7 ± 2.3 (0.9–4.0)||5.1 ± 2.6 (2.9–7.0)||3.7 ± 2.7 (1.2–5.3)|
|BASDAI (0–10 scale)||3.4 ± 2.1 (1.6–4.5)||5.9 ± 1.9 (4.4–6.8)||4.4 ± 2.3 (2.3–5.6)|
|BAS-G (previous week; 0–10 scale)||2.6 ± 2.1 (1–4)||5.9 ± 2.4 (4–7)||4.2 ± 2.7 (2–6)|
|FACIT (0–52 scale)||14.5 ± 10.1 (6–21)||26.5 ± 11.4 (18–32)||19.6 ± 12.2 (8–27)|
|ASQoL (0–18 scale)||4.4 ± 4.6 (0.5–7)||10.4 ± 4.8 (6–13)||6.9 ± 5.6 (1–10)|
Univariate analysis demonstrated that all patient-reported outcome measures were associated with being in a PASS condition after adjusting for age, sex, disease duration, and educational level. In stepwise logistic regression analyses, being in PASS was significantly and independently associated with increasing age (Exp[B] 1.05; 95% confidence interval [95% CI] 1.02, 1.09; P = 0.003), lower patient global disease activity (Exp[B] 0.79; 95% CI 0.66, 0.94; P = 0.008), and better function (BASFI; Exp[B] 0.72; 95%CI 0.60, 0.86; P < 0.001). The contribution of all variables included in the model to variance in PASS status was 37%.
Thresholds of patients' reported outcomes for being in PASS.
The PASS estimates for the patient-reported outcomes as defined by the 75th percentile of the cumulative distribution are listed in Table 2. Patients who considered their state satisfactory rated their BASDAI score <4.8 on the 0–10 NRS scale (Figure 1). This cutoff led to a false positive misclassification of 30% among patients who considered their state as unsatisfactory. Similarly, patients who considered their state satisfactory rated their back pain score <5, which was close to the threshold for the BASDAI. Patients who considered themselves to be in a satisfactory state rated their BASFI score <4 on the 0–10 NRS scale. However, the distribution of BASFI scores at which patients considered themselves in an unsatisfactory state was large, with some patients considering their state as unsatisfactory even with a BASFI score of 0, indicating that a 75th percentile threshold results in more patients being false positively misclassified as being PASS positive (40%) compared with the 75th percentile threshold for the BASDAI (Table 2 and Figure 2). False positive misclassification when applying the 75th percentile threshold ranged from 18% for fatigue (item 1 of the BASDAI) to 57% for stiffness (item 6 of the BASDAI). PASS cutoffs determined by ROC analysis demonstrated somewhat lower values, although the cutoff for the BASDAI was similar at 4.7 (Table 2).
|Parameter||ROC cutoff||Sensitivity||Specificity||PASS + 75th percentile threshold||False positive, %|
|ASAS core outcomes|
|Patient global (NRS)||4.5||69.0||71.2||5.0||42|
|Spinal stiffness (NRS)†||3.5||73.7||63.2||5.0||57|
|Total back pain (NRS)||3.5||78.5||60.4||5.0||46|
|Nocturnal back pain (NRS)||3.5||74.8||62.4||5.0||44|
Examination of the construct validity of the PASS.
Of the 257 patients who answered the question as to whether they needed to consult their rheumatologist, 80 (31.1%) reported a need to consult their rheumatologist (Table 3). Only 23.9% of PASS positive patients reported a need for consultation as compared with 63.7% of PASS negative patients (P < 0.001). Of the 271 patients who answered the question regarding flare status, 133 (49.1%) reported being in a current flare. Only 29.6% of PASS positive patients reported being in a current flare as compared with 76.8% of PASS negative patients (P < 0.001). Consequently, PASS status reflected both patients' need to consult their rheumatologist and the presence of current disease flare (71% and 73% correctly classified, respectively). Only 36 (12%) of all patients met the ASAS criteria for remission. Of the 165 patients who reported being in PASS condition, 34 (20.6%) also met the ASAS criteria for remission as compared with only 2 (1.8%) of 114 patients that were PASS negative (P < 0.001). After adjusting for age and sex, PASS was significantly contributory to QOL (ASQoL; B = −5.99; 95% CI −7.16, −4.08).
|PASS +||PASS −|
|Flare||47/159 (29.6)||86/112 (76.8)|
|Need doctor||15/155 (23.9)||65/102 (63.7)|
|ASAS partial remission||34/165 (20.6)||2/114 (1.8)|
In this study, we validated the relevance of the PASS concept for patients with AS using several independent approaches. In addition, we assessed the thresholds at which patients considered themselves to be in PASS for the core measures of patient-reported outcomes (ASAS) (15), comprising morning stiffness, the BASFI, total and nocturnal spinal pain, global disease activity, and fatigue, and for patient global well-being, the ASQoL and BASDAI. Results demonstrate that in a cross-sectional population, the majority of patients (58%) consider themselves to be in an acceptable state for the next few months.
It should be realized that the concept of the PASS is quite new in rheumatology and that the specific wording and the construct validity of the question deserve attention. For example, our question formulated a time frame to evaluate the present health state for the next several months. A time frame that is too short might not be relevant and might allow the inordinate intrusion of short-term stressful events unrelated to disease as confounding factors in patients' perceptions of their current health status. It is likely that a time frame of several years or even the remainder of a patient's life might constitute a more rigorous hurdle for acceptable disease status. A drawback to this approach, however, is that this could be significantly influenced by a patient's age.
In this study, PASS positive patients reported significantly fewer disease flares and less need to consult their rheumatologist. Moreover, PASS status contributed significantly to QOL. Function (BASFI) and global disease activity were the major contributors to PASS, and for clinicians treating patients with AS, this finding is not surprising. Although educational status has been identified to influence overall health status (16, 17), we were unable to show any impact on PASS status. This may be because most patients had achieved a relatively high level of education. Previous work has indicated that because patients with AS develop symptoms early in life, they tend to strive for less physically demanding occupations that require higher levels of education (18). Overall, the present data support the view that the PASS represents a clinically relevant concept.
In contrast, the agreement with external concepts such as experience of flare, need for a doctor, and remission was only moderate. However, a visit to the rheumatologist may be necessary for reasons unrelated to AS or to obtain a repeat prescription for medication. The relatively low percentage of patients in remission in the PASS positive group is not surprising because patients might consider themselves in PASS even though they do not reach the level of remission. It is very reassuring that only 2 patients stated that they were not in PASS while they were in remission, indicating the high specificity of the PASS concept.
The majority of patients reported being in a PASS condition. This finding could reflect some degree of selection bias according to demographic variables, such as education, that could affect a patient's willingness to respond to the questionnaire. However, multivariate analysis did not identify an independent effect of obvious demographic factors. Disease activity status seems an unlikely source of selection bias because one would anticipate that patients with more active disease would be more likely to respond to the questionnaire, thereby leading to a lower proportion of patients in PASS. The total explained variance by all included measures was only 37%. It is not surprising that PASS was independently associated with patient global assessment of disease activity and function. It is likely that other factors such as psychological and environmental factors unrelated to disease influence a patient's perception of acceptability of symptom state. The association with age could, for example, reflect several interrelated factors such as adaptation to the functional consequences of the disease, modification in work environment, increased appreciation that this is a slowly progressive disease with diminished concern over its consequences, improved personal economic circumstances, and improved knowledge with increased self efficacy. In future studies, the role of coping, self efficacy, anxiety and depression, and treatment expectations should be examined. Coping and treatment expectations are also factors that might account for the rather high percentage of patients who considered themselves in PASS and have been identified in prior studies of unrelated disorders to significantly impact perceived health status (19, 20). Furthermore, prior studies have shown that patients with AS demonstrate superior coping skills (21).
Adaptation might also explain high PASS thresholds for fatigue, total and nocturnal pain, and global disease activity. Apart from adaptation, patient expectations might explain the unexpected high thresholds. Recently, anti–tumor necrosis factor (anti-TNF) therapy has demonstrated large beneficial effects on the clinical symptoms of patients with AS, while previous treatment options for the disease were limited to NSAIDs and exercise therapy. It is likely that perceptions of the larger beneficial effects of treatment will influence patients' concept of PASS. In most Canadian provinces, anti-TNF therapy is not accessible through the public formularies. Consequently, few patients have become aware of or experienced this treatment and its major impact on symptoms. It is therefore possible that PASS estimates may differ in patients who have experienced the benefits of anti-TNF therapies.
The percentage of patients misclassified as being in PASS according to the 75th percentile threshold despite scores below the thresholds varied from 18% for fatigue to 57% for stiffness. The effect of specificity when the threshold was set at a sensitivity of 75% was reflected in the somewhat lower cutoff values defined by ROC analysis where cutoffs are based on optimal values for sensitivity and specificity. The relatively low sensitivity and specificity of cutoffs for these self-reported outcomes likely reflect the importance of both disease-related and unrelated psychosocial confounders as discussed above (22).
Three previous studies have assessed PASS in patients with a rheumatic disease. The first study modeled thresholds for the main patient self-reported outcomes in patients with hip or knee osteoarthritis after a 4-week period of treatment with NSAIDs (5). In this study, the choice of the 75th percentile of the cumulative distribution for the cutoff in the outcome of interest was based on both expert opinion and statistical modeling using logistic regression to target the point at the flattening of the curve at which most patients considered themselves in a satisfactory condition. The data from patients who considered their state as unsatisfactory were similarly modeled and the thresholds from both groups correctly classified patients according to PASS status. In contrast to our study, no significant impact of age was noted. Thresholds for pain scores and patient global scores were also substantially less in the osteoarthritis study than those noted in our study. This might reflect factors such as involvement of weight-bearing joints and differences in coping associated with much shorter disease duration (4.8 years for patients with knee osteoarthritis and 3.4 years for those with hip osteoarthritis, respectively). A significant difference in this study design from our study was that patients in the osteoarthritis study were asked the PASS question after completion of a 4-week period of treatment with an NSAID, whereas our study had a cross-sectional design. Selection of more symptomatic patients and experience of improved health status with the introduction of a new therapy likely influenced the PASS evaluation.
The second study was a cross-sectional postal survey of patients with spondylarthritis that estimated the BASDAI cutoff for acceptable pain and morning stiffness by plotting the ROC (23). Approximately half (55.3%) of patients reported inadequate symptom control, with a BASDAI score of 3.9 having the best combination of sensitivity (74.6%) and specificity (72.4%) for discriminating between patients in the 2 groups. The question posed to patients focused specifically on effectiveness of the current treatment in relieving pain and morning stiffness and therefore differed substantially from the question posed in our study, emphasizing the importance of standardization of wording if this concept is to be developed further.
The third study has been reported in abstract form and posed a question similar to that posed in our study but assessed patients randomized to 2 doses of celecoxib and diclofenac (8). PASS estimates for pain (3.4), patient global (3.6), BASDAI (3.5), and BASFI (3.1) were lower than estimates in our study, although selection of more symptomatic patients and expectation of improved health status may have influenced PASS estimates.
The thresholds defined by PASS could also be used to select patients for interventions. The threshold for BASDAI in our cross-sectional cohort was 4.8. However, the most recent clinical trials of treatments for patients with AS have used a BASDAI cutoff of 4 as a key inclusion criterion for entry into the study (24, 25). In addition, the ASAS working group has also recommended that a BASDAI cutoff of 4 be used as the primary inclusion criterion for institution of anti-TNF therapy after standard therapies have failed (26). It is important to note that these recommendations are in fact arbitrary and that the relationship between the BASDAI score and objective parameters of disease activity has not yet been clearly defined.
In conclusion, our evaluation of a cross-sectional cohort of patients from rheumatology practices serving northern Alberta has shown that the majority of patients with AS consider themselves in a satisfactory health state. The concept and wording of the question asking about satisfactory health state was validated by concomitant assessment of patients' desire to consult their rheumatologist, self-reported flare status, and being in remission. Estimates of PASS according to thresholds for the individual AS self-reported measures will require further study with different questions and in different data sets, countries, clinical settings, methods of analysis, and languages.
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 design. Dr. Maksymowych.
Acquisition of data. Dr. Maksymowych, Ms Richardson, and Ms Mallon.
Analysis and interpretation of data. Drs. Maksymowych, van der Heijde, and Boonen.
Manuscript preparation. Drs. Maksymowych, van der Heijde, and Boonen.
Statistical analysis. Drs. Maksymowych and Boonen.
- 8Determination of the patient acceptable symptom state in outcome criteria in ankylosing spondylitis [abstract]. Ann Rheum Dis 2005; 64 Suppl 3: 97., , , , , , et al.
- 16Predictors of self-rated health status among Texas residents. Prev Chronic Dis 2005; 2: A12., , .
- 25Ankylosing Spondylitis Study for the Evaluation of Recombinant Infliximab Therapy Study Group. Efficacy and safety of infliximab in patients with ankylosing spondylitis: results of a randomized placebo-controlled trial (ASSERT). Arthritis Rheum 2005; 52: 582–91., , , , , , et al, and the