Abstract
Objective. Raynaud phenomenon (RP) in systemic sclerosis (SSc) could be influenced by clinical phenotype, environmental factors (e.g., season), and personal factors (e.g., coping strategies and ill-health perceptions). We studied the relative influence of a range of putative factors affecting patient-reported assessment of SSc-RP severity.
Methods. SSc patients were enrolled at UK and US sites. Participants completed the 2-week Raynaud Condition Score (RCS) diary alongside collection of patient demographics, clinical phenotype, the Coping Strategies Questionnaire, Pain Catastrophizing Scale, Scleroderma Health Assessment Questionnaire (SHAQ), and both patient/physician visual analog scale (VAS) assessments for RP, digital ulcer disease, and global disease. Environmental temperature data were obtained at each site. A second RCS diary was completed 6 months after enrollment.
Results. We enrolled 107 patients (baseline questionnaires returned by 94). There were significant associations between RCS diary variables and both catastrophizing and coping strategies. There were significant associations between RCS diary outcomes and both environmental temperature and season of enrollment. Age, disease duration, sex, disease subtype, smoking, and vasodilator use were not associated with RCS diary outcomes. The best-fitting multivariate model identified the patient RP VAS, SHAQ pain VAS, and SHAQ gastrointestinal VAS subscales as the strongest independent predictors of the RCS.
Conclusion. Patient-reported assessment of SSc-RP severity is associated with a number of factors including pain, catastrophizing, and coping strategies. The effects of seasonal variation in environmental temperature on SSc-RP burden has implications for clinical trial design. Treatments targeting SSc-RP pain and the development of behavioral interventions enhancing coping strategies may reduce the burden of SSc-RP.
- RAYNAUD PHENOMENON
- SYSTEMIC SCLEROSIS
- OUTCOME MEASURES
- PATIENT-REPORTED OUTCOMES
- CLINICAL TRIALS
- VALIDATION STUDIES
Raynaud phenomenon (RP) describes episodic excessive vasoconstriction of the digital microvasculature in response to cold exposure and/or emotional stress1. It is the most common manifestation of systemic sclerosis (SSc) and a major cause of disease-related morbidity2,3,4. The severity of the underlying digital obliterative microangiopathy and the relative efficacy of vasodilator medications are likely to contribute to the wide interindividual variation in the severity and effect of SSc-RP5. A number of additional factors contribute to SSc-RP burden, including seasonal variation in environmental exposure to cold and the positive steps taken by patients to avoid or ameliorate the conditions responsible for SSc-RP symptoms4,6,7. Symptom habituation and adaptation further moderate the effect of SSc-RP symptoms4. The Raynaud Condition Score (RCS) diary is currently the preferred endpoint for SSc-RP clinical trials8. Collected over a 1- to 2-week period, the RCS diary provides an estimate of the mean daily frequency of SSc-RP attacks, the mean daily duration of SSc-RP attacks, and a mean daily assessment of the effect/severity of SSc-RP symptoms [applied as either an 11-point numerical rating scale (NRS) or 100-mm visual analog scale VAS)]. The mean daily frequency and duration of RP attacks during 2-week RCS diary collection have been relatively consistent across studies (between 3–4 attacks per day with a mean daily aggregate duration of 30–90 min/day, equating to an average duration of ∼15–20 min per attack3). Similarly, the mean RCS in patients with SSc is typically ∼4.4/10 on an 11-point NRS3. Much of these data have been obtained in clinical trial settings, typically undertaken during winter and often mandating a minimum threshold number of RP attacks in the period prior to study entry3. Establishing treatment efficacy using RCS diary variables has been challenging, with clinical trials of promising vasodilator therapies yielding negative or modestly positive findings at best9,10,11,12,13. The high placebo response and poor agreement between RCS diary variables and objective assessments of digital perfusion have caused consternation5,14. Additional concerns about the RCS diary have been raised among patients and SSc experts7,15. A thorough understanding of the factors contributing to RCS diary outcomes could provide insight into its performance as an endpoint in clinical trials, influence future SSc-RP clinical trial design, and support the development of novel approaches to SSc-RP management. The Raynaud Symptom Study (RSS) is a multicenter longitudinal study designed to assess the features and determinants of RP symptoms in SSc. We have recently reported the clinical significance of RP symptom characteristics in SSc16. This second report from the RSS focuses on the relative influence of putative factors including clinical phenotype, patient demographics, coping strategies, catastrophizing, and seasonal variation in environmental temperatures on patient-reported assessment of RP severity.
MATERIALS AND METHODS
Patients
SSc patients fulfilling the 2013 American College of Rheumatology/European League Against Rheumatism classification criteria for SSc17 were enrolled at routine clinical care visits from SSc clinics in Bath, UK, and Utah, USA, between April 2015 and January 2017. All patients spoke English. The study received ethical approval at each site (Bath REC 15/LO/1521 and Utah IRB #80665) and all participants provided informed written consent.
Clinician case report form (CRF)
A clinician CRF collected information on patient demographics (age, sex, ethnicity, disease duration based on time since first non-RP symptom), smoking history, clinical phenotype, and autoimmune serology. The clinical phenotype was established using documented evidence from the case notes of gastroesophageal reflux disease symptoms, puffy fingers, sclerodactyly, digital ulcers (DU), digital pitting, telangiectases, pulmonary arterial hypertension, interstitial lung disease, and autoantibody specificity. Relevant comorbidities and vasoactive medication use were documented. Clinicians completed 100-mm VAS for the physician’s global assessment, physician RP severity, and physician DU severity.
Patient questionnaires
Each participant received an RSS questionnaire containing the Scleroderma Health Assessment Questionnaire [SHAQ; comprising the HAQ–Disability Index (HAQ-DI) and SSc-specific 150-mm VAS subscales18], the 1-item Coping Skills Questionnaire (CSQ19; a validated abridged version of the original CSQ20), the Pain Catastrophizing Scale (PCS)21, and separate 100-mm VAS assessments for SSc-RP severity, patient’s global assessment, and DU severity (Supplementary Data, available from the authors on request). The 7 items of the 1-item CSQ are scored using a 7-point NRS (0–6, ranging from “never do” to “always do that”), with each representing distinct domains pertaining to the following: diverting attention, reinterpreting pain sensations, catastrophizing, ignoring sensations, praying and hoping, coping self-statements, and increasing behavioral activities19,20. Patients were dichotomized for each domain according to low coping strategies (score 0–2) and high coping strategies (score 3–6). The 13-item PCS was developed to investigate mechanisms by which catastrophizing affects pain experiences21. Each item is scored using a 5-point NRS (0–4, ranging from “not at all” to “all the time”). A composite score (0–52) was derived and a cutoff of 30 applied to dichotomize the group into copers and catastrophizers (based on earlier work identifying a score of 30 as corresponding to the 75th percentile in samples of chronic pain patients)21. Subscales for rumination (items 8–11), helplessness (items 1–5, 12) and magnification (items 6, 7, 13) were also derived from the PCS, as previously described21. Participants were instructed on completion of the 2-week RCS diary from which we derived the mean daily RCS, mean daily frequency of RP attacks, and mean aggregate daily duration of RP attacks (providing a minimum of 10 days out of 14 had been completed satisfactorily). Participants completed a second RCS diary 6 months following enrollment.
Weather data
The daily maximum and minimum temperature from Bath, and Salt Lake City, Utah, weather stations was obtained using UK Meteorological Office data for April 2015 through July 2017.
Statistical analysis
Descriptive statistics are defined where applicable. The chi-square test was used to compare observed frequencies across 2 or more categories. The unpaired t test was applied when comparing continuous data between groups for patient demographics. Mann-Whitney U and Kruskal-Wallis tests were applied to examine distributions of scores across multiple independent samples as appropriate. Spearman ρ correlation coefficients were used to assess the relationship between independent continuous variables. Multiple linear regression models assessed the combined effect of several variables on the mean daily RCS as the major response variable. A univariate simple linear regression model was first developed before establishing the best-fitting multivariate model according to all possible combinations of the variables found to be significant at p < 0.01 in the univariate analysis. The baseline RCS diary variables were used for analyses examining associations with baseline questionnaire outcomes (e.g., relationship with CSQ and PCS). The effect of season was assessed by pooling RCS diary returns and categorizing patients according to season of enrollment (winter: December 1–February 28, spring: March 1–May 31, summer: June 1–August 31, and fall: September 1–November 30). Local Meteorological Office data during the period of RCS diary collection were used to further examine the relationship between environmental temperature exposure and SSc-RP symptoms. The corresponding mean daily maximum and minimum temperatures were calculated within each period of RCS diary collection.
RESULTS
Patient demographics and missing data
The RSS enrolled 107 patients with SSc (57 in Bath and 50 from Utah). Ninety-four patients (82 female, 14 patients with diffuse cutaneous SSc) returned completed baseline questionnaires. The patient demographics and clinical phenotype of the cohort are summarized in Table 1. The 2 cohorts were similar and we did not consider the lower age of the Utah cohort (mean of 56.4 yrs vs 65.1 yrs) clinically meaningful or likely to have influenced our pooled analyses. The CSQ was adequately completed by 87 participants and the PCS by 84 participants. Baseline RCS diaries were returned by 88 participants (with at least 1 of the 3 RCS diary variables being adequately completed in 86 subjects). Sixty-eight subjects returned the 6-month diary (mean of 198 days ± SD 76 between diaries), allowing a total pooled analysis of up to 154 RCS diary returns. Adequately completed baseline and 6-month RCS diaries were available for 66 patients (70.2%). A full breakdown of missing data is available as Supplementary Data (available from the authors on request).
Associations between patient coping strategies on SSc-RP symptom burden
Higher scores for “praying and hoping” and “catastrophizing” domains of the CSQ were associated with significantly higher RCS (p < 0.05; Table 2), indicating a relationship between these thoughts and higher burden of RP. Significantly higher RCS were also identified among participants reporting the adoption of coping strategies to “reinterpret symptoms,” and “coping self-statements” (Table 2). Efforts to effectively “ignore sensations” was associated with lower RCS (not statistically significant). There was no significant relationship between “increased behaviors” (doing other activities despite symptoms) or “diverting attention” and the distribution of the RCS. Fewer statistically significant associations were identified between coping strategies and either the frequency/duration of SSc-RP attacks, although trends were present that mirrored the findings with the RCS (Table 2).
Association between catastrophizing and SSc-RP symptom burden
The relationship between catastrophizing and RCS diary outcomes was replicated using the PCS data. When dichotomizing the group (< or > a composite score of 30), “copers” had significantly lower median (interquartile range) mean daily RCS compared to the “catastrophizers” [1.7 (0.6–3.7) vs 4.7 (2.8–6.9), p < 0.01; Table 2]. Similar trends were observed for the mean daily frequency and duration of SSc-RP attacks. There was a moderate positive correlation between the total PCS and RCS across the cohort (Spearman ρ 0.42, p < 0.01; Table 3). This relationship was strongest for domains concerning “helplessness” (ρ 0.47) and “magnification” (ρ 0.43) when compared to “rumination” (ρ 0.35). No significant correlation was identified between the total PCS and frequency/duration of RP attacks (Table 3). Significant correlation coefficients were also identified between PCS total scores (and subdomains) and patient RP VAS (ρ 0.35), patient DU VAS (ρ 0.33), and patient global VAS (ρ 0.47). A weak positive correlation was identified between the total PCS and physician RP VAS (ρ 0.23, p < 0.05), but there were no other correlations with physician assessments. There was a positive correlation between the total PCS and the HAQ–DI (ρ 0.42) and each of the SHAQ subscales with the exception of the SHAQ breathing VAS (Table 3). Across all the analyses, the association between patient-reported outcome (PRO) instruments and PCS was strongest for the domains concerning “helplessness.” There was no relationship between total PCS and disease duration, although a weak negative correlation with age (ρ −0.27, p < 0.05) may indicate partially successful adaptation with advancing years.
Association between RCS diary responses and environmental temperature
Using pooled data from the 154 RCS diary returns (combined baseline and 6-month data), there was a weak negative correlation between the mean daily RCS and both mean daily maximum and minimum temperatures (Spearman ρ −0.22, p < 0.01 for both analyses). There were weak negative correlations between the mean daily frequency of RP attacks and mean daily maximum and minimum temperatures (Spearman ρ −0.27, p < 0.01 for both analyses). Similar trends were observed for mean daily duration of attacks (ρ −0.26 and −0.25, respectively, p < 0.01). The association between environmental temperatures and RCS diary outcomes was further supported when examining the distribution of RCS diary responses according to season of enrollment, with significantly higher RCS diary variables observed in winter compared to summer (Table 4).
Overall determinants of the RCS
A multivariate model was built to evaluate the relative contribution of all the relevant factors (weather, coping, other PRO instruments, clinical features, etc.) on the RCS. A parametric approach was necessary for building the multivariate model, but the univariate findings were consistent with the nonparametric analyses presented earlier (Table 2 and Table 3). Simple linear regression identified strongly significant associations (p < 0.01) between the RCS and the PCS, several of the CSQ domains [particularly “catastrophizing” (p < 0.001) and “coping self-statements” (p = 0.0044)], patient RP VAS, patient DU VAS, patient global VAS, HAQ-DI (and all SHAQ subscales other than breathing VAS). Univariate analysis did not identify significant relationships between RCS and physician assessments, age, disease duration, environmental temperatures, sex, smoking history, vasodilator use, history of DU, or disease subset (Table 5). The individual correlations between RCS and PCS scores were all significant using both Spearman ρ and simple linear regression. Only total PCS was incorporated into the multivariate model because the total PCS was perfectly collinear with the sum of the subscores and the PCS subscores were highly correlated (Pearson correlation > 0.7 for all pairs). The best-fitting multivariate model included only the patient RP VAS, the SHAQ pain VAS, and SHAQ gastrointestinal (GI) VAS when assessing all possible combinations of the variables found to be significant at the p < 0.01 level in the univariate analyses (Table 5). Other model selection techniques also identified these 3 predictors. The final multivariate model suggests increases in each of these PRO instruments were significantly associated with increases in mean daily RCS. None of the coping strategy scores were selected in the final model, indicating that we would expect 2 patients with similar SHAQ and RP VAS scores but different coping strategies to have similar RCS diary responses (Table 5).
DISCUSSION
We report the findings of a large study investigating the factors influencing self-report of RP in SSc. The overall burden of SSc-RP symptoms is not a simple linear relationship with the extent of digital vasoconstriction but the complex interplay of factors including but not limited to pain perception, coping strategies, catastrophizing, and seasonal variation in weather.
To our knowledge, our study is the first to examine the relationship between coping and patient-reported SSc-RP symptoms. Our findings confirm an earlier consensus among SSc experts, in which 90% considered coping strategies to be an important determinant of RCS diary outcomes15. Catastrophizing appears to be important with patients who report this behavior (using both the CSQ and PCS instruments), consistently reporting a higher burden of SSc-RP symptoms. The direction of causality cannot easily be determined but the identified associations between SSc-RP severity and coping strategies (such as catastrophizing) could be used to develop novel behavioral approaches to enhance resilience to reduce the effect of RP. A recent large qualitative study of SSc-RP patients independently identified a diverse range of coping strategies (including diverting attention, ignoring sensations, and coping self-statements) that individuals with SSc report adopting to lessen the burden of RP symptoms4. These coping strategies formed an important component of an emergent theme around “adaptation” in the patient experience of SSc-RP, and future interventions could capitalize on these observations4. Cognitive-behavioral interventions that modify catastrophizing (specifically concerning feelings of helplessness and tendency toward symptom magnification) could be used to reduce the burden of SSc-RP. The direction of the relationship between the adoption of coping strategies around “reinterpretation” and “coping self-statements” and RCS diary outcomes was somewhat unexpected (with patients reporting a higher effect of SSc-RP despite the use of coping strategies within these domains). Nonetheless, strategies to help patients to think about their RP symptoms in more neutral terms (reinterpretation), to desist from catastrophizing thoughts (such as “I can’t stand it anymore”), or to devise positive coping self-statements (such as “No matter how bad it gets, I can do it” or “It won’t last much longer”) may help lessen the burden of SSc-RP. A similar approach has been shown to modify pain endurance in patients with other forms of chronic pain22. Intriguingly, the method chosen for inducing experimental pain in this work (a cold pressor test involving the immersion of hands into cold-water baths) closely resembles the conditions and physiological responses accountable for SSc-RP symptoms. Other coping strategies might be less helpful in SSc-RP. For example, cognitive-behavioral interventions targeting “ignoring sensations” and “increasing behavioral activities” might result in excessive exposure to activities that might exacerbate peripheral vasoconstriction23. Previous behavioral interventions for RP have examined approaches such as biofeedback that from a modern perspective are deemed ineffective. Behavioral interventions focusing on the modification of catastrophizing and coping strategies could be a potentially effective but hitherto neglected area of therapeutics for RP symptoms. Resiliency training is being increasingly used as an intervention to modify quality of life and function in people affected by chronic disease24.
Our findings also suggest that interventions targeting SSc-RP pain might be as important as efforts to promote peripheral vasodilation. Our multivariate analysis identified the patient pain VAS (from the SHAQ), the patient RP VAS, and the SHAQ GI VAS as independent determinants of the RCS. While the association with the SHAQ GI VAS might represent a genuine association between RP severity and GI involvement in SSc, it is also possible this reflects an important shared contribution of pain/illness perception in both RP and GI severity self-report. Indeed, the SHAQ breathing VAS (which has no conceptual associations with pain) was the only SHAQ subscale not associated with RCS within the univariate analysis. Similarly, the SHAQ breathing VAS was the only subscale not to correlate with PCS variables. A complex nonlinear relationship between the severity of digital vasculopathy and pain perception may explain the poor agreement between subjective (RCS diary) assessment of RP burden and objective assessment of digital vascular function in SSc14,25.
In contrast, putative factors such as age, disease duration, disease subset, vasodilator use, smoking history, or history of DU do not appear to be associated with patient-reported severity and effect of SSc-RP symptoms. The original validation work of the RCS diary identified differences in RCS (but not mean daily frequency or duration of RP attacks) in patients with and without DU26. This study used data from a clinical trial that recruited patients only during the winter months, examined differences in RCS diary variables in patients with “active” DU (rather than a history of DU), and importantly, used RCS item wording that encouraged patients to consider the effect of “digital sores” when choosing their RCS26.
Our findings raise additional issues relevant to the design and interpretation of SSc-RP clinical trials. We have confirmed a previously reported association between seasonal variation in environmental temperature and RP symptom burden, although our findings suggested a lower burden of RP symptoms in winter than those previously reported in a relatively smaller study of 18 patients with SSc6. Our findings might be of value for future clinical trial power calculations, particularly regarding studies countenancing enrollment outside winter. The influence of seasonal variation is not unexpected but has often been overlooked in RP clinical trial design and interpretation. For example, concerns are frequently raised about the magnitude of the placebo response in RP clinical trials with 1 study identifying a > 50% improvement in RCS in over one-fifth of patients following placebo administration5. The pooled analysis for this estimation was undertaken using data from 3 randomized controlled trials that enrolled at Northern hemispheric sites during winter with primary endpoint analysis in the spring5. Changes in environmental temperature could be an important contributing factor to the placebo effect (a major hurdle to demonstrating efficacy within the treatment arm) and could be modified through the design of shorter RP clinical trials. Efforts under way to devise a novel PRO instrument for SSc-RP that is not reliant on diary collection may support novel clinical trial design that facilitates this27.
Our present study benefits from being a comparatively large multicenter study, but being primarily a cross-sectional study and lacking objective assessment limit the extent to which we can fully examine the determinants of RCS diary outcomes. About 30% of participants did not return the 6-month RCS diary. Study attrition in a longitudinal study of self-administered questionnaires was expected and the majority of our analyses used baseline data alone, for which there were few missing data. Our study has highlighted a number of factors influencing RCS diary outcomes and builds on recent work examining the opinions of patients and experts toward the RCS diary as a clinical trial endpoint7,15. The present study has highlighted a number of factors contributing to SSc-RP symptom burden (such as seasonal variation in environmental temperature) that will help us better interpret RCS diary outcomes and inform future clinical trial design, and may help develop novel behavioral approaches for the management of SSc-RP.
Acknowledgment
We thank the family of Dr. John Glyn and the Royal College of Physicians for awarding Dr. John Pauling the 2015 John Glyn Bursary, which supported this collaborative partnership.
- Accepted for publication November 12, 2018.
REFERENCES
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.
- 7.
- 8.
- 9.
- 10.
- 11.
- 12.
- 13.
- 14.
- 15.
- 16.
- 17.
- 18.
- 19.
- 20.
- 21.
- 22.
- 23.
- 24.
- 25.
- 26.
- 27.