Can we use patient-reported outcomes for remote monitoring in SSc?

In a general SSc population, ePROMs assessing general health in the context of the SF-36 are not associated with skin and/or lung progression. Dear Editor, Systemic sclerosis (SSc) has a heterogeneous nature: while a small minority develops rapidly progressive disease with severe complications, a substantial proportion of patients experience relatively mild disease [1, 2]. Since it remains difficult to predict disease trajectories for individual patients, extensive annual evaluation (including, but not limited to: clinical assessment, laboratory tests, pulmonary function tests and imaging) is recommended for all SSc patients [3]. Currently, healthcare systems around the world are faced with big challenges, such as growing demand, limited capacity and rising costs [4]. To provide continuous, high-quality care, we need affordable and sustainable solutions. In the Netherlands, in rheumatoid arthritis and spondylarthritis patients, remote (or home-) monitoring has been shown to effectively reduce number of outpatient clinic visits and associated healthcare costs [5, 6]. In order to explore the feasibility of remote monitoring in SSc (for example via telephonic consultation), we chose a pragmatic approach to analyse whether two simple questions assessing patients’ general health could identify patients with progressive disease. For this purpose, we used the first two questions of the Dutch version of the 36-Item Short Form Health Survey (SF-36). These comprise: ‘In general, how would you rate your health?’ (SF-1) and ‘Compared with one year ago, how would you rate your health in general?’ (SF-2), with possible responses on a 5-point ordinal scale. For the current analysis, progressive disease was defined by either worsening of mRSS ≥ 5 points and ≥ 25%, and/or ≥ 10% relative decrease in forced vital capacity (FVC), and/or ≥10% relative decrease in diffusing capacity for carbon monoxide (DLCO), compared with the previous visit. We focused on skin and lung parameters because skin progression is strongly associated with overall disease progression and pulmonary function parameters capture both lung involvement and general health status [1, 7]. We analysed data from 767 consecutive SSc patients from the Leiden Combined Care in Systemic Sclerosis (CCISS) cohort who fulfilled ACR/EULAR 2013 criteria and who had complete SF-36 data in ≥1 visit. Of the included patients, 78% (n = 596) was female, mean age was 55.2 (SD 13.9) years, 22% (n = 168) had a diffuse cutaneous subset and median disease duration since first non-Raynaud symptom was 2.2 (IQR 0.5–8.0) years (Table 1). Distribution of responses to SF-1 and SF-2 is shown in Supplementary Fig. S1. Baseline characteristics of SSc patients in the CCISS cohort. Parametric data are reported as mean ± SD, non-parametric data as median (IQR). Immunosuppressives include methotrexate, mycophenolate mofetil, corticosteroids, hydroxychloroquine, cyclophosphamide and azathioprine. Stable disease is defined as patients who had ≥ 3 consecutive progression-free visits. SSc: systemic sclerosis; ACA: anti-centromere antibody; GAVE: gastric antral vascular ectasia; ILD: interstitial lung disease; PAH: pulmonary arterial hypertension; FVC: forced vital capacity; DLCO: diffusing capacity for carbon monoxide. Skin and/or lung progression was observed in 392 patients (51.1%). After correcting for age and sex, both SF-1 and SF-2 responses were not associated with disease progression (Table 2). However, in the context of remote monitoring, it is of key importance to select the right subgroup of eligible patients. For example, early and active patients need frequent follow-up regardless of patient-reported outcome measures (PROMs). Therefore, we performed a subgroup analysis in ‘stable’ patients, defined as patients who had three consecutive ‘disease progression free’ visits, and then analysed the fourth subsequent visit. In this subgroup (n = 144 patients), the percentage of patients who met the progression criteria decreased to 20%. In order to evaluate discriminative ability of SF-1 and SF-2 for skin and/or lung progression in this stable subpopulation, we calculated sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for the ‘worst’ responses vs ‘neutral’ and ‘good’ responses (Supplementary Table S1). This yielded a sensitivity of 65.5%, a specificity of 53.0%, a PPV of 26.0% and a NPV of 85.2% for SF-1, and a sensitivity of 58.6%, a specificity of 60.0%, a PPV of 27.0% and a NPV of 85.2% for SF-2. Since in many observational studies a 10% DLCO decline is considered less meaningful than a 15% DLCO decline, we repeated both analyses including this threshold which did not change the results (Supplementary Table S2). Considering our ultimate aim of optimizing healthcare by safely extending outpatient clinic intervals for patients with stable disease, one might argue that an NPV of 85% could be acceptable for remote monitoring in this subgroup. However, this approach would lead to undetected skin and/or lung progression in three out of twenty patients. Association of SF-1 and SF-2 response with skin and/or pulmonary progression. Coefficients are derived from mixed-effects logistic regression analysis, which included age and sex as fixed effects and a random intercept and slope for each patient. Strikingly, the distribution of the answers was very comparable for SSc patients with progression and SSc patients without progression, indicating that patient-reported health is not determined by signs like skin worsening or pulmonary function decrease that might urge intensification of treatment according to the physician. This could be explained by moderate correlation between skin and/or lung progression and other disease domains such as pain, fatigue, functional limitations and gastrointestinal problems. Another important consideration is that fulfilling progression criteria does not necessarily translate into the need for a change of treatment. In the current study, we sought to study whether ePROMs could identify the subgroup of patients in which repetitive diagnostic testing including pulmonary function test and mRSS is needed. One could argue that initiation or modification of treatment would be considered a more clinically relevant end point. In conclusion, we found no significant association between SF-36 patient-reported general health and skin and/or pulmonary function worsening in SSc. Future research in remote monitoring in SSc should investigate alternative—or complementary—remote monitoring strategies that reflect the full spectrum of disease heterogeneity in SSc. This paper supports the development of comprehensive prediction models for remote monitoring in stable SSc. Supplementary material is available at Rheumatology online. A data dictionary will be made available upon request. Deidentified individual participant data could be made available upon reasonable request. Data sharing will have to follow appropriate regulations by the Dutch law. No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article. Disclosure statement: The authors have declared no conflicts of interest. Ethical approval and patient consent statement: The Leiden CCISS cohort was previously approved by the local Ethics Committee (CME no. B16.037, REU 043/SH/sh, P09.003/SH/second). All patients who are registered in the CCISS gave written informed consent for participation.