Seasonal Effect on Fatigue, Pain and Dryness in Primary Sjögren's Syndrome

Pierre-Marie Duret; Nicolas Meyer; Alain Saraux; Valérie Devauchelle-Pensec; Raphaele Seror; Véronique Le-Guern; Claire Larroche; Aleth Perdriger; Jean Sibilia; Vianney Guardiolle; Xavier Mariette; Jacques-Eric Gottenberg


Arthritis Res Ther. 2020;22(39) 

In This Article


Patient Selection

This study analysed patients from the French nationwide multicentre pSS cohort (Assessment of Systemic Signs and Evolution in Sjögren's Syndrome) ASSESS (n = 395),[12] established in 2006, with an available 5-year prospective follow-up and from three randomised placebo-controlled trials of infliximab (TRIPSS; n = 103; follow-up of 22 weeks; 7 visits),[13] rituximab (TEARS; n = 120; follow up of 24 weeks; 6 visits)[14] and hydroxychloroquine (JOQUER; n = 120; follow up of 48 weeks; 4 visits)[15] (Figure 1). All patients fulfilled the American-European Consensus group criteria for pSS.[16] At each visit, in all studies, visual analogue scales (VASs) of patients for pain, fatigue and dryness were collected. Objective assessments of dryness (Schirmer test and unstimulated salivary flow) were collected in the ASSESS cohort at enrolment, second and fifth year of follow-up.

Figure 1.

Study flow diagram. pSS, primary Sjögren's Syndrome; VAS, visual analogue scale; *≥ 1 VAS missing data

Since ASSESS is a non-interventional "natural history" cohort and all the three RCTs did not show any significant treatment effect, a combined analysis of all the available data was conducted.

Statistical Analysis

Data combined from the four previous studies were aligned according to the days, months and seasons of the year. The dataset was collected over 10 years, from 2001 to 2002 (TRIPSS), 2008 to 2011 (TEARS), 2008 to 2012 (JOQUER) and from 2007 to 2014 in the ASSESS cohort.

Patients for whom one of the three VAS criteria (fatigue, pain or dryness) was missing were not included in the analysis.

Quantitative variables are described with the mean and its standard deviation (SD). Normality was assessed using the Shapiro-Wilk test.

Mean VAS for fatigue, pain and dryness were compared first across months and, secondly, in a separate model, across seasons using linear mixed models.

Given the availability, in the ASSESS cohort, of data concerning symptomatic treatments of dryness and systemic immunomodulatory treatments (Additional file 1: Table S1), analyses were adjusted on two potential confounding factors which were age and treatment regimens assumed by patients. Linear mixed models were adjusted on symptomatic treatments of dryness (topical serum eye drops and pilocarpine hydrochloride) and systemic immunomodulatory treatments. The latter included hydroxychloroquine, corticosteroids, methotrexate, leflunomide, azathioprine, mycophenolate mofetil, cyclophosphamide (CYC) and rituximab (RTX) (CYC and RTX were considered as current treatment if prescribed in the 6 months prior to examination).

The models were fitted with a fixed time effect (month or season effect, in distinct models), with a random subject effect used to take into account the repeated structure of the data. In order to detect a cyclic seasonal effect, a cosinus transformation and a time shift was also applied to the time effect in separate models. Statistical analyses were performed using R3.3.1 with the LME4 and HGLM libraries. All statistical tests were two-sided and p values ≤ 0.05 were considered statistically significant.