The impact of sex and disease classification on patient-reported outcome measures in axial spondyloarthritis: a descriptive prospective cross-sectional study

Background The aim of this study was to explore the impact of sex and disease classification on outcomes in axial spondyloarthritis (axSpA) patients, including both radiographic (r-) axSpA and non-radiographic (nr-) axSpA, in males and females, respectively. Methods AxSpA patients were consecutively recruited from two rheumatology outpatient university clinics. We explored how sex and axSpA disease classification affected patient-reported outcome measures (PROMs). General linear models were used to investigate if there was an association between the continuous variables and each of the main effects of interest (sex and axSpA classification), as well as the possible interaction between them. Categorical outcome measures were analyzed with the use of logistic regression with the same fixed effects. We analyzed the relationship between tender point count (TPC) and the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI). The prevalence of extra-articular manifestations (EAMs) and the Charlson Comorbidity Index (CCI) were determined. Results According to the protocol, a total of 100 outpatients with axSpA were enrolled (r-axSpA males 30, r-axSpA females 10, nr-axSpA males 25, nr-axSpA females 35). The BASDAI scores appeared higher among nr-axSpA females (median [Q1; Q3], 47 [21; 60]) compared with the combined median for the 3 other subgroups 25 [12; 25]. Female sex was associated with a higher number of tender point count (TPC, P < 0.001). TPC and BASDAI were correlated for female nr-axSpA patients (r = 0.44, P = 0.008) and male nr-axSpA patients (r = 0.56, P = 0.003). Being classified as nr-axSpA was associated with a lower SF-36 Mental Component Summary (median for the 4 subgroups: nr-axSpa females 46.7, nr-axSpA males 52.3 vs. r-axSpA males 56.9 and r-axSpA females 50.4). EAMs were frequent (up to 50%). The CCI was low in all 4 subgroups, and no difference in the CCI between the subgroups was observed (P = 0.14). However, male sex had a significant impact on the CCI (P = 0.03). Conclusions In summary, patients with r-axSpA, regardless of sex, appeared less affected on most PROMs compared with nr-axSpA patients. However, female sex was associated with a higher number of TPC. TPC could possibly confound disease activity outcomes such as BASDAI, and one can consider different thresholds for defining high disease activity depending on the patient’s sex. Trial registration The trial is registered and approved by the Region of Southern Denmark’s Ethics Committee (S-20150219). Registered 19 February 2015.


Background and rationale
Spondyloarthritis (SpA) is a heterogenous group of chronic rheumatic diseases; it can be dominated by peripheral joint involvement, classified as peripheral SpA (pSpA), or by inflammatory back pain, classified as axial SpA (axSpA). Furthermore axSpA is subdivided into two stages: Nonradiographic (nr-axSpA) and radiographic (r-axSpA, formerly known as ankylosing spondylitis, AS). Next to the spinal and articular symptoms, many patients with axSpA also have extra-articular manifestations (EAMs; e.g. uveitis, enthesitis, psoriasis, inflammatory bowel disease) which contribute to reduced health-related quality of life (2). We anticipate that EAMs among patients with axSpA are underestimated and therefore undertreated.
Besides EAMs, axSpA patients have a higher risk of comorbidities, including diabetes mellitus, renal disease, cardiovascular disease (CVD), and pulmonary disease (3)(4)(5). Awareness of comorbidities among clinicians is important in view of their role for treatment choices for axSpA patients. Other medical specialists and care professionals do of course also have their place. AxSpA disorders are generally diagnosed more often in men than in women (6). In reference to the most representative form of these disorders, r-axSpA, three male cases are documented for every female case (6). In contrast to r-axSpA, nr-axSpA patients show little difference in prevalence among males and females (7). Nevertheless, little is known of the differential clinical expression of axSpA between males and females, and females are often underrepresented in clinical research (8,9). The higher incidence of r-axSpA in males compared to females has stimulated research on gender differences in axSpA. Few studies have analysed gender as a prognostic factor for course of disease in axSpA patients (8,10,11) and there is a need to determine the particularities and severity of axSpA in the female gender.

Aims
The overall objective of this cross-sectional study is to present a characterisation of axSpA patients, to Statistical Analysis Plan (SAP) illuminate patient-reported outcome measures, and ultimately elucidate factors associated with disease outcome.

Hypotheses
We hypothesise that the amount of self-reported disease activity is more pronounced in women than in men (i.e. Bath Ankylosing Spondylitis Disease Activity Index [BASDAI (12)], and Bath Ankylosing Spondylitis Functional Index [BASFI (13)]).

Objectives
The primary objective of this study is to analyse the impact of gender on self-reported disease activity in axSpA patients based on the type of clinical diagnosis.
Secondary objectives are: a. To determine the relationship between the continuous measure of muscular tenderpoints (TP (14)) and the self-reported disease activity (i.e. BASDAI).
b. To determine the prevalence of EAMs and comorbidities, as defined by the Charlson comobidy index, in axSpA patients. c. To compare the impact on health-related quality of life of axSpA patients, specifically in terms of how axSpA classification differentially affects aspects of physical, mental and social wellbeing.

Sample size considerations
This study was designed as an exploratory study. A sample size of 100 individuals was considered sufficient to test the hypothesis: 'is there a difference between reported BASDAI scores in axSpA patients by men vs women? For a two-sample pooled t test of a normal mean difference with a two-sided significance level of 0.05, assuming a common standard deviation of 1.5 BASDAI units, a total sample size of 100 assuming a balanced design (1:1 male-female ratio) has sufficient power (91%) to detect a mean difference of 1 BASDAI units. All descriptive statistics and tests will be reported in accordance with the recommendations of the "Enhancing the QUAlity and Transparency Of health Research" (EQUATOR) network (15): the STROBE Statement (1). We consider p values less than 0.05 (and 95% confidence intervals excluding the null) to be statistically significant.

Study design
The study is designed as a cross-sectional study with prospective enrolment of axSpA patients. Information about the patients and their exposures were collected at a single centre at one visit according to the clinical standards in Denmark. Examinations were conducted on the same day. The inclusion period was between 1 st of April 2016 and 30 th of March 2018.

Study participants
Patients with axSpA seen in the Department of Rheumatology, Odense University Hospital, Svendborg/Odense were recruited. We used the ASAS classification criteria for axSpA (imaging arm) (16), and the modified New York criteria for r-axSpA (17). To be considered for inclusion, participants must have the ability and willingness to give written informed consent and to meet the requirements of the prespecified protocol. Participants were excluded from the study if any of the following criteria were present: (1) no consent, (2) does not understand Danish, (3) age < 18 years.

Statistical interim analyses and stopping guidance
Preliminary results have been presented at the EULAR congress in 2017 in Madrid (18). 80 participants were included at that time point.

Timing of final analysis
The first main report/publication of the trial will be prepared when the SAP is completed, and the research team has approved it (anticipated to be November 2018).
Urine samples were collected in this study to analyse whether the urine electrolytes were altered and related to nephrolithiasis in axSpA patients. The results will not be part of the proposed manuscript, but will be presented in a separate future paper.

Confidence intervals and P values
Level of statistical significance and use of confidence intervals: All applicable statistical tests will be 2-sided and will be performed using a 5% statistical significance level. No correction for multiple testing will be conducted; instead the results will be interpreted with caution in the context multiplicity.

Protocol deviations
To specify the aim of this study and to clarify the statistical analyses, some deviations from the prespecified protocol were made.

Aims:
Protocol: The overall objective of this cross-sectional study is to perform a characterisation of SpA patients, to identify clinical phenotypes of SpA and ultimately elucidate factors predictive of disease outcome.

SAP:
The overall objective of this cross-sectional study is to present a characterisation of axSpA patients, to illuminate patient-reported outcome measures, and ultimately elucidate factors associated with disease outcome.

Eligibility
The number of ineligible patients will be reported with reasons for ineligibility.

Recruitment
A flowchart template modified from http://www.consort-statement.org comprising the number of people screened, eligible, consented, and included in the analyses will be used (Figure 1).

Collected variables
The following information was collected at the study visit ( Table 1)

Patient-reported outcome measures (PROMs)
Touchscreen questionnaires were used for PROMs at the inclusion visit. The Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) is the patient's evaluation of disease activity. The BASDAI consists of a 6item scale to measure severity of fatigue, spinal and peripheral joint pain, localised tenderness, morning stiffness (duration and severity), using the VAS scale (0-100 mm). To give each symptom equal weighting, the average of the two scores relating to morning stiffness is taken. The resulting 0 to 500 score is divided by 5 to give a final 0-100 BASDAI score. Scores of 40 or greater indicate suboptimal control of disease (12).
Physical function was addressed using the Bath Ankylosing Spondylitis Functional Index (BASFI). The BASFI is a 10-item scale on which respondents rate the degree of difficulty they have in performing certain tasks, using visual analogue scales (VAS) from 0 (easy) to 100 (impossible). The mean of the 10 responses is the BASFI score (13). Pain was assessed by a Visual Analogue Scale (VAS) pain last week/spine/at night/due to AxSpA on a 0-100 mm scale. The intensity of fatigue was assessed by a VAS scale (0-100 mm). Healthrelated quality of life (HRQoL) was addressed using the Medical Outcomes Study SF-36; SF-36 is a generic health status questionnaire that was developed as a tool to compare various aspects of health status across a general and broad patient population (20)(21)(22). The SF-36 examines eight general health domains: physical functioning, role limitations due to physical health problems, bodily pain, general health perceptions, vitality, social functioning, role limitations due to emotional problems and mental health. Furthermore, a physical and mental component summary score will be calculated. We used the Danish version of SF-36 (23) which uses a 4-week recall period. The PGA was assessed by a single question with a range from 0 to

Clinical examination
Spinal mobility was addressed by the BASMI. The BASMI is a composite index of spinal mobility, and is used internationally in research and clinical practice and is recommended by the ASAS (24). RAA performed a 44 swollen/tender joint count, a SPARCC enthesitis score (25) and a tender point count (26).

Para-clinical assessment
Blood samples as specified in the protocol was collected by a trained laboratory technician and treated according to set procedures. Results on inflammation (C-reactive protein, Plasma Calprotectin), Human Leukocyte Antigen B27 (HLA-B27) and faecal calprotectin will be presented in this study.

Analysis methods
This study was designed as a descriptive cross-sectional study. The characteristics of the participants will be described for each gender and axSpA classification using descriptive statistics. Means and SDs or medians and IQRs and Binary outcomes will be presented as numbers and corresponding percentages. The primary variable of interest is BASDAI coded as a continuous variable. The primary analyses will be based on General Linear Models (GLM), investigating if there is an association between BASDAI and each of the covariates of interest (gender and axSpA classification). Furthermore, a test for interaction between gender and axSpA classification will be performed. The model will include gender, axSpA classification and interaction between them as fixed effects. Categorical outcome measures will be analysed with the use of logistic regression with the same fixed effects. A sensitivity analysis will be performed to assess the robustness of the primary analyses, by imputation missing data with the grand mean. Secondary analyses will include the same model using the secondary outcome measures (see Table 1).
The relationship between the continuous measure of tender point count and the patient-reported disease activity score (BASDAI) will be analysed by scatter plots with Spearman rank correlation coefficients. We will stratify on gender (Figure 2A), and axSpA classification ( Figure 2B) (Appendix A). All models will be tested for their assumptions.

Missing data
The number of participants with missing observations will be reported for each outcome variable. We will use a simple manual imputation by replacing missing data with the grand mean of the variables. The presentation and interpretation of health-related quality of life from SF-36 is complex, and the impact of patterns of disease can be difficult to evaluate (27). We will use 'Spydergrams' which provide a visual method to examine the eight domains of health-related quality of life scored from 0 -100 simultaneously in a single figure (27); using spydergrams will enable us to compare the impact on health related quality of life of axSpA patients, specifically in terms of how axSpA classification differentially affects aspects of physical, mental and social wellbeing (Figure 3).

Statistical software
The analyses will be conducted using STATA (version 15). The program code for the primary analysis in the statistical program: .glm BASDAI i.Gender i.axSpA-classification, family (gamma) link (identity) .glm BASDAI i.Gender##i.axSpA-classification, family (gamma) link (identity) Figure 1. Numbers of axSpA patients who were screened, included in the study, and included in the analyses. Flow chart template modified from http://consort-statement.org.

Demographics
Males, n = Females, n = Males, n = Females, n = Effect of gender Effecf of axSpA classification Age, years