Factors Associated With Initiation of Biologics in Patients With Axial Spondyloarthritis in an Urban Asian City


Wan Yu Png, BSc (Pharm)(Hons); Yu Heng Kwan, BSc (Pharm)(Hons); Yi Xuan Lee, BSc (Pharm)(Hons); Ka Keat Lim, MPharm, MSc; Eng Hui Chew, BSc (Pharm)(Hons), PhD (Pharm); Nai Lee Lui, MBBS, MMed, MRCP, FAMS; Chuen Seng Tan, BSc, MSc, PhD; Julian Thumboo, MBBS, MMed, MRCP, FAMS, FRCP; Truls Østbye, MD, MPH, MBA, PhD, FFPH; Warren Fong, MBBS, MRCP, FAMS


J Clin Rheumatol. 2019;25(2):59-64. 

In This Article


Study Design and Population

This prospective longitudinal study used data from a dedicated registry, the PREcision medicine in SPondyloarthritis for better Outcomes aNd Disease remission (PRESPOND) from a tertiary referral center in Singapore General Hospital from January 2011 to July 2016. Recruited patients were followed up for 54 months or up to the point in time when biologics were initiated. Demographics characteristics, clinical data, and patient-reported outcomes (PROs) were collected at point of recruitment and subsequent visits to the clinic. All patients recruited into the PRESPOND registry were older than 18 years and fulfilled the 2009 Assessment of SpondyloArthritis International Society criteria for Spondyloarthritis. Patients with peripheral spondyloarthritis such as psoriatic arthritis and inflammatory bowel disease–associated arthritis were excluded from this study to maintain sample homogeneity because these conditions have several prognostic and therapeutic differences.[10] This study was approved by the institutional review board, and all subjects provided written informed consent.

Study Variables

The dependent variable was whether patients were initiated on biologics at subsequent visits to the clinic. The various demographics, clinical variables, and PRO variables collected at baseline were the independent variables.


Demographic data collected include duration of disease, age, sex, ethnic group, education level, type of housing, and employment status.

Clinical Variables

Clinical data collected include comorbidities, responsiveness to NSAIDs, and laboratory tests, which include C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR).

PRO Variables

Patient-reported outcome variables collected include Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) and Short-Form 36 Health Survey (SF-36).

Bath Ankylosing Spondylitis Disease Activity Index is a self-administered questionnaire consisting of six 10-cm horizontal visual analog scales to measure the severity of fatigue, spinal and peripheral joint pain, localized tenderness, and morning stiffness over the past week. The final BASDAI score ranges from 0 to 10, with higher score indicating higher disease activity.[11] Patients with BASDAI score of greater than or equal to 4 were considered as having high disease activity.

Short-Form 36 Health Survey is a self-administered questionnaire consisting of 36 items that assess 8 health concepts: physical functioning, physical role functioning, bodily pain, general health perceptions, vitality and energy, social role functioning, emotional role functioning, and mental health. Two summary measures of health-related quality of life were then derived: physical component summary (PCS) and mental component summary (MCS), each ranging from 0 to 100, with lower score indicating worse health state.[12] The scores in this study were standardized to overall population norm using the norm-based scale algorithms (mean, 50 [SD, 10]). As there are no previously established cutoff points for defining high or low PCS and MCS scores, they were dichotomized using the mean standard score of 50.[13] Norm-based scale of PCS and MCS scores of greater than 50 were considered as high, and scores of less than or equal to 50 were considered as low in the respective summary scores.

Statistical Analysis

Patients who have missing or incomplete data were excluded from statistical analyses. Descriptive statistics were used to summarize patient's baseline characteristics and clinical variables. For categorical variables, count and percentages are presented, whereas for continuous variables, mean and SD are presented. Differences in baseline characteristics between patients who were initiated on biologics and those who were not were compared using χ 2or Fisher exact test for categorical baseline variables and t test or Wilcoxon rank sum test for continuous variables where appropriate. p < 0.05 was considered statistically significant.

Logistic regression analyses were conducted with status of patients initiated on biologics as a binary outcome variable. Univariable and multivariable logistic regression analyses were performed, with the odds ratios (ORs), 95% confidence intervals (CIs), and p values reported. Univariable logistic regression analyses were performed on all variables identified from literature review, which include duration of disease; age; sex; ethnic group; education level; type of housing (as a proxy for household income); employment status; elevated CRP; ESR; response to NSAIDs; comorbidities, which include hyperlipidemia, hypertension, peptic ulcer disease (PUD), skin symptoms, and past surgery; and PRO variables that include BASDAI and SF-36 (PCS and MCS).

A backward modeling approach was used to reach the final multivariable model. Elimination started with all univariate variables in the model. At each step, variables were evaluated where variables with the largest p value in the multivariable analyses was removed one by one. Factors that were statistically significant at baseline were retained and on the basis that they have been reported in literature and been in use in clinical practice. The BASDAI score was retained in the final model regardless of its p value in the multivariable model because of strong literature evidence of its association with biologics initiation.[14,15] After attaining the final model, likelihood ratio test was used to compare the initial model and the final model. Analyses were performed for patients with complete baseline information using STATA SE version 12.1 for Windows (StataCorp, College Station, TX, 2012).