Are health status and treatment burden associated with non-response bias in patients with chronic spinal pain a prospective cohort study

Background: Chronic neck or back pain (spinal pain) is prevalent and among the leading causes of disability worldwide [1]. It often coexists with other chronic conditions, increasing both individual and societal burdens [2]. The coexistence of two or more chronic conditions is often termed multimorbidity [3], and individuals with multimorbidity are frequently underrepresented in research due to exclusion from clinical trials [4]. As multimorbidity is associated with greater personal burdens, poorer health, and reduced quality of life [3], these individuals may be less likely to participate in studies, introducing non-response bias. This occurs when those who do not respond differ in important ways from those who do, potentially skewing research findings [5]. Successive wave analyses, a method involving multiple reminders to non-respondents, can help assess non-response bias by comparing characteristics of later responders with those of initial responders [6]. It is unclear, however, whether self-reported factors related to multimorbidity, such as treatment burden and health status, contribute to this bias. Objectives: To investigate whether self-reported treatment burden, health-related quality of life, health status, and the number of comorbidities, are associated with non-response bias in patients with chronic spinal pain. Methods: This prospective observational open-cohort study enrolled patients referred to Aalborg University Hospital for spinal pain from June 1, 2023 to April 1, 2024. Patient information was obtained from medical records and electronic questionnaires. Patient reported outcome measures (PROMs) were collected at baseline and at 3-month and 6-month follow-up. To investigate non-response bias using the successive wave analysis method, patients were categorized into groups based on their response patterns to the questionnaires: immediate responders (Group 1), those responding after one reminder (Group 2), those responding after two reminders (Group 3), all respondents (Group 4), and non-respondents (Group 5). Associations between response patterns and self-reported treatment burden (MTBQ), health-related quality of life (EQ-5D-5L), health status (EQ-5D-5L VAS), and the number of comorbidities were assessed at baseline, 3-month follow-up, and 6-month follow-up. The MTBQ is a 10-item measure of challenges in managing multimorbidity, including medication use and healthcare coordination [7], while EQ-5D-5L assesses five health dimensions on a 5-point scale, and EQ-5D-5L VAS rates overall health on a 100mm scale from 0 (worst) to 100 (best) [8]. Statistical analyses included one-way ANCOVA analyses and mixed regressions with repeated measures. Additionally, sex and age differences between respondents and non-respondents to the questionnaires were examined using chi-square test and independent samples t-tests. Results: A total of 360 respondents and 280 non-respondents were included in the study. See Table 1 and Table 2. Among the respondents 68.6% met the criteria for multimorbidity, and nearly half of the respondents (45.9%) reported a medium to high treatment burden. One-way ANCOVA analyses revealed no significant differences between groups at baseline based on MTBQ, EQ-5D-5L, EQ-5D-5L VAS, or number of comorbidities. The mixed models of repeated measures revealed no significant differences over time in the outcomes. However, non-respondents were younger compared with the respondents, with a mean difference of 5 years (95% CI: -7.8 to -2.2, p-value: