Are Multimorbidity and Related Factors Associated With Non-response Bias in Patients With Spinal Pain?

Backgrpund:

The interaction between back pain, multimorbidity, and treatment burden can negatively affect patients' participation in treatment pathways and thus their prognosis. However, this has not been sufficiently investigated. A challenge in data collection through, for example, questionnaires is to ensure representative responses from a patient group with expected high treatment burden and lower health literacy. Selective participation can lead to non-response bias and affect conclusions about the impact of multimorbidity on prognoses for individuals with back pain. Therefore, the aim to investigate whether the number of chronic diseases, the treatment burden associated with multimorbidity, and health-related quality of life contribute to non-response bias in this specific population group.

Objectives:

The objectives of this study will be threefold:

1. To examine baseline differences in number of chronic diseases, treatment burden and health-related quality of life among individuals categorized into groups based on their response patterns at baseline.
2. To examine the differences between the aforementioned groups over time (baseline, 3 months, and 6 months).
3. Assess differences in age and sex distribution between responders and non-responders referred for assessment of back pain to questionnaires evaluating their condition.

Through these methods, the aim to uncover any potential non-response bias, which can provide insights for future research. This examination will aid in determining the most effective approach to investigating the influence of multimorbidity on individuals with back pain.

Methods:

In this observational cohort study, data is gathered from patients referred to the Rheumatology Department of Aalborg University Hospital for back pain. Conducted from June 2023 to April 2024, the study recruits participants through electronic links and reminders via E-Boks. Information is sourced from three channels: the department's booking plan (providing details on age and gender of non-responding patients), medical records, and electronic questionnaires. Patient-reported outcomes are collected at baseline, and during 3- and 6-month follow-ups using REDCap.

Patient characteristics include:

• From electronic questionnaire: smoking habits, alcohol consumption, Body mass index, highest achieved education, marital status and work status.
• From medical records: sex, chronic health conditions and use of pain medication,
• Please refer to the section "outcome measures" for details regarding the PROMS.

In this study, multimorbidity is defined as the simultaneous presence of at least two chronic conditions within an individual. The percentage of patients meeting these criteria will be reported, along with a table detailing the frequency of each specific disease.

The patients referred to the department during the time period will be divided into a total of 5 groups.

Group 1) Patients who responded to the first invitation to participate Group 2) Patients who responded to the second invitation to participate Group 3) Patients who responded to the third invitation to participate

In addition, two groups are formed:

Group 4) Patients who dit not respond to any inviations to participate Group 5) an amalgamation of people from the three groups above

Statistical analyses:

Aim 1) To assess baseline differences between groups 1,2 and 3 one-way ANOVA with number of diseases as primary outcome and MTBQ and EQ-5D-5L as secondary outcomes will be employed.

Aim 2) To assess between-group differences at baseline, 3- and 6 months follow up a repeated measurement analysis (RM ANOVA) will be employed with number of diseases as primary outcome and MTBQ and EQ-5D-5L as secondary outcomes.

Additionally, patient characteristics across the three groups will be presented in a table, with appropriate statistical tests used to measure baseline differences (e.g., chi-square test for categorical variables). Results of the repeated measures ANOVA will be visually depicted using bar charts accompanied by error bars.

Imputation will be carried out for missing data if considered appropriate.

Additionally, chi-square test will be employed to assess any differences between group 4 and 5 in terms of gender distrubution. Unpaired t-test will be employed to assess any difference between group 4 and 5 in terms of age.