Ethical approval and study design
Ethical approval for a large multi-centre prospective cohort study, the Singapore KneE osteoarthriTis CoHort (SKETCH) study, was granted by the National Healthcare Group Domain Specific Review Board (NHG DSRB) Singapore (Reference number: WHC/2020–00076). The SKETCH protocol has been published previously elsewhere23. This manuscript reports findings related to the OA-QI questionnaire and is written in alignment with the Strengthening the Reporting of Observational studies in Epidemiology (STROBE) guidelines24.
Participants and recruitment
Participants were recruited from two different public hospitals within the National Healthcare Group, Singapore, between June 2021 and February 2022. Specifically, recruitment occurred when people with knee osteoarthritis attended an orthopedic or physiotherapy appointment at either hospital. Participants were eligible if they were independent community mobilizers, met the National Institute for Health and Care Excellence (NICE) clinical diagnostic criteria for knee osteoarthritis6 and were conversant in English or Mandarin. Participants were excluded if they; i) had an alternate diagnosis of knee OA (e.g., referred pain from the hip or spine), ii) had other forms of knee arthritis (e.g., inflammatory, posttraumatic), iii) had moderate to severe cognitive impairment (e.g., dementia), iv) had received a previous knee arthroplasty, v) were wheelchair bound, vi) were pregnant or vii) had medical conditions that would medically interfere with rehabilitation involvement (e.g., decompensated heart failure, stroke with significant deficit, or end-stage renal failure). The participants were informed of the study directly at their orthopedic or physiotherapy appointment and had the option to engage immediately during their appointment visit or via a web-based form (FormSG) after their appointment at their own convenience.
Survey questions
After confirming eligibility and providing informed consent, the survey consisted of 3 sections: i) participant characteristics, ii) knee osteoarthritis symptoms and iii) the OA-QI questionnaire. First, the participants reported their age, height and weight, gender, ethnicity, occupational status (employed, unemployed, homemaker, retired), education level (no formal education, elementary school, high school, diploma holder, degree holder, others) and housing type (public housing, condominium, terrace house, others). The participants then reported the site of their symptomatic knee osteoarthritis (left/right/both) and how long they had a diagnosis of knee osteoarthritis. Joint-specific pain and function were then investigated via the Knee Osteoarthritis Outcome Scale-12 (KOOS-12)25. The KOOS-12 is frequently used in knee osteoarthritis research and consists of three sections of four items related to pain, function (activities of daily living and sport/recreation), and quality of life (QOL). The participants responded to each item via a five-point Likert scale (0 represents no knee problems, and 4 represents extreme knee problems). Section scores and total KOOS-12 scores were then calculated for each participant before being normalized on a 0–100 scale. Finally, the participants completed the 17-item OA-QI questionnaire18. The OA-QI questionnaire requires participants to respond to varying questions about their experience of care for their knee osteoarthritis, such as “yes”, “no” or “do not remember/not relevant”. The questions are broad and include topics related to referrals, clinical assessment, and treatments. The OA-QI is aligned with international clinical practice guidelines and has adequate content and construct validity in populations with knee osteoarthritis26,27.
Sample size estimation
The sample size was estimated with the objective of exploring the factors associated with the provision of more or less comprehensive care via G*Power 3.1.9.4. Given a type I error of 0.05 with 80% power of study, we needed data for at least 264 participants to detect a small effect size of Cohen’s f2= 0.03 in the multiple linear regression model.
Statistical analyses
Participant responses were deidentified, given a unique identifier number and stored on a secure web-based data management application (Research Electronic Data Capture [REDCAP]). The data were cleaned and analysed by one member of the research team (LCJ) via SPSS version 26.0. The participants’ weight and height were used to calculate their body mass index (BMI kg/m2). The BMI, age and KOOS-12 scores are presented as the means and standard deviations for each outcome. Other summarized participant demographic data are presented descriptively (frequency [n=] and proportion [%]).
For the OA-QI questionnaire, only ‘yes’/’no’ responses were considered to calculate the achievement rates (i.e., any responses of ‘Don’t remember’, ‘Not overweight’, ‘No such problems’, ‘No pain/discomfort’ or ‘Not severely troubled’ were excluded from analysis). Three types of achievement rates were then calculated; i.e., per item (i.e., the number of ‘yes’ responses/number of participants who responded ‘yes’ or ‘no’ for that particular OA-QI item) presented as % and 95% CI, ii. Overall mean per-item OA-QI achievement rate (calculating the mean achievement rate [% and 95% CI] for all 17 OA-QI items), and iii. The mean per-person achievement rate (number of items that an individual reported ‘yes’/number of ‘yes’ or ‘no’ responses they provided) is presented as the mean and 95% CI. The mean achievement rate per person provides a measure of the ‘comprehensiveness’ of treatment.
Linear regression modelling was used to explore the factors associated with the per-person and per-item OA-QI achievement rates. Simple linear regression was used to screen the variables, and variables with p<0.200 were included in the variable selection process via the stepwise variable selection method. The multicollinearity and interaction of the final model were checked, and heteroskedasticity was checked via the Bruesch-Pegan/Cook-Weisberg test.
On the other hand, multiple linear regression was used to explore the ability of the per-person OA-QI achievement rate and per-item OA-QI achievement rate to predict the 1-year change in functional outcomes, adjusting for baseline age, sex, BMI and KOOS total scores. All the tests were two sided, and statistical significance was denoted by p< 0.05.