Study design and setting
We studied prospectively gathered data on a consecutive cohort of patients that underwent surgical treatment of ulnar-sided wrist pathology between September 2017 and June 2020 at Xpert Clinics, The Netherlands. All surgeons at our institution are certified by the Federation of European Societies for Surgery of the Hand and/or fellowship trained.
After their first consultation with a hand surgeon, all our patients were invited to be included in the Hand and Wrist Cohort, a routine outcome measurement system. Upon agreement, they received secure web-based questionnaires using GemsTracker . Three reminders were sent to the patients for each round of questionnaires. The exact research setting of our study group has been reported previously .
We report this study using the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement . International Review Board approval was obtained from the ethics committee of the Erasmus Medical Center Rotterdam (NL/sl/MEC-2018–1088). All patients provided written informed consent for their data to be anonymously used in this study.
We identified adult patients in the Hand and Wrist Cohort scheduled for surgical treatment for ulnar-sided wrist pathology by the treatment codes: USO; TFCC reinsertion; pisiformectomy; u-head implant; and Adams procedure. Generally, conservative treatment was initiated by a short period of immobilisation, followed by a rigorous program of wrist exercises. After careful anamnesis, physical examination, and imaging (e.g. MRI, CT, or wrist arthroscopy), surgical treatment was considered if symptoms persisted for more than three months or if patients did not want further nonoperative management. Surgical treatment was directed to the type of pathological structure(s) focussing on the reconstruction of the anatomy . Exclusion criteria for this study were: 1) patients younger than 18 years; 2) patients with incomplete demographic and psychosocial data; 3) patients who underwent surgical procedures that were less prevalent than 30 in the dataset after applying exclusion criteria 1 and 2.
After the first consultation with a hand surgeon, sociodemographic characteristics were routinely collected. The variables evaluated in this study were age, sex, type of work, symptom duration, and whether the dominant side was affected.
The Pain Catastrophising Scale(PCS) was used to measure pain catastrophising [20, 21]. We evaluated the PCS total score, ranging from 0–52 (0 = no catastrophising behaviour; 52 = severe catastrophising behaviour). A PCS total score ≥ 30 is considered a clinically relevant level of catastrophising.
The Patient Health Questionnaire-4 (PHQ), a combination of the PHQ-2 and Generalised Anxiety Disorder (GAD)-2, measured psychological distress (anxiety and depression) . We evaluated the PHQ total score, ranging from 0–12 (0 = no psychological distress; 12 = severe psychological distress). A PHQ total score of 6–8 is considered a “yellow flag”, and ≥ 9 a “red flag”.
The Brief Illness Perception Questionnaire(B-IPQ) was used to measure how patients perceive their illness [23, 24]. As recommended , we evaluated the subscales (range 0–10) separately. For the items Personal control (how much control patients feel they have over their illness), Treatment control (how much patients think their treatment will help their illness), and Comprehension (how well patients understand their illness), a higher score is better. For all other items: Consequences (how much the illness affects the patients’ life), Timeline (how long patients expect their illness to last), Identity (how much patients contribute symptoms to their hand condition), Concern (How concerned patients are), and Emotion (how much the patients are emotionally affected by the illness) a lower score is better. The Credibility/Expectations Questionnaire (CEQ) measured the patients’ expectations and credibility of the treatment outcomes . It consists of 6 questions, and the scores range from 3–27 (3 = low expectations and credibility; 27 = high expectations and credibility). In this study, we only evaluated the Expectancy subscale. Due to collinearity concerns, the Treament control subscale was not evaluated. We used the validated Dutch language versions (DLV) for all questionnaires [21, 24, 26, 27].
The Patient Rated Wrist/Hand Evaluation-DLV (PRWHE) measured patient-reported pain and dysfunction before surgery and one year after surgery [28, 29]. The 1-year follow-up duration was chosen based on the alignment of clinical follow-up, the pathophysiology of the condition and expected treatment effect . This was in line with international recommendations . We evaluated the PRWHE total score, ranging from 0–100 (0 = no pain and dysfunction; 100 = severe pain and dysfunction). Outcomes from some of the included patients in this study have been evaluated before [31, 32].
For a fixed regression model with an R2 deviation from zero with an effect size F2 of 0.10, α of 0.05, 16 predictors and a sample size of 423, we reached a power of 99,5%. For the model at follow-up with the same effect size, α, 18 predictors, and 253 patients, we reached a power of 87,9%
We checked continuous data for normal distributions with histograms and quantile–quantile plots. Normally distributed data were displayed as mean values with standard deviations (SD), and skewed data were displayed as mean values with inter-quartile ranges (IQR).
We performed a hierarchical multivariable linear regression to investigate the relative contribution of different variables to the explained variance in the amount of pain and dysfunction before surgery (PRWHE total score; model 1) and after surgery (model 2). For model 1, we consecutively added sociodemographics, scheduled surgical procedure, PCS + PHQ scores, and B-IPQ scores. For model 2, we also stepwise added the CEQ expectation subscale and the preoperative PRHWE score. For model 3, the outcome was the change score in PRWHE before and after surgery. Similar to other studies, PCS and PHQ scores were added simultaneously but separately from B-IPQ since pain catastrophising and psychological distress have been studied more extensively [10, 12, 13]. Outcome expectations were only evaluated in the outcome models (model 2) . The variable treatment was added in the models as an instrumental variable for diagnosis/pathology.
Regression coefficients (B) with corresponding 95% confidence intervals and standardised coefficients (β) are reported for all variables. To illustrate the explained variance of different models, R2, adjusted R2, and significance of F change are reported per step. All linear regression model assumptions were checked and satisfied.
We calculated Pearson correlation coefficients and variance inflation factors (VIF) to investigate whether the correlation between psychosocial variables did not bias our estimates. We interpreted the Pearson correlation coefficients as suggested by Hinkle et al.  and VIF by Gareth .
Because data were collected during daily clinical practice, and participation in the routine outcome measurement was voluntary, missing data were expected. We tested for significant differences in demographic characteristics and, when available, preoperative questionnaire scores between patients who completed all questionnaires (complete responders) versus patients who completed none or only some (non/partial responders) using analysis of variance or Chi2 tests.
The sample size was calculated using GPower 3.1 . All other analyses were performed in software package R, version 3.6.1. For all tests, a p-value smaller than 0.05 was considered statistically significant.