The study was conducted as part of FINNISH trial (Clinicaltrial no NCT01998048, 28/11/2013). Approval for conducting this study was obtained from the Hospital District of Southwest Finland (T145/2013, decision TO1/004/13, 16/09/2013). Written informed consent was obtained from all the patients willing to participate the study. We conducted the study according to the revised Declaration of Helsinki by The World Medical Association [10].
Translation process
A two-way translation of the original WOSI score was carried out as described by Beaton et al. [11]. Two native Finnish speaking shoulder surgeons and one professional Finnish translator translated the original English questionnaire to Finnish (stage I, forward translation). Then synthesis was made and possible discrepancies were resolved (stage II, synthesis). Equality of sense, rather than vocabulary, was given priority. Thereafter two native English speaking translators (one with specific expertise in medical terminology) back-translated the questionnaire to English (stage III, back translation). A consensus meeting with all five translators, a scoring methodologist, and a senior expert member was held, and the final translation was agreed on (stage IV, expert committee review). The whole process and all decisions were documented. Minor changes were made during the adaptation process to improve cultural relevance.
Pre-testing
The translated Finnish WOSI was adapted into an electronic user interface and pre-tested (stage V, pre-testing) by the first author (SE) for comprehensibility with 10 healthy medical and non-medical personnel volunteers at Turku University hospital.
Study population
Sixty two male patients undergoing stabilizing surgery for anterior shoulder instability were recruited at eight institutions in Finland during 2013–2017. The inclusion criteria for this trial were as follows: male patient under 25 years of age, clinically documented instability after traumatic shoulder dislocation and patient’s willingness for operative treatment. Exclusion criteria were as follows: non-congruency of the joint on imaging investigations, concomitant injuries affecting the shoulder, previous surgery of the ipsilateral shoulder, intellectual disability or disability to co-operate, patient’s denial. Patients were requested to answer the electronic WOSI twice, with a two-week interval just before elective surgery.
The patients were operated on by experienced shoulder surgeons with an arthroscopic Bankart method. Postoperatively the arm was immobilized in a simple sling for three weeks and thereafter physiotherapy training was commenced. The electronic WOSI was repeated at three months and one year postoperatively in addition to Subjective shoulder value (SSV) [12], Oxford shoulder instability score (OSIS) [13] and Constant score (CS) [14].
Subjective shoulder value (SSV) is a score where patient’s subjective shoulder assessment is expressed as a percentage of a totally normal shoulder (that would be 100%) [12]. Oxford shoulder instability score (OSIS) is a 12-item questionnaire where the questions cover different areas of life such as shoulder dislocations, trouble with dressing, worst or usual level of pain, avoidance of activities or prevented activities of importance, interference with work, social life, hobbies or lifting, avoided positions in bed at night or simply shoulder on your mind, adding up to 48 points in total representing the worst clinical result [13]. Constant score (CS) is a 100-point score for assessing the shoulder condition. Questionnaire’s 4 parameters are pain (15), activities of daily living (20), range of motion (40) and power (25) adding up 100 points in total representing the best clinical result [14]. An additional dichotomous anchor question: is your shoulder better or worse after surgery, was asked at three months. A flow chart for this study is presented in Fig. 1.
Statistical analysis
Reliability
The internal consistency of WOSI and subdomains was calculated using Cronbach’s Alpha ((p / (p – 1)) (1 – (∑σi2 / σX2)), where p = number of items in WOSI (here, p = 21), σi2 = variance of the ith item, and σX2 = variance of the observed WOSI total score), with a value above 0.8 regarded as excellent [15]. The reliability was tested comparing the test-retest results through Bland-Altman method and calculating the intraclass correlation coefficient (ICC) for the total score, and the four domains. ICC above 0.75 was regarded as excellent, and below 0.4 as poor [15]. The measurement error (standard error of measurement, SEM) was estimated by calculating minimal detectable change (MDC) with the following equations: SEM = SD × √ (1 – ICC(3,k)), where ICC(3,k) is equal to Cronbach’s Alpha [16] and MDC = SEM x 1.96 x √ 2, where 1.96 is derived from the 95% confidence interval and √2 from 2 measurements at preoperative timepoint [17].
Validity
The construct validity was assessed through calculating Pearson’s correlation coefficient for SSV, OSIS and CS. We hypothesized a high positive correlation between WOSI domains and SSV, and a negative correlation between WOSI domains and OSIS. A correlation above 0.6 was regarded excellent and correlation below 0.3 poor [15]. 80% agreement was anticipated to achieve sufficient construct validity.
Responsiveness
The clinical outcome data was analyzed using primarily the analysis of variance of repeated measurements preoperatively, at three months and one year postoperatively. The proportion of responses hitting the lowest or highest limit of WOSI exceeding 15% was regarded as a significant floor and ceiling effect [18]. Both effect size (mean change divided by baseline standard deviation, ES) and standardized response mean (mean change divided by standard deviation of the change, SRM) were calculated, and we anticipated a large effect size. The ability of WOSI to detect change after the intervention (internal responsiveness), was determined by pooled ES and SRM. In addition, the ability of WOSI to detect a clinically important change with reference to an external anchor (external responsiveness) was determined by area under the receiver-operating characteristic (ROC) curve using the trapezoidal method [19]. The triggering change was determined by maximizing Youden’s J (sensitivity + specificity – 1). We used the dichotomous anchor question at three months to calculate the minimal clinically important difference (MCID). The triggering change in the WOSI associated with the response in the anchor question was determined by analysis of (1) the mean change, and (2) receiver-operating characteristic (ROC) analysis [20].
All analyses were performed using R version 3.6.3 (R Foundation for Statistical Computing, Vienna, Austria).