Changes and thresholds in the Oxford Shoulder Score following shoulder arthroplasty: Minimal clinically important difference,minimal important and detectable changes,and patient-acceptable symptom state

Abstract

Background

The minimal clinically important difference, minimal important change, minimal detectable change and patient-acceptable symptom state are poorly defined for the Oxford Shoulder Score following shoulder arthroplasty. The study's aim was to calculate their values.

Methods

One hundred patients underwent shoulder arthroplasty and completed pre and 1-year postoperative Oxford Shoulder Score. Patient satisfaction was assessed at 1-year using a visual analogue scale from 0 to 100: ‘very satisfied’ (>80), ‘satisfied’ (>60–80), and ‘unsatisfied’ (≤60). The difference between patients recording ‘unsatisfied’ (n = 11) and ‘satisfied’ (n = 16) was used to define the minimal clinically important difference. MIC_cohort was calculated as the change in Oxford Shoulder Score for those satisfied (>60). Receiver-operating characteristic curve analysis was used to determine the MIC_individual and patient-acceptable symptom state. Distribution-based methodology was used for the minimal detectable change.

Results

The minimal clinically important difference was 6.9 (95% confidence interval 0.7–13.1, p = 0.039). The MIC_cohort was 11.6 (95% confidence interval 6.8–16.4) and MIC_individual 13. The minimal detectable change was 6.6 and the patient-acceptable symptom state was defined as ≥29.

Discussion

The minimal clinically important difference and minimal important change can assess whether there is a clinical difference between two groups and whether a cohort/patient has had a meaningful change in their Oxford Shoulder Score, respectively. These were greater than measurement error (minimal detectable change), suggesting a real change. The patient-acceptable symptom state can be used as a marker of achieving satisfaction.

Keywords

Oxford Shoulder Score shoulder arthroplasty minimal clinically important difference minimal important change minimal detectable change patient-acceptable symptom state Level of evidence: level III diagnostic study.

Introduction

Shoulder arthroplasty (SA) is a cost-effective intervention for patients with end-stage glenohumeral osteoarthritis (OA) and other shoulder pathologies.¹ Recently, the number of SAs has increased exponentially,² with figures expected to rise even further following the coronavirus disease 2019 pandemic. Patient-reported outcome measures (PROMs) are commonly used in orthopaedic research to ascertain information from the patient's perspective. These have been shown to be reliable, valid and sensitive to clinical change.³ One such measure is the Oxford Shoulder Score (OSS), a 12-question PROM tool that focusses on shoulder pain and function.⁴

The OSS was first introduced by Dawson et al. in 1996 to assess patients undergoing shoulder operations other than stabilisation,⁴ and was revised in 2009 to give a total score out of 48 (with increasing scores indicating lower impairment).⁵ It has since been clinically validated against the Constant shoulder score and the SF-36,⁵ as well as in other languages.⁶ The OSS is popular given its simplicity, reliability and high internal consistency,⁴ but is sparsely reported in the literature for SA.⁷ A comparison of pre and postoperative scores can be used to look for improvement following surgery, but statistically significant changes may not necessarily represent clinical benefit from the patients’ perspective.⁸

The minimal clinically important difference (MCID) is a newly introduced concept that can help with the interpretation of these numerical PROM scales, to better reflect the clinical relevance of a treatment.^9,10 It is the minimal change in a scoring measure that is perceived by the patient to be beneficial (or harmful) relative to those who perceived to have not had postoperative clinical change.¹⁰ This is different to the minimal important change (MIC) which is the change in the scoring measure for a cohort (or individual patient) that perceive they have experienced an improvement in their outcome.¹¹ These definitions are often used interchangeably and can cause misunderstanding in the literature.¹¹ In addition to MCID and MIC, other psychometric measures include the minimal detectable change (MDC)¹¹ and patient-acceptable symptom state (PASS).¹² The MDC can be defined as the smallest change in an individual's score that is likely to be beyond the measurement error, and represents a true change.¹¹ The PASS is a threshold value beyond which patients consider themselves well and are satisfied.¹²

Only one study by Nyring et al.¹³ has reported an MCID value of 4.3 points for the OSS following SA, but the authors acknowledged that their sample size was small. The MIC and PASS remain undefined for the OSS in patients undergoing SA. Therefore, the aim of this study was to identify meaningful changes in the OSS (1-year following SA) by determining the MCID, MIC, MDC, and PASS in a cohort of patients having SA.

Methods

During a 5-year period (January 2016 to December 2020), 245 patients underwent SA at the study centre. SA was defined as any one of the following procedures: Anatomical (total), reverse (total), hemiarthroplasty (stemmed or resurfacing) or revision. Patients completing pre and 1-year postoperative functional questionnaires were identified from an established arthroplasty database held at the study centre. A retrospective analysis of this data was then performed.

As part of routine follow up, the patients had received postal questionnaires at designated intervals after their operation (at 6-month, 1-year and 2-year timepoints). These were sent with pre-paid envelopes so that the patients could return them to the unit. The study centre's research co-ordinators attempted to contact any patient by telephone who had not returned their questionnaire in time or had returned them incomplete, asking for more information or clarification of their answers before their next follow-up was due. If the patients were uncontactable, they were sent a repeat questionnaire in the post.

Outcomes measured

The OSS was recorded preoperatively and at 1-year postoperatively. The OSS consists of 12 questions assessed on a Likert scale with values from 0 to 4. A summative score is then calculated, where 48 is the best possible score (least symptomatic), and 0 is the worst possible score (most symptomatic).

Patient satisfaction was assessed by asking the patient ‘How do you rate your overall satisfaction with the outcome of your operation?’ at 1-year follow up. The response was recorded using a visual analogue scale (VAS) from zero (not satisfied) to 100 (very satisfied). To allow for anchor-based assessment of the OSS, the VAS was categorised into: ‘very satisfied’ (>80), ‘satisfied’ (>60 to 80), ‘neutral’ (≥40 to ≤60), ‘dissatisfied’ (20 to <40), ‘very dissatisfied’ (<20). These categories were arbitrarily defined based on a typical five-point Likert satisfaction scale. Patients who were ‘very satisfied’ or ‘satisfied’ were defined as ‘satisfied patients’. All other patients were defined as ‘unsatisfied’ and there were no differences in the change in OSS between these ‘neutral’, ‘dissatisfied’ or ‘very dissatisfied’ anchor groups (Table 1).

Table 1.

Differences in the mean change (95% CI) in the OSS according to the level of patient satisfaction 1-year after SA for the study cohort.

Versus	Level of satisfaction
Versus	Very satisfied	Satisfied	Neutral	Dissatisfied	Very dissatisfied
Very satisfied	-	12.4 (5.3 to 19.4) p < 0.001	22.0 (8.9 to 35.1) p < 0.001	14.5 (−3.8 to 32.8) p = 0.250	19.0 (7.2 to 30.8) p < 0.001
Satisfied	−12.4 (−19.4 to −5.3) p < 0.001	-	9.6 (−4.7 to 23.9) p = 0.557	2.1 (−17.0 to 21.3) p = 0.999	6.6 (−6.5 to 19.7) p = 0.999
Neutral	−22.0 (−35.1 to −8.9) p < 0.001	−9.6 (−23.9 to 4.7) p = 0.557	-	−7.5 (−29.6 to 14.6) p = 0.999	−3.0 (−20.1 to 14.1) p = 0.999
Dissatisfied	−14.5 (−32.8 to 3.8) p = 0.250	−2.1 (−21.3 to 17.0) p = 0.999	7.5 (−14.6 to 29.6) p = 0.999	-	4.5 (−16.9 to 25.9) p = 0.999
Very dissatisfied	−19.0 (−30.8 to −7.2) p < 0.001	−6.6 (−19.7 to 6.5) p = 0.999	3.0 (−14.1 to 20.1) p = 0.999	−4.5 (−25.9 to 16.9) p = 0.999	-

p-values are for an ANOVA with Bonferroni correction for multiple testing. OSS: Oxford Shoulder Score; ANOVA: analysis of variance; SA: shoulder arthroplasty.

Excluded patients

In total, 145 patients were excluded from the study as they had not completed preoperative (n = 80) or 1-year postoperative (n = 61) OSS or level of satisfaction for their SA (n = 4). All 80 patients with an incomplete preoperative OSS, had missing responses for all 12 questions. For those with an incomplete 1-year postoperative OSS, 59 had wholly missing responses, one had 7 missing responses (questions 6–12) and one had just one missing response (question 7). Although up to two missing responses can be accepted for the Oxford Hip and Knee Scores,¹⁴ no specific guide for the OSS exists. Nevertheless, this patient also did not complete a satisfaction score and had to be excluded. There was no significant difference in the sex (p = 0.694), age (p = 0.091), preoperative OSS (p = 0.616) or postoperative OSS (p = 0.292) between those fully completing (n = 100) and not completing (n = 145) their questionnaires.

Minimal clinically important difference

The MCID was primarily defined according to patient overall satisfaction. Using the anchor-based method, the MCID was defined as the difference in the mean OSS change between patients responding with ‘unsatisfied’ compared to those responding with ‘satisfied’ for level of satisfaction.

A secondary approach, a common distribution-based method, was used to estimate the MCID for comparison.^15,16 This was based on the standard deviation (SD) of the study cohort's change in OSS, divided by half. Thus, distribution-based estimate of MCID = 0.5 × SD (change in OSS).

Minimal important change

The MIC for a cohort was defined as the change in the OSS for ‘satisfied patients’. Receiver-operating characteristic (ROC) curve analysis was used to determine the MIC for an individual and was defined as the threshold value of change in the OSS that was predictive of patient satisfaction.

Minimal detectable change

A distribution methodology based on the standard error of measurement was used to calculate the MDC90. The ‘90’ indicates a 90% confidence interval (CI); that a change greater than this is real, and not due to intrinsic variability of the OSS. The standard error of measurement (the range in which a patient's true score lies) is the error associated with the measuring tool. The standard error of measurement was calculated using the SD for the change in the OSS (from the study cohort) and the reliability of the OSS: standard error of measurement = SD × √1-reliability. A previously established Cronbach's alpha of 0.92 for test-retest reliability for the OSS was used.⁴ The MDC was then calculated by multiplying the standard error of measurement by √2 (representing two separate occasions in which to measure change) and by a z value which represents the chosen CI. To establish the 90% CI of the OSS corresponds to a value of 1.65, hence: MDC90 = standard error of measurement × √2 × 1.65.

Patient-acceptable symptom state

The PASS was identified using ROC curve analysis to identify ‘satisfied patients’ from ‘unsatisfied’ patients according to their postoperative OSS.

Statistical analysis

Information on patient demographics, pre and postoperative OSS and patient satisfaction scores were extracted and tabulated in a database. Statistical Package for Social Sciences version 17.0 (SPSS Inc., Chicago, IL, USA) was then used for all data analysis. Data was assessed for normality and parametric tests where appropriate. Scalar variables were assessed using either unpaired Student's t-test, or one-way analysis of variance (ANOVA) with correction for multiple testing (Bonferroni) (Table 1). A Chi-square or Fishers exact test were used to assess sex and procedure differences between groups. ROC curve analysis was used to identify a threshold (point of maximal sensitivity and specificity) in the mean OSS change that was predictive of patient satisfaction for the change in the OSS (MIC) and the postoperative OSS (PASS). The area under the ROC curve (AUC) ranges from 0.5, indicating a test with no accuracy in distinguishing whether a patient is satisfied, to 1.0 where the test is perfectly accurate identifying all satisfied patients. Significance was set as a p-value of <0.05.

Ethics and approval

The project idea was conceived by members of the Academic Surgical Unit (ASU) at the study centre. The data used in the project had already been collected and stored on the institution's secure database system, which was available to all ASU clinicians. As no additional patient contact was required, the study was performed as a service evaluation without the need for formal ethical approval. However, the project required registration with the institution's audit department before data collection could commence. The collation of data was carried out in accordance with the GMC guidelines for good clinical practice and the Declaration of Helsinki.

Results

Study cohort characteristics

The study cohort consisted of 100 patients undergoing SA with complete pre and 1-year postoperative data that met the inclusion criteria. This included 26 (26%) male patients and 74 (74%) female patients, with an overall mean age of 73.5 (SD 7.9, range 47–89) years. All the operations had all taken place electively. More operations were for left-sided disease (58 patients). Six were revision cases. Of the 94 primary arthroplasty procedures, 48 were anatomical (total), 37 were reverse (total) and nine were hemiarthroplasty (four stemmed, five resurfacing) cases. The indications for each were for glenohumeral OA, cuff-tear arthropathy and sequalae of trauma accordingly. There were no significant differences in the preoperative or postoperative OSS between the primary and revision procedures (p > 0.327, Student's t-test), or according to implant design (p > 0.832, ANOVA), so these were amalgamated together for calculation and referred to as ‘SA’.

Overall

There was a greater improvement in the OSS with increasing level of patient satisfaction at 1-year, but this was not significantly different between ‘neutral’, ‘dissatisfied’ and ‘very dissatisfied’ patients (Table 1).

Minimal clinically important difference

There were 73 patients who were ‘very satisfied’, 16 patients who were ‘satisfied’ and 11 patients who were ‘unsatisfied’ (‘neutral’/‘dissatisfied’/‘very dissatisfied’). There was no significant difference in the patients’ demographics, procedure, or preoperative OSS between the ‘unsatisfied’ and ‘satisfied’ groups (Table 2). The MCID for the OSS (calculated as the difference in the mean OSS change between the ‘unsatisfied’ and ‘satisfied’ groups) was 6.9 (95% CI 0.7–13.1, p = 0.039). In contrast, the estimate of MCID based on the distribution approach was 5.6, as the SD for the change in OSS was 11.2. So, 0.5 × 11.2 = 5.6.

Table 2.

Patient demographics and pre-operative functional scores according to group.

Demographic	Descriptive	Satisfaction		Odds ratio/difference	95% CI		p-value*
Demographic	Descriptive	Neutral/dissatisfied/very dissatisfied (n = 11)	Satisfied (n = 16)	Odds ratio/difference	Lower	Upper	p-value*

Gender (M/F) (n, % of group)	Male	4 (51.6)	2 (44.6)	OR 4.00	0.58	27.41	0.187
Gender (M/F) (n, % of group)	Female	7 (48.4)	14 (55.4)	OR 4.00	0.58	27.41	0.187

Mean age (years: mean, SD)		72.3 (6.2)	72.3 (11.2)	0.0	-6.4	6.3	0.992**
Procedure	Primary	10	15	OR 0.67	0.37	11.9	0.658
	Revision	1	1
Oxford score	Pre-operative	15.6 (7.0)	17.9 (9.2)	Diff 2.3	-3.2	7.9	0.489**

*Fishers exact test unless otherwise stated, **unpaired t-test.

Minimal important change

There were 89 patients who were ‘satisfied’ or ‘very satisfied’ with their outcome. The mean change, pre to postoperative, in the OSS for these ‘satisfied patients’ was 11.6 (95% CI 6.8 to 16.4) and was defined as the MIC_cohort (Table 3). The MIC_individual was identified using ROC analysis to identify ‘satisfied patients’ from ‘unsatisfied’ patients according to change in their OSS. The change in the OSS was demonstrated to be a reliable and significant predictor of patient satisfaction with an AUC of 88.7% (95% CI 79.4–98.0, p < 0.001) (Figure 1). The maximal point of sensitivity and specificity for predicting satisfaction, which was 82%, corresponded to a change in the OSS of 13 points or more (Figure 2).

Table 3.

Mean preoperative and postoperative OSS and change in the score according to the patients’ level of satisfaction with their outcome 1-year following SA.

Level of satisfaction	Mean preoperative OSS (SD)	Mean postoperative OSS (SD)	Mean change in OSS (95% CI)	p-value*
Very satisfied (n = 73)	17.5 (7.8)	41.5 (7.3)	24.0 (22.9 to 26.0)	<0.001
Satisfied (n = 16)	17.9 (9.2)	29.6 (8.7)	11.6 (6.8 to 16.4)	<0.001
Neutral/dissatisfied/very dissatisfied (n = 11)	15.6 (7.0)	20.4 (9.5)	4.7-2.3 to 11.8	0.165

OSS: Oxford Shoulder Score; SA: shoulder arthroplasty

*Paired t-test.

MDC90

The SD of the mean postoperative OSS was 10 points, and the known test-retest reliability was defined as for the postoperative OSS to be 0.92 (Cronbach's alpha coefficient). The standard error of measurement was 2.8 (SD [10] multiplied by the square root of one minus the test-retest reliability [0.283]). The MDC90 was then calculated to be 6.6 points (SEM × 1.41 × 1.65) i.e., 90% of patients scoring more than this will have experienced a real change that is beyond measurement error.

Patient-acceptable symptom state

The PASS was identified using ROC curve analysis to identify ‘satisfied patients’ (n = 89) from ‘unsatisfied’ patients (n = 11) according to their postoperative OSS. The postoperative OSS was demonstrated to be a reliable and significant predictor of patient satisfaction with an AUC of 92.1% (95% CI 85.5–98.7, p < 0.001) (Figure 1). The maximal point of sensitivity and specificity for predicting satisfaction, which was 84%, corresponded to a postoperative OSS of 29 points or more (Figure 3).

Discussion

This study has defined the MCID, MIC, MDC and PASS for the English version of OSS after SA. The MCID for the OSS using the satisfaction anchor question was 6.9 points and 5.6 points when the distribution methodology was employed. MIC for a cohort of patients was 11.6 points and for an individual this was slightly greater at 13 points or more. The MDC90 of 6.6 points suggested that a value lower than this may have fallen within measurement error. A postoperative OSS of 29 points or more was identified as the PASS, which was predictive of patient satisfaction following SA.

The MCID for OSS has been described by only a small number of studies previously.^17,18 A systematic review by Jones et al.¹⁷ reported a mean value of 6 (range 5–6.9) from four studies, in patients with varying pathologies and treatment modalities. The current study's estimate of MCID was in keeping with such studies, but as MCID values have been shown to vary for conditions and treatment,^19,20 it is important to establish a better understanding of how this specifically relates to SA.

At present, Nyring et al.¹³ are the only authors reporting the MCID for OSS in patients having SA for glenohumeral OA. Their estimate for MCID (4.3) was lower than that in the current study. Their small sample size (n = 40) used for the anchor-based calculation may have accounted for such differences in the results. Furthermore, they only assessed patients who underwent anatomical total SA, which has been shown in the literature to be associated with significantly greater pre and postoperative outcome scores;^20,21 variables that can affect the MCID value obtained.

The MCID is useful to define as it provides an insight into clinical relevance, and so it can help with clinical decision-making.²⁰ Achievement of MCID has been shown to correlate with patient satisfaction as well.²² However, the several other methods that have been used to derive MCID values in the literature have caused confusion amongst clinicians as to which is the best to use.¹⁵ As the MCID is a patient-centred concept, anchor-based questioning may be preferable to other methods,¹⁰ as it compares the change in a PROM scale for a particular outcome, with another independent criterion that assesses for improvement (the anchor) e.g., a 5-point global rating scale.¹⁰ This demonstrates whether a patient has experienced clinical improvement or not. Such clinically relevant information cannot be captured through consensus or distribution methods alone,^10,23 but the MCID calculated in this way may be affected by the validity of the anchor used.¹⁰ Additionally, as shown by this study's distribution-based estimate for MCID, a discrepancy in values can occur when using a different method (5.6 versus 6.9). In such instance, the anchor-based value should be the one used, as it has been obtained from patient-centred data.¹⁰ As well as being clinically relevant, it was greater than the MDC90 (6.6) in this study and was therefore indicative of real change, whereas the distribution-based estimate of MCID was not and could have represented measurement error.

After subclassification into anchor groups, the MCID is derived through further analytical techniques.²⁰ Similar to Nyring et al.,¹³ this study used a mean difference method, calculating the difference in the mean change of OSS between the ‘satisfied’ and ‘unsatisfied’ groups. Although other techniques exist, this calculation method is good for comparing two transitional groups, and particularly useful in clinical trials involving intervention and control groups.²⁰

However, in contrast to Nyring, this study provided a different stance on the anchor question by focussing on patient satisfaction initially. Patients were asked to rate their overall satisfaction with the outcome of their operation and their response was recorded using a VAS from zero (not satisfied) to 100 (very satisfied). The VAS was then subclassified into anchor groups: ‘very satisfied’ to ‘very dissatisfied’. A similar way of converting a VAS score to anchor groups (for MCID calculation) has been previously reported by authors like Fan et al.²⁴ Although Likert scale anchor-based questions that focus on satisfaction exist,²⁵ the VAS is able to strongly reflect the patient's current state at the time of questioning, in comparison to their recalled preoperative state.²⁶ Furthermore, the VAS has been shown to be superior to Likert scales from a ceiling effect and construct validity perspective.²⁶ Therefore, this suggests that this method may be more suitable for anchor-based calculation and could help to improve the validity of the MCID value derived. On the other hand, neither the VAS or Likert scales have been deemed suitable to measure the amount of improvement over time.²⁶

Adopting different anchoring modalities for calculating the MCID has undoubtedly led to variability in the MCID values obtained across studies.²⁰ Although anchor-based questioning is seen as the best method to use, there are still limitations associated with the process. For example, the questions can be affected by recall bias where patients are unable to accurately remember their preoperative health status,²⁰ especially when using anchor questions that ask the respondent to assess their level of improvement in comparison to their preoperative state.^15,26 This may favour the use of a satisfaction-based anchor question, like the one used in this study instead. To add to the confusion, similar terms have also been used in the literature to express clinically significant changes in PROMs. These include the MIC and MDC. Such terms are related to the MCID but are in fact separate entities. It is important that clinicians are aware of these differences, so that there is more consistent reporting of these terminologies between studies.

Some authors have used the MIC to actually describe the MCID^27–29. The MIC differs from the MCID by only focussing on the clinically improved group or individual. Conversely, the MCID assesses whether an intervention has made a clinically important difference between two groups of patients, and therefore can be used to power studies.³⁰ Despite this, the MIC is still a useful tool to define; a change equal to or greater than the value represents a real change recognised by the group or individual. For cohort studies looking at a group of patients over time, this study demonstrated that a mean increase in OSS of 11.6 points would represent a clinically significant improvement in their symptoms after SA. However, a slightly higher increase of 13 points or more (absolute change) would be needed to show a clinically significant improvement for an individual patient. The ability of the MIC to identify changes that are clinically meaningful at an individual level is of particular benefit in clinical practice,²⁰ for example in the outpatient setting.

Similarly, the MDC (also known as the SDC- smallest detectable change) has been a term associated with both the MCID and MIC (or even considered as a distribution-based method for estimating MCID²⁰). The MDC is the smallest change in a PROM scale that can be detected when accounting for the measurement error. Due to this, it is a solely statistical phenomenon and like other distribution-based methods, cannot be used for detecting clinically meaningful changes alone.²³ Instead, the MDC should be used in conjunction with the MCID and MIC to test the reliability of the calculated values. An MCID or MIC value below the MDC may indicate a measurement error. In this study, the MDC90 for the OSS was calculated as 6.6 points. As the distribution-based estimate of MCID in this study (5.6 points) was below the MDC value, it could not be reliably used. This may have represented intrinsic error of the OSS and as previously discussed, it would not have held any clinical relevance regardless. However, the calculated values of MCID (6.9 points- using the anchor-based approach) and MIC (11.6 points- cohort and 13 points- individual) were greater than the MDC90, and so it was possible to say with 90% confidence that these changes in scores were beyond measurement error and constituted real measurable changes that were clinically relevant.

The MDC identified in the current study for OSS was comparable to that of Van Kampen et al.³¹ who calculated a value of 6.0 points. Their study differed by having established a 95% CI and their population having varying shoulder complaints. Nevertheless, their MCID score of 6.0 equalled their MDC and was therefore verified as a clinically important change. However, this was an example of the terms MCID and MIC being used equivocally in a study. Cross-cultural adaptation of the OSS has given a wide range of values for the MDC: 3.15 in Polish,³² 7.18 in Portuguese³³ and 18.8 in Persian,³⁴ but accompanying MIC and MCID values were not identified in these studies.

The PASS is a cut-off score beyond which patients consider themselves well and satisfied. This can also give an indication about the proportion of patients who achieved a clinical improvement.³⁵ Two approaches are used to identify the PASS – the 75th percentile method and ROC analysis, with the latter approach favoured and typically giving lower estimates.¹² In this study, the maximal point of sensitivity and specificity for predicting satisfaction was 84%, and this resulted in a PASS of 29 points or more. The only other authors to report a PASS for OSS were Christie et al.,³⁶ who defined this as 26 points, but this was based on the old scoring system for OSS. This now corresponds to a value of around 34 points, which is 5 points greater than the current study.³⁷ However, they only assessed patients with rheumatic diseases of shoulder. To the best of authors knowledge, this is the first study to identify the PASS for the OSS, specific to patients who have had SA.

There were several limitations to the current study that should be considered. The retrospective design meant that high numbers of patients were excluded (n = 145). However, there were no significant differences between the included and excluded groups in terms of sex, age, pre and postoperative OSS. This did not represent high loss to follow up for all patients, but rather the limited data available at 1-year follow-up. Similarly, only 11 patients were deemed ‘unsatisfied’. This is a small number, but one that can be expected in a cohort of 100 patients as arthroplasty satisfaction rates tend to be high.³⁸ Nevertheless, there were no significant differences between the ‘unsatisfied’ and ‘satisfied’ groups (Table 2). Furthermore, the use of a VAS presented a more favourable way of classifying anchor groups and may have given a more valid calculation of MCID.²⁶ However, further research is undoubtedly required to validate the findings in this study. A larger sample size would be needed to minimise any statistical fragility that may have been encountered from having just 11 ‘unsatisfied’ patients, and to allow for a greater number of them to be factored into the calculation.

The effect of combining different procedures (anatomical, reverse, hemi and revision) as ‘SA’ was another factor that should be taken into consideration. MCIDs vary for different treatment modalities,²⁰ and so calculated values for each subgroup procedure would have been preferred. However, this would have necessitated a longer study period to allow for enough numbers in each category. It is also difficult to account for all the influential factors in a cohort's characteristics, and so this may not be clinically relevant. Nevertheless, there were no significant differences in the preoperative or postoperative OSS between the primary and revision procedures or according to implant design. Although some would argue that a revision case is a different entity from a primary one, a study has shown little difference in calculated MCID values between revision and primary groups for other orthopaedic procedures.³⁹ Carreon et al. concluded that this may ‘simplify the interpretation of clinical improvement’.³⁹ Regardless, this study has given a baseline from which further work can be conducted to establish whether this is truly the case for all subgroup SA procedures in relation to the OSS.

Choosing the timepoint for follow-up was a further limitation of the study. Analysing just 1-year outcome and satisfaction scores only provided a snapshot assessment, which may have not reflected clinical states in the longer term. However, this is the commonest endpoint for most clinical studies. Clement et al. showed that no clinically important differences were observed with the Oxford knee scores between one and two years after total knee arthroplasty.⁴⁰ Additionally, the OSS has been shown to exhibit a ceiling effect at five years and three years, but not at six months following SA.⁴¹

Finally, it is also important to recognise the limitations that may arise from the interpretation of the values calculated in this study. The MCID, MIC, MDC and PASS values can only be generalisable to patients who have had SA, not other shoulder interventions. MCIDs are also commonly reported to one decimal place in the literature, as they are calculated mean values. However, this represents a false level of precision clinically for PROM scores like the OSS, as they are based on integer values only. Therefore, it would be reasonable to interpret this study's MCID as 7 (rounded up from 6.9) in the clinical setting. A change in OSS of 7 points is greater than the MDC (6.6), which suggests that this is beyond measurement error. This is also greater than the mean MCID value (6.9) calculated in the study, which suggests that the difference is clinically significant to a patient. A change in OSS of 6 points, on the other hand, would not be clinically relevant and possibly erroneous.

The PASS score of ≥29 points was calculated using ROC curve analysis, but an OSS of 29/48 at 1-year following SA does not appear to correspond to a satisfied state at face value. This may reflect how the PASS is a baseline-dependent tool.⁴² Those with less severe symptoms (higher OSS) preoperatively would require higher scores at 1-year follow-up to be in a satisfactory symptom state.⁴² The mean preoperative OSS for ‘satisfied patients’ was not too dissimilar to those who were ‘unsatisfied’ in this study (Table 3). As this is the first report of a PASS score for the OSS in SA, further studies would be needed to validate our findings, and once again larger scale datasets would be required to allow for greater baseline variation.

In conclusion, the estimates for MCID and MIC can be used to assess whether there is a clinical difference between two groups undergoing SA and whether a cohort/patient has had a meaningful change in their OSS, respectively. The MDC90 of 6.6 points suggests a value lower than this may fall within measurement error and supports the ability of the OSS to detect the defined MCID and MIC. The identified PASS threshold in the postoperative OSS can be used as a marker of achieving patient satisfaction following SA.

Footnotes

Contributorship

All persons designated as authors qualify for authorship as follows: PL contributed for the conceptualisation, methodology, data curation, writing – original draft and writing – review and editing. IA contributed for the resources, data curation and writing – review and editing. VA contributed for the supervision and writing – review and editing. NDC contributed for the methodology, software, validation, formal analysis, writing – original draft and writing – review and editing. VP contributed for the supervision, conceptualisation, methodology, resources and writing – review and editing. Each author has approved the final version of the manuscript.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. The paper is NOT based on a previous communication to a society or meeting

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Guarantor

ORCID iD

Perry Liu

References

Cregar

Beletsky

Cvetanovich

, et al. Cost-effectiveness analyses in shoulder arthroplasty: a critical review using the quality of health economic studies (QHES) instrument. J Shoulder Elbow Surg 2021; 30: 1007–1017.

Farley

Wilson

Daly

, et al. The incidence of shoulder arthroplasty: rise and future projections compared to hip and knee arthroplasty. JSES Open Access 2019; 3: 44.

Fitzpatrick

Fletcher

Gore

, et al. Quality of life measures in health care. I: Applications and issues in assessment. BMJ 1992; 305: 1074–1077.

Dawson

Fitzpatrick

Carr

Questionnaire on the perceptions of patients about shoulder surgery. J Bone Joint Surg Br 1996; 78: 593–600.

Dawson

Rogers

Fitzpatrick

, et al. The Oxford Shoulder Score revisited. Arch Orthop Trauma Surg 2009; 129: 119–123.

Frich

Noergaard

Brorson

Validation of the Danish version of Oxford shoulder score. Dan Med Bull 2011; 58: A4335.

Padua

de Girolamo

Grassi

, et al. Choosing patient-reported outcome measures for shoulder pathology. EFORT Open Rev 2021; 6: 779–787.

Cook

CE.

Clinimetrics corner: the minimal clinically important change score (MCID): a necessary pretense. J Man Manip Ther 2008; 16: E82–E83.

Copay

Subach

Glassman

, et al. Understanding the minimum clinically important difference: a review of concepts and methods. Spine J 2007; 7: 541–546.

10.

McGlothlin

Lewis

RJ.

Minimal clinically important difference: defining what really matters to patients. JAMA 2014; 312: 1342–1343.

11.

Terwee

Peipert

Chapman

, et al. Minimal important change (MIC): a conceptual clarification and systematic review of MIC estimates of PROMIS measures. Qual Life Res 2021; 30: 2729–2754.

12.

Kvien

Heiberg

Hagen

KB.

Minimal clinically important improvement/difference (MCII/MCID) and patient acceptable symptom state (PASS): what do these concepts mean?

Ann Rheum Dis 2007; 66: iii40–iii41.

13.

Nyring

MRK

Olsen

Amundsen

, et al. Minimal clinically important differences (MCID) for the western Ontario osteoarthritis of the shoulder index (WOOS) and the Oxford shoulder score (OSS). Patient Relat Outcome Meas 2021; 12: 299–306.

14.

Murray

Fitzpatrick

Rogers

, et al. The use of the Oxford hip and knee scores. J Bone Joint Surg Br 2007; 89: 1010–1014.

15.

Mouelhi

Jouve

Castelli

, et al. How is the minimal clinically important difference established in health-related quality of life instruments? Review of anchors and methods. Health Qual Life Outcomes 2020; 18: 36.

16.

Wright

Hannon

Hegedus

, et al. Clinimetrics corner: a closer look at the minimal clinically important difference (MCID). J Man Manip Ther 2012; 20: 160–166.

17.

Jones

Togashi

Heckmann

, et al. Minimal clinically important difference (MCID) for patient-reported shoulder outcomes. J Shoulder Elbow Surg 2020; 29: 1484–1492.

18.

Angst

Schwyzer

Aeschlimann

, et al. Measures of adult shoulder function: disabilities of the arm, shoulder, and hand questionnaire (DASH) and its short version (QuickDASH), shoulder pain and disability index (SPADI), American shoulder and elbow surgeons (ASES) society standardized shoulder. Arthritis Care Res 2011; 63: S174–S188.

19.

Holmgren

Öberg

Adolfsson

, et al. Minimal important changes in the constant-murley score in patients with subacromial pain. J Shoulder Elbow Surg 2014; 23: 1083–1090.

20.

Maredupaka

Meshram

Chatte

, et al. Minimal clinically important difference of commonly used patient-reported outcome measures in total knee arthroplasty: review of terminologies, methods and proposed values. Knee Surg Relat Res 2020; 32: 19.

21.

Flurin

Marczuk

Janout

, et al. Comparison of outcomes using anatomic and reverse total shoulder arthroplasty. Bull Hosp Jt Dis 2013; 71 : 101–107.

22.

Geoghegan

Mohan

Cha

EDK

, et al. Achievement of a minimum clinically important difference for back disability is a suitable predictor of patient satisfaction among lumbar fusion patients. World Neurosurg 2021; 152: e94–e100.

23.

Turner

Schünemann

Griffith

, et al. The minimal detectable change cannot reliably replace the minimal important difference. J Clin Epidemiol 2010; 63: 28–36.

24.

Fan

, et al. How much improvement can satisfy patients? Exploring patients’ satisfaction 3 years after total knee arthroplasty. J Orthop Surg 2021; 16: 89.

25.

Sun

Luo

, et al.

What constitutes a clinically important change in mayo elbow performance index and range of movement after open elbow arthrolysis?

Bone Jt J. 2021;103-B:366–372.

26.

Brokelman

RBG

Haverkamp

van Loon

, et al. The validation of the visual analogue scale for patient satisfaction after total hip arthroplasty. Eur Orthop Traumatol 2012; 3: 101–105.

27.

Sørensen

van Tulder

Johannsen

, et al. Responsiveness and minimal important change of the Oxford shoulder score, EQ-5D, and the fear-avoidance belief questionnaire physical activity subscale in patients undergoing arthroscopic subacromial decompression. JSES Int 2021; 5: 869–874.

28.

Ekeberg

Bautz-Holter

Tveitå

, et al. Agreement, reliability and validity in 3 shoulder questionnaires in patients with rotator cuff disease. BMC Musculoskelet Disord 2008; 9: 68.

29.

Christiansen

Frost

Falla

, et al. Responsiveness and minimal clinically important change: a comparison between 2 shoulder outcome measures. J Orthop Sports Phys Ther 2015; 45: 620–625.

30.

Chuang-Stein

Kirby

Hirsch

, et al. The role of the minimum clinically important difference and its impact on designing a trial. Pharm Stat 2011; 10: 250–256.

31.

van Kampen

Willems

van Beers

LWAH

, et al. Determination and comparison of the smallest detectable change (SDC) and the minimal important change (MIC) of four-shoulder patient-reported outcome measures (PROMs). J Orthop Surg 2013; 8: 40.

32.

Bejer

Szczepanik

Płocki

, et al. Translation, cross-cultural adaptation and validation of the Polish version of the Oxford shoulder score in patients undergoing arthroscopic rotator cuff repair. Health Qual Life Outcomes 2019; 17: 91.

33.

Gonçalves

Caldeira

Rodrigues

, et al. Cross-cultural adaptation and validation of the Portuguese version of the Oxford shoulder score (OSS). Acta Reumatol Port 2018; 43: 102–108.

34.

Naghdi

Nakhostin Ansari

Rustaie

, et al. Simple shoulder test and Oxford shoulder score: Persian translation and cross-cultural validation. Arch Orthop Trauma Surg 2015; 135: 1707–1718.

35.

Harris

Brand

Cote

, et al. Guidelines for proper reporting of clinical significance, including minimal clinically important difference, patient acceptable symptomatic state, substantial clinical benefit, and maximal outcome improvement. Arthrosc J Arthrosc Relat Surg 2023; 39: 145–150.

36.

Christie

Dagfinrud

Garratt

, et al. Identification of shoulder-specific patient acceptable symptom state in patients with rheumatic diseases undergoing shoulder surgery. J Hand Ther 2011; 24: 53–61.

37.

Marks

Grobet

Audigé

Validity, responsiveness and minimal important change of the EQ-5D-5L in patients after rotator cuff repair, shoulder arthroplasty or thumb carpometacarpal arthroplasty. Qual Life Res 2021; 30: 2973–2982.

38.

Puzzitiello

Nwachukwu

Agarwalla

, et al. Patient satisfaction after total shoulder arthroplasty. Orthopedics 2020; 43: e492–e497.

39.

Carreon

Bratcher

Canan

, et al. Differentiating minimum clinically important difference for primary and revision lumbar fusion surgeries: clinical article. J Neurosurg Spine 2013; 18: 102–106.

40.

Clement

Afzal

Demetriou

, et al. There is no clinically important difference in the Oxford knee scores between one and two years after total knee arthroplasty: the one-year score could be used as the benchmark timepoint to assess outcome. Knee 2020; 27: 1212–1218.

41.

Singh

Haque

Taub

, et al. Floor and ceiling effects in the Oxford shoulder score : An analysis from the national joint registry. Bone Jt J 2021; 103-B: 1717–1724.

42.

Ingelsrud

Terluin

Gromov

, et al.

Which Oxford knee score level represents a satisfactory symptom state after undergoing a total knee replacement?

Acta Orthop 2021; 92: 85–90.