A Novel Methodology for Establishing Best Values for MCID,SCB,and PASS Thresholds for Rotator Cuff Repair

Abstract

Background:

The heterogeneity of threshold values for the minimal clinically important difference (MCID), substantial clinical benefit (SCB), and Patient Acceptable Symptom State (PASS) as reported in the rotator cuff repair literature undermines the utility of these concepts. This systematic review identifies studies with published threshold values and proposes a methodologic framework for determining which values should be used for rotator cuff repair moving forward.

Purpose:

To provide recommendations for the MCID, SCB, and PASS thresholds of commonly utilized patient-reported outcome measures for rotator cuff repair, as well as recommendations for how these thresholds should be calculated moving forward.

Study Design:

Systematic review; Level of evidence, 3.

Methods:

All studies reporting MCID, SCB, and PASS threshold values after rotator cuff repair published between January 1, 2000, and May 31, 2022, were extracted via systematic review. The following data were collected: follow-up duration and patient attrition; reported threshold values; and data relevant to threshold calculation, including method, anchor questions and responses, area under the curve, and confidence intervals. The authors prioritized values calculated with an anchor question over those calculated without one, values from receiver operator characteristic analysis over those from mean change and logistic regression, and anchor questions with multiple response options over those with binary response options.

Results:

In total, 41 studies were included in the systematic review: 37 (90%), 11 (27%), and 16 (39%) reported MCID, SCB, and PASS thresholds, respectively. In addition, 12 studies calculated threshold values using anchor-based methods, and 6 calculated threshold values through distribution-based methods. The authors made recommendations for each threshold reported by at least 4 studies: for MCID, American Shoulder and Elbow Surgeons (ASES) = 21, visual analog scale for pain = 1.5, single assessment numeric evaluation (SANE) = 12, University of California at Los Angeles shoulder score = 6, and Constant-Murley score = 5.5; for SCB, ASES = 26 and SANE = 20; and for PASS, ASES = 78, visual analog scale for pain = 1.7, SANE = 71, and Constant-Murley score = 23.3.

Conclusion:

With standardized MCID, SCB, and PASS threshold values for rotator cuff repair surgery, these concepts hold enormous potential to power future comparative studies, guide reimbursement policy, and aid patient decision-making. Future research on novel MCID, SCB, and PASS threshold values should collect preoperative and 12-month postoperative patient-reported outcome measure data. Anchor questions should pertain to overall satisfaction with surgery and have multiple specific answer choices. These data should be correlated by receiver operator characteristic analysis, and any threshold values should then be compared with the standard error of the mean or minimal detectable change to ensure statistical significance.

Keywords

shoulder rotator cuff patient-reported outcome measures minimal clinically important difference substantial clinical benefit Patient Acceptable Symptom State threshold cutoff.

The collection and reporting of patient-reported outcome measure (PROMs) data have led to the proliferation of distinct threshold values for the minimal clinically important difference (MCID), substantial clinical benefit (SCB), and Patient Acceptable Symptom State (PASS) in the rotator cuff repair literature. These thresholds are used to ascribe clinical significance to improvements in PROM scores attributed to surgery or after surgery. However, there is no consensus regarding the most appropriate MCID, SCB, and PASS methodologies and threshold values for common PROMs after rotator cuff repair. The heterogeneity of reported threshold values frustrates efforts to make meaningful comparisons between studies and techniques. Underpinning this heterogeneity is the wide array of acceptable threshold value calculation methods, which results in significant variability in reported values, sometimes even for the same patient data.⁶

There are 2 sets of approaches to MCID, SCB, and PASS threshold calculation. Anchor-based approaches correlate PROM values with responses to an external question, called an anchor, to ascribe clinical significance.⁷ Distribution-based methods ascribe clinical significance purely from statistical characteristics of the cohort of patient data without correlation to an anchor question.¹⁹ The purpose of this systematic review was to evaluate the reported values for common MCID, SCB, and PASS thresholds and to propose a means for evaluation based on underlying methodology. We then present our recommended values for the MCID, SCB, and PASS thresholds for the Constant-Murley score (CMS), American Shoulder and Elbow Surgeons (ASES) score, visual analog scale for pain (P-VAS), Single Assessment Numeric Evaluation (SANE), and University of California at Los Angeles (UCLA) shoulder score.

Methods

Study Inclusion

This systematic review was performed in accordance with the PRISMA guidelines (Preferred Reporting Items for Systematic Reviews and Meta-analyses). The PubMed, Embase, Ovid, Cochrane, and Google Scholar databases were searched for articles published from January 1, 2000, to May 31, 2022, inclusive. Included studies reported at least 1 MCID, SCB, or PASS value for at least 1 PROM. For studies that reported on multiple shoulder pathologies or on operative and nonoperative management of rotator cuff tears, we included only those with unique MCID, SCB, or PASS threshold values for the cohort that underwent operative repair of rotator cuff tear. The required minimal follow-up for all patients in all studies was 12 months. Non-English, abstract-only, review, and editorial articles were excluded, as were case reports and cadaveric and animal studies. Studies reporting only minimal detectable change or standard error of measurement were not included in this analysis. Study screening was performed by 2 authors (A.C.L. and J.C.), with all disagreements settled by discussion with the senior author (R.L.P.).

Data Extraction

The type of study, country of investigation, and indications for rotator cuff repair were recorded into a predefined Microsoft Excel workbook. MCID, SCB, and PASS threshold values for each PROM and their calculation methods were also extracted. When a single study reported multiple values for the same threshold—whether at multiple time points, with multiple distinct calculation methods, or for distinct patient cohorts—all values were recorded. The methods used to calculate each threshold value were classified as anchor based, distribution based, or referencing a previous study. When a study calculated new threshold values, the exact methodology was also recorded. When available, anchor questions, their response options, and the options that corresponded to positive and negative anchors were extracted. For studies that referenced the existing literature, the referenced study cohort was recorded as “only rotator cuff repair patients” or “other.” Patient attrition, defined as the proportion of patients initially evaluated who were subsequently lost to follow-up, was recorded at each time interval for which thresholds were reported. When available, the area under the curve (AUC), the quantity of patients with positive and negative anchor responses, and the confidence intervals for individual thresholds were also recorded.

Methodology for Ranking Values

After data extraction, we developed a system based on methodology to determine which threshold values to recommend from the rotator cuff repair literature. This system was developed over multiple meetings among the authors and with reference to how MCID, SCB, and PASS thresholds are calculated and used in orthopaedic surgery and in other medical specialties (eg, pulmonology, oncology, clinical epidemiology).^{7,22,30,31,38,40} We recommended values for all MCID, SCB, and PASS thresholds reported by at least 4 studies. We propose the following set of value judgments, which are examined further in the Discussion section. We also excluded MCID, SCB, and PASS thresholds values calculated before 12-month follow-up.

MCID, SCB, and PASS thresholds calculated with an anchor question were of higher quality than those calculated without an anchor question or those referenced from the literature.

Thresholds calculated by receiver operator characteristic (ROC) analysis with an acceptable AUC (>0.7) were of higher quality than those calculated from other anchor-based methods, such as the mean improvement among patients with a positive anchor response (mean change method) or the difference in mean improvements for patients of each anchor response (linear regression method).

Thresholds calculated by ROC analysis without an acceptable AUC (<0.7) were the lowest quality among the thresholds calculated by anchor-based methods.

Anchor questions with >2 response options were superior to those with a binary response option.

Anchor questions that asked about overall satisfaction were superior to those that focused on a single aspect of recovery, such as pain or function.

Among thresholds calculated without an anchor question, those that used half of the effect size, where the effect size is equal to the standard deviation of the preoperative values, were considered superior to those that used other techniques, such as a different fraction of the effect size or another statistical characteristic of the patient cohort.

Results

Study Characteristics

Forty-one studies (6331 rotator cuff repairs) met the inclusion criteria: 37 (90%), 11 (27%), and 16 (39%) reported MCID, SCB, and PASS threshold values, respectively. Among all studies, the most commonly used PROMs were the ASES, CMS, P-VAS, SANE, and UCLA scores. Thirteen studies calculated novel MCID, SCB, or PASS threshold values using their own patient data. Study characteristics are found in Table 1. Of these studies, 9 prospectively included consecutive patients who underwent rotator cuff repair, while 4 included only those patients with PROM scores at final follow-up. Ten studies used their 12-month postoperative data to calculate MCID, SCB, and PASS thresholds; however, 2 of these 10 studies calculated thresholds at other time points for the same patient cohort. One of these 2 studies reported too few negative responses to the anchor question at 12 months to calculate a meaningful anchor-based MCID and recommended the values calculated from patient responses at 3 months.¹⁸ The second study cited MCID thresholds for the Oxford shoulder score, CMS, and UCLA score at 12 and 24 months and found them to be nearly identical for both anchor questions used.³⁹ Four studies used an anchor-based method and a distribution-based method to calculate MCID thresholds for the same patient data set, allowing for direct comparison between methods. Eleven studies reported novel MCID thresholds; 5, novel SCB thresholds; and 5, novel PASS thresholds.

Table 1

Characteristics of All Studies That Calculated New MCID, SCB, or PASS Threshold Values ^a

Study	No. of Eligible Patients	Follow-up, mo	Lost to Follow-up, %	Country	Evidence and Study Type	Metric	Score	Methods
Kukkonen (2013)¹⁸	802	3, 12	3	Finland	Basic science: validation of outcomes instrument	MCID	CMS	ROC analysis, mean change (positive anchor), effect size
Gagnier (2018)¹⁰	222	15	91	United States	Basic science: validation of outcomes instrument	MCID	ASES, WORC	Mean change (positive anchor)
Cvetanovich (2019)⁸	355	12	19	United States	Basic science: validation of outcomes instrument	MCID, SCB, PASS	ASES, CMS, SANE	ROC analysis, effect size
Gowd (2019)¹²	89	12	—	United States	Level 3: cohort study	MCID, SCB, PASS	ASES, CMS	ROC analysis
Xu (2020)³⁹	327	12	6	Singapore	Level 3: cohort study	MCID	CMS, OSS, UCLA	Linear regression
		24	32
Haunschild (2020)¹⁴	101	12	—	United States	Level 3: cohort study	SCB, PASS	PROMIS-UE	ROC analysis
	105					MCID		Effect size
Kim (2020)¹⁵	92	12	11	South Korea	Level 3: cohort study	PASS	ASES, P-VAS, SANE, UCLA	ROC analysis
						MCID, SCB		ROC analysis, mean change (positive anchor)
Tashjian (2020)³⁵	202	12	—	United States	Basic science: validation of outcomes instrument	MCID	ASES, P-VAS, SST	Mean change (positive anchor)
Marks (2021)²⁴	153	12	3	Switzerland	Level 3: cohort study	MCID	EQ-5D-5L	ROC analysisMean change (positive anchor)Effect size
Pagan-Conesa (2021)²⁸	110	12	17	Spain	Level 3: prospective therapeutic study	MCID	CMS, P-VAS	Mean change (all) minus 0.5 × SD change (all)
Malavolta (2022)²⁰	329	12	12	Brazil	Basic science: validation of outcomes instrument	MCID	ASES, UCLA	ROC analysis, effect size
Kim (2022)¹⁶	117	24	—	South Korea	Level 3: cohort study	PASS	ASES, P-VAS, SANE	ROC analysis
Tramer (2022)³⁷	198	18	15	United States	Level 3: cohort study	MCID, SCB	PROMIS-D, PROMIS-PI, PROMIS-UE	ROC analysis

Studies without “lost to follow-up” data defined their cohorts as only those who followed up at the denoted interval. ASES, American Shoulder and Elbow Surgeons; CMS, Constant-Murley score; EQ-5D-5L, EuroQol Group Health-Related Quality of Life Metric; MCID, minimal clinically important difference; OSS, Oxford shoulder score; PASS, Patient Acceptable Symptom State; PROMIS-D, Patient-Reported Outcomes Measurement Information System – Depression; PROMIS-PI, Patient-Reported Outcomes Measurement Information System – Pain Interference; PROMIS-UE, Patient-Reported Outcomes Measurement Information System – Upper Extremity Function; P-VAS, visual analog scale for pain; SANE, single assessment numeric evaluation; SCB, substantial clinical benefit; SST, simple shoulder test; UCLA, University of California at Los Angeles; WORC, Western Ontario Rotator Cuff.

MCID, SCB, and PASS Calculation Methods

Studies sometimes used several methods to calculate thresholds (Table 2). All novel SCB and PASS thresholds and some MCID thresholds were calculated via an anchor-based method, with a majority of the studies using an anchor-based method for at least 1 threshold calculation (92%; 12/13). Of the 12 studies that used an anchor-based method, 75% (9/12) correlated their clinical scores to anchor question responses using ROC analysis. All of these studies chose the threshold value that maximized the Youden index, defined as sensitivity plus specificity minus 1. Five studies used the mean change method, whereby the MCID or SCB threshold is set equal to the mean pre- to postoperative improvement among the patients with a positive anchor. One study utilized linear regression, whereby the mean score improvement was calculated for each response to the anchor question, and the MCID was set to the mean difference between adjacent response means.³⁹

Table 2

Methods Used to Calculate New MCID, SCB, or PASS Threshold Values Within Included Studies ^a

Methods	Description	Benefits	Drawbacks
Anchor based
ROC analysis	Creation of a curve with sensitivity and specificity of all possible thresholds plotted	Can choose the best threshold value. AUC metric assesses fit between PROM and anchor question. Applicable to MCID, SCB, and PASS.	Requires patients with positive and negative anchor responses for a high-quality result.
Mean change (positive anchor)	Mean pre- to postoperative improvement among patients with a positive anchor response	Easy to calculate; requires only positive anchor responses.	High risk of overestimation of MCID and SCB; cannot be applied to PASS.
Linear regression	Mean of the difference in pre- to postoperative improvement between patients with adjacent anchor responses (ie “very good” vs “excellent”)	Requires fewer patients than other anchor-based methods.	Result greatly influenced by the number of anchor response choices.
Distribution based
Effect size	Half the standard deviation of all preoperative patient scores.	Does not require an anchor question. For MCID, produces similar values to ROC analysis for many interventions.	Not applicable to SCB or PASS. No correlation with individual patient improvement.
Mean change variant	One half a standard deviation less than the mean improvement among all patients	Does not require an anchor question.	Large variation in MCID based on patient selection and abilities of surgeon.

AUC, area under the curve; MCID, minimal clinically important difference; PASS, Patient Acceptable Symptom State; PROM, patient-reported outcome measure; ROC, receiver operator characteristic; SCB, substantial clinical benefit.

Six studies used a distribution-based approach to calculate at least 1 MCID threshold. PASS and SCB thresholds were not calculated by distribution-based approaches. The most common threshold definition, used in some capacity by all 5 studies, was half the standard deviation of the preoperative scores. Kukkonen et al¹⁸ also reported an MCID value set to one-fifth the standard deviation of the preoperative scores. Another study defined the MCID of the Constant and P-VAS scores to half a standard deviation below the mean improvement for all patients in the cohort.²⁸

Four studies reported distinct values for the same MCID threshold using anchor- and distribution-based methods.^8,18,20,24 Of these 7 direct comparisons, 4 saw a higher value from the anchor-based method and 2 saw a higher value from the distribution-based method. One article reported 4 values total using 2 distinct anchor-based and 2 distinct distribution-based methods. The anchor-based values were higher and lower than the distribution-based values.¹⁸ Kukkonen et al¹⁸ and Marks et al²⁴ compared ROC analysis with the anchor-based mean change approach for MCID calculations. Both studies cited higher values using the mean change approach than ROC analysis.

The thresholds derived from “mean improvement” in score, whether of the patients with the positive anchor (anchor based) or of all patients (distribution based), were higher than those calculated by other methods. For example, the largest CMS MCID threshold value of 44.45 was derived from the mean pre- to postoperative improvement of all patients,²⁸ whereas other values ranged from 2 to 16.4. Similarly, the 2 highest ASES MCID threshold values, both calculated from the mean pre- to postoperative improvement among patients with a positive anchor response, were 39 and 27.1,^10,35 whereas the remaining values from all other methods ranged from 6.1 to 21.

Anchor Questions and Answers

All 6 studies that calculated a PASS value used a binary yes/no anchor question. Of 11 studies that reported an MCID or SCB value, a majority (82%; 9/11) used anchor questions with >2 answer choices (Table 3). The number of response choices ranged from 4 to 15, and all included neutral and positive response options; most also included negative response options. For studies using anchor-based methods to calculate MCID, 9 studies offered the patient response options on a Likert scale. For the 8 studies that defined a positive anchor, the positive anchor response or responses were a subset of those that were more positive than neutral but not the most strongly positive response. Xu et al³⁹ did not define a positive or negative anchor for their linear regression method. All 5 studies that used anchor questions to calculate SCB thresholds offered responses on a Likert scale and defined the positive anchor as the most strongly positive response option or options.

Table 3

Anchor Questions and Response Options for Each Study That Calculated an MCID, SCB, or PASS Threshold Using an Anchor-Based Method ^a

Study	Question	No. of Options	Response Options	Metric	Positive Anchor	Negative Anchor
Kukkonen (2013)¹⁸	Is the shoulder better or worse after the operation compared with the preoperative state?	2	Yes, no	MCID	Yes	No
Gagnier (2018)¹⁰	Were patients better, worse, or the same after treatment, with change quantified on a 5-point ordinal rating scale	11	–5 to 5	MCID	1 to 2	—
Cvetanovich (2019)⁸	Since your surgery, has there been any change in the pain in your shoulder?	15	–7 (great deal worse) to 7 (great deal better)	MCID	2 to 4	–1 to 1
				SCB	5 to 7	–1 to 1
	Taking into account all activities you have done during your daily life, your level of pain, and also your functional impairment, do you consider that your current state is satisfactory?	2	Yes, no	PASS	Yes	No
Gowd (2019)¹²	Since your surgery, has there been any change in the overall function of your shoulder?	15	–7 (great deal worse) to 7 (great deal better)	MCID	2 to 4	–1 to 1
				SCB	5 to 7	–1 to 1
	Taking into account all activities you have done during your daily life, your level of pain, and also your functional impairment, do you consider that your current state is satisfactory?	2	Yes, no	PASS	Yes	No
Xu (2020)³⁹	Has the surgery met your expectation so far?	7	1 (yes, totally) to 7 (no, not at all)	MCID	—	—
	How would you rate the overall results of your treatment?	6	1 (excellent) to 6 (terrible)
Haunschild (2020)¹⁴	Since your surgery, has there been any change in the overall function of your shoulder?	15	–7 (very great deal worse) to 7 (very great deal better)	SCB	5 to 7	–1 to 1
	Taking into account all activities you have done during your daily life, your level of pain, and also your functional impairment, do you consider that your current state is satisfactory?	2	Yes, no	PASS	Yes	No
Kim (2020)¹⁵	Since your shoulder rotator cuff repair, please rate your response to treatment	4	A (none) to D (excellent)	MCID	C	A or B
				SCB	D	A or B
	Are you satisfied with present symptoms and functional state?	2	Yes, no	PASS	Yes	No
Tashjian (2020)³⁵	Since your shoulder rotator cuff repair, please rate your response to treatment	4	A (none) to D (excellent)	MCID	C	—
Marks (2021)²⁴	Perceived change in quality of life at follow-up as compared with the baseline status	5	–2 (much worse) to 2 (much better)	MCID	1	–2, –1, 0
Malavolta (2022)²⁰	Taking into consideration all the activities of your daily life, pain level, and overall function, how do you consider your status in comparison to the preoperative period?	2	“Satisfied and better” or “dissatisfied and worse”	MCID	Satisfied and better	Dissatisfied and worse
Kim (2022)¹⁶	Are you satisfied with your current condition in terms of pain and function?	2	Yes, no	PASS	Yes	No
Tramer (2022)³⁷	Compared with before surgery, how has the functioning of your shoulder changed? Compared with before surgery, how much does pain in your shoulder impact your life today? Compared with before surgery, how has your mental health changed?	5	–1 (worse) to 3 (complete improvement)	MCID	2	–1 to 1
				SCB	3	–1 to 1

MCID, minimal clinically important difference; PASS, Patient Acceptable Symptom State; SCB, substantial clinical benefit.

Of the 19 anchor questions used, 7 inquired about general patient satisfaction with surgery without suggesting any specific criteria; 5 inquired about general patient satisfaction by asking for a combined evaluation of several criteria; and 8 asked for patient satisfaction with regard to a specific dimension of clinical function.

Recommended MCID, SCB, and PASS Threshold Values

Our recommended values can be found in Table 4. All recommended values were calculated by ROC analysis at 12-month follow-up, with the value chosen that maximized the Youden index. The range of values is among all 41 studies reporting MCID, SCB, and PASS thresholds.

Table 4

Recommended Values for Commonly Used MCID, SCB, and PASS Thresholds ^a

			Recommendation
	No. of Studies	Measurement Range	Value	Study	Method
MCID
ASES	26	0-100	21	Kim (2020)¹⁵	ROC
CMS	17	0-100	4.6	Cvetanovich (2019)⁸	ROC
P-VAS	9	0-10	1.5	Kim (2020)¹⁵	ROC
SANE	6	0-100	12	Kim (2020)¹⁵	ROC
UCLA	5	0-35	6	Kim (2020)¹⁵	ROC
SCB
ASES	9	0-100	26	Kim (2020)¹⁵	ROC
CMS	4	0-100	5.5	Cvetanovich (2019)⁸	ROC
SANE	4	0-100	20	Kim (2020)¹⁵	ROC
PASS
ASES	13	0-100	78	Kim (2020)¹⁵	ROC
SANE	6	0-100	71	Kim (2020)¹⁵	ROC
CMS	5	0-100	23.3	Cvetanovich (2019)⁸	ROC
P-VAS	5	0-10	1.7	Kim (2020)¹⁵	ROC

ASES, American Shoulder and Elbow Surgeons; CMS, Constant-Murley score; MCID, minimal clinically important difference; PASS, Patient Acceptable Symptom State; P-VAS, visual analog scale for pain; ROC, receiver operator characteristic; SANE, single assessment numeric evaluation; SCB, substantial clinical benefit; UCLA, University of California at Los Angeles.

Discussion

Calculation Methods

Among the studies, MCID, SCB, and PASS thresholds values were determined by correlation of PROM scores to a clinical question (anchor based) or through statistical characteristics of the scores (distribution based).^7,19 In our analysis, we recommended thresholds calculated from anchor-based methods over distribution-based methods because they correlated PROM scores, which represent a compilation of very specific functional and clinical outcomes, to an anchor question—a single question of broader clinical significance. Anchor-based methods are widely recommended over distribution-based methods when appropriate anchor questions and responses are available.^6,24,30,38 Distribution-based methods may be used when no prospectively collected anchor data are available or when the available anchor data are not relevant to the clinical outcome of interest.

Anchor-Based Methods

There was significant variability in how PROM scores are correlated to anchor responses in the rotator cuff literature. The most common method—and the one that we recommend—is ROC analysis.²⁵ In this method, a confusion matrix is calculated for each of a large number of randomly generated threshold values, with the positive and negative anchor responses considered the true result. The ROC curve is then created, which plots the true positive fraction (sensitivity) against the false positive fraction (1 – specificity), with each point representing these values for a distinct confusion matrix.¹³ From this graph, the MCID, SCB, or PASS value is typically the threshold that maximizes the Youden index, although some studies in the arthroplasty literature have used other values, such as the value with maximum sensitivity and at least 80% specificity or the value that maximizes the sensitivity and specificity while keeping them equal.^{5,11,23,29,32,41} We consider ROC analysis to be the gold standard because it allows for the determination of the maximally discriminatory MCID, SCB, or PASS threshold for the data set.

ROC analysis has limitations. The AUC signifies how well the PROM scores predict the response to the anchor question, with an AUC >0.7 considered acceptable and >0.8 considered excellent.²² Among studies, we prioritized other anchor-based calculation methods over ROC analyses with AUC <0.7 owing to the poor predictive capability of the PROM for the anchor question. Cvetanovich et al⁸ did not recommend the MCID SANE threshold that they calculated from ROC analysis for this reason. High-quality ROC curves also require an adequate number of positive and negative anchors to provide high-resolution sensitivity and specificity data. For surgery such as rotator cuff repair, a dearth of negative anchor responses can increase the uncertainty in the AUC value when calculating MCID and PASS thresholds. Tashjian et al³⁵ reported only 14 negative anchor responses from 202 total responses and opted to set their MCID to the mean improvement among patients with a positive anchor for this reason. Likewise, the 92% satisfaction rate at 12 months reduced the statistical power for Kukkonen et al,¹⁸ leading the authors to instead recommend their 3-month follow-up threshold. One potential solution is to calculate thresholds at shorter postoperative intervals, although satisfaction at short-term follow-up does not necessarily predict satisfaction at long-term follow-up.² Another solution is to set high standards of satisfaction for positive anchor responses, although this negates the purpose of the MCID threshold.

One method that avoids the need for negative anchor responses is the mean change method. The longitudinal mean change method is performed by calculating the mean or median improvement in PROM among the patients who indicate a positive response to the anchor question. The comparative mean change method sets the threshold of interest to the difference between the improvement of patients with a positive anchor response and the improvement of patients with a negative anchor response. In this review, several studies used the longitudinal mean change method to calculate the MCID and SCB. We found that thresholds derived from this method were consistently higher than those calculated by all other methods.^{10,15,18,24,35} We suggest that this may be due to anchor answer choices that were not sufficiently granular to distinguish a minimal improvement from a substantial improvement or better. Indeed, 1 of the studies utilizing the longitudinal mean change method used a binary yes/no anchor question while 3 of the other 4 had only 2 possible positive responses to the anchor question. As such, this method may overestimate the MCID value by classifying patients as “minimal clinically important difference” from their anchor response when they may in fact have improvement that is significantly greater than “minimal.” Pagan-Conesa et al²⁸ used a distribution-based variation of this method, in which the MCID was calculated from the mean improvement of all patients, minus half the standard deviation of the improvement. This also yielded a clinical threshold value significantly greater than the other studies.

The comparative mean change method was used to calculate an MCID threshold in only 1 study from 2013 but has since become a popular method to calculate the MCID in the total knee arthroplasty literature.^4,17,18,34 While this method provides useful information, we do not believe that it accurately represents the purpose of the MCID, which is to correlate improvements in PROM to global clinical improvement for the individual patient.

The final anchor-based method represented in our review was linear regression. Xu et al³⁹ posed the question “How would you rate the overall results of your treatment?” with 6 answer choices ranging from “excellent” to “terrible.” The mean pre- to postoperative MCID improvement was then pooled for each answer choice. The difference in improvement between adjacent answer choices (ie, between “excellent” and “very good”) was calculated and the mean reported as the MCID. Like the longitudinal mean change method, this method does not require many negative anchors to yield a precise result. It also requires a smaller patient cohort than ROC analysis because all patient data are used in the analysis. However, this method lacks relevance to individual patients (similar to the comparative mean change method) and is highly dependent on the number of anchor responses offered. Furthermore, this method assumes that the difference in improvement between adjacent anchor responses is roughly equal, which was not demonstrated. Xu et al found that the difference between the PROM improvement for “very good” and “excellent” was much greater than the difference between “fair” and “poor” or “terrible.”

Anchors are external indicators that attribute clinical meaning to changes in PROMs. In other fields of medicine, anchors may be clinical measurements, such as laboratory tests or vital signs, or physician evaluations.³¹ In the field of orthopaedic surgery, we commonly use anchor questions to assess improvements in pain, function, and global satisfaction with surgery. When choosing an anchor question, we recommend assessing overall patient satisfaction, as this implicitly evaluates multiple aspects of patient recovery with a single question. Malavolta et al²⁰ provide a high-quality example with their anchor question “Taking into consideration all the activities of your daily life, pain level and overall function, how do you consider your status in comparison to the preoperative period?” For anchor questions designed to calculate MCID and SCB thresholds, we recommend a minimum of 2 positive, 2 negative, and 1 neutral response. Binary yes/no anchor questions fail to distinguish patients with greater-than-minimal improvement from those with substantial improvement, which can overestimate the MCID and underestimate the SCB. Several studies^8,12,14 included a neutral anchor response, 7 positive responses, and 7 negative responses, ranging from “No change” to “A very great deal better.”

Distribution-Based Methods

The most common distribution-based approach was to define the MCID by the “effect size,” defined as a fraction of the standard deviation of preoperative values (preop SD).⁷ An MCID equal to 0.20 × preop SD has been proposed for “small” effects, 0.50 × preop SD for “moderate” effects, and 0.80 × preop SD for “large” effects.⁷ Across the orthopaedic literature, 0.50 × preop SD is most commonly used to calculate the MCID threshold. This value has been shown to be very similar to the MCID calculated by anchor-based methods for a broad range of interventions, in and beyond the field of orthopaedic surgery.²⁷ Among the 5 studies that defined MCID in terms of effect size, all used the 0.50 value. Kukkonen et al¹⁸ also reported the MCID derived from the 0.20 value. Four studies calculated MCID thresholds using ROC analysis and an effect size of 0.5 × preop SD for the same patient data. Three direct comparisons resulted in a higher value with the effect size calculation while 4 resulted in a higher value with the ROC analysis.^8,18,20,24 Because of the extensive empiric literature demonstrating similarities in the results from these 2 methods, an effect size of 0.5 × preop SD is the distribution-based method that we recommend for calculation of MCID thresholds. No distribution-based methods were used to calculate SCB or PASS thresholds among the studies. We have no recommendations regarding distribution-based methods to calculate SCB or PASS thresholds.

Any MCID, SCB, or PASS threshold must be greater than the measurement precision of the PROM to have clinical significance. Therefore, despite the limitations of distribution-based approaches to MCID, SCB, and PASS calculation, statistical analysis of patient data plays an important role in the calculation of new threshold values. There are 2 important concepts at play: the standard error of the mean (SEM) and the minimal detectable change (MDC). The SEM represents the variability in score attributed to the imprecision of a specific PROM and is equal to preop SD } $\sqrt{1 - r}$ , where r is the test-retest reliability coefficient for the PROM and exists on a continuum from −1 (perfectly inverse correlation) to 1 (perfect correlation).⁶ Any change less than the SEM is more likely due to random chance than an actual clinical improvement. The MDC represents the minimal difference between 2 scores that is statistically significant with 95% confidence. It is equal to $1.96 \times \sqrt{2} \times SEM$ . Of the studies, only 1 calculated an SEM or MDC value associated with its recommended threshold.²⁰ That study found the anchor-based MCID threshold exceeded the MDC value while the distribution-based value did not. While there are logistical challenges with calculating test-retest reliability postoperatively, test-retest reliability values have been proposed for most common PROMs used in the rotator cuff repair literature.^1,26,36 Future studies should incorporate either SEM or MDC when proposing new MCID, SCB, and PASS thresholds.

Follow-up

Of the 41 studies, 19 reported follow-up at a single time point (12 months), while 8 others reported PROM scores at 12 months and at least 1 subsequent time point. In 6 of these 8 studies, the postoperative scores leveled off at 12 months.^{3,9,21,33,39,42} Furthermore, 2 demonstrated that the incidence of large retears after rotator cuff repair decreases significantly after 12 months.^2,16 For these reasons, we endorsed the thresholds recommended by Kim et al¹⁵ at 12-month follow-up over those recommended by Kim et al¹⁶ at 24-month follow-up, despite the same anchor question and response choices, the same calculation method (ROC analysis), and very similar patient cohort.

Limitations

We identified articles to include in our systematic review using current terminology. The concept of the MCID is described by the phrase “minimal important difference” and “minimal important change” in older literature, when the terminology was not as well defined. For this reason, it is possible that articles that merited inclusion in our systematic review may have been inadvertently omitted. Our review did not exclude studies or threshold values based on the clinical indications for surgery. As such, some studies included only patients with full-thickness tears while others included full- and partial-thickness rotator cuff tears. Similarly, some studies included patients who underwent other indicated procedures, such as subacromial decompression or biceps tenodesis. This heterogeneity in indication and intervention may have contributed to the range of MCID, SCB, and PASS values in this review. Finally, the specific wording of individual anchor questions may bias PROM scores. With few exceptions, nearly every study that calculated its own threshold values used unique wording in the anchor questions. In the future, a standard set of anchor questions to assess each aspect of recovery may remove this source of variability.

Conclusion

This study proposes standardized MCID, SCB, and PASS threshold values for rotator cuff repair surgery, as well as a framework for assessing novel thresholds values and best practices for future studies. Future research on novel MCID, SCB, and PASS threshold values should collect preoperative and 12-month postoperative PROM data. Anchor questions should pertain to overall satisfaction with surgery and have multiple specific answer choices. These data should be correlated by ROC analysis, and any threshold values should then be compared with the SEM or MDC to ensure statistical significance.

Footnotes

Submitted March 19, 2024; accepted June 25, 2025.

The authors have declared that there are no conflicts of interest in the authorship and publication of this contribution. AOSSM checks author disclosures against the Open Payments Database (OPD). AOSSM has not conducted an independent investigation on the OPD and disclaims any liability or responsibility relating thereto.

ORCID iDs

Alexander C. Lee

Robert L. Parisien

An online CME course associated with this article is available for 1 AMA PRA Category 1 Credit™ at https://education.sportsmed.org/Public/Catalog/Home.aspx?CourseSearch=1&Criteria=9&Option=25. In accordance with the standards of the Accreditation Council for Continuing Medical Education (ACCME), it is the policy of The American Orthopaedic Society for Sports Medicine that authors, editors, and planners disclose to the learners all financial relationships during the past 12 months with any commercial interest (A ‘commercial interest’ is any entity producing, marketing, re-selling, or distributing health care goods or services consumed by, or used on, patients). Any and all disclosures are provided in the online journal CME area which is provided to all participants before they actually take the CME activity. In accordance with AOSSM policy, authors, editors, and planners’ participation in this educational activity will be predicated upon timely submission and review of AOSSM disclosure. Noncompliance will result in an author/editor or planner to be stricken from participating in this CME activity.

References

Blonna

Scelsi

Marini

, et al. Can we improve the reliability of the Constant-Murley score? J Shoulder Elbow Surg. 2012;21(1):4-12.

Chona

Lakomkin

Lott

, et al. The timing of retears after arthroscopic rotator cuff repair. J Shoulder Elbow Surg. 2017;26(11):2054-2059.

Ciccotti

Horan

Nolte

Elrick

Millett

PJ.

Outcomes after arthroscopic rotator cuff repair using margin convergence versus superior capsular reconstruction: should candidates for margin convergence be treated with superior capsular reconstruction?

Orthop J Sports Med. 2021;9(11):23259671211050624.

Clement

Weir

Deehan

Meaningful values in the Short Form Health Survey-36 after total knee arthroplasty—an alternative to the EuroQol five-dimension index as a measure for health-related quality of life: minimal clinically important difference, minimal important change, patient-acceptable symptom state thresholds, and responsiveness. Bone Joint Res. 2022;11(7):477-483.

Connelly

Galea

Rojanasopondist

, et al. Patient Acceptable Symptom State at 1 and 3 years after total knee arthroplasty: thresholds for the Knee injury and Osteoarthritis Outcome Score (KOOS). J Bone Joint Surg Am. 2019;101(11):995-1003.

Copay

Subach

Glassman

Polly

Schuler

TC.

Understanding the minimum clinically important difference: a review of concepts and methods. Spine J. 2007;7(5):541-546.

Crosby

Kolotkin

Williams

GR.

Defining clinically meaningful change in health-related quality of life. J Clin Epidemiol. 2003;56(5):395-407.

Cvetanovich

Gowd

Liu

, et al. Establishing clinically significant outcome after arthroscopic rotator cuff repair. J Shoulder Elbow Surg. 2019;28(5):939-948.

Dabija

Pennings

Archer

, et al. Which is the best outcome measure for rotator cuff tears? Clin Orthop Relat Res. 2019;477(8):1869-1878.

10.

Gagnier

Robbins

Bedi

Carpenter

Miller

BS.

Establishing minimally important differences for the American Shoulder and Elbow Surgeons score and the Western Ontario Rotator Cuff Index in patients with full-thickness rotator cuff tears. J Shoulder Elbow Surg. 2018;27(5):e160-e166.

11.

Galea

Florissi

Rojanasopondist

, et al. The Patient Acceptable Symptom State for the Harris Hip Score following total hip arthroplasty: validated thresholds at 3-month, 1-, 3-, 5-, and 7-year follow-up. J Arthroplasty. 2020;35(1):145-152.e142.

12.

Gowd

Cvetanovich

Liu

, et al. Preoperative mental health scores and achieving Patient Acceptable Symptom State are predictive of return to work after arthroscopic rotator cuff repair. Orthop J Sports Med. 2019;7(10):2325967119878415.

13.

Hajian-Tilaki

Receiver operating characteristic (ROC) curve analysis for medical diagnostic test evaluation. Caspian J Intern Med. 2013;4(2):627-635.

14.

Haunschild

Gilat

, et al. Establishing the minimal clinically important difference, patient acceptable symptomatic state, and substantial clinical benefit of the PROMIS Upper Extremity questionnaire after rotator cuff repair. Am J Sports Med. 2020;48(14):3439-3446.

15.

Kim

Kholinne

, et al. Minimal clinically important difference, substantial clinical benefit, and patient acceptable symptomatic state after arthroscopic rotator cuff repair. Am J Sports Med. 2020;48(11):2650-2659.

16.

Kim

Kholinne

, et al. Retear bigger than preoperative tear size would lead to treatment failure after rotator cuff repair. J Shoulder Elbow Surg. 2022;31(2):310-317.

17.

Kim

Koh

Choi

Seo

Minimal clinically important differences for patient-reported outcomes after TKA depend on central sensitization. J Bone Joint Surg Am. 2021;103(15):1374-1382.

18.

Kukkonen

Kauko

Vahlberg

Joukainen

Aarimaa

Investigating minimal clinically important difference for Constant score in patients undergoing rotator cuff surgery. J Shoulder Elbow Surg. 2013;22(12):1650-1655.

19.

Lydick

Epstein

RS.

Interpretation of quality of life changes. Qual Life Res. 1993;2(3):221-226.

20.

Malavolta

Yamamoto

Bussius

, et al. Establishing minimal clinically important difference for the UCLA and ASES scores after rotator cuff repair. Orthop Traumatol Surg Res. 2022;108(2):102894.

21.

Manderle

Gowd

Liu

, et al. Time required to achieve clinically significant outcomes after arthroscopic rotator cuff repair. Am J Sports Med. 2020;48(14):3447-3453.

22.

Mandrekar

JN.

Receiver operating characteristic curve in diagnostic test assessment. J Thorac Oncol. 2010;5(9):1315-1316.

23.

Maria Knoph

Ivar SØNbØ

Elisabeth

LIE

Tore Kristian

Identification of cutpoints for acceptable health status and important improvement in patient-reported outcomes, in rheumatoid arthritis, psoriatic arthritis, and ankylosing spondylitis. J Rheumatol. 2010;37(1):26.

24.

Marks

Grobet

Audige

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(10):2973-2982.

25.

Metz

CE.

Receiver operating characteristic analysis: a tool for the quantitative evaluation of observer performance and imaging systems. J Am Coll Radiol. 2006;3(6):413-422.

26.

Michener

McClure

Sennett

BJ.

American Shoulder and Elbow Surgeons standardized shoulder assessment form, patient self-report section: reliability, validity, and responsiveness. J Shoulder Elbow Surg. 2002;11(6):587-594.

27.

Norman

Sloan

Wyrwich

KW.

Interpretation of changes in health-related quality of life: the remarkable universality of half a standard deviation. Med Care. 2003;41(5):582-592.

28.

Pagan-Conesa

Garcia-Ortiz

Salmeron-Martinez

Moya-Martinez

Lopez-Prats

Diagnostic ultrasound shows reversal of supraspinatus muscle atrophy following arthroscopic rotator cuff repair. Arthroscopy. 2021;37(10):3039-3048.

29.

Paulsen

Roos

Pedersen

Overgaard

Minimal clinically important improvement (MCII) and patient-acceptable symptom state (PASS) in total hip arthroplasty (THA) patients 1 year postoperatively. Acta Orthop. 2014;85(1):39-48.

30.

Revicki

Hays

Cella

Sloan

Recommended methods for determining responsiveness and minimally important differences for patient-reported outcomes. J Clin Epidemiol. 2008;61(2):102-109.

31.

Revicki

Erickson

Sloan

, et al. Interpreting and reporting results based on patient-reported outcomes. Value Health. 2007;10(suppl 2):S116-S124.

32.

Ruopp

Perkins

Whitcomb

Schisterman

EF.

Youden Index and optimal cut-point estimated from observations affected by a lower limit of detection. Biom J. 2008;50(3):419-430.

33.

Schlegel

Abrams

Angelo

, et al. Isolated bioinductive repair of partial-thickness rotator cuff tears using a resorbable bovine collagen implant: two-year radiologic and clinical outcomes from a prospective multicenter study. J Shoulder Elbow Surg. 2021;30(8):1938-1948.

34.

Singh

Fiedler

Huang

, et al. Patient Acceptable Symptom State for the forgotten joint score in primary total knee arthroplasty. J Arthroplasty. 2022;37(8):1557-1561.

35.

Tashjian

Shin

Broschinsky

, et al. Minimal clinically important differences in the American Shoulder and Elbow Surgeons, simple shoulder test, and visual analog scale pain scores after arthroscopic rotator cuff repair. J Shoulder Elbow Surg. 2020;29(7):1406-1411.

36.

Thigpen

Shanley

Momaya

, et al. Validity and responsiveness of the single alpha-numeric evaluation for shoulder patients. Am J Sports Med. 2018;46(14):3480-3485.

37.

Tramer

Yedulla

Ziedas

, et al. Risk factors for failure to achieve minimal clinically important difference and significant clinical benefit in PROMIS computer adaptive test domains in patients undergoing rotator cuff repair. J Shoulder Elbow Surg. 2022; 31(7):1416-1425.

38.

Wells

Beaton

Shea

, et al. Minimal clinically important differences: review of methods. J Rheumatol. 2001;28(2):406-412.

39.

Chen

Lie

HME

Hao

Lie

DTT

. Minimal clinically important difference of Oxford, Constant, and UCLA shoulder score for arthroscopic rotator cuff repair. J Orthop. 2020;19:21-27.

40.

Yendluri

Alexanian

Lee

, et al. A novel methodology for establishing minimum clinically important difference and substantial clinical benefit thresholds for patient-reported outcome measures following reverse total shoulder arthroplasty. J Shoulder Elbow Surg. Published online March 13, 2025.

41.

Youden

WJ.

Index for rating diagnostic tests. Cancer. 1950;3(1):32-35.

42.

Zhang

Tan

Lie

DTT

. Outcomes of arthroscopic rotator cuff repair in stiff shoulders are comparable to non-stiff shoulders when combined with manipulation under anesthesia. Arthroscopy. 2020;36(12):2954-2961.