The Effect of Trauma History and Symptom Duration on Repair Integrity in 2335 Consecutive Arthroscopic Rotator Cuff Repairs

Abstract

Background:

Few studies have assessed the impact of trauma history and preoperative symptom duration on cuff integrity after arthroscopic rotator cuff repair (RCR).

Purpose:

To assess the hypothesis that acute, traumatic rotator cuff tears are less likely to retear after arthroscopic RCR compared with chronic, atraumatic tears.

Study Design:

Cohort study; Level of evidence, 3.

Methods:

We conducted a post hoc analysis of prospectively collected data for 2335 consecutive patients who underwent primary arthroscopic RCR and were evaluated for retear on ultrasound 6 months postoperatively. A single-row knotless repair technique was used for all patients. The cohort was divided into patients who recalled a specific event that instigated their symptoms (“traumatic” group) and those who did not (“atraumatic” group). Chi-square test was utilized to assess the difference in retear rate between the traumatic and atraumatic groups. Multivariate logistic regression analyses were performed to identify independent predictors of retear, and receiver operating characteristic curve analysis was used to evaluate the accuracy of the regression equations.

Results:

The traumatic and atraumatic groups consisted of 1489 and 846 patients, respectively. There was no significant difference in retear rate between the traumatic and atraumatic groups (13% and 11%, respectively; P = .14). In the entire cohort, trauma history and preoperative symptom duration were not predictive of retear. In the traumatic group, larger tear size area was the strongest independent predictor of retear (area under the curve [AUC], 0.76; 99% CI, 0.70-81), followed by longer operative time (AUC, 0.69; 99% CI, 0.64-0.74), older patient age (AUC, 0.68; 99% CI, 0.63-0.73) and full-thickness tear (AUC, 0.66; 99% CI, 0.61-0.71). In the atraumatic group, larger tear size area was the strongest independent predictor (AUC, 0.76; 99% CI, 0.68-0.83), followed by older patient age (AUC, 0.67; 99% CI, 0.59-0.75) and full-thickness tear (AUC, 0.66; 99% CI, 0.58-0.73).

Conclusion:

Trauma history and preoperative symptom duration did not affect cuff integrity 6 months after arthroscopic RCR. More important factors associated with enhanced repair integrity included smaller tear size and younger patient age.

Retear of the rotator cuff is one of the main postoperative barriers to patient recovery after arthroscopic rotator cuff repair (RCR). ²¹ Recent reported rates of retear range from 8% to 19% at 6 months postoperatively.^17,25,27 The influence of trauma history and preoperative symptom duration on cuff integrity after arthroscopic RCR is unclear.

Abechain et al ² contended that deterioration of the rotator cuff is “essential” in the etiopathogenesis of “slow and progressive” rotator cuff tears that do not have clear traumatic causes. Thus, traumatic rotator cuff tears are “believed to have better healing potential” than atraumatic tears because there are supposedly “less degenerative changes of tendon.” ¹ However, few studies have tested this hypothesis. Cho and Rhee ⁵ conducted a cohort study on 55 shoulders with a history of trauma and 114 shoulders without and reported similar retear rates 6 months after arthroscopic RCR (18% and 25%, respectively; P = .98).

The effect of preoperative symptom duration on cuff integrity after RCR is also debated. In a retrospective study on 286 patients, Harada et al ¹¹ found no significant difference in the mean symptom duration between the intact group and retear group (9.3 months and 12.6 months, respectively; P = .08). Contrastingly, in a case series study on 282 patients, Kim and Kim ¹⁵ found that the retear rate was 9% among patients who experienced symptoms for <12 months preoperatively and 20% among those with symptoms lasting ≥12 months (P = .006).

In view of the conflicting findings surrounding trauma history and preoperative symptom duration, we aimed to design a study to (1) determine if there is a difference in retear rate after arthroscopic RCR between traumatic and atraumatic tears, (2) determine if trauma history and preoperative symptom duration are independent predictors of retear after arthroscopic RCR, and (3) identify other factors that predict retear for traumatic and atraumatic tears.

According to Loew et al, ²⁰ it is important to distinguish between traumatic rotator cuff lesions and degenerative lesions, because the “outcome of rotator cuff repair is better in traumatic than in degenerative” tears. Therefore, it is hypothesized that the retear rate at 6 months after arthroscopic RCR will be lower for traumatic tears compared with atraumatic tears.

According to Zumstein et al, ³⁰ “chronic [rotator cuff] tears show a rather unfavourable healing milieu,” leading to a “rather high failure rate.” Thus, it is hypothesized that longer preoperative symptom duration will be a predictor of retear 6 months after arthroscopic RCR.

Methods

Study Design

This was a post hoc cohort study using prospectively collected data about patients who underwent primary arthroscopic RCR and were assessed for retear 6 months postoperatively. Ethics approval was obtained from the South Eastern Sydney Local Health District Human Research Ethics Committee.

Inclusion and Exclusion Criteria

Patients who underwent primary arthroscopic RCR by the senior author (G.A.C.M.) and returned to the clinic 6 months postoperatively for ultrasonography were included. Patients were excluded if they had a (1) revision repair; (2) repair requiring a polytetrafluoroethylene patch; (3) isolated subscapularis repair; or (4) repair performed concomitantly with stabilization, capsular release, or resection for calcific tendonitis. Patients with severe muscular atrophy were not offered RCR. Biceps tenodesis and acromioplasty were occasionally performed concurrently with RCR, and these patients were not excluded from the investigation.

Initial Visit

At the initial preoperative visit, all patients completed a questionnaire that asked if they could recall a specific event that instigated their symptoms. The date on which the patient's symptoms began was recorded. After a shoulder examination was performed using a previously described protocol, ²⁴ an experienced sonographer confirmed the diagnosis of a rotator cuff tear by ultrasound using a Siemens ACUSON S2000 system.

Surgical Technique

Patients were administered an interscalene nerve block and were lightly sedated before being placed in the beach-chair position for arthroscopic RCR. For all patients, a single-row knotless technique was used to perform an undersurface and/or bursal repair of the torn rotator cuff tendon, as described previously. ²⁸ In summary, the rotator cuff tendon was pierced with a single inverted mattress suture using an Opus SmartStitch suturing device (ArthroCare). The ends of the mattress suture were passed through an Opus Magnum anchor (ArthroCare), which was subsequently positioned in an anchor hole on the greater tuberosity of the humerus. The TensionLock winding mechanism (ArthroCare) was utilized to tightly wind the sutures into the anchor and attach the tendon to the greater tuberosity of the humerus. This process was repeated if multiple anchors were needed. Partial-thickness tears were completed to full-thickness tears intraoperatively by the surgeon before repair. Cuff characteristics were assessed intraoperatively by the surgeon and rated on a 4-point scale (fair = 1, good = 2, very good = 3, excellent = 4).

Rehabilitation

Patients were discharged on the same day of surgery, with an UltraSling (DJO) and a small abduction pillow supporting the surgical arm. Pendulum exercises were performed in the first 8 days after surgery. Passive shoulder movements were performed from 8 days to 4 weeks postoperatively. At 6 weeks, the sling was discontinued and patients commenced isometric strengthening exercises and active shoulder movement. After 3 months, patients commenced TheraBand exercises and were permitted to perform overhead activities. Patients were usually permitted to return to full work duties and sport 6 months postoperatively.

Postoperative Evaluation

All patients included in this study attended a follow-up clinic 6 months after surgery to assess the integrity of the repaired tendon on the ACUSON S2000 ultrasound system. The same experienced sonographer who performed the preoperative ultrasound also performed the postoperative ultrasound. The sonographer was not blinded to the patient's clinical history. Retear was defined as a full-thickness or partial-thickness defect in the tendon. Partial-thickness tears are defects that do not penetrate the entire thickness of the tendon, leaving a portion of tendon fibers intact and attached to the greater tuberosity of the humerus.

According to a meta-analysis by Gyftopoulos et al, ¹⁰ ultrasound and magnetic resonance imaging (MRI) have comparable diagnostic accuracy for the detection of both partial- and full-thickness retears after RCR, with a mean sensitivity and specificity of 81% and 83% for MRI compared with a mean sensitivity and specificity of 84% and 91% for ultrasound. However, ultrasound is more cost-efficient and easily accessible.

Post Hoc Analysis

The study cohort was divided into a traumatic and an atraumatic group. If patients responded “yes” and were able to recall a traumatic event when asked, “Do you recall a specific event that instigated your symptoms?” on the preoperative questionnaire, they were allocated to the traumatic group. Those who responded “no” were allocated to the atraumatic group. This methodology is consistent with previous studies related to this topic. A recent cohort study by Baum et al ⁴ categorized patients’ tears as either traumatic or degenerative based on the patient's medical history; those who recalled a sudden onset of symptoms after a traumatic event and did not have previous shoulder surgery were allocated to the traumatic group, while patients with chronic shoulder pain and no recountable history of trauma or surgery were allocated to the degenerative group. Allocating patients to a traumatic or atraumatic group based on the patient's clinical history was also performed in other studies by Cho and Rhee, ⁵ Paul et al, ²³ and Tan et al. ²⁶

Preoperative symptom duration was defined as the time that had elapsed from the date of symptom onset (in the case of atraumatic tears) or the date of the initiating event (in the case of traumatic tears) to the date of RCR.

We also evaluated the effect of other variables such as patient age, tear size, operative time, public/private hospital setting, and use of workers’ compensation on cuff integrity to create predictive models for retear in both the traumatic and the atraumatic groups, in alignment with the third aim of our investigation. In Australia, public hospitals are owned and managed by state governments and funded by Australian taxpayer contributions. Private hospitals are managed by private organizations and require private health insurance or out-of-pocket payments for treatment.

Statistical Analysis

Chi-square test was used to assess the difference in retear rate between the traumatic and atraumatic groups. Fisher’s exact test was used to determine if there were significant differences in retear rates between groups separated according to preoperative symptoms. A 2-way analysis of variance (ANOVA) was also performed to determine if there was a significant interaction effect between trauma history and symptom duration on retear rate. Spearman correlation was used to determine factors associated with retear in the entire cohort and in the traumatic and atraumatic groups. Multivariate logistic regression analyses were then performed to identify independent predictors of retear. Receiver operating characteristic (ROC) curve analysis and Youden J statistic were used to evaluate the accuracy of the regression equations and to determine threshold values for independent predictors of retear in the traumatic and atraumatic groups.

Results

Cohort

During the study period of January 2005 to December 2022, a single surgeon (G.A.C.M.) performed 4042 arthroscopic RCRs. Of these, 1050 were excluded because concomitant procedures were performed at the time of surgery (stabilization, capsular release, resection for calcific tendonitis), 273 were excluded because they were revision repairs, 142 were excluded because a polytetrafluoroethylene patch was utilized, and 28 were excluded as isolated subscapularis tears. Of the remaining 2549 patients, 214 were excluded because they did not attend their 6-month follow-up appointment to assess cuff integrity (yielding a follow-up compliance rate of 91.6%). After accounting for these exclusions, 2335 patients were included in the study cohort.

Patient Demographics

The patient demographics for the traumatic and atraumatic groups are summarized in Table 1. A total of 1489 patients recalled a specific event that instigated their symptoms (the traumatic group) while 846 patients did not recall such an event (the atraumatic group). On average, those in the traumatic group were younger, had larger tears, and experienced a shorter duration of symptoms preoperatively. The traumatic group also contained a significantly higher portion of patients with full-thickness tears, patients who underwent RCR in a private hospital, and patients who claimed workers’ compensation. Intraoperatively, there were no significant differences in surgeon-ranked tissue quality, tissue mobility, or repair quality between the traumatic and atraumatic groups. Falling was the most common mechanism of injury reported in the traumatic group (n = 602; 40%), followed by lifting or moving objects (n = 228; 15%) and motor vehicle accidents (n = 108; 7%).

Table 1.

Patient Demographics (N = 2335) ^a

Variable	Traumatic	Atraumatic	P
Patients, n	1489	846
Age, y	58.0 ± 0.3	61.6 ± 0.4	<.001
Preoperative symptom duration, mo	22 ± 1.4	28 ± 1.9	.01
Tear size area, mm2	387 ± 14	290 ± 11	<.001
Operative time, min	21 ± 0.3	19 ± 0.4	.003
Tear type, n (%) b			<.001
Full thickness	919 (61.7)	449 (53.1)
Partial thickness	536 (36.0)	372 (44.0)
Cuff characteristics c
Tissue quality	3.2 ± 0.02	3.2 ± 0.03	.79
Tissue mobility	3.4 ± 0.02	3.4 ± 0.03	.16
Repair quality	3.5 ± 0.02	3.5 ± 0.02	.46
Hospital type, n (%)			<.001
Private	1392 (93.5)	769 (90.9)
Public	97 (6.5)	77 (9.1)
Workers’ compensation, n (%)	220 (14.8)	38 (4.5)	<.001

^aData are presented as mean ± SEM, unless otherwise indicated. Statistically significant if P < .05. Unpaired Student t tests were used for parametric data including patient age, tear size, operative time, and preoperative symptom duration. A Mann-Whitney U test was used for cuff characteristics. Chi-square tests were used for categorical variables including tear type, hospital setting, and workers’ compensation.

^bSome patients had missing data for this variable.

^cCuff characteristics were assessed intraoperatively by the surgeon and rated on a 4-point scale (fair = 1, good = 2, very good = 3, excellent = 4).

Trauma History and Retear

There was no significant difference in retear rate between the traumatic and atraumatic groups (Table 2). However, the mean retear size among patients who retore their rotator cuff after RCR was higher in the traumatic group (P = .02) (Table 2). The mean retear size was similar to the mean initial tear size (Table 1) for both the traumatic and the atraumatic groups.

Table 2.

Retear Rate and Retear Size in Traumatic and Atraumatic Groups (N = 2335) ^a

Variable	Traumatic (n = 1489)	Atraumatic (n = 846)	P
Retear cases	195 (13.1)	93 (11.0)	.14
Retear size, mm2	365.6 ± 24.9	281.9 ± 26.2	.02

^aData are presented as mean ± SEM or n (%). A chi-square test was used to compare the retear rates in the traumatic and atraumatic groups. An unpaired Student t test was used to compare retear size. Statistically significant if P < .05.

Of the 93 retears in the atraumatic group, 1 was partial thickness. Of the 195 retears in the traumatic group, 4 were partial thickness.

Preoperative Symptom Duration and Retear

In both the traumatic and the atraumatic groups, there was no consistent trend that summarized the relationship between symptom duration and retear rate (Table 3 and Figure 1).

Table 3.

Number of Patients With Intact and Retorn Rotator Cuffs in Groups Separated According to Preoperative Symptom Duration (n = 2129) ^a

Symptom Duration, mo	Traumatic		Atraumatic
	Intact	Retorn	Intact	Retorn
≤1	214	37	41	5
>1 but ≤3	255	39	94	11
>3 but ≤6	175	31	128	16
>6 but ≤12	165	12	115	13
>12 but ≤24	142	24	131	7
>24	239	32	176	27

^aSome patients did not write down the date symptoms began on the questionnaire at the initial preoperative visit. Due to these missing data, the number of patients included in this table is slightly less than the total study sample size of 2335 patients.

Figure 1.

Retear rate in groups separated according to preoperative symptom duration for traumatic and atraumatic tears (n = 2129). Fisher exact tests were used to compare retear rates between all groups. Some patients did not write down the date symptoms began on the questionnaire at the initial preoperative visit. Due to these missing data, the number of patients included in this figure is slightly less than the total study sample size of 2335 patients. *P < .05. P < .05 indicates significance.

A 2-way ANOVA was also performed and confirmed that neither trauma history nor symptom duration had a significant effect on retear rate (P = .89 and P = .67, respectively), and there was no significant interaction effect between trauma history and symptom duration on retear rate (P = .61) (Table 4).

Table 4.

Results of 2-Way ANOVA With Retear Rate as the Dependent Variable ^a

Source	Type 3 Sum of Squares	df	Mean Square	F	P
Corrected model	28.958 b	274	0.106	1.006	.46
Intercept	6.117	1	6.117	58.238	.00
Trauma history	0.002	1	0.002	0.019	.89
Symptom duration	18.338	184	0.100	0.949	.67
Symptom duration * trauma history	8.897	89	0.100	0.952	.61
Error	194.739	1854	0.105
Total	254.000	2129
Corrected total	223.697	2128

^aF value represents the ratio of the between-group variability and the within-group variability. P values are statistically significant if P < .05. df, degrees of freedom. *indicates the interaction effect between symptom duration and trauma history.

^bR ² = 0.129 (adjusted R² = 0.001).

Multivariate Logistic Regression Analysis

In the entire cohort, trauma history and preoperative symptom duration were not predictive of retear (Table 5).

Table 5.

Independent Predictors of Retear Ranked by Strength of Predictive Value (Wald Statistic) From Multivariate Logistic Regression Analysis ^a

Predictor	Wald Statistic	P b	More Likely to Retear if
Entire cohort (N = 2335)
Patient age	32.3	<.001	Older patient
Full/partial	25.8	<.001	Full-thickness tear
Tear size area	23.3	<.001	Larger tear size area
Public/private	21.1	<.001	Public hospital
Case number	10.7	.001	Lower case number
Operative time	6.5	.01	Longer operative time
Workers’ compensation	5.2	.02	Workers’ compensation claimed
Traumatic/atraumatic c	0.9	.33	NS
Preoperative symptom duration c	0.4	.60	NS
Traumatic group (n = 1489)
Patient age	31.7	<.001	Older patient
Public/private	30.3	<.001	Public hospital
Full/partial	25.9	<.001	Full-thickness tear
Tear size area	19.6	<.001	Larger tear size area
Operative time	11.7	.001	Longer operative time
Case number	7.3	.007	Lower case number
Workers’ compensation	6.6	.01	Workers’ compensation claimed
Preoperative symptom duration c	0.2	.63	NS
Atraumatic group (n = 846)
Tear size area	17.7	<.001	Larger tear size area
Patient age	11.4	.001	Older patient age
Case number	6.4	.01	Lower case number
Full/partial	6.1	.01	Full-thickness tear
Preoperative symptom duration c	0.6	.43	NS

^aNS, not significant.

^bP values are derived from multivariate logistic regression analysis and are statistically significant if P < .05.

^cAlthough not found to be independent predictors of retear in multivariate logistic regression analysis, data shown are relevant to the aims of this study.

In the traumatic group, older patient age, surgeries performed in a public hospital, larger tear size area, longer operative time, lower case number (ie, surgeries performed earlier in the surgeon's career), and cases involving workers’ compensation were all independent predictors of retear, while preoperative symptom duration was not (Table 5). The retear rate declined as surgeon experience increased, peaking at 30% after approximately 600 traumatic RCR cases but gradually decreasing since then (despite a slight rise in recent cases) (Figure 2).

Figure 2.

Moving mean graph displaying the relationship between case number and retear rate in the traumatic and atraumatic groups (N = 2335).

In the atraumatic group, larger tear size, older patient age, lower case number, and full-thickness tears were independent predictors of retear in the atraumatic group, while preoperative symptom duration was not. The retear rate declined as case number increased, peaking at approximately 26% after 600 atraumatic RCR cases but gradually decreasing since then (Figure 2).

Clinical Application of Multivariate Regression Analysis

Using the predictors identified from the multivariate regression analyses (Table 5), regression equations were formulated to calculate an individual's risk of retear 6 months after arthroscopic RCR. Equations 1 and 2 can be used for traumatic and atraumatic tears, respectively. Both models were excellent fits for the raw data in this study, with likelihood ratio test statistics of 216 and 74 for equations 1 and 2, respectively (P < .001).

Equation 1 (Traumatic Tears)

Let p be the probability of retear 6 months after arthroscopic RCR.

$$\mathrm{Logit}p=\mathrm{lo}{g}_e\left(\frac{p}{1-p}\right)=(-0.000403\times \mathrm{case}\mathrm{number})+(0.0541\times \mathrm{age}\mathrm{in}\mathrm{years})+(1.63\times 1\mathrm{if}\mathrm{full}\mathrm{thickness}\mathrm{or}0\mathrm{if}\mathrm{partial}\mathrm{thickness})+(0.000644\times \mathrm{tear}\mathrm{size}\mathrm{area}\mathrm{in}m{m}^2)+(0.0241\times \mathrm{operative}\mathrm{time}\mathrm{in}\mathrm{minutes})+(-2.26\mathrm{for}1\mathrm{if}\mathrm{private}\mathrm{hospital}\mathrm{or}0\mathrm{if}\mathrm{public}\mathrm{hospital})+(0.632\times 1\mathrm{if}\mathrm{worker}s\mathrm{’}\mathrm{compensation}\mathrm{or}0\mathrm{if}\mathrm{none})+(-4.88)$$

Equation 2 (Atraumatic Tears)

Let p be the probability of retear 6 months after arthroscopic RCR.

$$\begin{array}{c}\multicolumn{1}{c}{\mathrm{Logit}p=\mathrm{lo}{g}_e\left(\frac{p}{1-p}\right)=(-0.000587\times \mathrm{case}\mathrm{number})}\\ \multicolumn{1}{c}{+(0.0489\times \mathrm{age}\mathrm{in}\mathrm{years})}\\ \multicolumn{1}{c}{+(0.910\times 1\mathrm{if}\mathrm{full}\mathrm{thickness}\mathrm{or}0\mathrm{if}\mathrm{partial}\mathrm{thickness})}\\ \multicolumn{1}{c}{+(0.00155\times \mathrm{tear}\mathrm{size}\mathrm{in}m{m}^2)+(-5.98)}\end{array}$$

Four hypothetical case scenarios were formulated to demonstrate the utility of these regression equations in calculating the probability of retear (Table 6). For traumatic tears, the probability of retear 6 months postoperatively was lower when the patient was young and had a small, partial-thickness tear that was repaired in a relatively short operation performed by a more experienced surgeon, in a private hospital without workers’ compensation. Similarly, for atraumatic tears, the probability of retear was lower when the patient was younger and had a small, partial-thickness tear that was repaired when the operating surgeon had more experience in RCRs. The mathematical analysis used to calculate retear probability for each scenario is shown in the Appendix.

Table 6.

Probability of Retear 6 Months Postoperatively in 4 Hypothetical Clinical Cases, Calculated Using Regression Equations ^a

Variable	Patient A	Patient B	Patient C	Patient D
	Traumatic	Traumatic	Atraumatic	Atraumatic
Patient age, y	55	65	65	75
Tear size area, mm2	350	500	200	400
Full/partial	Partial	Full	Partial	Full
Case number	1200	800	2000	700
Public/private a	Private	Private	N/A	N/A
Operative time, min	20	40	N/A	N/A
Workers’ compensation	No	Yes	N/A	N/A
Probability of retear, % b	1.9	40.2	2.5	23.3

^aN/A, not applicable. N/A variables were not found to be independent predictors of retear in the atraumatic group and were therefore not included in the regression equation for the referenced group.

^bCalculations using regression equations are shown in the Appendix.

ROC Curve Analysis

In our ROC curve analysis, the area under the curve (AUC) for the traumatic group's regression equation was 0.82 (99% CI, 0.78-0.86), indicating that this equation can accurately predict 82% of retears (Figure 3). Regarding the regression equation for the atraumatic group, the AUC was 0.79 (99% CI, 0.72-0.85), indicating that this equation can accurately predict 79% of retears (Figure 4).

Figure 3.

Receiver operating characteristic curve for regression equation of traumatic group with area under the curve (AUC) data for independent predictors of retear 6 months post–arthroscopic rotator cuff repair. OP_time, operative time; WC1, workers’ compensation claimed.

Figure 4.

Receiver operating characteristic curve for regression equation of atraumatic group with area under the curve (AUC) data for independent predictors of retear 6 months post–arthroscopic rotator cuff repair.

In the traumatic group, tear size area was the most accurate predictor of retear (AUC, 0.76; 99% CI, 0.70-0.81), followed by operative time (AUC, 0.69; 99% CI, 0.64-0.74) and patient age (AUC, 0.68; 99% CI, 0.63-0.73) (Figure 3). Similarly, in the atraumatic group, tear size area was the most accurate predictor (AUC, 0.76; 99% CI, 0.68-0.83), followed by patient age (AUC, 0.67; 99% CI, 0.59-0.75) (Figure 4).

Predictive Thresholds

The ROC curves were used to determine a threshold value for each independent predictor at which retear was more likely to occur (Table 7). The threshold value was defined as the value at which the Youden J statistic was greatest. In the traumatic group, threshold values were 368 mm² for tear size area, 56 years for patient age, and 18 minutes for operative time. In the atraumatic group, threshold values were 297 mm² for tear size area and 70 years for patient age.

Table 7.

Predictive Threshold Values of Independent Predictors of Retear

Independent Predictor	Threshold Value	Youden J Statistic a
Traumatic group
Tear size area	368 mm2	0.42
Patient age	56 y	0.25
Operative time	18 min	0.31
Atraumatic group
Tear size area	297 mm2	0.36
Patient age	70 y	0.30

^aSensitivity (%) + specificity (%) – 100.

Discussion

The major finding of this study was that trauma history and preoperative symptom duration were not independent predictors of retear 6 months after arthroscopic RCR. There was no significant difference in the retear rate between patients with or without a history of trauma, nor was there a significant correlation between preoperative symptom duration and retear rate for both traumatic and atraumatic tears.

Smaller studies have also found a similar risk of retear between traumatic and atraumatic tears. Paul et al ²³ conducted a cohort study with 61 patients and also reported no significant difference in cuff integrity 2 years after RCR between patients with traumatic and nontraumatic tears. However, the mean age of the traumatic and atraumatic groups differed significantly, and the authors did not conduct a multivariate regression analysis to isolate the effect of trauma on cuff integrity. Our study utilizes multivariate regression analysis and a much larger sample size of 2335 patients to more definitively conclude that trauma history does not influence cuff integrity, supporting the findings of a cohort study with 421 patients by Baum et al. ⁴ Our findings align with a smaller study from our institute by Tan et al, ²⁶ which also found that trauma history was not an independent predictor of retear.

In our study, surgeon-ranked tissue quality did not differ significantly between the traumatic and nontraumatic groups (Table 1), providing a possible explanation for similarity in the retear rates between the 2 groups. This observation is consistent with the results presented by Aagaard et al, ¹ who found no significant difference in Bonar score between tendons harvested from patients with a clear history of trauma and those without.

In alignment with our findings, smaller studies by Lobo-Escolar et al ¹⁹ and Harada et al ¹¹ have reported that preoperative symptom duration does not affect cuff integrity after arthroscopic RCR. However, Kim and Kim ¹⁵ found that patients who experienced shoulder symptoms for <12 months before surgery were significantly less likely to retear compared with those who experienced symptoms for ≥12 months. It is important to note that Kim and Kim excluded patients with partial-thickness tears and patients with traumatic rotator cuff tears, compromising the external validity of the study.

While trauma history and symptom duration did not affect cuff integrity, we identified other important predictors of retear that were not identified in the previous study from our institute by Tan et al. ²⁶ Advanced patient age, larger tear size area, full-thickness tear, and lower case number were all predictive of retear 6 months postoperatively in both the traumatic and the atraumatic groups, corroborating the findings of previous studies.^6,8,9,18 Additionally, the performance of RCR in a public hospital and longer operative time negatively affected cuff integrity in the traumatic group, supporting the findings of numerous studies.^5,7,17,29 To our knowledge, this is the first study to assess the impact of workers’ compensation on cuff integrity after RCR. We found workers’ compensation to be a predictor of retear in the traumatic group, in alignment with other studies that have reported poorer functional outcomes of RCR among patients with workers’ compensation.^12,13,16 The regression equations and predictive thresholds identified in Table 7 can be utilized to provide patients with a quantitative understanding of the impact of these variables on retear risk.

Limitations

There are limitations to this study. Data about the exact date of symptom onset or the presence of a specific initiating event is subject to recall bias, as this information could only be acquired by asking patients questions about their shoulder in the preoperative questionnaire. There was no assessment of fatty infiltration using MRI in our study. We did not offer RCR to patients with severe muscular atrophy who would likely fall into the atraumatic group if they were to undergo RCR and be included in the study. We did not examine the effect of other potential risk factors such as smoking and diabetes status on cuff integrity. Additionally, it is likely that some patients retored their rotator cuff after the 6-month follow-up period. However, numerous studies have demonstrated that retear most commonly occurs within the first 6 months after surgery, with an insignificant rise in retear occurrence from 6 months to 2 years postoperatively.^3,14,22

Moreover, different authors may define traumatic and atraumatic differently, affecting the comparability of results between different studies. It is possible that a portion of the traumatic tears in our study may be acute-on-chronic tears, with a degree of preexisting asymptomatic degeneration before a traumatic event caused further tendon injury and characteristic rotator cuff tear symptoms. Nonetheless, our definition is based on the patient's medical history, in alignment with other studies by Cho and Rhee, ⁵ Paul et al, ²³ Baum et al, ⁴ and Tan et al. ²⁶

Conclusion

In summary, trauma history and symptom duration do not influence cuff integrity after arthroscopic RCR, rejecting our hypothesis. Health professionals can advise patients that their risk of retear is affected by other factors such as patient age, tear size, and surgeon experience.

Appendix

Calculations of Probability of Retear 6 Months Postoperatively Using Regression Equations

Patient A:

Using equation 1:

$$\begin{array}{c}\multicolumn{1}{c}{\mathrm{Logit}p=\mathrm{lo}{g}_e\left(\frac{p}{1-p}\right)=(-0.000403\times 1200)+(0.0541\times 55)+(1.63\times 0)+(0.000644\times 350)+(0.0241\times 20)}\\ \multicolumn{1}{c}{+(-2.26\times 1)+(0.632\times 0)+(-4.88)}\end{array}$$

$$\mathrm{lo}{g}_e\left(\frac{p}{1-p}\right)=-3.9407$$

$$\frac{p}{1-p}=e^{-3.9407}$$

$$\frac{p}{1-p}=0.019435$$

$$p=\frac{0.019435}{1+0.019435}$$

$$p=0.019064$$

Therefore, the probability of patient A retearing the tendon 6 months postoperatively is 1.91%.

Patient B:

Using equation 1:

$$\begin{array}{c}\multicolumn{1}{c}{\mathrm{Logit}p=\mathrm{lo}{g}_e\left(\frac{p}{1-p}\right)=(-0.000403\times 800)+(0.0541\times 65)+(1.63\times 1)+(0.000644\times 500)+(0.0241\times 40)}\\ \multicolumn{1}{c}{+(-2.26\times 1)+(0.632\times 1)+(-4.88)}\end{array}$$

$$\mathrm{lo}{g}_e\left(\frac{p}{1-p}\right)=-0.3979$$

$$\frac{p}{1-p}=e^{-0.3979}$$

$$\frac{p}{1-p}=0.67173$$

$$p=\frac{0.67173}{1+0.67173}$$

$$p=0.40182$$

Therefore, the probability of patient B retearing the tendon is 40.2%.

Patient C:

Using equation 2:

$$\mathrm{Logit}p=\mathrm{lo}{g}_e\left(\frac{p}{1-p}\right)=(-0.000587\times 2000)+(0.0489\times 65)+(0.910\times 0)+(0.00155\times 200)+(-5.98)$$

$$\mathrm{lo}{g}_e\left(\frac{p}{1-p}\right)=-3.6655$$

$$\frac{p}{1-p}=e^{-3.6655}$$

$$\frac{p}{1-p}=0.025591$$

$$p=\frac{0.025591}{1+0.025591}$$

$$p=0.024953$$

Therefore, the probability of patient B retearing the tendon is 2.50%.

Patient D:

Using equation 2:

$$\mathrm{Logit}p=\mathrm{lo}{g}_e\left(\frac{p}{1-p}\right)=(-0.000587\times 700)+(0.0489\times 75)+(0.910\times 1)+(0.00155\times 400)+(-5.98)$$

$$\mathrm{lo}{g}_e\left(\frac{p}{1-p}\right)=-1.1934$$

$$\frac{p}{1-p}=e^{-1.1934}$$

$$\frac{p}{1-p}=0.30319$$

$$p=\frac{0.30319}{1+0.30319}$$

$$p=0.23265$$

Therefore, the probability of patient B retearing the tendon is 23.3%