Does fat infiltration of the rotator cuff muscles contribute to increased hidden blood loss (HBL) and other risk factors during arthroscopic rotator cuff repair for rotator cuff injuries?

Abstract

Objective

This study aimed to compare hidden blood loss (HBL) in patients undergoing arthroscopic rotator cuff repair based on the degree of rotator cuff fatty infiltration, and to determine whether fatty infiltration is associated with increased HBL and other risk factors.

Methods

A retrospective analysis included 141 patients undergoing arthroscopic rotator cuff repair between January 2023 and March 2025. Patients were grouped by rotator cuff fatty infiltration severity (Grades 0–4). Demographics and blood parameters were recorded. Visible blood loss was quantified, and hidden blood loss (HBL) was calculated using preoperative hematocrit (Hctpre) and postoperative hematocrit (Hctpost) to assess total blood loss. Relationships with HBL were analyzed using Pearson/Spearman correlation, Mann-Whitney U test, and multivariate linear regression to identify independent risk factors.

Results

Multivariate linear regression analysis identified the following independent risk factors for HBL: Hctpost (β = -482.318. 95% CI: -814.825to -149.811, p = 0.005), intraoperative blood loss (β = 0.679, 95% CI: 0.444 to ,0.915, P < 0.001), the classification of fatty infiltration (β = 28.279, 95% CI: 13.774to 42.783, P < 0.001), and the grading of tear size (β = 20.954, 95% CI: 3.580 to 38.329, p = 0.018). Notably, the HBL increased significantly with the classification of fatty infiltration(P < 0.001), with median (IQR) values as follows: Stage 0: 293.65(234.20-331.08) ml, Stage 1: 339.66(264.78-376.54) ml, Stage 2: 355.21(268.05-450.66) ml, Stage 3:462.92(340.57-517.23) ml, and Stage 4: 512.23(431.12-575.98)ml.

Conclusion

Patients exhibiting higher classification of fatty infiltration in the rotator cuff muscles, as well as those with more extensive rotator cuff tears, may be associated with increased levels of hidden blood loss (HBL) during arthroscopic rotator cuff repair. Factors such as Hctpost, intraoperative bleeding volume, severity of fatty infiltration, and tear size have been suggested as potential independent risk factors for HBL. Close perioperative monitoring of these parameters could contribute to improved patient safety during arthroscopic rotator cuff repair procedures.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12891-025-09310-7.

Background

Rotator cuff tears (RCTs) represent a widespread musculoskeletal disorder with considerable global prevalence, contributing significantly to shoulder pain and functional impairment [1, 2]. This condition markedly affects patients’ daily lives, frequently leading to reduced range of motion (ROM), diminished quality of life (QoL), and substantial limitations in activities of daily living (ADL) and occupational functions [3–6]. The incidence of Rotator cuff tears exhibits a strong correlation with advancing age; studies indicate that approximately 25% of individuals over 60 years are affected, with prevalence escalating to over 45% in those beyond 70 years of age [7]. For patients requiring surgical management, arthroscopic rotator cuff repair (RCR) has emerged as a well-established intervention, consistently demonstrating favorable to excellent subjective outcomes and high rates of patient satisfaction in clinical reports [8].

However, sports medicine physicians frequently overlook the undetectable blood loss that occurs during arthroscopic rotator cuff repair surgery. This blood loss, commonly referred to as hidden blood loss (HBL), can lead to unexpected postoperative complications.The primary cause of HBL in this context is the extravasation of blood into the interstitial tissue, and its volume is not accounted for in either intraoperative estimated blood loss (EBL). Excessive HBL may negatively impact the patient’s overall condition, potentially resulting in postoperative hemodynamic instability, delayed need for blood transfusion, and prolonged recovery. A comprehensive understanding of all sources of blood loss during arthroscopic rotator cuff repair, along with accurate measurement of HBL, is therefore essential to prevent such complications.

While HBL has been well-documented in spine and lower limb arthroplasty, these findings may not be directly extrapolatable to shoulder arthroscopy [9]. This is due to fundamental differences in anatomy, rich collateral circulation, surgical positioning (beach-chair position), and the routine non-use of a tourniquet, all of which result in distinct perioperative hemodynamics.Currently, evidence on HBL following shoulder arthroscopy is scarce, and there is a particular lack of systematic analysis regarding its association with specific rotator cuff pathologies, such as tear size and the extent of fatty infiltration. Crucially, the relationship between fatty infiltration—a unique degenerative marker for rotator cuff tears—and HBL could not be investigated in lower limb studies.The novelty of this study lies in its comprehensive prospective quantification of HBL in arthroscopic rotator cuff repair, with a specific focus on elucidating its quantitative relationships with both tear size and fatty infiltration classification. However, detailed prospective data on HBL and its specific determinants in this surgical context are limited. This research therefore addresses an important aspect of perioperative blood management in shoulder surgery by providing evidence to help identify patients at high risk for significant HBL.

From a pathophysiological perspective, fatty infiltration may influence HBL through several mechanisms. First, it is hypothesized that fatty infiltration, which is frequently accompanied by tendon degeneration, may also involve local inflammatory responses and elevated levels of angiogenic factors such as VEGF. These factors could promote pathological angiogenesis, potentially resulting in the formation of fragile, structurally incomplete blood vessels that are more prone to rupture during surgery and difficult to control with electrocoagulation. Second, another speculative mechanism involves pro-inflammatory cytokines (e.g., IL-6 and TNF-α), which are potentially elevated with fatty infiltration. By impairing platelet function and interfering with coagulation, these cytokines couldpotentially increase the risk of intraoperative and postoperative microvascular bleeding.intraoperatively and postoperatively. Third, in cases of severe fatty infiltration (Goutallier stage ≥ 3), the tendon quality is markedly compromised. Based on these mechanisms, we hypothesize a positive correlation between fatty infiltration classification and HBL, though this requires validation in shoulder arthroscopy.

Methods

Inclusion and exclusion criteria

Inclusion criteria

 

1. The symptoms and signs were consistent with the clinical manifestations of rotator cuff injury, including anterior-lateral or lateral shoulder pain and significant nocturnal pain;

2. Confirmed rotator cuff injury with imaging evidence, with or without fatty infiltration of the rotator cuff muscles;

3. The selected surgical method was arthroscopic rotator cuff repair;

4. All surgical procedures were performed by the same experienced surgeon;

5. Availability of complete medical records, including detailed documentation of comorbidities, medications, and intraoperative details.

Exclusion criteria

1. Patients diagnosed with hematological disorders, significant hepatic or renal dysfunction, or requiring long-term anticoagulant therapy;

2. Patients who received antiplatelet agents (e.g., aspirin, clopidogrel) or regular non-steroidal anti-inflammatory drugs (NSAIDs) within 10 days prior to surgery; or anticoagulants (e.g., warfarin, Rivaroxaban, Dabigatran), or those (e.g.,venous thromboembolism, and secondary prevention of coronary artery disease)who required any form of bridging anticoagulation therapy(e.g., low molecular weight heparin) during the perioperative period.

3. Patients who received corticosteroid injections into the affected shoulder within 3 months prior to surgery;

4. Patients who received intravenous or topical tranexamic acid or other hemostatic agents during the perioperative period;

5. Patients who received open surgical repair for rotator cuff injury;

6. Patients who underwent revision rotator cuff surgery.

Data extraction

A total of 141 patients diagnosed with rotator cuff injuries and treated at Hebei Provincial People’s Hospital between January 2023 and March 2025 were enrolled in this study. The patients ranged in age from 17 to 81 years, with a mean age of 57.83 ± 10.01 years. The cohort included 97 females and 44 males. Based on the severity of fat infiltration in the rotator cuff muscles, patients were categorized into five groups: no fat infiltration (n = 22), stage 1 fat infiltration (n = 51), stage 2 fat infiltration (n = 40), stage 3 fat infiltration (n = 18), and stage 4 fat infiltration (n = 10). Fat infiltration was classified using a 0–4 classification system. Demographic and clinical data, including gender, age, height, weight, hypertension (defined as blood pressure ≥ 140/90 mmHg), diabetes mellitus, duration of surgery, length of hospital stay, and American Society of Anesthesiologists (ASA) physical status classification, were collected. Additionally, perioperative parameters such as intraoperative blood loss (IBL), preoperative and postoperative hematocrit levels, total blood loss (TBL), and hidden blood loss (HBL) were also recorded.

The size of rotator cuff tears was classified during surgery according to the Cofield classification system [10]. Tears were categorized as follows: small (≤ 1 cm), medium (> 1 to 3 cm), large (> 3 to 5 cm), or massive (> 5 cm).Fatty infiltration of the supraspinatus muscle was assessed using magnetic resonance imaging (MRI) and classified according to the Goutallier classification [11]. Evaluation was performed on the sagittal Y-view, specifically on the most lateral image where the scapular spine remains in contact with the scapular body. Infiltration was staged as:Stage 0(completely normal muscle with no fatty streaks) stage 1 (presence of minor fatty streaks), stage 2 (significant fatty infiltration but muscle still predominates over fat), stage 3 (equal amounts of fat and muscle), and stage 4 (fat exceeds muscle).

Calculation of HBL

Perioperative HBL was determined according to the methodology described in references [12, 13]. The 0total blood volume (TBV, in ml) was first estimated based on the patient’s sex, height (m), and body weight (kg) using the following formulae:

For male patients:

For female patients:

The total predicted blood loss was then calculated using the equation:

where Hctpre denoted the preoperative hematocrit, postoperative hematocrit (Hctpost) represented the hematocrit measured on postoperative day 1 or 2, and Hctave was the average of Hctpre and Hctpost. This timing was selected because it represented the critical period when hemodilution due to fluid shifts and resuscitation was most pronounced, and it aligned with the standard clinical practice for postoperative monitoring in our institution, allowing for consistent data collection. Furthermore, it was a frequently adopted timeframe in the existing literature on HBL after orthopedic surgery, facilitating comparisons across studies [14].

Finally, the HBL was computed. It was important to clarify that no postoperative drainage tubes were used in any of the patients in this study. Therefore, the HBL was calculated using the following simplified formula: HBL = Total predicted blood loss + Total blood transfusion volume − Intraoperative blood loss (IBL).

Here, the total transfusion volume was converted into milliliters based on the standard that one unit of concentrated red blood cells was equivalent to 200 ml. Intraoperative blood loss (IBL) was defined as the sum of the volume absorbed by soaked gauzes and the fluid collected in the negative-pressure suction container (after subtracting the volume of irrigation fluid used). Minimal wound exudation on the dressings was considered negligible and therefore excluded from the calculation.

Statistical analysis

All statistical analyses were conducted using SPSS version 24.0.Continuous data were presented as mean ± standard deviation or median (interquartile range), as appropriate based on their distribution. Appropriate statistical methods were employed to analyze the relationships between various factors and HBL. The correlations between continuous variables and HBL were examined using Pearson or Spearman correlation analysis. For binary variables, the Mann-Whitney U test was used to compare HBL differences between groups. Variables that showed significant associations in univariate analyses (P < 0.10)were subsequently entered into a multivariable linear regression model to identify independent predictors of HBL.Variables that did not meet this significance threshold were excluded. Multivariate linear regression analysis was performed to determine the independent risk factors associated with HBL. Potential multicollinearity was assessed by examining the variance inflation factor (VIF), and variables with a VIF > 5 were considered to indicate severe multicollinearity and were evaluated for exclusion. Additionally, a correlation matrix was examined as a supplementary measure; pairwise correlations with a coefficient exceeding 0.7 were flagged to aid in the interpretation of potential collinearity between specific variables, but the VIF remained the primary criterion for exclusion. These criteria collectively established the final predictive model. A P-value < 0.05 was considered statistically significant in all statistical tests. For subgroup analyses, if the data were normally distributed and exhibited homogeneity of variances, a one-way analysis of variance (ANOVA) was used for comparisons across multiple groups.If the ANOVA results indicated a statistically significant difference, post-hoc pairwise comparisons were performed using the Tukey HSD test.If the data violated the assumptions of normality or homogeneity of variances, the Kruskal-Wallis H test was employed. Subsequently, if the Kruskal-Wallis H test showed a significant difference, Pairwise comparisons were performed using the Mann-Whitney U test. To control for the increased risk of Type I error due to multiple comparisons, the Bonferroni correction was applied. To assess the stability and potential overfitting of the multivariate linear regression model, internal validation was performed using the bootstrap method with 1000 resamples. And a random seed was set to 20,251,001 to ensure complete reproducibility of the results. This technique was employed because it is particularly valuable for assessing model reliability with limited sample sizes, as it does not rely on stringent distributional assumptions. It works by repeatedly sampling from the original dataset with replacement to create numerous simulated samples, thereby empirically estimating the sampling distribution and stability of the regression coefficients. This technique allows for the estimation of the accuracy (bias, precision) of the regression coefficients by simulating the process of repeated sampling from the population. The bootstrap-corrected coefficients and the bias-corrected and accelerated (BCa) 95% confidence intervals were calculated and reported. The BCa interval is a refined method that adjusts for both bias and skewness in the bootstrap distribution, providing a more accurate and reliable confidence interval estimate than standard methods.

Perioperative fluid management

To minimize the potential confounding effect of perioperative fluid balance on hematocrit measurements and the subsequent calculation of HBL, a standardized perioperative fluid management protocol was rigorously followed for all enrolled patients. As part of this protocol, all patients received standardized antibiotic prophylaxis with 2 g of cefazolin in 100mL of normal saline intraoperatively and an additional dose postoperatively. Intraoperatively, fluid resuscitation was primarily conducted using lactated Ringer’s solution, guided by continuous monitoring of hemodynamic parameters (including blood pressure and heart rate) and the maintenance of a urine output greater than 0.5 mL/kg/h. The administration of colloids and blood products was strictly contingent upon meeting predefined clinical criteria. In the postoperative phase, patients received a standardized regimen of maintenance intravenous crystalloids (at a rate of 1.5–2.0.5.0 mL/kg/h). Additionally, as part of the standardized analgesic protocol, patients received ketorolac tromethamine for pain control, administered every 8 h. Intravenous fluids were continued until adequate oral intake was resumed. Fluid balance, encompassing total intake (including antibiotic, analgesic, and maintenance fluids) and output, was meticulously monitored and recorded for the initial 24-hour period following surgery. This comprehensive protocol was designed to ensure a consistent and physiologically appropriate fluid status across the patient cohort, thereby enhancing the reliability of HBL estimation.

Results

A retrospective analysis was performed on data from 141 patients diagnosed with rotator cuff injuries who underwent rotator cuff repair surgery between January 2023 and March 2025. All demographic characteristics are presented in Table 1.

Table 1.

Patient baseline characteristics

Variables	Cases (n = 141)
Age (years)	57.83 ± 10.01
Sex (n)
Male	44
Female	97
Height (m)	1.62 ± 0.08
Weight (kg)	69.61 ± 15.12
BMI (kg/m2)	26.21 ± 4.04
Underlying disease (n)
Hypertension	55
Diabetic mellitus	40
ASA classification (n)
II	106
III	35
Operative time (min)	84.04 ± 32.15
Hospitalization time (d)	7.66 ± 2.63
Classification of fat infiltration in the rotator cuff muscles(n)
0	22
1	51
2	40
3	18
4	10
Grading of rotator cuff tear size(n)
small	34
medium	58
large	40
massive	9
PT	10.65 ± 0.63
PT INR	0.92 ± 0.06
PLT	229.67 ± 52.55
TBL	454.56 ± 171.50
HBL	355.31 ± 119.08
Intraoperative bleeding	98.78 ± 70.67

Continuous data are presented as mean ± standard deviation.The following classification systems were used: (1) ASA physical status: Class 1, A normal healthy patient; Class 2, A patient with mild systemic disease; Class 3, A patient with severe systemic disease that is not incapacitating; Class 4, A patient with an incapacitating systemic disease that is a constant threat to life. (2) Fatty infiltration of the rotator cuff muscles: Stage 0, Completely normal muscle with no fatty streaks; Stage 1, Presence of minor fatty streaks; Stage 2, Significant fatty infiltration but muscle still predominates over fat; Stage 3, Equal amounts of fat and muscle; Stage 4, Fat exceeds muscle. (3) Rotator cuff tear size: Small (≤1 cm), Medium (>1 to 3 cm), Large (>3 to 5 cm), Massive (>5 cm)

BMI Body mass index, ASA American Society of Anesthesiologists, PT Prothrombin time, INR International normalized ratio, PLT Platelet count, TBL Total blood loss, HBL Hidden blood loss

Statistical analysis revealed several factors significantly associated with HBL. As shown in Table 2, Pearson or Spearman correlation analysis of HBL in patients demonstrated that Hctpost (P < 0.001), intraoperative bleeding (P < 0.001), classification of fat infiltration (P < 0.001), grading of rotator cuff tear size (P = 0.008), and the number of anchor screws (P < 0.001) were significantly correlated with HBL. Additionally, Mann-Whitney U tests showed that HBL was significantly higher in patients who underwent acromioplasty (P = 0.037) and intraoperative release (P = 0.042), as well as in those with hypertension (P = 0.011).

Table 2.

Correlates of hidden blood loss (HBL) in arthroscopic rotator cuff repair

Variables	Group	HBLmedian [IQR] (mL)	Correlation Coefficient	p
Gender	-	-	−0.047	0.577
Age	-	-	0.131	0.121
Weight (kg)	-	-	0.095	0.262
Height (cm)	-	-	0.062	0.469
BMI	-	-	0.026	0.763
Hypertension	-	-	0.214	0.011
Diabetic mellitus	-	-	0.119	0.161
Operation time	-	-	−0.059	0.484
Hospitalization time	-	-	−0.048	0.574
ASA classification	-	-	0.083	0.330
Hctpre	-	-	−0.103	0.226
Hctpost	-	-	−0.388	< 0.001
PT	-	-	0.012	0.886
PT INR	-	-	0.026	0.759
PLT	-	-	−0.018	0.831
Intraoperative bleeding	-	-	0.60	< 0.001
Total blood volume	-	-	0.078	0.361
Classification of fat infiltration in the rotator cuff muscles	-	-	0.435	< 0.001
Grading of rotator cuff tear size	-	-	0.222	0.008
The number of anchor screws	-	-	0.343	< 0.001
acromioplasty	No	332.360(260.450,378.035)	0.191	0.037
	Yes	355.205(296.048,474.608)
Intraoperative release	No	330.630(263.225,385.910)	0.184	0.042
	Yes	360.755(285.485,477.015)

BMI Body mass index, ASA American Society of Anesthesiologists, Hctpre preoperative hematocrit, Hctpost postoperative hematocrit, PT Prothrombin time, INR International normalized ratio, PLT Platelet count, IQR Interquartile Range

The results of the multivariate linear regression analysis are presented in Table 3. Although the correlation matrix indicated associations between intraoperative bleeding and both tear size grade (r = 0.263) and fat infiltration (r = 0.396), the variance inflation factors (VIFs) for the three variables were low (range: 1.304–1.599), well below the threshold of 5. This confirms that despite their clinical importance and some correlation, severe multicollinearity was not present in the final model, and each variable contributes independent information. However, when the number of anchors was included alongside the rotator cuff tear size in the multiple linear regression model, variance inflation factor analysis detected severe multicollinearity (VIF were 6.183 and 7.302,respectively, above the accepted threshold of 5). Given that severe multicollinearity undermines the stability and interpretability of regression coefficient estimates, following established statistical principles, only the rotator cuff tear size was retained in the final predictive model. Multivariate linear regression analysis revealed that Hctpost (P = 0.005), intraoperative bleeding (P < 0.001), classification of fat infiltration in the rotator cuff muscles (P < 0.001), and grading of rotator cuff tear size (P = 0.043) had significant effects on HBL in patients.Hctpost (P = 0.005) exerted a statistically significant negative effect on patients’ HBL. Intraoperative bleeding (P < 0.001), classification of fat infiltration in the rotator cuff muscles (P < 0.001), and grading of rotator cuff tear size (P = 0.043) were all significantly and positively associated with patients’ HBL. However, hypertension (P = 0.401), acromioplasty (P = 0.668) and Intraoperative release (P = 0.612)did not demonstrate a statistically significant relationship with HBL.

Table 3.

Multivariate linear regression analysis of independent risk factors for hidden blood loss (HBL)

Coefficients for HBL	Unstandardized		Standardized	t	p	VIF
	Beta	SE	Beta
Hypertension	5.189	6.154	0.055	0.843	0.401	1.101
Hctpost	−492.838	172.697	−0.191	−2.854	0.005	1.171
Intraoperative bleeding	0.665	0.124	0.395	5.371	< 0.001	1.406
Classification of fat infiltration in the rotator cuff muscles	27.447	7.562	0.257	3.830	< 0.001	1.304
Grading of rotator cuff tear size	21.886	10.731	0.160	2.040	0.043	1.599
acromioplasty	8.087	18.784	0.034	0.431	0.668	1.630
Intraoperative release	−9.966	19.590	−0.042	−0.509	0.612	1.771

Hctpost postoperative hematocrit, VIF Variance Inflation Factor

The bootstrap internal validation with 1000 replicates demonstrated remarkable stability and robustness of the model estimates. The bootstrap-corrected mean coefficients were identical to the original coefficients derived from the initial model (Table 4). Furthermore, the bias-corrected and accelerated (BCa) 95% confidence intervals for both fatty infiltration [12.951,43.268] and tear size [3.735,38.174] did not include zero.In practical terms, the exclusion of zero from the BCa 95% confidence interval provides strong evidence that the variable is a statistically significant independent predictor at the 5% significance level. This finding offers robust evidence that these associations are not due to overfitting or chance findings in the particular sample, reinforcing the validity of our model.

Table 4.

Multivariable linear regression analysis for factors associated with hidden blood loss (HBL) with bootstrap validation

Variables	Original Model			Boostrop Validation
	B(95%CI)	β	P-value	B Mean	Bias	BCa 95%CI
Hctpost	−482.318(−814.825,−149.811)	−0.187	0.005	−482.318	−5.279	(−840.444,−146.370)
Intraoperative bleeding	0.679(0.444,0.915)	0.403	< 0.001	0.679	−0.003	(0.461,0.879)
Classification of fat infiltration in the rotator cuff muscles	28.279(13.774,42.783)	0.265	< 0.001	28.279	−0.440	(12.951,43.268)
Grading of rotator cuff tear size	20.954(3.580,38.329)	0.153	0.018	20.954	−0.210	(3.735,38.174)

The bootstrap mean coefficients for significant variables were identical to the original coefficients, indicating exceptional model stability

Hctpost postoperative hematocrit, B unstandardized coefficient, β standardized coefficient, CI Confidence interval, BCa Bias-corrected and accelerated

To further elucidate the relationship, a subgroup analysis comparing HBL across the different fatty infiltration stages was performed. Since the HBL data across the fatty infiltration stage groups did not meet the assumption of normality, the Kruskal-Wallis H test was employed for comparisons. The results showed a highly statistically significant difference in HBL among the different fatty infiltration stages (H(4) = 28.661, p < 0.001). Pairwise comparisons were performed using the Mann-Whitney U test. To control for the increased risk of Type I error due to multiple comparisons, the Bonferroni correction was applied, resulting in an adjusted significance level of p < 0.005. Specific results were as follows: statistically significant differences in HBL were found between Stage 0 and Stage 3 (p < 0.001), Stage 0 and Stage 4 (p < 0.001), Stage 1 and Stage 3 (p = 0.002), Stage 1 and Stage 4 (p = 0.001), and Stage 2 and Stage 4 (p = 0.005). The difference between Stage 0 and Stage 2 was on the borderline of statistical significance (p = 0.006). The remaining pairwise comparisons (Stage 0 vs. 1, p = 0.018; Stage 1 vs. 2, p = 0.157; Stage 2 vs. 3, p = 0.069; Stage 3 vs. 4, p = 0.133) did not reach the adjusted significance level. Overall, a trend of increasing HBL with advancing fatty infiltration stages was observed (Table 5).

Table 5.

Comparison of hidden blood loss (HBL) across fatty infiltration classification

Classification of fat infiltration in the rotator cuff muscles	n	HBLmedian [IQR] (ml)	P-value (vs. 0)	P-value (vs. 1)	P-value (vs. 2)	P-value (vs. 3)	P-value (vs. 4)
0	22	293.65(234.20,331.08)	N	0.018	0.006	< 0.001	< 0.001
1	51	339.66(264.78,376.54)	0.018	N	0.157	0.002	0.001
2	40	355.21(268.05,450.66)	0.006	0.157	N	0.069	0.005
3	18	462.92(340.57,517.23)	< 0.001	0.002	0.069	N	0.133
4	10	512.23(431.12,575.98)	< 0.001	0.001	0.005	0.133	N

Fatty infiltration of the rotator cuff muscles: Stage 0, Completely normal muscle with no fatty streaks; Stage 1, Presence of minor fatty streaks; Stage 2, Significant fatty infiltration but muscle still predominates over fat; Stage 3, Equal amounts of fat and muscle; Stage 4, Fat exceeds muscle

HBL Hidden blood loss, IQR Interquartile Range

To further elucidate the relationship, a subgroup analysis comparing HBL across the different tear sizes was performed. A one-way ANOVA confirmed a statistically significant overall difference (F(3, 137) = 7.517, p < 0.001). Post-hoc analysis using the Tukey HSD test revealed a distinct pattern: there were no statistically significant differences in HBL between small, medium, and large tears (all p > 0.05). However, HBL in massive tears was significantly greater than in each of the other three groups (small vs. massive: mean difference = −193.66 ml, p < 0.001; medium vs. massive: mean difference = −175.83 ml, p < 0.001; large vs. massive: mean difference = −160.34 ml, p = 0.001). This indicates that the significant increase in HBL is primarily driven by the presence of massive tears.

To enhance clinical interpretability, we calculated the percentage of HBL relative to total blood loss (HBL%) for each fatty infiltration stage (Table 6). The HBL% was 82.51%(73.51%−86.77%) for Stage 0, 83.02%(77.56%−86.06%) for Stage 1, 84.42(75.24%−86.42%) for Stage 2, 76.09%(71.33%−81.24%) for Stage 3, and 73.21%(65.77%−81.98%) for Stage 4.

Table 6.

Comparison of hidden blood loss (HBL) by rotator cuff tear size

Tear Size	n	HBL(ml)	P-value (vs. Small)	P-value(vs. Medium)	P-value (vs. Large)	P-value(vs. Massive)
Small	34	326.17± 114.44	N	0.881	0.577	< 0.001
Medium	58	344.00± 94.25	0.881	N	0.906	< 0.001
Large	40	359.48± 125.21	0.577	0.906	N	0.001
Massive	9	519.82± 138.99	< 0.001	< 0.001	0.001	N

Continuous data are presented as mean ± standard deviation. Rotator cuff tear size: Small (≤ 1 cm), Medium (> 1 to 3 cm), Large (> 3 to 5 cm), Massive (> 5 cm)

HBL Hidden blood loss

To enhance clinical interpretability, we calculated the percentage of HBL relative to total blood loss (HBL%) for each fatty infiltration stage in Table 7. The HBL% was 82.51%(73.51%-86.77%) for Stage 0, 83.02%(77.56%-86.06%) for Stage 1, 84.42(75.24%-86.42%) for Stage 2, 76.09%(71.33%-81.24%) for Stage 3, and 73.21%(65.77%-81.98%) for Stage 4.

Table 7.

Hidden blood loss percentage (HBL%) by fatty infiltration classification

Fatty Infiltration Classification	n	HBL%median [IQR]
Stage 0	22	82.51(73.51–86.77)
Stage 1	51	83.02(77.56–86.06)
Stage 2	40	84.42(75.24–86.42)
Stage 3	18	76.09(71.33–81.24)
Stage 4	10	73.21(65.77–81.98)

HBL% = (Hidden Blood Loss/Total Blood Loss) × 100%. Fatty infiltration of the rotator cuff muscles: Stage 0, Completely normal muscle with no fatty streaks; Stage 1, Presence of minor fatty streaks; Stage 2, Significant fatty infiltration but muscle still predominates over fat; Stage 3, Equal amounts of fat and muscle; Stage 4, Fat exceeds muscle

HBL Hidden blood loss, IQR Interquartile Range

Kruskal-Wallis H test revealed statistically significant differences among groups (H(4) = 14.559, p ＝ 0.006). Post-hoc comparisons with Bonferroni correction (α = 0.005) showed that HBL% in Stage 1 was significantly higher than in Stage 3 (p = 0.004), and the difference between Stage 2 and Stage 4 approached significance (p = 0.005). No other pairwise comparisons were significant after correction (Table 8).

Table 8.

Pairwise comparisons of hidden blood loss percentage (HBL%) between stages

Comparison	Mann-Whitney U	p-value	Significant after Bonferroni Correction (α = 0.005)
Stage 0 vs. Stage 1	549	0.885	No
Stage 0 vs. Stage 2	411	0.670	No
Stage 0 vs. Stage 3	132	0.075	No
Stage 0 vs. Stage 4	63	0.058	No
Stage 1 vs. Stage 2	958	0.620	No
Stage 1 vs. Stage 3	250	0.004	Yes
Stage 1 vs. Stage 4	116	0.006	No
Stage 2 vs. Stage 3	209	0.010	No
Stage 2 vs. Stage 4	88	0.005	Yes (Borderline)
Stage 3 vs. Stage 4	75	0.494	No

Overall significance: Kruskal-Wallis H(4) = 14.559, p = 0.006. Bonferroni correction applied for 10 comparisons

As shown in Fig. 1, HBL demonstrates a significant increasing trend with higher classification of fat infiltration. Moreover, patients with massive rotator cuff tears have significantly greater HBL compared to those with small or medium tears (Fig. 2).

Fig. 1.

Comparison of Hidden Blood Loss (HBL) across different fat infiltration classification

Fig. 2.

Comparison of Hidden Blood Loss (HBL) across different rotator cuff tear sizes

Discussion

This study confirms that the classification of fatty infiltration is an independent predictor of perioperative HBL. Our research did not directly investigate the underlying biological mechanisms. Nevertheless, the strong association we observed supports the pathophysiological hypotheses proposed in the existing literature. For instance, theories such as pathological angiogenesis and microenvironment inflammation provide a plausible explanatory framework for our finding that more severe fatty infiltration leads to greater HBL. Consequently, our clinical findings provide indirect support for these potential mechanisms at the human level and justify further investigation. Most importantly, the findings of this study add a new dimension to the field of HBL research. Previous studies on hidden blood loss (HBL) have primarily focused on external factors such as operative time. In contrast, our work is the first to systematically demonstrate, in the context of shoulder arthroscopy, that the preoperatively assessed classification of fatty infiltration is an independent and quantifiable predictor for perioperative HBL. This finding shifts the focus of risk assessment to the patient’s intrinsic tissue characteristics. This implies that by using routinely available preoperative MRI scans, clinicians can identify patients at high risk for HBL prior to surgery. This capability provides a direct rationale for implementing individualized perioperative management strategies, such as preoperative anemia optimization and enhanced postoperative monitoring.

An in-depth analysis of the post-hoc pairwise comparisons revealed a non-linear progression pattern. Specifically, no significant difference in HBL was found between low-classification infiltration stages (Stage 0 vs. Stage 1), suggesting a limited impact of fatty infiltration on bleeding during the early disease stages. However, a substantial increase in HBL began to emerge when infiltration progressed to Stage 2 and beyond, evidenced by the comparison between Stage 0 and Stage 2 being on the borderline of statistical significance. Most importantly, significant differences were observed between Stages 3 & 4 and nearly all lower stages (Stages 0, 1, & 2), highlighting severe fatty infiltration as a critical threshold for a sharp increase in HBL risk. A noteworthy finding was that no significant difference was found between Stage 3 and Stage 4 themselves, potentially indicating a “plateau effect” where the impact on HBL stabilizes once fatty infiltration exceeds a certain severity threshold (e.g., Stage 3). They underscore the importance of precise preoperative assessment of the fatty infiltration stage. For patients identified with Stage 2 or higher, particularly Stages 3 and 4, on imaging, surgeons should consider them a high-risk population for HBL and develop individualized blood management strategies. This may include the prophylactic use of antifibrinolytic agents (e.g., tranexamic acid) and preparing for appropriate intraoperative monitoring and blood transfusion.

The grading of rotator cuff tear size also demonstrated a positive correlation with HBL. Subgroup analysis refined this relationship, revealing that the significant increase in HBL is primarily driven by massive tears, with no statistically significant differences observed among small, medium, and large tears. This correlation is likely multifactorial. Massive tears are often accompanied by tendon retraction, adhesion, and concomitant fatty infiltration. Consequently, these complex tears typically require more extensive surgical procedures such as joint capsule release, synovectomy, and acromioplasty, which inevitably lead to greater microvascular injury and tissue trauma. Compared with smaller tears, massive tears involve a significantly larger operative field, making it more challenging to achieve complete hemostasis with electrocoagulation at all potential bleeding sites. Moreover, the repair of massive tears typically necessitates the placement of a greater number of anchors and sutures. Drilling more bone tunnels can lead to increased bleeding from the cancellous bone of the humeral head, which has a rich blood supply. We speculate that this blood, along with postoperative capillary bleeding from the larger repaired wound area, is more likely to seep into the joint cavity and surrounding soft tissues. This may contribute to the formation of what can be conceptualized as a “hidden blood loss pool” in the subacromial space or glenohumeral joint. Therefore, while tear size is a significant predictor overall, clinicians should be particularly vigilant about the risk of substantial HBL specifically in patients with massive rotator cuff tears.

The study uncovered a critical finding: while the number of anchors was significantly correlated with HBL, it exhibited severe statistical collinearity with rotator cuff tear size. This phenomenon is not a methodological limitation but a statistical confirmation of an inherent clinical logic. Specifically, the number of anchors used in rotator cuff repair is not an independent or random decision; rather, it is a direct, quantitative response by the surgeon to the most fundamental pathological characteristic—the size of the rotator cuff tear. Larger, more complex tears intrinsically require a greater number of anchors to achieve biomechanically sound fixation. Consequently, our final model retained rotator cuff tear size as the more fundamental driver. This ensures statistical robustness and allows our conclusions to focus more accurately on the underlying pathophysiological mechanism of HBL, rather than on mere surgical technical details. This positioning endowed the findings of this study with broader scientific significance and clinical generalizability.

Univariate analysis showed that both intraoperative release and acromioplasty were associated with higher HBL. However, the key multivariable regression analysis revealed that after adjusting for the effects of fat infiltration and rotator cuff tear size, these intraoperative procedures were not independent predictors of HBL. This finding has clear clinical rationale: intraoperative release and acromioplasty are typically necessary measures implemented when addressing more severe and complex rotator cuff injuries. Therefore, their impact on HBL essentially stems from the more severe underlying pathological conditions they address, rather than the procedures themselves. The final model of this study demonstrates that fat infiltration and tear size are more fundamental pathophysiological drivers than specific surgical maneuvers.

HBL is a common phenomenon occurring during the perioperative period in various orthopedic surgical procedures. For instance, Smorgick et al. [15] reported that in posterior spinal fusion surgery, the average total blood loss was 1439 ml, with HBL constituting 42% of this volume. Similarly, in a study involving patients undergoing transforaminal lumbar interbody fusion (TLIF) for degenerative lumbar disease, Xu et al. [16] found an average HBL of 362.8 ml, accounting for 47% of the total blood loss.In our study, patients with rotator cuff injuries undergoing arthroscopic rotator cuff repair experienced a mean HBL of 355.31 mL, which accounted for 78.17% of the total blood loss (TBL). Despite its significant contribution to overall blood loss, HBL is frequently overlooked. Excessive HBL can negatively impact the patient’s systemic condition, potentially leading to postoperative hemodynamic instability, the need for blood transfusion, and delayed recovery. Therefore, during the perioperative management of arthroscopic rotator cuff repair, careful monitoring of HBL is essential, particularly in patients with fatty infiltration of the rotator cuff muscles, to ensure patient safety.

Our analysis of HBL% revealed a clinically important and non-linear relationship with fatty infiltration severity. Contrary to initial expectations, the highest proportions of HBL were observed in moderate infiltration stages (Stages 1–2) rather than in the most severe stages (Stage 4). This pattern may be explained by a “dilution effect” in advanced disease. Patients with severe fatty infiltration (Stages 3–4) typically present with more extensive tendon damage and require more complex surgical repairs, leading to substantially greater total blood loss. While the absolute volume of HBL remains clinically significant in these patients, its proportion relative to the dramatically increased total blood loss may be relatively lower. This finding highlights that moderate fatty infiltration (Stages 1–2) represents a critical phase where HBL constitutes the most prominent component of total blood loss. Clinicians should be particularly vigilant during this stage, as the high proportion of hidden loss may lead to underestimation of true blood loss if based solely on visible measures. The decrease in HBL% in severe stages does not diminish the importance of monitoring HBL but rather reflects the overwhelming nature of total blood loss in these complex cases.

When comparing the HBL in this study with values reported for spine and hip/knee arthroplasty, it is crucial to emphasize the fundamental differences inherent to shoulder arthroscopy that dictate its distinct bleeding pattern. Firstly, from an anatomical and hemodynamic perspective, the shoulder region possesses a rich collateral circulation, and surgery is typically performed in the beach-chair position almost invariably without a tourniquet. This stands in stark contrast to lower limb surgeries, which are often performed supine with a tourniquet, creating a model of controlled ischemia-reperfusion injury [17]. Secondly, regarding the nature of the procedure, shoulder arthroscopy is a truly minimally invasive surgery focused on soft tissue repair. In contrast, joint arthroplasty involves extensive bone cutting and marrow cavity exposure, which continuously releases tissue factors and procogulants into the systemic circulation, potentially triggering systemic coagulopathy and fibrinolysis [18]. Consequently, HBL in hip/knee arthroplasty may largely stem from systemic coagulation dysfunction, while HBL in shoulder arthroscopy is predominantly attributed to persistent local oozing from the surgical field. Finally, and most significantly, our core finding links HBL to a pathology specific to the rotator cuff—fatty infiltration—a relationship that cannot be investigated in studies of the lower limb joints. In summary, our findings underscore that the dominant mechanisms of HBL differ across anatomical sites and procedure types, necessitating independent risk assessment and management strategies in the realm of shoulder arthroscopy.

The primary findings of this study are that the fat infiltration in the rotator cuff muscles and rotator cuff tear are two strong and independent predictors of HBL following arthroscopic rotator cuff repair. The robustness of this association is underscored by the bootstrap validation, which showed no bias in the point estimate and yielded a narrow confidence interval that excluded zero. This suggests that the effect of the fat infiltration in the rotator cuff muscles and rotator cuff tear is both nstatistically significant and clinically relevant, with a high degree of certainty.

This study has several limitations. First, as a retrospective study with a relatively small sample size, it may not fully reflect the characteristics of the general population. Further prospective studies with larger sample sizes are required to confirm these findings. Second, perioperative blood dilution resulting from intravenous fluid administration may cause certain laboratory parameters to be underestimated. Third, Some patients experience significant postoperative wound bleeding that is not accounted for in the standard measurement of postoperative blood loss. This omission may compromise the accuracy of bleeding volume calculations based on Hctpost levels, thereby affecting the reliability of the conclusions. Fourth, our assessment of HBL depends on mathematical formulas and indirect measurement methods, which may not precisely reflect the actual magnitude of blood loss.Fifthly, the assessment of fatty infiltration using the Goutallier classification, while clinically relevant, is semi-subjective and might lack the sensitivity of more advanced quantitative MRI techniques to detect subtle associations.Sixly, as a single-center study, the generalizability of our findings may be influenced by specific surgical protocols and patient demographics. Future prospective, multi-center studies utilizing objective fat quantification methods are warranted to validate our conclusions.Seventhly, the calculation of HBL based on hematocrit changes is inherently susceptible to confounding factors, primarily the timing of measurement and perioperative fluid balance. While the conventional timing of measurement (postoperative days 1 or 2) may lead to an underestimation due to ongoing bleeding beyond this period, it is also during this critical window that fluid shifts and administrations can cause hemodilution, potentially leading to an overestimation. Although we implemented a standardized fluid management protocol (as described in the Methods) to mitigate the impact of fluid balance and enhance the internal consistency of our data, it is impossible to completely eliminate individual physiological variations. Therefore, the calculated HBL values should be interpreted as a robust clinical estimate within the context of standardized care, rather than an absolute precise measure.Eighth, the single-surgeon, single-center nature of this study, while promoting procedural consistency, may limit the external validity and generalizability of our findings to other settings with different surgical techniques and patient populations. Ninth,, the sample size was relatively small, particularly within the subgroups of patients with higher classification of fatty infiltration (e.g., Goutallier Classification 4, n = 10), which may affect the statistical power and robustness of the conclusions for these specific cohorts.Tenth, Although we endeavored to minimize the impact of medication on HBL by strictly excluding patients who were on or required perioperative bridging anticoagulation therapy, as a retrospective study, we did not perform systematic preoperative coronary CTA screening on all enrolled patients. Therefore, we cannot entirely rule out the presence of undiagnosed, stable coronary artery disease patients. However, such undiagnosed patients are typically not on anticoagulants; thus, we believe their potential impact on HBL is relatively limited and likely randomly distributed across different groups, which mitigates the bias to our core conclusions to some extent.Eleventh, an important limitation of this study is the lack of subgroup analyses for different tear types (e.g., partial-thickness vs. full-thickness tears, or tears of different configurations). Different tear types likely involve varying repair techniques and surgical dissection, which could exert distinct effects on HBL. We acknowledge that our findings regarding the relationship between tear size and HBL primarily reflect trends within the full-thickness tear population. Future studies that incorporate more detailed tear classification will be able to delineate more precisely the complex relationship between rotator cuff pathology and perioperative blood loss.

Conclusions

Patients exhibiting higher classification of fatty infiltration in the rotator cuff muscles, as well as those with more extensive rotator cuff tears, may be associated with increased levels of HBL during arthroscopic rotator cuff repair. Factors such as Hctpost, intraoperative bleeding volume, severity of fatty infiltration, and tear size have been suggested as potential independent risk factors for HBL. Close perioperative monitoring of these parameters could contribute to improved patient safety during arthroscopic rotator cuff repair procedures.

Authors’ contributions

**Wenyi Li** : Writing–original draft, Validation, Resources, Methodology, Investigation, Conceptualization. **Zekai Sun: ** Writing – review & editing, Writing – original draft, Data curation, Conceptualization, Supervision. **Shangju Gao: ** Writing – review & editing, Investigation, Resources Conceptualization. **Fantao Meng: ** Methodology, Investigation, Conceptualization. **Yingjie Zhang: ** Methodology, Investigation, Conceptualization.

Funding

This work was supported by The People’s Government of Hebei Province [grant numbers ZF2024020].

Data availability

Data is provided within the manuscript or supplementary information files.

Declarations

Ethics approval and consent to participate

This retrospective study was reviewed and approved by the Institutional Review Board (IRB) of Hebei General Hospital (Approval No. 2025-LW-0196). The IRB specifically reviewed and approved the inclusion of participants under the age of 18. As the research involved only the analysis of pre-existing, anonymized data, the IRB formally waived the requirement for obtaining informed consent. All procedures were performed in accordance with the ethical standards of the Declaration of Helsinki.

Consent for publication

All data presented in this manuscript are anonymized and no individual can be identified from the details provided. Therefore, consent for publication is not required.

Competing interests

The authors declare no competing interests.