Clinical Nomogram Predicts Full Thickness Rotator Cuff Tears Using Examination Findings

A prospective clinical investigation evaluating 357 patients presenting with suspected rotator cuff injuries has demonstrated that a nomogram integrating objective strength assessments with routine clinical history can predict full-thickness rotator cuff tears with high diagnostic accuracy. In the cohort evaluated at a specialized shoulder clinic, magnetic resonance imaging verified full-thickness tears in 166 participants, representing an overall diagnostic prevalence of 46.5 percent. Utilizing multivariable logistic regression following initial variable screening, the investigators established that a quantitative deficit in shoulder strength during standardized maneuvers served as the most powerful individual predictor of full tendon discontinuity, followed by daily pain severity, direct trauma history, advancing age, occupational physical labor, and symptom duration. In the primary training cohort, the resulting nomogram achieved an area under the receiver operating characteristic curve of 0.910 alongside an internal concordance index of 0.890. These findings suggest that structured clinical variables collected during initial examinations can provide an accurate framework for stratifying tear probability prior to obtaining advanced magnetic resonance imaging.

Full-thickness rotator cuff tears represent a major cause of shoulder dysfunction and progressive physical disability, frequently necessitating surgical intervention to restore tendon integrity and prevent irreversible structural deterioration. While magnetic resonance imaging remains the established reference standard for diagnosing structural disruption across the rotator cuff, universal imaging at first clinical presentation is constrained by elevated procedural expenses, limited equipment availability, and protracted waiting lists. Conversely, failure to recognize full-thickness tears promptly risks diagnostic delay and worse post-operative outcomes, as untreated tears may enlarge over time and develop progressive fatty infiltration. In current orthopedic practice, clinicians evaluating suspected rotator cuff pathology must determine which individuals warrant immediate diagnostic imaging and which can safely undertake conservative treatment. Previous prognostic investigations focused primarily on isolated risk factors or derived predictive algorithms from specialized anatomical measurements on radiographs, rather than providing clinicians with an accessible, quantified estimate of individualized tear probability at the initial point of care.

To develop a practical clinical prediction instrument, the researchers designed a prospective single-center diagnostic study that consecutively screened 476 patients presenting with shoulder pain between May 2025 and September 2025 at a regional orthopedic referral center. From this initial screening population, senior shoulder specialists diagnosed 390 individuals with rotator cuff injuries based on standardized physical examination maneuvers, including the Jobe test, the external rotation resistance test, and the abdominal compression test. Applying predefined protocol criteria, the study excluded 33 patients due to confounding factors, such as prior shoulder surgery on the affected limb, concomitant fractures, infections, systemic inflammatory disorders like rheumatoid arthritis, neurological impairments altering motor function, and contraindications preventing magnetic resonance scanning. The final analytical sample comprised 357 patients, who were subsequently partitioned at a 7:3 ratio into an internal derivation training set of 249 patients and an independent internal test set of 108 patients, with comparable baseline clinical distributions across both cohorts.

Standardized diagnostic confirmation was established for every enrolled participant using a dedicated 3.0 Tesla magnetic resonance scanner with a specialized shoulder coil. The standardized imaging protocol incorporated T1-weighted, T2-weighted, and proton density-weighted fat-suppressed acquisitions across oblique coronal, oblique sagittal, and axial anatomical planes at a slice thickness of three millimeters with a 0.3-millimeter interslice gap. Full-thickness rotator cuff tear was strictly defined as complete structural discontinuity of tendon fibers extending from the articular surface through to the bursal surface. All acquired imaging series were independently evaluated by two senior shoulder specialists with more than five years of musculoskeletal imaging experience who were blinded to clinical data, with discordant readings conclusively resolved through secondary adjudication by a third specialist. Diagnostic evaluation revealed 166 full-thickness tears, while the remaining 191 cases included 103 partial-thickness tears, 82 cases of isolated tendinopathy, and six non-rotator cuff conditions originally misdiagnosed on physical examination.

Candidate predictor variables collected during baseline examinations encompassed demographic factors, medical comorbidities, pain characteristics, and specialized physical findings. Demographic and historical data included patient age, biological sex, limb dominance, specific trauma preceding symptom onset, occupational physical labor classified as mild or heavy, hypertension, diabetes, coronary heart disease, hyperlipidemia, and tobacco smoking. Pain was quantified through two distinct clinical dimensions: daily pain score, measured on a 10-centimeter visual analog scale assessing average symptoms during daily activities, and pain time, defined as the duration in months from initial pain onset to specialist evaluation. Beyond binary determinations of positive or negative findings during the Jobe, abdominal compression, and external rotation resistance tests, the investigators introduced continuous quantitative metrics, recording the maximum pain intensity produced during testing as the physical examination pain score and patient-level shoulder strength as the lowest percentage of contralateral force observed across the three maneuvers.

To avoid model overfitting while retaining informative clinical variables, the researchers implemented least absolute shrinkage and selection operator regression across 15 candidate predictors within the derivation cohort. Penalized regression analysis eliminated diabetes mellitus, hyperlipidemia, and tobacco smoking due to near-zero predictive coefficients, retaining ten candidate parameters for subsequent regression modeling. Subsequent multivariable logistic regression identified six independent predictors displaying statistically significant associations with confirmed full-thickness tears. Decreased physical examination shoulder strength demonstrated the strongest predictive association, conferring an odds ratio of 3.5 for each 20 percent reduction relative to the healthy contralateral shoulder. An acute history of direct trauma increased the odds of a full tear by 3.1-fold, while occupational physical labor yielded an odds ratio of 2.4. Furthermore, each two-point increment in daily pain score increased tear odds by 2.2-fold, each decade of advancing age elevated odds by 1.9-fold, and each six-month increase in symptom duration was associated with a 1.5-fold increase in tear probability.

The multivariable coefficients were converted into a clinical nomogram that enables practitioners to plot individual predictor values on calibrated axes and aggregate corresponding points to calculate a predicted probability of full-thickness tendon disruption. In the training cohort, the nomogram demonstrated robust discrimination with an area under the receiver operating characteristic curve of 0.910 and an associated 95 percent confidence interval spanning from 0.88 to 0.95. At the optimal diagnostic threshold, the nomogram achieved an empirical sensitivity of 88 percent and a specificity of 82 percent. Internal validation through 1,000 bootstrap resamples demonstrated minimal model optimism, yielding an apparent area under the curve of 0.907 and an optimism-corrected area under the curve of 0.892. Calibration curves confirmed strong agreement between predicted probabilities and observed clinical frequencies in both derivation and test cohorts, while decision curve analysis indicated positive net clinical benefit across a broad spectrum of decision thresholds.

The quantitative dominance of physical examination shoulder strength over examination pain intensity provides meaningful insight into the biomechanical presentation of tendon disruption. While average daily pain scores correlated independently with tear presence, localized pain elicited during physical examination maneuvers failed to maintain predictive significance in multivariable modeling. The authors noted that profound weakness on resisted isometric testing directly reflects substantial loss of mechanical tendon integrity, whereas localized pain is prone to distortion by subjective pain tolerance, emotional distress, or transient analgesic usage. Consequently, patients exhibiting high pain levels alongside preserved physical strength remain relatively unlikely to harbor full-thickness tears, whereas individuals demonstrating disproportionate weakness despite modest pain scores exhibit high structural risk. Quantifying strength as an objective percentage deficit against the asymptomatic limb therefore supplies diagnostic information that surpasses conventional binary categorizations of strength loss.

To assist clinical decision-making, the authors translated the continuous predictive output of the nomogram into three operational risk tiers designed to guide diagnostic workflows. Patients classified within the high-risk stratum, defined by a predicted tear probability exceeding 70 percent, exhibit high model specificity and should be prioritized for prompt magnetic resonance imaging and early surgical evaluation to avoid progression and fatty degeneration. Conversely, individuals falling into the low-risk category with a calculated probability below 20 percent may safely undertake an initial trial of non-operative conservative management lasting eight to twelve weeks without immediate cross-sectional imaging, reserving magnetic resonance studies for instances of persistent or deteriorating symptoms. Patients assigned to the intermediate-risk range of 20 to 70 percent represent a heterogeneous population in which clinical decisions should incorporate patient preferences, functional expectations, and symptom severity alongside physician judgment.

Several methodological limitations must be considered when interpreting the performance and clinical applicability of the proposed nomogram. The model was constructed and internally validated within a single tertiary orthopedic referral institution where the baseline prevalence of full-thickness tears was 46.5 percent, a figure reflecting an enriched referral population that may not reflect primary care or general community settings. Additionally, shoulder strength was evaluated semi-quantitatively against the contralateral limb by a single experienced examiner without formal testing of inter-observer or intra-observer reliability. Because the model entails an approximate 12 percent false-negative rate and an 18 percent false-positive rate, it cannot entirely replace comprehensive clinical assessments. The investigators emphasize that independent external validation across diverse, multi-center healthcare environments and broader primary care cohorts remains necessary before the nomogram can be adopted as a standardized triage instrument in general clinical practice.

Subject of Research: Clinical prediction modeling for full-thickness rotator cuff tears

Article Title: Development and validation of a model to predict full thickness rotator cuff tear

Article References: Wang, T., Tian, Q., Lu, B., Guo, J., Yin, B., & Zhu, Y. (2026). Development and validation of a model to predict full thickness rotator cuff tear. Archives of Orthopaedic and Trauma Surgery, 146(1), Article 355. https://doi.org/10.1007/s00402-026-06521-y

Image Credits: AI Generated

DOI: 10.1007/s00402-026-06521-y

Keywords: rotator cuff tear, clinical prediction model, nomogram, magnetic resonance imaging, shoulder strength assessment, physical examination, orthopedic surgery, least absolute shrinkage and selection operator, multivariable logistic regression, trauma history, daily pain score, symptom duration, internal validation, diagnostic triage, tendon discontinuity, receiver operating characteristic