Prediction of Shoulder Pain in Youth Competitive Swimmers: The Development and Internal Validation of a Prognostic Prediction Model

Abstract

Background: Knowledge of predictors for shoulder pain in swimmers can assist professionals working with the athlete in developing optimal prevention strategies. However, study methodology and limited available data have constrained a comprehensive understanding of which factors cause shoulder pain. Purpose: To investigate risk factors and develop and internally validate a multivariable prognostic model for the prediction of shoulder pain in swimmers. Study Design: Cohort study; Level of evidence, 2. Methods: A total of 201 pain-free club- to international-level competitive swimmers were followed for 2 consecutive seasons. The cohort consisted of 96 male (mean ± SD age, 13.9 ± 2.2 years) and 105 female (13.9 ± 2.2 years) swimmers. Demographic, sport-specific, and musculoskeletal characteristics were assessed every 6 months. Swim-training exposure was observed prospectively. Shoulder pain interfering with training was the primary outcome. Multiple imputation was used to cope with missing data. The final model was estimated using multivariable logistic regression. We applied bootstrapping to internally validate the model and correct for overoptimism. Results: A total of 42 new cases of shoulder pain were recorded during the study. Average duration of follow-up was 1.1 years. Predictors included in the final model were acute:chronic workload ratio (odds ratio [OR], 4.31; 95% CI, 1.00-18.54), competitive level (OR, 0.19; 95% CI, 0.06-0.63), shoulder flexion range of motion, posterior shoulder muscle endurance (OR, 0.96; 95% CI, 0.92-0.99), and hand entry position error (OR, 0.37; 95% CI, 0.16-0.91). After internal validation, this model maintained good calibration and discriminative power (area under the receiver operating characteristic curve, 0.71; 95% CI, 0.60-0.94). Conclusion: Our model consists of parameters that are readily measurable in a swimming setting, allowing the identification of swimmers at risk for shoulder pain. Multivariable logistic regression showed the strongest predictors for shoulder pain were regional competitive swimming level, acute:chronic workload ratio, posterior shoulder muscle endurance, and hand entry error.

Improving Clinical Prognostic Model Methodology: Letter to the Editor

“…Prediction models can play an important role in identifying individuals at risk of experiencing adverse events. In their recent paper, Feijen and colleagues 4 describe the development of a tool to identify swimmers at increased risk for shoulder pain. The authors should be commended for their design and collection of a prospective cohort specifically tailored for prognostic model development, the imputation of missing data, using the full data set for the development and internal validation of the model, and using the Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD) guidelines for methodological reporting.…”

“…Briefly, these include the lack of full reporting of missing data and the missing data mechanism (eg, any assumptions and what was included in the imputation model), lack of a reported intercept to enable other researchers to implement or validate the model, weak assessment of calibration (eg, the Hosmer-Lemeshow test has long been disregarded for assessing calibration), and unclear implementation of bootstrapping methods for internal validation (eg, important to replay all modeling steps, including any variable selection steps), evaluating shrinkage (eg, degree of overfitting and to shrink the regression coefficients), and optimism (eg, adjust model performance measures by this optimism due to any overfitting). 2,9 It is not the motivation of these authors to critique the overall well-designed prospective study of Feijen et al 4 However, as clinical prognostic models can have a direct influence on patient and athlete health, using best practice methods is imperative to improve patient and athlete outcomes, potentially having a direct effect on the athletes' health, well-being, and careers. The authors would be interested in collaborating on redeveloping this model utilizing these methodological considerations to compare and contrast model performance and risk factor inferences.…”

“…Having an adequate sample size for developing a prediction model is an important design feature to mitigate against the risk of overfitting (ie, fitting idiosyncrasies in the data). In their study, Feijen et al 4 fit their model on a cohort of 201 participants with 42 events (individuals with shoulder pain). Using sample size formulae for developing prediction models 6 (though not used in the study by Feijen et al) based on information reported in their study, namely an outcome prevalence of 42 of 201 (0.208), 7 predictor parameters, and the model Cox-Snell R 2 of 0.1 (converting from the reported McFadden R 2 of 0.14), the authors would require 595 participants with 125 events to develop their model.…”

“…7,8,10 Approaches to consider when examining nonlinearity include restricted cubic splines and fractional polynomials. 8,10 Within Feijen et al’s 4 study, further assessment and accounting for the potential nonlinear relationship between acute/chronic workload and pain may have increased model performance. For example, in a study by Carey et al 1 the relationship between athlete acute/chronic workload and injury was shown to have a U shape.…”

“…It is not the motivation of these authors to critique the overall well-designed prospective study of Feijen et al 4 However, as clinical prognostic models can have a direct influence on patient and athlete health, using best practice methods is imperative to improve patient and athlete outcomes, potentially having a direct effect on the athletes’ health, well-being, and careers. The authors would be interested in collaborating on redeveloping this model utilizing these methodological considerations to compare and contrast model performance and risk factor inferences.…”