File
Raab AM1, de Groot S2, Berlowitz D3, Post MWM4, Adriaansen J4, Hopman M5, Mueller G1
1Swiss Paraplegic Centre, Clinical Trial Unit, Nottwil, Switzerland, 2Amsterdam Rehabilitation Research Center, Reade, Netherlands, 3Institute for Breathing and Sleep, Austin Health, Heidelberg, Australia, 4Center of Excellence for Rehabilitation Medicine, Brain Center Rudolf Magnus, University, Medical Center Utrecht and De Hoogstraat Rehabilitation, Utrecht, Netherlands, 5Department of Physiology, Radboud University Nijmegen, Nijmegen, Netherlands
Background: A spinal cord injury (SCI) leads to lesion dependent impairment of the respiratory function. Therefore, the models used in able-bodied persons for prediction of respiratory function cannot be transferred to persons with SCI. Since respiratory complications are still a leading cause of death in this population, an individual prediction of respiratory values with consideration of the lesion level would be valuable. In 2012 a reference value calculator was developed to predict lesion dependent respiratory function values in people with SCI. To develop these initial models, 440 people were tested and 150 of these were between six months and two years post injury; a period when lung function is known to improve after a SCI. Thus, we did not know how accurate these models are for people with chronic SCI. Furthermore, there is a lack of validation of these models.
Purpose: To test the accuracy of the 2012 models in persons with chronic SCI and if necessary develop and validate new models to predict respiratory function.
Methods: A retrospective, multicenter cohort study was conducted. Participants were recruited from 10 rehabilitation centers (Switzerland, Netherlands, Australia). Eligible participants were people with a chronic (>2 years), motor complete SCI at C4-T12 and >18 years of age. The outcome measure was respiratory function, determined by lung function and respiratory muscle strength. Candidate predictors were gender, age, height, weight, time post injury, lesion level and smoking, all obtained from medical records before each measurement.
The project is divided into three phases, 1)validation of the models published in 2012, 2)development of new statistical models, 3)validation of the new models.
Phase 1 and 3 were tested by comparing predicted values with actually measured respiratory function values (intraclass-correlation (ICCs), standard error of measurement (SEMs), Bland&Altman). Phase 2 was performed with respiratory function parameters as dependent variables and candidate predictors as independent variables (multiple regression analysis).
Results: 613 participants were included in the study.
The 2012 models were not found to be accurate for all respiratory function parameters using the data form people with chronic injuries. The mean differences between the measured and predicted respiratory function values were high for peak expiratory flow (0.53±1.80L/s), inspiratory (13.05±31.25cmH2O) and expiratory muscle strength (11.41±33.57cmH2O), the ICCs ranged from 0.40 to 0.64 and the SEMs from 0.35 to 13.64.
For the new models the significant predictors for all respiratory function parameters were lesion level, gender and weight (R2 0.69 to 0.78). Time post injury and age had an additive negative effect only on one respiratory function parameter. Smoking history conferred no predictive power.
For the newly developed models the mean differences for respiratory muscle strength were at about 4±36cmH2O, the ICCs between 0.28 and 0.53 and the SEMs between 0.29 and 11.56.
Conclusion(s): The new models for prediction of respiratory function of people with chronic injuries showed better accuracy especially for respiratory muscle strength.
Implications: These prediction models may help clinicians identify patients at higher risk for respiratory complications. Clinicians can then deliver targeted interventions to these patients to prevent reduced respiratory function and ultimately, respiratory complications and death.
Keywords: Spinal cord injury, prediction model, respiratory function
Funding acknowledgements: This project is funded by Wings for Life, a spinal cord research foundation, grant number WFL-CH-017/14.
Purpose: To test the accuracy of the 2012 models in persons with chronic SCI and if necessary develop and validate new models to predict respiratory function.
Methods: A retrospective, multicenter cohort study was conducted. Participants were recruited from 10 rehabilitation centers (Switzerland, Netherlands, Australia). Eligible participants were people with a chronic (>2 years), motor complete SCI at C4-T12 and >18 years of age. The outcome measure was respiratory function, determined by lung function and respiratory muscle strength. Candidate predictors were gender, age, height, weight, time post injury, lesion level and smoking, all obtained from medical records before each measurement.
The project is divided into three phases, 1)validation of the models published in 2012, 2)development of new statistical models, 3)validation of the new models.
Phase 1 and 3 were tested by comparing predicted values with actually measured respiratory function values (intraclass-correlation (ICCs), standard error of measurement (SEMs), Bland&Altman). Phase 2 was performed with respiratory function parameters as dependent variables and candidate predictors as independent variables (multiple regression analysis).
Results: 613 participants were included in the study.
The 2012 models were not found to be accurate for all respiratory function parameters using the data form people with chronic injuries. The mean differences between the measured and predicted respiratory function values were high for peak expiratory flow (0.53±1.80L/s), inspiratory (13.05±31.25cmH2O) and expiratory muscle strength (11.41±33.57cmH2O), the ICCs ranged from 0.40 to 0.64 and the SEMs from 0.35 to 13.64.
For the new models the significant predictors for all respiratory function parameters were lesion level, gender and weight (R2 0.69 to 0.78). Time post injury and age had an additive negative effect only on one respiratory function parameter. Smoking history conferred no predictive power.
For the newly developed models the mean differences for respiratory muscle strength were at about 4±36cmH2O, the ICCs between 0.28 and 0.53 and the SEMs between 0.29 and 11.56.
Conclusion(s): The new models for prediction of respiratory function of people with chronic injuries showed better accuracy especially for respiratory muscle strength.
Implications: These prediction models may help clinicians identify patients at higher risk for respiratory complications. Clinicians can then deliver targeted interventions to these patients to prevent reduced respiratory function and ultimately, respiratory complications and death.
Keywords: Spinal cord injury, prediction model, respiratory function
Funding acknowledgements: This project is funded by Wings for Life, a spinal cord research foundation, grant number WFL-CH-017/14.
Topic: Neurology
Ethics approval required: Yes
Institution: Swiss Paraplegic Centre Nottwil
Ethics committee: Ethics committee northwest and central Switzerland
Ethics number: 2014/13120
All authors, affiliations and abstracts have been published as submitted.
